AIG IN HINDSIGHT ... · ... (“Maiden Lane II”) and credit default swap operations (“Maiden Lane III”). In particular, ... senior credit default swap portfolio.

NBER WORKING PAPER SERIES

AIG IN HINDSIGHT

Robert L McDonaldAnna Paulson

Working Paper 21108httpwwwnberorgpapersw21108

NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue

Cambridge MA 02138April 2015

David Autor Ben Chabot Larry Cordell Mark Finn Scott Frame Chiang-Tai Hsieh Yilin HuangArvind Krishnamurthy Anil Kashyap Andreas Lehnert Debbie Lucas David Marshall Richard MillerRichard Rosen David Scharfstein Robert Steigerwald and Tim Taylor provided helpful commentaryand feedback as did seminar participants at Case Western and the Federal Reserve Banks of NewYork and Chicago We are grateful to Kyal Berends Mike Mei and Thanases Plestis for excellentresearch assistance The views presented here are solely our own and do not reflect those of the FederalReserve Bank of Chicago the Board of Governors of the Federal Reserve System or the NationalBureau of Economic Research

At least one co-author has disclosed a financial relationship of potential relevance for this researchFurther information is available online at httpwwwnberorgpapersw21108ack

NBER working papers are circulated for discussion and comment purposes They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications

copy 2015 by Robert L McDonald and Anna Paulson All rights reserved Short sections of text notto exceed two paragraphs may be quoted without explicit permission provided that full credit includingcopy notice is given to the source

AIG in HindsightRobert L McDonald and Anna PaulsonNBER Working Paper No 21108April 2015JEL No E00G01G18G2

ABSTRACT

The near-failure on September 16 2008 of American International Group (AIG) was an iconic momentin the financial crisis Two large bets on real estate made with funding that was vulnerable to bank-runlike behavior on the part of funders pushed AIG to the brink of bankruptcy AIG used securities lendingto transform insurance company assets into residential mortgage-backed securities and collateralizeddebt obligations ultimately losing at least $21 billion and threatening the solvency of the life insurancecompanies AIG also sold insurance on multi-sector collateralized debt obligations backed by realestate assets ultimately losing more than $30 billion These activities were apparently motivated bya belief that AIGrsquos real estate bets would not suffer defaults and were ldquomoney-goodrdquo We find thatthese securities have in fact suffered write-downs and that the stark ldquomoney-goodrdquo claim can be rejectedUltimately both liquidity and solvency were issues for AIG

Robert L McDonaldKellogg School of ManagementNorthwestern University2001 Sheridan RoadEvanston IL 60208and NBERr-mcdonaldnorthwesternedu

Anna PaulsonFederal Reserve Bank of Chicago230 S LaSalle StreetChicago IL 60604AnnaPaulsonchifrborg

1 Introduction

The near-failure on September 16 2008 of American International Group(AIG) was an iconic moment of the financial crisis AIG a global insuranceand financial company with $1 trillion in assets lost $993 billion during2008 (American International Group Inc 2008a p 194) and was rescuedwith the help of the Federal Reserve the Federal Reserve Bank of NewYork and the Treasury The rescue played out over many months andinvolved the extension of loans the creation of special purpose vehiclesand equity investments by the Treasury with the government assistanceavailable to AIG ultimately totaling $1823 billion The decision to rescueAIG was controversial at the time and remains so AIGrsquos fate also providedan important touchstone in discussions of financial reform AIG motivatedthe enactment of new rules governing non-bank financial institutions aswell as rules about the treatment of financial derivatives

In this paper we begin with an overview of AIGrsquos main corporate fi-nancial indicators from 2006-2009 However most of the attention paidto AIGmdashand our focusmdashconcerns the two main activities that caused theinsurance company to be driven to the edge of bankruptcy by falling realestate prices and mortgage foreclosures AIGrsquos securities lending businessand its credit default swap business Although much of the discussion con-cerning AIG has centered on its credit default swap business we will showthat losses from its securities lending business were of a similar magnitudeOn September 16 2008 the cumulative losses from these two activities wereon the order of $50 billion and both appear to have played important rolesin AIGrsquos near-failure (as also emphasized by Peirce (2014) and Taibbi (2011Chapter 3))

We then turn to a description of the government rescue of AIG includ-ing the special purpose vehicles that the New York Fed created to deal withthe assets related to AIGrsquos securities lending (ldquoMaiden Lane IIrdquo) and creditdefault swap operations (ldquoMaiden Lane IIIrdquo) In particular we examine thewrite-downs on the assets in these portfolios from each assetrsquos inceptionto October 2014 AIGrsquos real estate positions were apparently motivatedby the belief that these investments would not default The analysis shedslight on a claim often made by AIG executives that their mortgage-relatedinvestments might have suffered a decline in their market value in the short-term but that they would pay off over time This claim implicitly attributesany price decline in such securities to short-term illiquidity The head ofAIGFP Joseph Cassano often referred to the mortgage-related securitiesthat AIG insured through credit default swaps as ldquomoney goodrdquo (for ex-

ample see American International Group Inc (2007c)) Mark Hutchingswho ran AIGrsquos securities lending business made similar statements aboutthe real-estate related investments financed by securities lending (Hutch-ings 2010) However this stark claim that assets were ldquomoney goodrdquo isnot borne out a number of AIGrsquos mortgage-related investments sufferedprincipal write-downs

In our concluding section we discuss the question of how to think aboutAIG as a financial firm

It is important to be clear about what we do not do in this paper Wedo not analyze AIGrsquos regulatory oversight prior to the crisis We discusswhat happened in the AIG rescue but we do not analyze alternative poli-cies or capital structures for a rescue We discuss the specific parties whobenefited most from the rescue but we do not address the broad questionof what might have happened to the financial system had AIG failed Therewas certainly reason for concern In testimony about the AIG rescue Fed-eral Reserve Chairman Ben Bernanke (Bernanke 2009) noted that AIG had$20 billion of commercial paper outstanding and $50 billion of exposureto other banks via loans lines of credit and derivatives Lehman Broth-ers had around $57 billion in commercial paper and its failure wreakedhavoc on money market mutual funds (Federal Deposit Insurance Corpo-ration 2011) Policymakers and academics have written extensively aboutpotential systemic consequences from the failure of a large interconnectedfinancial firm like AIG for example V V Acharya Gale and Yorulmazer(2011) Brunnermeier and Pedersen (2009) Kacperczyk and Schnabl (2010)Duarte and Eisenbach (2014) and Ellul et al (2014) among many others

2 AIG Financials 2006-2009

AIG was an international insurance conglomerate with four main linesof business 1) General Insurance including propertycasualty and com-mercialindustrial insurance 2) Life Insurance and Retirement includingindividual and group life insurance and annuities 3) Asset Managementincluding private banking brokerage and investment advisory servicesand 4) Financial Services including a capital markets division consumerfinance and aircraft leasing Looking at that list of lines of business it isnot at all obvious why AIG had significant exposure to risks from fallingreal estate prices and default rates on subprime mortgages

Each year public firms must file with a 10K report with the Securitiesand Exchange Commission with an in-depth presentation of its financial

position In its 2007 10K report AIG listed $106 trillion in assets Amer-ican International Group Inc (2007d p 130) Table 1 presents financialindicators for 2006ndash09 which help to put AIGrsquos 2008 performance intoperspective The firm was showing some reasons for concern in 2007 in-cluding losses in the Financial Services division and unrealized losses inits credit default swap business But in 2008 AIG lost money in all of itsmain lines of business with the largest losses in the Life Insurance andFinancial Services divisions In both cases the losses stemmed from heavybets on real-estate-related financial products The Life Insurance divisionlost money primarily because of securities lending ($21 billion in losses)where life insurance company assets were loaned in exchange for cash thatwas used to invest in mortgage-related securities In the case of financialservices AIG had written credit default swaps on mortgage-related bondslosing $286 billion in 2008 American International Group Inc (2008a p265) The securities lending business will be discussed in the next sectionthe credit default swap business will be discussed in the section after thatAIGrsquos reported 2008 revenue of $111 billion incorporates the losses fromsecurities lending credit default swaps and other sources

3 AIGrsquos Securities Lending Business

During 2008 AIGrsquos life insurance subsidiaries lost approximately $21 billionfrom securities lending in which the life insurance subsidiaries loaned outassets and invested the proceeds in risky assets including assets backed bysubprime residential mortgage loans In this section we discuss AIGrsquos se-curities lending activity which created unique problems because of its linksto AIGrsquos state-regulated life insurance subsidiaries Recently Peirce (2014)has examined the securities lending business in detail We argue that it isimpossible to evaluate the potential consequences of an AIG failure withoutunderstanding AIGrsquos life insurance and securities lending activities

31 What Is Securities Lending

In a securities lending transaction one party borrows a security from an-other and deposits collateral typically cash with the securities lender Theborrower may use the security as part of a short-selling strategy or to de-liver a particular security to a customer The securities lender invests thecash collateral and earns a yield from these investments less a rebate paidto the securities borrower Absent default the lender remains the eco-

Table 1 AIG financial indicators by operating segment 2006-2009 $ Billions

Item 2006 2007 2008 2009

Revenues 11339 11006 1110 9600Earnings 1405 620 -9929 -1231

Realized capital gains 011 -359 -5548 -686Unrealized CDS losses (AIGFP) 0 -1147 -2860 142

Operating IncomeGeneral Insurance 1041 1053 -575 017

Life Insurance amp Retirement Services 1012 819 -3745 204Financial Services 038 -952 -4082 052

Asset Management 154 116 -919 NAAssets

General Insurance 16700 18171 16595 15473Life Insurance amp Retirement Services 55096 61316 48965 55349

Financial Services 20249 19398 16706 13282Asset Management 7828 7727 4685 NA

Source American International Group Inc (2008a p 71 194 and 225) andAmerican International Group Inc (2009 p72 195 and 230)Notes In 2009 results from asset management activities were included inthe Life Insurance amp Retirement Services category Revenue is composed ofpremiums and other income net investment income realized capital gains(or losses) and unrealized CDS losses Earnings are equal to net income (orlosses) as reported on AIGrsquos consolidated statement of income Realizedcapital gains are primarily comprised of sales of securities and other in-vestments foreign exchange transactions changes in the fair value of nonAIGFP derivative instruments that do not qualify for hedge accountingtreatment and other-than-temporary impairments on securities Unreal-ized CDS losses are the unrealized market valuation loss on AIGFPrsquos supersenior credit default swap portfolio Operating income is equal to pre-taxincome (or loss) for each business segment Assets are equal to year-endidentifiable assets for each business segment

nomic owner of the security that is on loan earning its return includingany dividend or coupon payments The cost to the security borrower is thedifference between the return the borrower could have earned investing thecash collateral and the rebate fee which is a market price determined bythe scarcity of the security on loan The term of a securities lending trans-action may extend for various periods up to several months but in manycases either party can terminate the transaction early The borrower canend the transaction by returning the security to the lender at which timethe lender must also return the cash deposit to the borrower A problemcan arise if many borrowers simultaneously decide to end transactions andthe securities lender does not have or cannot raise sufficient cash to meetthese demands in a timely fashion1

32 Characteristics of AIGrsquos Securities Lending

AIGrsquos securities lending activities were conducted ldquoprimarily for the benefitof certain AIG insurance companiesrdquo (AIG 2007b p 108) These activi-ties were centralized in a non-insurance subsidiary AIG Global SecuritiesLending (GSL) which served as an agent for AIGrsquos subsidiary life insur-ance companies The life insurance companies provided securities primar-ily corporate bonds to GSL These securities were loaned to banks andbroker-dealers in return for cash collateral that was invested by GSL Theinvestment proceeds were used to fund the rebate to the security borrowerand the remainder was split 50-50 between GSL and the insurance compa-nies Nearly all of AIGrsquos security loans had a one month term (Hutchings2010)2

AIG expanded its securities lending rapidly in the run-up to 2008 Atthe end of 2003 the firm had less than $30 billion in securities lendingoutstanding At the peak in 2007Q3 AIG had securities lending outstandingof $884 billion (American International Group Inc 2007e p 2) AIG hadsecurities lending of $70 billion the second quarter of 2008 which then fellalmost to zero by the fourth quarter of 2008

AIG consistently lent more than 15 percent of its domestic life insuranceassets in 2007 for example the figure was 19 percent By comparison

1Securities lending transactions are very similar to repurchase agreements as discussedin Adrian et al (2013) For additional background on securities lending see Aggarwal Saffiand Sturgess (2012) and Bank of England (2010)

2Term arrangements can be fixed or indicative If they are indicative they can beterminated early without penalty (Bank of England 2010) We do not have informationabout whether AIGrsquos arrangements were fixed or indicative

Metlife another active insurance securities lender never had more than 10percent of its domestic life insurance assets on loan

Typically securities lending collateral is invested in short-term highlyliquid securities A firm cannot easily lend its securities for cash collateral ifpossible borrowers of those securities fear that their cash collateral may notbe secure However AIG invested a substantial portion of the cash collateralit received from securities borrowers in longer term illiquid instrumentsincluding securities dependent on the performance of subprime residen-tial mortgages At the end of 2007 65 percent of AIGrsquos securities lendingcollateral was invested in securities that were sensitive either directly orindirectly to home prices and mortgage defaults These securities includedsecurities backed by residential and commercial mortgages as well as secu-rities backed by credit card auto and home equity loans It also includedcollateralized debt obligations which are structured financial instrumentsthat are backed by a pool of financial assets often the riskier tranches ofmortgage-backed securities Cash flows to collateralized debt obligationsare divided into tranches ranked from junior to senior Any losses arefirst allocated to the more junior tranches until their value is exhausted astructure which offers protection to senior tranches

Of the remainder of AIGrsquos securities lending collateral 19 percent wasinvested in corporate bonds and 16 percent was in cash or other short-term investments American International Group Inc (2007d p 108) Forcomparison a Risk Management Association Survey (Risk ManagementAssociation 2007) of securities lenders shows that on average 33 percent oflending proceeds was invested in mortgage-backed securities asset-backedsecurities (a broad category of securities backed by securities like credit cardreceivables and auto loans) and collateralized debt obligations 42 percentin corporate bonds and 25 percent in cash and short-term investments

AIGrsquos use of securities lending collateral to purchase residentialmortgage-backed securities and collateralized debt obligations is similarto the broader phenomenon described in Krishnamurthy Nagel andOrlov (2014) of financial firms using short-term funding like repurchaseagreements and securities lending to fund assets that had previously beenfunded through insured bank deposits AIGrsquos investments of securitieslending collateral in real-estate-related instruments accelerated after2005 On the other hand AIGFP decided to stop increasing its exposureto real-estate-related risk near the end of 2005 It took some time toimplement this decision however and deals that were in the pipeline werecompleted and as a result AIGFPrsquos real estate exposure continued to growIn addition some of the collateralized debt obligations that AIGFP insured

were ldquoactively managedrdquo which meant that the manager of the securitycould replace maturing refinanced and defaulting mortgages with newones including the particularly default-prone mortgages that were madein 2006 and 2007

The AIG securities lending business was characterized by a large liq-uidity and maturity mismatch Securities borrowers can demand the returnof their cash collateral on short notice However AIG was investing thiscash in long-term assets whose market values and liquidity could vary sub-stantially in the short run As long as AIG could make new security loanswhen existing ones came due it could maintain its investments in long-run illiquid assets But an arrangement based on a liquidity and maturitymismatch like this one is clearly vulnerable to bank-run dynamics Thesecurity borrowers have incentives that are similar to bank depositors wholack deposit insurance Depositors will rush to withdraw cash when theyare concerned about their bankrsquos solvency They want to make sure thatthey get their funds before the bank runs out of money Similarly secu-rity borrowers who are worried about the AIGrsquos ability to return their cashon demand are likely to ask for it to be returned And efforts to satisfythese demands will further erode AIGrsquos liquidity and generate losses thatwill prompt other securities borrowers to demand the return of their cashcollateral

Indeed before AIG was rescued on September 16 2008 securities lend-ing counterparties began to terminate these lending agreements Standardand Poorrsquos Moodyrsquos and Fitch all lowered AIGrsquos credit rating in May orJune 2008 AIG announced large second-quarter losses on August 6 2008The possibility of further losses and still-lower credit ratings appears tohave accelerated the counterpartiesrsquo efforts to reduce their securities lend-ing exposure to AIG Because the combination of falling real estate pricesand higher mortgage foreclosures had reduced the market price of securi-ties tied to these underlying assets and because it did not have access toother sources of liquidity AIG was unable to generate sufficient funds tomeet redemption requests and to return the cash collateral Moreover itslosses on securities lending threatened the regulatory capital positions ofAIGrsquos life insurance subsidiaries a point we discuss later and one that isalso emphasized by Peirce (2014)

Like many episodes during the crisis AIGrsquos securities lending problemscan be viewed through the lenses of both liquidity and solvency AIGsummed up its dilemma with respect to securities lending with considerableunderstatement in its 2008 10K report (American International Group Inc2008a) ldquoDuring September 2008 borrowers began in increasing numbers

to request a return of their cash collateral Because of the illiquidity in themarket for RMBS [residential mortgage-backed securities] AIG was unableto sell RMBS at acceptable prices and was forced to find alternative sourcesof cash to meet these requestsrdquo On Monday September 15 2008 aloneAIG experienced returns under its securities lending programs that led tocash payments of $52 billion (American International Group Inc 2008ap 4)

On September 16 2008 AIG received ldquoalternative sources of cashrdquo fromthe Federal Reserve Bank of New York The cash was initially in the formof loans However the New York Fed soon set up several limited liabilitycompanies as financial vehicles to handle its rescue of AIG In December2008 Maiden Lane II purchased AIGrsquos remaining portfolio of residentialmortgage back securities in which it had invested securities lending col-lateral for $205 billion a 48 discount relative to their par value of $393billion According to the Congressional Oversight Panel (2010 p 45) AIGrsquossecurities lending counterparties demanded the return of $24 billion in cashcollateral between September 12 and September 30 2008 Ultimately AIGreported losses from securities lending in excess of $20 billion in 2008

33 Securities Lending and Bankruptcy

What would have happened to AIGrsquos insurance companies and securitieslending counterparties in the event of an AIG bankruptcy Generallyif a securities lender seeks bankruptcy protection the borrower simplytakes ownership of the security that it borrowed any additional claimsassociated with the transaction would be resolved in bankruptcy Thevalue of the security on loan is marked to market daily and the collateralis adjusted accordingly so any additional claims if a security lender goesbankrupt would typically be small Because securities lending transactionsare exempt from the ldquoautomatic stayrdquo provisions of the bankruptcy codemdashthat is the rule that once bankruptcy has been declared creditors cannotmove to collect what they are owedmdashresolving these securities lendingtransactions should be fast and straightforward

However AIGrsquos securities lending was conducted largely on behalf of itslife insurance companies which were regulated at the state level If AIG haddeclared bankruptcy the resolution of claims related to securities lendingwould likely have depended on the actions of state insurance regulatorsWhen a life insurance company cannot meet its financial obligations astate insurance commissioner will take control of the companyrsquos operations

and place it in receivership3 Federal bankruptcy law does not apply toinsurance companies although the actions taken under state receivershipstatutes are generally patterned after federal bankruptcy However certainimportant exceptions to this practice may have been material for AIG in2008

If AIG had sought bankruptcy protection state insurance commission-ers would probably have seized AIGrsquos insurance subsidiaries Dinallo (2010)In these circumstances the status of securities lending transactions mighthave varied depending on where a particular AIG insurance subsidiary waslocated As of 2008 of the ten states where AIGrsquos life insurance subsidiarieswere located only Texas had passed a version of the Insurer Receiver-ship Model Act (IRMA) written by the National Association of InsuranceCommissioners (NAIC) which allows securities lending and other quali-fied financial contracts to receive the same exemption from the automaticstay provisions in an insurance resolution that would apply in bankruptcy4

Texas-domiciled companies supplied the securities for 58 percent of AIGrsquossecurities lending However the legal treatment of counterparties to theremaining 42 percent of the securities supplied by life insurers located inother states would have been uncertain in an insurance insolvency AIGrsquos2007 10K points out that ldquothe securities on loan as well as all of the assetsof the participating companies are generally available to satisfy the liabilityfor collateral receivedrdquo (American International Group Inc 2007d p 108)

An additional protection for some securities borrowers would havearisen from a unique aspect of AIGrsquos lending program Rather than thetypical practice of requiring collateral of 102 percent of the value of thesecurity being lent AIG began lending securities with less than 100 percentcollateral with the AIG parent company making up the difference to theinsurance subsidiary (American International Group Inc 2008a p 3) AIGseems to have accelerated this practice as its liquidity issues grew moreacute For example in an August 14 2008 email a Federal Reserve Bank ofNew York employee noted that ldquoCSG (Credit Suisse Group) does not needthe securities it borrows but instead AIG is using the deals to raise cashAs such CSG is looking to take a haircut on AIGrsquos securities as opposed toposting cash to AIG in excess of the securities value which is the market

3The state receivership process has three stages 1) conservation 2) rehabilitation and 3)liquidation The receivership process can involve transfers of blocks of assets and liabilitiesto other companies If the company cannot be rehabilitated or sold it is declared insolventand the commissioner liquidates the company and distributes assets or the proceeds fromasset sales to approved claimants in the manner prescribed by the statersquos receivership laws

4See Fitch Ratings (2006) and Law360 (2012)

standardrdquo (Federal Reserve Bank of New York 2008) By 2008 AIG hadalso boosted rebate fees paid to securities borrowers and was making losseson securities lending arrangements but felt this was warranted in order toavoid a ldquorun on the bankrdquo scenario Hutchings (2010)

When the borrowing firm does not post enough cash to fund ldquosubstan-tially all of the cost of purchasing replacement assetsrdquo then from an ac-counting perspective the transaction will be treated as a sale rather than asa securities lending transaction American International Group Inc (2008ap 166) reported losses of $24 billion on securities lending transactions thathad to be reclassified as ldquosalesrdquo in 2008

Overall this analysis suggests that losses for AIGrsquos securities lendingcounterparties would have been small had AIG sought bankruptcy protec-tion and if the counterparties were able to take possession of the securitiesthat they had borrowed Securities borrowers who held securities worthmore than the cash they were due from AIG would not have suffered lossesin an AIG bankruptcy barring uncertainties associated with state insur-ance law Note that this conclusion only takes into account the potentialfor direct losses Counterparties needing to unwind or liquidate positionsquickly might have suffered indirect losses as well

34 Impact of Securities Lending on AIGrsquos Domestic Life Insur-ance Subsidiaries

The losses for life insurance companies engaged in securities lending canbe attributed to two factors losses on sales of assets incurred when thosesecurities were sold for cash when borrowed securities were being returnedand unrealized mark-to-market losses on similar assets that had not yet beensold Together these losses put AIGrsquos domestic life insurance companiesunder considerable regulatory pressure Life insurance regulators establishminimum levels of capitals that take into account each companyrsquos assetrisk insurance risk market risk interest rate risk and business risk (alongwith an adjustment to account for the fact that these risks are not perfectlycorrelated) When capital falls below a certain threshold state insuranceregulators are required to intervene to protect policyholders

Looking at their official end-of-the-year balance sheets AIGrsquos life insur-ance subsidiaries appear to have made it through 2008 with a comfortablecushion of capital relative to regulatory minimums However these figuresinclude over $19 billion in capital infusions in the third and fourth quartersof 2008 that were only possible because of the rescue of AIG Table 2 showsthe capital positions of the eleven AIG life insurance subsidiaries that had

more than $5 billion in assets at the end of 2007 For each company the tableshows 2007 assets and the share of those assets that were on loan throughAIGrsquos securities lending business securities lending losses in 2008 and thecompanyrsquos regulatory capital as of the end of 2008 both with and withoutthe capital infusions made possible by the rescue Eight of these elevencompanies would have had negative capital without the capital infusionsThe rescue funds recapitalized the life insurance companies and kept themsolvent despite their securities lending losses This ultimately benefitedAIGrsquos life insurance policy holders

The urgency of the problems in AIGrsquos life insurance subsidiaries is re-flected in the rapidity with which they were recapitalized By September30 2008 just 14 days after the initial loan to AIG $133 billion of the loanproceeds from the Federal Reserve Bank of New York had already gone to-ward recapitalizing the life insurance subsidiaries (Congressional OversightPanel 2010 p 84) Ultimately at least $58 billion of the total governmentassistance to AIG went to addressing problems related to securities lending$19 billion in capital infusions to the life insurance subsidiaries to addresssecurities lending losses $367 billion to repay collateral to securities lend-ing counterparties ($195 billion from Maiden Lane II plus $172 billion fromthe revolving credit facility that the New York Fed established in the initialstages of the rescue) as well as an additional $31 billion from the revolv-ing credit facility to repay securities obligations (Congressional OversightPanel 2010 p 237)

4 AIGrsquos Credit Default Swap Portfolio

We now turn to a discuss AIGrsquos credit default swap business with the goalof understanding the position in which AIG and its counterparties foundthemselves on September 16 2008

41 Credit Default Swaps

A credit default swap is a derivative financial instrument that behaves likean insurance contract on a bond or a similar financial security The writer ofthe credit default swap who is the insurance seller promises to pay to thebuyer of a credit default swap the difference between the market value andthe par value of the insured bond if a ldquocredit eventrdquo occurs (For presentpurposes setting aside the sometimes arcane details of these contracts itis sufficient to think of a credit event as the failure of the bond to make a

Table 2 The role of the rescue in recapitalizing AIGrsquos Life Insurance Subsidiaries $ Million

2007 2008Company State Assets ($) Assets Realized Sec Post-rescue Regulatory Regulatory

Loaned in Lending Capital Capital CapitalSec Lending Losses Infusions $ with rescue without rescue

ALICO DE 101632 45 470 967 4332 3365VALIC TX 63999 151 3563 3621 2940 -681AIG Annuity TX 50553 397 7109 6048 3242 -2806Am General Life TX 33682 313 3790 3084 2844 -240SunAmerica Life AZ 39455 271 2281 1366 4805 3439AIG SunAmericaLife

AZ 35072 61 425 281 1317 1036

AIG Life DE 10790 236 870 679 465 -214Am Gen Life ampAccident

TN 9134 339 977 786 594 -192

First SunAmerica NY 6479 303 654 947 550 -397Am International NY 7093 351 771 801 458 -343United States Life NY 5315 251 395 456 305 -151Total AIG Life 364770 190 21305 19036 22393 3357

Source Authorsrsquo calculations from insurance regulatory filings accessed through SNL Financial and March 5 2009Hearing before the Senate Committee on Banking Housing and Urban Affairs httpwwwgpogovfdsyspkgCHRG-111shrg51303pdfCHRG-111shrg51303pdf (page 43) Table includes details for active securities lendingparticipants with assets of at least $5 billion The ldquoTotal AIG Liferdquo row includes all AIG life insurance subsidiariesCapital is regulatory capital

promised payment as in a default) There are two ways that the writerof a credit default swap like AIG can suffer a loss Obviously a loss canoccur if a credit event means that the bond or security no longer makes itspromised payments But in addition a loss can occur when the probabilityof a future credit event rises and so the price of buying a new credit defaultswap for protection against that loss also rises In this case the firm thatoriginally sold the credit default swap at a lower price has suffered a losson a mark-to-market basis and that loss is incorporated in its accountingstatements The use of mark-to-market accounting was controversial duringthe financial crisis (Heaton Lucas and McDonald 2010) but it is standardpractice for most derivatives Mark-to-market losses on AIGrsquos credit defaultswap contracts were $286 billion in 2008 (American International GroupInc 2008a p 265)

42 AIGrsquos Credit Default Swaps

As of December 31 2007 AIG had written credit default swaps with a no-tional value of $527 billion These swaps were written on corporate loans($230 billion) prime residential mortgages ($149 billion) corporate debt-collateralized loan obligations ($70 billion) and multi-sector collateralizeddebt obligations ($78 billion) (American International Group Inc 2007dp 122) (AIG also had an additional $15 trillion of other derivative expo-sures including over $1 trillion in interest rate swaps) The credit defaultswaps written on multi-sector collateralized debt obligations proved themost troublesome Again a collateralized debt obligation is a financialsecurity backed by an underlying stream of debt payments which can befrom mortgages home equity loans credit card loans auto loans and othersources The payments on this security are then divided into tranches sothat junior tranches will bear losses before senior tranches domdashallowing thesenior tranches to receive a higher credit rating It is even possible to create acollateralized debt obligation by combining tranches of other collateralizeddebt obligations a so-called CDO-squared AIG insured collateralized debtobligations backed by a variety of assets but including a substantial sharebacked by mortgages including both residential and commercial as wellas prime subprime and Alt-A (which fall between prime and subprime onthe risk spectrum) (American International Group Inc 2008a p 139)5 Itis important to realize that AIGrsquos credit default swap exposure resulted in

5Details of one of AIGrsquos multi-sector credit collateralized debt obligation are summarizedin Appendix A Source documents for many CDOs are available online at httpfciclawstanfordeduresourcestaff-data-projectscdo-Library

a ldquoone-wayrdquo bet on real estate that is a decline in real estate prices anda rise in foreclosures would impose costs on AIG but a rise in real estateprices or a fall in foreclosures would not benefit AIG In contrast market-making financial firms (like a stockbroker-dealer) typically seek to hedgeany significant directional exposure so that they make profits regardless ofwhether the price of the underlying asset (say the price of a stock) rises orfalls

American International Group Inc (2007d p 122) characterized $379billion of its credit default swaps (out of $527 billion) those on corporateloans and prime residential mortgages as used for ldquoregulatory capital reliefrather than risk mitigationrdquo primarily by European banks These do notappear to have been especially risky in its 2008 10-K (American Interna-tional Group Inc 2008a p 118) AIG reported a mark-to-market loss of$379 million on this portfolio 01 percent of the notional value MoreoverAIG expected that the swaps would be terminated by the counterpartiesonce they were operating under the Basel II capital rules (American Inter-national Group Inc 2007d p 122) This suggests that the counterpartybanks considered themselves compliant with Basel II although they werenot yet regulated under those rules

AIG began originating multi-sector credit default swaps in 2003 at atime when the firm was rated AAA Over half of AIGrsquos cumulative issuancesof credit default swaps however occurred after the firmrsquos credit rating wasdowngraded twice in 2005 AIGFP reportedly decided to stop originatingcredit default swaps in December 2005 at which point they still had $80billion of commitments (Polakoff 2009 p 5)6

43 Collateral and Variation Margin

AIGrsquos credit default swap contracts were traded over-the-countermdashthat isdirectly with counterpartiesmdashas opposed to being traded on an exchangeand cleared through a clearinghouse The standard master agreement forover-the-counter derivatives is provided by the International Swaps andDerivatives Association and includes a credit support annex which speci-fies how counterparty credit risk will be addressed Both the master agree-ment and annex can be customized when negotiating a deal

By construction many derivatives contracts have zero market value atinception this is generally true for futures swaps and credit default swaps

6The AIG portfolio of credit default swaps does not directly show up on its balance sheetdue to accounting conventions

When a position has zero market value the two parties to a contract can bymutual consent exit the contract without any obligation for either to makeany further payment to the other Note that one or both parties may beusing the contract to hedge a position in which case exiting would leave atleast one party with unhedged risk to consider

As time passes and prices move a contract initiated with zero marketvalue will generally not remain at zero market value Fair value will bepositive for one counterparty and negative by an exactly offsetting amountfor the other In such cases it is common for the negative value party tomake a compensating payment to the positive value counterparty Sucha payment is referred to as margin or collateral the two terms mean thesame thing7 Collateral can flow back and forth as market values changeIt is important to note that this transfer of funds based on a market valuechange is classified as a change in collateral and not as a payment Thereason is that the contract is still active so collateral is held by one partyagainst the prospect of a loss at the future date when the contract maturesor makes payment on a loss If the contract ultimately does not generatethe loss implied by the market value change the collateral is returnedThe accounting treatment of collateral recognizes this description and thereporting of collateral on the balance sheet depends upon the existenceof a master netting agreement When full variation margin is regularlyexchanged the value of the contract is in effect regularly reset to zeromeaning that the counterparties can agree to exit the contract without anyfurther payments

44 AIGrsquos Collateral Practices

The post-crisis investigation shed light on AIGrsquos collateral arrangementswith various counterparties Most of the credit default swap contractswritten by AIG did not call for full exchange of variation margin Ratherthey carried a wide range of collateral provisions (details are summarizedin American International Group Inc (2007a) and American InternationalGroup Inc (2007b) and standard collateral practices are discussed in ISDA(2010)) Some contracts made no provision for any exchange of collateralMost often AIG would make collateral payments only if the decline in valueof the insured assets exceeded some predefined threshold These thresholds

7Technically payments due to market value changes are variation margin Another use ofcollateral is to protect against possible future market value changes This kind of collateralcalled ldquoinitial marginrdquo or the ldquoindependent amountrdquo was typically not used in OTC marketsin dealer-to-dealer transactions prior to the crisis and is not relevant for discussing AIG

often depended on AIGrsquos credit rating which meant that a corporate ratingsdowngrade could lead to a large required collateral payment

Selected examples from December 2007 (American International GroupInc 2007b) illustrate agreements ranging from full mark-to-market to an 8percent threshold with various credit rating triggers for AIG and in somecases for the underlying collateral Here are three examples GoldmanSachs had 44 transactions with AIG with a total notional value of $1709billion The threshold (level of market value change required to trigger acollateral payment) was ldquo4 as long as AIGFP is rated in the AAAa cate-goryrdquo (American International Group Inc 2007b p 4) Societe Generalehad 38 transactions with AIG with a total notional value of $1864 billionThe threshold was ldquo8 as long as AIGFP is rated AAAa2 and ReferenceObligation is rated at least in the AAAa category the Threshold is reducedbased on a matrix that takes into account lower ratings of AIGFP andorthe Reference Obligationrdquo (American International Group Inc 2007b p6) Finally RBS had four transactions with AIG with a total notional valueof $135 billion AIG had to make variation payments for any market valuechange the threshold for these was zero (American International GroupInc 2007b p 6)

The assets underlying the multi-sector collateralized debt obligationswere not easily traded As a consequence there were running disagree-ments between AIG and its counterparties later documented by the FederalCrisis Inquiry Commission about their mark-to-market value at any giventime and hence the amount of collateral that AIG owed counterparties

Because many of the AIG credit default swap agreements did not in-clude full payment of mark-to-market variation margin AIG could and didaccumulate unpaid losses An unpaid variation amount is economicallyequivalent to a loan from the counterparty to AIG If AIG has $1 billion inunpaid variation margin it is as if AIG borrowed $1 billion from the coun-terparty In addition a party accumulating unpaid losses may be unwillingto exit a derivatives contract because doing so would force it to make fullcollateral payments Presumably this is why the credit support annex ofswap agreements will often contain provisions that allow the purchaser ofa credit default swap to terminate the agreement if the issuer of the swapexperiences a credit downgrade

AIG had first reported a loss on its written credit default swaps in2007 losing $115 billion on all such swaps for the yearmdash$111 billion inthe fourth quarter alonemdashwith 98 percent of the total coming from creditdefault swaps on multi-sector collateralized debt obligations (American

Table 3 Evolution of collateral calls and collateral posted for AIGrsquos CDS onMulti-Sector CDOs $ Millions

Goldman Sachs Societe Generale All Counterparties Total

Date Call Posted Call Posted Call Posted Shortfall6302008 7493 5913 1937 1937 15780 13241 25399122008 8979 7596 4280 4008 23441 18922 4519

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

Source AIGGoldman Sachs Collateral Call Timeline FCIC httpfciclawstanfordedudocumentsview2172AIG was downgraded on September 15 2008 as a result of which manymulti-sector CDS counterparties were contractually entitled to additionalcollateral

International Group Inc 2007d p 83)8 Losses continued in 2008 Table 3depicts the evolution of collateral calls between June and September 2008for Goldman Sachs and Societe Generale (AIGrsquos two largest credit defaultswap counterparties) as well as for all counterparties combined As ofJune 30 2008 counterparties had called $1578 billion and AIG had posted$1324 billion The totals climbed gradually until on September 12 totalcalls amounted to $2344 billion with AIG having posted $1892 billionThus prior to the rescue AIG had already provided almost $20 billion tocounterparties

The effect of ratings triggers is evident in a comparison of collateral callsfor September 12 2008 and those for September 15 2008 the day on whichall three ratings agencies downgraded AIG below AA- Total collateral callsincreased by $86 billion to $32 billion AIGrsquos collateral shortfall rose from$45 billion to $124 billion Societe Generalersquos call on that day rose by 55

8AIGrsquos credit default swap business was barely disclosed prior to 2007 The phraseldquosuper seniorrdquo referring to tranches of collateralized debt obligations appears four times inthe 2006 annual report and 114 times in 2007 ldquomulti-sectorrdquo does not appear in 2006 butappears 23 times in 2007 ldquoCDOrdquo (for collateralized debt obligation) appears twice in 2006and 93 times in 2007 AIGrsquos 2006 annual report discloses that it had written $4836 billionin credit default swaps but provides no details whereas the 2007 report reports notionalvalues of credit default swap by category AIGrsquos first public disclosure of credit defaultswaps written on the multi-sector collateralized debt obligations came on August 9 2007during a second-quarter earnings call (Financial Crisis Inquiry Commission 2011 p 268)The lack of disclosure is surprising given that the credit default transactions increased thesize of AIGrsquos balance sheet by 50 percent in economic terms

billion

45 What Would Have Happened to Credit Default Swap Coun-terparties if AIG Had Declared Bankruptcy

If AIG had declared bankruptcy on September 16 2008 what would havebeen the direct effect on credit default swap counterparties It is of courseimpossible to answer this question definitively but some straightforwardobservations are possible

AIG had 21 counterparties for its multi-sector credit default swaps Ofthose nine had collateral calls exceeding $500 million and six of thosemdashGoldman Sachs Societe Generale Merrill UBS DZ Bank and Rabobankmdashhad a difference greater than $500 million between the collateral they hadrequested and the amount AIG had posted Table 4 shows these collateralshortfalls for the six largest counterparties to AIGrsquos multi-sector credit de-fault swaps as of September 16 2008 and also shows the shortfall relativeto shareholder equity for each counterparty Of the $114 billion that AIGowed to counterparties on its credit default swaps on September 16 2008these six banks accounted for $10 billion

If AIG had defaulted the counterparty banks to the credit default swapson the multi-sector collateralized debt obligation would have likely facedthree direct consequences First the banks would have kept the collateralalready posted by AIG This is a result of the rule mentioned earlier thatderivatives are exempted from the automatic stay in bankruptcy (for discus-sion see Edwards and Morrison (2005) and Bolton and Oehmke (forthcom-ing)) Second the banks would have been treated as general creditors forany collateral that had been requested but AIG had not yet posted Thirdthe banks would have retained the asset or position that had been hedgedby the defaulted credit default swap

Assuming that assets were valued correctly and that the September 152008 downgrade of AIG to an A rating eliminated remaining any remainingthresholds that might have further increased collateral calls the economiccost of an AIG default for its counterparties would be equal to the collateralshortfall that is the difference between called and posted collateral Howsignificant would this shortfall have been for the counterparty banks Ascan be seen in Table 4 even for the six banks that were individually owedmore than $500 million in no case did the shortfall exceed 10 percent oftheir equity capital

However comparing the actual loss with counterparty equity may betoo sanguine because it assumes that counterparties would simply absorb

Table 4 AIGrsquos counterparties for CDS on multi sector CDOs CollateralShortfall relative to equity and Asset Sales necessary to maintain pre-shortfall equity to asset ratio $B

Total Shareholder Collateral Shortfall ImpliedAssets Equity Shortfall Equity Asset Sales

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

Goldman Sachs 10818 456 25 541 585Societe Generale 16944 560 42 756 1281Merrill Lynch 8758 384 10 270 236UBS 17845 415 10 241 430DZ Bank 6770 106 07 700 474Rabobank 8940 450 06 131 117

Total asset sales to return to pre AIG shortfall equity to assets 3124

Source Financial Crisis Inquiry Commission (2010) and author calculationsusing 2008 Q2 and Q3 financials Goldman Sachs Merrill Lynch and UBSassets shareholders equity and tier 1 capital come from 2008-Q3 financialstatements Societe Generale DZ Bank and Rabobank values come from2008-Q2 financial statements Column [5] is equal to column [4] multipliedby column [1] and represents the assets sales that would be necessary ifthe AIG collateral shortfall from column [3] was realized and the firm inquestion chose to preserve its original equity to asset ratio

the loss This assumption faces at least three potential problems FirstBrunnermeier and Pedersen (2009) and Duarte and Eisenbach (2014) amongothers emphasize the possibility of fire-sale spillovers Institutions mightrespond to the loss in capital by selling assets in order to return to theirpre-loss leverage ratios This could lower asset prices and lead to mark-to-market losses at other firms who might in turn sell assets to get backto target leverage ratios Our back of the envelope calculations presentedin Table 4 suggest that if these six banks had chosen to respond by sellingassets to get back to their pre-AIG default debt to equity ratios they wouldhave needed to sell $312 billion in assets Second the cancellation of thecredit default swaps would leave many of the counterparties with unhedgedexposure to real estate risk Retaining this risk could reduce the capacityfor risk-taking Third even if one concludes that counterparties could haveabsorbed losses due to an AIG failure other market participants would nothave known at the time who was exposed and in what amount For thisreason the failure of any large financial firm can be stressful for the financialsystemmdasha conclusion that is not particular to credit default swaps or AIG

Another consequence of AIGrsquos failure would have been cancellation ofthe $387 billion of other credit default swaps mainly held by Europeanbanks Collateral calls related to these positions totaled just $500 millionon September 16 2008 (Congressional Oversight Panel 2010 p 42) andas noted above the institutions were apparently anticipating the swaps toexpire when they adopted Basel II capital rules The cancellation of theseswaps would have created a capital deficiency but it is not clear that thiswould have been economically important In any event European financialregulators would have had the option to forebear for a time with enforcingthe capital rules thus allowing a period for adjustment

Overall how much did the rescue of AIG benefit its multi-sector creditdefault counterparties Some media reports suggest that $62 billion intaxpayer funds were paid to AIGrsquos multi-sector credit default swap coun-terparties (for example Orol (2010)) In fact the direct counterparty benefitfrom the rescue is smaller We can divide the payments to AIGrsquos creditdefault swap counterparties into three categories

First there are collateral payments AIG made prior to the rescue Thesepayments would have been retained by counterparties in a bankruptcyand therefore cannot be attributed to the rescue These payments totaled$224 billion with $185 billion associated with multi-sector collateralizeddebt obligations that became part of the Maiden Lane III Fed-created spe-cial purpose vehicle (see also Congressional Oversight Panel (2010 p 93))Second there are collateral payments made by AIG after the rescue These

payments could only be made because of the rescue and clearly offset lossesthat counterparties would have sustained in the absence of a rescue Thisamount provides a lower bound on the assistance received by counterpar-ties to the credit default swaps due to the rescue AIGrsquos 2008 10-K reportstotal collateral payments for credit default swaps of $401 billion for 2007and 2008 suggesting that $177 billion was paid after the rescue (As con-firmation of this amount the Congressional Oversight Panel (2010 p 93)found that collateral payments of $165 billion were made after the rescuefor the assets that became part of Maiden Lane III) Finally Maiden LaneIII made cash payments of $268 billion in exchange for the assets that AIGhad insured These payments were equal to the estimated fair market valueof the assets at the time (SIGTARP 2009) While there may not have beenmany buyers for these assets even at 47 of face value in the fall of 2008 itis inappropriate to consider the entire amount of the price that Maiden LaneIII paid for the credit default swap as a direct benefit to the counterpartiesIndeed as we discuss in the next section this portfolio of assets appreciatedand was later sold for a modest gain

5 Performance of Maiden Lane Assets

The Federal Reserve Bank of New York created several special purposevehicles as part of the rescue of AIG Among them Maiden Lane II pur-chased the remaining securities lending invested collateral from AIG andMaiden Lane III acquired from AIGFPrsquos counterparties the collateralizeddebt obligations that AIG had insured This acquisition terminated theassociated credit default swaps Maiden Lane II was funded by a $195billion loan from the New York Fed and $1 billion from AIG that wouldabsorb the first $1 billion in losses Maiden Lane III was funded by aloan from the New York Fed of $243 billion and $5 billion in equityfrom AIG (Congressional Oversight Panel 2010 p 87 91) The NewYork Fed has thoroughly documented the resulting cash flows at httpwwwnewyorkfedorgmarketsmaidenlanehtml These data in combi-nation with information from various other sources allow us to examinehow the value of these securities evolved both while they were held in theMaiden Lane vehicles and afterward

Table 5 Summary statistics for assets in Maiden Lane 2 and Maiden Lane 3portfolios

Maiden Lane 2 Assets Maiden Lane 3 Assets

Min Median Max Min Median MaxNotional (mm$) 002 3100 26600 004 20100 540000Purchase percentage 001 056 099 010 048 094Sale percentage 000 058 102 003 049 096Gain (mm$) -7050 153 7640 -17200 3680 77900Return ( Gain

Purchase Price minus 1) -095 013 406 -085 035 124Benchmark return -015 022 023 003 021 023Return less Benchmark return -118 -007 384 -091 014 102

Source Authorsrsquo calculations using data from the Federal Reserve Bank ofNew York and Markit Notes Purchase percentage is the ratio of the pricepaid for each asset to its notional value Sale percentage is the ratio of theprice received for each asset to its notional value All dollar values are inmillions For Maiden Lane II the benchmark is ABXHEAAA06-1 whilefor Maiden Lane 3 the benchmark is 70 ABXHEAAA06-1 plus 30CMBXNAAAA1-1 an index of commercial mortgage backed obligations

51 Maiden Lane II and III Performance

The New York Fed managed the Maiden Lane vehicles and assets withthe goal of selling the assets once markets stabilized Both Maiden Lanevehicles were ultimately liquidated for a total gain of $95 billion While heldin the Maiden Lane vehicles the underlying securities paid interest and alsorepaid principal and experienced writedowns both of which reduced theirface value They were ultimately sold by auction The Maiden Lane II assetswere bought in December 2008 for $205 billion (53 of par value) returned$89 billion in interest and principal while held and the residual claims weresold for $151 billion (51 of par) for a non-annualized return of 169 Thesecurities were sold principally in 2011 and 2012 Table 5 summarizes thesize purchase and sale discount and returns of the individual Maiden LaneII and III securities There is significant variation in the size and discountsof securities

It is not obvious whether the overall return of 169 is ldquogoodrdquo giventhe risk of the assets We can ask however whether the Maiden Lane secu-rities performed especially well or poorly compared to a broader universeof residential real estate To perform this comparison while controlling fordifferent liquidation dates we use as a benchmark an index of AAA secu-ritized subprime mortgage loans originated in the last six months of 2005

the ABXHEAAA06-1 index The median security in Maiden Lane II hada 13 return and underperformed the ABX by 7 It is worth noting thatAIG had begun to sell its securities lending collateral prior to the creation ofMaiden Lane II and the securities acquired by the special purpose vehiclewere likely the poorest assets

The securities in Maiden Lane IIImdashprimarily the multisector collateral-ized debt obligation that AIG had insured through its credit default swapsmdashwere bought in November and December 2008 for $293 billion (474 ofpar) returned $171 billion in interest and principal and were sold for $226billion (498 of par) for a non-annualized return of 351 The securi-ties were sold primarily in 2012 The median security in Maiden Lane IIIreturned 35 exceeding the benchmark return by 14 Returns on theMaiden Lane III securities were greater than those on Maiden Lane II evenafter adjusting for the return benchmark (The benchmark for Maiden LaneIII was 70 ABXHEAAA06-1 and 30 CMBXNAAAA1-1 an index ofcommercial mortgage backed obligations We obtained almost identicalresults using this benchmark and using ABX alone)

52 Post-Maiden-Lane Performance

Table 6 shows the performance of the securities lending invested collateralportfolio that eventually became part of Maiden Lane II and the supersenior tranches of the collateralized debt obligations that were insured byAIGFP and eventually became part of Maiden Lane III9 The table providesinformation at four points when the securities were originated (variousdates) when the Maiden Lane vehicles were created when the securitieswere sold from the Maiden Lane vehicles (various dates) and as of October2014 (or the most recent prior date for which information is available)Thirty-six percent of the Maiden Lane II securities and fifty-nine percent ofthe Maiden Lane III securities in the table have experienced write-downs Asizeable share of write-downs have occurred during the post-Maiden Laneperiod As explained earlier senior tranches will be the last to experienceactual losses and for this reason actual losses in these tranches will appearlater and will likely increase over time With approximately one-third ofprincipal still outstanding future substantial writedowns for the assets inboth Maiden Lanes II and III remain possible

9Figures reported in Table 6 reflect the full outstanding amount for any security thatwas included in Maiden Lane II or III and not the share of the security purchased by thosevehicles Please see the notes to Table 6 for additional details

Table 6 Aggregate Performance of Maiden Lane Assets Originationthrough October 31 2014

DateOrigination Maiden Lane Most

Purchase Sale Recent

ML2 Notional ($ billions) 1377 859 626 432ML2 Amortization ($ billions) 000 518 726 874ML2 Write-down ($ billions) 000 005 25 70ML2 Write-down Since Start () 000 004 18 51ML2 securities with Write-downs () 000 05 175 360ML3 Notional ($ billions) 825 688 458 295ML3 Amortization ($ billions) 000 137 310 431ML3 Write-down ($ billions) 000 000 57 99ML3 Write-down Since Start () 000 000 69 120ML3 securities with Write-downs () 000 000 472 590Source Authorsrsquo calculations based on data from the Federal Reserve Bankof New York and from summaries derived from Intex data Analysis usingthe Intex data was performed by Larry Cordell and Yilin Huang of theFederal Reserve Bank of PhiladelphiaNotes Data were available for each of the 855 securities in Maiden Lane IIand 146 of the 155 securities in Maiden Lane III accounting for 97 percentof the original Maiden Lane III face amount Omitted securities were eithernot present in the Intex data (7 securities) or had partially missing data (2securities)Origination is the date the security was created Beginning of MaidenLane is the approximate time at which the asset was purchased by a MaidenLane Maiden Lane Sale is the approximate time at which the asset wasa sold by a Maiden Lane and ldquoMost Recentrdquo refers to information as ofOctober 31 2014 or the most recent prior data available (Some assetsmatured or were written down completely prior to October 31 2014 Oncea security has been paid off or written down completely no additionaldata are reported for it) Figures reflect the full outstanding amount forany security that was included in Maiden Lane II or III and not the shareof the security purchased by those vehicles For example Maiden Lane IImight have owned 10 of a particular security and 100 of the outstandingamount of the security is used to compute the figures in the table

Reported write-downs to date are 51 percent of the original face valueof the securities that ended up in Maiden Lane II and 12 percent for MaidenLane III These estimates were calculated from information provided byLarry Cordell and Yilin Huang from the Federal Reserve Bank of Philadel-phia following the methodology in Cordell Huang and Williams (2012)The Maiden Lane III assets are harder to assess because issuers of collat-eralized debt obligations do not report writedowns prior to maturity It isthus necessary to look for writedowns on the individual instruments con-stituting the collateralized debt obligation The fact that the Maiden Lane IIand III assets have suffered write-downs means that we can reject the starkclaim that they were ldquomoney-goodrdquo

6 Was AIG Special

Given the drama surrounding AIG it is natural to ask how AIG comparedto other financial firms at the time Was AIG unusual in its risk-taking orwas it just unlucky It turns out that AIG resembled some large banks inimportant respects its real estate holdings were comparable to those ofCitigroup and Bank of America banks which also received considerableofficial support in 2008 and 2009 In addition AIGrsquos financing of its realestate positions was fragile and prone to runs in times of financial difficultyMaking a comparison with other firms requires first that we assess AIGrsquosposition prior to the rescue especially its exposure to housing A notablefeature of AIG was its large position in written credit default swaps and weneed to take these into account when comparing firms

61 A Comparison of AIG with Other Financial Firms

Issuing a credit default swap is economically equivalent to borrowing inorder to finance the purchase of the same risky bond that the credit defaultswap would insure To see this suppose that you have excellent creditthat you borrow $50 at a 5 rate of interest and that you use the proceedsto buy $50 in one-year bonds that might default and which consequentlypay a 15 rate of interest If the bonds pay in full you have a $5750asset ($50 times 115) offset by a $5250 liability ($50 times 105) and you will haveearned the 10 interest differential ($5) However if the bonds lose $20for example you have a $30 asset and a $5250 liability ndash you have a lossof $2250 This pattern of gains and losses is precisely that faced by theseller of a credit default swap on the bonds If the bonds pay in full the

seller earns the credit default swap premium ($5) and if the bonds defaultthe credit default swap seller bears the loss ($2250) which is paid to thebondholder10

To relate this to AIG consider the simplified example of a firm with $100in assets and $90 of debt and therefore $10 of equity The firm has an asset-to-equity ratio of 101 (that is $100$10) This firm now sells a credit defaultswap on $50 of mortgage-backed securities In the contract the buyer ofthe credit default swap agrees to make an annual payment of $5 and theseller bears the loss if the mortgage-backed securities fail The economicresult is the same as if the firm had $150 in assets ($100 plus the $50 inmortgage-backed securities insured by the credit default swap) financedwith $140 in debt $50 of which is implicit in the credit default swap Theissuance of a credit default swap implicitly changes assets and debt but notequity

This was approximately AIGrsquos situation the firm as a whole had $106trillion of assets and about $964 billion in liabilities at the end of 2007 so ithad equity of $96 billion It issued $527 billion in credit default swaps Itwas therefore economically equivalent to a firm with $159 trillion in assetsand $96 billion in equity Taking into account the credit default swapsAIGrsquos ratio of assets to equity was 161 rather than 111

AIG was not the only financial firm with off-balance sheet real estateholdings Citigroup Bank of America and JPMorgan Chase all had off-balance-sheet asset-backed commercial paper conduits used to fund realestate holdings (V Acharya Schnabl and Suarez 2013) The effective assetto equity ratio for these banks was also higher than reported

Table 7 compares AIGrsquos total real estate exposure with Citigroup Bankof America and JPMorgan Chase and with that of another large insurancecompany Metlife After adjusting the balance sheets as discussed abovewe find that AIGrsquos real estate exposure was 24 of assets comparable tothat of Bank of America (32) and Citigroup (21) AIGrsquos effective realestate holdings were almost four times its book equity

62 Was AIG A Bank

Banks typically employ short-term financing to fund holdings of long-termilliquid assets AIG did have some explicit short-term financing in particu-lar $20 billion of commercial paper But AIGrsquos illiquid real estate positions

10In economic terms a credit default swap is economically equivalent to a purchase of theinsured asset financed by issuing floating rate debt (Duffie 1999) For a general discussionof credit default swaps see McDonald (2013 Chapter 27)

Table 7 Real Estate Exposure and Leverage for AIG MetLife CitigroupBank of America and JPMorgan Chase year-end 2007 $B

AIG MetLife Citigroup BofA JPM

Real Estate on balance sheet 1533 1175 3775 5149 1386Off-balance sheet ABCP 975 542 634CDS on Residential Mortgage 1490CDS on Multi-Sector CDOs 780Total real estate exposure 3803 1175 4750 5691 2020Reported Assets 10605 5586 21876 17157 15621Adjusted Assets 15875 5586 22852 17700 16256Equity 958 352 1136 1468 1232Reported AssetsEquity 111 159 193 117 127Adjusted AssetsEquity 166 159 201 121 132Real estate as of AdjustedAssets

24 21 21 32 12

Real estate as of Equity 397 334 418 388 164

Sources MBS and real estate values for MetLife Citigroup Bank ofAmerica and JPMorgan Chase are from 2007-Q4 bank holding com-pany data accessed via the Federal Reserve Bank of Chicago web-site httpwwwchicagofedorgwebpagesbankingfinancial_institution_reportsbhc_datacfm Values for off-balance sheet ABCP come fromPhilipp Schnablrsquos website httppagessternnyuedu~sternfinpschnablAll other values are from 2007 10K filings and authorsrsquo calculations

Notes AIGrsquos real estate exposure includes investment of securities lend-ing collateral in real estate related assets CDS on residential mortgageand multi-sector CDOs are the notional value of AIGrsquos credit default swapportfolio (by asset class) Equity is equal to total shareholdersrsquo equity asreported in the companiesrsquo 2007 10K filings on their consolidated balancesheets Reported assets are equal to total assets as reported on the com-paniesrsquo consolidated balance sheets Adjusted assets are equal to reportedassets plus off-balance sheet ABCP and in AIGrsquos case their credit defaultswap portfolio Please note that in addition to the CDS on residentialmortgages and multi sector CDOs listed above this also includes CDS oncorporate loans ($230 billion) and CDS on corporate debtcorporate CLOs($70 billion) Reported leverage is equal to reported assets divided byequity Adjusted leverage is equal to adjusted assets divided by equity

were also financed in a way that was not as transparently fragile as demanddeposits but which could create large liquidity needs if AIG suffered losses

As discussed earlier AIGrsquos securities lending agreements had a rela-tively short maturity and could be subject to early termination As AIGsuffered downgrades and as the real estate investments made with secu-rities lending proceeds suffered losses securities lending counterpartiesbecame increasingly likely to terminate these agreements culminating ina $52 billion redemption request on September 15 2008 This desire bycounterparties to unwind their exposure to AIG resembled a bank run ascounterparties sought to unwind the positions rather than be left with collat-eral and possibly involved in lawsuits AIG effectively used collateralizedshort-term financing to buy real estate assets

Although the mechanism was different AIGrsquos multi sector credit defaultswap positions also suffered from something akin to a bank run AIGrsquosCDS counterparties could not unilaterally terminate credit default swapagreements but they were entitled to collect collateral as the values ofinsured assets declined and these counterparty rights could sometimes beaccelerated if AIGrsquos credit rating was lowered When AIG was downgradedon September 15 2008 collateral calls on AIGrsquos multi sector credit defaultswaps increased by $86 billion as a result

While AIG was not literally a bank it undeniably had bank-like char-acteristics as it employed financing (both explicit and implicit) that wassubject to termination and cash demands when asset values fell

7 Conclusions

Insurance companies are traditionally less vulnerable to financial crisesthan banks in large part because they have relatively low-risk assets anddo not rely heavily on short-term funding However AIG made itself vul-nerable in a number of ways Notably AIGrsquos near-failure was a result oftwo outsized bets on real estate both of which generated large needs forliquidity First AIG used securities lending to transform insurance com-pany assets into residential mortgage-backed securities and collateralizeddebt obligations ultimately losing $21 billion and threatening the solvencyof its life insurance subsidiaries On one day in 2008 AIG was required topay $52 billion in cash to satisfy redemption requests Second AIG issuedcredit default swaps on real-estate-backed multi-sector collateralized debtobligations ultimately losing more than $30 billion and facing a one-day$86 billion collateral demand due to a downgrade its credit rating Secu-

rities lending and writing credit default swaps were both ldquocarry tradesrdquothat is bets that long-term assets would earn a higher return than the short-term cost of funding AIGrsquos use of financial markets to transform itselffrom a traditional insurance company to a bank-like firm ultimately proveddisastrous

The rescue of AIG had many beneficiaries The broader financial systemwas spared the unpredictable consequences of a large and complicated firmfailing at a time when financial markets were very fragile Direct beneficia-ries of the rescue included the life insurance subsidiaries that received $20billion in capital infusions protecting their policy holders The counterpar-ties to the credit fault swaps AIG had sold on multi-sector credit defaultobligations were also beneficiaries although their direct benefit was the$177 billion in collateral payments made after the rescue rather than muchlarger figures that have sometimes have been emphasized In additionto addressing problems with securities lending and the multi-sector creditdefault swap portfolio rescue funds provided to AIG directly benefited nu-merous other counterparties including AIGrsquos employees holders of AIGrsquoscommercial paper and other AIG debt holders and repo counterpartiesstates and municipalities who had AIG sponsored Guaranteed InvestmentAgreements as well as defined contribution pension plans holding stablevalue wraps issued by AIG

AIGrsquos near failure is often described as a liquidity event that is it founditself in 2008 holding a number of mortgage-based securities that wereimpossible to sellmdashexcept perhaps at unreasonably low ldquofire salerdquo pricesBut AIG sustained a loss of $99 billion in 2008 exceeding the firmrsquos end of2007 equity of $96 billion (AIG 2008b p 36) raising the question of whetherit experienced a liquidity problem a solvency problem or both Despite itsreliance on fragile sources of funding AIG had no specialized liquidityrisk committee until 2007 (AIG 2007b p 99) It is tempting to attributethis to the companyrsquos insurance origins together with senior managementsrsquobelief that the real estate related investments were ldquomoney goodrdquo Ourexamination of the performance of AIGrsquos underlying real estate securitiesindicates that AIGrsquos problems were not purely about liquidity While wecannot say whether prices in 2008 were ldquocorrectrdquo the assets representedin both Maiden Lane vehicles have experienced substantial write-downswith the possibility of more in the future With hindsight it may seemobvious that AIGrsquos real estate assets were not ldquomoney goodrdquo and wouldsuffer real losses However the belief that they would not suffer lossesand that liquidity would not be a problem was an important factor in theircreation and purchase by AIG and others

References

Acharya Viral V Douglas Gale and Tanju Yorulmazer (2011) ldquoRolloverRisk and Market Freezesrdquo In Journal of Finance 664 pp 1177ndash1209

Acharya Viral Philipp Schnabl and Gustavo Suarez (2013) ldquoSecuritizationWithout Risk Transferrdquo In Journal of Financial Economics 107 pp 515ndash536

Adrian Tobias et al (2013) ldquoRepo and Securities Lendingrdquo Federal ReserveBank of New York Staff Report 529

Aggarwal Reena Pedro A C Saffi and Jason Sturgess (2012) ldquoThe Role ofInstitutional Investors in Voting Evidence from the Securities LendingMarketrdquo Georgetown McDonough School of Business Research PaperNo 2012-07

American International Group Inc (2007a) AIG Status of Collateral CallPostings httpfciclawstanfordedudocumentsview7

mdash (2007b) AIG Super Senior Credit Transactions ndash Principal Collateral Provi-sions httpfciclawstanfordedudocumentsview1140

mdash (2007c) American International Group Investor Meeting ndash Final http wwwfcicgovdocumentsview1139

mdash (2007d) Form 10-K 2007 for the fiscal year ended December 31 2007 Re-trieved from httpwwwsecgovedgarshtml

mdash (2007e) Form 10-Q quarterly report for the fiscal quarter ended September 302007 Retrieved from httpwwwsecgovedgarshtml

mdash (2008a) Form 10-K 2008 for the fiscal year ended December 31 2008 Re-trieved from httpwwwsecgovedgarshtml

mdash (2008b) Schedule A to Shortfall Agreement Between Maiden Lane III LLC andAIG Financial Products Corp httpstaticreuterscomresourcesmediaeditorial20100127Schedule20Apdf

mdash (2009) Form 10-K 2010 for the fiscal year ended December 31 2009 Retrievedfrom httpwwwsecgovedgarshtml

mdash (2010) Information Pertaining to the Multi Sector CDS Portfoliohttpfciclawstanfordedudocumentsview53

Bank of England (2010) Securities Lending An Introductory Guide http wwwbankofenglandcoukmarketsDocumentsgiltssl_intro_green_9_10pdf

Bernanke Ben S (2009) Testimony Before the Committee on Financial ServicesUS House of Representatives Washington DC March 24 2009

Bolton Patrick and Martin Oehmke (forthcoming) ldquoShould Derivatives BePrivileged in Bankruptcyrdquo In Journal of Finance

Brunnermeier Markus K and Lasse Heje Pedersen (2009) ldquoMarket Liquid-ity and Funding Liquidityrdquo In Review of Financial Studies 226 pp 2201ndash2238

Congressional Oversight Panel (2010) Oversight Report The AIG RescueIts Impact on Markets and the Governmentrsquos Exit Strategy GovernmentPrinting Office

Cordell Larry Yilin Huang and Meredith Williams (2012) ldquoCollateralDamage Sizing and Assessing the Subprime CDO Crisisrdquo Federal Re-serve Bank of Philadelphia url httppapersssrncomsol3paperscfmabstract_id=1907299

Dinallo Eric (2010) ldquoWhat I Learned at the AIG Meltdown State regulationwasnrsquot the problemrdquo In Wall Street Journal

Duarte Fernando and Thomas M Eisenbach (2014) ldquoFire-Sale Spilloversand Systemic Riskrdquo Staff Report 645 Federal Reserve Bank of New York

Duffie Darrell (1999) ldquoCredit Swap Valuationrdquo In Financial Analysts Journal551 pp 73ndash87

Edwards Franklin R and Edward Morrison (2005) ldquoDerivatives and theBankruptcy Code Why the Special Treatmentrdquo In Yale Journal on Reg-ulation 22 pp 91ndash122

Ellul Andrew et al (2014) ldquoMark-to-Market Accounting and Systemic RiskEvidence from the Insurance Industryrdquo In Economic Policy 29 pp 297ndash341

Federal Deposit Insurance Corporation (2011) ldquoThe Orderly Liquidationof Lehman Brothers Holdings Under the Dodd Frank Actrdquo In FDICQuarterly 52

Federal Reserve Bank of New York (2008) FRBNY Email re AIG Meeting withOTS httpfciclawstanfordedudocumentsview2113

Financial Crisis Inquiry Commission (2010) AIGGoldman-Sachs CollateralCall Timeline httpfciclawstanfordedudocumentsview2172

mdash (2011) The Financial Crisis Inquiry Report Final Report of the NationalCommission on the Causes of the Financial and Economic Crisis in the UnitedStates httpfcic-staticlawstanfordeducdn_mediafcic-reportsfcic_final_report_fullpdf

Fitch Ratings (2006) Legal Status of Derivative Counterparties in UInsur-ance Company Insolvencies Remains Murky Increases Risk httpswwwfitchratingscomcreditdeskreportsreport_framecfmrpt_id=266110

Heaton John C Deborah Lucas and Robert L McDonald (2010) ldquoIs Mark-to-Market Accounting Destabilizing Analysis and Implications for Pol-icyrdquo In Journal of Monetary Economics 571 pp 64ndash75

Hutchings Mark (2010) 2010-06-22 FCIC staff audiotape of interview with MarkHutchings American International Group Inc httpfciclawstanfordeduinterviewsview313

ISDA (2010) Market Review of OTC Derivative Bilateral Collateralization Prac-tices httpwwwisdaorgc_and_apdfCollateral-Market-Reviewpdf

Kacperczyk Marcin and Philipp Schnabl (2010) ldquoWhen Safe Proved RiskyCommercial Paper During the Financial Crisis of 2007-2009rdquo In Journalof Economic Perspectives 241 pp 29ndash50

Krishnamurthy Arvind Stefan Nagel and Dmitry Orlov (2014) ldquoSizing UpRepordquo In Journal of Finance 696 pp 2381ndash2417

Law360 (2012) Expanding Insurance Regulation One State At A Timehttpwwwlaw360comarticles295760expanding-insurance-regulation-one-state-at-a-time

McDonald Robert L (2013) Derivatives Markets 3rd Boston MA Pearson-Addison Wesley

Orol Ronald D (2010) Geithner Paulson defend $182 billion AIG bailout httpwwwmarketwatchcomstorygeithner-paulson-defend-182-bln-aig-bailout-2010-01-27 MarketWatch

Peirce Hester (2014) ldquoSecurities Lending and the Untold Story in the Col-lapse of AIGrdquo Working Paper 14-12 Mercatus Center George MasonUniversity

Polakoff Scott (2009) Statement of Scott M Polakoff to the House Committee onBanking Housing and Urban Affairs March 5 httpwwwoccgovstaticnews-issuancesotstestimonyots-testimony-ts171-03-05-2009pdf

Risk Management Association (2007) Securities Lending Quarterly AggregateComposite url http wwwrmahqorg securitieslending quarterly-aggregate-composite

SIGTARP ndash Office of the Special Inspector General For the Troubled AssetRelief Program (2009) Factors Affecting Efforts to Limit Payments to AIGCounterparties url httpwwwsigtarpgovAudit20ReportsFactors_Affecting_Efforts_to_Limit_Payments_to_AIG_Counterpartiespdf

Taibbi Matt (2011) Griftopia Random House

Appendices

A Example of a Collateralized Debt ObligationAdirondack 2005-1

Adirondack was a $15 billion multi-sector CDO created by Gold-man Sachs in 2005 and the AAA-rated tranches were insuredby AIG The Adirondack 2005-1 Pitchbook is available at httpfciclawstanfordedudocumentsview2306 and the Offering Circu-lar at httpfciclawstanfordedudocumentsview2284The Adirondackprospectus stated that a AAA rating for the senior tranches was a precon-dition for issuance and that the proceeds from issuance would be usedto purchase RMBS (787) CMBS (92) CDOs (81) ABS (1) REITs(3) and Synthetics (121 of which 294 are RMBS 217 are CMBSand 489 are CDOs) See p 25 of the Offering Circular

Table 8 shows the various tranches of the CDO the interest they wouldpay and the maturity date of each tranche The CP notes (in the last rowof the table) contained a put agreement under which Societe Generale wasobligated to buy additional A-1 LT-a notes at par with the proceeds used torepay the notes AIG wrote CDS on the two senior most tranches the A-1LT-a Floating Rate tranche which paid 32 basis points above LIBOR andthe CP Notes which were intended to pay LIBOR

Tranche Description Amount ($b) Interest (LIBOR + ) DueA-1 LT-a Floating Rate 2675 032 034 after

July 20082040

A-2 Floating rate 608 040 2040B Floating rate 577 058 2040C Floating rate 304 140 2040D Floating rate 243 275 2040E Floating rate (optional) 50 gt=275 2040CP Notes 10701 0 2040A-1 LT-b Floating rate 032 034 after

July 20082040

Total 15158

Table 8 Details of Adirondack 2005-1 offering Source Adirondack 2005-1CDO Offering Circular FCIC

B Notes on Data

Important data sources include

bull the Maiden Lane documentation available from the New York Fedincluding spreadsheets reporting prices and dates for the MaidenLane transactions

bull Annual (10-K) and quarterly (10-Q) reports for AIG and other financialinstitutions

bull numerous documents available via the FCIC website

bull Proprietary data from Intex which provides ldquothe source data for theuniverse of publicly issued private-label mortgage-backed securities(MBS) as well as as publicly traded structured-finance ABS CDOsrdquo(Cordell Huang and Williams 2012 p 3) Actual calculations withthe Intex data were performed by Larry Cordell and Yilin Huang ofthe Federal Reserve Bank of Philadelphia Numbers in this paper arebased on results of their calculations

bull Proprietary data from Markit which provides CDS price quotesspecifically the ABX and CMBX series used in Table 5

bull Proprietary data from SNL Financial which provides access to theregulatory filings of US domiciled insurance operating companiesused in constructing Table 2

bull Proprietary data from Bloomberg which provides historical ratingsfor AIG (Table 9) and for the Maiden Lane II and III securities (Figure1)

C Notes on the Maiden Lane Securities

C1 Information from the New York Fed

The New York Fed data related to the Maiden Lane transactions and vehiclesare publicly available at httpwwwnewyorkfedorgmarketsmaidenlanehtml For each security held in a Maiden Lane the New York Fed suppliesspreadsheets that report

bull Security description

bull Date when acquired (settlement date)

bull Date when sold (settlement date)

bull The face value of the security and purchase price both when acquiredand sold

bull Net cash flow received while the security was held in a Maiden Lane

bull The identity of the counterparties

Important characteristics of the securities however can only be inferredusing proprietary commercial data

The Maiden Lane data reports the price at which securities were boughtand sold and the total cashflow earned from holding the security It is thuspossible to compute the return earned while the securities were held by theMaiden Lanes and the NY Fed spreadsheets do this Unfortunately it is notpossible to compute actual writedowns while the securities were held inthe Maiden Lanes and the NY Fed spreadsheets do not report performanceafter the securities are sold

Given additional data such as that available from Intex we can de-compose the reported return into four components interest principal re-payments capital gains and writedowns We can also track subsequentperformance To understand what the New York Fed data does and doesnot allow us to infer let Ft be the face value of a security at time t with F0denoting the face value at origination The face value can change either be-cause there have been cumulative principal repayments Rt or cumulativewritedowns Wt There are also cumulative interest payments It

Suppose that a CDO or an RMBS with face value Ft on date t is purchasedfor x1F1 on date 1 and sold at date 2 for x2F2 with xt representing the percentof face value at which the security trades Total net cash flows between time1 and 2 C2 are due to interest and principal repayments C2 = ∆I2 + ∆R2

where ∆ denotes the change between time 1 and time 211 Over time theface value changes as a result of cumulative principal repayments Rt andcumulative writedowns Wt By definition principal changes are due torepayments and writedowns

F2 = F1 minus ∆R2 minus ∆W2 (1)

The return on the security is

Return = x2F2 + ∆R2 + ∆I2 minus x1F1 (2)

The Maiden Lane data reports this return To interpret this number we canrewrite equation (2) as

Return = x1(F2 minus F1) + ∆R2 + ∆I2 + (x2 minus x1)F2

= x1(minus∆R2 minus ∆W2) + ∆R2 + ∆I2 + (x2 minus x1)F2

= minusx1∆W2 + ∆R2(1 minus x1) + ∆I2 + (x2 minus x1)F2

To interpret this expression

minusx1∆W2 Writedowns are a loss but only a loss against the price paid for thesecurity (x1) We report cumulative writedowns Wt and cumulativeprincipal repayments (ldquoamortizationrdquo) in Table 6

∆R2(1 minus x1) Principal is repaid at $1 which is a gain relative to the dis-counted purchase price of x1

∆I2 Interest is worth $1 for each dollar paid

(x2 minus x1)F2 There can be a capital gain or loss on the remaining principal(Principal that is not remaining has been written down for a loss orpaid off at a gain)

If we know either writedowns or principal repayments for the MaidenLane securities the other can be inferred

C2 Other information

There are also numerous formerly confidential documents published bythe FCIC and otherwise relating to the Maiden Lane securities Thesenotably include American International Group Inc (2010) and American

11For simplicity we incorporate interest on principal repayments into It

International Group Inc (2008b) American International Group Inc (2010)tells us the origination dates for the multi-sector CDOs that AIG insuredand it reveals that most of them had at least some subprime mortgageexposure American International Group Inc (2008b) reports losses on asecurity-level basis for securities in Maiden Lane III at the time the securitieswere purchased for the vehicle

C3 Credit Rating History for Maiden Lane Securities

Credit ratings for the individual RMBS and CDOs held in the Maiden Lanevehicles are plotted in Figure 1 This figure complements Table 6 showingthat ratings declines accompanied the writedowns

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

Initial ratingLatest rating

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Figure 1 Credit ratings for the assets in Maiden Lanes II and III at origina-tion and latest available Low numbers correspond to high ratings a ldquo1rdquo isAAA and 22 is the lowest rating Source Bloomberg

D AIGrsquos Credit Rating History

AIGrsquos rating history is in Table 9

Date Rating Firm Rating3302005 SampP AA+6032005 SampP AA5082008 SampP AA-9152008 SampP A-3312005 Moodys Aa15022005 Moodys Aa25222008 Moodys Aa39152008 Moodys A23152005 Fitch AA+5022005 Fitch AA5082008 Fitch AA-9152008 Fitch A

Table 9 History of Fitch Moodyrsquos and SampP credit ratings for AIG 2005 -2008 Prior to March 2005 AIG was rated AAA by all three rating agenciesSource Bloomberg

Introduction

AIG Financials 2006-2009

AIGs Securities Lending Business

What Is Securities Lending

Characteristics of AIGs Securities Lending

Securities Lending and Bankruptcy

Impact of Securities Lending on AIGs Domestic Life Insurance Subsidiaries

AIGs Credit Default Swap Portfolio

Credit Default Swaps

AIGs Credit Default Swaps

Collateral and Variation Margin

AIGs Collateral Practices

What Would Have Happened to Credit Default Swap Counterparties if AIG Had Declared Bankruptcy

Performance of Maiden Lane Assets

Maiden Lane II and III Performance

Post-Maiden-Lane Performance

Was AIG Special

A Comparison of AIG with Other Financial Firms

Was AIG A Bank

Conclusions

Example of a Collateralized Debt Obligation Adirondack 2005-1

Notes on Data

Notes on the Maiden Lane Securities

Information from the New York Fed

Other information

Credit Rating History for Maiden Lane Securities

AIGs Credit Rating History

AIG in HindsightRobert L McDonald and Anna PaulsonNBER Working Paper No 21108April 2015JEL No E00G01G18G2

ABSTRACT

The near-failure on September 16 2008 of American International Group (AIG) was an iconic momentin the financial crisis Two large bets on real estate made with funding that was vulnerable to bank-runlike behavior on the part of funders pushed AIG to the brink of bankruptcy AIG used securities lendingto transform insurance company assets into residential mortgage-backed securities and collateralizeddebt obligations ultimately losing at least $21 billion and threatening the solvency of the life insurancecompanies AIG also sold insurance on multi-sector collateralized debt obligations backed by realestate assets ultimately losing more than $30 billion These activities were apparently motivated bya belief that AIGrsquos real estate bets would not suffer defaults and were ldquomoney-goodrdquo We find thatthese securities have in fact suffered write-downs and that the stark ldquomoney-goodrdquo claim can be rejectedUltimately both liquidity and solvency were issues for AIG

Robert L McDonaldKellogg School of ManagementNorthwestern University2001 Sheridan RoadEvanston IL 60208and NBERr-mcdonaldnorthwesternedu

Anna PaulsonFederal Reserve Bank of Chicago230 S LaSalle StreetChicago IL 60604AnnaPaulsonchifrborg

1 Introduction

Item 2006 2007 2008 2009

AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



1 Introduction

Item 2006 2007 2008 2009

AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Item 2006 2007 2008 2009

AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Item 2006 2007 2008 2009

AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Item 2006 2007 2008 2009

AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



AZ 35072 61 425 281 1317 1036

TN 9134 339 977 786 594 -192

9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9152008 10072 7596 9833 4320 32013 19573 124409162008 10065 7596 9818 5582 33879 22445 11434

billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



billion

9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



9162008 ([3][2]) ([4]times[1])[1] [2] [3] [4] [5]

6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



6 Was AIG Special

62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



62 Was AIG A Bank

24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



24 21 21 32 12

7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



7 Conclusions

References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



References

Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Appendices

July 20082040

Total 15158

B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



B Notes on Data

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



2002 2004 2006 2008 2010 2012 2014

Maiden Lane 2

2002 2004 2006 2008 2010 2012 2014

Maiden Lane 3

Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information



Introduction














Was AIG Special


Was AIG A Bank

Conclusions


Notes on Data



Other information

