1) What is a credit default swap (CDS)? Consider the cash-flow pattern of the CDS in Exhibit 2 ofFrom free lunch to black hole: Credit Default Swaps at AIG. Can you set up a trading strategy using long and short positions of the default-free floating rate bond and the defaultable floating rate bond that replicates the cash flow pattern of the CDS? Given the coupon rate on the defaultable floating rate bond, can you determine the CDS rate based on the law of one price?

2) Review Exhibits 3-6 ofFrom free lunch to black hole: Credit Default Swaps at AIG. What was AIG's role in the process of mortgage securitization? How did the Bistro deal work? How does the mortgage securitization work? What are super-senior tranches and what is their purpose?

3) How risky are BBB-tranches in the first layer of securitization?

4) How risky are super-senior tranches in the second layer? Should one view a super-senior tranche as safer than an AAA-rated corporate bond?

5) Is it possible that AIG could experience big losses on its super-senior CDS positions? Are collateral calls justified?

For the exclusive use of O. Oztekin, 2017. case W04C41 August 6, 2015 From Free Lunch to Black Hole: Credit Default Swaps at AIG Alan Frost, executive vice president for AIG Financial Products (AIGFP), was on vacation when he received a disturbing email from Andrew Davilman at Goldman Sachs in the evening of July 26, 2007 (see Exhibit 1). In the years leading up to 2007, AIGFP had written $75 billon notional value of credit default swaps (CDS) tied to subprime mortgage-backed securities (MBS). Through these CDS, AIGFP provided credit insurance to its counterparties, meaning that AIGFP would absorb losses on the MBS underlying the CDS if homeowners defaulted on their mortgages.1 Goldman Sachs was the counterparty for more than $20 billion notional value of these CDS contracts.2 For several years, AIG had profited handsomely from these contracts and had judged that there was a virtually zero risk that AIG would ever have to make a payment on these CDS. But now, as housing prices started to fall, subprime borrowers began to default in greater numbers, and market values of the subprime MBS underlying the CDS dropped substantially. In Goldman Sachs' view, these circumstances were a clear indication that the marked-to-market value of the CDS contracts with AIG had shifted in Goldman Sachs' favor. Referring to the CDS contract terms, Goldman Sachs therefore asked AIG to provide collateral. On July 27, the day after Frost received the heads-up from Davilman about the forthcoming margin call, Goldman Sachs formalized its demand for collateral by sending AIGFP a collateral invoice requesting that the company provide it with collateral worth $1.8 billion.3 American International Group In 2007, American International Group (AIG) was one of the biggest insurance companies in the world. It operated in 130 countries and employed close to 100,000 people. AIG was founded in 1919 as an insurance agency in Shanghai. In wake of the revolution in China, the company moved to New York City in 1949. AIG went public in 1969, with Maurice R. Greenberg as CEO. Greenberg led the company through an enormous expansion until he was forced out in 2005 amid investigations into accounting irregularities by the New York Attorney General. Greenberg remained a large shareholder in AIG, and contested the charges of accounting irregularities and the circumstances of his forced departure from AIG. After Greenberg's departure, Martin J. Sullivan, a long-time AIG employee, took over as CEO.4 Published by WDI Publishing, a division of the William Davidson Institute (WDI) at the University of Michigan. 2015 Stefan Nagel. This case was written by Stefan Nagel (Michael Stark Professor of Finance at the Ross School of Business) at the University of Michigan to be the basis for class discussion rather than to illustrate either the effective or ineffective handling of a situation. Secondary research was performed to accurately portray information about the featured organization. Company representatives were not involved in the creation of this case. This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Exhibit 1 Email from Andrew Davilman to Alan Frost Source: Financial Crisis Inquiry Commission. . AIG Financial Products Earlier, looking for new business opportunities, AIG created its AIG Financial Products (AIGFP) subsidiary in 1987. AIG was looking for ways to profitably exploit its AAA credit rating and its strong capital base. Its objective in starting AIGFP was to participate in the growing over-the-counter markets for financial derivatives.5 It hired a group of traders from Drexel Burnham Lambert, the firm that had pioneered the junk bond market in the U.S. The firm was in deep trouble at the time amid allegations of insider trading and investigations by Rudy Giuliani, the U.S. attorney for the Southern District of New York.6 One member of this group was Joseph Cassano, who first served as chief financial officer of AIGFP and, from 2001, as head of AIGFP.7 AIG located its AIGFP unit in London, but its business was nevertheless subject to U.S. regulation. Because AIG owned a small savings and loans institution, it was able to choose the Office of Thrift Supervision (OTS) as its federal regulator for its non-insurance businesses (which included AIGFP). OTS was therefore also responsible for supervision of the AIGFP subsidiary (AIG's U.S. insurance business was supervised by state regulators). Since AIGFP agreed to supervision by OTS, it avoided regulation by the U.K. Financial Services Authority (FSA). OTS, typically responsible for supervision of small savings and loans institutions, had little expertise in the type of business conducted by AIGFP.8 AIGFP engaged in transactions with interest-rate swaps, derivatives on commodities, and many other derivatives products. In the 1990s, AIGFP expanded its activities into the nascent market for credit default swaps (CDS). AIGFP's earnings rose from $150 million in 1993 to $323 million in 1998 and to $758 million in 2001.9 2 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Credit Default Swaps Credit default swaps (CDS) were invented in the 1990s as an instrument for transferring credit risk between financial institutions. For example, a bank that made a loan to a company could use CDS to transfer to another party (e.g. an insurance company) the risk that this company might default on the loan. The bank in this example is the credit-protection buyer and the insurance company is the credit-protection seller. To compensate the insurance company for assuming the default risk of this loan, the bank makes a small periodic payment, as long as the company does not default on the loan. In the event of default, the insurance company is required to make a payment to the bank to cover the bank's losses on the loan. The same idea was soon applied to corporate bonds, mortgage-backed securities, and other securities, for which CDS provided credit protection in the event that the borrower defaulted on the obligations under these securities. More precisely, a CDS pays out to the credit-protection buyer if a credit event occurs with respect to the underlying reference entity (e.g., the company that received the loan in the above example). What exactly constitutes a credit event is defined in the terms of the CDS contract. A credit event can involve outright default of the underlying reference entity, but it can also be an event involving some restructuring of the debt, for example. Many CDS are conducted on the basis of terms laid out in a master agreement by the International Swaps and Derivatives Association (ISDA). As an example, suppose that IBM has a floating-rate bond outstanding with more than seven years remaining until maturity that trades at par. Suppose party A, the credit-protection buyer, enters into a CDS with party B, the credit-protection seller, with seven years maturity. The notional value of the CDS contract is $1 million. In the CDS contract, the two counterparties agree that as long as IBM does not default on this bond during the next seven years, A pays B an annual payment of u% (the \"CDS rate,\" which would be specified as a specific number in the contract) of the notional value of $1 million. In the event that IBM defaults, B makes a payment to A equal to the notional value of $1 million times the \"loss given default\" and the contract terminates. For example, if IBM defaults, and subsequent to the default event, the IBM bonds are trading at 60% of their face value, then the loss given default is 40%, and the required payment would be 40% times $1 million = $400,000. Thus, the credit-protection buyer would be fully insured against the default risk inherent in a $1 million position in these IBM bonds. Exhibit 2 illustrates these cash flows patterns schematically for a floating-rate bond with face value F, time of default d, CDS rate u, CDS maturity T, and market value Pd of the underlying bond in the event of default. Financial institutions had a variety of motivations for entering into CDS. Part of it was to lay off credit risk to institutions that could better diversify these risk exposures. For example, if a bank makes a large loan to a company, this bank may have a large concentrated credit risk exposure to this single counterparty. By entering into a CDS, the bank can transfer all or part of this credit risk exposure to another financial institution that is in a better position to bear the risk. For many CDS transactions, the motivation, however, was driven by regulation. For example, European bank regulators calculated the amount of equity capital a bank needed to hold as a buffer against possible losses based on the risk exposure of the bank's assets. By laying off credit risk to an entity like AIG through a CDS transaction, the bank could reduce the amount of capital it was required to hold. To the extent that the payments that AIG charged for taking on the credit risk were lower than the cost that banks perceived for putting up more equity capital against risk exposures, banks were happy to enter such trades. At the time, this credit-risk transfer was also welcome by regulators, as it transferred credit risk away from highly levered banks, which enjoyed, to some extent, implicit and explicit government (i.e., taxpayer) support to institutions like AIG outside the regulated banking sector that had large amounts of equity capital and which were perceived at the time not to enjoy similar incentivedistorting government support. 3 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Exhibit 2 CDS Cash Flows from Perspective of Credit-Protection Buyer Source: Created by the author of the case. Until the early 2000s, AIG's credit risk exposure through its CDS transactions was mostly to corporate credit. However, this took a drastic turn, as financial innovation in the mortgage market involved AIG as a key player. Synthetic Securitization The set of participants in the credit default swap market was relatively narrow. Most of the counterparties in credit default swaps were banks (including investment banks), hedge funds, and some insurance companies. In the mid-1990s, several banks were developing ideas on how to transfer credit risk from bank balance sheets to a broader investor population. In 1997, J.P. Morgan set up its groundbreaking transaction called BISTRO (Broad Index Secured Trust Offering). In a BISTRO structure, a bank that originated a portfolio of loans bought credit protection for this portfolio from J.P. Morgan via a CDS. J.P. Morgan in turn bought credit protection on this portfolio of loans from a special purpose vehicle (SPV), a shell company set up by J.P. Morgan only for the purpose of this BISTRO transaction.10 For example, in the first BISTRO deal in 1997, the SPV takes on the credit risk of a loan portfolio worth $9.7 billion.i In return, the SPV receives a regular payment (based on the CDS rate) from J.P. Morgan. If a credit event occurs on the underlying portfolio, the SPV has to make a payment that compensates J.P. Morgan for the loss in value on the loans associated with the credit event. To have assets available that can be used for these payments in case of credit events, the SPV issues notes in capital markets. The amount raised from the issuance of these notes is invested in safe securities such as government obligations until it is needed to make any payments on the CDS due to credit events in the underlying portfolio of loans that the SPV insured. Thus, just like in traditional securitization, the credit risk of the underlying loans was transferred from the originating bank to capital market investors, but in this case not by transferring the actual loans from the originating bank's balance sheet, but instead by just transferring the credit risk of these loans, using CDS contracts (see Exhibit 3). This process is called synthetic securitization.11 i The originating bank would bear the first 1.5% loss on the underlying loan portfolio. The author ignores this feature in this illustrative discussion. 4 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 In the first BISTRO deal, the SPV issued only $700 million worth of notes, an amount much smaller than the notional value of the loan portfolio ($9.7 billion) that the SPV insured. Up to losses of $700 million, any credit losses on the loan portfolio would be borne by the capital market investors that bought the notes issued by the SPV. If losses on the loan portfolio exceeded $700 million, the capital market investors would be wiped out, and J.P. Morgan would bear the additional losses. The loans underlying BISTRO deals were, on average, of investment-grade quality, and hence the risk that losses on a diversified portfolio of such loans might exceed $700 million was judged to be miniscule. This risk was seen as smaller than the risk of loss on a typical AAA credit. For this reason, this remaining credit exposure that J.P. Morgan was exposed to was labeled to be \"better than triple-A\" or \"super-senior.\"12 Exhibit 3 BISTRO Deal Source: Created by the author of the case. This super-senior risk was mostly correlation risk: For the losses on the loan portfolio to exceed $700 million, many borrowers would have to default at the same time. Assessing the super-senior risk therefore required a judgment about the degree of correlation in borrowers' defaults. Based on a long history of data on corporate defaults, the correlation of defaults appeared sufficiently low so that super-senior losses seemed extremely unlikely. Partly because of pressure from regulators in some jurisdictions to hold some amount of reserve capital against this super-senior risk, and partly because of concerns about the potentially huge (even though unlikely) magnitude of the risk exposure, J.P. Morgan was looking for some counterparty to insure this supersenior risk. A suitable counterparty would have to have a strong balance sheet and a triple-A credit rating. Enter AIG. AIG's Financial Products division was willing to insure J.P. Morgan's super-senior risk. AIGFP would enter into a CDS contract with J.P. Morgan to assume the super-senior risk and, in return, earn a small periodic fee. For the first deals, AIGFP was paid 0.02% of the insured amount per year. In the initial BISTRO deal, this would amount to $1.8 million per year for insuring a notional value of $9 billion of super-senior risk exposure. In subsequent deals, AIGFP managed to contract a higher rate of 0.11%.13 Mortgage Securitization and Collateralized Debt Obligations In the early 2000s, Wall Street started applying the same credit-risk transfer technology not just to corporate credit, but also to consumer credit such as automobile loans, student loans, credit-card debt, 5 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 and mortgages. These consumer loans were transferred from originating lenders to capital market investors through the creation of asset-backed securities (ABS) or mortgage-backed securities (MBS). Until the early 2000s, most securitization of mortgages was carried out by government-sponsored enterprises Fannie Mae and Freddie Mac, which took on the credit risk of the mortgages that were securitized.14 In the early 2000s, however, Wall Street started to securitize mortgages that did not satisfy the eligibility criteria required by Fannie Mae and Freddie Mac, for example, because borrowers had low credit scores and/ or lacked documentation of income, the loan-to-value ratios were high, or the size of the mortgage exceeded set limits. (See Exhibit 4). Exhibit 4 Role of AIG in Mortgage Securitization Source: Created by the author of the case. In these \"private-label\" mortgage securitizations, MBS are issued by an SPV, and the SPV uses the proceeds from MBS issuance to purchase mortgages from an originating bank (or from an intermediary that was \"warehousing\" the loans for a brief period). The MBS are often issued as tranches that differ in credit quality. If credit losses occur on the underlying portfolio of mortgages, the (unrated) equity (or \"first-loss\") tranche bears the first losses. In the example shown in Exhibit 4, the BBB-tranche starts to suffer any losses only if the equity tranche is completely wiped out. The AAA-tranche is the last one to suffer losses, which occurs only if all the other lower-rated tranches have been wiped out. The AAA-tranche would typically account for the bulk of the notional value of the MBS, often around 80%. Thus, roughly speaking, investors in the AAA-tranche suffer losses only if the losses on the underlying portfolio of mortgages exceed 20% of the combined face value of the underlying loans.15 6 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 While there was plenty of demand for higher-rated tranches, BBB-rated tranches were difficult to sell. As a consequence, it became common in the 2000s that the BBB-tranches were used as raw material for a second layer of securitization, with the goal of \"manufacturing\" more highly rated paper. In this second layer, an SPV issues so-called collateralized debt obligations (CDOs), which are also issued in tranches, and the proceeds from the issuance of these debt obligations are used to purchase the BBB-tranches of the MBS from the first layer of securitization (CDO of asset-backed securities (ABS)).16 CDOs backed by a pool of BBB-tranches are \"mezzanine CDOs\" (this is the case shown in Exhibit 4). In some cases, CDOs were backed by a pool of higher-rated tranches and called \"high-grade CDOs.\" Often, CDOs would be set up with a mixed pool of underlying assets, partly consisting of tranches from mortgage securitization and partly with tranches from securitization of other consumer credit or corporate loans (a \"multi-sector CDO\").17 The CDO tranches were typically structured to include a super-senior tranche. For example, the supersenior tranche might account for 70% of the combined face value of all tranches and the AAA-tranche might account for 10%. Roughly speaking, the AAA-tranche bears losses if the credit losses on the underlying assets (BBB-tranches from the first layer of securitization) exceed 100% - 80% = 20%. But due to the presence of the super-senior tranche, the AAA-tranche is wiped out if the losses exceed 100% - 70% = 30%, the so-called \"attachment point.\" Thus, the AAA-tranche is a lot riskier than it would be in the absence of the super-senior tranche. Nevertheless, ratings agencies were willing to rate it AAA.18 In this example, the super-senior tranche would experience losses only if the credit losses on the underlying assets exceeded the attachment point of 30%. For a proper risk assessment, though, it is also important to consider that the assets underlying the CDO are themselves from a tranched structure. In a CDO on BBB-rated MBS, losses of 3-4% on the loans underlying the MBS can be enough to wipe out the entire BBB-tranche. In this case, all tranches of the CDO, including the super-senior, would become worthless. The super-senior tranche was typically retained by the financial institution that sponsored the creation of the CDO. According to one estimate, banks held around $216 billion worth of super-senior tranches in 2007.19 While some financial institutions decided to bear the super-senior risk, others were looking for ways to insure at least part or all of it and they approached AIGFP. As in the original BISTRO deals described above, AIGFP was willing to enter into CDS contracts to sell super-senior credit protection.20 As subprime lending and the volume of private-label MBS expanded enormously in 2004 and 2005 (see Exhibit 5), the notional volume of super-senior risk that AIG insured quickly grew to $79 billion (see Appendix A). The asset composition of the multi-sector CDOs also changed substantially during these years. While the share of subprime MBS was very small initially, by 2006 a large part contained subprime MBS. About a third of AIG's super-senior exposure from multi-sector CDOs was from mezzanine CDOs, i.e., where the MBS and ABS assets backing the CDO were rated BBB, and two thirds of the \"high-grade\" variety where the assets underlying the CDO were rated A or AA.21 (See Exhibit 5.) 7 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Exhibit 5 Global CDO Issuance and U.S. Private-Label Residential MBS Issuance ($bn) 800.0 700.0 600.0 500.0 400.0 300.0 200.0 100.0 0.0 2000 2001 2002 CDO (Global) 2003 2004 2005 2006 Private-label RMBS (US) 2007 Source: Securities Industry and Financial Market Association (SIFMA). . Goldman Sachs was AIG's largest counterparty for multi-sector CDO super-senior CDS, with a total notional amount of $23 billion of credit-protection written by AIG. The second largest counterparty was Socit Gnrale, followed by Merrill Lynch (see Exhibit 6). Exhibit 6 AIG's Main CDS Counterparties for Multi-Sector CDO Super-Senior Transactions Counterparty Notional Amount ($bn) Goldman Sachs 23.0 Socit Gnrale 18.6 Merrill Lynch 9.9 UBS 6.3 Calyon 4.5 Bank of Montreal 1.6 Royal Bank of Scotland 1.4 Wachovia 0.8 Deutsche Bank 0.6 Source: Created by the author based on data gathered from: Financial Crisis Inquiry Commission, 200712-31_AIG_Status_of_Collateral_Call_Postings. 31 Dec. 2007. . End of the House Price Boom In the months leading up to June 2005, as CEO Greenberg was forced out during investigations, the major ratings agencies downgraded AIG from its prized triple-A rating. Until the downgrade, AIGFP's counterparties had been willing to enter into CDS contracts without requiring AIGFP to put up any collateral. AIGFP's promise to make payments on the CDS contracts in case of a credit event was sufficient until that point. But after the loss of the triple-A rating, counterparties asked for collateral. In 2005, the required collateral 8 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 amounts were small, though, as the risk of a credit event on the super-senior risk seemed remote and the super-senior tranches were valued at par.22 In December 2005, AIGFP stopped writing credit-protection on super-senior tranches of multi-sector CDOs. Thus, no new super-senior risk exposures would be added to the existing ones in AIGFP's portfolio of CDS. However, AIG did not undertake any steps to hedge the existing super-senior risk exposures. In AIG's view, the risk that it would ever have to make a payment on the super-senior CDS was still negligible.23 In early 2006, house prices took a drastic turn, as illustrated in Exhibit 7 with the S&P/Case-Shiller U.S. National House Price Index. Exhibit 7 S&P/Case-Shiller U.S. National House Price Index Source: S&P/Case-Shiller. . During the years leading up to 2006, housing prices in the U.S. had doubled within less than a decade. During those years of rising house prices, defaults on subprime mortgages had been minimal. These low default rates were due to the particular design of subprime mortgages. A typical subprime mortgage features low monthly payments (based on a low \"teaser\" rate) for an initial period of two to three years. Then the payments reset to a much higher rate. For many subprime borrowers, paying the higher rates after the reset date would have been impossible, given their typically precarious financial situation. However, as long as house prices were rising, subprime borrowers could simply re-finance before the reset and take out a new mortgage, now based on the appreciated home value, again with an initial low \"teaser\" rate. But as house prices began to decline, this refinancing option disappeared. Defaults on subprime mortgages accelerated quickly. The rise in the frequency of subprime mortgage defaults severely impacted the value of mortgagebacked securities that contained subprime material, particularly the lower-rated tranches. However, it was generally rather difficult to obtain reliable market prices for MBS tranches, as there was little trading in these securities, and the market was over-the-counter with little price transparency for parties other than the dealers in this market. In January 2006, this changed with the introduction of the ABX.HE indices by Markit Group, Ltd.24 9 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 The ABX.HE indices track the prices of CDS contracts written on baskets of subprime residential MBS. Specifically, each ABX.HE index tracks CDS contracts written on baskets of tranches with a specified rating (the rating at the time the MBS were issued), and of a specified vintage. For example, the ABX.HE BBB-06-01 index tracks a basket of CDS contracts on BBB-tranches of subprime MBS issued in the second half of 2005. Every six months, a new vintage of the index is created. For example, the ABX.HE BBB-06-02 index started in July 2006, and it references BBB-tranches of subprime MBS issued in the first half of 2006. For the first time, the publication of the ABX.HE indices allowed a wide range of market participants to observe information on the pricing of credit risk in subprime MBS tranches. Exhibit 8 plots four vintages of the ABX.HE BBB indices. Exhibit 8 ABX.HE index for BBB-tranches of subprime MBS Source: Stanton, R., and N. Wallace. \"ABX.HE Indexed Credit Default Swaps and the Valuation of Subprime MBS.\" UC Berkeley Working Paper. 2009. . While \"single-name\" CDS contracts on corporate debt were typically quoted in terms of a CDS rate that represents the periodic payment that the credit-protection buyer has to make, the ABX.HE index was typically quoted in terms of a price that approximately resembles the implied price at which the underlying MBS would be expected to trade relative to their par value, given the pricing of the CDS contracts that enter into the calculation of the ABX. For example, an ABX index value of 80 would imply, approximately, that the underlying MBS (assuming they were issued at par) should trade at \"80 cents on the dollar,\" or 80% of their par value. As Exhibit 8 shows, the four vintages of the ABX.HE-BBB index showed little price movement during 2006, but they began to decline in 2007. By mid-2007, they had fallen to levels between 40 and 60 cents on the dollar. At the same time, ratings agencies downgraded hundreds of subprime MBS.25 The ABX index provided for the first time a relatively transparent view on prices at which market participants were willing to conduct trades of CDS on subprime MBS with various ratings. However, it was still difficult to extract information from the ABX indices about the valuation of CDO tranches, because the 10 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 CDOs were built from pools of subprime MBS with particular ratings (often BBB). Thus, the ABX indices shed light on the valuation of MBS that entered into the MBS pool underlying CDOs, but did not allow someone to directly infer the valuation of tranched exposures to this pool. In February 2007, Markit Group introduced a new index, called TABX.HE, which addressed this issue. The TABX.HE BBB referenced 40 CDS on subprime MBS with initial BBB ratings in two consecutive vintages of the ABX.26 While the ABX represents CDS written directly on these subprime MBS individually, the TABX represents CDS on tranched exposure to a pool of these MBS. Just like in a typical CDO, one of the tranches is a supersenior tranche, labeled the 35-100 tranche, which is the most senior tranche and accounts for 65% of the combined face value of all tranches. Effectively, the CDS underlying the TABX.HE index represent a synthetic CDO, and the index values of this newly constructed TABX.HE index provided, for the first time, a relatively transparent perspective on the valuation of super-senior CDO tranches. Exhibit 9 presents end-of-month values of the super-senior tranche of the TABX.HE constructed from CDS on the subprime MBS pools underlying the 06-02 and 07-01 vintages of the ABX. By the end of July 2007, the index level had fallen to around 55 cents on the dollar. Exhibit 9 TABX.HE Super-Senior Tranche (35-100) (Vintage ABX 06-2 and 07-1) Source: Bloomberg. Collateral Calls While AIGFP executives were on vacation in the summer of 2007, they received disturbing news. On July 27, 2007, Goldman Sachs asked AIGFP to send $1.8 billion of collateral (see Exhibit 10). Goldman Sachs had a total notional amount of $23 billion of super-senior CDS outstanding with AIG, and it requested collateral for $12 billion of those. These included CDS written on super-senior tranches of CDOs with subprime MBS for which the CDS contract allowed Goldman Sachs to request collateral.ii On August 2, 2007, after some further ii For example, Goldman Sachs did not yet request any collateral for the $5.2 billion notional super-senior CDS on Abacus CDOs, because the credit support annex of these CDS specified that Goldman Sachs could request collateral only if the junior tranches of these CDOs were downgraded, which had not happened yet by August 2007 (see Financial Crisis Inquiry Commission, 2007_07_27 Collateral Call, July 27, 2007). 11 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 analysis, Goldman Sachs revised its collateral call down to $1.2 billion. Goldman Sachs valued the $12 billion notional value of super-senior CDO tranches underlying the CDS at 80 to 97 cents on the dollar.27 AIGFP sent $450 million to Goldman Sachs on August 10, 2007, substantially less than the amount requested. Negotiations between the two counterparties kept going on. AIGFP still maintained that there was essentially zero risk that it could ever lose money on the CDS if it held them to maturity. In AIG's earnings conference call on August 9, 2007, Joe Cassano said \"...it is hard for us, without being flippant, to even see a scenario within any kind of realm of reason that would see us losing $1 in any of those transactions.\"28 Exhibit 10 Goldman Sachs Collateral Invoice from July 27, 2007 Source: Financial Crisis Inquiry Commission, 2007_07_27 Collateral Call, July 27, 2007. . Goldman Sachs argued that market values of the underlying super-senior tranches had fallen. The source of the disagreement with AIGFP was that no actual trading was taking place in these super-senior 12 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 tranches.29 Hence, \"true\" prices of actual transactions were not available. Goldman Sachs used prices of the TABX (trading was still taking place in the CDS that the TABX was constructed from), and bid/ask quotes for super-senior tranches it obtained from other dealers and from its own trading desks, to estimate the mark-to-market value of AIGFP's super-senior CDS. While some dealers had revised down their quotes, some dealers still maintained, according to Goldman Sachs, that \"... super-senior CDOs should be worth about par (i.e., one hundred cents on the dollar). However, when we asked if they would take on additional risk by trading at those levels, they refused. We believed that these underwriters had large amounts of super-senior CDOs in their own inventories and thus had incentives to maintain higher prices than the market genuinely reflected.\"30 AIGFP continued to dispute Goldman Sachs' valuations. But AIGFP's efforts were hampered by the fact that it did not have its own internal pricing model to value the super-senior tranches on which it had sold credit protection. AIGFP had to rely on price quotes from third-party dealers. These dealers provided quotes that were often higher than Goldman Sachs' quotes, but they were reluctant to provide firm quotes at which they would actually stand ready to buy a significant amount. Conclusion Until this point, AIGFP never had to make any cash payouts on super-senior CDS. Did the price declines in the subprime mortgage MBS market, and the margin calls that resulted from this decline, represent just a temporary liquidity problem that would go away if markets went back to a normal mode of operation soon? Or could the situation deteriorate further, with margin calls escalating far beyond the amounts that AIGFP requested to post at this point in August 2007? AIGFP's models suggested that the chance of a loss on the super-senior CDS was virtually nil. After all, AIGFP'S CDS provided credit insurance for the safest tranches of CDOs. How could these super-senior tranches possibly suffer any default losses? Or did AIGFP's models underestimate the risk of super-senior default losses? (For additional insight see Appendices B and C). 13 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG Appendices Appendix A Consolidated Balance Sheet of AIG, December 2006 14 This document is authorized for use only by Ozde Oztekin in 2017. W04C41 For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG Appendix A (continued) Source: AIG Form 10-K 4th quarter of 2006. Securities and Exchange Commission. . 15 This document is authorized for use only by Ozde Oztekin in 2017. W04C41 For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG Appendix B Consolidated Income Statement of AIG, December 2006 Source: AIG Form 10-K 4th quarter of 2006. Securities and Exchange Commission. . 16 This document is authorized for use only by Ozde Oztekin in 2017. W04C41 For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG Appendix C AIG's Perspective on Its Super-Senior Exposure in August 2007 17 This document is authorized for use only by Ozde Oztekin in 2017. W04C41 For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Appendix C (continued) Source: AIG Residential Mortgage Presentation. 9 Aug. 2007 (financial figures are as of June 30, 2007). . 18 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. From Free Lunch to Black Hole: Credit Default Swaps at AIG W04C41 Endnotes 1 Financial Crisis Inquiry Commission. The Financial Crisis Inquiry Report. Jan. 2011. p. 268. Accessed 19 June 2015. . 2 Financial Crisis Inquiry Commission. 3 Financial Crisis Inquiry Commission. 4 Boyd, Roderick. Fatal Risk: A Cautionary Tale of AIG's Corporate Suicide. John Wiley & Sons, Hoboken, NJ, 2011. Ch. 6-7. 5 Boyd. 6 Paltrow, Scot J. \"The Drexel Agreement to Settle: The Prosecutor Giuliani Has Fans, Foes in War on White-Collar Crime.\" Los Angeles Times. 22 Dec. 1988. Accessed 19 June 2015. 7 Lewis, Michael. \"The Man Who Crashed the World.\" Vanity Fair. August 2009. Accessed 19 June 2015. . 8 Lewis. 9 Teitelbaum, Richard, and Hugh Son. \"Unwinding at AIG Prompts Pasciucco to Ponder Systemic Failure.\" Bloomberg. 1 July 2009. Accessed 17 June 2015. . 10 Lanchester, John. \"Outsmarted: High Finance vs. Human Nature.\" The New Yorker. 1 June 2009. Accessed 19 June 2015. . 11 Lanchester. 12 Lanchester. 13 Tett, Gillian. Fool's Gold: How Unrestrained Greed Corrupted a Dream, Shattered Global Markets and Unleashed a Catastrophe. Simon & Schuster. New York. 2009. pp. 51-56. 14 Green, Richard K. and Susan M. Wachter. \"The American Mortgage in Historical and International Context.\" Journal of Economic Perspectives. 2005. Vol. 19, no. 4, pp. 93-114. 15 Tett. 16 Tett. 17 Tett. 18 Tett. 19 \"Put Out.\" The Economist. 6 Dec. 2007. Accessed 17 June 2015. . 20 Tett. 21 AIG Form 10-Q 3rd Quarter 2008. Securities and Exchange Commission. p. 116. Accessed 19 June 2015. . 22 Lewis. 23 Lewis. 24 Tett. 25 CNBC.com. \"Moody's Downgrades Residential Mortgage-Backed Securities.\" 10 July 2007. Accessed 17 June 2015. . 26 Markit Financial Information Services. \"Current Prices.\" Accessed 17 June 2015. . 27 Financial Crisis Inquiry Commission. p. 267. 28 Financial Crisis Inquiry Commission. p. 268. 29 Financial Crisis Inquiry Commission, Ch. 14. 30 Financial Crisis Inquiry Commission, 2010-08-31 Goldman Sachs - Valuation and Pricing Related to Initial Collateral Calls on Transactions with AIG, August 31, 2010. . 19 This document is authorized for use only by Ozde Oztekin in 2017. For the exclusive use of O. Oztekin, 2017. Established at the University of Michigan in 1992, the William Davidson Institute (WDI) is an independent, non-profit research and educational organization focused on providing private-sector solutions in emerging markets. Through a unique structure that integrates research, field-based collaborations, education/training, publishing, and University of Michigan student opportunities, WDI creates long-term value for academic institutions, partner organizations, and donor agencies active in emerging markets. WDI also provides a forum for academics, policy makers, business leaders, and development experts to enhance their understanding of these economies. WDI is one of the few institutions of higher learning in the United States that is fully dedicated to understanding, testing, and implementing actionable, private-sector business models addressing the challenges and opportunities in emerging markets. This document is authorized for use only by Ozde Oztekin in 2017