How J.P. Morgan Invented the Credit Default Swap, 1994

A Weekend in Boca Raton

In June 1994 about eighty people from J.P. Morgan's swaps group flew to the Boca Raton Resort and Club on the Florida coast for one of the off-sites that Peter Hancock, who ran the bank's derivatives business, had built into the department's calendar. Gillian Tett's account of the weekend, drawn from interviews with the participants a decade and a half later, describes the usual mixture of whiteboard sessions and poolside drinking, and one question that kept returning between them: interest rate risk and currency risk could both be stripped out of a loan and sold to somebody else, so why not the risk that the borrower simply failed to pay (Tett, 2009).

Nobody left Boca Raton with a finished product. What the group carried back to 60 Wall Street was a conviction that credit — the oldest risk in banking, and the one every bank had always held to maturity because there was no other choice — could be turned into a traded instrument. Within four years that conviction had produced the credit default swap, the synthetic collateralised debt obligation, and a market that grew from nothing to a notional $62 trillion before it broke. The instrument did not cause the collapse of 2008, but it decided where the damage landed.

The Exxon Problem

A concrete case arrived that autumn. On 16 September 1994 a federal jury in Anchorage returned $5 billion in punitive damages against Exxon for the grounding of the Exxon Valdez in Prince William Sound five years earlier. Exxon asked J.P. Morgan, its longstanding banker, for a credit line of up to $4.8 billion to cover the contingency.

Morgan could hardly refuse a client of that standing. Extending the line was nonetheless close to worthless in accounting terms: under the 1988 Basel Accord, a corporate loan carried a 100 per cent risk weighting and therefore an 8 per cent capital charge, which meant roughly $384 million of equity locked against a facility priced at a few basis points. Hancock's team proposed something new. Rather than decline the business or eat the capital, Morgan would keep the line on its books and pay a third party a fee to assume the default risk. Blythe Masters, then in her mid-twenties and running credit derivatives marketing, took the structure to the European Bank for Reconstruction and Development in London, which agreed to sell the protection in exchange for an annual premium.

Two things made the trade work. EBRD was comfortable with Exxon's credit, having no existing exposure to it, and Morgan could argue to its regulators that the risk it now held was EBRD's rather than Exxon's — and a multilateral development bank attracted a 20 per cent risk weighting instead of 100. Capital against the facility fell by roughly four-fifths. A relationship loan that destroyed shareholder value became one that created it, and the difference was a contract that had not existed the year before.

What Basel Made Expensive

Regulatory arbitrage was not a side effect of the credit default swap. It was the commercial engine.

Basel I had been written in 1988 to stop banks from competing on thin capital, and it did so with a blunt instrument: four risk buckets, applied identically to a AAA-rated utility and a distressed manufacturer. Any bank that could move exposure from the 100 per cent bucket to the 20 per cent bucket without actually reducing its economic risk earned free capital. Credit derivatives did that on paper in a single line. American regulators effectively blessed the technique in supervisory guidance issued during 1996, which set out when a credit derivative could be recognised for capital purposes, and the market's growth dates from that recognition rather than from Boca Raton.

BISTRO and the Synthetic Balance Sheet

Selling risk one name at a time was slow. In December 1997 a team under Bill Demchak launched the Broad Index Secured Trust Offering, known inside the bank as BISTRO, which referenced roughly $9.7 billion of credit exposure across 307 companies on Morgan's books. A special purpose vehicle sold about $700 million of notes in tranches, with investors in the lowest slices absorbing the first defaults and investors in the senior slices protected by everything beneath them.

Morgan thereby laid off the risk of a $9.7 billion portfolio while raising less than a tenth of that in cash. The gap — the great majority of the pool, sitting above the funded notes — was hedged with a super-senior swap deemed so remote from loss that no one expected to pay on it. Rating agencies assigned the senior tranches investment-grade ratings on the basis of diversification and historical default correlations, and the structure was promptly copied across Wall Street. Every synthetic CDO of the following decade, including the mortgage-referenced deals that detonated in 2007, descended from that December transaction.

What the copies changed was the reference pool. Morgan had used its own corporate loans, which it knew intimately and continued to service. Later issuers assembled pools of residential mortgage-backed securities chosen precisely because they were correlated, and sold the risk to buyers who read a rating rather than a loan file. Darrell Duffie's assessment for the Bank for International Settlements, written as the crisis was breaking, noted that credit risk transfer had genuine stabilising properties in principle while warning that the instruments had outrun the infrastructure — the confirmation backlogs, the absence of central clearing, the opacity of who held what (Duffie, 2008).

Writing the Rules, and Removing Them

A contract that pays out when a borrower defaults needs an agreed definition of default. The International Swaps and Derivatives Association supplied one in its 1999 Credit Derivatives Definitions, revised substantially in 2003 after restructuring clauses produced disputes over whether a negotiated debt exchange counted as a credit event. Standard documentation turned a bespoke agreement into something approaching a security.

Regulatory questions were settled in the other direction. Brooksley Born, chair of the Commodity Futures Trading Commission, issued a concept release in May 1998 asking whether over-the-counter derivatives needed oversight. Alan Greenspan, Robert Rubin and Arthur Levitt opposed her publicly; Congress imposed a moratorium on CFTC action that autumn; Born left office in June 1999. Mark Brickell, an ISDA chairman who had come from Morgan's swaps desk, spent those years arguing that the industry's own documentation and collateral practice were a better discipline than statute. The Commodity Futures Modernization Act, signed on 21 December 2000, placed credit default swaps outside the reach of both the CFTC and the SEC and pre-empted state gaming and bucket-shop laws that might otherwise have voided contracts written by parties with no exposure to the underlying borrower.

That last provision mattered more than it looked. Insurance requires an insurable interest; a credit default swap, freed from those statutes, did not. Anyone could buy protection on a company whose bonds they had never owned, which is what turned a hedging tool into a market where the notional amount outstanding on a given borrower could exceed its debt several times over.

From Hedge to Position

Doubling roughly every year between 2001 and 2007 is not the growth path of a hedging tool. Most of that notional represented positions taken by dealers and hedge funds against one another, offsetting and re-offsetting, rather than banks laying off loans they had made. Gross notional overstated real exposure by a wide margin — a bought and a sold contract on the same name mostly cancel — but the contracts were bilateral, so cancellation was economic rather than legal, and every trade left a counterparty who had to be good for it.

Official opinion was admiring. Speaking to the Federal Reserve Bank of Chicago's annual conference on bank structure in May 2005, Greenspan told his audience that "the use of a growing array of derivatives and the related application of more-sophisticated approaches to measuring and managing risk are key factors underpinning the greater resilience of our largest financial institutions." Warren Buffett had reached the opposite conclusion in his 2002 letter to Berkshire Hathaway shareholders, calling derivatives "time bombs, both for the parties that deal in them and the economic system" and "financial weapons of mass destruction." Both men were describing the same instrument, and the difference between them was a judgement about whether risk that had been dispersed was risk that had been reduced.

René Stulz, reviewing the evidence after the crisis, concluded that credit default swaps were neither the villain of popular account nor the stabiliser of the official one: they transferred risk efficiently when counterparties were sound, and they concentrated it catastrophically when the largest sellers turned out to be a handful of firms with no capacity to pay (Stulz, 2010).

Cassano's One Dollar

One firm proved the point. AIG Financial Products, a London-based unit of the American insurer run by Joseph Cassano, had been selling protection on the super-senior tranches of other people's CDOs since 1998, collecting premiums of perhaps fifteen basis points on paper that carried an implicit AAA. By the end of 2007 AIGFP had written roughly $533 billion of notional credit protection, of which about $78 billion referenced multi-sector CDOs stuffed with subprime mortgages.

On an investor conference call on 9 August 2007, with the asset-backed market already seizing, Cassano offered a summary that has outlived him: "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 one dollar in any of those transactions."

He was arguably right about ultimate defaults and entirely wrong about the mechanism that killed his employer. AIG's contracts carried collateral triggers tied to market prices and to AIG's own credit rating. As CDO marks fell through 2007 and 2008, Goldman Sachs and others demanded cash against positions that had not yet lost a penny to actual default, and the downgrades of 15 September 2008 triggered the rest. The Federal Reserve's $85 billion credit facility of the following day, later expanded toward $182 billion, was extended to an insurance company whose regulated insurance subsidiaries were solvent. What failed was a derivatives book written in London against a parent company's credit rating.

What the Auctions Revealed

Fear through September 2008 ran on the assumption that Lehman Brothers' bankruptcy would trigger payouts the system could not absorb. Something near $400 billion of gross notional referenced Lehman. The auction held on 10 October 2008 set the recovery rate at 8.625 cents on the dollar, and when the offsetting positions were netted the actual cash that changed hands came to roughly $5.2 billion, settled without a failure.

Netting, in other words, worked. The Financial Crisis Inquiry Commission's verdict was nonetheless severe: over-the-counter derivatives, and credit default swaps in particular, "contributed significantly" to the crisis by amplifying losses through collateral calls, by concentrating exposure in a few dealers, and by making it impossible for any regulator to see where the risk sat (FCIC, 2011). The instrument passed its settlement test and failed its transparency test in the same month.

After the Bang

Reform followed the diagnosis. ISDA's Big Bang Protocol of April 2009 hardwired auction settlement into every contract and created Determinations Committees to rule on credit events; standard coupons of 100 and 500 basis points replaced bespoke pricing, which made positions fungible and therefore clearable. ICE Clear Credit began clearing index trades in March 2009, and Title VII of the Dodd-Frank Act, signed on 21 July 2010, made central clearing and trade reporting mandatory for standardised swaps. Notional outstanding fell for a decade, to under $10 trillion by 2019 on BIS figures, much of that reduction coming from portfolio compression rather than closed positions.

Arguments over the instrument continued in specific cases. A Determinations Committee ruled in March 2012 that the Greek debt exchange constituted a restructuring credit event, settling contracts at 21.5 cents and answering a question that had hung over the eurozone's long sovereign crisis for two years. In 2012 J.P. Morgan itself lost about $6.2 billion in its Chief Investment Office when Bruno Iksil accumulated a position in the CDX.NA.IG.9 index large enough that the market could see him coming. Nine years later the same family of instruments — total return swaps rather than credit default swaps, but the same problem of a position invisible to everyone except the firm's counterparties — produced the Archegos blowup.

Blythe Masters, by then head of J.P. Morgan's commodities business, offered her own defence in 2008: "I do believe CDSs have been miscast, much as poor workmen tend to blame their tools." The survey literature has largely come around to a version of that view, finding that credit default swaps improved price discovery and lowered borrowing costs while creating incentives for lenders to stop caring about the borrowers they had insured (Augustin et al., 2014).

Both claims can hold at once, because the instrument the Boca Raton group imagined and the market it became were not the same object. Hancock's people wanted to sell the risk of loans their bank had underwritten and would continue to monitor. What grew instead was a market in which the buyer of protection frequently had no loan, the seller frequently had no capital, and the reference borrower was never told. Morgan's own team understood the distinction well enough that the bank largely stayed out of subprime synthetics and emerged from 2008 as a buyer of wreckage rather than a piece of it. Demchak went on to run PNC; Cassano left AIG in February 2008 with the collateral calls already running, and was never charged.