Carmakers’ downgrades lead to risk management pile-up

When US car giants General Motors (GM) and Ford found themselves relegated to junk status earlier this summer, there was turmoil. Speculation had long been rife about the downgrades but the two are among the biggest global corporate borrowers, with an estimated $450bn in long-term debt.

Together with their finance arms, GMAC and Ford Motor Credit, they were among the most commonly traded credit derivative names and were widely included in ‘standardised’ tranches based on tradeable credit default swap (CDS) indices. They also constituted the largest single-name exposures in bespoke and managed collateralised debt obligations (CDOs).

Just before the downgrades, JPMorgan estimated that GM and Ford represented well over $10bn of long-term credit risk in existing CDOs alone. Also, a significant widening in the issuers’ spreads had already compelled some credit derivatives players to sell out of, or hedge, their positions. CDO managers and investors had also begun looking at their mark-to-market sensitivity.

Negative mark-to-market

Due to auto companies’ widespread inclusion in structured credit portfolios, most CDO tranches had already experienced a negative mark-to-market following widening of the auto sector’s spread before the downgrades.

Managers had to decide whether to hedge the names – by buying protection on them and reducing CDO tranche subordination – and investors had to determine whether to keep their positions long, given the higher volatility of these tranches.

Clearly not all of them did so. Immediately the names were downgraded, there were rumours spilling into the media of significant pain being felt, of massive trading losses and of hedge fund melt-downs. It looks as if most of the rumours were overblown as there have been no collapses so far. But there has certainly been pain, judging by the data emerging in some banks’ second-quarter trading results and in hedge fund exits.

Why was this, when there was plenty of time to prepare for the move? The credit spreads on the two issuers blew out for only a relatively short period, reaching a four-day peak in mid-May, and shrinking back to early April levels by mid-June.

Had bankers and investors seriously misjudged the market, or was the danger rather more insidious? Were they mismanaging or mispricing their risks? The evidence certainly suggests so.

Julien Turc, senior quantatitive strategist at SG CIB in London, says: “What I would call the ‘crisis’ was mostly related to correlation risks and less directly to GM and Ford, which held out relatively well. The downgrades were really not a surprise, and people had plenty of time to prepare for the [rating] agencies’ moves. What was more of a surprise to some was that these changes meant that their correlation assumptions no longer held.”

According to Mr Turc, these players were betting on very high correlation between credit names – relying on the assumption that all the spreads in a given basket of exposures would move, to some degree, in unison. But the GM and Ford spreads blew up on a standalone basis, and these entities were not hedged against that occurring. Moreover, Mr Turc says, these banks had perhaps underpriced the idiosyncratic risk in correlation, and overpriced the systemic risk.

“Those that did not have to react to the sudden changes would have been alright. Those that reacted to the correlation disruption by selling off or hedging their equity tranches at the worst point would have taken the biggest hits.”

So why were the dealers positioned like this? Largely it is because, in recent years, there has been good demand from investors to take risk on intermediate – or so-called mezzanine – tranches of credit risk through bespoke synthetic CDOs. According to Barclays Capital, dealers sold approximately $132bn of rated bespoke mezzanine CDOs to buy-and-hold investors over the past six years.

Dealers typically managed the associated risks by selling single-name CDS protection equal to the “delta” of the tranches. By doing so, dealers protected themselves against systemic or non-idiosyncratic small changes in spreads.

As long as correlation went up, the position would have increased in value, but as soon as the contained spreads begun to move in an uncorrelated fashion, implied correlation decreased or a name defaulted, the position would start losing money.

Lorenzo Isla and Jeff Meli, credit analysts at Barclays Capital in London, estimate that between one-quarter and one-third of these “long correlation” positions were transferred from dealers to hedge funds. The risk profile of the positions that the hedge funds assumed after hedging their own positions were similar to the dealers’ own – they would make money so long as spreads continued to move together or if implied correlation increased, and lose money if idiosyncratic volatility or implied correlation decreased. In other words, both dealers and hedge funds were in the same position when the rating agencies came along, downgraded GM and Ford and both names blew out in isolation.

Supply and demand mismatch

Unsurprisingly, this supply and demand mismatch caused correlation to reprice when risk shifted away from the mezzanine to the riskier, or so-called “equity”, tranches – causing losses in the long correlation portfolios of dealers and hedge funds. And, these losses prompted dealers to try (for the first time since CDO tranches began trading) to reduce their correlation positions significantly, which they found difficult as there was scant appetite for it. Mr Isla and Mr Meli say: “We believe this accelerated the re-pricing of equity and mezzanine risk, as correlation changed as a market-clearing mechanism to deal with this imbalance.”

Mr Turc agrees. “Both banks and hedge funds have correlation exposure, so in a sense they faced similar issues at the time of the disruption,” he says.

At this point, it is interesting to draw a parallel with the equity derivatives market, which boasts a wider range of active participants than the credit market. SG CIB, the investment banking arm of Société Générale, has extensive equity derivatives-based structured product business, much of which is geared towards retail. This activity, in isolation, would leave the bank with correlation risk.

Offsetting products

One way that SG’s equity derivatives group hedges its equity correlation risks is by developing offsetting products, and by tapping different investor sectors with distinct appetites. For instance, back in 2002, the bank began trading correlation swaps with hedge funds to unwind the risk that its equity derivatives structuring groups had assumed from putting together retail-aimed structured products.

Arnaud Sarfati, head of pricing and new products at SG CIB Equity Derivatives, says: “Hedge funds were already very familiar with, and trading, volatility as an asset – so it was an obvious next step to introduce correlation swaps to this sector. This was particularly useful as it dispersed risks between trading rooms and hedge funds, and many others have since followed, and the market developed. More recently, both we and others have started trading dispersion swaps as another means of doing the same thing.”

Though some of the higher-profile losses related to the GM and Ford downgrades and the so-called correlation skew were initially associated with the hedge fund community, these players were, in a sense, better off than the dealers.

Hedge funds can take more diversified bets, whereas many banks are positioned against client demand. When buy-side demand dried up, they would have faced some significant pressures.

Mr Turc adds: “Some hedge funds are held by partners, while some others managed to stabilise their investor base. Such players were able to absorb the short-term volatility.”

Moreover, he says that many of the large, experienced correlation hedge funds were not rushing to unwind their positions.

“Rather it was the newcomers, or intermediate-sized players in the credit markets – such as the convertible funds that have taken to correlation trading only recently, and that had large equity tranche exposures – that were the real sellers. Their lack of experience in this market counted against them.”

According to Farid Amellal, global head of credit derivatives at BNP in London, some dealers managed to weather the storm without too much damage while others were less fortunate. Those banks that manage their correlation books and credit exposures with smoothing, long-term risk management views will have been best off.

“They have not been the victims of panic reactions – as they will not have been driven to taking rash decisions. The problem is that a lot of the new market entrants are still experimenting with risk management technology, and parameters and do not appear to treat the risks in a controlled and measured manner.

Some of these younger and recent comers to credit correlation markets have been surprised by the violence of the movements, and might have panicked when things became dislocated. These reactions actually increased the losses and exaggerated the movements, and would also have resulted in trading losses,” Mr Amellal says.

What lessons have been learnt?

So did the so-called correlation market learn some lessons about risk management? Indeed it did, according to Alan Shaffron, head of European credit derivatives at Citigroup in London, who says: “The disruption certainly proved that, however good models are, you also need to take into account several other factors when running correlation trading books.” He adds, however, that the pricing and risk models were not necessarily wrong.

Most market participants have made significant advances over the past few years in the ways that they model correlation risk and have constantly reworked parameters. They will likely have to keep refining the models and producing new iterations as time goes on. But he does not believe there were any fundamental issues in the actual models.

“After all, the models were not intended to predict how correlation assumptions could change,” says Mr Saffron. “What some market participants did learn is that relying on models alone means that you are effectively trading with only one eye open. You really have to look at the full picture: flows, investor base preferences and positions, attitudes to risk and leverage.”

That is not all. Mr Amellal says that it appears “almost certain” that management at certain banks intervened and insisted traders cut positions at the worst possible time, creating a snowball effect and triggering new losses at other houses and hedge funds. He believes the entities that have taken the largest losses are those that suffered a dislocation in confidence and understanding between the trading and management function, at houses that have rushed into the credit markets or focused less on risk management and more on sales and distribution.

“Management confidence in traders’ ability and their risk management techniques is crucial, but it is not unusual for liquidation decisions to be made beyond the trading desks. It is a constant dilemma in any entity, but far less so at banks where management have greater familiarity with the risks and products and confidence in traders.”