Derivatives 2005 - Correlation change underway
Broader credit spreads have come back quickly from the effect of hedging of correlation losses. Although the slump in credit correlation in May caused losses for both dealers and their hedge fund clients alike, dealers say that models were not to blame. Nevertheless, models will ultimately change. Jean Haggerty reports.
With May in mind, dealers are waiting to see what impact Delphi’s bankruptcy filing has in terms of tranche downgrades. Rising raw material and labour costs prompted Delphi, the largest US auto supplier, to file for bankruptcy on October 8.
Delphi has the potential to rock the markets in a similar fashion as the GM and Ford downgrades did this spring. Fitch says that Delphi featured in 35% of its bespoke synthetic collateralised debt obligations (CDOs) and Standard & Poor’s estimates that Delphi is in 50% of its European deals. At default, Delphi could have significant repercussions for tranche markets.
If a default does indeed cause a wave of two and three notch downgrades, as S&P’s report suggests, some dealers think selling mezzanine and senior tranches might be at least as intense as in equity. S&P noted that a single default in one of the most commonly referenced names would cause nearly half of all outstanding synthetic CDOs to be downgraded by an average of two and a half notches.
What happened this May was an aberration, dealers say, when supply and demand was simply out of whack. “Unfortunately, models do not say how the market will move and our business is driven by supply and demand [in different traches],” said Mark Stainton, head of credit correlation at Deutsche Bank in London. Deutsche Bank is using the same models for valuing correlation as it was this spring.
“Some people say models were wrong and that they could not handle [the repricing], but you have to remember that models are only as good as their inputs,” said Torben Botts, senior correlation trader at ABN AMRO in London.
Although there has been much discussion about the problems with the models - or even the inputs to the models - much of this is misplaced, said Matt King, a credit strategist at Citigroup.
“The observation that people are starting from is that equity sold off much more than the model said it ought to, and mezzanine rallied much more than the model said it ought to. But the reason for this is that everyone in the market was hugely long equity and hugely short mezzanine. Had the positions been reversed, then equity would have outperformed and mezzanine underperformed,” King said.
Therefore, the problem is neither with the model nor with the inputs to the model, but with correlation traders' overreliance on models, he added.
“It is the golden lesson of strategy. No trade is a good trade when 150 people have got it on before you. Valuation, or what the model tells you, is only one part of the puzzle,” King said.
Ultimately, models will change, but not because of May, dealers say, noting that the approach to pricing and hedging correlation has been evolving for some time now. Gaussian Copula models are still used, but in recent years models have changed from an implied correlation to a base correlation framework.
Base correlation is the implied correlation of bootstrapped equity tranches on credit portfolios. Effectively, it is a common language for talking about correlation skew, and its big selling point is its consistency across the capital structure. Banks have been using base correlation approaches in-house for years. Increasingly, they have been sharing their valuation techniques with investors. JP Morgan in March 2004 made public its approach to calculating implied correlations from standardised tranche spreads.
According to dealers, the change from an implied correlation to a base correlation framework altered the way market participants hedge and look at risk. Unlike implied correlations, base correlations are constant when measuring credit spread deltas. When the credit correlation market took root in mid-2000, the correlation numbers that were used were estimates and it was not clear how different dealers were looking at correlation. By demanding more transparency in the correlation market, accountants precipitated the changes that set in motion the creation of individual tranche markets. US accounting standard EITF 02-3, which took effect in early 2003, required dealers to use independent price verification and recognise income from derivatives upfront. Even though it was initially intended to apply only to energy derivatives, EITF 02-3 covers all over-the-counter derivatives transactions.
“What happened in May will force a lot of people to rethink whether they are happy with their models and how it handles risk and market movements. We never saw anything like what we saw in May,” Botts said (see chart).
For now, today’s models suit their purposes well enough and only slight differences exist between the models in use by various dealers.
According to Oldrich Masek, co-head of global structured credit and alternative investment at JPMorgan in London, the next move for the industry is to use market-based implied correlation instead of model-based correlation.
"This significantly changes the competitive landscape for asset managers of single tranche programmes. The better managers need not only be able to mange credit risk but also understand the trading of correlation and its impact on the valuation of tranches - this will certainly create tiering in the market in terms of their performance and the type of structures they can suitably manage," he added.
According to Masek, it is hard to say how many managers have the capability to manage credit correlation. A plethora of managers who have been historically long-only focused and who have limited credit derivatives experience have become involved in the market, he noted. In light of this, JP Morgan has opted to focus on working with a few managers and to do fewer but more innovative products.
"The only way for managers to demonstrate that they understand correlation is to gain experience trading bespoke tranches," Masek said. In the meantime, investors need to come to grips with all of the different risks that they are running, he added.
For the bulk of the last year and a half, long-only, Single-A to Triple-A, mostly mezzanine deals reigned. In recent months, however, the focus on leveraged super-senior deals has intensified. Although leveraged super-senior deals have only been a force in the market for five to six months, credit correlation players are already referring to them as the product of the year. A leveraged super-senior deal takes the senior tranche and levers that, normally by seven to 10 times, to deliver a higher coupon.
“The downside of the trade is that it can be terminated if its triggers are hit,” ABN’s Botts said.
In general, the popularity and increased activity in super senior trades has concentrated attention at the higher end of the capital structure. “There is some appeal for mezzanine, but various features can still be added to deals to enhance the coupon,” Botts said. Dealers have jumped on leveraged super-senior trades, but linking the coupon to different parts of the capital structure is also a possibility. “You can also link an interest rate or equity component to a credit derivative,” he noted, adding that ABN has pegged credit contingent products as a future growth spot.
“Say spreads are tight and correlation is not helping the mezzanine [tranche], you can add an interest rate range accrual to enhance a coupon or you can add a constant maturity credit default swap to enhance the coupon,” Botts said.