The ever insightful Alea points us in the direction of rising correlation on credit indices.
Reuters reports that correlation - a measure of systemic risk - appears to be increasing:
Correlation sets a price on the market’s fear that a few bad apples will default versus its fear of an economic cataclysm.
Correlation on the 5-year investment-grade Markit iTraxx Europe index is now running at 45 per cent. At low correlations, the risk environment is one in which corporate defaults are individual and idiosyncratic - like Enron or Parmalat. The “few bad apples” scenario. But when correlations are high, the problem is systemic - the likelihood is of multiple, interlinked defaults. More of a bad apple tree scenario.
But the correlation figures Reuters reports aren’t “real-world”. They’re implied by market spreads - and consequently are susceptible to technical factors - trading driven, rather than fundamentally driven. Spreads are observed on a basket of CDO tranches.
At low correlations, CDO capital structures function well: idiosyncratic “bad apple” defaults in the underlying portfolio naturally impact the equity tranche, which performs its function as a buffer for more senior tranches. All well, good and how it should be.
But at high correlations, a CDO’s tranching comes under pressure. At high correlations, defaults occur system wide - the whole apple tree is bad. Not only does the equity tranche bear risk, as it should do, but others do too, since the large number of defaults eats right up the waterfall.
At high correlations then, the risklessness of a CDO’s AAA tranche, versus its equity tranche, is diminished. Accordingly, at high correlations, the spread on CDO equity tranches decreases (their relative, added value is less) while the spread on senior tranches increases (they’re relatively riskier). Alea has this graph:
Flipping that around then, you can imply correlation from the market by comparing the current spreads on CDO tranches. Thus Reuters’ figures.
What then might be the technical factors causing higher implied correlation? There are two big ones. On the one hand, the holders of super-senior CDO positions are looking desperately to sell. On the other hand, hedge funds, the holders of equity and mezz notes, are looking to stay put.
To tackle the second of those first: Hedge funds holding firm. Many are doing this because they have executed - and are executing - profitable delta-hedging trades on their CDO mezz and equity debt. In a delta-hedge, the fund sells protection on the CDO equity tranche and buys protection on the iTraxx Europe. That delivers a big cash payment up front (a big bonus now prime brokers are tighter-fisted) and generates returns whether spreads rise and fall. It’s basically a trade made on the assumption that implied correlation isn’t reflecting real-world correlation.
Back then to the first technical factor - that super-senior holders are looking to sell. In turn, there are three influences on that:
Firstly, as Felix Salmon observes, many super-senior positions have been leveraged by banks to deliver better returns (LSS conduits). That has distorted the risk involved in holding super-seniors.
Secondly, trouble with monoline insurers is threatening to force unwinding of negative basis trades. Banks are thus again, desperate to exit super-seniors.
And thirdly, at a fundamental level, investors are discovering that senior CDO tranches are actually far more risky than their ratings ever gave light to. Even for prices slightly below other AAA-rated debt, they’re still not worth it.
That last - in current markets perhaps utterly obvious - point is worth one final tangent. In a way it cuts right to the heart of the whole subprime fiasco. How did CDO tranches get AAA ratings? Because most rating models don’t - or didn’t - have a correlation metric.
A Triple-A CDO tranche rating reflects the chance of say - 50 - totally unlinked, totally individual, idiosyncratic, defaults occurring, each with its own exceptional circumstances. Of course that is triple-A, because it would be very unlikely to happen. What it doesn’t reflect is a slew of defaults as a result of a system-wide trend or pattern of behaviour - like subprime lending.
