Tranches of trepidation

Recent correlation and spread dynamics in the credit market have made some dealers nervous that the tranche market may soon become decidedly treacherous, mirroring the events of the 2005 correlation crisis. Navroz Patel reports

Stress in the US housing and mortgage sector is causing concern among some participants in the US tranched credit derivatives market. Painful memories of the 2005 correlation crisis, when idiosyncratic risk in the form of downgrades and spread widening among auto sector credits caused huge mark-to-market losses, have been reawakened.

This time around, credit default swaps (CDSs) referencing the debt of four US homebuilders (Dallas-based Centex, Miami-based Lenmar, Michigan-based Pulte Homes and Pennsylvania-based Toll Brothers) and two financial services firms (California-based Countrywide and Pennsylvania-based Radian) have been identified by traders and analysts as a possible source of idiosyncratic risk and future significant tranche price volatility. With a 40% recovery rate, the current CDX North America investment grade index equity (0-3%) tranche can tolerate six defaults before mark-to-market losses completely consume principal.

"There have been definite similarities between the pricing of risk in the current market and the period leading up to the 2005 correlation crisis," says Jeffrey Rosenberg, head of credit strategy research at Banc of America Securities in New York. "That's not to say we see the market evolving into another crisis, but the disconnect between credit fundamentals and tranche pricing is noteworthy to say the least."

Rosenberg argues that equity tranches are currently being priced at levels that suggest a very benign credit market, despite the significant idiosyncratic risk associated with subprime mortgage-related stress and ongoing leveraged buyout activity. On March 23, five-year CDX index equity (0-3%) tranche correlation rose to 15% for the first time since July 29, 2005, in the aftermath of the correlation crisis. Equity correlation entered 2007 trading at a level of 12.2% on January 2.

In effect, increasing correlation equates to lower premium for taking on the first-loss risk. High correlation is positive for equity tranches, as it means it is more likely that either no credits will default or all will default - and as even a single default will prompt mark-to-market losses on an equity tranche, higher correlation is associated with a lower expected loss. In the current market, each percentage point increase in correlation lowers the premium received by around 60 basis points.

The perception of risk among some participants in the correlation market didn't even appear to have been shaken as volatility initially erupted across credit and equity markets on February 27 (Risk April 2007, page 22). On February 28, equity correlation traded at 13.7%, around the same level it had traded during the week or so before the market shake-up. Indeed, equity correlation stayed below the 13.7% level for most of the ensuing week, hitting a low of 12.3% on March 5.

"Underlying credit markets saw a major shift in the pricing of risk, but the following day correlation demand came in reflective of pre-February 27 levels," explains Rosenberg. "The mentality appeared to be: spreads will have to tighten, there can't be corporate defaults and why would one enter a losing trade like buying protection?" Figure 1 illustrates how equity tranche correlation has broadly trended upwards since the third quarter of last year.

The questions raised by Rosenberg and his colleague Glen Taksler as to whether equity tranches are much riskier than their price suggests are part of a growing concern in the market. Several hedge fund managers and analysts who spoke to Risk say that in recent weeks they have received a steady stream of enquiries from major investors and clients who wonder if they were somehow about to be blindsided by the re-emergence of idiosyncratic risk.

For their part, dealers believe a new correlation crisis is unlikely. They point to the relative absence of wild price swings in late February and March and robust liquidity as evidence that the correlation market has matured into a more balanced and stable business.

"We are not yet seeing genuine distress among the US homebuilders that trade in CDSs and are included in structured credit portfolios - they are still fundamentally investment-grade credits trading at not terribly wide spreads," says Fahad Roumani, head of correlation risk management at JP Morgan in London. "The situation is very different to US autos in May 2005, both in terms of position concentrations and the level of stress in the sector, and so far the impact on the correlation market has been muted."

The current situation is also very different in that one of the main hedging practices that came spectacularly unstuck in 2005 - namely, delta hedging of the equity tranche with the mezzanine tranche - is largely out of favour. While contemporaneous models had suggested that a leverage of 16 times (that is, buying mezzanine protection in an amount 16 times larger than the notional of equity protection sold) would leave the position hedged, the rise of idiosyncratic risk meant this prediction was to prove flawed. Instead, the actual required leverage turned out to be around 35 times, leaving most hedgers reeling from huge mark-to-market losses. Two years ago, this strategy lost 23% in the space of six weeks.

Over that same April 1 to mid-May period in 2005, the still-popular index-hedged equity strategy lost 8%. The model-predicted leverage ratio when hedging equity with index protection buying was around six times, but the empirical leverage required during the 2005 correlation crisis turned out to be around negative five. This meant that not only did hedgers have on the wrong amount of index protection, they also needed to sell index protection rather than buy it.

The depth and breadth of participation in the correlation market is certainly a lot stronger now than it was two years ago, with hedge funds and bank proprietary trading desks much more active. "Market-making correlation desks hold less outright correlation risk now than before May 2005, but do have basis risks between index and bespoke portfolio positions, and across different types of bespoke portfolios," Roumani says.

Brian Zalaznick, global head of fixed-income structuring at Barclays Global Investors in San Francisco, agrees that hedge funds have become much bigger buyers of equity tranche risk. "Dealers are still very hungry to do equity-type managed transactions with managers with good track records in corporate credit," he adds.

One popular way dealers have sought to offload equity risk since late 2005 is through credit constant proportion portfolio insurance (CPPI) transactions. These deals have a strategy that dynamically allocates assets between a cash portfolio and credit protection selling portfolio, with leverage increased through the opportunistic selling of additional protection.

Other vehicles for offloading equity risk include: structures that use premiums received to build up a subordination fund so that the equity can get a rating; and zero-coupon equity, where the investor receives no premium and instead is paid a portion of notional upfront, and can collect the remaining notional minus any losses at the end of the deal.

Although the major dealers say they have been successful in widely distributing a large amount of the equity tranche risk they had previously accumulated as a consequence of rampant client demand for mezzanine product, some participants suggest that data implies the market has not distributed risk as evenly as some firms claim.

In the second half of 2005, post-correlation crisis, and until around the end of last year, correlation remained at relatively low levels, indicating that equity tranche protection buyers may have been more active in the market. This is suggestive of a disciplined reduction of equity correlation exposure by dealers, concedes Banc of America's Rosenberg. On September 29 2006, for instance, equity correlation traded at 10.6%, compared with 9.25% at the end of the third quarter of 2005; it was largely rangebound between 7% and 11% in the interim. "But for the most part since mid-2006, equity tranche spreads have been tightening, and that implies the discipline may have gone away. Certainly, somewhere in the system, large long correlation positions are being held," adds Rosenberg.

JP Morgan's Roumani believes outright correlation risk is now concentrated mainly with hedge funds and proprietary trading desks. Some correlation traders tell Risk that while structuring desks can lay off a lot of correlation risk, some choose not to as they view risk retention as likely to be profit-turning position.

"Dealers' correlation axe is still present, albeit in a much less pronounced way than was the case a couple of years ago," says Stephen Siderow, co-founder and president of New York-based hedge fund manager BlueMountain Capital Management. "Nonetheless, the balance of flows is such that the demand for rated mezzanine risk is still much greater than the demand for unrated equity risk. Equity is relatively cheap because of the demand from dealers to offload that risk."

This structural imbalance proved to be especially advantageous to BlueMountain in March. As credit spreads widened, dealers began to pre-position their books by shorting equity tranche risk in anticipation they would transact a significant volume of bespoke mezzanine deals for large institutional investors. This dealer behaviour during a spread sell-off coincided with the hedge fund manager's launch of a new equity tranche-focused vehicle.

"We had believed there would be opportunities for putting risk on in our new fund over the course of a six- to nine-month investment period," explains Siderow. "As it turned out, we put around 75% of the risk on by the end of March."

BlueMountain launched the $88 million BlueHorizon II fund, which is focused on absolute value opportunities in CDO equity, on March 1. "The market is much deeper and more liquid than it was 18 months ago. Back then, we could only trade at $10 million or $20 million notional size," says Siderow. "Now, we are able to do trades of $50 million or $100 million without problem."

BlueMountain is cognisant of the dangers posed to traditional delta-hedged tranche strategies by idiosyncratic risk, and actively manages that risk with single-name CDSs. This kind of approach means that while a delta-neutral amount (on a notional basis) of index and single-name short trades is entered into, the precise distribution of that delta across these trades reflects a specific view on individual credits. In other words, each specific name can have a different delta and be hedged to a different extent.

This technique may be beneficial given the current crop of poorly performing subprime mortgage-related and homebuilder investment-grade names in US portfolios. "Arguably, the fact that the housing-related names currently included in tranches are of relatively high quality makes them a greater risk to equity tranche holders than was the case with the auto sector names in 2005," says Rosenberg. "So, if the spreads on housing-related names were to suddenly widen, it could amount to a lot of negative convexity and a lot of losses." Figure 2 illustrates the proportion of overall index risk that market pricing currently attributes to the equity tranche.

Dealers are correct in their assertion that they have a much broader base of clients to whom they can lay off equity correlation risk, says BlueMountain's Siderow. But in his opinion, there remains a big open question over this evolution: "How many of these newer players truly understand correlation risk, and will they stay in the market at a time of great stress?"

If newer, perhaps less savvy, players do bolt for the exit door at a time of market stress, some of the major players are waiting in the wings to snap up equity tranches when correlation levels bottom out. As one correlation trader puts it, a number of major hedge funds are "still kicking themselves" for not buying cheap equity tranches during the correlation crisis of 2005. "I know for a fact that they would dive in and put their capital to work massively if we see that kind of dramatic price action again," he says.