Dealers are issuing huge volumes of single tranche collateralized debt obligations based on untested correlation assumptions and this could mean everything from the pricing of the deals to the hedge is wrong, according to several investors. Frederick Horton, managing director at Trust Company of the West, said CDO dealers' trading books "look like an accident waiting to happen." Risk management of bespoke CDO tranches is an opaque process and no-one really knows how to manage correlation, he added. Horton made the comments in a speech at the ABS West conference earlier this month but declined to elaborate when contacted afterward.
The investors and managers are concerned that models used for managing risk and pricing deals are based on a paucity of historical data. It is difficult for managers to bring value to single tranche deals because the pricing of aspects such as substituting credits is not transparent. When credits in bespoke tranches are substituted, dealers also need to adjust their delta-hedge of the portfolio, which they factor into the price, explained an analyst. When determining the costs of the trade a dealer will take into account its entire credit portfolio, including whether it wishes to take more exposure to the particular credit being pitched. Not only is this a process that dealers will not wish to share with CDO managers, it is one that will create greatly divergent prices between firms depending on whether the new credit is attractive for their trading book, explained the analyst.
Brian Colgan, fixed-income manager with a portfolio of some USD40 billion at WestLB in New York, is also uncomfortable with correlation assumptions saying it is one factor delaying him from purchasing single tranche synthetic deals.
CDO dealers, not surprisingly, think the correlation assumptions are accurate. Alex Reyfman, global head of credit derivatives research at Bear Stearns in New York, said there is a liquid and transparent market for correlation and the risk profile of correlation books that are marked-to-market is pretty clear. Large market moves should not bring any surprise profit and loss changes, he added.
Another strategist explained that most dealers are long correlation so if spreads all move in the same direction, banks will make money. In the worst case scenario, a credit default does not lead to a general widening of spreads. This is because the dealer will have to pay out on the default, but will not gain from spreads on the other positions widening. This happened when Parmalat defaulted because the market decided it was an isolated case. The banker, however, estimated that around 10% of the dealers' portfolios would have to default without spreads on the remainder of the portfolio being affected for there to be a serious risk to the CDO houses.