RiskMetrics Journal - Winter 2009

Feb 9, 2009

We are pleased to bring you the 2009 issue of the RiskMetrics Journal. This issue contains four papers representing a diverse set of issues that our research team has encountered over the last twelve months.

In the first paper, Christopher Finger presents empirical tests on variations of the standard model for tranched credit derivatives, or synthetic CDOs. There is a rich literature of new model proposals or extensions, but little empirical work focusing on the typical application of the model. Christopher examines a dataset of historical tranche and underlying credit prices, starting from the onset of the credit index tranches as liquid instruments, and including much of the market upheaval of the last year. He performs a backtesting exercise, evaluating the standard model against both simpler and more complex alternatives. This exercise challenges two commonly accepted notions: first, that a sophisticated pricing model is needed to describe the relationship between tranche and underlying, and second, that the standard model can be extended to provide useful hedging information. While the standard model does stand up to the first challenge, its extensions fail on the second.

Our second paper also takes on an ambitious empirical study and challenges some commonly held notions. In this paper, Gilles Zumbach investigates the evolution of large covariance matrices. Having previously investigated volatility on the same datasets, Gilles turns to the joint problem of volatility and correlation. Certainly, one challenge in the study is to define meaningful ways to reduce the quantity of information, so that we can gain some intuition about how covariance evolves broadly. Gilles shows that the spectrum of the covariance matrix is actually quite static, with most of the interesting dynamics restricted to a small number of eigenvalues. This in itself would seem to support methods such as principal components analysis, where we concentrate on a small number of stable directions for correlation or covariance. Gilles’s deeper investigation shows, however, that the directions associated with the important eigenvalues change quite a bit, supporting an approach where correlation and volatility are both dynamic quantities.

In our third paper, we return to the pricing of credit derivatives. Luis O’Shea examines the relatively new market for loan-only credit default swaps (LCDS). After introducing this derivative and discussing a few salient differences between LCDS and traditional corporate CDS, Luis presents a number of pricing frameworks. He shows that even after adjusting for the most obvious difference between LCDS and CDS—that is, that the expected recovery rate on LCDS is significantly higher—a significant basis remains between these two derivatives. For firms that issue both loans and bonds, the LCDS market seems to imply higher default probabilities. Accounting for the next difference between the products—that LCDS can cancel without defaulting—does not appear to explain this basis either. Luis proposes a joint model for CDS and LCDS that posits uncertainty in a firm’s initial credit state, and shows that this model can indeed account for much of the basis we observe.

Finally, Stéphane Daul and Elena Gutíerrez Vidal present a set of case studies on insurance liability replication. Under modern valuation conventions, insurance firms increasingly use market-consistent techniques to value both their assets and liabilities. Consequently, the market sensitivities of both sides of the balance sheet contribute to economic capital and surplus-at-risk. Computing such risk measures, or for that matter managing the market risk sensitivities at all, presents significant practical issues, stemming largely from the complexity of insurance products. One means to address these difficulties is to represent the insurance liability portfolios using standard financial instruments. Stéphane and Elena present a procedure to construct such replicating portfolios, and examine a number of practical issues with the procedure through two actual insurance liability portfolios.

Christopher C. Finger