Credit Models and the Crisis: A Journey into CDOs, Copulas, Correlations and Dynamic Models

1

The Ingenious Hidalgo Don Quixote of La Mancha, Miguel de Cervantes, 1605 and 1615.

2

See primarily Shreve (2008) and also, for example, Donnelly and Embrechts (2009). Szego (2009) gives a much broader overview of the crisis, with some critical insight that is helpful in clarifying the more common misconceptions.

3

See, for example, Torresetti and Pallavicini (2007) and the Fitch Ratings report “First Generation CPDO: Case Study on Performance and Ratings”, published before the crisis in April 2007, stating “[. . . ] Fitch is of the opinion that the past 10 years by no means marked a high investment grade stress in the range of AAA or AA.”

4

By Oliver Biggadike, 24 November 2009, Bloomberg.

5

See, for example, the FBI Mortgage Fraud Report, 2007, www.fbi.gov/publications/fraud/mortgage_fraud07.htm.

6

See “Request for Comment: Update to Global Methodologies and Assumptions for Corporate Cash Flow CDO and Synthetic CDO Ratings”, 18 March 2009, Standard & Poors.

7

The attachment of the CDX tranches are slightly higher, reflecting the average higher perceived risk (as measured, for example, by the CDS spread or balance sheet ratios) of the liquid investment grade North American names.

8

The default count distribution is obtained via a homogeneous finite pool one-factor Gaussian Copula: the homogeneous 10-year probability of default calibrated from the DJ-iTraxx S5 CDS term structure was 9.3%.

9

Let us stress again that the actual outstanding notional in each period would ...