Working Paper: NBER ID: w12210
Authors: Francis A. Longstaff; Arvind Rajan
Abstract: We study the pricing of collateralized debt obligations (CDOs) using an extensive new data set for the actively-traded CDX credit index and its tranches. We find that a three-factor portfolio credit model allowing for firm-specific, industry, and economywide default events explains virtually all of the time-series and crosssectional variation in CDX index tranche prices. These tranches are priced as if losses of 0.4, 6, and 35 percent of the portfolio occur with expected frequencies of 1.2, 41.5, and 763 years, respectively. On average, 65 percent of the CDX spread is due to firm-specific default risk, 27 percent to clustered industry or sector default risk, and 8 percent to catastrophic or systemic default risk. Recently, however, firm-specific default risk has begun to play a larger role.
Keywords: Collateralized Debt Obligations; CDOs; Pricing; Credit Risk; Default Risk
JEL Codes: G1
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| firm-specific default risk (G32) | CDX spread (Y10) |
| industry-wide default risk (G33) | CDX spread (Y10) |
| economy-wide default risk (G33) | CDX spread (Y10) |
| market events (G14) | perceived risk of industry-wide defaults (G33) |
| market events (G14) | perceived risk of economy-wide defaults (E44) |
| three-factor portfolio credit model (G11) | CDX index tranche prices (G19) |
| three-factor portfolio credit model (G11) | RMSE for pricing individual index tranches (G12) |