Fitch Introduces First Statistically Valid U.S. Multiborrower CMBS Model

Fitch Ratings said late Friday that it has introduced the first statistically valid U.S. CMBS multiborrower rating model for evaluating the credit risk of commercial mortgage pools.

According to the company, the development of the new model follows an extensive study of over 32,000 CMBS fixed-rate conduit loans originated between 1994 and 2001, totaling $173 billion, and how they performed through the end of last year.

“The maturation of the CMBS market has produced quality data which facilitates the development of robust quantification of the key variables affecting probability of default, probability of loss and loss severity,” said Managing Director and CMBS group head Susan Merrick.

Fitch’s new model uses over 500,000 different simulations, now includes statistical methods to identify the greatest contributing factors for probability of default, probability of loss and loss severity, along with the more sophisticated method of quantifying the effect of portfolio composition on a pool’s credit risk profile.

Perhaps the most unique attribute of Fitch’s new model is that it breaks the credit risk of an individual loan into 3 components that have multiple contributing quantitative variables, which are as follows:

• Probability of Default
• Probability of Loss
• Loss Severity

Each of these variables are based on multivariate regression analysis using loan attributes and historical CMBS performance data. The subordination levels for all rating categories are derived from a pool’s simulated loss distribution, which reflects the correlation among loans.

During the model calibration period, Fitch conducted an analysis of over 20 Fitch-rated deals issued this year using the new model to compare subordination levels, as well as on 20 CMBS deals that it did not rate in 2006. Both tests yielded similar subordination levels as in Fitch’s initial analysis, however there were specific concentrations and correlation among deals with the largest variances that are more accurately captured with the new model.

“Subordination levels on some CMBS deals came out slightly better while some fared worse, further proving that Fitch’s new model provides a more accurate characterization of credit risk and will further improve transparency in the CMBS market,” said Merrick.

Fitch said it will begin using its new multiborrower CMBS model on new transactions beginning March 1 of this year.