Many banks have ended up owning complex structured products where trading has ground to a halt. How can a financial institution efficiently mark these complex securities to market using state of the art risk technology? A regional bank in the United States summarizes the process of best practice FAS 157 valuation in its December 31, 2008 10-k filing with the Securities and Exchange Commission.
The Kamakura On-Line Processing Service brings the full power of the Kamakura Risk Manager software system and the KRIS default probability service to clients who don't have the time or the staff to install or operate the system for themselves. Best practice FAS 157 valuations for collateralized debt obligations was summarized in this 10-k report for December 31, 2008:
For those securities that cannot be priced using quoted market prices or observable inputs a Level 3 valuation is determined. Given the conditions in the debt markets, the absence of observable transactions in the secondary and new issue markets, and the overall inactivity of the market we determined that some of our trust preferred securities should be classified within Level 3 of the fair value hierarchy. In certain situations we use independent third parties to help prepare the valuations for some of our trust preferred securities.
The approach used to determine the fair value of some of our trust preferred securities involved the following steps:
|
|
•
|
|
Estimation of the credit quality of the collateral using average probability of default values for each issuer (adjusted for rating levels);
|
|
|
•
|
|
Consideration of the potential for correlation of default probabilities among issuers within the same industry (e.g. banks with other banks);
|
|
|
•
|
|
Assumption of loss given default was assumed of 95% (i.e. a 5% recovery);
|
|
|
•
|
|
Forecasting of cash flows for the underlying collateral and application to each collateralized debt obligation (CDO) tranche to determine the resulting distribution among securities;
|
|
|
•
|
|
The discounting of expected cash flows to calculate the present value of the security;
|
|
|
•
|
|
Modeling of the calculations were modeled in several thousand scenarios using a Monte Carlo engine with use of the average price for valuation purposes.
|
The effective discount rates are highly dependent upon the credit quality of the collateral, the relative position of the tranche in the capital structure of the CDO and the prepayment assumptions. The approach used to fair value our other multi-issuer CDOs involved the use of a model that was developed by a leading risk management solution providers. The approach used to determine the fair value of these CDOs involved the following steps:
|
|
•
|
|
The one and five year default probability was determined for each issuer in the pool based on the Kamakura Risk Information Services model;
|
|
|
•
|
|
The default probability for insurance issuers was developed using 22 macro factors which drive the default for mid-size insurance companies through the use of a logistic regression model;
|
|
|
•
|
|
Twenty-seven macro factors were candidate variable for macro driven default rates; and
|
|
|
•
|
|
100,000 Monte-Carlo simulations were run in annual time stops until maturity to derive a fair market value.
|
The approach utilized by our consultant uses a multifactor default model incorporating market/macro economic factors as well as unsystematic factors. This approach establishes the line between market and credit risk and provides a framework for dynamic instead of static market and credit risk modeling. Reduced form default probabilities by our consultants are seen as the most modern and most accurate approach to default probability assessment. The use of 100,000 scenarios was done to minimize statistical error.
We evaluated current defaults and deferrals from trustee reports, structural support within the CDO, and the coupon rate at the Cusip level compared to the coupon on the tranche. In evaluating these factors we examined the trustee reports to determine current payment history and the structural support that existed within the CDO at year-end. We incorporate the modeling for evaluating future deferrals and defaults and coupon rates based on the current swap curve to project future cash flows. Several scenarios were done involving different levels of liquidity risk. Because of the lack of an active market, the determinations of fair value assume that market participants would utilize the same assumptions in determining a price.
For more information on how to access the Kamakura On-Line Processing Service ("KOPS") please contact info@kamakuraco.com.
Donald R. van Deventer
Kamakura Corporation
Honolulu, March 17, 2009