Kamakura Releases Basel II Credit Model Test Results, Advanced Kamakura Default Probabilities KDP Semiannual Model Upgrade Cycle Announced

HONOLULU, May 12, 2003: Kamakura Corporation announced today that it is releasing to clients its latest results on credit model performance in accordance with the Basel II guidelines from the Basel Committee on Banking Supervision. Kamakura, one of the world's leading suppliers of default probabilities and risk management software, also announced that it has completed its first semi-annual upgrade of its three default probability models. Kamakura currently distributes a Merton structural model default probability (KDP-ms), a reduced form default probability based on work by Robert Jarrow and Sudheer Chava (KDP-jc), and a hybrid model KDP-jm which adds the Merton default probability KDP-ms to the Jarrow-Chava inputs. All three of the models have been substantially enhanced based on inputs from Kamakura's default probability client base, with impressive improvements in model performance.

"Our day to day work with large institutional clients on KDP default probabilities has led to some important insights that have allowed us to substantially improve the results that Sudheer Chava and I derived previously. Our earlier work was based on a study of all listed companies in the United States from 1963 to 1998. The new improvements reflect evaluations of a larger number of defaults, defaults in more countries and a longer time period," said Robert A. Jarrow, Kamakura Managing Director for Research and Ronald and Susan Lynch Professor of Investment Management at Cornell University. "We have added a number of additional explanatory variables to the original mix of accounting ratios, relative size, and equity-related data and found a very impressive boost in accuracy as measured by the ROC accuracy ratio." The ROC accuracy ratio is a standard statistical measure of model performance, with a score of 100 indicating a perfect model and a score of 50 indicating a model no better than random chance.

"The basic KDP-jc scored an accuracy ratio of 92.74, as Kenji Imai and I discussed in our new book Credit Risk Models and the Basel Accords," said Dr. Donald R. van Deventer, Kamakura's Chairman and Chief Executive Officer. "The new versions of KDP-jc and the hybrid model KDP-jm are scoring much higher than this on our commercial data base with data from 1989-2003, a more difficult environment to model than the 1963-1998 research data." Commenting on the Kamakura Merton model, Dr. van Deventer added, "We have enhanced the mapping from the theoretical Merton default probabilities to a KDP-ms Kamakura Merton default probability that is carefully benchmarked to the actual defaults experienced over the 1989-2003 period. Our clients are well aware that the Merton model, if not carefully mapped based on actual experience, tends to produce default probabilities that are unreasonably bunched at levels that are too low or too high to be realistic. We are very pleased with our results in this regard."

The Basel II credit model tests provide Kamakura clients with quantitative model performance measures using advanced statistical tests that are consistent with the highest standards of statistical practice. The proposed new Basel Accords require that banks consistently and accurately assess model performance. The research report will be made available to Kamakura clients on May 31.

Become a bobsguide member to access the following

1. Unrestricted access to bobsguide
2. Send a proposal request
3. Insights delivered daily to your inbox
4. Career development