Kamakura Corporation announced today that it has added implied debt ratings on 16,000 public companies in 29 countries to its Kamakura Risk Information Services default probability and default correlation service.
The addition of ratings brings the number of default methodologies on the KRIS web site to five. In addition to ratings, the KRIS default probability service includes the fourth generation and third generation Jarrow-Chava "reduced form" default models, a hybrid default model which adds the Merton default probability as an input, and the Kamakura Merton default technology. Kamakura's web site now includes implied ratings for 10 times more companies among these 16,000 firms than the rating agencies themselves cover with a formal debt rating.
"Rating agencies incur a very significant "change cost" when it comes to ratings upgrades and downgrades, which makes ratings changes very infrequent," said Warren Sherman, Kamakura President and Chief Operating Officer. "Citigroup, for example, has had the same rating since 1997 in one of the most turbulent economic periods in U.S. history. KRIS users now have access to a daily-updated rating on 16,000 companies which takes into account the behavior of one of the two major rating agencies over the last 15 years. The analysis of ratings behavior, which is documented in an appendix to the KRIS 4.1 Technical Guide, shows that the KRIS reduced form default probability term structure, company size and macro economic factors are the key predictive variables in explaining rating agency behavior."
The expanded KRIS service now includes the full probability distribution of potential ratings for all public companies on KRIS, as well as a "most likely" rating and the probability of being "not rated". The uncertainty in rating agency behavior is attributable to the long periods of stability in ratings during times of great changes in default probabilities, financial ratios, stock price inputs, and macro economic factors. The Kamakura version 4.1 default probabilities have a 99.00% accuracy in predicting default of rated public companies. This ROC accuracy ratio is well above that of public debt ratings at any time horizon. This accuracy differential also contributes to the uncertainty in rating agency behavior. The probability of being "not rated" or "rated" is based on 2.2 million monthly observations in the KRIS default data base from January 1990 to the present. The predicted rating, conditional on being rated, is based on 263,000 monthly observations of public debt ratings over the same time period.