Systematic and Idiosyncratic Risk in Middle-Market Default Prediction: A Study of the Performance of the RiskCalc™ and PFM™ Models

by Roger M. Stein of Moody's|KMV,
Ahmet E. Kocagil of Moody's|KMV,
Jeff Bohn of Moody's|KMV, and
Jalal Akhavein of Moody's|KMV

February 2003

Abstract: Clients of MKMV (legacy Moody 's Risk Management Services and legacy KMV) have often requested information on the relative performance of the legacy KMV Private Firm Model™ (PFM) and the legacy MRMS RiskCalc™ models. Embedded within such requests is a desire to understand better the relative importance of the various drivers that each model uses to predict default. These include equity market information on an industry and regional basis, in the case of PFM, and localized middle market financial statement data, which is used to parameterize and calibrate localized models, in the case of RiskCalc. Having successfully merged these two legacy organizations, we are now able to test these models rigorously and begin to answer some of these questions.

In this Special Comment we present the results of our analysis of the relative performance of RiskCalc and PFM. We conducted a series of performance tests using native middle-market data from a number of geographies in Moody 's Credit Research Database (CRD) which contains valid financial statement and loan performance information on over 900,000 unique middle market firms worldwide, including over 60,000 defaulted firms.

Download paper (3,583K PDF) 40 pages