Revisiting Credit Scoring Models in a Basel 2 Environment


By Edward I. Altman

Abstract

This paper discusses two of the primary motivating influences on the recent development/revisions of credit scoring models, i.e., the important implications of Basel 2’s proposed capital requirements on credit assets and the enormous amounts and rates of defaults and bankruptcies in the US in 2001-2002. Two of the more prominent credit scoring techniques, Z-Score and KMV’s EDF models, are reviewed. Finally, both models are assessed with respect to default probabilities in general and in particular to the infamous Enron debacle. In order to be effective, these and other credit risk models should be utilized by firms with a sincere credit risk culture. ##

Introduction

Around the turn of the new millennium, credit scoring models have been remotivated and given unprecedented significance by the stunning pronouncements of the new Basel Accord on credit risk capital adequacy - - the so-called Basel 2 (see Basel [1999] and [2001]). Banks, in particular, and most financial institutions worldwide, have either recently developed or modified existing internal credit risk systems or are currently developing methods to conform with best practice systems and processes for assessing the probability of default (PD) and, possibly, loss-given-default (LGD) on credit assets of all types. Coincidentally, defaults and bankruptcies reached unprecedented levels in the United States in 2001 and have continued in 2002. Indeed, companies that filed for bankruptcy/reorganization under Chapter 11 with greater than $100 million liabilities reached at least $240 billion in liabilities in 2001 (even with Enron’s understatement at the time of filing) and there were 39 firms that filed for protection under the US bankruptcy code with liabilities greater than $1 billion (see Panel A).

The pace of these large bankruptcies has continued in 2002 with another 23 firms of such great size filing in the first six months. In the public bond arena, over $63 billion of U.S. domestic public debt defaulted in 2001 and the default rate on US high yield bonds was almost a record 9.8% (see Altman and Arman [2002]). And, in the first six months of 2002, the default rate is 6.97% and with WorldCom (July 2002) included, has already broken the record default rate for a single calendar year. This paper primarily discusses a model developed by the author over 30 years ago, the so-called Z-Score model, and its relevance to these recent developments. In doing so, we will provide some updated material on the Z-Score model’s tests and applications over time as well as some modifications for greater applicability. We also discuss an alternative widely used credit risk model, known as the KMV approach, and compare both KMV and Z-Score in the now infamous Enron (2001) bankruptcy debacle. The paper is not meant to be a comparison of all of the well known and readily available credit scoring models, such as Moody’s RiskCalc® or the ZETA® scoring model. Finally, we summarize a recent report (Altman, Brady, Resti and Sironi, [2002]) on the association between aggregate PD and recovery rates on defaulted credit assets.