Credit Risk Modelling: Current practices and applications

by the Basle Committee on Banking Supervision

April 1999

Summary and objectives: Over the last decade, a number of the world's largest banks have developed sophisticated systems in an attempt to model the credit risk arising from important aspects of their business lines. Such models are intended to aid banks in quantifying, aggregating and managing risk across geographical and product lines. The outputs of these models also play increasingly important roles in banks' risk management and performance measurement processes, including performance-based compensation, customer profitability analysis, risk-based pricing and, to a lesser (but growing) degree, active portfolio management and capital structure decisions. The Task Force recognises that credit risk modelling may indeed prove to result in better internal risk management, and may have the potential to be used in the supervisory oversight of banking organisations. However, before a portfolio modelling approach could be used in the formal process of setting regulatory capital requirements for credit risk, regulators would have to be confident not only that models are being used to actively manage risk, but also that they are conceptually sound, empirically validated, and produce capital requirements that are comparable across institutions. At this time, significant hurdles, principally concerning data availability and model validation, still need to be cleared before these objectives can be met, and the Committee sees difficulties in overcoming these hurdles in the timescale envisaged for amending the Capital Accord.

Models have already been incorporated into the determination of capital requirements for market risk. However, credit risk models are not a simple extension of their market risk counterparts for two key reasons:

• Data limitations: Banks and researchers alike report data limitations to be a key impediment to the design and implementation of credit risk models. Most credit instruments are not marked to market, and the predictive nature of a credit risk model does not derive from a statistical projection of future prices based on a comprehensive record of historical prices. The scarcity of the data required to estimate credit risk models also stems from the infrequent nature of default events and the longer-term time horizons used in measuring credit risk. Hence, in specifying model parameters, credit risk models require the use of simplifying assumptions and proxy data. The relative size of the banking book - and the potential repercussions on bank solvency if modelled credit risk estimates are inaccurate - underscore the need for a better understanding of a model's sensitivity to structural assumptions and parameter estimates.
• Model validation: The validation of credit risk models is fundamentally more difficult than the backtesting of market risk models. Where market risk models typically employ a horizon of a few days, credit risk models generally rely on a time frame of one year or more. The longer holding period, coupled with the higher confidence intervals used in credit risk models, presents problems to model-builders in assessing the accuracy of their models. By the same token, a quantitative validation standard similar to that in the Market Risk Amendment would require an impractical number of years of data, spanning multiple credit cycles.

Download paper (290K PDF) 65 pages