Measuring and managing credit portfolio risk

This paper describes a model for calculating the expected losses as well as economic capital required to support an arbitrary portfolio of credit exposures. It does so by explicitly modelling both the marginal and absolute conditional loss distributions for any arbitrary portfolio of credit exposures; these portfolio loss distributions can be made conditional on the current state of the economy given the counterparty's country, industry and rating. The conditioning relationships between the probability of a credit event (e.g. credit rating migrations or defaults) and the current state of the economic cycle are based on empirical regularities observed in historical data. This model differs from other credit portfolio models in several important aspects: First, it models the actual, discrete loss distribution, dependent upon the number and size of credits, as opposed to using a normal distribution or mean-variance approximations; this allows the model to explicitly tabulate a 'large exposure premium' in terms of risk adjusted capital for less diversified portfolios. Second, the losses (or gains) are measured on a marked-to-market basis for credit exposures which cannot be liquidated (e.g, most loans or OTC trading exposure lines) as well as those which can be liquidated prior to the maximum maturity of the exposure; these loss distributions can therefore be tabulated for any time horizon, including one which coincides with an organisations planning and budgeting process Third, the tabulated loss distributions are conditional on the current state of the economy rather than being based on the unconditional or 20 year averages which do not reflect the portfolio's true current risk. Finally, a multi-factor model of systematic default risk, as opposed to a single factor model based on asset volatilities and CAPM or public rating histories, is explicitly estimated based on empirically observed 'regional- and sectoral-betas', allowing the model to mimic the actual default correlations between industries and regions at the transaction as well as portfolio level.(2)