The effects of estimation error on measures of portfolio credit risk

This paper uses Monte Carlo simulations to assess the impact of noisy input parameters on the accuracy of estimated portfolio credit risk. Assumptions about input quality are derived from the distribution of historical sample statistics commonly used in default risk modelling. The resulting estimation error in the distribution of portfolio losses is considerable. Losses that are judged to occur with a probability of 0.3% may actually occur with a probability of 1%. The paper also shows that estimation error leads to biases in value at risk estimates and significance levels of backtests. The biases can be corrected by analysing predictive distributions which average over the unknown parameter values. (C) 2003 Elsevier Science B.V. All rights reserved.