Measuring Portfolio Loss Using Approximation Methods

Abstract

One of the approaches to determining and quantifying the credit risk of a loan portfolio is by obtaining the distribution of losses of the portfolio and determining the risk quantities from such distributions. In this paper, we describe the challenges to using this approach and illustrate a practical solution where simulation methods are used to obtain loss distribution for a two obligor portfolio. This is then extended to ten and hundred obligor portfolios. Existing probability distributions with specified parameters are then used to approximate the loss distributions obtained. Using such parameters of the existing probability distributions, we obtain the risk quantities associated with the loan portfolio including Expected and Unexpected losses. We realized that depending on the confidence interval for which we measure the Unexpected Loss, Stress Losses are needed to account for the total loss of the portfolio