Weighted Average Correlation Matrices Method for Correlation Stress Testing and Sensitivity Analysis

Abstract

Until the onset of recent financial crises, the relative utility of correlation stress testing and sensitivity analysis were commonly underestimated. Changes in correlation can seriously influence positions of both market risk and credit risk. This article aims to present the use of a weighted average correlation matrix as a new and simple methodology for correlation stress testing and sensitivity analysis. This method allows a sensitivity level to be set to a given percentage for both upward and downward adjustments of a correlation matrix, which can be applied to the analysis of both market and credit portfolios. To the best of my knowledge, for the first time a systematic methodology that allows for the downward adjustment of an overall correlation matrix has been proposed. Its simplicity and utility give this methodology great potential to become a common practice for correlation stress testing and sensitivity analysis in the future.