Estimation of intra-sector asset correlations

Christian Meyer

Abstract

ABSTRACT

Asset correlations are an intuitive and therefore popular way to incorporate event dependence into event risk, eg, default risk and modeling. Due to the long time horizon and large confidence level, classical backtesting is usually not an option for event risk models. Validation of the parameters used is therefore essential. If these are estimated from historical data, the first thing to do is examine the statistical properties of the estimators applied. In this paper we study in detail the estimation of intra-sector asset correlations by separation of the cross-sectional dimension and the time dimension.