Evaluation of Correlation Forecasting Models for Risk Management

摘要

Volatility and correlation forecasts are of paramount importance in modern risk management systems. The forecasting performance of volatility models in the context of risk management applications has extensively been investigated in the open literature. However, in spite of the plethora of correlation forecasting models, their impact on the VaR accuracy calculation has not yet been explicitly examined. In this paper, traditional and modern correlation forecasting techniques are compared using standard statistical loss functions, as well as, VaR based economic loss functions. Historical data on portfolios consisting of stocks, bonds and currencies are used for the purposes of this study.