The SNP-DCC model: a new methodology for risk management and forecasting

Abstract

This paper generalizes the Dynamic Conditional Correlation (DCC) model of En- gle (2002) to incorporate a flexible non-Gaussian distribution based on Gram-Charlier expansions. The resulting semi-nonparametric (SNP)-DCC model admits a separate estimation of, in a first stage, the individual conditional variances under a Gaussian distribution and, in the second stage, the conditional correlations and the rest of the density parameters, thus overcoming the known "dimensionality curse" of the mul- tivariate volatility models. Furthermore the proposed SNP-DCC model solves the negativity problem inherent to truncated SNP densities providing a parametric struc- ture that may accurately approximate a target heavy-tailed distribution. We test the performance of a SNP-DCC model with respect to the (Gaussian)-DCC through an empirical application of density forecasting for portfolio asset returns data. Our results show that the proposed multivariate model provides a better in-sample fit and forecast of the portfolio returns distribution, being thus useful for financial risk forecasting and evaluation.