A multivariate semi-parametric portfolio risk optimization and forecasting framework

We develop a novel multivariate semi-parametric modelling approach to portfolio Value-at-Risk (VaR) and Expected Shortfall (ES) forecasting. Differently from existing univariate semi-parametric approaches, the proposed framework involves explicit modelling of the dependence structure among portfolio asset returns through a dynamic conditional correlation (DCC) parameterization. Model parameters are estimated through a two-step procedure based on the minimization of strictly consistent scoring functions derived from the Asymmetric Laplace (AL) distribution. The proposed model is consequently denominated as DCC-AL. The performance of the DCC-AL in risk forecasting and portfolio optimization is assessed by means of a forecasting study on the components of the Dow Jones index for an out-of-sample period from December 2016 to September 2021. The empirical results support the proposed model class, comparing to a variety of existing approaches, in terms of risk forecasting and portfolio optimization.