Forecasting co-volatilities via factor models with asymmetry and long memory in realized covariance

Modeling covariance structures are known to suffer from the curse of dimensionality. In order to avoid the problem for forecasting, the paper proposes a new factor multivariate stochastic volatility (fMSV) model for realized covariance measures that accommodates asymmetry and long memory. Using the basic structure of the fMSV model, the paper extends the dynamic correlation MSV model, the conditional/stochastic Wishart autoregressive models, the matrix-exponential MSV model, and the Cholesky MSV model. Empirical results for 7 financial asset returns for the US stock returns indicate that the new fMSV models outperform existing dynamic conditional correlation models for forecasting future covariances. Regarding the forecasting performance for one-day, five-day and ten-day horizons, the recommended specification among the new fMSV models is the stochastic Wishart autoregressive specification with asymmetric effects for the periods during and after the global financial crisis, while the Cholesky fMSV model with long memory and asymmetry displays the best performance for periods without the financial turbulence.