Title: A vector index-augmented heterogeneous autoregressive model for forecasting realized covariance matrices
Authors: Alain Hecq - Maastricht University (Netherlands) [presenting]
Abstract: A multivariate model for the elements of realized covariance matrices is proposed where each equation follows a heterogeneous autoregressive model that is augmented with a common index structure. Our modelling can accommodate both idiosyncratic and common dynamics of realized covariances. We offer a switching algorithm to maximise the Gaussian likelihood of our model. Since the maximum likelihood estimator may perform poorly when the dimension of the covariance matrix becomes large, we also propose some algorithms for regularized estimation. The proposed approach is evaluated through both simulations and empirical examples.