Title: A multiplicative dynamic model for realized covariance matrices
Authors: Manuela Braione - Universite catholique de Louvain (Belgium)
Giuseppe Storti - University of Salerno (Italy)
Luc Bauwens - Universite catholique de Louvain (Belgium) [presenting]
Abstract: A class of multiplicative dynamic models for realized covariance matrices is introduced. The multiplicative structure enables consistent three-step estimation of the parameters, starting by covariance targeting of a scale matrix. The dynamics of conditional variances and correlations are inspired by specifications akin to the consistent dynamic conditional correlation model of the multivariate GARCH literature, and estimation is performed by the quasi-maximum likelihood method, assuming a Wishart conditional distribution. Simulations show that in finite samples a three-step estimator has smaller bias and root mean squared error than a one-step estimator when the cross-sectional dimension increases. An empirical application illustrates the flexibility of the model in a low-dimensional setting, and another one illustrates its effectiveness and practical usefulness in high dimensional portfolio allocation strategies.