Estimation and Empirical Performance of Non-Scalar Dynamic Conditional Correlation Models
Forthcoming in Computational Statistics and Data Analysis
68 Pages Posted: 11 Mar 2014 Last revised: 21 Feb 2015
Date Written: March 11, 2014
A method capable of estimating richly parametrized versions of the dynamic conditional correlation (DCC) model that go beyond the standard scalar case is presented. The algorithm is based on the maximization of a Gaussian quasi-likelihood using a Bregman-proximal trust-region method that handles the various non-linear stationarity and positivity constraints that arise in this context. The general matrix Hadamard DCC model with full rank, rank equal to two and, additionally, two different rank one matrix specifications are considered. In the last mentioned case, the elements of the vectors that determine the rank one parameter matrices are either arbitrary or parsimoniously defined using the Almon lag function. Actual stock returns data in dimensions up to thirty are used in order to carry out performance comparisons according to several in- and out-of-sample criteria. Empirical results show that the use of richly parametrized models adds value with respect to the conventional scalar case.
Keywords: multivariate volatility modeling, dynamic conditional correlations (DCC), non-scalar DCC models, constrained optimization, Bregman divergences, Bregman-proximal trust-region method.
JEL Classification: C13, C32, G17.
Suggested Citation: Suggested Citation