ABSTRACT. This paper discusses a new dynamic model for dealing with large dimensional realized covariance matrices. The proposed modelling strategy is based on a two-step approach: first an univariate model is fitted to each series of realized variances; second, a multivariate model is fitted to the series of realized correlation matrices. In both steps, in order to match the properties of the data to be analyzed, different specifications of the volatility and correlation models can be considered, resulting in a very flexible model structure. The model parameters can be estimated by the Quasi Maximum Likelihood method. Moreover, the model is applicable to very large matrices since estimation can be also done by the Composite Likelihood method. Finally, an application to a set of fifty S&P500 assets is presented.
Modelling vast dimensional realized covariance matrices
STORTI, Giuseppe
2011
Abstract
ABSTRACT. This paper discusses a new dynamic model for dealing with large dimensional realized covariance matrices. The proposed modelling strategy is based on a two-step approach: first an univariate model is fitted to each series of realized variances; second, a multivariate model is fitted to the series of realized correlation matrices. In both steps, in order to match the properties of the data to be analyzed, different specifications of the volatility and correlation models can be considered, resulting in a very flexible model structure. The model parameters can be estimated by the Quasi Maximum Likelihood method. Moreover, the model is applicable to very large matrices since estimation can be also done by the Composite Likelihood method. Finally, an application to a set of fifty S&P500 assets is presented.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
