A1251
Title: Variable ordering in a Cholesky- multivariate stochastic volatility model
Authors: Martina Zaharieva - CUNEF Universidad (Spain) [presenting]
Ping Wu - University of Strathclyde (United Kingdom)
Abstract: Modeling the time-varying structure of the covariance matrix of financial time-series has been widely studied both theoretically and empirically in the finance literature. Recently, the Cholesky-decomposition-based approach to multivariate stochastic volatility has been established as a flexible and interpretable alternative. Despite that, the issue of order dependence in this type of model has not been the primary focus. The aim is to propose a prior for variable ordering in a Cholesky-type multivariate stochastic volatility model, where a natural ordering of the time series is not available. The approach to forecasting a large cross-section of stock returns is applied.
