A1024
Title: Factor regression multivariate realized stochastic volatility model and its application to high dimensional return data
Authors: Tsunehiro Ishihara - Takasaki City University of Economics (Japan) [presenting]
Abstract: For the high dimensional multivariate asset return series modeling, multivariate stochastic volatility models are popular and exhibit high forecasting performance. A multivariate stochastic volatility model with observed factors and some extensions of the model are proposed. Of particular interest is the computation time. In this setting, the computational time increases in proportion to the dimension of returns. Additionally, intraday information is also introduced via realized covariance to the model. Estimation and forecasting can be performed in parallel in each univariate model. An empirical illustrative example is presented using 33 stock return series and market-, size-, and value-based factors, as well as their realized covariances. Some forecasting performance results are also shown.
