Title: Forecasting multivariate realized volatility using time varying coefficient models
Authors: Laleh Tafakori - University of Melbourne (Australia) [presenting]
Hans Manner - University of Graz (Austria)
Bing Liu - University of Melbourne (Australia)
Abstract: Modelling- and forecasting- multivariate realised volatility plays an indispensable role for option pricing, portfolio allocation and risk management. The existing models for realised volatility may perform well in-sample but in general their out-of-sample forecasts are often biased. We aim to build a model for realised volatility with improved forecasting performance by accounting for the fact that the multivariate realised covariances are the only estimates of the true variance and by introducing time varying parameters. With its more accurate forecasting, our model holds the promise to empirically more accurate pricing models and improved financial decision-making.