Title: Improved forecasting of realized variance measures
Authors: Hans Manner - University of Cologne (Germany) [presenting]
Jeremias Bekierman - University of Cologne (Germany)
Abstract: The problem of forecasting realized volatility measures is considered. These measures are known to be highly persistent, but also to be noisy estimates of the underlying integrated variance. This fact has been recently exploited to extend the heterogeneous autoregressive (HAR) model by letting the model parameters vary over time depending on estimated measurement errors and show that their model leads to improved forecasts. We propose an alternative specification that allows the autoregressive parameter of the HAR model for log-volatilities to be driven by a latent gaussian autoregressive process that potentially also depends on the estimated measurement error. The model can be estimated straightforwardly using the Kalman filter. Our empirical analysis considers the realized volatilities of 40 stocks from the the S\&P 500 estimated using three different observation frequencies. Our in-sample results show that the time-varying parameters resulting from our specification are more flexible than those of the previous model and consequently provides a better model fit. Furthermore, our model generates superior forecasts and consistently outperforms the competing models in terms of different loss functions and for various subsamples of the forecasting period.