Title: The role of jump activity and signed jumps in forecasting realized volatility
Authors: Rodrigo Hizmeri - Lancaster University (United Kingdom) [presenting]
Marwan Izzeldin - Lancaster University Management School (United Kingdom)
Anthony Murphy - Federal Reserve Bank of Dallas (USA)
Mike Tsionas - Lancaster University (United Kingdom)
Abstract: The gains from various jumps classifications are examined: signed, finite and infinite jumps in volatility forecasting. Using a Heterogeneous Autoregressive (HAR) framework, we illustrate gains from considering various jumps specifications. We consider different sampling frequencies and microstructure noise and document the impact on the model forecasting performance. We find that on average, jumps improve volatility forecasts. Negative jumps are more important for short horizons whilst positive jumps achieve greater gains at longer horizons. Controlling for market microstructure noise at higher frequencies, results in substantial out-of-sample improvements for both short and longer horizons. Findings from the use of a model averaging approach based on the selected models by the model confidence set, outperforms the benchmark model under the various scenarios considered.