The Economic Value of Volatility Timing with Realized Jumps
49 Pages Posted: 12 Mar 2014 Last revised: 21 Aug 2015
Date Written: January 29, 2015
This paper comprehensively investigates the role of realized jumps detected from high frequency data in predicting future volatility from both statistical and economic perspectives. Using seven major jump tests, we show that separating jumps from diffusion improves volatility forecasting both in-sample and out-of-sample. Moreover, we show that these statistical improvements can be translated into economic value. We find a risk-averse investor can significantly improve her portfolio performance by incorporating realized jumps into a volatility timing based portfolio strategy. Our results hold true across the majority of jump tests, and are robust to controlling for microstructure effects and transaction costs.
Keywords: high frequency data, jumps, asset allocation, volatility forecasting, realized volatility
JEL Classification: C58,C53,G11
Suggested Citation: Suggested Citation