We uncover the structure of volatility in financial markets using ultra high-frequency data. Our recent work (Christensen et al. 2014) suggests that volatility over short time intervals may differ from what a vast amount of prior research has indicated and that jumps in asset prices account for only about one percent of the total return variation. A reduced role for jumps has many important implications, because jumps have a profoundly distinct impact on option pricing, risk management, and asset allocation. Motivated by this, the objective of this research project is to build a foundation necessary to analyse volatility using data at the highest sampling frequency. We propose to extend the extant literature by designing new tools for drawing inferences from underlying continuous and discontinuous volatility components using bootstrap resampling and subsampling of high-frequency data. Moreover, to link our findings to the underlying causes of volatility, such as macroeconomic fundamentals and corporate earnings announcements, we will pursue the ideas of Andersen et al. (2003) by modelling the response of asset prices to news releases so that we can analyse how surprises in news announcements feed into the price discovery process, either in the form of jumps in prices or bursts in volatility.
The main result is an econometric method suited to solve many of the problems that are currently preventing largescale analysis of financial volatility at the highest sampling frequency, e.g., by allowing one to perform a hypothesis test of the presence of jump risk in financial markets. The research will improve our capacity to (i) measure the level of risk that players in the financial markets are currently exposed to and to (ii) predict how that risk is expected to evolve in the near-future, e.g., to comply with regulatory standards or to avoid large losses by timely and prudent risk management. The literature on high-frequency financial volatility is a flourishing area, and we expect that top-ranked journals in finance will find such research interesting to publish.
Early Stage Resercher working on the project: Giorgio Mirone