HIGH-FREQUENCY ESTIMATION OF ITÔ SEMIMARTINGALE BASELINE FOR HAWKES PROCESSES
Y. POTIRON*, O. SCAILLET**, V. VOLKOV*** and Y. YU****
* Keio University ** Université de Genève and Swiss Finance Institute *** University of Tasmania **** Northwestern University
Abstract
We introduce Hawkes self-exciting processes with a baseline driven by an Itô semimartingale with possible jumps. We study three measures of the intensity of these Hawkes processes: the integrated intensity, the integrated baseline, and the integrated volatility of the baseline. The statistics are based on empirical averages and preaveraging of local Poisson estimates constructed from high-frequency quote times. We also propose feasible statistics induced by central limit theory with in-fill asymptotic theory, and develop a test for the absence of a Hawkes component and a test for a constant baseline. An empirical application to high-frequency data of the E-mini S&P500 futures contracts finds rejection of both the null hypothesis of no Hawkes excitation and that of constant baseline contributing to the formal identification of patterns in high-frequency trading activity.
Keywords: Hawkes tests, in-fill asymptotics, high-frequency data, Itô semimartingale, self-exciting process, time-varying baseline.
JEL: C14, C22, C41, C58, G00, G13.