An Econometric Defence of Pure-Jump Price Dynamics

Abstract

Pure-jump stochastic processes are shown to be capable of explaining many empirical features of high-frequency asset prices. A simple pure-jump process can match the empirical bipower, realized variance and jump detection statistics of Andersen, Bollerslev and Dobrev (2007) at the two-minute frequency. A multi-frequency analysis of Spyder returns shows the theoretical predictions can also be aligned reasonably accurately with the empirical evidence across more than one sampling frequency.

Keywords: Jumps, High-Frequency Prices, Bipower Variation, Realized Variance

JEL Classification: C10, G10