Two Tests for Jumps in High Frequency Financial Time Series : Simulation and Empirical Application

IR Web Site

Search this article

Abstract

application/pdf

We often observe significant discontinuous variations, so-called jumps, in financial time series but empirically it is not easy to distinguish between a large variation and a discontinuous jump. The two tests have been proposed for detecting jumps in continuous diffusion process by using discrete data of high frequency financial time series. One is proposed by Barndorff-Nielsen and Shephard (2006) and the other was proposed by Lee and Mykland (2008). The former test is aimed to see if a time series is a jump diff usion process. In other word it can detect whether the process contains jumps or not globally. On the other hand, the latter test can detect the local jump arrival time and the size of realized jump. In this article we briefly introduce the two tests, show the empirical applications results, and examine the performance and applicability of the two tests. Furthermore we examine the performance of LM test by Monte Carlo experiment and real data analysis in particular.

Journal

Details 詳細情報について

Report a problem

Back to top