wp9-11.pdf (3.28 MB)
Testing for co-jumps in high-frequency financial data: an approach based on first-high-low-last prices
journal contributionposted on 2022-11-04, 03:49 authored by Yin Liao, Heather M. Anderson
This paper proposes a new test for simultaneous intraday jumps in a panel of high frequency financial data. We utilize intraday first-high-low-last values of asset prices to construct estimates for the cross-variation of returns in a large panel of high frequency financial data, and then employ these estimates to provide a first-high-low-last price based test statistic to detect common large discrete movements (co-jumps). We study the finite sample behavior of our first-high-low-last price based test using Monte Carlo simulation, and find that it is more powerful than the Bollerslev et al (2008) return-based co-jump test. When applied to a panel of high frequency data from the Chinese mainland stock market, our first-high-low-last price based test identifies more common jumps than the return-based test in this emerging market.
Classification-JELC12, C22, C32, G12, G14
Working Paper Series Number9/11