Jump-Robust Volatility Estimation using Nearest Neighbor Truncation

We propose two new jump-robust estimators of integrated variance based on high-frequency return observations. These MinRV and MedRV estimators provide an attractive alternative to the prevailing bipower and multipower variation measures. Specifically, the MedRV estimator has better theoretical efficiency properties than the tripower variation measure and displays better finitesample robustness to both jumps and the occurrence of “zero” returns in the sample. Unlike the bipower variation measure the new estimator allows for the development of an asymptotic limit theory in the presence of jumps. Finally, it retains the local nature associated with the low order multipower variation measures. This proves essential for alleviating finite sample biases arising from the pronounced intraday volatility pattern which afflict alternative jump-robust estimators based on longer blocks of returns. An empirical investigation of the Dow Jones 30 stocks and an extensive simulation study corroborate the robustness and efficiency properties of the new estimators. ∗Torben Andersen: Northwestern Univ.; NBER; CREATES; [email protected] Dobrislav Dobrev: Federal Reserve Board of Governors, [email protected] Ernst Schaumburg: Northwestern Univ., [email protected] We are grateful to Federico Bandi, Luca Benzoni, Jean Jacod, Per Mykland, Roel Oomen, Neil Shephard, Viktor Todorov, Lan Zhang, and, in particular Mark Podolskij and Kevin Sheppard for their insights. We also thank participants at the Singapore Management University Conference in Honor of P.C.B. Phillips, July 2008, the CREATES Volatility Symposium, Aarhus, August 2008, and the Chicago/London Conference on Financial Markets, December 2008, for their comments. The views in this paper are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or of any other person associated with the Federal Reserve System. Andersen gratefully acknowledges financial support from the NSF through a grant to the NBER and by CREATES funded by the Danish National Research Foundation.