ON THE ROBUSTNESS OF RANGE‐BASED VOLATILITY ESTIMATORS

Abstract

Abstract We empirically examine Parkinson's range‐based volatility estimate in the federal funds market, which is unique because institutional regulations create a predictable pattern in interday volatility. We find that range‐based volatility estimates and standard deviations produce the expected volatility pattern. We also find that at trading pressure points where microstructure noise should be greatest, range‐based estimates are less than the standard deviations. Thus, we support the argument that range‐based volatility estimates remove the upward bias created by microstructure noise. We find that the Parkinson method is the most efficient range‐based volatility measure among a set of alternates in this market.