Estimation of Asset Volatility and Correlation Over Market Microstructure Noise in High-Frequency Data

Abstract

Accurate measurement of asset return volatility and correlation is an important problem in financial econometrics. The presence of market microstructure noise in high-frequency data complicates such estimations. This study extends a prior application of a model-based volatility estimator with autocorrelated market microstructure noise to estimation of correlation. The model is applied to a high-frequency dataset including a stock and an index, and the results are compared to some existing models. This study supports previous findings that including an autocorrelation factor produces an estimator potentially less vulnerable to market microstructure noise, and finds that the same is true about the extended correlation estimator that is introduced here.