A1659
Title: Dependent microstructure noise and integrated volatility estimation from high-frequency data
Authors: Merrick Zhen Li - University of Amsterdam (Netherlands) [presenting]
Roger Laeven - University of Amsterdam (Netherlands)
Michel Vellekoop - University of Amsterdam (Netherlands)
Abstract: We develop econometric tools to study integrated volatility with potentially time-dependent microstructure noise in high-frequency data. In this context, we first develop consistent estimators of the variance and covariances of noise using a variant of realized volatility. Next, we adapt the pre-averaging method and derive a consistent estimator of the integrated volatility, which converges stably with optimal rate $n^{(-1/4)}$ to a mixed Gaussian distribution. In a finite sample analysis, we find that the second moments of noise and the integrated volatility induce biases to each other, and we propose novel two-step estimators to correct the ``interlocked'' bias. Our extensive simulation studies demonstrate the excellent performance of our estimators. Empirically, we find strong evidence of positively autocorrelated noise in the examined stocks and show the considerable accuracy gains achieved by our new estimators.