Consistent Realized Covariance for Asynchronous Observations Contaminated by Market Microstructure Noise

Abstract

This paper proposes a consistent estimator for the realized covariance of high frequency and asynchronous assets’ returns that are contaminated by microstructure noise. The main contribution is the introduction of the pseudoaggregation which transforms the observations into series with the same number of data points without incurring in any loss of information. This in turn makes it possible to construct an unbiased and consistent covariance estimator by merging techniques from the literature relating to the asynchrony and the market microstructure contamination. Monte Carlo simulations confirm the theoretical results and highlight the outstanding performance of the proposed estimator.

Keywords: Realized Covariance, Asynchronous Observations, Market Microstructure, Sub-sampling, Pseudo-Aggregation

JEL Classification: C32, G0, G1