Liquidity-Based Estimation of Spot Volatility Under Microstructure Noise

Abstract

Recent literature on realized volatility suggests that the observed price process of an asset may be decomposed into two parts: the unobservable, efficient price process and microstructure noise. In this article we present a methodology to sequentially estimate spot volatility from noisy data by separating these components. We use different liquidity-based measures, traded volume and quoted spread, for the noise variance of single price observations. Nonlinear Kalman filters provide us with sequential estimates of the unobservable price process and its parameters. Our approach is implemented in a continuous-discrete state space model to cope with irregular trading frequencies.

Keywords: Volatility Estimation, Realized Volatility, Bayesian Parameter Estimation, Non-linear Kalman Filter, High Frequency Data, Market Microstructure

JEL Classification: C22, C53