Robust Price Discovery to Heavy-Tailed Market Shocks

Abstract

We show that conclusions drawn from widely used measures of price discovery are highly sensitive to the presence of price outliers in the calculations. We demonstrate using simulation studies however that the long-run information share (LFS) measure of price discovery location proposed by Kim and Linn (2022), coupled with Bayesian estimation of a Vector Error Correction Model (VECM) allowing for outliers, provides the most robust and reliable metric for evaluating price discovery in the presence of outliers. A separate empirical analysis of the spot and futures prices of non-ferrous metals shows the pervasive presence of price outliers. Implementation of our proposed estimation of a VECM using Bayesian methods allowing for outliers and the subsequent calculation of LFS, provides strong evidence that both spot and futures markets for non-ferrous metals contribute significantly to the price discovery process when daily price data are employed.

Keywords: Price discovery, Cointegration, Outliers, Robust estimation, Heavy-tailed distributions