Robust information share measures with an application on the international crude oil markets

This paper proposes a robust estimation approach of the popular information share using a robust vector error correction model. Via simulation studies, we show that the proposed measures lead to more accurate estimates than the existing measures in the presence of outliers. The proposed measures are then investigated using high-frequency crude oil prices data of the West Texas Intermediate and Brent over November 2019–October 2020. The results suggest that the abnormally large price movement in April 2020 may cause biased estimates of the ordinary measures, whereas the robust measures produce rather consistent results.