Testing Independence and Conditional Independence in High Dimensions via Coordinatewise Gaussianization

Prof. Ilya ARCHAKOV
Assistant Professor
Department of Economics
York University

Date:    17 October 2025 (Thursday)
Time:   12:30-14:00
Venue: E22-4015
Host:    Prof. Jun YU, UMDF Chair Professor of Finance and Economics, Chair Professor of Finance and Economics

Abstract
We construct a measure of association between random variables that is based on their similarity in both direction and magnitude. Under special conditions the proposed measure becomes an unbiased and consistent estimator of the linear correlation coefficient for which the sampling distribution is available. The latter is intrinsically insensitive to heavy tails and outliers which facilitates robust inference for correlations. The measure can be extended to a higher dimensional setting where it can be interpreted as an index of joint similarity between multiple random variables. We inspect the empirical performance of the proposed measure with financial returns at high and low frequencies. Then, we use it to formulate a new specification of the multivariate GARCH model.

Speaker
Prof. Ilya Archakov is an assistant professor in the Department of Economics at York University (Toronto, Canada).  He obtained Ph.D. in Economics from the European University Institute (EUI) in Florence (Italy). Prior to 2023, Prof.Archakov was a postdoctoral researcher in the Department of Statistics and Operations Research (ISOR) at the University of Vienna. His research interests include financial econometrics, multivariate and high-dimensional statistics, estimation and modeling of correlations and correlation matrices, high-frequency data, and market microstructure analysis.

All are welcome!