Abstract:In this paper, we study distance covariance, Hilbert–Schmidt covariance (aka Hilbert–Schmidt independence criterion [In Advances in Neural Information Processing Systems (2008) 585–592]) and related independence tests under the high dimensional scenario. We show that the sample distance/Hilbert–Schmidt covariance between two random vectors can be approximated by the sum of squared componentwise sample cross-covariances up to an asymptotically constant factor, which indicates that the standard distance/Hilbert–Schmidt covariance based test can only capture linear dependence in high dimension. Under the assumption that the components within each high dimensional vector are weakly dependent, the distance correlation based $t$ test developed by Székely and Rizzo (J. Multivariate Anal. 117 (2013) 193–213) for independence is shown to have trivial limiting power when the two random vectors are nonlinearly dependent but component-wisely uncorrelated. This new and surprising phenomenon, which seems to be discovered and carefully studied for the first time, is further confirmed in our simulation study. As a remedy, we propose tests based on an aggregation of marginal sample distance/Hilbert–Schmidt covariances and show their superior power behavior against their joint counterparts in simulations. We further extend the distance correlation based $t$ test to those based on Hilbert–Schmidt covariance and marginal distance/Hilbert–Schmidt covariance. A novel unified approach is developed to analyze the studentized sample distance/Hilbert–Schmidt covariance as well as the studentized sample marginal distance covariance under both null and alternative hypothesis. Our theoretical and simulation results shed light on the limitation of distance/Hilbert–Schmidt covariance when used jointly in the high dimensional setting and suggest the aggregation of marginal distance/Hilbert–Schmidt covariance as a useful alternative.

A Test of General Asymmetric Dependence

New general dependence measures: construction, estimation and application to high-frequency stock returns

Generalized spectral tests for serial dependence

Asymmetries in Stock Returns: Statistical Tests and Economic Evaluation

Asymptotic Distribution Theory for Nonparametric Entropy Measures of Serial Dependence

A low variance consistent test of relative dependency

A New Class of Nonsymmetric Multivariate Dependence Measures

Independence test via mutual information in the presence of measurement errors

Distance-based and RKHS-based dependence metrics in high dimension

A nonparametric Bayesian test of dependence

Testing for Pairwise Serial Independence Via the Empirical Distribution Function

Randomization Tests for Equality in Dependence Structure

Tail Dependence Measure for Examining Financial Extreme Co-Movements

Measures of conditional dependence for nonlinearity, asymmetry and beyond

Nonparametric Testing for Asymmetric Information

A NEW PERSPECTIVE ON ROBUST M-ESTIMATION: FINITE SAMPLE THEORY AND APPLICATIONS TO DEPENDENCE-ADJUSTED MULTIPLE TESTING

Generalized Independence Test for Modern Data

On high-dimensional tests for mutual independence based on Pearson’s correlation coefficient

Rearranged dependence measures

RANK-BASED INDICES FOR TESTING INDEPENDENCE BETWEEN TWO HIGH-DIMENSIONAL VECTORS

Measuring Dependence in a Set of Asset Returns