Optimal Rank-based Procedures for Testing Randomness against Panel Bilinear Dependence

Abstract

Based on the local asymptotic normality (LAN) property, we are proposing parametric and nonparametric locally and asymptotically optimal (in the Le Cam-Hájek sense) procedures for the problem of testing white noise against the general panel super-diagonal bilinear dependence. The proposed rank-based tests are asymptotically most stringent and valid under a wide class of densities. The asymptotic relative efficiencies (AREs), with respect to the corresponding pseudo-Gaussian procedure, are computed under various densities. Simulation studies are conducted to illustrate the finite-sample performance of the proposed tests.