Specification Tests for Dynamic Models Using Multipliers

Abstract

Extending the multiplier central limit theorem and resampling bootstrap to statistics and empirical processes of pseudo-observations, it is shown how to build asymptotically independent copies of statistics and empirical processes to perform statistical tests. Application to parametric and semi-parametric specification tests for the innovations of dynamic models are given. Finally, the performance of some multipliers-based tests is studied for testing the null hypothesis of Gaussian innovations for GARCH models.

Keywords: Multipliers, bootstrap, p-value, empirical process, pseudo-observations, time series, copula

JEL Classification: C12, C14, C15