Parente, P. M. D. C. and Smith, R. J.

Quasi-Maximum Likelihood and the Kernel Block Bootstrap for Nonlinear Dynamic Models

Journal of Time Series Analysis

Vol. 42(4) pp. 377-405 (2021)

Abstract: This article applies a novel bootstrap method, the kernel block bootstrap (KBB), to quasi-maximum likelihood (QML) estimation of dynamic models with stationary strong mixing data. The method first kernel weights the components comprising the quasi-log likelihood function in an appropriate way and then samples the resultant transformed components using the standard ‘m out of n’ bootstrap. We investigate the first-order asymptotic properties of the KBB method for QML demonstrating, in particular, its consistency and the first-order asymptotic validity of the bootstrap approximation to the distribution of the QML estimator. A set of simulation experiments for the mean regression model illustrates the efficacy of the kernel block bootstrap for QML estimation.

Keywords: Bootstrap, heteroskedastic and autocorrelation consistent inference, quasi-maximum likelihood estimation

JEL Codes: C14, C15, C22

Author links: Richard J Smith  

Publisher's Link: https://doi.org/10.1111/jtsa.12573