Quasi‐maximum likelihood and the kernel block bootstrap for nonlinear dynamic models

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