Testing for the Presence of Structural Change and Spatial Heterogeneity

TL;DR

This paper introduces a nonparametric bootstrap-based method to detect structural change and spatial heterogeneity in spatial-temporal models, providing robust confidence intervals and improved detection capabilities.

Contribution

It develops a novel nonparametric testing procedure using bootstrap and forward search algorithms to identify structural change and spatial heterogeneity.

Findings

The proposed test effectively detects structural change in simulations.

The method accurately identifies spatial heterogeneity.

Bootstrap confidence intervals improve robustness of the tests.

Abstract

In a spatial-temporal model, structural change and/or spatial heterogeneity can easily affect estimation of parameters. Following the spatial-temporal model in [1], we develop a nonparametric procedure for test-ing the presence of structural change and spatial heterogeneity using bootstrap techniques and the forward search algorithm. The time series bootstrap can filter the effect of temporary structural change in the con-struction of a confidence interval for the temporal parameter. The forward search will also facilitate the construction of a robust confidence interval for the spatial parameter. These confidence intervals are then used in deciding on the null hypothesis that there is no structural change/spatial heterogeneity. Simulation studies illustrate the ability of the proposed test procedure in detecting presence of structural change and spatial heterogeneity under certain conditions.