Robust change-point detection in panel data

TL;DR

This paper introduces a nonparametric, robust change-point detection test for panel data that remains effective under heavy-tailed distributions and asymptotic conditions.

Contribution

It develops a new nonparametric test for detecting changes in location in panel data, with proven asymptotic properties under dependence and large N, T.

Findings

Test is robust to heavy-tailed distributions

Asymptotic distribution derived for large N, T

Simulations confirm effectiveness in practical scenarios

Abstract

In panel data we observe a usually high number N of individuals over a time period T. Even if T is large one often assumes stability of the model over time. We propose a nonparametric and robust test for a change in location and derive its asymptotic distribution under short range dependence and for N, T tending to infinity. Some simulations show its usefulness under heavy tailed distributions.