Changepoint Estimation for Dependent and Non-Stationary Panels

Abstract

The changepoint estimation problem of a common change in panel means for a very general panel data structure is considered. The observations within each panel are allowed to be generally dependent and non-stationary. Simultaneously, the panels are weakly dependent and non-stationary among each other. The follow up period can be extremely short and the changepoint magnitudes may differ across the panels accounting also for a specific situation that some magnitudes are equal to zero (thus, no jump is present in such case). We introduce a novel changepoint estimator without a boundary issue meaning that it can estimate the change close to the extremities of the studied time interval. The consistency of the nuisance-parameter-free estimator is proved regardless of the presence/absence of the change in panel means under relatively simple conditions. Empirical properties of the proposed estimator are investigated through a simulation study.