Estimation of common breaks in linear panel data models via screening and ranking algorithm

In this paper, we consider the estimation of common breaks for linear panel data models by means of screening and ranking algorithm. For static and dynamic panel data models, we estimate the regression coefficients using covariance estimation and generalized method of moments, respectively, and apply a screening and ranking algorithm on this basis. The possible break points are first screened by constructing local statistics based on the coefficient estimators, then further screened by the thresholding rule, and finally the final break points are screened by the information criterion. Monte Carlo simulations demonstrate that the proposed methods work well in finite samples. We apply the screening and ranking algorithm to study the influence of rural residents' consumption demand on China's economic growth using a panel of 31 provinces from 2005 to 2023 and find a break point in the model.