研究期間：10208~10307;In this project I develop threshold regression methods for panel data with varying degrees of cross-section dependence. Different common factor models, including weak, semi-weak/strong, strong, and nonstationary factors, are employed for characterizing correlations across individuals. I first demonstrate inference problems for failing to consider the presence of cross-section dependence in panel threshold models. A two-step procedure is suggested in the project. First, I use the common correlated effects (CCE) transformation developed by Pesaran (2006) to remove unobserved common factors, and then minimize the concentrated sum of squared errors of the transformed model to yield threshold and slope estimates. I will try to construct a likelihood ratio test for the presence of threshold effects in cross-sectionally dependent panel data. Asymptotic properties of the estimation and testing procedures will be explored in the project. Monte Carlo simulations will be presented to evaluate the finite sample performance of the proposed methods. To illustrate the usefulness of the proposed estimators and tests, I will conduct an empirical application to reexamine the threshold effects in the relationships among inflation, output growth, and income inequality.