Change point estimation in panel data with time‐varying individual effects
提出在短面板中通过普通最小二乘法一致估计多个变点的方法，无需事先去除个体效应，小样本性质优于一阶差分法，并给出检验变点来源的两种检验，应用于环境库兹涅茨曲线和美国房价预期。
Summary Existing panel data methods remove unobserved individual effects before change point estimation through data transformations such as first‐differencing. In this paper, we show that multiple change points can be consistently estimated in short panels via ordinary least squares. Since no data variation is removed before change point estimation, our method has better small‐sample properties compared to first‐differencing methods. We also propose two tests that identify whether the change points found by our method originate in the slope parameters or in the covariance of the regressors with individual effects. We illustrate our method via modeling the environmental Kuznets curve and the US house price expectations after the financial crisis.