Social and Behavioral Sciences Commons^™

Testing For Breaks In Cointegrated Panels, Chihwa Kao, Lorenzo Trapani, Giovanni Urga

Center for Policy Research

We investigate the issue of testing for structural breaks in large cointegrated panels with common and idiosyncratic regressors. We prove a panel Functional Central Limit Theorem. We show that the estimated coefficients of the common regressors have a mixed normal distribution, whilst the estimated coefficients of the idiosyncratic regressors have a normal distribution. We consider strong dependence across the idiosyncratic regressors by allowing for the presence of (stationary and nonstationary) common factors. We show that tests based on transformations of Wald-type statistics have power versus alternatives of order