Empirical Decision in Panel Unit Root Testing: Comparing Simulated and Asymptotic Critical Values

Abstract

A comparison of asymptotic and simulated critical values for panel unit root tests is made to address the problem of size distortions under asymptotic critical values and how to achieve the stabilized size under the simulated critical values. Our results suggest that the simulated critical values are more appropriate to avoid size distortion, especially in finite panel dimensions, and to get the stabilized size around nominal size of 5%. This finding holds whether panel dimensions are small, medium or large.     

Keywords: Asymptotic Critical Value, Oversized, Undersized, Simulated Critical Value, Size Distortion.