Stationary bootstrapping for panel cointegration tests under cross-sectional dependence

Abstract

Stationary bootstrapping is applied to panel cointegration tests which are based on the ordinary least-squares estimator and the seemingly unrelated regression (SUR) estimator of the residual unit root. Large sample validity of stationary bootstrapping is established. A finite sample experiment reveals that size performances of the bootstrap tests are much less sensitive to cross-sectional correlation than those of existing tests and a test based on the SUR estimator has substantially better power than existing tests.

Keywords:

• cointegration test
• cross-sectional dependence
• panel data
• seemingly unrelated regression estimator
• stationary bootstrapping

Mathematics Statistics Classification:

• 62G10

Acknowledgements

The author is very grateful for an associate editor and a referee whose comments improved the paper considerably. This study was supported by Basic Research Program (2012-001361) and Science Research Center program (2011-0030811) through the National Research Foundation of Korea (NRF) funded by the Ministry of Education Science and Technology.