Abstract
This article derives the limiting distributions of the Ordinary Least Squares (OLS) and Least Square Dummy Variable (LSDV) estimators in both spurious and cointegrated panel regressions. The limit theories employed in this article are different from those of Kao (Citation1999) and Phillips and Moon (Citation1999), in which the time dimension of the panel is fixed.
Acknowledgements
I wish to thank Katsuto Tanaka and Taku Yamamoto for many helpful comments and suggestions and the Hitotsubashi COE program for financial support.
Notes
1 Hsiao (Citation1986) and Baltagi (Citation1995) reviewed much of the earlier research on large N but small T panels.
2 Harris and Tzavalis (Citation1999) derived asymptotic unit root tests for first-order autoregressive panel data models assuming that T is fixed.
3 Simulations are based on 5000 replications with N = 200.
4 The same matrix is used to generate the error processes in the spurious panel model.
5 Results for T = 25, 50 for both experiments are available upon request.
6 Tanaka (Citation1996) notes that the matrix C may be called a random walk generating matrix.