Testing Stationarity in Panel Data

Economics

Further details

This paper extends the tests of Hadri (2000) for the null of stationarity against the alternative of unit roots in panel data with independent errors over t to the more interesting case of heterogeneous and serially correlated errors over t. The tests are asymptotically similar under the null. The asymptotic distributions of the statistics are derived under the null and are shown to be normally distributed. Their moments are derived analytically. Finite sample sizes and powers are considered in a Monte-Carlo experiment. We compare the effects on the tests of four spectral methods, two parametric and two non-parametric, for estimating the "long-run variance". In general, the non-parametric estimators fair better than the parametric ones. The Monte Carlo results show clearly that the power of the tests increases substantially with N and T.