AComparative study of unit root tests with panel data and a new simple test（13）

G. S.Maddala and Shaowen Wu

(H. D.Thoreau: Walden)

Correlated errors and the bootstrap alternative

  1)      Recallthat the properties of all three tests are based on the assumption that theerror terms are not cross-correlated. When this assumption is violated thederived distributions for these test statistics are no longer valid.

  2)      Theyall suffer from nuisance parameter problems. More specifically, if there is cross-correlationin the data, as noted earlier, the distributions of the test statistics are notthe same as before and are not known.