AComparative study of unit root tests with panel data and a new simple test（15）

G. S.Maddala and Shaowen Wu

(H. D.Thoreau: Walden)

Tests based onBonferroni inequality

1，Forthe correlated errors case we argued that, for both the Fisher test statistic andthe IPS test statistic, the distribution involves nuisance parameters, but bootstrapmethods could be used to make inferences.

2，Thereis, however, one test statistic that is equally applicable in both theuncorrelated and correlated cases. This is the statistic based on the Bonferroniinequality (see Alt, 1982) and discussed in Dufour and Torres (1996). Thesetests have also been discussed earlier in Savin (1984) and Dufour (1990).