Testing for a Unit Root in a Time Series with a Changing Mean: Corrections and Extensions

This note provides a correction to the treatment of the asymptotic distribution of tests for a unit root for the additive outlier model presented in Perron (1990). It is shown that the tests, as stated for that case, have asymptotic distributions that depend on the correlation structure of the data even if the appropriate order of the autoregression is selected. The authors present a simple modification that yields statistics with the same asymptotic distributions (free of nuisance parameters) as stated earlier.