Hi,

I would like to test a time series variable for unit root. I have one model with 4 impulse dummies, one model with impulse and step dummies, and one last model with only step dummies (I have estimated the dummies with Autometrics in PcGive). In addition to the dummies I have several lags of the variable of the endog. variable.

1. Does the model with only step dummies imply structural breaks? (I´m a bit confused what the difference is between step and impulse dummies). If not - can I use the ADF-test to test for unit root when including only impulse dummies? If I use the optimal lags when including impulse dummies, my variable is I(0) with the ADF test.

2. I´m looking for a unit root test for multiple breaks. Does anyone know of a test for multiple breaks + a less technical explanation of the test?

3. Or does anyone know of a textbook which describes the GLS based test in a less technical way?
I have found one; the GLS based unit root test with multiple structural breaks by Carrion-i-Silvestre, J., Kim, D., & Perron, P. (2009).

4. Is it possible to perform such a test in Stata?