Am 11.01.2017 um 02:16 schrieb Allin Cottrell: If I understand George correctly, he is looking for the combination of a unit root and a structural break test, presumably with the structural break under the alternative hypothesis. Then this perron97.inp script would be the thing to look at I guess, not necessarily the other sources Allin mentioned. It is true this could be made into a function package, it wouldn't be too difficult. Maybe the most tedious work would actually be to collect all the non-standard critical values or to think about a clever way how to provide those to the user. BTW, if I remember correctly there are some subtle issues involved like a shift dummy for the levels needing to be augmented by further impulse dummies for the lagged differences and so forth. In the original articles that wasn't always done I think. cheers, sven