Hi Balaji,

For two structural breakdown, you can run Lumsdaine and Papell (1997) test.

Ya, that what i mentioned as AO & IO (Additive Outlier & Innovative Outlier models) above. And looking for >2 breaks.

Thanks lanouar, i can't download it, as it is $25. I hope will get this soon.

Thanks Balazs, I, atleast now, am trying to study such breaks in context of behavior in different regressors. And i'm not convinced on accuracy of forecasting the series, because of simultaneous relation b/w variables.

Have a look at my 2004 Economics Letters paper, "Bayesian evidence on the structure of unemployment." I run Bayesian versions of ADF tests with up to 5 unknown breaks. There's a working paper version on my profile page.

Bai-Perron test is an alternative to Quandt-Andrews' test in estimating regression with structural break and detection of break. For purpose of testing for unit root with structural break, the perron's test should be applied. You need to clarify what you need in your research. Typically, >2 breaks point is difficult to find in empirical studies unless you have very long span of data. Make sure that in your research, you really need more than 2 breaks test.

Lanour, your paper "Breaks or long memory behavior: An empirical investigation" is very interesting

What about the same question applied to panel data? Unit Root testing with Structural breaks =>2 in Panel Data

Hello, am using a data from 1981-2016, am sure there was a break in 2000. But because Zivot-Andrew Andrew can only choose one break, it chose 1994. Am not interested in what happened in 1994, am more interested in what happened in 2000. How do I make Zivot-Andrew check the break date of year 2000