I have time series for 35 years and want to detect unknow structural breaks as suggested by Kim and Perron (2009). Kindly help me which software should be suitable to conduct this analysis and how to command..??
Yes it is possible to test for stationarity by a unit root test, i.e. Augmented Dickey Fuller test or Phillip Peron test, using eviews. Eviews is one of most user-friendly softwares for time series analysis.
The ADF and other test are old -generation tests. I saw that there are some news unit root test (with structural) breaks in Eviews 10. Try the last version. Personally i hadn't the occasion to do it so i don't know if fits your desire.
Dear Nazia, there is currently no formal procedure for implementing Kim and Perron (2009) in EVIEWS, not even an add-in for such test. However, I suggest you follow Prof. Dave Giles blog "Econometric Beat", he might have a useful step on how to trick EVIEWS into implementing the test. Meanwhile, you may which to download and install Zivot-Andrews (ZA) add-in for performing unit root test for unknown break. If your ZA test reveals structural break, then you may have to apply the "Break Point Unit Root Test" which is based on modified ADF that allows for a break, usually found in EVIEWS VERSIONS higher than 8.
Nazia, when MULTIPLE unknown structural breaks are in the arena for a time series, and you wanna test stationarity, the BEST way is Narayan-Popp 2010 nonstationarity test since NP2010 uses innovational outliers technique, and perfectly locate position of structural breaks as an ON THE FLY during the execution of the test.
Narayan-Liu 2015 is the version that handles the possibility of heteroskedasticity in the series as well.
Superiority of NP2010 test to earlier structural break tests (including Zivot-Andrews, Lee-Strazicich etc.) can be found in Narayan's 2013 paper.
You can find the Gauss code for NP2010 test in Narayan's RG page. Gauss 18 trial student version is free. You may go with this path.
Since I did not like that code, I wrote an R code (I'm an useR; we think therefore we R!) to perform NP2010 test with ggplot2 to present the results visually.