So it seems like my data has a structural break in it, using Box-Jenkins methodology and evaluation such as ADF test wont give appropriate forecasts/results if I do not correct for the structural break in the data.

I don't understand how to go about fixing the structural break and then doing the ARIMA procedure of identification, estimation, diagnostic checks ext.

I know one way is to dummy out variables....however how do I include this in the proc arima step? I can create dummy variables in the data set but its not like including the dummy's in my data set will do anything to the ARIMA process because the arima process doesn't do anything about additional variables in the dataset except for the one variable I am trying to forecast.

  • $\begingroup$ Can't you simply estimate 2 ARIMA, one before and one after the break? $\endgroup$
    – Louis. B
    Nov 20 '15 at 2:59
  • $\begingroup$ Also see Bai and Perron (2003) it may help to confirm that you have a break in your data... $\endgroup$
    – Louis. B
    Nov 20 '15 at 2:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.