Is it ok to use the Bai-Perron method for detection of structural breaks in a cointegrating relationship? If so, is it correct to include all the long-run independent variables and allow for breaks only in the constant (and trend)?
Bai Perron is for monthly data set, for single series, this have nothing to do with co-integration. it is better to visualize the series to get better understanding
Thanks, but Zivot-Andrews is not for cointegration as far as I know, and the Gregory-Hansen package in Stata can only detect one break date.
You can see the paper by Joakim Westerlund and David Edgerton in
Journal of Time Series Analysis, Vol. 28, No. 2, pp. 188-224, March 2007
You must analyze he graphs of time series and to make a simple regresion. Then use Bai-Perron o Quandt-Andrews. After consider unit root test with structural breaks (ZV, Perron or in Eviews ADF with structural breaks). If you have I(1) with structural breaks use Johansen, Mosconi and Nielsen test in the break-points (one o two breaks) or Gregory-Hansen in RATS with one break point. In some test you need to append the break point like a dummy variable.
- Badges
- Science topic
- Similar topics
- Correction