Structural Breaks

[Manuel Rodriguez]

I have a big set of time series. All of them run through 40 years and comprise 19 variables over 26 countries.

I will like to know if there is a Pre-estimation test to run to detect structural breaks. Like those post estimations test

estat sbknown

estat sbsingle

I am aiming at demostrating cointegration relationships and a VECM.

Thanks in advanced.

[Nicolás Ronderos]

Hi,

For detecting the existence and the date of a structural break you can use the Bai-Perron (1998) test in a previous stage. However there are some unit root test (you want to see if there are cointegration) that take into account the strunctural breaks and determines endogenously the date of the break, I think that the determination of the break date reduces the power of the test but I am not sure. As an advice i recommend you to read the work of Johansen (2000, cointegration analysis in the presence of structural breaks in the deterministic trend) , in there is stated another functional form for the VECM that allows the modeling of the structural breaks, I think it would become tedious with 19 variables. Good luck!

Regards,