Gregory Hansen Test for Cointegration

[Rose Banda]

Hi Everyone,

I am trying to check for structural breaks in my model before running the ARDL using the Gregory Hansen test for integration command

ghansen y1 x1 -x5

I get the error message "more than 4 right hand side variables not supported."

I have also tried

estat sbsingle

and I get the message saying "insufficient observations at the specified trim level"

Is there another way I can test for structural breaks before running the ARDL and if so whats the command for it?

Thanks

[Rose Banda]

Hi All,

I have come across some published papers where the run the structural break test using ghansen and remove a few independent variables to ensure 4 right hand variables are in the model. I will run it this way for now, but if anyone has a better approach please do suggest one. I am working with time series with 25 observations


Thanks

[Rose Banda]

Hi again,

I have performed the Gregory Hansen Cointergration test successfully to check for breaks in 1. constant 2. the constant and trend. However when I try to check for breaks in the constant and slope using the command

ghansen depvar indepvars, break(regime) lagmethod(bic)

I get this:
adf(): 3301 subscript invalid
main(): - function returned error
<istmt>: - function returned error

I don't understand what this means even after reading the suggested r(33011) by stata. What am I doing wrong?

My 3 of my vars where non stationary at level and then stationary at first difference while the rest where stationary at level and also at first difference. And I have 25 observations (no gaps in time series). I don't know if this is affecting the result of the last command.

[Rose Banda]

Also found the solution