Hello all guys:
This is my very third post, I'll try to explain myself better to see if someone knows if this is doable in STATA.
I'm running an ADF test using rolling windows and recursive samples, specifically this command:
rolling, window(4) recursive clear start(0) end(178) : dfuller lgpr
lgpr is just a logarithm of a price to rent ratio
When you use a rolling window and recursive samples for the ADF test, what you do is running the (sup) ADF test regression on subsamples of the data in a recursive fashion.
The size of my total sample is 178 and what STATA is doing right now is running the regression from t=0 to t=1, then t=0 to t=2, t=0 to t=3 ... until t=0 to t=178.
The matter is that I don't want the starting point fixed at 0, I would need this point changing within a feasible range, i.e. from 0 to 170.
So in this rolling ADF (generalized sup) the regression would be ran from t=0 - t=2, t=0 to t=2, so on until t=0 to t=178. But also it would change the starting point from t=0 to t=1, all the recursive regression again until t=178. Then change the starting point again from t=1 to t=2, all the recursive regression again. And so on until the starting point is t=178.
The graphs above show my problem. The first graph is what STATA allow me to test, and the second in the right is what I want to do.
My sample interval in this case would be [0,178].
There is no need to use Dickey-Fuller tests to know how it works a rolling window and the recursive regression. I just need to know how to process this through STATA, if possible.
The hypothetic command could be something like:
rolling, window(4) recursive clear start(0, 178) end(178) : dfuller lgpr
?
Is there any way to perform this?
Thank you very much,
Ignacio
This is my very third post, I'll try to explain myself better to see if someone knows if this is doable in STATA.
I'm running an ADF test using rolling windows and recursive samples, specifically this command:
rolling, window(4) recursive clear start(0) end(178) : dfuller lgpr
lgpr is just a logarithm of a price to rent ratio
When you use a rolling window and recursive samples for the ADF test, what you do is running the (sup) ADF test regression on subsamples of the data in a recursive fashion.
The size of my total sample is 178 and what STATA is doing right now is running the regression from t=0 to t=1, then t=0 to t=2, t=0 to t=3 ... until t=0 to t=178.
The matter is that I don't want the starting point fixed at 0, I would need this point changing within a feasible range, i.e. from 0 to 170.
So in this rolling ADF (generalized sup) the regression would be ran from t=0 - t=2, t=0 to t=2, so on until t=0 to t=178. But also it would change the starting point from t=0 to t=1, all the recursive regression again until t=178. Then change the starting point again from t=1 to t=2, all the recursive regression again. And so on until the starting point is t=178.
The graphs above show my problem. The first graph is what STATA allow me to test, and the second in the right is what I want to do.
My sample interval in this case would be [0,178].
There is no need to use Dickey-Fuller tests to know how it works a rolling window and the recursive regression. I just need to know how to process this through STATA, if possible.
The hypothetic command could be something like:
rolling, window(4) recursive clear start(0, 178) end(178) : dfuller lgpr
?
Is there any way to perform this?
Thank you very much,
Ignacio
Attached Files
Comment