Page 1 of 1

VEC and impulse reponse

Posted: Mon May 20, 2013 7:46 am
by fboehlandt
Hi everyone,
I am estimating a vector autoregressive model from four index series. Since the four series are ~I(1) and jointly cointegrated, I opted for a VEC. When conducting an impulse response test, the model appears somewhat unstable (i.e. the shock does not die away gradually). I suspect this comes about as a result of the series being cointegrated. Is there a way to conduct an impulse response test that accounts for the cointegrating relationship?
Thanks for your input

Re: VEC and impulse reponse

Posted: Mon May 20, 2013 7:56 am
by fboehlandt
From Userguide II:
For stationary VARs, the impulse responses should die out to zero and
the accumulated responses should asymptote to some (non-zero) constant.
As expected, the impulse reponse should die out for stationary VARs only. I find it difficult to interpret the impulse reponse for VECs. Is there an appropriate way to conduct the test?