Impulse Response Function in VECM
Posted: Thu May 04, 2017 12:36 pm
Hi,
I have estimated a VECM for 10 variables with 4 cointegrating equations, and am consequently using IRFs as my primary source for interpreting the impact of each of the system's variables on y.
However, I am confused by the interpretation of the EViews IRF output. All of my variables are natural logs, and the IRFs show the accumulated response to generalised one s.d. innovations. However, I am primarily interested in the impact of a 1% change in Xi on y, expressed in terms of an x% change in y. Are the impulse response functions, at present, expressing the results as an x s.d. change in y? E.g. a 1 s.d. increase in Xi leads to an 0.8 s.d. change in y after 10 periods? If this is the case, is there a simple way to make EViews express the results in terms of percentage changes?
Thanks in advance!
I have estimated a VECM for 10 variables with 4 cointegrating equations, and am consequently using IRFs as my primary source for interpreting the impact of each of the system's variables on y.
However, I am confused by the interpretation of the EViews IRF output. All of my variables are natural logs, and the IRFs show the accumulated response to generalised one s.d. innovations. However, I am primarily interested in the impact of a 1% change in Xi on y, expressed in terms of an x% change in y. Are the impulse response functions, at present, expressing the results as an x s.d. change in y? E.g. a 1 s.d. increase in Xi leads to an 0.8 s.d. change in y after 10 periods? If this is the case, is there a simple way to make EViews express the results in terms of percentage changes?
Thanks in advance!