3
$\begingroup$

enter image description here

I want to study the impulse response function and the variance decomposition by fitting a VAR model. The lag length criteria gave me this result. What's the problem?

$\endgroup$
  • $\begingroup$ Your model is miss-specified if it includes lag 0. If the 0 is in fact the first lag, then there is no problem, just pick the lag that has highest value according to some criterion (that would be one lag). Remember that this procedure assumes same data also. $\endgroup$ – Dole Apr 7 '16 at 17:46
  • $\begingroup$ I don't get your answer. How can a model be miss-specified I didn't estimate the model. I'm seeking for the optimal lag to fit a lag model. I have 4 stationnary. I made them stationnary ps: I'm using eviews $\endgroup$ – user109464 Apr 7 '16 at 18:03
  • $\begingroup$ @Dole, 0 lag may mean no lags, not inclusion of the zero lag. $\endgroup$ – Richard Hardy Apr 7 '16 at 18:10
  • $\begingroup$ Yes, it means that the optimal lag is zero. What could be the cause of such a result? $\endgroup$ – user109464 Apr 7 '16 at 18:16
  • $\begingroup$ @user109464 That would mean mere constant maximizes the criterion, since none of the VAR model should be left. Again, VAR model can't be estimated with 0 lags, so I hope you have correct specification (IE 1 = first lag, 2 = first and 2nd lag etc.) No contemporaneous terms should be allowed. $\endgroup$ – Dole Apr 7 '16 at 18:19
2
$\begingroup$

Based on the criteria (lowest AIC, lowest BIC, etc.), zero lag is preferred, which means a model with just an intercept but no lags. From such a model you will not be able to obtain impulse response functions, while variance decomposition will be trivial (none of the variables explains the variance). The selection of zero lag suggests that VAR might not be a good model for your data, so you might want to explore alternative models.

$\endgroup$
-1
$\begingroup$

If zero lag presents mean there is no contribution of past value of the exogenous variable. So it is not a time series problem It is simple regression

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.