When analyzing the results of a VAR model with 12 lags, should i take care for the coefficients of all 12 lags, or should i only look up for the coefficients of the last (12) lag?
Asked
Active
Viewed 36 times
0
-
What do you want to do with these coefficients? – Richard Hardy Feb 18 '17 at 17:39
-
I mean, when i am interpreting the coefficients in the analysis. Should i explain the coefficients in the other lags, or only the last lag? – vannila Feb 19 '17 at 17:13
-
VAR coefficients do not have clear interpretations, unless it is a structural VAR model. It is easier to interpret impulse response functions and forecast error variance decomposition instead. But in general, all lags are equally important. – Richard Hardy Feb 19 '17 at 17:22
-
This was very helpful. Thank you for the clarification Richard. – vannila Feb 19 '17 at 17:25
-
See also [this](http://stats.stackexchange.com/questions/40905/arima-model-interpretation/63072#63072) which applies not only to ARIMA but also to VAR. – Richard Hardy Feb 19 '17 at 17:32
-
Thanks again. And when it comes to interest rate, does it matter if in the model they are included as percentage, or a number divided by 100 (ex. 4.3% od 0.043)? – vannila Feb 19 '17 at 17:37
-
That does not matter for estimation, but you have to interpret the results accordingly. – Richard Hardy Feb 19 '17 at 18:19