Hi, I have a question regarding the treatment of exogenous variables in a VECM.
I am using Eviews 12 and all seven variables are I(1).
I have estimated a VAR (with all variables in levels) that contains 5 endogenous and 2 exogenous variables and a constant. From this specification it says that the optimal lag length for the VAR according to the information criterions is around 2 lags.
I then use the Johansen cointegration test using option (3) - intercept (no trend) in CE and test in VAR. I specify the lag intervals as 1,1 given that the VAR optimal lag length was 2 (if someone could clarify that I have specified the lag interval correctly by subtracting off 1 lag that would be great). I also put the two exogenous variables in exogenous variables box (not including the constant). The two test results suggest one co-integrating relation exists.
When I now turn to estimate the VECM (the one cointegrating relation has no restrictions - all five endogenous variables are included in the equation with the first enodog variable normalised to one) I now want to ask
(1): whether putting in (1,1) for the "lagged intervals for endogenous" is correct,
(2): does putting the two exogenous variables into the 'exogenous variables' impact the estimation results of the VECM, because the cointegrating part of the VECM output does not include these variables - its just the same output as the first step of the Johansen procedure. The exogenous variables only appear in the second step VAR in first differences. I have been told that the way you estimate the co-integrating vector differs from its estimation in the VECM? Are you able to confirm whether the exogenous variables are being incorporated correctly into the VECM estimation and how the first part of the VECM output is estimated?
(3): If the optimal lag length from the VAR is 2, do I need to also specify 2 or 1 (or no) lags for the exogenous variables (which are I(1)) when entering them into the VECM 'exogenous variable box'. Does this change if the exogenous variables were instead I(0)?
(4):When i click on 'proc' -> 'make residuals' after estimating the vecm, can i confirm how these residuals are obtained? Are they the reduced form residuals coming from the second step VAR in first differences? They differ from the residuals obtained from my original VAR estimation in levels.
Cheers
VECM exogenous variables
Moderators: EViews Gareth, EViews Moderator
Who is online
Users browsing this forum: No registered users and 2 guests