Historical decomposition on VECM vs VAR

[sk1234]

Hi,

I am wondering whether it is normal to get such different historical decomposition results when performed on a VECM versus a VAR. I have checked that the VAR is stable and that the equivalent VECM is also stable so I am unsure why this is happening. Please see images attached. The VAR one looks normal, but the VECM looks off (for all variables not just the one attached).

Both use the same identification process (Choleksy df adjusted) with the same ordering.

I am using eviews 12. The same problem occurs with eviews 13.

Cheers,