If my memory is correct, we can obtain the impulse response function (irf) with bivariate VECM .
However, I read some researcher just difference the variables under the pre-specified relation.
For example, consumptions and corporate earnings are cointegrated with beta=1 in their arguments , so they just differenced the consumption with itself and the consumption with the corporate earnings to obtain irf in levels.
My question are 1. Does the method make sense? 2. If so, how can I get irf?
Thanks in advance