(This is able to entail jogging a baseline regression for both of those sets of data independently, then generating comparisons of baseline-predicted consumption with genuine consumption for both equally sets of information individually, then combining the effects collectively at the end. This post