This paper is concerned with ex ante and ex post counterfactual analyses in the case of macroeconometric applications where a single unit is observed before and after a given policy intervention. It distinguishes between cases where the policy change affects the model's parameters and where it does not. It is argued that for ex post policy evaluation it is important that outcomes are conditioned on ex post realized variables that are invariant to the policy change but nevertheless influence the outcomes. The effects of the control variables that are determined endogenously with the policy outcomes can be solved out for the policy evaluation exercise. An ex post policy ineffectiveness test statistic is proposed. The analysis is applied to the evaluation of the effects of the quantitative easing (QE) in the UK after March 2009. It is estimated that a 100 basis points reduction in the spread due to QE has an impact effect on output growth of about one percentage point, but the policy impact is very quickly reversed with no statistically significant effects remaining within 9-12 months of the policy intervention.