SUMMARY
As we emphasized in the general methodological sheet on public policy evaluation, ex-post impact assessment involves identifying the causal links between a policy and the changes observed. This means comparing the effects of a policy with a fictitious situation in which no policy would have been implemented. This requires the evaluator to determine a reference situation, a control group or counterfactual. To do this, it is necessary to correct for biases both observable and unobservable by the evaluator, otherwise there is a risk of arriving at conclusions that are not robust: erroneous or inverse causal links, underestimated or overestimated impacts, etc.
Keywords: causality, difference-in-differences, estimation, evaluation, randomized controlled experiments, impact, method, matching methods, public policy, discontinuity regressions, instrumental variables
Illustration: www.modernisation.gouv.fr/
Publication is reserved for founding members and partners of the A&NMT Chair. If you have access, log in to your partners area to read the rest.
