Evaluating Nationwide Health Interventions : Malawi's insecticide-treated-net distribution programme

We evaluate Malawi's main malaria prevention campaign, a nationwide insecticide-treated-net distribution scheme, in terms of its effect on infant mortality. Methodologically, evaluating such nationwide health interventions is particularly difficult. There is no contemporaneous comparison group that has not been subject to the intervention. Moreover, common environmental trends, the availability of new drugs and a variety of other health improving measures used at the same time imply that the often advocated before-after estimator is not a good choice. We propose an alternative estimator that can be used if the intervention influences health through its effect on individual health-seeking behaviour but has no other effect on the outcome. We also suggest some plausibility checks and falsification tests to assess the validity of the identifying assumptions that we impose in applications. Using the estimator proposed we find that Malawi's insecticide-treated-net distribution campaign reduced all-cause child mortality by about 1 percentage point, which corresponds to about 40% of the total reduction in infant mortality from 8.2% to 5.4% over the study period.