What should we use as a measure of malaria infection risk? Implications from infant mortality during the Liberian Civil War

What should we use as a measure of malaria infection risk when examining its impacts on socio-economic outcomes? While endemicity is a natural candidate, it is usually endogenous. An indirect risk measure relying on climatic determinants of malaria and/or vector ecology is more exogenous; however, it may not reflect all factors representing malaria risk. This study addresses this question by focusing on the Liberian civil war. Taking a difference-in-differences approach, it first shows increased infant mortality following wartime pregnancy in malaria-endemic areas, based on respondents' full birth histories provided by the Liberian Demographic and Health Survey (2007 and 2013). This mortality effect is robust and plausible from numerous perspectives, highlighting a country's and/or the citizens' diminished capability to control malaria transmission in wartime. Nevertheless, this effect does not hold once malaria suitability indexes are used as a measure of malaria risk. However, these indirect measures also did not have a relationship supposed to be 'correct' with a community's temperature and precipitation. Taking these findings together with those provided by prior studies, these direct and indirect measures of malaria risk may be consistent across broad regions in explaining economic outcomes, but not necessarily within a particular country. This remark may serve as an important caution, particularly when analysing countries affected by the large-scale armed conflict and the resulting population displacement.