Detailed geographic information, conflict exposure, and health impacts

•We estimate impacts of war exposure on children’s health using geographic information on household distance to conflict sites.•Results using a regional measure indicate reductions for exposed children of 0.25 SD in Ethiopia and 0.73 SD in Eritrea.•Results using GPS information on household distance to conflict are 2–3 times larger compared to measures at regional levels.•Results show children nearest to conflict face the largest negative effects (0.72 SD in Ethiopia and 1.37 SD in Eritrea).•Negative war impacts on children diminish as the household's distance from the conflict increases. We expand existing research by estimating the impact of exposure to conflict on children’s health outcomes using geographic information on households’ distance from conflict sites—a more accurate measure of shock exposure than the traditional approach using regional-level information—and compare the impact of exposure in utero versus after birth. The identification strategy relies on exogenous variation in the conflict’s geographic extent and timing. Using multiple waves of survey data from Ethiopia and Eritrea, we find that conflict-exposed children have significantly lower height-for-age. With GPS information that enables accounting for households’ distance from the conflict sites, negative impacts of conflict exposure are two to three times larger than if exposure is measured at the imprecise regional level. Results are robust to addressing potential exposure misclassification due to migration happening between the war and the survey collection date.