<p>Nunn and Qian (Am Econ Rev 104:1630–1666, 2014) find that an increase in US food aid increases the incidence of civil conflicts. However, their identification strategy (based on shift-share instrumentation) is invalid because the shares used to construct the shift-share instrument do not add up to one. Even when the shifts are exogenous, countries with larger shares systematically have higher values of the instrument, introducing bias if these countries have different unobservables that affect conflict. The exogeneity of the instrument can be restored by adding a sum of shares control interacted with year fixed effects. In Nunn and Qian (Am Econ Rev 104:1630–1666, 2014), the sum of shares control is time-invariant and country-specific, naturally giving rise to a common factor model representation. We replicate Nunn and Qian (Am Econ Rev 104:1630–1666, 2014) using common factor models. We find evidence that US food aid reduces the incidence of conflicts. Our baseline estimates suggest a 10% increase in US food aid would result in a 0.3 percentage point decrease in the incidence of conflicts (or a 1.5% decrease at the sample mean). Our results are robust to first-differenced and dynamic model specifications. We also find that this conflict-reducing effect does not disappear after the Cold War. We also find that US food aid decreases the incidence of small-scale conflicts, the incidence of civil conflicts, and the onset of civil conflicts.</p>