We exploit \(0.5^{\circ}\times0.5^\circ\) raster data to document how exceedances of the local \(90^{th}\) percentile thresholds for daily maximum and minimum temperatures affect conflict in mainland Southeast Asia. We show that conflict incidence increases with extreme high maximum temperature days and decreases with extreme high minimum temperature days. This implies that failing to control for extreme minimums understates the effects of extreme maximums. Moreover, as the frequency of extreme maximums and minimums is expected to increase together with average temperatures, the countervailing effects at both tails of the temperature distribution offset one another in mean-temperature regressions, helping to explain earlier inconclusive findings for the region. We also show that the effects of extreme maximums and minimums differ by conflict type, actors involved and affected populations. Thus, even in the absence of an aggregate mean-temperature effect, a rising frequency of extreme temperature days may generate complex distributional conflict incidence.
