Global markets and the differential effects of climate and weather on conflict

Both climate and weather have been attributed historically as possible drivers for violence. Previous empirical studies have either focused on isolating local idiosyncratic weather variation or have conflated weather with spatially coherent climatic changes. This paper provides the first study of the differential impacts of climate and weather variation by employing methods developed in earlier work linking the El Nino Southern Oscillation (ENSO) with the onset of civil conflicts. By separating the effects of climate from local weather, we are able to test possible mechanisms by which atmospheric changes can cause violence. It is generally difficult to separate the effect of year-to-year climate variations from other global events that might drive conflict. We avoid this problem by examining the set of tropical countries that are strongly teleconnected to ENSO. For this region, the ENSO cycle parallels the common year-to-year pattern of violence. Using ENSO, we isolate the influence of climatic changes from other global determinants of violence and compare it with the effect of local weather variations. We find that while climate affects the onset of civil conflicts in teleconnected countries, local weather has no significant effect. Productivity overall as well as across major sectors is more affected by local weather than by climatic variation. This is particularly evident in the agricultural sector where total value and cereal yield decline much greater from a 1°C increase in local temperature than a 1°C increase in ENSO. However, when examining the effect on food prices, we find that ENSO is associated with a large and statistically significant increase in cereal prices but no effect from hotter local temperatures. Altogether, this evidence points toward the ability of global and regional commodity markets to insure against the effects of local weather variation and their limitations in containing losses from aggregate shocks such as El Nino events. We posit that conflict reacts to climate and not weather because climatic events trigger not only local agricultural losses but also increased food prices as a result of an aggregate decline in output. This is because in an open economy, idiosyncratic weather variation alone would not lead to higher prices. These results are informative in understanding the impacts of anthropogenic global change, which would yield variation exhibiting spatial coherence beyond the extent of existing markets.