Published on Development Impact

Climate change and conflict

This page in:

I was at the Centre for the Study of African Economies conference this week, and Ted Miguel gave a fascinating keynote lecture.   The talk is based on a paper with coauthors Marshall Burke and Solomon Hsiang where they look at the effects of climate change on conflict.    And it was fascinating because they pull together a range of different evidence to build the case that if we care about conflict we ought to be really worried about climate change.  

Let’s start at the individual level.   Here the evidence the use comes from 2 psychology studies.    The basic result is that individuals get more aggressive or violent when they experience an increase in temperatures.    These are definitely cocktail party worthy – one study uses Dutch police officers who are more likely to shoot at an intruder (a virtual intruder, in training), when they are randomly assigned to a hot room.   The second study had folks in a mall stopping intentionally in a mall parking lot when the light was green.   People behind them were more likely to honk, and honk persistently, on a hotter day.   (Keep in mind this is deviations from the normal temperature – as a Sicilian friend was pointing out to me, there might be an argument that people who always live in hot temperatures may be particularly calm).         

Then, we go to the archeological.    Here they cite a number of studies which conclude that exceptionally hot or dry periods were associated with the collapse of civilizations.   This includes societies with incomes similar to those of poor countries today such as the Mayans and the 9th century Tang dynasty.   But as Miguel and his coauthors point out, these are not comprehensive reviews of all societal collapses, but rather focus on one particular civilization.

So this is where it gets even more interesting.   What they do is to not only review the gamut of studies out there (and this is a field evolving by the month), but obtain 16 different datasets, reanalyze data from 11 papers and reinterpret the results from 6 others.    And they bring to bear a common method, using year and location fixed effects to look at the impact of temperature and precipitation deviations. 

What do they find?   The 21 empirical studies that focus on temperature show a positive association with higher temperature and violence.   Fourteen of the 16 rainfall studies also show this relationship. And these relationships appear to be pretty statistically significant.  

They go on to explain this more detail by looking at specific papers at different levels of geographic focus.    They start at the village level, revisiting Miguel’s witch killing paper but using temperature instead of rainfall.   They then move to the regional level, with a paper by O’Lauglin et. al. that shows higher temperature is associated with more violence in East Africa.   This is followed by the country level and a paper by Burke which shows that higher temperature increases civil war risk in Sub-Saharan Africa.   Finally, a paper by Hsiang et. al. uses variation in El Nino to show that higher temperature is associated with more civil conflict in the tropics.  

These effects are not small.   Miguel and co. estimate that a 1 standard deviation change in climate is associated with an 11 percent increase in intergroup conflict.   And to help benchmark this, forecasts estimate a 3 standard deviation change for Sub-Saharan Africa by 2050.  

So what is the channel here?   They cite a number of studies which show that there are clear links between economic growth and output and climate, so economic contraction is likely one channel.   Add to this the psychology work (which is still being explored as you can see from the experiments above) and you have another.  

Now of course, climate variability doesn’t fully explain all of the conflict we see out there.   But this relationship is sufficiently large as to be seriously sobering.       


Markus Goldstein

Lead Economist, Africa Gender Innovation Lab and Chief Economists Office

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000