Currently reading an interesting paper by Eleanora Nillesen & Philip Verwimp on whether agricultural shocks (namely rainfall shocks) increased people’s likelihood to actively participate in Burundi’s civil war. They find that, whilst negative shocks to the price of coffee (the country’s principle cash crop) didn’t increase rebel recruitment, drought shocks were positively correlated with recruitment, perhaps underlining the greater role of agriculture in helping households manage risk – it could be a greater blow to lose consumption crops than to receive a lower yearly payment for a cash crop. (It was especially interesting to read this in light of Heather Sarsons’ recent work [PDF] questioning the use of rainfall as an instrument for wages from agricultural labor, based on new data from India. She suggests that, unsurprisingly, rainfall may affect people’s participation in political protests through channels other than the creation of grievance & reduced opportunity cost of involvement. Since N&V weren’t using rainfall as an instrument, this critique doesn’t directly apply to their work, but it’s still useful to think through the multiplicity of ways in which rainfall affects people’s lives in low income countries.)
N&V also included a small methodological note which I found especially telling in re: the social importance of land in Burundi. The sample for the data underlying this paper was drawn from households who completed the 1998 Burundi Priority Survey, which was a joint project of the Burundi Institute of Statistics & Economic Studies and the World Bank. Interestingly, despite the fact that N&V collected a second round of data in 2007, after multiple years of war, they noted only 13% attrition from their original sample. As they write, “In Burundi the pressure [on] land is extraordinary high… As a result people may have only have fled [from the conflict] at the very last minute, if there was no other option, and return[ed] immediately after the violence…to ensure their claim to land. Most often, our survey team would find the households in the same location as in 1998” (p. 6). Pretty remarkable.