USA Wall, Dire Dawa, Ethiopia: Child leaning against wall made of USAID food aid containers in the flood-destroyed area of Bahere Tsege in Dire Dawa. Photo: Liz Lucas/Oxfam America

A new study reminds us that even simple aid tools can have complex effects.

Does food aid cause civil conflict? A new study finds that it does, in certain circumstances. But the answer to the question is far from a straightforward yes or no.

Before rushing to make broad conclusions, it’s important to understand exactly what the study (ungated here), published in the American Economic Review, analyzed and what it found. In very short form, here’s what it found:

• A 10% increase in food aid delivery increases conflict by nearly 0.7%.
• Food aid seems to extend civil, domestic conflicts in cases where recipient countries have been long-time food aid recipients and where conflict has been recurring for a long time.
• Food aid doesn’t appear to cause conflicts to start.
• It doesn’t seem to matter if the conflicts are big or small.
• It doesn’t seem to have any effect on inter-state conflict (rather than intra-state).

Now, both conflict and food aid are pretty complicated. So, it took a lot of clever thinking to design a study that could analyze the effect of food aid on conflict. Drawing a simple correlation between food aid and conflict (more food aid = more conflict) is obviously flawed because conflict creates the need for food aid. I’ll spare you the details, but the authors found a credible way to measure whether increased food aid results in increased conflict. I’m not a statistician, but it’s a pretty good study marshaling a lot of data and using some interesting and innovative techniques.

Issues and concerns:

The study authors propose that the mechanism for food aid to lead to more conflict is that belligerents benefit from the food aid, which extends their capacity to fight. There’s a simple truth here: conflict is more likely to continue if the belligerents are alive. To the extent that food aid helps keep belligerents alive, it helps keep the conflict alive. That’s an oversimplification, but that’s the gist of it.

There’s a lot of anecdotal and case-study evidence that food aid is diverted by or controlled by belligerents. The authors of the study, in their discussion, tend to focus on examples when insurgents and rebels capture food aid which is used to support their troops, build goodwill among local populations, and sold to raise funds. This sort of diversion is more likely for food aid that is transported over vulnerable territory.

But in truth, most food aid during much of the study period (1971-2006) was not delivered as humanitarian aid, but was sent as economic assistance to friendly cold-war ally governments. From 1970 to 1990, 60% of food aid was sent as “title I” concessional sales for economic and market assistance rather “title II” for emergencies and development projects. In other words, the US sent crops to other country governments, who took control of the shipments, often selling them for cash. Food aid was a major aid program; it wasn’t used as food, it was used as a fundraising mechanism.

Further questions:

So, does the proposed mechanism for impact on conflict still make sense if the food aid was never delivered as assistance, and therefore was never hijacked or diverted by belligerents, or sold by governments for cash or for other special projects? This might not change the results of the study, but it would change the implications and the understanding of how the food aid causes conflict than about whether aid causes conflict.

This leads to a larger question why food aid seemed to cause more conflict in some conditions, and not in others. Why doesn’t food aid have an impact on inter-state conflict? Why only for cases where conflict has been ongoing? Do the authors have a hypothesis for this? Or could it be that the data and methods do not offer sufficient statistical detail to dig into these questions? Without more explanation, it seems very dangerous to draw any conclusions, much less recommendations, from this study.

The study points to the need for understanding negative impacts and trying to reduce them. If a lot of food aid is hijacked, how to reduce this “leakage” and ensure that control for humanitarian purposes and targets are maintained.

Thinking about the implications of this study, there’s a further analysis needed to operationalize its implications. The fact that food does – by some mechanism – extend conflict does not automatically mean we shouldn’t send food aid. The question will become: is the net benefit of the food aid higher than the cost? Do the lives saved and improved by food aid outweigh the lives lost to conflict and associated misery and impoverishment? That won’t be an easy analysis.

Overall, I welcome this study in particular and, more importantly, the broader goal to spend time analyzing not just the direct impacts of aid, but the indirect, secondary, and structural results. It’s important not to draw too much from this study, but it’s a great reminder that even pretty simple aid tools, like food aid, can have complex and contradictory impacts.