But Isn't Doing Something Better than Doing Nothing?

Sadly, not always. 

On a recent trip home, I visited my mother's Dining for Women group, where her friends raise money for small projects in developing countries, while learning more about the surrounding issues. I went to provide some background on education in Kenya. After telling me how much I've grown since the last time they've seen me, the members of the group asked for my thoughts about various aid or charity programs that they've heard about. After I hesitantly admitted some skepticism about the ability of a lot of interventions to generate noticeable improvements in well-being, one looked at me and asked, somewhat incredulously, "But isn't doing something better than doing nothing?" At the time, I responded vaguely, with "Well, I'm not sure," and wandered off to get another delicious homemade brownie. 

I get it. It's really uncomfortable to sit by when people are suffering. And this discomfort should motivate action. But this doesn't mean that all well-intentioned somethings are better than nothing. 

Yesterday, the American Economic Review  published two excellent articles that clearly demonstrate this point. Each paper identified a situation in which there appears to be strong evidence that aid resulted in an increase in violence. 

Demonstrating such causality isn’t easy. Simply comparing places that get aid with places that don't will not show the impact of aid, because aid is often targeted to places that are worse off. Finding that these places are more violent would merely demonstrate correlation, not a cause-effect relationship. Each of the papers uses a brilliant identification strategy to get around this issue, with devastating findings. 

Nathan Nunn and Nancy Qian, in their paper US Food Aid and Civil Conflict, exploit a powerful determinant of food aid that has nothing to do with the place getting the aid. It turns out that an excellent predictor of how much food aid the US sends is how big the wheat surplus is in the US. The paper shows that when there's more rain in the US, there's more wheat production, and the US sends more food aid abroad. So the authors compare places that happened to have a famine during a year of good rains in the US (therefore, more likely to get food aid) with places that happened to have a famine during a year of bad rains in the US (less likely to get food aid). They find that this aid increases both the likelihood and the duration of civil conflict.

Benjamin Crost, Joseph Felter, and Patrick Johnston , in their paper Aid under Fire: Development Projects and Civil Conflict, estimate the impact of a World Bank-funded project in the Philippines in which eligibility for the program is determined by a fixed poverty threshold. Municipalities above this level are ineligible, while those below are eligible. The poorest and the richest are different, but those very close to the cut-off are very similar. By comparing subsequent violence in these municipalities, the authors find that this aid program increased the number of casualties. 

Both of these papers are also important because they address causes of conflict over which someone has control. Poverty or natural resources may also cause variation in violence, but they're a bit harder to change. Aid is a policy lever. 

To be clear, this definitely does not imply that all aid is bad, and there is certainly plenty of evidence of benefits.

Bottom line, though, is that those in a position to determine aid allocations should take seriously the warnings in these two papers. At least in some instances, doing something may be worse than doing nothing.