Friday, June 6, 2014

Does Foreign Aid Prolong Civil Conflicts?

When other nations send foreign aid with the intention of helping those enmeshed in a civil war, does the aid make the conflict worse? The question has been asked by some with high credibility in these matters, like the 1999 Nobel Peace Prize winner Médecins Sans Frontières. In reflecting on their attempts to help those in the conflicts in Chad and Darfur, the organization wrote:
We are unable to determine whether our aid helps or hinders one or more parties to the conflict … it is clear that the losses—particularly looted assets—constitutes a serious barrier to the efficient and effective provision of assistance, and can contribute to the war economy. This raises a serious challenge for the humanitarian community: can humanitarians be accused of fueling or prolonging the conflict in these two countries.
Nathan Nunn and Nancy Qian offer this quotation as a starting point for their investigation of the relationship between "US Food Aid and Civil Conflict," which appears in the June 2014 issue of the American Economic Review (104:6, pp. 1630–1666). (The AER isn't freely available on-line, but many readers will have access through library or personal subscriptions.) They find that "an increase in US food aid increases the incidence and duration of civil conflicts."

There is considerable anecdotal evidence that foreign aid may prolong civil conflict. Here's Nunn and Qian (citations and footnotes omitted):

Humanitarian aid is one of the key policy tools used by the international community to help alleviate hunger and suffering in the developing world. The main component of humanitarian aid is food aid. In recent years, the efficacy of humanitarian aid, and food aid in particular, has received increasing criticism, especially in the context of conflict-prone regions. Aid workers, human rights observers, and journalists have accused humanitarian aid of being not only ineffective, but of actually promoting conflict. These qualitative accounts point to aid stealing as one of the key ways in which humanitarian aid fuels conflict. They highlight the ease with which armed factions and opposition groups appropriate humanitarian aid, which is often physically transported over long distances through territories only weakly controlled by the recipient government. Reports indicate that up to 80 percent of aid can be stolen en route. Even if aid reaches its intended recipients, it can still be confiscated by armed groups, against whom the recipients are typically powerless. In addition, it is difficult to exclude members of local militia groups from being direct recipients if they are also malnourished and qualify to receive aid. In all these cases, aid ultimately perpetuates conflict. A large body of qualitative evidence shows that such cases are not rare, but occur in numerous contexts.
Of course, a bunch of anecdotes don't prove the general case that aid increases civil conflict. As social scientists like to say, "data" is not the plural of "anecdote." But figuring out a persuasive statistical approach for investigating whether food aid causes additional conflict is tricky. After all, if civil conflict and dysfunctional political and economic institutions lead to a situation where it looks like more aid is needed, then there might be a positive correlation between conflict and aid--but the conflict is causing the need for aid, rather than the aid causing additional conflict.

In a perfect world for social science, there could be experiments with foreign aid, where aid would be randomly given for some civil conflicts but not for other equivalent civil conflicts, and over a period of a few decades researchers could then study the results. In the real world, the challenge is to find some way of looking at the existing evidence that can approximate this thought experiment of the effects of random variation. Nunn and Qian offer an approach that is a nice illustration of a method called "two-stage least squares," and which I'll try to explain here in words.

The authors focus on two reasons why food aid for countries can vary that are not related to whether the country is currently experiencing a civil conflict. One is the size of U.S. agricultural harvests; bigger US harvests are correlated with  more food aid. The other is how likely a specific country is to receive food aid in any given year over the 36-year period of the data, from 1971-2006. These trick is first to estimate how much of the year-to-year variation in humanitarian food aid for any given country is determined by these two factors. This calculation will determine what share of the rise and fall in food aid is determined by U.S. weather, which can be viewed as an event that is increasing or decreasing food aid randomly whether there is a civil conflict or not. (This is the "first stage" of a two-stage least squares approach.)

Next, calculate whether these random rises and falls in food aid are correlated with civil conflict in the recipient countries. (This is the "second stage" of two-stage least squares.) Nunn and Qian summarize the results:

[T]he 2SLS [two-stage least squares] estimates identify a large, positive, and statistically significant effect of US food aid on the incidence of civil conflict, but show no effect on the incidence of interstate conflict. The estimates imply that increasing US food aid by 1,000 metric tons (MT) (valued at $275,000 in 2008) increases the incidence of civil conflict by 0.25 percentage points. For a country that receives the sample mean quantity of US food aid of approximately 27,610 MT ($7.6 million in 2008) and experiences the mean incidence of conflict (17.6 percentage points), our estimates imply that increasing food aid by 10 percent increases the incidence of conflict by approximately 0.70 percentage points. This increase equals approximately 4 percent of the mean incidence of conflict. 
In more detailed statistical work looking at large-scale and small-scale civil conflicts, as well as whether conflicts are starting or the length of conflicts, they find "these findings suggest that the primary effect of food aid is to prolong the duration of smaller-scale civil conflicts."

Perhaps the bottom line goes without saying, but I'll say it anyway: These results certainly don't prove by themselves that food aid is overall counterproductive, or overall a bad policy idea. They do suggest that food aid, regardless of its humanitarian intentions, has a mixture of effects, and that sensible public policy will seek ways to tilt the balance toward the good and away from the bad.