Cash for conflicts

By S.H. | LONDON

THIS week’s Free exchange column looks at the latest research on foreign aid and economic growth. While we report that there is mounting evidence suggesting that aid has a positive effect on growth, we argue that cost-benefit analyses of development cash are much needed. If the positive impact on growth is small relative to the costs, spending money on aid may not make much sense. There are alternative policies to aid that also alleviate poverty and boost incomes, but perhaps may do so at a lower cost.

So what are the costs of foreign aid? One important negative effect of aid, according to new research, is civil conflict. Benjamin Crost of University of Colorado Denver and his co-authors looks in a recent paper at development projects and civil-conflict deaths in the Philippines. So-called community-driven development (CDD) programmes have become increasingly popular in recent years. The idea behind CDD is that much of the control over aid projects should rest with the communities rather than with western bureaucrats since the former know more about the context and the needs of the local population. Intuitively this sounds like a sound approach. In 2012, the World Bank supported some 400 CDD programs in about 100 countries amounting to $30 billion.

The researchers studied the impact of the Philippines’ flagship CDD program, which was implemented in 2003. The World Bank explains the purpose of the project as follows.

It strengthens community participation in local governance and develops local capacity to design, implement, and manage development activities. Community grants are used to support the building of low-cost, productive infrastructure such as roads, water systems, clinics, and schools.

The cash grants were sizeable and amounted to roughly 15% of an average municipality’s annual budget. To prioritise the neediest communities, eligibility for the program was restricted to the poorest 25% of the municipalities in the most impoverished provinces of the Philippines. Mr Crost and co-authors were therefore able to compare the municipalities who were barely eligible with those that just missed the threshold to see what the outcomes of the program were.

Because falling right above or below the poverty threshold is just a matter of a few dollars, there is little reason to expect that the municipalities around the 25th income percentile are much different from each other. After all, if we look at a large number of individuals who fall just below and just above the 25th income percentile in a country, we would expect them to have, on average, similar socioeconomic backgrounds, educational attainment, professions, and so on.

This makes the researchers’ inquiry akin to a randomised controlled trial where some municipalities basically were “treated” with the CDD program and others were not. Strikingly, the results show that the municipalities that were just eligible for the CDD program experienced a large increase in conflict-related casualties compared to those who were just above the 25th income percentile. The authors find that that most of these deaths occur in attacks during the preparatory stages of the program. They surmise that insurgents try to sabotage aid projects, because if government-supported projects succeed they may weaken the public support of the insurgents.

Some may argue that outcomes like conflict or corruption are hard to avoid when large sums of aid money are involved. Perhaps non-pecuniary aid like food would deliver safer returns with a lower risk of unintended side-effects? Alas, another recent paper suggests that such aid may not do much better.

Nathan Nunn of Harvard University and Nancy Qian of Yale University have studied the impact of American food aid on civil conflicts from 1971 to 2006 in 125 developing countries. Just as with the research on foreign aid and economic growth, the analytical dilemma is to determine causality. Perhaps the United States is more prone to send food aid to conflict-shattered countries. Or maybe the American government sanctions regimes who do not manage to curb internal violence by reducing aid flows.

To avoid mistakenly capturing any causal effect that conflicts may have on aid, the researchers looked at American weather and wheat production. Year-to-year differences in wheat yields are largely determined by weather. And since the United States government stabilises prices by purchasing wheat from farmers at a fixed price, it accumulates excess reserves of wheat in years when the weather has been favourable. Much of this wheat surplus will then be shipped off to poor countries as food aid.

Mr Nunn and Ms Qian thus find that annual wheat yields in the United States are positively related to the size of American food-aid shipments in the following year. By looking at records of how often individual developing countries have received food aid from America in the past, the researchers are able to identify instances where recipients have seen increases in food aid that likely can be attributed solely to favourable weather conditions (which have nothing to do with civil conflicts in developing countries).

The authors find that boosts in food aid, on average, increases both the incidence and the duration of civil conflict. The results suggest that a 10% increase in American food-aid shipments to an average recipient country increases the incidence of conflict by about 4%. The study does not reveal what the exact mechanism behind this negative effect is, but the authors speculate based on historical cases that armed groups steal the aid. During the Nigeria-Biafra civil war in the 1960s, for instance, rebel leader Odumegwu Ojukwu took the food aid to feed his soldiers, which many reckon helped prolong the conflict for years.

What these two papers on foreign aid and civil conflict illustrate is that aid policy has different effects on the recipient country. Some of them are unintentional and may therefore easily be overlooked. The point is that one cannot study just one outcome from aid (such as economic growth) and draw a general conclusion that aid “works” or “does not work”. Policymakers need to better consider a range of outcomes—both intended and unintended—when they decide on aid budgets, and they should ask themselves if the total benefits outweigh the aggregate costs. Generous inputs with a good purpose does not automatically lead to great outcomes.