The future of humanitarian assistance and security policy in chaotic places such as Syria and Iraq could rest on a single question: Does aid in conflict zones promote peace or war? It seems intuitive to assume that hunger and exposure push people to violence and that aid should, therefore, lead to peace. This idea has been the bedrock of scores of “hearts and minds” campaigns dating back to the Cold War, which have invested billions of dollars on the principle that assistance can buy compliance and, eventually, peace.

Yet recent evidence indicates that sending aid into conflict-affected regions can actually worsen violence in some cases. Over the past decade, our research collective, the Empirical Studies of Conflict Project (ESOC), has conducted a suite of studies in conflict zones to test this relationship. Among other countries, we studied the Philippines, a state riven by a variety of long-term conflicts in areas with limited governmental control. Our findings provide several lessons on how infusions of aid work in poorly governed spaces.

On the broadest level, we found that the type of the aid program matters greatly. Certain kinds of assistance—targeted, low-profile, conditional transfers delivered directly to needy families—appear to decrease conflict. By contrast, large and more high-profile projects, such as initiatives to improve infrastructure, may empower insurgents and exacerbate hostilities. These findings, we hope, can help inform an improved approach to aid that would both relieve suffering and help promote long-term stability.


The assumption that aid reduces conflict is grounded in intuitive notions of alleviating grievances, providing legitimate employment, and creating reciprocal gratitude. But recent research challenges these notions. The economists Nathan Nunn and Nancy Qian, for example, have analyzed cross-national data and concluded that food aid often fuels violence. The authors used changes in the size of U.S. food aid deliveries (which corresponded to fluctuations in U.S. wheat production levels) to measure the effect of such aid on violence, while controlling for a range of other influences. Their findings were stark: greater food aid appeared to raise both the incidence and the duration of civil conflicts.

One possible reason for this relationship is that food aid deliveries in contested areas gave insurgent groups something to steal and to fight over. Indeed, recent history offers ample examples of how radical groups can entrench themselves by capturing aid. (Hamas in the Gaza Strip, for instance, has been known to rebrand international aid as its own, and Shiite militias in Iraq did the same with aid funded by the United States.) Nunn and Qian’s findings applied only to ongoing hostilities in countries with a recent history of civil strife—meaning that food aid was not found to spark conflict where none existed—and to subnational violence, not wars between nations.

Although studies such as this make clear that the relationship between aid and conflict is not as straightforward as intuition might suggest, they are limited by their macro-level approach. Nunn and Qian, as well as numerous other researchers, have focused on differences among large-scale aid dumps, comparing experiences across countries. This approach can offer a useful overview, but it masks the true complexity of aid’s effects at the local level.

Our data from the Philippines allowed ESOC to take a micro-level view, studying variation in aid activity and violence across municipalities and over time, using precise information about perpetrators and victims. We were able to incorporate a wide range of variables into our analysis—we looked at both Communist and Islamist insurgencies, for example—as well as various types of aid, from infrastructure projects to conditional cash transfers to private investment. Our methodology varied as well, from observational analysis to randomized experiments. Together, these methods yielded four key lessons.  

First and perhaps most surprising, we found that creating jobs did not inhibit violence. A common belief is that gainfully employed men are less likely to participate in insurgencies because they’re too busy working or because the incomes they earn quell their discontent. In this view, the opportunity cost of violence is too high for those with jobs, and therefore donors should simply fund programs that generate employment opportunities.

However, our analysis of employment and violence in the Philippines (co-authored by Michael Callen) challenged this logic, pointing to the opposite conclusion: violence, in fact, increased with employment. A province that experienced larger than average year-to-year increases in employment typically also showed greater than average increases in violent incidents, as recorded by the Philippine military. This relationship was not unique to the Philippines; it paralleled the findings of our earlier research in Iraq and Afghanistan.

Although our data do not allow us to put a finger on the exact source of this correlation, a possible reason could be, ironically, that jobs help improve people’s lives. Local residents who see their fortunes rise could feel empowered to resist the government and its allies—for example, by refusing to sell information about insurgents—if such cooperation isn’t specified as a precondition of employment. But it’s also possible that the link between employment and violence roots elsewhere. It could be that governments seek to reduce violence by adopting tougher measures (running checkpoints, conducting sweeps) that damage the local economy and, hence, employment opportunities. More research is needed to inform employment-generating programs of the future. 

Our second finding is that increased private investment, too, can increase violence. Governments and nongovernmental organizations are not the only sources of economic activity in conflict zones. Local officials often encourage civilian investment in an attempt to deescalate hostilities. Yet these efforts sometimes do the opposite and play into the hands of insurgents—for instance, when new investment presents new opportunities for militants to extort money from local businesses.

We examined this kind of investment-fueled development in the Philippines (in a study coauthored by Ethan Kapstein and Erin Troland) by using a proxy variable for investment: detailed information on the value of industrial building permits. At the outset, investment in areas heavily affected by conflict was predictably low. But a subsequent influx of funds led to more violence, not less. For example, in 2004 a Japanese construction firm broke ground on a nine-figure project in Subic Bay, replacing a former U.S. naval base with a cargo port  and shipbuilding facility. That year, violent incidents spiked in the surrounding province, Zambales. Notably, most of these incidents were initiated by the government, as both police and military security forces cracked down on local militants. It’s possible that by adopting tougher measures, the government aimed to protect jobs and tax revenue that the infrastructure project held in store. In any case, this additional complication further shatters the expectation that investment can create quick harmony.

Third, aid in the form of public works programs can also increase violence. Community-driven development programs have become a popular form of development assistance around the world, including in conflict-ridden countries. These programs give communities block grants for infrastructure improvement projects that villagers themselves propose. The way in which the Philippine government implemented one such public works program—by using a specified poverty threshold to determine which communities were eligible for aid—allowed us to take a quasi-random approach in our research, comparing villages just over the mark with those just under it. Aside from either barely qualifying for aid or just missing the threshold, these communities showed no major differences. ESOC analyzed data on the villages’ participation in the program together with detailed information on the level of violence in the area and data on which groups suffered most and at whose hands.

The study, which one of us co-authored with Benjamin Crost and Patrick Johnston, provided strong evidence that public works programs exacerbated hostilities in communities that implemented them, significantly raising casualty rates. Notably, those casualties were concentrated among government forces, and most were recorded early in the aid program’s life, before the funds were even distributed. Violence was often unleashed by groups with strong ideological motivations (such as Islamist and Communist insurgents) rather than criminal gangs, which were also present. It appears likely, therefore, that in perpetrating violence, insurgent groups aimed to sabotage state-sponsored projects out of fear that their successful implementation would shift popular support in the government’s favor, undermining their own influence.

Our final conclusion was more hopeful: targeted, low-profile aid can decrease violence and weaken rebel influence. One form of assistance that appears somewhat effective in deescalating conflicts is conditional cash transfers: small payouts given directly to households in exchange for carrying out certain actions, such as vaccinating their children and keeping them in school.

ESOC researchers (including Crost and Johnston) were able to estimate the effect of this type of aid on Philippine communities by using data from a World Bank program that provided funds to randomly selected households across 130 villages in 2009–10. Villages that received conditional cash transfers saw a substantial reduction in the incidence of violence. Moreover, the influence of insurgent groups in these villages also decreased, according to the Philippine military’s assessments. The latter finding is significant because political insurgencies, such as the armed wing of the Philippine Communist Party, often take advantage of the governance vacuum in conflict zones by assuming public functions: levying taxes and providing basic services. A reduction in one such group’s influence may therefore signal a longer-term effect of development aid in strengthening government control than a mere drop in violent incidents would indicate, since a lull in hostilities could be temporary.


Although these outcomes may seem inconsistent, they fit a pattern suggested by our prior exploration of aid and rebellion in Iraq, which we laid out in a paper titled “Can Hearts and Minds Be Bought?” In that research, we created a model of insurgency as a contest with three actors: rebels seeking political change through violence, a government trying to minimize violence through some combination of service provision and hard counterinsurgency, and civilians deciding whether or not to share information about insurgents with government forces, guided by which side might best govern their neighborhood. 

When we tested that model using data on violence against U.S. and Iraqi forces in 2004–8, we found a pattern similar to the one in the Philippines. Increased development spending correlated with increased violence, on average. Yet when aid was conditioned on community characteristics that predicted violence—such as sectarian divisions and a history of prior conflict—small-scale assistance tended to reduce hostilities. Further research (with Erin Troland) found that this violence-reducing effect grew stronger when aid projects ran small in size, received more input from development experts during their planning stages, and were secured by sufficient government forces during implementation.

If the findings in Iraq and the Philippines offer one overarching message, it is this: aid, investment, and job creation are not magic bullets guaranteed to bring peace to conflict-affected spaces. Following the “do no harm” principle in delivering aid is a difficult task that requires a thoughtful and informed approach.

Aid programs applied bluntly or indiscriminately in the absence of security are likely to create perverse effects. But programs that are modest in scope and carefully planned—conducted in consultation with development specialists and local communities, coordinated with local security forces, and possibly conditioned on cooperation with those forces—can succeed in winning some community support. This support, even from a small fraction of the community, can be enough to ensure a steady flow of information about insurgents, which would in turn allow government forces to ensure the aid program’s secure implementation. In this sense, the objectives of development assistance and security complement each other.

The U.S. military has moved beyond the carpet-bombing techniques of World War II. Today it has smart missiles and drones designed to minimize collateral damage. The humanitarian side of counterinsurgency—aid programs and private investment—should become just as smart. Small, targeted projects will produce better outcomes than carpeting conflict regions with cash. Both the needs of the vulnerable populations and Washington’s strategic goals will be better served by a surgical approach to aid that takes into account the often counterintuitive truths about the economics of conflict.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now
  • ELI BERMAN is a Research Director at the University of California Institute on Global Conflict and Cooperation and Professor of Economics at the University of California, San Diego. JOSEPH H. FELTER is a Senior Research Scholar at the Center for International Security and Cooperation at Stanford University. JACOB N. SHAPIRO is Associate Professor of Politics and International Affairs at the Woodrow Wilson School of Public and International Affairs at Princeton University.
  • More By Eli Berman
  • More By Joseph H. Felter
  • More By Jacob N. Shapiro