Does foreign aid have an impact on violence — on wars, on street fights, on random attacks?

The answer is that yes, it does — in two opposite, and perhaps counterintuitive, ways. On the one hand, aid can provide jobs and resources. And that, in turn, can reduce the incentives for people to engage in violent actions.

Yet it can have the reverse effect as well. “Aid can also increase conflict by introducing something to fight over,” says Austin Wright, a data scientist at the University of Chicago who works at the intersection of public policy and statistics. He’s referring to resources like roads and supplies paid for by the foreign assistance. In other words, “things that are of value to control.”

The termination of USAID — America’s premiere aid agency — gave researchers another angle to explore. Does the sudden withdrawal of aid funding have an impact on conflict?

In a study published in the journal Science, Wright and his colleagues conclude that the abrupt dismantling of USAID led to an uptick in overall conflict in places within Africa that have received aid compared to those that have not.

“The rapid collapse of what is probably the most sophisticated humanitarian assistance program in human history had enormous consequences on the ground, undermining livelihoods and therefore leading to a surge in violence,” Wright concludes.

The near instantaneous evaporation of assistance “took away the livelihoods, it undermined economic productivity,” he explains, thereby weakening the incentives that people might have to refrain from violence. And at the same time, “it did not yet eliminate what the actors were fighting over. And so this is what creates the chaos and the violence that we end up observing.”

By way of example, Wright refers to protests that broke out at the Kakuma refugee camp in northwest Kenya in July 2025. The roughly 300,000 refugees there depended on food and other services paid for by USAID. Wright says, “After the cuts, food distributions were sharply reduced, and refugees took to the streets,” throwing rocks and setting things on fire. One person was killed. “It is exactly the kind of incident our results capture.”

NPR reached out to the U.S. State Department for comment. Spokesperson Tommy Pigott replied in part, “One of the biggest problems with this ‘report’ is that it fundamentally ignores what is actually happening in Africa. The Trump administration has made unprecedented progress towards the advancement of peace on the continent. Unlike the previous administration … the Trump administration has reinvigorated our assistance programs to focus on efficiency, effectiveness, and partnership.”

The interplay of aid and conflict

The researchers considered a map of USAID funds disbursed at the state or provincial level preceding the termination of the agency.

“And then we overlay that with conflict activity,” he says, in the form of armed clashes, protests, riots and violence against civilians. These data from the ten months before and after early 2025 came from a detailed independent database of violent conflict called the Armed Conflict Location and Event Data. “So we don’t just see that a conflict occurred in a place,” Wright says. “We see exactly when the conflict occurred.”

And they looked to see how any change in that violent activity may have corresponded to the elimination of USAID investments.

The team found that places that once received more assistance tended to experience more conflict once that aid abruptly disappeared, often because those who had relied on that aid had become more vulnerable to recruitment by armed groups. “When aid is withdrawn suddenly, the economic opportunities disappear fast: wages dry up, clinics close, food programs stop,” explains Wright.

“But the things worth fighting over (infrastructure, territory, political power, ethnic grievances, geopolitical tensions) don’t vanish overnight,” he continues. That meant that in the short term, loss of opportunities, alongside an animation of the individual motivations to fight, caused violence to increase.

In addition, the conflict included combat between armed groups, public demonstrations that spontaneously turned hostile, and deliberate acts of violence that targeted noncombatants — such as rebels attacking a village. 

Wright thinks of the conflict arising in environments that are low on economic opportunity but high on grievance. “This is an arrow that points directly at the timing of the shutdown,” he says.

The findings are “convincing”

The one exception that he and his colleagues found was in places with government that support stronger constraints on the executive leader. These are “settings where the president or equivalent actor cannot unilaterally declare war [or] bypass elected congressional bodies,” says Wright.

Such institutions helped their constituents weather the storm of the sudden funding withdrawal. This meant that there was less of an impact on conflict there. Wright cites Nigeria’s $200 million supplementary health budget and South Africa’s decision to help cover gaps in AIDS and HIV treatment as two examples.

Researchers not involved in the study caution this is a challenging topic to study. “Conflict and the sources of conflict — it’s very complex,” says Andy Solow, a statistician at the Woods Hole Oceanographic Institution who wasn’t involved in the study. “I think it’s just generally speaking difficult to get a definitive answer. But you do the best you can, and I think that they did.”

Solow did raise a couple concerns with the analysis, including that conflict can be contagious, allowing it to spread. This means that perhaps the individual violent events that the researchers were taking note of in their analysis may have been interrelated.

“Those are technical issues,” he says. “But they’re not likely to overturn the basic result, which is that the cutoff led to an increase in conflict. I believe their results. They’re convincing.”

As the authors point out, the increase in violence may well be the legacy of the decision to terminate USAID suddenly. It’s a reality that Wright says is concerning since “recent conflict is the single best predictor of future conflict. Once violence escalates, it tends to be self-reinforcing.”

That means that even if the aid were to be resurrected, the situation is unlikely to improve as suddenly as it unraveled. That is, says Wright, “the damage from this period of increased violence would not simply undo itself.”