No foreign policy seems more inherently benign than humanitarian military intervention. It is rooted in the altruistic desire to protect innocents from violent death. It appears feasible, given the military superiority of Western forces over those in developing countries where most violent conflict occurs. And the only obvious costs are a modest financial commitment and the occasional casualty. For these reasons, in the wake of the world’s failure to prevent violence in the Balkans and Rwanda, US President Bill Clinton declared in June 1999 the doctrine that bears his name: “If the world community has the power to stop it, we ought to stop genocide and ethnic cleansing.” In December 2001, a distinguished international panel went a step further and declared the existence of a “Responsibility to Protect”—suggesting that the failure to intervene by those capable of doing so might even breach international law.

But a more sophisticated analysis calls into question the value of humanitarian military intervention, even when judged by its own explicit standard of saving lives. For two reasons the benefits of such intervention are much smaller, and the costs much greater, than commonly recognized. First, most violence is perpetrated faster than interveners can realistically arrive to stop it. Second, as economists could have predicted, but few humanitarians have acknowledged, the intervention regime actually exacerbates some conflicts through what I have labeled “the moral hazard of humanitarian intervention.” In light of these two dynamics, an increase in intervention does not save as many lives as commonly claimed. Unless the West adopts a number of reforms outlined at the end of this article, more intervention might actually lead to a net increase in killing.

Killers are Quicker than Interveners

In the high-profile conflicts of the 1990s, most violence was perpetrated far more quickly than commonly realized. In Bosnia, although the conflict dragged on for more than three years, the majority of ethnic cleansing was perpetrated in the spring of 1992. By the time Western media arrived on the scene later that summer, Serb forces already occupied two-thirds of the republic and had displaced more than one million residents. In Rwanda, at least half of the eventual half-million Tutsi victims were killed in the first three weeks of genocide. When Croatia’s army broke a three-year cease-fire in August 1995, it ethnically cleansed virtually all of the more than 100,000 Serbs from the Krajina region in less than a week. In Kosovo, when Serbian forces switched from a policy of counter-insurgency to ethnic cleansing in March 1999, in response to NATO’s decision to bomb, most of their cleansing occurred in the first two weeks, and they managed to cleanse 850,000 Albanians, half the province’s total. In East Timor, following a 1999 vote for independence, Indonesian-backed militias damaged the majority of the province’s infrastructure and displaced most of the population in little more than a week.

Even less well recognized is the fact that logistical obstacles impose significant delays on military intervention, humanitarian or otherwise, even where a strong political will exists. For example, after Iraqi President Saddam Hussein invaded Kuwait in August 1990, and US President George Bush ordered an immediate deployment to defend Saudi Arabia (Operation Desert Shield), it took nine days for the first unit of 2,300 US troops to reach the area of conflict. Another week was required before the unit was sufficiently prepared to venture beyond its makeshift base. Thus, even with the vital national security interest of oil at stake, it took the United States more than two weeks to deploy and begin operations of a relatively tiny force. The reasons are numerous, but stem mainly from three factors: modern militaries cannot operate without their equipment, their equipment is extremely heavy, and there are limits to the rate at which such equipment can be airlifted to remote countries.

Indeed, the airlift to Saudi Arabia was much easier than a typical humanitarian intervention because the Arabian peninsula is closer to the United States and has better airfields than most conflict zones. Even short-distance interventions can be bedeviled by the combination of poor airfields and weighty forces. For example, when the United States deployed just 24 Apache helicopters from Germany to nearby Albania (Task Force Hawk) for the 1999 Kosovo intervention, it required 17 days. Two factors explain this: poor Albanian airfields and a US Army doctrine that mandated a force of 5,350 personnel and their equipment to operate, maintain, and protect the helicopters. In total, this task force weighed 22,000 tons, necessitating 500 flights of large, modern C-17 cargo aircraft.

Intervening in Africa, where most of today’s violent conflicts take place, is even harder due to bad airfields and the farther distance from western military bases. Had the United States tried to stop the Rwandan genocide, it would have required about six weeks to deploy a task force of 15,000 personnel and their equipment. This time estimate is conservative, because analogous past interventions to Haiti, Panama, and the Dominican Republic actually required somewhat larger forces. Unfortunately, this means that by the time Western governments learned of the Rwandan genocide and deployed an intervention force, the vast majority of the ultimate Tutsi victims would already have been killed.

The fact that much civil violence is carried out more quickly than intervention forces can arrive is by itself no excuse for failing to intervene; some lives can be saved even by belated intervention. However, it does mean that the benefits of humanitarian intervention are far smaller than commonly realized. This is important to remember as we turn to the unexpected costs of humanitarian intervention.

Intervention Exacerbates Violence

The most counterintuitive aspect of humanitarian military intervention is that it sometimes may cause the very tragedies it is intended to prevent. The explanation for this starts from the little known but empirically robust fact that genocidal violence is usually a state retaliation against substate groups for launching armed secession or revolution. Most groups are deterred from such armed challenges by the fear of state retaliation. In the 1990s, however, the regime of humanitarian military intervention changed this calculus, convincing some groups that the international community would intervene to protect them from retaliation, thereby encouraging armed rebellions. As events played out, these armed challenges did provoke genocidal retaliation, but intervention arrived too late to save many of the targets of retaliation. Thus, the intervention regime—intended to insure against risks of genocide and ethnic cleansing—inadvertently encouraged risk-taking behavior that exacerbated these atrocities. This is the classic dynamic of moral hazard, which is an inherent drawback of insurance systems.