Military Strategy or Poverty Reduction?: Investigating the Provincial Allocation of Aid to Afghanistan

Recent U.S. military interventions, relying on the coordination of military and humanitarian activities, have sparked an important debate over the appropriate use of development assistance.

November 27, 2012

Jake Douglas

Recent U.S. military interventions, relying on the coordination of military and humanitarian activities, have sparked an important debate over the appropriate use of development assistance. In Iraq and Afghanistan, many U.S. policymakers have viewed economic development as indispensable to U.S. combat efforts. Following the 2001 invasion of Afghanistan, then Secretary of State Colin Powell remarked, “I am serious about making sure we have the best relationship with the NGOs who are such a force multiplier for us, such an important part of our combat team.”


U.S. counterinsurgency (COIN) strategy explicitly calls for the harmonization of military and economic initiatives. The 2007 Army/Marine Corps Counterinsurgency Field Manual stresses the need for “maximum unity of effort” along all “logical lines of operations,” including reconstruction and economic development. Improving economic conditions is seen as crucial. It is believed that insurgents will “exploit a lack of employment or job opportunities to gain active and passive support for their cause.” The standard operating procedures manual for the Commanders’ Emergency Response Program (CERP)—an initiative that allows military commanders, at their discretion, to fund development projects in their areas of operational oversight—is unapologetically titled “Money as a Weapon System.”


Many NGOs and commentators deride what they call the securitization, militarization, or weaponization of aid. They highlight a loss in the neutrality and impartiality of humanitarian work (Macrae and Leader 2001; Shannon 2008; Terry 2011). They also point to the dangers of subordinating development goals to a security agenda (Howell and Lind 2009; Woods 2005). Others have sounded the alarm about the ineffectiveness of CERP-inspired projects. See, for example, Center for Global Development’s Rethinking US Foreign Assistance Blog, where Justin Sandefur posts on “Aid, WHAM, and Afghanistan.”


One of the most serious allegations about the militarization of aid is the formation of a so-called ‘hierarchy of victims,’ in which development activities in strategically important areas are funded at the expense of the strategically insignificant, but poorer areas. By assigning priority to military needs, the conventional wisdom is that securitized aid crowds out humanitarian and poverty reduction efforts.


I recently investigated whether the provincial distribution of U.S. aid in Afghanistan more closely aligns with military priorities or humanitarian need. Using Spearman’s R and Pearson’s R correlations, I find a significant, positive relationship between the provincial distribution of U.S. aid in Afghanistan and a province’s level of combatant violence. By contrast, I find some evidence of a negative correlation between the level of poverty by province and the allocation of aid.

The figures above depict the positive relationship between combatant violence and the allocation of aid by province. I operationalize combatant violence using a 2007 ISAF count of coalition fatalities by province and a 2008 provincial count of attacks by armed opposition groups against any target. The plots in the top row reflect significant, positive correlations between the log count of combatant fatalities and two different measures of per capita development assistance: (1) CERP and Provincial Reconstruction Team (PRT) spending, and (2) total government and donor spending. The lower plot shows a similarly significant relationship between the number of attacks and CERP and PRT spending.


So, there is some evidence of a significant, positive relationship between level of provincial violence and receipt of aid per capita in Afghanistan. By contrast, poverty—measured here as the inverse of household consumption, as reported in Afghanistan’s only comprehensive poverty profile, the 2007-08 National Risk and Vulnerability Assessment (NRVA)—is negatively correlated with total government and donor spending. When comparing provincial household consumption against the allocation of total government and donor spending, Spearman’s R and Pearson’s R correlations both indicate that this relationship is significant at the p=0.05 level. The negative association lends support to the claim that aid in Afghanistan is not distributed on the basis of development and poverty reduction objectives.


But there is an interesting twist: the provincial poverty rate is positively correlated with USAID spending in 2009, and this correlation is significant at the p=0.07 level. Thus, USAID seems to do a better job of targeting its provincial aid allocation according to economic and humanitarian needs—or at least they did  in 2009. This is an important complement to the more dire findings reported by Justin Sandefur and Danny Cutherell, who constructed an a measure of aid distribution by lumping together 2010 USAID and CERP/PRT aid flows.


The lack of reliable, sub-national data has meant that, with very few exceptions, only partial analyses of the spatial determinants of aid distribution are possible. Thus, researchers who study the geographic dispersion of aid can only draw tentative conclusions. This is especially true for studies on aid in combat zones like Iraq and Afghanistan. However, independent, spatial assessments of how donor agencies and military organizations target their scarce resources should become more feasible as aid become more transparent and the practice of geocoding aid projects becomes more prevalent. Interested readers should also keep their eyes peeled for the release of the 2010-11 NRVA, which is scheduled for the end of this year.

No items found.

Jake Douglas (‘14) is a former research assistant for the Making Reform Incentives Work project at the College of William and Mary. 

The views expressed here are those of the authors alone, and do not necessarily reflect the views of the institutions to which the authors belong.