Subscribe to E-Bulletin Donate to CEGA

Can Hearts and Minds Be Bought? Public Service Spending and Insurgent Violence in Iraq

Development Challenge

Conflict periods are often characterized by cycles of economic, social, and institutional instability. As such, they can destroy infrastructure and disrupt a government’s capacity to provide essential public services to its citizens, undermining economic and political development. Reconstruction is complicated by the increasing irregularity of political violence, which is often characterized by insurgent strikes rather than conventional state-on-state warfare. An important question for foreign policy and conflict management is how best to counteract insurgent groups while providing basic governance to conflict areas. Spending on reconstruction projects is one method employed to deliver aid in conflict zones. However, few studies have evaluated the cost-effectiveness of reconstruction spending on improving public service provision and stability. 


The setting for the evaluation is the 2003-2011 U.S. invasion and occupation of Iraq (commonly known as the Iraq War). During the conflict, Coalition Forces (the U.S. and allied countries) joined the Iraqi government to engage in counterinsurgent attacks against rebel forces and to allocate reconstruction funds. From 2003 to 2007, the U.S. spent a total of about $30 billion on reconstruction programs. This research evaluates the effects of the Commander’s Emergency Response Program (CERP), a $2.9 billion fund used by the U.S. military to implement small-scale, local reconstruction projects. The research also analyzes the effectiveness of an increase in troop strength in 2007 (the “surge”) by 30,000 combat troops, and a change in military strategy that emphasized greater engagement with local leaders. This setting provides an opportunity to evaluate the effects of a shift in military strategy on social and economic outcomes. 

Evaluation Strategy

The authors’ primary research question was whether CERP spending reduced insurgent violence, defined as attacks per capita against Coalition and Iraqi government forces. The authors modeled the conflict, using a game theoretic model, as a contest consisting of three players: the government, the rebels, and civilians. In the game, the government wants to reduce violence through counterinsurgency measures and spending for public services; the rebels want to impose costs on the government through insurgent attacks; and civilians possess information about the rebels, which is valuable to the government’s counterinsurgency strategy. In the game, the key to the government’s success is acquisition of information from civilians to more effectively deploy counterinsurgent measures.

To evaluate the effectiveness of spending on reducing violence, the authors combined three panel datasets: (i) incidents of violence against Coalition and Iraqi government forces, (ii) reconstruction spending at the project level, and (iii) district-level community characteristics. Data on oil reserves and infrastructure measures were also used. The researchers analyzed all 104 districts in Iraq over ten six-month periods between January 2004 and December 2008, resulting in a total sample size of 1,040 observations. They test three main hypotheses in their analysis:

(1)    Conditional on local characteristics, CERP spending reduces violence. The researchers utilized panel data to estimate the effect of a change in CERP spending on a change in violence using the “First-Differences” method. Assuming that pre-existing trends do not change over time, multiple time periods should eliminate any omitted characteristics that might bias the results.

(2)    The violence-reducing effect of reconstruction spending is greater when the government has better information on community needs and preferences. To test this second hypothesis, the authors employed an interrupted time series analysis to determine the effect of CERP spending on violence in both the pre-surge (2004-2006) and post-surge periods (2007-2008), during which deployment involved more contact with community members. The difference in the effect of a change in CERP spending in both the pre-surge and post-surge periods should determine the effect of the surge (more troops and more community engagement) on spending effectiveness.

(3)   Across communities, the same characteristics that predict violence also predict CERP spending. The authors compare the magnitude and significance of the determinants of violence, such as Sunni vote share, with the determinants of CERP spending, to test both the strength and direction of any correlation.

Results and Policy Implications

The study results suggest that, conditional on community characteristics, CERP spending is effective in reducing violence, particularly in the post-surge period. After the surge, each additional $100 dollars in per capita CERP spending predicted 159 fewer violent incidents per 100,000 residents per half-year (the average was 58.6 incidents per 100,000 people), suggesting that this may be a cost-effective violence-reduction program. Furthermore, results show that smaller projects implemented by teams who had local knowledge were five times more effective than were larger projects at reducing violence. Finally, the authors found the same characteristics that predict violence also predict government spending on services, supporting the allocation of CERP spending to areas that are more predictably violent. These results provide preliminary evidence that public service provision during periods of conflict can reduce attacks on military forces. More studies in other settings will be necessary to infer whether these constructive results were due to characteristics of the Iraqi setting (i.e., a reasonably stable governing structure), or are generalizable to other conflicts.


Analysis of data for the period 2004-2008