Many governments in very low-income countries struggle with insurgency. One means employed to reduce political violence has been to direct resources toward providing employment opportunities for potential insurgent recruits, based on the conjecture that fully employed young men are less likely to participate in insurgent violence.1This investigation examines the relationship between unemployment and insurgent violence, as a means of testing that conjecture. The results also bear on the larger question of how development can best proceed in areas afflicted by violence and insecurity, both of people and of property.
While the theory noted above is commonly accepted by policymakers, there are other competing viewpoints. One model2 suggests that the greater the economic gains associated with controlling an area, the greater the effort rebels will invest in violent capture.3 The means by which violence is suppressed may also reduce employment rates, if they impede flows of labor, goods or resources. On the other hand, the “hearts and minds” approach states that the key predictor of violence is attitudes of the population toward both government and insurgents. Those attitudes may be affected by employment rates, among other factors.4
In this study, the researchers consider all three theories in their attempt to identify the conditions that incentivize political violence in Afghanistan, Iraq, and the Philippines. The authors note, however, that the insurgency in Iraq was substantially more intense than in Afghanistan or the Philippines. Even when measured in incidents per thousand, the civilian casualty rate was higher in Iraq than the Philippines. Furthermore, provinces in the Philippines are larger than districts in Iraq and much larger than those in Afghanistan, suggesting that estimates for Iraq and Afghanistan may be more precise.
The primary goal of this study is to test whether violence increases as unemployment increases in areas with active insurgencies. Using survey data on unemployment and two newly available measures of insurgency — rates of attacks per capita against government and allied forces, and violence leading to civilian deaths — the authors test the dominant theory that unemployment drives insurgent violence.
The researchers use a variety of data for each country which capture the date, location, and intensity of insurgent activity, measured as the rate of attacks per capita against government forces and their allies. This study also examines whether civilians were killed during insurgent attacks against coalition or government forces. The lack of detailed unemployment data in all three countries limited the strength of analysis. To address this limitation, researchers focused their analysis on periods for which local unemployment data were available. For Afghanistan, researchers used five waves of household surveys from September 2008 through September 2009. Three household surveys captured unemployment data at the district level in Iraq in 2004 and 2005. In the Philippines, researchers used provincial-level unemployment rates captured by labor force surveys from 1997-2003 and 2006.
Results and Policy Implications
For all three countries, increases in employment were found to predict increased violence, even after controlling for a wide range of possible factors using time and space fixed effects. For both Iraq and the Philippines, the authors rejected the hypothesis that violence increases as unemployment increases, although the results are more significant in the Philippines. Once researchers included controls for all fixed effects, they also rejected this hypothesis for Afghanistan, regardless of the measure of violence used.
The authors offer three theories to explain their findings. First, insurgent violence may increase in economically advantaged periods and areas because of predation: the return to violence in those areas increases. Second, government security efforts such as blockades may prevent the transportation of goods and services, suppressing economic activity and insurgent violence. And finally, government and allied forces might rely on civilian intelligence about insurgents to limit violence, implying that counterinsurgents may operate more effectively in areas with high unemployment because the cost of information is lower.
These findings do not necessarily refute the conjecture that fully employed males are less likely to engage in political violence, but it does imply that this theory cannot be the dominant mechanism in play. Neither does the paper refute a “hearts and minds” approach to understanding insurgency, though it does suggest that short term changes in employment levels are not the major driver of attitudes in that mechanism.
The authors’ findings offer several pertinent policy implications. First, a negative correlation between unemployment and violence suggests that a strategy of emphasizing short-term job creation may not be effective. Second, the findings suggest that policymakers need a better understanding of how, when, and where aid spending helps to reduce political violence prior to developing and implementing programs.
Analysis of data for the period 1997-2009
1 General Peter Chiarelli, U.S. Army Commander of Multinational Forces in Iraq, in a Dec. 8, 2006, press briefing.
2 Fearon, James (2008) “Economic Development, and Civil War.” Institutions and Economic Performance, ed. E. Helpman. Cambridge, MA: Harvard University Press.
3 Dube and Vargas (2008) find that violence increases in oil-rich areas of rural Colombia when the price of oil increases.
4 This view is largely based on the British experience in Malaya, the French in Algeria, and the U.S. in Vietnam; Trinquier (1961), Taber (1965), Galula (1964), Clutterbuck (1966), Thompson (1966), Kitson (1971), and Popkin (1979).
Photo Credit: Eli Berman