Are airstrikes an effective tool against insurgent organizations?
Despite the question’s historical and contemporary relevance, we have few dedicated studies, and even less consensus, about airpower’s effectiveness in counterinsurgency wars. To answer this question, we draw on declassified United States Air Force records of nearly 23,000 airstrikes and non-lethal shows of force in Afghanistan (2006-11), satellite imagery, and a new SQL-enabled form of dynamic matching to estimate the causal effects of airstrikes on insurgent attacks over variable temporal and spatial windows. Evidence consistently indicates that airstrikes markedly increase insurgent attacks relative to non-bombed locations for at least 90 days after a strike. Civilian casualties play little role in explaining post-strike insurgent responses, however. Instead, these attacks appear driven by reputational concerns, as insurgent organizations step up their violence after air operations to maintain their reputations for resolve in the eyes of local populations.
Follow-on studies will investigate airpower’s effects in contemporary (Iraq, 2003-09) and historical cases.
“Bombing to Lose? Airpower and the Dynamics of Violence in Counterinsurgency Wars.” pdf.
PhotoCredit: Emin Menguarslan/Anadolu Agency/Getty Images
Can humanitarian assistance reduce popular support for insurgents?
This project investigates whether humanitarian assistance can reduce support for insurgents among populations that have been harmed by direct or indirect counterinsurgent actions. We use a “nested” natural experiment approach to assessing how the Afghan Civilian Assistance Program II (ACAP II), a USAID-funded non-monetary assistance program, has affected harmed individuals’ mental resiliency, economic prospects, and support for combatants in Afghanistan (2010-12). We draw on individual level survey experiments among 2,700 respondents as well as patterns of violence in over 900 villages to assess program effectiveness.
How do we explain variation in battlefield performance in conventional war?
This project investigates the sources of battlefield performance in conventional wars (including civil wars) fought between 1800 and 2011. A new dataset of conventional wars and combatants — the product of over 80 coders working in 21 languages over 6 years — forms the empirical backbone of these efforts. New data on battlefield practices, including desertion and defection among soldiers, is also marshalled to demonstrate that the sources of wartime military effectiveness lie in a state’s prewar identity type. More specifically, the nature and severity of exclusion directed toward internal populations by regimes seeking to legitimate their domestic rule helps condition a state’s military performance even before the first battle is joined. A mixture of crossnational data, natural experiments, and paired historical cases are used to test this argument against alternative explanations.
“Why Armies Break: Explaining Mass Desertion in Conventional War” pdf.
We are currently engaged in a large-scale data collection effort designed to track patterns of violence against civilians in Afghanistan over time (2008-15). Drawing on multiple data sources, we aim to construct a village-level dataset recording by both the Taliban and the International Security Assistance Force (ISAF). We are interested not only in explaining patterns of civilian victimization but also assisting the targeting of economic and other assistance to the neediest areas in Afghanistan.
“A (fighting) season to remember in Afghanistan,” The Monkey Cage (20 October 2014), here
We draw on multiple survey experiments to measure civilian attitudes toward combatants in Afghanistan. Taken together, these projects have investigated four different drivers of these attitudes: (1) coethnicity; (2) exposure to violence; (3) receipt of economic assistance; and (4) sudden security “shocks” in the form of ISAF base closures. The FieldLab is currently invested in efforts to demonstrate how survey data can be used to predict survey violence and how panel data can be used to track changes over time in Afghanistan.
“Can Civilian Attitudes Predict Civil War Violence?” With Kosuke Imai and Kentaro Hirose. pdf.
“Coethnic Bias and Wartime Informing.” With Kosuke Imai and Yuki Shiraito. pdf.
“Comparing and Combining List and Endorsement Experiments: Evidence from Afghanistan.” American Journal of Political Science, 58:4 (October 2014), 1043-1063. With Graeme Blair and Kosuke Imai. pdf
“Explaining Support for Combatants in Wartime: A Survey Experiment in Afghanistan.” American Political Science Review, 107:4 (November 2013). With Graeme Blair and Kosuke Imai. pdf.
Is there a connection between unemployment and insurgent violence? We draw on new data from USAID’s Community Development Program (CDP), a large-scale employment program in Afghanistan, to examine whether economic assistance can decrease insurgent violence by reducing unemployment. The CDP was implementated in 2010-12 across many of the most violent districts in Afghanistan, creating a difficult test for assessing the effectiveness of the program. We implement a new form of dynamic spatial point process modeling to estimate the effects of the program over space and time. We are especially interested in measuring spillover from work sites into other neighboring villages and areas.