Space-Time Granger Analysis of the War in Iraq: A Study of Coalition and Insurgent Action-Reaction

Abstract

We investigate insurgent-coalition interaction using the WikiLeaks dataset of Iraq war logs 2004–2009. After a review of existing theoretical interventions on the dynamics of insurgency and presenting a baseline model of violent events, we test a conceptual model of reciprocity using an innovative space-time Granger causality technique. Our estimation procedure retains predicted probabilities of reaction in response to a previous opponent's action across different temporal and spatial configurations in Iraq and in Baghdad. Our conclusions about conflict in Iraq are based on these profiles of risk—what we call space-time signatures. We find strong evidence of “tit-for-tat” associations between coalition/Iraq forces on one side and insurgents/militants on the other. Specifically, we find that the action-reaction association varies strongly by majority ethnic region across Iraq and in Baghdad, by urban and nonurban location, and within Sunni-dominated areas, by district income. While violence is strongly temporally dependent in the same location, the effect of distance varies significantly across the different subsets of the Iraq data.

Keywords:

• Baghdad
• Granger causality
• Iraq
• spatial analysis
• WikiLeaks war logs

Acknowledgments

We thank Nils Weidmann and Idean Salehyan for sharing their spatial representation of ethno-sectarian populations among Baghdad's neighborhoods, Nancy Thorwardson for preparing the publication graphics from our drafts, and our anonymous reviewers for helpful suggestions. Any errors are ours. Replication code is available on the International Interactions dataverse page at http://dvn.iq.harvard.edu/dvn/dv/internationalinteractions and at http://www.colorado.edu/ibs/johno/iraq/. Authors are graduate research assistant, post-doctoral research associate, and professor of distinction in Geography, respectively.

Notes

¹Following Williams and Simpson (2008, 194), we use “ethno-sectarian” to characterize Sunni, Shi'a, and Kurdish communities, fully acknowledging the complexity of debates surrounding how “ethnicity” is defined.

²Repression can be associated with lower levels of violence, as Lyall (2009) has found for the aftermath of targeted bombing campaigns by Russian forces in Chechnya (see also Bueno de Mesquita 2005).

³ http://www.gadm.org/

FIGURE 1 Violent events per 1,000-people district in Iraq, 2004–2009.

⁴ http://www.iauiraq.org/reports/WFP%20Baseline%20Food%20Security%20in%20Iraq%20%20Sept%202004.pdf

⁵ http://www.ornl.gov/sci/landscan/

⁶ http://gulf2000.columbia.edu/

⁷ http://www.iauiraq.org/reports/WFP%20Baseline%20Food%20Security%20in%20Iraq%20%20Sept%202004.pdf

⁸Image and data processing were conducted by NOAA's National Geophysical Data Center. DMSP data were collected by US Air Force Weather Agency.

⁹There were foreign fighters in the Iraq insurgency. Dramatic tactical and social rifts between domestic Sunni and international Sunni insurgents existed, however, as in Fallujah during 2004 (Mocktaitis 2008:127). The foreign contingencies of the insurgency were also largely jihadist.

¹⁰As in Goldstein (1991:196), Rajmaira and Ward (1990:463), and in the work of others developing upon Richardson (1960).

¹¹In the traditional model, it is the opponent's action at t. With a time aggregation of three days, we are wary of endogenous effects within the same period and prefer a minimum lag of one period.