After the Rain: Rainfall Variability, Hydro-Meteorological Disasters, and Social Conflict in Africa Cullen Hendrix and Idean Salehyan University of North Texas Climate Change and Security Conference, Trondheim, NO 21-24 June 2010
Is there a link?
Climate change and conflict Scholars, policymakers are concerned that climate change, environmental degradation will lead to conflict Weak and ambiguous findings with respect to interstate, intrastate war Other forms of social conflict may be more likely Interstate wars are rare Insurgency entails significant start-up costs, government is (comparatively) well-armed The state may not be an appropriate target Interstate war Intrastate war Protests, rioting, communal conflict, government repression, strikes
Different in kind, not necessarily scale Kenya 2007: marred elections lead to widespread ethnic rioting, 800-1,500 deaths, political compromise Nigeria 1998: Ife-Modakeke ethnic violence kills 3,000 near Osun 2001-2008: Muslim-Christian religious violence kills >1,200 in Jos Zambia Protests in the early 1990s spur multiparty elections Event death totals are higher than many civil conflicts Peter Andrew/Reuters
Theory and causal mechanisms Conflict between consumers of water Water shortage/extr eme abundance Price disputes between rural producers and urban consumers Migration out of affected areas Reduction in state revenues and patronage opportunities Political unrest Poor macroeconomic outcomes
New data Social Conflict in Africa Database (SCAD) Covers 47 African countries, 1990-2009 Over 6,000 events Protests, riots, strikes, government repression, communal violence, intra-government violence Augments existing data sources on inter-, intrastate war Online database, visualization tools expected August 2010, hosted by Strauss Center, UT-Austin
Data generation Searches of AP and AFP (1990-2009) Demonstrations, riots, strikes, repression, anti-government violence, extra- government and intra-government violence Contains information on: Start and end dates Escalation Actors and targets Repression Number of participants Number of deaths Location Issue(s)
Social conflict indicators for 47 African countries, 1990-2009 Variable Obs. Mean σ Min. Max. Civil Conflict Onset 843 0.05 0.22 0 1 Total Events 901 6.91 11.55 0 95 Nonviolent Events 901 3.65 5.53 0 42 Violent Events 901 3.26 7.06 0 67 Government- Targeted Events Nongovernment- Targeted Events 901 3.58 5.53 0 40 901 3.32 7.43 0 74
Measures of water scarcity/abundance Different measure of rainfall shocks Deviations from long-term panel means (1979-2008), standardized by the standard deviation of panel mean Mean = 0, σ = 1 A growth year after a decline year could still be a net - year A decline year after a growth year could still be a net + Estimating impact of hydrometeorological disasters: floods and droughts Counts based on CRED data
Estimation Civil conflict onset: logit with clustered errors, controls for temporal dependence Events: negative binomial with fixed effects, period dummies, time trend Independent Variables Standardized Rainfall Deviations (linear and squared, present and lagged), Droughts (count), Floods (count) Control Variables Lagged DV, Polity2, Polity2 2, GDP per capita and GDP Growth, Population, Population Growth, Conflict
Rainfall Rainfall and civil conflict onset, 1990-2007 +3 SD +2 SD +1 SD Mean -1 SD -2 SD -3 SD -63% -49% -30% +40% +101% +190% 0 0,02 0,04 0,06 0,08 0,1 pr(conflict onset), controls at means
-3,5-3,3-3,1-2,9-2,7-2,5-2,3-2,1-1,9-1,7-1,5-1,3-1,1-0,9-0,7-0,5-0,3-0,1 0,1 0,3 0,5 0,7 0,9 1,1 1,3 1,5 1,7 1,9 2,1 2,3 2,5 2,7 2,9 3,1 3,3 3,5 % Change in Number of Events Rainfall and social conflict events, 1990-2007 300 250 200 Estimated Effect 95% CI 150 100 50 0 Standardized Rainfall Deviation
Rainfall Rainfall and social conflict events, by type, 1990-2007 +2 SD +1 SD -1 SD -2 SD Nongovernment-Targeted Events Government-Targeted Events Violent Events Nonviolent Events Total Events 0 % 10 % 20 % 30 % 40 % 50 % 60 % Percent Increase in Events from Country Mean
Conclusions Civil conflict more likely in wetter years Strong, curvilinear effect of rainfall on social conflict events Violent, nongovernment-targeted conflict more responsive to rainfall abundance Nonviolent, government-targeted conflict more abundant rainfall scarcity No effects for droughts, floods Reporting bias? Politics? Bergholt and Lujala measure probably better Extensions Interactions with political, economic, sociodemographic factors Explaining the strength of these correlations in the cross-section Correlation Strong (r > 0. 45) Moderate (0.45 > r > 0. 2) Negative (r < -0. 2)
Supplemental slides
Disaggregate? Growing body of work which disaggregates conflict Do local disasters foster local conflicts? However effects may be geographically distant Rural drought affects urban food prices Migration Effects on public employees No reason to expect conflict to be confined to areas with water shortages