CLIMATE CHANGE AND CIVIL WARToni Sipic, Mikesell Lab, University of Oregon

Introduction

• IPCC identified climate change as a threat to social stability

• Climate Change– Change in average temperature and

precipitation patterns– Change in the number and severity of climate

change related disasters• Goal

– Identify and measure the potential effects of climate change that lead to violent conflict with a goal of informing policy

Introduction

Climate Change and Civil War Civil war measures

Onset or start Incidence or occurrence Duration

Climate change related disasters Income shock channel Such events decrease the opportunity costs of

using violent conflict as a method to assert power over allocation of scarce resources

Other channels

Literature Overview

• Civil War Literature (Unrelated to Climate Change)– Blattman and Christopher (2010) Literature

Review– Fearon and Laitin (2003) Civil War Onset– Fearon (2004) Civil War Duration

• Average Temperature and Precipitation Effects on Civil War– Hendrix and Glaser (2007) average precipitation

in Africa – Burke et al. (2009) average precipitation and

temperature in Africa– Zhang et al. (2006) Low average temperatures

in China– Tol and Wagner (2010) Low average

temperatures in Europe– Miguel et al. (2004)

Literature Overview

Extreme Weather Effects on Civil War Nel and Righarts (2008) single index of

disasters on civil war onset Besley et al. (2009) theoretical model

for the civil war incidence, some empirics

No research on the impacts of climate change on civil war duration

Contribution

First study to look at the differentiated effects of variety of climate change related disasters on various measures of civil war

Quantifying the costs of climate change As such it can inform climate change

mitigation and adaptation policies

Data

Civil War Fearon and Laitin (2003) Civil War Onset

161 countries, 1945-1999 Fearon (2004) Civil War Duration

73 countries, 1945-1999 Natural Disasters

Centre for Research on the Epidemiology of Disasters (CRED) Emergency Events Database (EM-DAT), a global database on natural and technological disasters, 1900 – Present

Droughts, epidemics, extreme cold temperatures, extreme hot temperatures, floods, storms and wildfires

Other controls from a variety of sources: World Development Indicators, Quality of Government

Summary statistics: Onset and Incidence

Summary statistics: Onset and Incidence

Summary statistics: Duration

Econometric Specifications: Onset and Incidence

Panel logit specification Random effects logit Conditional (Fixed Effects) Logit

Significant loss of observations Fixed-Effects Linear Probability Model

Econometric Specification: Onset• Onset

• Disasters are: droughts, epidemics, extreme cold temperatures, extreme hot temperatures, floods, storms and wildfires

• Occurrence of civil war in the previous year, income per capita lagged one year, log(population density), log(percent of country’s terrain classified as “mountainous”), states with noncontiguous territories, oil-producing country, newly formed state, Polity IV, anocracy

itit XDisasterOnsetCivil_War_ :O1 Model it10it

ittit

s

XDisasterOnsetCivil_War_ :O8 Model3

0s-it0it

Results

Since the complete set of results is so voluminous, I will focus on just the key estimated coefficients For each 7 disaster types (and the joint

model) there are 8 regressions, and 3 types of specifications, for a total of 192 regressions

Only the significant coefficients on the disaster variables are summarized in these tables

Individual Disaster Events Regression Results: Onset

Variables Significant Coefficients Models Significant Coefficients ModelsDroughtt-1 0.589* O3 -- --Extreme Coldt 1.226** to 1.438*** O1,O2,O3,O4 1.209*** to 1.381*** O5,O6,O7,O8Extreme Heatt-1 1.098* to 1.427** O2,O3,O4 1.273* to 1.581** O6,O8Extreme Heatt-2 1.091* O3 -- --Extreme Heatt-3 1.102* O4 1.135* O8Epidemict 0.299* to 0.380** O1,O2,O3,O4 0.346* to 0.379** O5,O6,O7,O8Stormt-2 0.235* to 0.316** O3,O4 0.353* O8Stormt-3 -- -- -0.511* O8Wildfiret 1.125** to 1.247*** O1,O2,O3,O4 1.133** to 1.234** O5,O6,O7,O8Control VariablesTime Fixed EffectsNote: Summary of selected coefficients from Appendix A Tables A1 through A7. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Table 7. Determinants of Civil War Onset : Significant Coefficients from Individual Event Panel Logit Models, 1945-1999 (n=6278)

Yes YesNo Yes

Individual Disaster Events Regression Results: Onset

Variables Significant Coefficients Models Significant Coefficients ModelsDroughtt-1 0.589* O3 -- --Extreme Coldt 1.226** to 1.438*** O1,O2,O3,O4 1.209*** to 1.381*** O5,O6,O7,O8Extreme Heatt-1 1.098* to 1.427** O2,O3,O4 1.273* to 1.581** O6,O8Extreme Heatt-2 1.091* O3 -- --Extreme Heatt-3 1.102* O4 1.135* O8Epidemict 0.299* to 0.380** O1,O2,O3,O4 0.346* to 0.379** O5,O6,O7,O8Stormt-2 0.235* to 0.316** O3,O4 0.353* O8Stormt-3 -- -- -0.511* O8Wildfiret 1.125** to 1.247*** O1,O2,O3,O4 1.133** to 1.234** O5,O6,O7,O8Control VariablesTime Fixed EffectsNote: Summary of selected coefficients from Appendix A Tables A1 through A7. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Table 7. Determinants of Civil War Onset : Significant Coefficients from Individual Event Panel Logit Models, 1945-1999 (n=6278)

Yes YesNo Yes

Individual Disaster Events Regression Results: Onset

Variables Significant Coefficients Models Significant Coefficients ModelsDroughtt-1 0.589* O3 -- --Extreme Coldt 1.226** to 1.438*** O1,O2,O3,O4 1.209*** to 1.381*** O5,O6,O7,O8Extreme Heatt-1 1.098* to 1.427** O2,O3,O4 1.273* to 1.581** O6,O8Extreme Heatt-2 1.091* O3 -- --Extreme Heatt-3 1.102* O4 1.135* O8Epidemict 0.299* to 0.380** O1,O2,O3,O4 0.346* to 0.379** O5,O6,O7,O8Stormt-2 0.235* to 0.316** O3,O4 0.353* O8Stormt-3 -- -- -0.511* O8Wildfiret 1.125** to 1.247*** O1,O2,O3,O4 1.133** to 1.234** O5,O6,O7,O8Control VariablesTime Fixed EffectsNote: Summary of selected coefficients from Appendix A Tables A1 through A7. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Table 7. Determinants of Civil War Onset : Significant Coefficients from Individual Event Panel Logit Models, 1945-1999 (n=6278)

Yes YesNo Yes

Many Disaster Events Regression Results: Onset

Variables Significant Coefficients Models Significant Coefficients ModelsDroughtt-2 -- -- -0.862* O7Extreme Coldt 0.913* to 1.205*** O1,O2,O4 0.949* to 1.226*** O5,O6,O8Extreme Heatt-1 -- -- 1.400* O8Epidemict 0.300* O1 0.345* to 0.391* O5,O6,O7,O8Stormt-2 0.358** O4 0.417** O8Stormt-3 -0.478* O4 -0.624** O8Control VariablesTime Fixed Effects

Table 8. Determinants of Civil War Onset : Significant Coefficients from All Events Panel Logit Models, 1945-1999 (n=6278)

Yes YesNo Yes

Note: Summary of selected coefficients from Appendix A Table A8. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Econometric Specification: Incidence• Incidence

• Disasters are: droughts, epidemics, extreme cold temperatures, extreme hot temperatures, floods, storms and wildfires

• Income per capita lagged one year, log(population density), log(percent of country’s terrain classified as “mountainous”), states with noncontiguous territories, oil-producing country, newly formed state, Polity IV, anocracy

itit XDisastersin_ProgresCivil_War_:P1 Model it10it

ittit

s

XDisastersin_ProgresCivil_War_:P8 Model3

0s-it0it

Individual Disaster Events Regression Results: Incidence

Variables Significant Coefficients Models Significant Coefficients ModelsExtreme Coldt 0.94** to 1.159*** P1,P2,P3,P4 -- --Extreme Heatt-1 1.008* to 1.029* P2,P3,P4 1.003* to 1.053* P6,P7,P8Extreme Heatt-2 1.526*** to 1.543*** P3,P4 1.511*** to 1.564*** P7,P8Extreme Heatt-3 1.265** P4 1.193*** P8Floodt 0.152* to 0.218*** P1,P2,P3 -- --Floodt-1 0.139* to 0.169** P2,P3 -- --Floodt-3 0.164* P4 -- --Stormt 0.142** P1 -- --Control VariablesTime Fixed Effects

Note: Summary of selected coefficients from Appendix D Tables D1 through D7. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Table 9. Determinants of Civil War Incidence : Significant Coefficients from Individual Event Panel Logit Models, 1945-1999 (n=6278)

Yes YesNo Yes

Many Disaster Events Regression Results: Onset

Variables Significant Coefficients Models Significant Coefficients ModelsExtreme Coldt 0.797* to 0.951** P1,P2,P3,P4 -- --Extreme Heatt-1 -- -- 1.022* to 1.06* P6,P7,P8Extreme Heatt-2 1.489** to 1.501** P3,P4 1.601*** to 1.627*** P7,P8Extreme Heatt-3 1.181* P4 1.258** P8Floodt 0.152* to 0.218*** P1,P2,P3 -- --Floodt-1 0.165** P1 -- --Control VariablesTime Fixed Effects No YesNote: Summary of selected coefficients from Appendix D Table D8. Only significant coefficients are presented. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Table 10. Determinants of Civil War Incidence : Significant Coefficients from All Events Panel Logit Models, 1945-1999 (n=6278)

Yes Yes

Econometric Specification: Duration

• Duration– I allow climate-change-related disasters to vary

over the duration of a war– Specifications

– Cox proportional hazards model– Exponential– Weibull– Gompertz– a discrete time proportional hazards model (Prentice-

Gloeckler, 1978) – Logistic regression models

Econometric Specification: Variables

Duration Disasters: droughts, epidemics, extreme

cold temperatures, extreme hot temperatures, floods, storms and wildfires

Controls: Coups, anti-colonial wars, Eastern Europe, “Sons of soil”, contraband production and trade, income, democracy,

Individual Disaster Events Regression Results: Duration

Models Droughtt

Cox -0.117*Exponential -1.377**Weibull -1.485**

Gompertz -1.415**

Discrete Time Proportional Hazards -1.400**Logit -1.460**

Table 11. Determinants of Civil War Duration Controlling for Frequency of Drought Events,

1945-1999 (n=1102)

Note: Summary of selected coefficients from Appendix G Tables G1 through G7. Only significant coefficients are presented. All six models include a full set of controls. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Many Disaster Events Regression Results: Duration

Models Droughtt

Cox -0.128**Exponential -1.377**Weibull -1.488**Gompertz -1.412**Discrete Time Proportional Hazards -1.418**Logit -1.462**

Table 12. Determinants of Civil War Duration Controlling for Frequency of All

Disaster Events, 1945-1999 (n=1102)

Note: Summary of selected coefficients from Appendix G Table G8. Only significant coefficients are presented. All six models include a full set of controls. Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.

Conclusion

First evidence on the impact of individual climate-change-related natural disasters on various measures of civil war Extreme cold events and epidemic outbreaks

increase the probability of civil war onset Previous years’ extreme heat events increase the

probability of civil occurring Drought events prolong civil war duration

Results can inform policy makers as they contemplate policies to mitigate and adapt to climate change

Future Research

Explore policy effectiveness in preventing and shortening civil war Government spending on disaster

preparedness Aid

Explore disaggregated data Indian province level data