Political Science
Eric Keels
Ph.D. Candidate
Department of Political Science
University of North Texas erickeels@my.unt.edu
Political Science
Negotiated settlements are often unstable
Post-civil war peacebuilding and civil war recurrence o Peacebuilding security objectives o Successful Agreements
Reduce Commitment Problems (Walter 2002; Fortna 2004; Hartzell and Hoddie 2003)
Share Information (Mattes and Savun 2010)
Raise costs for Defection (Mattes and Savun 2009) o Peacebuilding and democratization
Incompatibility between security and democracy objectives
Political Science
Democratization may generate post-civil war instability o Paris (2004), Jarstad and Sisk (2008), Autesserre (2009), Brancati and
Snyder (2013)
Democracy objectives may undermine post-war security
Security objectives may undermine post-war democracy
How can post-war governments maintain security while instituting democratic reforms?
Political Science
Post-Civil War Democratization and Security o Joshi et al (2015)
Primary focus on elections rather than democracy more broadly o Democracy Requires:
Broad participation
Open competition
Meaningful translation of votes into national representation
Theory: Electoral Reform and
Peace Duration
Political Science
Revisions to electoral laws act as a signal to opponents that the incumbent government is committed to the peace process
Revisions to electoral laws make it more difficult for dissidents to renege on peace process
Political Science
Electoral laws have historically been a way for governments to limit political competition
Laws determine who is allowed to participate (vote or run for office) and how votes are translated into representation (electoral systems)
Post-war elections may occur under the old electoral system which limits competition
Political Science
Electoral reforms are costly for elites in the incumbent government o Past systems may have ensured access to goods and services o Elites shoulder considerable risk by alienating their supporters
Implementing these reforms act as a costly signal that elites are interested in the peace process
H1: The implementation of electoral reforms called for in peace agreements reduces the risk of civil war recurrence.
Political Science
Electoral reforms should make the initial post-civil war election more stable o Armed militant groups have an incentive to abandon the peace process after losing (Brancati and Snyder 2013) o Militant groups rely on some civilian support o Electoral reforms make conventional participation more attractive than supporting violent dissent
H2: First post-war elections that occur after electoral reforms are implemented reduce the risk of civil war recurrence as compared to first post-war elections that occur when no electoral reform has taken place.
Political Science
Post-Civil War Comprehensive Peace Agreements from 1989-
2010 o UCDP Conflict Termination Dataset (Kreutz 2010) o UCDP Peace Agreement Dataset (Högbladh 2011) o Peace Accords Matrix (Joshi and Darby 2012)
Dependent Variable: Civil War Recurrence (0,1)
Independent Variable: Electoral Reform o 0=No reform implemented, 1= Reform initiated, 2=Partial implementation, 3=Fully implemented
Independent Variable: First Post-War Election*Electoral
Reforms
Political Science
Controls o Agreement Provisions: Disarmament, Demobilization, and Reintegration.
Transitional Power-Sharing and Peacekeeping o Conflict Variables: Conflict Intensity, Conflict Duration, Ethnic Conflict,
Incompatibility with Government o Government Controls: Polity II o Economic Controls: Logged GDP Per Capita
Unit of Analysis: Post-war Country Year
Model: Weibull Duration Model o Robustness checks with Cox Proportional Hazard Models and
LogLogistic Duration Models o Used Selection Models for robustness
Political Science
Substantive
Results
H1
Electoral
Reforms
H2
Electoral
Reforms *First
Election
Predicted Impact on Recurrence
Actual Impact on
Recurrence
Statistically
Significant?
Reduction in Risk of Recurrence
Negative
Negative
Negative
Negative
Yes
(Reject H0 by
95%)
Yes
(Reject H0 by
99%)
97.2%
99.9%
Political Science
Table II:
Hazard Ratios for each stage of implementation
Electoral Reform
Implementation
Electoral Reforms
Initiated
Reduced Risk by
69.7%
Electoral Reforms
Mostly Implemented
Reduced Risk by
90.9%
Electoral Reforms
Fully Implemented
Reduced Risk by
97.2%
Political Science
Figure I: The Impact of Electoral Reforms on the
Survival of Peace Following Negotiated Settlements
Political Science
Examine rates of Pro-Government and
Antigovernment violence after civil wars o SCAD dataset (Salehyan et al 2011) o Peace agreements in Africa and Latin America o Measured simple inclusion of Electoral Reform in peace agreements o Reduced rates pro-government violence by 76.8% o Reduced rates of antigovernment violence by 81% o Results remain robust when only looking at election violence
Political Science
Post-Civil War Electoral Reforms o Reduce the risk of civil war recurrence
Implementing these costly reforms signals a willingness to abide by the peace process o Ensures that post-war elections are stable
Reduces the ability of dissidents to abandon the peace process by allowing conventional participation o Results remain robust when looking at post-war event data
Questions?
Eric Keels
Ph.D. Candidate
Department of Political Science
University of North Texas erickeels@my.unt.edu
Political Science
Political Science
Descriptive
Statistics
Total Peace Agreements Electoral Reform
Inclusion
Average
Implementation
Minimum
Implementation
Maximum
Implementation
37 57% 1.37
0 3
Political Science
Table I
Weibull Hazard Model on Risk of Peace Failure
Implemented Disarmament
Implemented Demobilization
Implemented Reintegration
Implemented Transitional Political Power-Sharing lnGDP Per Capita
Major War
War Duration
Government Incompatibility
Polity II
Peacekeeping Operation
First Post-war Election
Implementation of Electoral Reforms
Electoral Reform*First Election
Ethnic War
Constant
P
Observations
𝑊𝑎𝑙𝑑 𝐶ℎ𝑖 2
Model 1
-1.156**
(0.470)
Model 2
-8.816***
(0.453)
1.883***
(0.702)
-8.756***
(1.305)
0.0391
(0.343)
-0.589
(0.470)
1.715
(1.070)
-1.296***
(0.111)
0.119
(0.753)
-0.157**
(0.0627)
0.179
(0.721)
-1.231**
(0.598)
-2.170***
(0.310)
-0.500
(0.762)
2.573
(3.292)
-0.500
(0.766)
2.629
(3.260)
0.78
6.04**
329
1.22
1736.95***
329
Coefficients Listed. Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
1.24
1634.08***
329
Model 3
-8.835***
(0.456)
1.872**
(0.753)
-9.193***
(1.846)
0.0168
(0.344)
-0.619
(0.457)
1.773*
(1.067)
-1.299***
(0.113)
0.140
(0.765)
-0.162**
(0.0653)
0.254
(0.725)
0.660
(1.180)
-1.214**
(0.594)
Model 4
-0.162**
(0.0650)
0.248
(0.720)
0.690
(1.200)
-1.193**
(0.599)
-7.147***
(1.012)
-0.497
(0.765)
2.618
(3.224)
-8.788***
(0.472)
1.660**
(0.713)
-8.167***
(0.716)
0.0158
(0.342)
-0.617
(0.450)
1.768*
(1.059)
-1.305***
(0.106)
0.150
(0.760)
1.24
2586.91***
329
Table IV
DDR Index Check
DDR Index
Transitional Political Power-Sharing lnGDP Per Capita
Major War
War Duration
Government Incompatibility
Polity II
Peacekeeping Operation
First Post-war Election
Implementation of Electoral Reforms
Electoral Reform*First Election
Ethnic War
Constant
Chi Squared
P
Observations
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Model 1
0.248
(0.720)
0.690
(1.200)
-1.193**
(0.599)
-7.247***
(1.012)
-0.497
(0.765)
2.643
(3.224)
-69.73***
(3.645)
0.0158
(0.342)
-0.617
(0.450)
1.768*
(1.059)
-1.328***
(0.104)
0.150
(0.760)
-0.162**
(0.0650)
2253.24***
1.24
329
Political Science
Integrated Army
DDR
GDP per Capita
Peacekeeping Operation
Polity II
Major War
War Duration
Ethnic Fractionalization
Electoral Reforms
Inverse Mills Ratio
Monopoly Ethnic Group
Political Power-Sharing
Constant
Observations
Gamma
Log Logistic Regression
Selection Model Selection Stage
(Probit Predicting Implementation)
1.700**
(0.705)
-0.483
(0.764)
-4.60e-05
(6.60e-05)
2.326**
(0.928)
Full Model
(Log Logistic Duration Model)
-2.009**
(0.931)
-1.418
(1.435)
0.00148**
(0.000696)
-4.814
(3.009)
-0.300
(0.254)
-2.400***
(0.651)
0.354**
(0.145)
-0.900
(1.078)
12.19***
(3.032)
-1.564**
(0.739)
-2.570*
(1.413)
0.121
(0.760)
-2.992***
(1.128)
371
18.88***
11.40***
(3.981)
308
381.62***
0.54
(0.30)
Robust standard errors in parentheses. Inverse Mills Ratio measures degree of selection effects.
*** p<0.01, ** p<0.05, * p<0.1
Political Science
Model I Bivariate Probit Model Predicting a) Choice to Include Electoral Reform and b) Choice to Return to Civil War
Equation Predicting Return to Violence
Implementation of Electoral Reforms
GDP Per Capita
War Duration
Major War
Peacekeeping Operation
Polity II
Transitional Power-sharing Government
Ethnic War
T1
T2
T3
Constant
-1.00
(0.41)**
-0.0003
(0.0001)***
0.10
(0.04)**
0.86
(0.32)***
0.16
(0.45)
-0.07
(0.03)**
0.38
(0.36)
0.44
(0.37)
-0.13
(0.04)***
0.03
(0.05)
-0.02
(0.03)
230.3
(113.6)**
Equation Predicting the Inclusion of Electoral Reforms in Peace
Agreements
Peacekeeping Operation
GDP Per Capita
Ethnic War
Polity II
Past Experience with Democracy
Constant
Rho
𝑊𝑎𝑙𝑑 𝐶ℎ𝑖
2
N
1.09
(0.49)**
0.00001
(0.00004)
-1.65
(0.50)***
-0.08
(0.04)*
0.28
(0.44)
0.55
(0.44)
-0.23(Not statistically different)
77.32***
357
Robust standard errors in parentheses.
*** p<0.01, **p<0.05, *p<0.1
Political Science
TABLE I
Negative Binomial Regression
Total Violence vs. Election Related Violence
Lagged Antigovernment Violence
Electoral Reform
Disarmament
First Election
Peacekeeping Operation
Ongoing Civil War
Transitional Power-Sharing
Infant Mortality Rate
Total Casualties
Duration
Lagged Polity II
Lagged Pro-Government Violence
Constant
𝐶ℎ𝑖 2
Alpha
Observations
Total
Antigovernment
Violence
Model 1
Election
Antigovernment
Violence
Model 2
Total
ProGovernment
Violence
Model 3
Election
ProGovernment
Violence
Model 4
0.274**
(0.114)
-1.660*
(0.858)
0.256
(0.201)
-0.856
(0.547)
-0.875
(0.631)
-0.405
(0.378)
0.430**
(0.204)
-0.00258
(0.00877)
1.48e-06**
(6.10e-07)
0.00282
(0.00231)
-0.0544
(0.0408)
-0.0270
(1.122)
75.00***
3.73
260
0.149
(0.161)
-4.359**
(2.050)
1.570***
(0.339)
1.893**
(0.785)
-1.407
(1.854)
-1.992**
(0.789)
1.579***
(0.540)
0.0100
(0.0267)
2.98e-06
(2.72e-06)
0.00792
(0.00592)
-0.0566
(0.0633)
-5.184*
(2.921)
188.62***
17.1
260
-1.460**
(0.590)
0.198
(0.244)
-0.155
(0.724)
-2.110***
(0.517)
-0.149
(0.512)
0.780***
(0.241)
0.0181
(0.0119)
2.70e-06***
(5.40e-07)
0.00204
(0.00272)
-0.0160
(0.0372)
0.522
(0.343)
-3.066*
(1.571)
56.79***
5.03
260
-5.152***
(1.339)
1.159***
(0.353)
4.013***
(1.115)
-2.683**
(1.109)
-1.822
(1.191)
1.517***
(0.431)
-0.0109
(0.0222)
3.62e-06*
(2.08e-06)
8.94e-05
(0.00527)
0.210***
(0.0754)
-0.134
(0.231)
-3.277
(2.608)
516.46***
20.3
260
Coefficients listed. Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Political Science