An Inquiry into the Use of Illegal Electoral Practices and Effects - - PowerPoint PPT Presentation

An Inquiry into the Use of Illegal Electoral Practices and Effects of Political Violence

Roxana Gutiérrez-Romero UAB, iiG

Objectives

2

• Violence instigated strategically in 2007 Kenyan

Elections?

• Consequences of electoral violence?

3

CSAE Surveys

• Two weeks before 2007 elections and summer 2008
• Nationally and regionally representative
• 1,207 Kenyans aged 18 and over

Media Monitoring

• On 24 hours basis 1 Dec 2007- 31 March 2008

– 4 Kenyan national TV stations – 2 major Kenyan newspapers – 6 Kenyan local radio stations

4

CSAE Surveys

One in ten received a threat A third reported political actors instigating violence before and after elections 23% reported gangs operating in communities. After elections up to 30% 20% suffered a personal impact

5

Death Toll

12 109 123 79 17 40 88 35 13 14 36 18 13 46 74 130 98 21 24 10 20 16 20

1 21 41 61 81 101 121 141

1/12/07 2/12/07 3/12/07 4/12/07 5/12/07 6/12/07 7/12/07 8/12/07 11/12/07 13/12/07 20/12/07 22/12/07 24/12/07 27/12/07 30/12/07 1/1/08 2/1/08 3/1/08 4/1/08 5/1/08 7/1/08 8/1/08 9/1/08 10/1/08 12/1/08 13/1/08 14/1/08 15/1/08 17/1/08 18/1/08 19/1/08 20/1/08 21/1/08 22/1/08 23/1/08 24/1/08 25/1/08 26/1/08 27/1/08 28/1/08 29/1/08 30/1/08 31/1/08 1/2/08 2/2/08 3/2/08 4/2/08 5/2/08 6/2/08 12/2/08 14/2/08 19/2/08 22/2/08 28/2/08 29/2/08 3/3/08 7/3/08 19/3/08 23/3/08 Pre-Electoral Violence Post-Electoral Violence

41 killed pre + 1,169 killed post election day

Source: Monitored Kenyan media

Probit Marginal Effects Before After PNU ODM Before After (1) (2) (3) (4) (5) (6) Expected Win Difference in Presidential Election 2007‡

• 0.132* -0.217** -0.043** -0.055**
• 0.091
• 0.314***

Constituency has more preference for PNU than ODM*Expected win difference 0.069** Constituency has more preference for ODM than PNU*Expected win difference 0.053 0.075 0.076** 0.003 0.132 Ethnicity (Kikuyu reference group) Luo 0.016 0.028 0.052

• 0.026**

0.048 0.024 Luhya

• 0.005
• 0.085*

0.018

• 0.022**

0.189** 0.144** Kamba

• 0.015 -0.164***
• 0.001
• 0.092*
• 0.201***

Meru

• 0.088* -0.255***
• 0.000

0.008

• 0.222***
• 0.259***

Kissi

• 0.017
• 0.066
• 0.016*
• 0.018*
• 0.034
• 0.060

Kalenjin 0.016 0.067 0.019

• 0.026**

0.015 0.050 Mijikenda 0.103 0.090

• 0.008
• 0.180***
• 0.241***

Somali

• 0.110
• 0.130**
• 0.162***
• 0.283***

Others 0.030

• 0.089*

0.020

• 0.027***
• 0.085**
• 0.125**

Ethnic hotspot (PNU reference group) Ethnic hotspot ODM

• 0.140**
• 0.020

0.045*

• 0.014
• 0.067
• 0.029

Ethnic hotspot ODM-K

• 0.052
• 0.014
• 0.029**
• 0.112**
• 0.073

Other

• 0.061
• 0.061

0.238 0.019

• 0.105**
• 0.056

Wealth Asset Index 0.133 0.186** 0.001 0.044 0.155* 0.150 Number of observations 1207 1207 1014 1094 1207 1207 Significance Level * p<0.10, ** p<0.05, *** p<0.001

‡ Index based on pre-election question:

"Which party do you think will win the National Presidential elections in December 2007?" Received threats from: Polititians Instigated Violence Heard Gangs Connected to Politics Operating

Changes in Perceptions Before- After the Disputed Elections

• Difference-in-difference estimator via propensity

score matching.

• Propensity score controlling for: respondent's

ethnicity, education, age, had land disputes right before the 2007 elections or in the 1997, 2002, 2005 elections, poverty

Since elections sometimes produce bad results, we should adopt

• ther methods

Parties should not be allowed to form on tribe or religion basis In deciding which party likes most, considers the ethnic origin of the party's leader Identify first and foremost as Ethnic It is sometimes necessary to use violence in support

• f a just cause

(1) (2) (3) (4) (5) Victim of violence*2008

• 0.092**

0.122**

• 0.004

0.126** 0.097** Year 2008 0.212*** 0.046 0.130***

• 0.102***

0.065** Victim of violence 0.007

• 0.123***
• 0.004
• 0.074**
• 0.020

Constant 0.088*** 0.662*** 0.501*** 0.218*** 0.128*** Number of observations 1352 1352 1352 1352 1352 Significance Level * p<0.10, ** p<0.05, *** p<0.001

In Sum

• Violence instigated strategically
• Desire for changing ethno-political system
• But still relying on ethnicity to assess parties
• More victims of violence identify on ethnic

terms

Introduction The Model Outside Interventions Conclusions

State strength, outside interventions, and the intensity of conflict

Katharina Wick 1 Erwin Bulte 2

1University of Vienna 2Wageningen & Tilburg University

March 18, 2013

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent,

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent, effort costs,

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent, effort costs, effectiveness in conflict

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent, effort costs, effectiveness in conflict and ruler’s time horizon.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent, effort costs, effectiveness in conflict and ruler’s time horizon. Main Message: Policies are not ’one size fits all’

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

In a Nutshell

Theoretically model conflict of two groups (ruler and rebels) over distribution of exogenous windfall rents. Evaluate effect of third party interventions with aim of influencing conflict

• utcome.

Considered interventions: Change in value of rent, effort costs, effectiveness in conflict and ruler’s time horizon. Main Message: Policies are not ’one size fits all’ Effect of policies depends on military strength, state strength and their interplay.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003).

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention?

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power) military strength:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power) military strength:

valuation of the prize (eg. trade embargo: Kimberley Process)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power) military strength:

valuation of the prize (eg. trade embargo: Kimberley Process) conflict effort costs (eg. availability of arms, affect

• pportunity costs)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power) military strength:

valuation of the prize (eg. trade embargo: Kimberley Process) conflict effort costs (eg. availability of arms, affect

• pportunity costs)

ability/efficiency on the battlefield (eg. transfer of military technology)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

Civil Conflict

After 1945: majority of conflicts are intrastate (Balch-Lindsey & Enterline 2000), majority of those in developing countries (Murdoch & Sandler 2002). Rising incidence and duration (Collier et al. 2003). Consequence: human misery, economic setback, international spillovers. Often distributive conflicts over rents. Economic factors determine viability. What can be done? Third Party Intervention? Potential Interventions:

state strength (eg. probability to stay in power) military strength:

valuation of the prize (eg. trade embargo: Kimberley Process) conflict effort costs (eg. availability of arms, affect

• pportunity costs)

ability/efficiency on the battlefield (eg. transfer of military technology)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions: ability or capacity of states to act (raise taxes, regulate, provide public goods), given a set of objectives and constraints (Acemoglu 2005)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions: ability or capacity of states to act (raise taxes, regulate, provide public goods), given a set of objectives and constraints (Acemoglu 2005) Pranab Bardhan, 2010: ’ability to credibly pre-commit’

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions: ability or capacity of states to act (raise taxes, regulate, provide public goods), given a set of objectives and constraints (Acemoglu 2005) Pranab Bardhan, 2010: ’ability to credibly pre-commit’ ’(a strong ruler can be thought of) a Stackelberg leader, in a model where the ruler maximizes his objective function subject to the reaction function of the ruled.’

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions: ability or capacity of states to act (raise taxes, regulate, provide public goods), given a set of objectives and constraints (Acemoglu 2005) Pranab Bardhan, 2010: ’ability to credibly pre-commit’ ’(a strong ruler can be thought of) a Stackelberg leader, in a model where the ruler maximizes his objective function subject to the reaction function of the ruled.’

Strong Ruler: Leader in a Stackelberg setup, state can influence actions of the rebels

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions In a Nutshell Civil Conflict State Strength

State strength

State Strength, different complementary definitions: ability or capacity of states to act (raise taxes, regulate, provide public goods), given a set of objectives and constraints (Acemoglu 2005) Pranab Bardhan, 2010: ’ability to credibly pre-commit’ ’(a strong ruler can be thought of) a Stackelberg leader, in a model where the ruler maximizes his objective function subject to the reaction function of the ruled.’

Strong Ruler: Leader in a Stackelberg setup, state can influence actions of the rebels Weak Ruler: Player in a Cournot setup, state cannot influence actions of rebels

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Setup

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Setup

Environment given by exogenous parameters and potential third party intervention

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Setup

Environment given by exogenous parameters and potential third party intervention Indefinitely repeated conflict game between ruler and rebels is played

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Stage Game

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Stage Game

Two groups, Ruler (R) and Rebels (E),

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Stage Game

Two groups, Ruler (R) and Rebels (E), choose their effort levels

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Stage Game

Two groups, Ruler (R) and Rebels (E), choose their effort levels so as to maximize their ’income’ πi = pi · Zi − ci · Xi,

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Stage Game

Two groups, Ruler (R) and Rebels (E), choose their effort levels so as to maximize their ’income’ πi = pi · Zi − ci · Xi, where probability of winning of group i is given by pi = ai · Xi ai · Xi + aj · Xj i = {E, R} Zi is valuation of prize of group i ci is unit effort cost of group i aj is bias of contest in favor of group j, aR = a, aE = 1

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Two dimensions worth distinguishing:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Two dimensions worth distinguishing: State Strength: Weak State

• vs. Strong State

Cournot Stackelberg

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Two dimensions worth distinguishing: State Strength: Weak State

• vs. Strong State

Cournot Stackelberg Military Strength (∼ Prize/Cost Ratio): a ZR

cR > ZE cE ZE cE > a ZR cR

Favorite State vs. Underdog State

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Two dimensions worth distinguishing: State Strength: Weak State

• vs. Strong State

Cournot Stackelberg Military Strength (∼ Prize/Cost Ratio): a ZR

cR > ZE cE ZE cE > a ZR cR

Favorite State vs. Underdog State In equilibrium: aX ∗

R > X ∗ E

X ∗

E > aX ∗ R

p∗

R > p∗ E

p∗

E > p∗ R Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Interaction in the Stage Game

Two dimensions worth distinguishing: State Strength: Weak State

• vs. Strong State

Cournot Stackelberg Military Strength (∼ Prize/Cost Ratio): a ZR

cR > ZE cE ZE cE > a ZR cR

Favorite State vs. Underdog State In equilibrium: aX ∗

R > X ∗ E

X ∗

E > aX ∗ R

p∗

R > p∗ E

p∗

E > p∗ R Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Conflict Outcomes: State as Underdog

Cournot Eqm Stack Eqm BRE BRR

Ρ

IsoprofitR R E

Ruler is disadvantaged, rebels are ’ready to fight’. ⇒ aX ∗

R < X ∗ E always

⇒ X ∗S

R

< X ∗C

R ,

X ∗S

E

< X ∗C

E

⇒ less conflict if ruler is strong.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Conflict Outcomes: State as Favorite

Cournot Eqm Stack Eqm BRE BRR

Ρ

IsoprofitR R E

Ruler is advantaged, rebels are ’ready to re- treat’. ⇒ X ∗

E < aX ∗ R always

⇒ X ∗S

R

> X ∗C

R ,

X ∗S

E

< X ∗C

E

⇒ ambiguous

• n

total conflict outcome

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device: If a ruler wants to be perceived as strong in the future, he needs to stick to X ∗S

R . Incentive to do so, depends on his regard for future

income.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device: If a ruler wants to be perceived as strong in the future, he needs to stick to X ∗S

R . Incentive to do so, depends on his regard for future

income.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device: If a ruler wants to be perceived as strong in the future, he needs to stick to X ∗S

R . Incentive to do so, depends on his regard for future

income.

For ruler to be credibly strong: δ > δcrit.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Setup Ingredients of the Stage Game Interaction in the Stage Game Credibility of a State

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device: If a ruler wants to be perceived as strong in the future, he needs to stick to X ∗S

R . Incentive to do so, depends on his regard for future

income.

For ruler to be credibly strong: δ > δcrit. Ruler is likely to be weak, whenever military strength (prize/cost ratio) differs a lot among the two groups, i.e. conflict is very uneven.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Objectives of a Third Party

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Objectives of a Third Party

Objectives of a third party can be of different sorts:

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Objectives of a Third Party

Objectives of a third party can be of different sorts: reduce resources wasted in conflict, (X ∗

E + X ∗ R ). Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Objectives of a Third Party

Objectives of a third party can be of different sorts: reduce resources wasted in conflict, (X ∗

E + X ∗ R ).

reduce open conflict intensity, min(X ∗

E , X ∗ R ): (one cannot fight alone). Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Objectives of a Third Party

Objectives of a third party can be of different sorts: reduce resources wasted in conflict, (X ∗

E + X ∗ R ).

reduce open conflict intensity, min(X ∗

E , X ∗ R ): (one cannot fight alone).

affect the distribution of expected conflict gains.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement)

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power → ruler becomes strong

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power → ruler becomes strong → less resources wasted in conflict, conflict intensity decreases, distribution?

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power → ruler becomes strong → less resources wasted in conflict, conflict intensity decreases, distribution? Increase a favorite state’s probability to stay in power

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power → ruler becomes strong → less resources wasted in conflict, conflict intensity decreases, distribution? Increase a favorite state’s probability to stay in power → ruler becomes strong →

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Changing the ’rules of the game’

Outside party can change the game that is played (the environment), by affecting the ruler’s regard for the future (probability of replacement) affecting relative military strength (prize, cost, effectiveness) non-marginally Examples: Increase an underdog state’s probability to stay in power → ruler becomes strong → less resources wasted in conflict, conflict intensity decreases, distribution? Increase a favorite state’s probability to stay in power → ruler becomes strong → ambiguous effect on resources wasted in conflict, conflict intensity decreases, distribution tilts towards ruler (repressive regime).

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

The Effect of Outside Interventions

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZE ↓, cE ↑ X

↓ ↓ ↓ ↓ X

↑ ↑ ↑ ↓ (X

? ? ? ↓ min(X

↑ ↑ ↓ ↓ Distribution tilts towards Ruler Ruler Ruler ? ZR ↓, cR ↑ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↓ (X

↓ ↓ ? ? min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels Rebels a ↓ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↑ (X

↓ ↓ ? ↑ min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels ? Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Ruler

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZR ↓, cR ↑ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↓ (X

↓ ↓ ? ? min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels Rebels

If ZR ↓, then XR ↓.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Ruler

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZR ↓, cR ↑ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↓ (X

↓ ↓ ? ? min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels Rebels

If ZR ↓, then XR ↓. Rebels react by decreasing or increasing their effort, depending on relative military strength.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Ruler

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZR ↓, cR ↑ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↓ (X

↓ ↓ ? ? min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels Rebels

If ZR ↓, then XR ↓. Rebels react by decreasing or increasing their effort, depending on relative military strength. Reduction in prize ZR may inadvertently lead to higher conflict intensity & more resources wasted.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Ruler

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZR ↓, cR ↑ X

↓ ↓ ↑ ↑ X

↓ ↓ ↓ ↓ (X

↓ ↓ ? ? min(X

↓ ↓ ↑ ↑ Distribution tilts towards ? ? Rebels Rebels

If ZR ↓, then XR ↓. Rebels react by decreasing or increasing their effort, depending on relative military strength. Reduction in prize ZR may inadvertently lead to higher conflict intensity & more resources wasted. Policies targeting the ruler are ’robust’ to state strength.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Rebels

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZE ↓, cE ↑ X

↓ ↓ ↓ ↓ X

↑ ↑ ↑ ↓ (X

? ? ? ↓ min(X

↑ ↑ ↓ ↓ Distribution tilts towards Ruler Ruler Ruler ?

Known that state is favorite, but state strength still plays crucial role.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Rebels

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZE ↓, cE ↑ X

↓ ↓ ↓ ↓ X

↑ ↑ ↑ ↓ (X

? ? ? ↓ min(X

↑ ↑ ↓ ↓ Distribution tilts towards Ruler Ruler Ruler ?

Known that state is favorite, but state strength still plays crucial role. If cE ↑, XE ↓.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Objectives of a Third Party Non-Marginal Interventions Marginal Interventions Targeting the Ruler Targeting the Rebels

Targeting the Rebels

policy

Change in

State as Weak Underdog, (1) State as Strong Underdog, (2) State as Strong Favorite, (3) State as Weak Favorite, (4) ZE ↓, cE ↑ X

↓ ↓ ↓ ↓ X

↑ ↑ ↑ ↓ (X

? ? ? ↓ min(X

↑ ↑ ↓ ↓ Distribution tilts towards Ruler Ruler Ruler ?

Known that state is favorite, but state strength still plays crucial role. If cE ↑, XE ↓. A strong state realizes rebels can now be suppressed even easier,XR ↑.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Conclusions

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Conclusions

Important to pay attention to not only direct effect of policy, but also indirect effects through strategic interaction.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Conclusions

Important to pay attention to not only direct effect of policy, but also indirect effects through strategic interaction. ⇒ Policies may have unintended consequences.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Conclusions

Important to pay attention to not only direct effect of policy, but also indirect effects through strategic interaction. ⇒ Policies may have unintended consequences. Some policies are more robust than others with respect to state strength, military strength as well as their interplay.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Conclusions

Important to pay attention to not only direct effect of policy, but also indirect effects through strategic interaction. ⇒ Policies may have unintended consequences. Some policies are more robust than others with respect to state strength, military strength as well as their interplay. Changing the ’rules of the game’ (affecting time horizon of the ruler, or prize/costs non-marginally) may be another lever for policy.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device. When is it then in the best interest of the ruler to behave strong (though this is not his best answer in a one-shot game)?

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device. When is it then in the best interest of the ruler to behave strong (though this is not his best answer in a one-shot game)? Given rebels play ’trigger strategy’ (believe ruler will play Stackelberg as long as he has never deviated, never believe him thereafter):

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device. When is it then in the best interest of the ruler to behave strong (though this is not his best answer in a one-shot game)? Given rebels play ’trigger strategy’ (believe ruler will play Stackelberg as long as he has never deviated, never believe him thereafter): If a ruler wants to be perceived as strong in the future, he needs to stick to R∗

S . Incentive to do so, depends on his regard for future income Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

Who Is A Strong State?

Stage Game is indefinitely repeated, no commitment device. When is it then in the best interest of the ruler to behave strong (though this is not his best answer in a one-shot game)? Given rebels play ’trigger strategy’ (believe ruler will play Stackelberg as long as he has never deviated, never believe him thereafter): If a ruler wants to be perceived as strong in the future, he needs to stick to R∗

S . Incentive to do so, depends on his regard for future income

→ A credible strong ruler has a high discount factor.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

The Critical Discount Factor

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

The Critical Discount Factor

For ruler to be credibly strong: δ > δcrit.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

The Critical Discount Factor

For ruler to be credibly strong: δ > δcrit.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction The Model Outside Interventions Conclusions Additional Material

The Critical Discount Factor

For ruler to be credibly strong: δ > δcrit.

Ruler is likely to be weak, whenever military strength (prize/cost ratio) differs a lot among the two groups, i.e. conflict is very uneven.

Katharina Wick , Erwin Bulte State strength, outside interventions, and the intensity of conflict

Introduction Data and Empirical Strategy Results Resilience Conclusion

War, Resilience, and Political Engagement in Africa

Ach Adhvaryu & James Fenske

Yale University and University of Oxford

March 2013

1 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Motivation Overview

Motivation

Do the effects of war linger on through changes in individuals’ preferences? (Voors et al., 2013) Political knowledge, engagement, and attitudes matter: Better informed voters turn out more and vote for better candidates (Banerjee et al., 2012); civic engagement builds social capital (Putnam, 2000); democracy is associated with growth (Barro, 1996; Tavaresa & Wacziarg, 2001) Does war beget war through the endogenous evolution of preferences?

If so, this link could help explain part of the “African growth tragedy” (Collier & Gunning, 1999)

2 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Motivation Overview

What we do

How does exposure to conflict in early life shape political beliefs and activities? We use data on political engagement, beliefs, and attitudes from the 2005 Afrobarometer (nationally representative survey from 17 sub-Saharan African countries). We merge this with spatial data on war intensity in the last 5 decades from the Peace Research Institute Oslo (PRIO). We find that exposure to conflict during early childhood (ages 0-14) has very small effects on later-life political engagement and attitudes.

3 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Motivation Overview

What we do (ctd.)

We show that this (non-)result:

Is robust to alternative specifications, measurements, data sources, and possible data problems. Is consistent with small (standardized) effects in most other studies.

Implications:

Individuals are, on average, resilient. Population-level estimates tell a different story than case studies focusing on directly affected individuals.

4 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Motivation Overview

1

Introduction

2

Data and Empirical Strategy

3

Results

4

Resilience

5

Conclusion

5 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

1

Introduction

2

Data and Empirical Strategy

3

Results

4

Resilience

5

Conclusion

6 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Data: War Location and Intensity

These data come from three UCDP/PRIO data sets, which geocode all deadly conflicts in Africa since 1947 and give the number of battle deaths and spatial extent for each conflict. We treat battle deaths as a measure of the intensity of conflict exposure. A respondent is “treated” with the conflict’s deaths if its spatial extent overlaps with the respondent’s survey cluster and it occurred during the specified ages for the respondent.

7 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Table 1A. Summary statistics: War exposure and controls

(1) (2) (3) (4) (5) Mean s.d. Min Max N PRIO Conflict Measures: Battle deaths (000s), ages 0-14 26.574 45.807 207.933 18,141 Battle deaths (000s), ages 0-14, per 1000 population 0.001 0.002 0.025 18,141 Battle deaths (000s), ages 0-14, per area 0.045 0.112 0.859 18,141 1(Battle deaths > 0, ages 0-14) 0.539 0.498 1 18,141 Years of conflict exposure, ages 0-14 4.053 5.045 15 18,141 MEPV Conflict Measures: War of Independence Score, ages 0-14 (avg) 0.137 0.544 3 17,755 International Violence Score, ages 0-14 (avg) 0.016 0.059 0.4 16,490 International War Score, ages 0-14 (avg) 0.026 0.078 0.267 16,490 Civil Violence Score, ages 0-14 (avg) 0.002 0.012 0.067 16,490 Civil War Score, ages 0-14 (avg) 0.254 0.795 4.8 16,490 Ethnic Violence Score, ages 0-14 (avg) 0.319 0.585 2 16,490 Ethnic War Score, ages 0-14 (avg) 0.287 0.730 2.8 16,490 Controls: Year of birth 1968 14.8 1875 1987 25,110 Age heaping 0.28 0.45 1 25,397 Female 0.50 0.50 1 25,397 Urban 0.38 0.49 1 25,397 Ethnic share in district 0.61 0.35 0.0027 1 25,397 8 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Battle deaths, Ages 0-14: Africa 1965, Nigeria 1965 & 75

9 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Data: Political Outcomes

Engagement and attitude responses come from the 2005 round of the Afrobarometer. This covers 25,397 individuals from 18 nationally-representative surveys of Sub-Saharan countries. We group these into 10 categories: voting, collective action, contact with political figures, refusal to pay bribes, interest in politics, deference to authority, support for democracy, support for equality, support for the rule of law, and trust.

10 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Data: Afrobarometer Examples

Which of the following statements is closest to your view? Choose Statement A or Statement B. A: All people should be permitted to vote, even if they do not fully understand all the issues in an election. B: Only those who are sufficiently well educated should be allowed to choose our leaders. There are many ways to govern a country. Would you disapprove or approve of the following alternatives: Only one political party is allowed to stand for election and hold office? (Five point scale, “strongly disapprove” to “strongly approve”)

11 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Table 1B. Sum. stats.: Standardized Outcome Groupings

(1) (2) (3) (4) (5) Mean s.d. Min Max N Voting 1

• 1.923

0.561 25,248 Collective action 1

• 1.693

4.654 23,981 Contact 1

• 0.709

5.223 22,571 Refusal of bribes 1

• 5.428

0.878 24,677 Interest 1

• 2.495

2.871 20,227 Deference 1

• 4.534

2.488 20,766 Democracy 1

• 4.657

1.853 17,127 Equality 1

• 4.118

2.269 23,668 Rule of law 1

• 4.467

1.621 21,842 Trust 1

• 2.513

1.938 10,291

Notes: See Table A.1 for individual components of each grouping. To generate standardized outcomes, each component is first de-meaned and normalized by its standard deviation. All standardized components within an outcome grouping are then summed, creating a standardized group variable, and normalized by SD again. This mean 0 variance 1 var is then used in mean effects analysis. 12 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Data Specification

Empirical Specification

yir = βexposureir + x′

irγ + δr + ηt + ǫir

yir is a measure of political engagement or attitudes. Respondent i lives in sub-national “region” r. exposureir is the respondent’s exposure to war between ages 0 − 14. δr is a region of residence dummy. We do not have region of birth, and will introduce robustness checks to deal with migration below. ηr is a year of birth dummy. xir is other controls: female, urban, district ethnic share, and dummies for own living standards, level of education, and

• ccupation.

We use OLS, and cluster standard errors by survey cluster. We report “standardized” coefficients.

13 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

1

Introduction

2

Data and Empirical Strategy

3

Results

4

Resilience

5

Conclusion

14 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 2A. Main Results: Effects of Early-life Conflict Exposure on Political Attitudes and Behaviors

(1) (2) (3) (4) (5) Voting Collective action Contact Refusal of bribes Interest Mean Effects: Battle Deaths (0-14) 0.00614

• 0.0126
• 0.000325

0.0236* 0.0143 (0.0109) (0.0117) (0.0134) (0.0128) (0.0131) Battle Deaths (0-14) per 1000 0.0176 0.00980

• 0.0119

0.0146 0.0232* (0.0124) (0.0130) (0.0125) (0.0137) (0.0121) Battle Deaths (0-14) per area 0.0352*** 0.0120 0.00283

• 0.0128

0.0204** (0.00942) (0.0103) (0.0108) (0.0118) (0.00949) Exposure dummy (0-14)

• 0.0577***
• 0.0117

0.0377

• 0.000964
• 0.0261

(0.0206) (0.0222) (0.0240) (0.0216) (0.0221) Years of exposure (0-14)

• 0.00852***
• 0.00502*
• 0.00326

0.00779*** 0.00110 (0.00260) (0.00266) (0.00282) (0.00292) (0.00266) IV Battle Deaths (0-14)

• 0.054
• 0.019
• 0.026

0.042 0.062* (0.041) (0.043) (0.039) (0.048) (0.034) First stage F-stat 103.59 102.21 104.71 101.84 101.18 Factors: Battle Deaths (0-14) 0.006 0.003

• 0.002

0.022*

• 0.003

(0.011) (0.011) (0.013) (0.013) (0.013) N 17,780 16,950 16,621 17,372 14,287 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each cell represents a regression of a political atti- tude/behavior (column headers) on a conflict measure (row headers). All regressions include dummies for region of residence, year

• f birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s

ethnicity in the district population is also used as a control. Standardized coefficients are reported. 15 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 2B. Main Results: Effects of Early-life Conflict Exposure on Political Attitudes and Behaviors

(6) (7) (8) (9) (10) Deference Democracy Equality Rule of law Trust Mean Effects: Battle Deaths (0-14) 0.0497*** 0.00996 0.00895 0.00262 0.0138 (0.0138) (0.0143) (0.0139) (0.0150) (0.0171) Battle Deaths (0-14) per 1000 0.0289**

• 0.00580

0.00645 0.00588 0.00892 (0.0131) (0.0135) (0.0138) (0.0142) (0.0141) Battle Deaths (0-14) per area 0.0336***

• 0.0121

0.00324

• 0.000884
• 0.00111

(0.0111) (0.0109) (0.0128) (0.0129) (0.0107) Exposure dummy (0-14) 0.0276 0.0128 0.00924 0.00276

• 0.0494*

(0.0230) (0.0254) (0.0211) (0.0220) (0.0261) Years of exposure (0-14) 0.00618**

• 0.000339

0.000609 0.00143

• 0.00440

(0.00307) (0.00296) (0.00281) (0.00304) (0.00344) IV Battle Deaths (0-14) 0.077**

• 0.055
• 0.024

0.091** 0.118*** (0.039) (0.043) (0.041) (0.046) (0.035) First stage F-stat 105.73 89.06 110.36 104.71 109.28 Factors: Battle Deaths (0-14) 0.017 0.009

• 0.013

0.012 0.018 (0.017) (0.013) (0.013) (0.014) (0.018) N 14,914 12,470 16,638 15,573 7,847 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each cell represents a regression of a political atti- tude/behavior (column headers) on a conflict measure (row headers). All regressions include dummies for region of residence, year

• f birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s

ethnicity in the district population is also used as a control. Standardized coefficients are reported. 16 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 3A. Heterogeneous Effects of Early-life Conflict Exposure by Time Since Exposure, Minority Status, and Gender

(1) (2) (3) (4) (5) Voting Collective action Contact Refusal of bribes Interest Time Since Exposure: Battle Deaths (0-14) x Age

• 0.021

0.003 0.012

• 0.030*

0.034** (0.015) (0.014) (0.016) (0.017) (0.015) Battle Deaths (0-14) 0.032

• 0.017
• 0.015

0.061***

• 0.028

(0.024) (0.021) (0.023) (0.023) (0.023) Minority Status: Battle Deaths (0-14) x Ethnic Share in District

• 0.000

0.012

• 0.013
• 0.021**

0.002 (0.008) (0.008) (0.009) (0.009) (0.009) Battle Deaths (0-14) 0.006

• 0.014
• 0.001

0.026** 0.015 (0.011) (0.012) (0.013) (0.013) (0.013) Gender: Battle Deaths (0-14) x Female 0.021 0.038** 0.001

• 0.014
• 0.024*

(0.014) (0.015) (0.015) (0.014) (0.014) Battle Deaths (0-14)

• 0.005
• 0.032**
• 0.001

0.031** 0.027* (0.014) (0.014) (0.016) (0.014) (0.015) N 17,780 17,319 16,621 17,372 15,709 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column by pair of rows above represents a regression

• f a political attitude/behavior (column headers) on battle deaths interacted with either age, ethnic share in district, or a female

dummy (see row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 17 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 3B. Heterogeneous Effects of Early-life Conflict Exposure by Time Since Exposure, Minority Status, and Gender

(6) (7) (8) (9) (10) Deference Democracy Equality Rule of law Trust Time Since Exposure: Battle Deaths (0-14) x Age

• 0.010
• 0.019
• 0.031*

0.021 0.011 (0.018) (0.016) (0.017) (0.017) (0.020) Battle Deaths (0-14) 0.062** 0.034 0.047*

• 0.024

0.000 (0.027) (0.026) (0.024) (0.026) (0.031) Minority Status: Battle Deaths (0-14) x Ethnic Share in District

• 0.010
• 0.020*
• 0.020**
• 0.017

0.012 (0.011) (0.010) (0.010) (0.011) (0.012) Battle Deaths (0-14) 0.049*** 0.011 0.008 0.004 0.015 (0.014) (0.014) (0.014) (0.015) (0.017) Gender: Battle Deaths (0-14) x Female

• 0.034**
• 0.005
• 0.034**

0.010

• 0.016

(0.017) (0.018) (0.016) (0.016) (0.018) Battle Deaths (0-14) 0.067*** 0.012 0.027

• 0.002

0.021 (0.016) (0.015) (0.016) (0.018) (0.019) N 14,914 12,470 16,638 15,573 7,847 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column by pair of rows above represents a regression

• f a political attitude/behavior (column headers) on battle deaths interacted with either age, ethnic share in district, or a female

dummy (see row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 18 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 4A. Heterogeneous Effects of Early-life Conflict Exposure by Conflict Type

(1) (2) (3) (4) (5) Voting Collective action Contact Refusal of bribes Interest War of Independence 0.262

• 0.197
• 0.595

0.0398

• 0.123

(0.300) (0.345) (0.363) (0.348) (0.282) International Conflict

• 0.0546***
• 0.0189
• 0.00624

0.0267

• 0.0140

(0.0160) (0.0155) (0.0152) (0.0164) (0.0151) Civil Conflict 0.0458* 0.0192 0.0201

• 0.0124

0.0494** (0.0265) (0.0270) (0.0246) (0.0301) (0.0237) Ethnic Conflict 0.0514*** 0.0417** 0.0765***

• 0.0654***

0.0591*** (0.0187) (0.0209) (0.0199) (0.0208) (0.0195) N 14,182 13,567 13,438 13,864 11,988 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column above represents a regression of a political attitude/behavior (column headers) on a conflict measure (row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 19 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 4B. Heterogeneous Effects of Early-life Conflict Exposure by Conflict Type

(6) (7) (8) (9) (10) Deference Democracy Equality Rule of law Trust War of Independence 0.572* 0.544 0.321 0.105

• 0.229

(0.335) (0.366) (0.317) (0.352) (0.548) International Conflict

• 0.00538
• 0.0403**
• 0.0204

0.0113 0.00394 (0.0180) (0.0185) (0.0168) (0.0165) (0.0166) Civil Conflict 0.0308

• 0.0550*

0.0152 0.0361 0.0776*** (0.0273) (0.0298) (0.0290) (0.0297) (0.0280) Ethnic Conflict

• 0.00914

0.0142 0.0327

• 0.00941
• 0.0511

(0.0226) (0.0229) (0.0220) (0.0227) (0.0356) N 12,076 10,007 13,467 12,530 6,227 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column above represents a regression of a political attitude/behavior (column headers) on a conflict measure (row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 20 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 5A. Heterogeneous Effects of Ethnic Conflict Exposure by Ethnic Share in District

(1) (2) (3) (4) (5) Voting Collective action Contact Refusal of bribes Interest Ethnic War (0-14) x Ethnic Share in District

• 0.0216**
• 0.0237**
• 0.0199*
• 0.00894

0.0229** (0.00976) (0.0103) (0.0104) (0.0118) (0.00965) Ethnic War 0.0239 0.0335* 0.0687***

• 0.0515***

0.0433** (0.0176) (0.0190) (0.0191) (0.0191) (0.0182) Ethnic Share in District 0.0350***

• 0.00458

0.0236**

• 0.0233*

0.0120 (0.0117) (0.0122) (0.0108) (0.0119) (0.0107) N 14,182 13,567 13,438 13,864 11,988 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column by pair of rows above represents a regression

• f a political attitude/behavior (column headers) on battle deaths interacted with either age, ethnic share in district, or a female

dummy (see row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 21 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Table 5B. Heterogeneous Effects of Ethnic Conflict Exposure by Ethnic Share in District

(6) (7) (8) (9) (10) Deference Democracy Equality Rule of law Trust Ethnic War (0-14) x Ethnic Share in District 0.0138

• 0.0134

0.00547 0.0294***

• 0.00619

(0.0111) (0.0114) (0.0101) (0.0108) (0.0119) Ethnic War

• 0.0174

0.0112 0.0226

• 0.0152
• 0.0819***

(0.0215) (0.0209) (0.0213) (0.0212) (0.0291) Ethnic Share in District 0.00867

• 0.00649

0.00597

• 0.00133

0.0187 (0.0131) (0.0138) (0.0116) (0.0123) (0.0141) N 12,076 10,007 13,467 12,530 6,227 Notes: ***Significant at 1%, **Significant at 5%, *Significant at 10%. Each column by pair of rows above represents a regression

• f a political attitude/behavior (column headers) on battle deaths interacted with either age, ethnic share in district, or a female

dummy (see row headers). All regressions include dummies for region of residence, year of birth, female, responses to current living standards, level of education, occupation, and urban. The share of the respondent’s ethnicity in the district population is also used as a control. Standardized coefficients are reported. 22 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Main results Heterogeneity Ethnic War Robustness

Robustness: Overview

Measurement: Binary war exposure; IV using alternative data; use factor analysis to aggregate outcomes. Migration: Remove the three largest cities from each country. Remove individuals living in regions where less than 10% of their ethnic group lives. Measure war at country or ethnicity

• level. Restrict to rural sample.

Specification: Enter separately at each age. Alternative functional

• forms. Alternative fixed effects. Add

rainfall. Afrobarometer: Used in several

• studies. Intra-Afrobarometer

correlations are sensible. PRIO: Used in several studies. Figure

• 1. Strongly correlated with

alternative measures of exposure.

23 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Early life war exposure: magnitudes in the literature Mechanisms and evidence

1

Introduction

2

Data and Empirical Strategy

3

Results

4

Resilience Early life war exposure: magnitudes in the literature Mechanisms and evidence

5

Conclusion

24 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Early life war exposure: magnitudes in the literature Mechanisms and evidence

Table 6. Results in the literature

Author(s) Outcome Treatment Coef SD(LHS) SD(RHS) Std Coef Annan et al Member of at least one group (women) (Table 2) Abduction as a child sol- dier

• 0.04
• Member of at least one

group (men) (Table 2) 0.01

• Antidemocratic

atti- tudes (women) (Table 2) 0.10

• Bellows, John and Ed-

ward Miguel Did you attend any com- munity meetings in the past year? (Table 3) Conflict Victimization Index 0.0775 0.49 0.34 0.054 Blattman Voted in 2005 (Table 5) Violent acts witnessed 0.04 0.5 1.7 0.136 Blattman and Annan Index

• f

psychological distress (Table 7) Index of violence experi- enced 0.15 2.4 3.1 0.194 Glenerster et al Road maintenance (Ta- ble 3) Ethnolinguistic fraction- alization

• 0.083

0.2 0.21

• 0.087

Leon Years of completed edu- cation (Table 4) Number of years exposed to violent events (early childhood)

• 0.066

2.82 1.97

• 0.046

Miguel et al Yellow cards (Table 2) Years of civil war (coun- try level) 0.0076 2.73 4.74 0.013 25 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion Early life war exposure: magnitudes in the literature Mechanisms and evidence

Mechanisms and evidence

The long run effects of violence are often small even if the short-run effects are large:

The effects of bombing campaigns against Japan, Nazi Germany, and Vietnam have all been erased within a few decades (Brackman, Garretsen, Schramm, 2004; Davis and Weinstein, 2002; Miguel and Roland, 2011; Waldinger, 2012). The effects of civil conflict are large in the short-run, but on average, affected countries catch up very rapidly after war ends (Cerra and Saxsena, 2008; Chen et al., 2008; Leon 2012). Even among child combatants, social reintegration and psychological recovery can occur quickly (Annan et al., 2010; Blattman, 2009).

Anthropology literature suggests that, on average, individuals exposed to war are resilient (Betancourt and Khan, 2009; Eggerman and Panter-Brick, 2010) For many outcomes, interacting with age shows moderate main effect that is dissipated over time (Table 3).

26 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion

1

Introduction

2

Data and Empirical Strategy

3

Results

4

Resilience

5

Conclusion

27 / 28

Introduction Data and Empirical Strategy Results Resilience Conclusion

Conclusion

Exposure to war in childhood has small or zero effects on a wide range of measures of political actions and attitudes. Our results are robust to several alternative specifications. They are not due to problems with either the data on war exposure or political

• utcomes.

Our results are consistent with other small effects from the literature

• n war exposure. The exception is the impact of child soldiering.

Our results seem sensible when considering the population as a whole, and the literatures on resilience and recovery to shocks over the long run.

28 / 28

Delivering justice to the poor

Theory and experimental evidence from Liberia Justin Sandefur1 Bilal Siddiqi2

1Center for Global Development 2Stanford University

Summary

The poor trade off customary system’s efficiency for formal law’s content

◮ Develop model of forum shopping under legal pluralism ◮ Predict demog. groups facing customary bias go to formal or nowhere ◮ Test model using data on 4,500 disputes in Liberia

Need solution that addresses this trade-off

◮ Paralegals trained in formal law act as advocates & mediators ◮ Offer low-cost, remedial justice a la customary law

RCT of paralegal intervention finds case-specific and downstream benefits

◮ Individual randomization of 420 “potential clients” across 76 villages ◮ High take-up from disputants who would have gone to formal or none ◮ Evidence of significant short-term downstream benefits

Summary

The poor trade off customary system’s efficiency for formal law’s content

◮ Develop model of forum shopping under legal pluralism ◮ Predict demog. groups facing customary bias go to formal or nowhere ◮ Test model using data on 4,500 disputes in Liberia

Need solution that addresses this trade-off

◮ Paralegals trained in formal law act as advocates & mediators ◮ Offer low-cost, remedial justice a la customary law

RCT of paralegal intervention finds case-specific and downstream benefits

◮ Individual randomization of 420 “potential clients” across 76 villages ◮ High take-up from disputants who would have gone to formal or none ◮ Evidence of significant short-term downstream benefits

Summary

The poor trade off customary system’s efficiency for formal law’s content

◮ Develop model of forum shopping under legal pluralism ◮ Predict demog. groups facing customary bias go to formal or nowhere ◮ Test model using data on 4,500 disputes in Liberia

Need solution that addresses this trade-off

◮ Paralegals trained in formal law act as advocates & mediators ◮ Offer low-cost, remedial justice a la customary law

RCT of paralegal intervention finds case-specific and downstream benefits

◮ Individual randomization of 420 “potential clients” across 76 villages ◮ High take-up from disputants who would have gone to formal or none ◮ Evidence of significant short-term downstream benefits

Outline

Model Forum shopping (observational variation) Legal aid (experimental evaluation)

Outline

Model Forum shopping (observational variation) Legal aid (experimental evaluation)

Model

Assumption #1. The custom is biased against certain social and demographic groups. Assumption #2. Formal justice is ‘leaky’—i.e., utility lost by the defendant does not fully accrue to the plaintiff

Model

Assumption #1. The custom is biased against certain social and demographic groups. Assumption #2. Formal justice is ‘leaky’—i.e., utility lost by the defendant does not fully accrue to the plaintiff

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff N C F

r r r

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff N C F

r r r r

• ❅

❅ ❅

Chief r C r C

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge r F r F

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F)

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff N C F

r r r r

• ❅

❅ ❅

Chief r C r C

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge r F r F

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F)

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r

4,586 Disputes

r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff N C F

r r r r

• ❅

❅ ❅

Chief r C r C

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge r F r F

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F) 2,081 Households

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r

4,586 Disputes

r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff 58%=N 38%=C F=4%

r r

1,736 Cases

r

180 Cases

r

• ❅

❅ ❅

Chief r C r C

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge r F r F

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F) 2,081 Households

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r

4,586 Disputes

r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff 58%=N 38%=C F=4%

r r

1,736 Cases

r

180 Cases

r

• ❅

❅ ❅

Chief 75%=r C r C=25%

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge 70%=r F r F=30%

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F) 2,081 Households

Game tree

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r

4,586 Disputes

r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff 58%=N 38%=C F=4%

r r

1,736 Cases

r

180 Cases

r

• ❅

❅ ❅

Chief 75%=r C r C=25%

r

=0.0uP(r C)=1.3

r

=0.0uP(r C)=1.6

r

• ❅

❅ ❅

Formal judge 70%=r F r F=30%

r

=0.0uP(r F)=1.1

r

=0.0uP(r F)=1.5 =0.0uD(r C)=1.5 =0.0uD(r C)=0.9 =0.0uD(r C)=2.0 =0.0uD(r F)=-.3 2,081 Households

Outline

Model Forum shopping (observational variation) Legal aid (experimental evaluation)

Predictions for forum shopping behavior and outcomes

• 1. Dispute types:

Plaintiffs are more likely to report the dispute (j = C, F) when harm h is high (e.g., violent crimes or land disputes)

• 2. Plaintiff characteristics:

Ceteris paribus, disadvantaged plaintiffs are more likely to report to the customary system.

• 3. Plaintiff-defendant pairings:

Plaintiffs facing greater bias β in the customary system (e.g., women suing men), will either not report, or go to the formal system.

• 4. Payoffs:

The customary system provides greater aggregate welfare than the formal system, in that the sum of the utilities of P and D is higher.

Prediction #1.

Land Debt Labor Property Assault Rape/GBV Murder Family

5 10 15 20 25 Reporting: % to formal sector 10 20 30 Frequency: % of total disputes

Predictions for forum shopping behavior and outcomes

• 1. Dispute types:

Plaintiffs are more likely to report the dispute (j = C, F) when harm h is high (e.g., violent crimes or land disputes)

• 2. Plaintiff characteristics:

Ceteris paribus, disadvantaged plaintiffs are more likely to report to the customary system.

• 3. Plaintiff-defendant pairings:

Plaintiffs facing greater bias β in the customary system (e.g., women suing men), will either not report, or go to the formal system.

• 4. Payoffs:

The customary system provides greater aggregate welfare than the formal system, in that the sum of the utilities of P and D is higher.

Predictions for forum shopping behavior and outcomes

• 1. Dispute types:

Plaintiffs are more likely to report the dispute (j = C, F) when harm h is high (e.g., violent crimes or land disputes)

• 2. Plaintiff characteristics:

Ceteris paribus, disadvantaged plaintiffs are more likely to report to the customary system.

• 3. Plaintiff-defendant pairings:

Plaintiffs facing greater bias β in the customary system (e.g., women suing men), will either not report, or go to the formal system.

• 4. Payoffs:

The customary system provides greater aggregate welfare than the formal system, in that the sum of the utilities of P and D is higher.

Predictions #2 and #3.

Formal over Customary None over Customary Formal over None Defendant - Plaintiff (β) (+) (+) (?) Gender bias 1.111***

• 0.276***

1.387*** (0.262) (0.074) (0.261) Income bias 0.550** 0.268** 0.282** (0.201) (0.103) (0.199) Elite bias 0.400 0.040 0.360 (0.197) (0.119) (0.203) Ethnic bias 0.092

• 0.015

0.107 (0.205) (0.085) (0.204) Plaintiff (u0) (+) (+) (?) Male 1.446***

• 0.195***

1.641*** (0.318) (0.107) (0.318) Non-farm employment 1.050 0.343 0.708 (0.271) (0.142) (0.269) Related to chief 0.732** 0.365** 0.367** (0.249) (0.135) (0.252) Ethnic majority

• 0.191

0.009

• 0.201

(0.239) (0.114) (0.240)

Predictions for forum shopping behavior and outcomes

• 1. Dispute types:

Plaintiffs are more likely to report the dispute (j = C, F) when harm h is high (e.g., violent crimes or land disputes)

• 2. Plaintiff characteristics:

Ceteris paribus, disadvantaged plaintiffs are more likely to report to the customary system.

• 3. Plaintiff-defendant pairings:

Plaintiffs facing greater bias β in the customary system (e.g., women suing men), will either not report, or go to the formal system.

• 4. Payoffs:

The customary system provides greater aggregate welfare than the formal system, in that the sum of the utilities of P and D is higher.

Prediction # 4.

Non-farm Related Ethnic Benchmark Male employment to chief majority Plaintiff .43 .45 .43 .44 .40

(.12)∗∗∗ (.12)∗∗∗ (.12)∗∗∗ (.12)∗∗∗ (.12)∗∗∗

Plaintiff ×β .11 .08 .10

• .25

(.09) (.17) (.13) (.12)∗∗

Plaintiff × Formal

• .16
• .25
• .15
• .28
• .12

(.21) (.21) (.21) (.25) (.21)

Plaintiff × Formal ×β 1.68

• .29
• .40

.54

(.56)∗∗∗ (.51) (.41) (.65)

Defendant .51 .53 .51 .60 .52

(.16)∗∗∗ (.16)∗∗∗ (.16)∗∗∗ (.17)∗∗∗ (.16)∗∗∗

Defendant ×β

• .03
• .11
• .34
• .10

(.18) (.28) (.23) (.29)

Defendant × Formal

• 1.13
• 1.17
• 1.14
• 1.20
• .91

(.28)∗∗∗ (.28)∗∗∗ (.28)∗∗∗ (.32)∗∗∗ (.30)∗∗∗

Defendant × Formal ×β

• 1.15
• .05

.22

• 1.36

(.57)∗∗ (.52) (.54) (.75)∗

Observations 940 940 940 940 940

• Adj. R2

0.070 0.082 0.067 0.075 0.070

Outline

Model Forum shopping (observational variation) Legal aid (experimental evaluation)

Experimental impact evaluation

Mobile paralegals

◮ Free legal aid: content of formal law, in more customary

trappings

◮ Summer of 2011, intensified round of ‘awareness raising’ ⇒

420 new cases across 76 villages. Individual-level randomization

◮ Panel of individuals ◮ Three-month (i.e., short-term) impacts

Add legal aid to the game: low φ, β = 1/2

❜

• ❅

❅ ❅

Defendant h = 0 h > 0

r r r ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ ✏ PPPPPPPPPPP P

Plaintiff N C F

r r r r

• ❅

❅ ❅

Chief r C r C

r

uP(r C)

r

uP(r C)

r

• ❅

❅ ❅

Formal judge r F r F

r

uP(r F)

r

uP(r F) uD(r C) uD(r C) uD(r F) uD(r F)

Predictions for the RCT

• 5. Take-up:

A disproportionate share of disputes taken to paralegals would, absent legal aid, not have been taken to any forum.

• 6. Treatment effects:

Legal aid will increase plaintiffs’ payoffs.

• 7. Heterogeneous treatment effects:

Plaintiffs facing disadvantageous bias in the customary system (e.g. women, subsistence farmers, minorities) will benefit more from legal aid.

Prediction #5. Take-up

61.8 34.6 3.6 49.6 14.8 35.7

0% 20% 40% 60% Full baseline sample Control group ex post None Customary Formal

Predictions for the RCT

• 5. Take-up:

A disproportionate share of disputes taken to paralegals would, absent legal aid, not have been taken to any forum.

• 6. Treatment effects:

Legal aid will increase plaintiffs’ payoffs.

• 7. Heterogeneous treatment effects:

Plaintiffs facing disadvantageous bias in the customary system (e.g. women, subsistence farmers, minorities) will benefit more from legal aid.

Predictions for the RCT

• 5. Take-up:

A disproportionate share of disputes taken to paralegals would, absent legal aid, not have been taken to any forum.

• 6. Treatment effects:

Legal aid will increase plaintiffs’ payoffs.

• 7. Heterogeneous treatment effects:

Plaintiffs facing disadvantageous bias in the customary system (e.g. women, subsistence farmers, minorities) will benefit more from legal aid.

Prediction #7. Heterogeneous treatment effects

Female Farmer Minority Uneducated Refugee Migrant x Z x P x Z x P x Z x P x Z x P x Z x P x Z x P Case results 0.236

• 0.091
• 0.026

0.053 0.041

• 0.034

(0.155) (0.218) (0.299) (0.163) (0.156) (0.146) Justice system 0.144*** 0.171*** 0.232 0.149*** 0.164*** 0.149*** (0.047) (0.044) (0.179) (0.051) (0.045) (0.044) Attitudes

• 0.055

0.033

• 0.122

0.084

• 0.030

0.029 (0.081) (0.065) (0.164) (0.089) (0.082) (0.071) Behavior 0.171**

• 0.001
• 0.254

0.064 0.039

• 0.018

(0.066) (0.064) (0.171) (0.095) (0.065) (0.064) Household wellbeing 0.122** 0.063

• 0.169

0.045 0.099 0.066 (0.059) (0.047) (0.162) (0.071) (0.076) (0.065)

Summary

The poor trade off customary system’s efficiency for formal law’s content

◮ Develop model of forum shopping under legal pluralism ◮ Predict demog. groups facing customary bias go to formal or nowhere ◮ Test model using data on 4,500 disputes in Liberia

Need solution that addresses this trade-off

◮ Paralegals trained in formal law act as advocates & mediators ◮ Offer low-cost, remedial justice a la customary law

RCT of paralegal intervention finds case-specific and downstream benefits

◮ Individual randomization of 420 “potential clients” across 76 villages ◮ High take-up from disputants who would have gone to formal or none ◮ Evidence of significant short-term downstream benefits