Week 04: Civil Wars

# Week 04: Civil Wars
## Violence against Civilians
<html>
<div style="float:left">

</div>
<hr color='#EB811B' size=1px width=800px>
</html>
### Danilo Freire
### 1st October 2019

---

<style>

.remark-slide-number {
  position: inherit;
}

.remark-slide-number .progress-bar-container {
  position: absolute;
  bottom: 0;
  height: 6px;
  display: block;
  left: 0;
  right: 0;
}

.remark-slide-number .progress-bar {
  height: 100%;
  background-color: #EB811B;
}

.orange {
  color: #EB811B;
}
</style>

# Last week we saw that...

* Grievance are constant, so civil wars need another triggering factor

* Opportunity: rebel groups revolt when they're able to

* Lootable resources, money from diasporas and hostile governments, and low costs of fighting (poverty) explain civil war onset 
]
---

# Last week we saw that...

* Also proxied by GDP, like C&H's opportunity variable

* Ethnic and religious heterogeneity don't matter once you control for income

* Factors that favour insurgency (mountainous terrain, oil, unstable regimes) increase the risk of civil wars
]

---

# Last week we saw that...

* Ethnicity is useful because politicians mobilise it

* Secession, rebellion, or group infighting

* Excluded population fight for the first two, included groups for the latter (sometimes for secession too)
]

---

# Last week we saw that...

* GDP per capita and population are the best predictors of civil conflicts

* Statistical significance is not the same as practical or predictive significance
]

---

# Violence against Civilians

<html><div style='float:left'></div><hr color='#EB811B' size=1px width=720px></html> 
---

# Kalyvas (1999)

# Some context

* Islamist groups vs the government

* Onset: government cancelled 1991 elections after they thought the Islamic Salvation Front (FIS) would win

* FIS was banned and members arrested, Islamist guerrillas rebelled against the government

* Two main rebel groups: Islamic Armed Movement (MIA) and Armed Islamic Group (GIA)
]
---

# Some context

* GIA declared war against FIS

* Pro-Army party won the next election (1997)

* MIA surrended and GIA were hunted down

* 1996-1999 peak violence; civil war ends in 2002 (government victory)
]
---

# Some context

* 44,000 to 100,000 deaths

* Salafist Group for Preaching and Combat (GSPC), a GIA dissident group, later affiliated with Al-Qaeda and continue fighting in the Maghreb
]
---

# Kalyvas (1999)

* Massacres are not acts of barbarism, but a war strategy

* Violence maximises civilian support under constrains

* Illustrates his case using the Algerian Civil War as an example
]
--

---

# Kalyvas (1999)

.font150[
* **Questions**:
	- Should political organisations kill civilians randomly?
	- Specifically, why would *insurgents*, who tend to be weaker, kill civilians?
]
--
.font150[
* *No*]
--
.font150[
* *To find infiltrated enemies and prevent civilian defection*

* Repression is effective:
	- Cheaper than selective benefits
	- Useful in areas and periods of contested rebel control
]
---

# Targets

* Civilians associated with other insurgents

* Defectors
]
--
.font150[
* **Question**: how to identify the enemies?
]
--
.font150[
* Access to reliable local information
]
---

# A theory of selective violence

* But rebels and gov't forces often don't have full control oves the territory

* They use coercion to elicit information from citizens and punish the enemies

* Constraints faced by denouncers provide a good way to model the process

* Key constraint is *the likelihood of retaliation*
]
---

# A theory of selective violence

.font150[
* In areas over which the rebels/gov't have full control, they can easily retaliate citizens

* In contested areas, citizens can switch sides and it's hard for parties to use selective violence

* Denunciation leading to selective violence is most likely where one actor exercises *dominant but incomplete control*

* **Question**: in which type of conflict actors *might not* need to solicit information to citizens?
]
--
.font150[
* Ethnic civil wars: private information is often available 
]
---

# Territorial control and violence

# Territorial control and violence

.font150[
* Violence is more likely in zones 2 and 4 
	- Dominant but incomplete control by the rebels
	- The same for incumbents

* As they lose control over the population, actors compensate by increasing the level of brutality

* *This process can happen elsewhere*, not only in Algeria

* The level of violence probably does *not* come from ideology, but from territorial control
]
---

# Questions?

<html><div style='float:left'></div><hr color='#EB811B' size=1px width=720px></html> 
---

# Humphreys & Weinstein (2006)

# Humphreys & Weinstein (2006)

.font130[
* "...the extent of civilian suffering varies across conflicts, over time within a conflict, and across geographic regions in countries that experience violence"

* Three factors:
	- Incentives for groups to be violent
	- Territorial control
	- *Poor internal cohesion*

* "Units composed of members with private goals, which organize into ethnically heterogeneous groupings and lack internal mechanisms to discipline behavior, exhibit relationships with noncombatants characterized by coercion and abuse"
]
---

# Humphreys & Weinstein (2006)

* Too much violence is counterproductive: "kills the goose that lays the golden egg"

* Too *little* violence is also counterproductive: rebels will lose without information and conquest

* There exist an optimal point where violence is effective: restrained violence + public goods production

* *When do groups find that equilibrium?*
]
---

# Hypotheses

.font150[
* H1: Abuse levels are likely to be higher in poorer areas 
	- Rationale: Fewer resources to extract in the long run, less need for constrain]
--

.font150[
* H2: Lower levels of abusiveness should be apparent in geographic zones in which warring factions are fighting in their home communities
	- Repeated interactions, social connections
]
---

# Hypotheses

.font140[
* H3: Lower levels of abusiveness should be apparent in geographic zones in which warring factions are predominantly of the same ethnic group as members of the local community
	- Same: repeated interactions, social connections
]
--

.font140[
* H4: Greater levels of contestation will be associated with higher levels of abuse
	- More competition, more violence: monopolies are efficient]
--

.font140[
* H5: Warring factions that recruit combatants using offers of private benefits are more likely to exhibit high levels of civilian abuse
	- Private incentives are not the same as public incentives: greed
]
---

# Hypotheses

* H7: More ethnically homogenous factions should exhibit lower levels of civilian abuse
	- Same: repeated interactions, social connections
]
--

.font150[
* H8: Factions with tight disciplinary structures are likely to be less abusive of civilian populations
	- Control over rogue soldiers]
---

# Conflict in Sierra Leone

* Blood diamonds

* Majority of crimes perpetrated by the Armed Forces Revolutionary Council (AFRC) and the Revolutionary United Front (RUF), supported by Charles Taylor (Liberia)

* UN sent a mission in 1999; it failed

* British Armed Forces intervened in 2000: conflict over by 2002
]
---

# Data

* Asked questions about characteristics of armed groups during all stages of the combat 
	- Problem: selective memory

* Units of analysis: 
	- same faction and chiefdom (FCH)
	- subfaction and district (SFD)
]
---

# Results

# Results

# Discussion

* Strong support for the hypothesis that internal discipline decreases violence

* Other variables have little or no effect
]
---

# Discussion

* Weak relationship between the extent of combatant–community ties and patterns of abuse

* Wealth is also nor correlated with more violence
	- Little variation in wealth, diamond wealth not captured by indices?

* No support for territorial control (Kalyvas)
]
---

# Questions?

<html><div style='float:left'></div><hr color='#EB811B' size=1px width=720px></html> 
---

# Balcells (2010)

# Balcells (2010)

* Many civil wars are fought for political reasons: regime change, self-determination, etc

* F&L equate all civil wars with guerrilla wars

* B&K affirm that about 50% of civil wars are conventional wars, such as the Spanish Civil War

]
---

# Balcells (2010)

* Irregular wars: contested areas

* **Hypothesis**: While violence in irregular wars in connected with military competition to gain territory, in conventional civil wars violence is caused by *prewar political cleavages*
]
---

# Spanish Civil War

# Theory

* Civilians are generally isolated from the battlefield
	- Indirect violence (bombings) is cost-effective

* In guerrilla wars, information about enemies is difficult
	- Selective violence

* Why use selective violence in conventional civil wars?

* Rebel groups target *highly mobilised supporters of the rival* 
]
---

# Hypotheses

* **Rivalry**:

* H1: The greater prewar competition approaches parity between groups, the higher the levels of violence perpetrated by the armed group controlling that locality (*rivalry*)
	- If the enemy has a large majority or a large minority, civilians restrain their armies
	- When groups have parity, strong supporter of each side push armies to fight
]
---

# Hypotheses
.font150[
* **Revenge**:

* H2: The higher the levels of violence perpetrated by an armed group controlling a locality at t1, the higher the levels of violence perpetrated by the rival armed group in the same locality during the subsequent time period (*revenge*)
	- Tit-for-tat 
]
---

# Results

# Results

# Discussion

* Presence of CNT (labour union) and Church centres increase violence

* Montainous terrain, proximity to France and to the sea decrease executions (exit)
]
---

# Discussion

* In locatities with high parity in the prewar, violence is used to change the status quo

* Local collaboration is not endonegous with territorial control, but to prewar mobilisation

* Grievances can explain civil war violence
]
---

# Questions?

<html><div style='float:left'></div><hr color='#EB811B' size=1px width=720px></html> 
---

# Cohen (2013)

# Cohen (2013)

* Three explanations for wartime sexual violence:
	- Greed
	- Ethnic hatred
	- Gender inequality

* It appears that those explanations are not true

* New hypothesis: *combatant socialisation*
]
---

# Rape and combatant socialisation

* *Rape as a socialisation device*

* Many combatants are abducted, so they create bonds of esteem by violating women

* Men are more likely to be killed during wartime, women and girls more likely to suffer nonlethal violence and to experience long-run consequences of war
]
---

# Hypotheses

.font150[
* **Opportunity/greed**: 
]
--
.font150[
*  H1: State collapse is correlated with higher state-perpetrated and insurgent-perpetrated levels of wartime rape
	- Collapse of social norms
]
--
.font150[
* H2: Rape by insurgents is more likely in conflicts whereinsurgent groups rely on material resources.
	- "Individual incentives"
]
---

# Hypotheses

.font150[
* **Ethnic hatred** 
]
--
.font150[
* H3: Ethnic wars are associated with conflict-wide rape
	- Humiliation 
]
--
.font150[
* H4: States or insurgents who perpetrate genocide are more likely to commit wartime rape than those who do not
	- Inflict bodily or mental harm to other group
]
--
.font150[
* H5: Secessionist wars, especially those featuring ethnic cleansing, are associated with insurgent-perpetrated rape
	- Ethnic cleansing or forced expulsion
]
---

# Hypotheses

.font150[
* **Gender inequality**
]
--
.font150[
* H6: Greater gender inequality is associated with conflict-wide rape
	- Acceptance of violence against women
	- Shames not only the victim, but male relatives
]
---

# Rape as combatant socialisation tool

* "Rite of passage"

* Creates and solidifies social status for men

* Act of camaraderie: increases group cohesion

* Gang rapists are less likely to have committed other crimes and are considered less pathological than private rapists
]
---

# Rape as combatant socialisation tool

.font150[
* H7: Insurgent groups that depend on abduction as a recruitment mechanism are more likely to perpetrate rape than groups that use other, more voluntary methods of recruitment

* H8: States that depend on pressganging as a recruitment mechanism are more likely to perpetrate rape than groups that use other, more voluntary methods of recruitment
]
---

# Results

# Discussion

* But supported for rebel-group rape

* Insurgent contraband funding is positively associated with rape

* Ethnic hatred arguments are not supported for conflict-wide or rebel-group rape

* Gender inequality also not supported (not shown)

* Maybe GI was already so widespread that it doesn't explain sexual violence 
]
---

# Discussion

# Discussion

# Discussion

* Associated with state collapse

* Lootable resources have a positive influence

* Common suppositions were not supported
]
---

# Questions?

<html><div style='float:left'></div><hr color='#EB811B' size=1px width=720px></html> 
---

# See you next week!