Civil wars are crucial events in world history

They are hard to define:

• Partisan bias (favour victims)
• Political bias (war is not politics)
• Urban bias (costly information, big narratives)
• Selection bias (overaggregation, lack of context)

Violence is both an outcome and an instrument of civil wars

Even barbaric acts can be rational

Quantitative measures are somewhat arbitrary

PRIO/UCDP and ACLED

• 25/1000 annual battle-related deaths
• "contested incompatibility over territory or government and one of the parties is the state"

Three waves of civil war

• Cold War: class-based conflicts, peasant rebellions
• 1991-2003: ethnic conflicts
• 2003-present: radical Islamism, but religion might not be the main cause

C&H argue political scientists put too much emphasis on motivations (grievances)

Grievances are more or less constant, so why do civil wars break only at a specific point in time?

Window of opportunity: people rebel when they can

Greed: atypical circumstances that can generate large returns

Yet C&H agree they cannot differentiate between opportunity and viability to express grievances

Ratio of primary exports to GDP

Relationship is non-linear

Countries with very low and very high ratios do not have civil wars

Question: why?

E.g.: Tamil Tigers funded by American-resident Tamils

Immigrants provide money (and skills?) rebel groups wouldn't have otherwise

Proxied by % of immigrants living in the US

• Algerian civil war (rebels supported by USSR/ France by the US)
• Soviet-Afghan war (Sunni Mujahideen supported by US/PAK/CN, etc; Shia Mujahideen supported by Iran)
• Syrian civil war (a mess!)

1) Income is low: GDP per capita

2) Weapons are cheap: previous wars

3) Government is weak: montainous terrain

4) Social cohesion: ethnic diversity

Ethnic and religious hatred: ethnic fractionalisation

Political repression: non-democracies

Political exclusion: ethnic dominance (majority group comprises 45-90% pop)

Economic inequality: Gini index, ratio top-to-bottom income

Factors that influence the opportunity for rebellion:

• Money from commodities
• Diaspora
• Low GDP
• Dispersed population

No evidence for grievance-based theories

Only ethnic dominance has a positive effect

Question: what do you think? Is their explanation convincing?

No difference between rebels and criminals

GDP, democracy, primary exports change very slowly over time... what triggers conflict onset?

Urban bias: no mention of rural dynamics

Passive role of the state in the conflict

Crude proxies for greed and grievances

Most cited paper in political science in the last 15 years

Remarkable data collection effort

Four main points:

• The end of the Cold War did not cause civil wars
• Controlling for income, ethnicity and religion don't matter
• Little support for grievance-based theories
• Factors that explain insurgency are the most relevant

Different from C&H, F&L argue that low GDP proxies for state capacity

Conflicts that kill at least 1,000 people, at least 100 per year, rebels or government forces

Include colonial wars

F&L are cautious about their data on empires

They analyse the data both including and excluding colonial wars

F&L criticise the idea of "clash of civilisations"

Deep-rooted ethnic grievances are not enough to explain civil war onset

Sceptical of "modernisation theory": limits to upward mobility cause people to revolt

Question the idea that discrimination (ethnic or economic favouritism to other groups) leads to conflict

Focus on small guerrilla wars

Guerrilla groups are weak compared to the government

Factors that facilitate rebel group survival are very important

• Montainous terrain (hideouts)
• Local knowledge
• Rural base
• Weak state governance

The high number of civil wars in the 1990s is the result of accumulation of previous conflicts, not the end of the Cold War

Ethnicities and grievances do not explain why internal wars occur

Factors that favour insurgency do: state weakness, low GDP, instability, and large population

Practical implications:

• Promoting democracy abroad doesn't work
• Cultural dialogue doesn't work either
• Little foreign states can do: GDP, state capacity take time

C&H: homo oeconomicus goes to war

Rebels are essentially criminals: profits from looting

Question: natural/lootable resources can fuel grievances too, can't they?

Resource-rich countries might have higher inequality, forced migrations, trade shocks, etc

Diasporas can also provide welfare to local communities and therefore reduce the likelihood of conflict

F&L: unclear which aspect of state capacity decreases civil war risk: Inclusive institutions, shared power, armed forces, economic redistribution?

Impossible to adjudicate between their theory (state capacity, proxied by GDP per capita) and C&H's (lootable resources, proxied by.... GDP per capita!)

As with C&H, national-level data obscures important within-country dynamics

Ethnicity does play a role in civil war outbreaks

Qualify previous literature

Offer better data on ethnic groups

Establish new mechanisms that link group grievances to conflicts

Focus on political dynamics of ethnic exclusion and competition

Institutionalist: political structures create incentives for players to act strategically

Configurational: the same institution provides different incentives according to the distribution of power

Ethnicity matters because the nation state uses it for legitimacy

Politicians have incentives to favour their co-ethnics

Ethnic favouritism is more likely in poor, young states (which need more legitimacy)

Secession: changing the territorial boundaries of a polity and can be pursued by both excluded and included groups

Rebellion: excluded segments of a population fight to shift the boundaries of inclusion

Infighting: elite disputes for the spoils of government

Predictions:

• ethnic exclusion breeds conflict
• more power-sharing increase conflict (coalitions)
• large states and those under indirect rule rebel more

"The likelihood of armed confrontation increases as the center of power becomes more ethnically segmented and as greater proportions of a states population are excluded from power because of their ethnic background"

"These conflicts are even more likely in incoherent states where the population is not accustomed to direct rule by the political center"

"Ethnicity is not an aim in itself, but the organizational means through which individuals struggle to gain access to state power"

Good idea to focus on ethnicity and power dynamics

However, still treats ethnicity as a fixed category

Ethnicity and conflict can be endogenous

Urban bias, maybe?

Do systemic factors play a role in civil war dynamics?

More specifically, what is the role of the international system in civil wars?

"Technologies of rebellion": ways that civil conflicts are fought

Disaggregating the types of conflict

The end of the Cold War caused a decline in irregular wars

Irregular warfare: guerrillas

Conventional warfare: army and rebels have similar power

Symmetric non-conventional (SNC) warfare: states unable/unwilling to fight ("primitive" wars)

Civil wars were understood as proxy wars

End of the Cold War brought changes to the int'l system:

• End of multiethnic states
• Emergence of new states
• Cheap weapons from the former USSR
• Weakening of client states
• No legitimising principle to the state (link with Wimmer et al.)

Contest F&L's results that the Cold War has no effect

Indeed the "technologies of rebellion" had been overlooked

However, the model doesn't seem very convincing

Almost nothing is significant; bad control variables

No testing of any mechanism:

• Material support
• Ideological cohesion
• Access to weapons

SNC is a poorly-defined category

Statistical models explain what has already happened (duh!)

How useful are they to predict what might happen in the future?

In-sample vs out-of-sample testing

Predictive accuracy: Possible means for testing the validity of a theory

False positives and false negatives

• Every probability above 0.5 is considered positive in the model
• There is nothing special about that number

Question:

• You are the leader of your country and you see that country X has 50% chance of going to war next year. If you are right, a civil war can be averted. But if you are wrong, you will invade another country, your citizens will die, and conflict might spread. What probability would make you go to war?

Statistically significant variables have little predictive power

How can we use these models to predict new conflicts?

Factors that explain previous civil wars may not explain future ones

Prediction can help us build more robust (although not necessarily more efficient) models for civil war prevention

Civil wars happen in poor, weak states

Windows of opportunity are more important than old grievances

Ethnic rivalries can lead to civil war outbreak, but under certain conditions

The Cold War has changed the way civil wars are fought

Many of the variables we know explain past wars, but can they explain future ones?