Conflict and Development II

The Consequences of Civil War

Carolina Torreblanca

University of Pennsylvania

Global Development: Intermediate Topics in Politics, Policy, and Data

PSCI 3200 - Spring 2026

Agenda

  1. Last Week: What Causes Conflict?
  2. Blattman and Annan (2010): The Consequences of Child Soldiering
  3. Violence and Political Participation
  4. Miguel, Saiegh, and Satyanath (2011): Civil War and the Soccer Pitch

Last Week

What Causes Conflict?

  • Poverty lowers the opportunity cost of fighting (Dal Bo and Dal Bo)
  • Wealth can also cause conflict if it increases the returns to fighting: rapacity
  • Other explanations: grievance, ethnic fractionalization, weak institutions
  • Conflict is both a cause and a consequence of poverty complicating causal analyses
  • We used instrumental variables to try to isolate the causal effect of income on conflict

Today: the other side. What does conflict do to people?

What Does Conflict Destroy?

Three Kinds of Destruction

  • Physical capital: roads, bridges, factories, livestock
  • Human capital: education interrupted, health damaged, skills lost
  • People themselves: psychology, behavior, trust

Most of what we know about the first two comes from macro data. To understand the third, we need micro evidence: what happens to individuals?

Blattman and Annan (2010)

Setting

  • Northern Uganda, late 1980s onwards
  • The Lord’s Resistance Army (LRA), led by Joseph Kony, waged a brutal insurgency
  • The LRA abducted children and young adults to serve as soldiers, porters, and laborers
  • Tens of thousands abducted, mostly boys aged 10-24

What do you think happens to these children in the long run?

Research Question

What are the long-run consequences of forced military service on education, economic outcomes, and psychological well-being?

Hypotheses

  • H1 (human capital): abduction interrupts education and skill formation \(\rightarrow\) lower earnings, worse economic outcomes
  • H2 (psychology): exposure to extreme violence \(\rightarrow\) lasting psychological distress, aggression, difficulty reintegrating

What would make this hard to study? Why can’t we just compare soldiers to non-soldiers?

The Identification Problem

  • You can’t just compare ex-combatants to non-combatants
  • People who fight may be systematically different: poorer, less educated, more aggressive
  • Selection bias: the same things that lead to recruitment may also affect outcomes

Strategy

The LRA’s abduction strategy was essentially random:

  • “Targets were generally unplanned and arbitrary; they raided whatever homesteads they encountered” (p. 887)
  • Strategy: abduct first, sort out later: keep all adolescent and young adult males
  • Rural Acholi households live in isolation, making them vulnerable to roving raiding parties
  • Within a community, who was taken was largely a matter of bad luck

What assumption does this rely on? What could go wrong?

Data

  • Survey of 741 male youth aged 14-30 in 2005-2006
  • 8 sub-counties in two war-affected districts (Kitgum and Pader)
  • Compared abductees to non-abductees within the same communities
  • Outcomes: years of education, literacy, daily earnings, employment type, psychological distress

Results: Human Capital

  • Abductees had 0.75 fewer years of education: a 10% reduction
  • Were 15 percentage points less likely to be functionally literate
  • Largest effects on the youngest abductees: those taken during critical schooling years

Why would age at abduction matter?

Results: Economic Outcomes

  • Abductees’ wages were 33% lower
  • Significantly less likely to be engaged in skilled work
  • More likely to be in subsistence agriculture
  • The education gap explains much (but not all) of the earnings gap

If education doesn’t explain the full gap, what else could?

Mechanism: Decomposing the Earnings Gap

  • Education accounts for a large share of the earnings difference
  • But psychological distress independently predicts worse outcomes, even conditional on education
  • Those who witnessed or committed more violence during captivity had worse outcomes regardless of schooling
  • It’s not just lost years of school: violence itself leaves a mark

Results: Psychological Costs

  • Abductees reported significantly higher levels of psychological distress
  • Higher rates of aggression and social difficulties
  • The intensity of violence experienced matters more than the duration of abduction
  • Someone abducted for 6 months who saw killings had worse outcomes than someone abducted for 2 years who did not

The Surprising Finding

Violence and Political Participation

In a companion paper, Blattman (2009) uses the same natural experiment in northern Uganda and finds something unexpected:

  • Abductees were more likely to vote than non-abductees
  • More likely to be community leaders
  • More engaged in collective action

Development outcomes worsen, but political participation increases

Why?

  • Exposure to violence may change people’s relationship to the state: they want to shape the political order that failed to protect them
  • War creates shared identity among survivors and a sense of collective purpose
  • Similar to what Bateson (2012) found with crime: victimization can be a mobilizing experience

How can the same experience destroy someone’s economic prospects and make them MORE politically engaged? What does that tell us about what conflict does to people?

Miguel, Saiegh, and Satyanath (2011)

Motivation

Blattman and Annan showed conflict changes people psychologically: more distress, more aggression

But that evidence comes from within the conflict zone

Can we detect behavioral consequences of war in a completely different setting, years later, thousands of miles away?

Setting

  • Professional soccer: people from very different countries compete under identical rules
  • Same referees, same leagues, same competition
  • Violent behavior (fouls, yellow and red cards) is precisely recorded
  • Skill (goals, assists) is also recorded separately

Research Question

Do players who grew up in countries experiencing civil war exhibit more violent behavior on the soccer pitch?

Before we go on: what do you think? What would you predict, and why?

Hypothesis

H1: civil war exposure during formative years increases propensity toward violent behavior, even outside of war

Mechanism: growing up surrounded by violence normalizes aggression as a way of dealing with conflict and competition

What alternative stories could explain the same pattern?

Data

  • Thousands of players in top European leagues (England, France, Germany, Italy, Spain)
  • Treatment: years of civil war in the player’s home country (1980-2005)
  • Outcomes: yellow cards, red cards, goals scored
  • Controls: player position, age, minutes played, team quality, league

Strategy

  • Compare players from different countries playing in the same league
  • Two players, same team, same position, same age: one from a country with 10 years of civil war, one from a peaceful country
  • Assumption: conditional on controls, nationality is not correlated with unobserved determinants of on-field aggression

What could violate this assumption?

Descriptive Evidence

Main Results

Reading the Table

  • More years of civil war \(\rightarrow\) significantly more yellow and red cards
  • No effect on goals scored: aggression, not skill
  • Holds controlling for position, age, league, team quality
  • Robust to dropping outliers (Colombia, Israel)

Why does the “no effect on goals” matter?

Placebo Tests in Observational Data

How do you build confidence in a result you can’t experimentally verify?

  • Ask: if my story is right, what else should be true?
  • And critically: what should NOT be true?
  • If your treatment affects an outcome it shouldn’t, your story has a problem
  • If it does NOT affect that outcome, that’s reassuring

A placebo test checks whether the pattern is specific to the mechanism you claim

Placebo Tests in This Paper

  • Should be true: civil war exposure \(\rightarrow\) more cards (aggression) \(\checkmark\)
  • Should NOT be true: civil war exposure \(\rightarrow\) more goals
    • If war predicted goals, something else is going on: playing style, desperation, not aggression
    • Result: no effect on goals \(\checkmark\)
  • Should be true: effect stronger for players younger during the civil war (more impressionable)

The “no effect on goals” is the placebo: it rules out stories where war just makes you a different kind of player

Threats

  • Selection: maybe only the most aggressive players from war countries make it to European leagues
  • Referee bias: could referees card players from certain countries more harshly?
  • Omitted variables: poverty, inequality, culture may correlate with both war and aggression

How would you design a test to rule out referee bias?

Connecting the Papers

  • Blattman and Annan: conflict destroys education, earnings, psychological health
  • Blattman (2009): but it also increases political engagement
  • Miguel, Saiegh, Satyanath: and it changes behavior years later, thousands of miles away
  • The common thread: war reshapes the people who lived through it

Discussion

Discussion

  • Blattman and Annan show economic devastation; Blattman (2009) shows political mobilization. Can both be true? What does that mean for post-conflict societies?
  • The soccer paper shows behavioral effects far from the conflict zone. What does this imply for refugee and diaspora communities?
  • If the consequences of conflict are this persistent, what kinds of interventions could actually help?

Next Class

Health and Development

  • Reading: Chen et al. (2013): air pollution and life expectancy in China
  • A new causal inference method: Regression Discontinuity Design