Experimentally testing the roots of poverty and violence: Changing preferences, behaviors, and outcomes


Christopher Blattman
Ramalee E. Pearson Professor of Global Conflict Studies, University of Chicago
Julian Jamison
Senior Behavioral Economist, World Bank Global INsights Initiative (GINI)
Margaret Sheridan
Assistant Professor, University of North Carolina


It is widely believed that poor and unemployed young men are more likely to fight, riot and rebel. In poor countries like Liberia, governments are especially fearful of the young, urban poor, who may be especially vulnerable to armed recruitment, rioting, or election violence. In addition to increased security, the most common policy recommendation is cash transfer and employment programs. This study uses a field experiment with high-risk young men in Liberia to answer four questions. First, is there a causal relationship between poverty and violence? While there are many theoretical rationales, microlevel evidence is sparse and inconclusive. Second, do poverty and violence have common behavioral roots? A growing literature argues that present bias, impulse control and lack of discipline can lead to poverty. But violent acts, both instrumental and emotional, may also be directly influenced by time preferences and self-control. If so, any poverty-violence correlation would be partly spurious. Third, can preferences like present bias and non-cognitive skills such as impulse control and conscientiousness be modified, thereby reducing poverty and social instability? Fourth, how do preferences and non-cognitive skills shape economic and political behavior, and how best to measure them?surveys, psychological tests, or incentivized games?

This study builds on two prior experiments that show (a) a reduced-form relationship between employment programs and violence; and (b) correlation between time preferences and both poverty and violence. They cannot, however, conclusively answer our questions. To test these, this study examines two cross-cutting interventions with 1000 poor urban street youth: (i) an 8-week Transformation Program (TP) using group cognitive behavior therapy; and (ii) an unconditional cash transfer. A random 25% of youth receive each individual intervention and a further 25% receive both; 25% remain as an untreated control group. Short-term evidence from a 100-person pilot of this intervention bears out our predictions. While the small sample results are seldom statistically significant, the directions and magnitudes are promising. The TP increases patience, impulse control, and self-discipline (measured through incentivized games as well as surveys); it increases investment and reduces poverty; and it lowers crime, political violence and interpersonal aggression. The cash transfer reduces poverty as well as aggression, and provides cautious support for two of the main causal mechanisms linking poverty to violence.

Intellectual merit: Patience, impulsivity and discipline are thought to be fixed in adults. We will provide experimental evidence of large and sustainable change through a real-world program. Second, in addition to showing these traits are malleable, their exogenous manipulation would provide clear evidence of a causal effect of time preferences on economic decision-making and poverty. The real-world nature of the experiment is important, as most aggression research is psychological, lab-based, and with US and European populations. Third, the psychological foundations of political violence are surprisingly weak or outdated, based largely on aggression research from the 1930s. The results will expand the psychological understanding of aggression, challenge conventional theories of political violence, and help initiate a field of behavioral politics, likely the first to examine violence. Fourth, there is little convincing evidence on the causal effect of poverty on violence, almost none of it experimental. This study tests this link directly, and compares interventions rooted in competing theories. Finally, there is little evidence comparing the performance of preference and non-cognitive skill measures from surveys, psychological tests, and incentivized games, and little comparison of neuropsychology and economics measures.

Broader impacts: First, the study will have a large impact on the design and targeting of youth employment and stabilization programs worldwide, especially in Africa and fragile states. Evidence on the efficacy and efficiency of unconditional cash transfers to such a high-risk group is novel and important. The transformation program will be manualized and publicly available online for ease of replication elsewhere. The intervention is designed to be highly scalable and inexpensive, and is a candidate for scaling-up in Liberia?a plan already in discussion with the government and NGOs. Second, the study will be an opportunity to provide research training and field experience to two graduate students, pre-graduate students, Liberian university graduates and masters students, and several members of the government and non-profit sectors in Liberia. Third, the comparative analysis of neopsyhological and economic measures of preferences and non-cognitive skills, as well as the comparison of survey, psychological and incentived measures, will be a broad public good informing questionnaire design and measurement across the behavioral sciences.