Can the wounds of war be healed? Experimental evidence on reconciliation in Sierra Leone

Civil wars divide nations along social, economic and political cleavages, often pitting one neighbor against another. To restore social cohesion in the aftermath of war, many countries undertake truth and reconciliation efforts. We examine the consequences of one such effort in Sierra Leone, designed and implemented by a Sierra Leonean NGO. As a part of this effort, community-level forums are set up in which victims detail war atrocities, and perpetrators confess to war crimes. We use random assignment to study its impact across 200 villages. We find that reconciliation has both positive and negative consequences. On the one hand, it leads to greater forgiveness of perpetrators, and strengthens social capital: social networks were larger and people were more community-oriented, contributing more to public goods in treated villages. On the other hand, these benefits came at a substantial cost: the reconciliation process also worsened psychological health, increasing depression, anxiety and post-traumatic stress disorder in these same villages. All of the effects, both positive and negative, also persisted for nearly three years after the intervention. Our findings indicate that policymakers need to restructure reconciliation processes in ways that reduce their negative psychological costs, while retaining their positive societal benefits. ∗We would like to thank John Caulker from Fambul Tok and Libby Hoffman from Catalyst for Peace for enabling us to conduct this study of Fambul Tok’s program. Many helpful comments were provided by Eric Dixon, James Fearon, Michael Gilligan, David Laitin, Sendhil Mullainathan, David Stasavage, and participants at the Michigan Development Seminar, Princeton-Warwick Political Economy Conference, Princeton Political Economy Seminar, Inter-American Development Bank, Working Group on African Political Economy meeting, the Villars Conflict and Development Conference, APSA 2014 and Econometric Society 2015 Winter Meetings. We are also grateful to Theresa Betancourt for providing valuable input on the psychometric survey measures. For outstanding research assistance in the field, we thank Anthony Mansaray, Ali Ahmed, Jessica Creighton, and Nadia Hasham (in chronological order of involvement). This project was funded by 3ie and the J-PAL Governance Initiative, and received extensive support from Innovations for Poverty Action. We thank them, without implicating them, for making this project possible. All errors and omissions are our own. †Oxford University. jacobus.cilliers@bsg.ox.ac.uk. ‡New York University and BREAD. odube@nyu.edu §World Bank. bsiddiqi@worldbank.org