Trust and conflict. Experimental evidence from Yemen

This paper explores the effects of conflict on trust towards individuals in the current Yemen civil conflict. We conduct lab-in-the-field experiments to measure in-group and out-group trust with a public good game and combine these data with the Uppsala Conflict Data Program data on the number of conflict-related casualties occurring in the participants' districts.We worked with participants from all over Yemen under 35 years old working for the World Bank RAWFD (Rural Advocates Working for Development) program across the whole country. All participants were requested to fill in a survey before the start of their contract and at the end of it. We randomly assigned them to play remote-distance public goods games either prior to their participation (October 2017) or after their participation (October 2018) in the RAWFD program. Each participant played two games and was paired with both a near partner and a far-distance partner. This mean playing with participants from the same side of the conflict (near and far ingroup partners) or from the opposite side (far outgroup partner). Our main outcome of interest is the difference between the amounts invested when paired with a near partner as compared to the amounts invested when paired with a far partner, from the same or opposite region. We call this difference "Near Preference". The main set of results shows that being exposed to the civil conflict per se doesn't affect people's behavior in the public good game. What matters is with whom they play. For those exposed to the conflict, playing with an outgroup person (far partner from the opposite region), increases near preferences by 93% compared to the sample mean. This is explained by a decrease in the amounts played when confronted with an outgroup person, rather than an increase in the amount played with a near ingroup partner (player from the same region). In turn, conflict exposure decreases near preferences by 3.2% of the sample mean when playing with a far ingroup partner (far partner from the same region). This is explained by an increase in the amounts played with far ingroup partners (far partners from the same region), rather than a decrease in the amounts played with near ingroup partners. One standard deviation increase in conflict exposure increases « solidarity » towards far ingroup partners by 8.2%. This is the first paper to study the effect of conflict on trust and cooperation with an experimental approach that makes people from the same and opposite sides of the conflict interacting together. It is also the first paper to address this question during an ongoing conflict and not in its aftermath. Our results have important implications for community cooperation, reconciliation and social capital reconstruction after the end of the conflict.