Cooperation in a fragmented society: Experimental evidence on Syrian refugees and natives in Lebanon

Michalis Drouvelis, Bilal Malaeb, Michael Vlassopoulos, and Jackline Wahba

Journal of Economic Behavior & Organization, Volume 187 (2021), Pages 176-191


This paper examines intra- and inter-group cooperation of Syrian refugees and host communities in Lebanon. While Syrian refugees share similar ethnic, linguistic and cultural backgrounds with Lebanese nationals, there are tensions between the two groups due to historical tensions and economic pressures.

The authors carry out a ‘lab-in-the-field experiment’ in Lebanon with Syrian refugees and Lebanese nationals to evaluate levels of cooperation between and within the two groups and the extent to which punishments applied to enforce cooperation are effective in increasing cooperation. Cooperation is measured in a social dilemma game—first without and then with punishment opportunities, as follows:

  • In the game without punishment, each participant is given 10 tokens and has to decide how many of them to keep and how many to contribute to ‘a project’ (a public good). Each token a participant keeps increases their own payoff by 1 experimental currency unit (ECU), while each token contributed to the public good increases the payoff of both players in the pair by 0.75 ECUs.
  • In the game with punishment, participants can see the contribution of the other player in the pair and are given the opportunity to assign up to 5 punishment points to the other player. Each punishment point costs the punisher 1 ECU and the recipient of the punishment 3 ECUs.
  • In the game with punishment, facilitators elicited each participant’s beliefs about the other player’s contribution behavior and expected punishment. Accurate beliefs were paid 1 Experimental Point exchanged at a pre-specified rate mentioned at the beginning of the instructions.

The 312 participants were randomly assigned to Lebanese-only, Syrian-only, or mixed sessions. At the beginning of each session, the facilitators conducted an ‘icebreaker’ that revealed subtle differences in spoken Arabic to make the nationalities of the participants clear, without explicitly stating the nationalities of the participants. The participants were assigned to randomly formed pairs and played, anonymously, three rounds of the game without punishment and three rounds with punishment.

Main findings:

  • Contributions to the public good are higher in homogeneous groups than in mixed groups. On average, randomly formed pairs in homogeneous sessions contribute more than those in mixed sessions, suggesting in-group cooperation is stronger. The reduction in cooperation in the mixed sessions is driven by the behavior of Lebanese participants, who tended to reduce their contributions in mixed sessions.
  • Homogenous groups have higher expectations of contributions to the public good. In homogeneous sessions, participants have similar beliefs about the other player’s contribution, and expect the other player to contribute a larger amount of tokens to the public good compared to mixed sessions. Compared to participants in the homogeneous sessions, participants in the mixed sessions expect that the other player in the pair will contribute less. This effect is primarily driven by the behavior of Lebanese participants.
  • There is a higher tendency for participants in mixed groups to condition their cooperation on how much they believe the other player in the pair will cooperate than in homogeneous groups. In the homogeneous sessions, cooperation appears to be driven more by an unconditional relationship between contribution and beliefs, which could be attributed to altruistic motives. In contrast, participants’ contribution behavior in the mixed sessions is more strongly conditioned on their beliefs about the other player’s behavior, suggesting a pattern of reciprocal behavior.
  • Mixed groups punish significantly less than homogenous groups. Participants punished both negative and positive deviations from their own contribution. There was a substantial degree of antisocial punishment (i.e. the sanctioning of players who behave prosocially), especially in Lebanese-only sessions. On average, randomly formed pairs in homogeneous sessions punished significantly more than those in mixed sessions, a result that is driven by Lebanese participants. This suggests that for the Lebanese participants, there is a lower willingness to punish out-group defectors and an inclination to punish in-group cooperators.
  • There were no differences in average net earnings between homogenous and mixed groups, when the cost of punishment is taken into account. Average earnings were significantly lower in mixed sessions compared to homogeneous sessions in the public goods game with punishment, but only at the contribution stage. There were no differences between homogeneous and mixed groups, when adjusting total earnings for the costs of punishment.

The authors conclude that the mixing of the two groups leads to lower contributions to the public good, and sanctions are not able to redress this lack of cooperation. They note that behavior is not symmetric across the two groups, as it is the host community that shows less cooperation toward the refugees. The authors call for future research to assess whether interventions aimed at increasing intergroup contact, trust and co-operation would help to reduce outgroup biases and increase public good provision, which are important for the well-being of refugees and host communities.


Lebanon | Syria