Trust, trustworthiness and social networks: Playing a trust game when networks are formed in the lab

This paper investigates the impact of network formation on trust and trustworthiness. We run a laboratory experiment where, in sequence, networks are generated endogenously within an anonymous group and subjects play a trust game. The experimental design includes two main treatments and a baseline: in the baseline subjects play a trust game with no networks being formed, in treatment NT the network building phase precedes the trust game, and in treatment TN the network game is played at the end. This allows us to identify the two main factors through which networks impact on trust and trustworthiness: information accrued to subjects through social interaction (when this occurs first) and reputation (when it follows). We find that in NT, the overall level of trust is lower but offers are directed to more trustworthy recipients. A common past history matters in determining whom to trust (information value of networks). In TN, continuation play enforces higher levels of trust and trustworthiness (reputation and enforced reciprocity). Profits that subjects make in the trust game are higher in the presence of social interaction, and significantly so when network formation informs the decision of whom to trust.