Résumé
We study to what extent identification does matter for trustfulness and trustworthiness to emerge in a population of players. Our experimental protocol is designed for isolating the effects of trustees’ identification. Trustees’ identification is a necessary condition for introducing a reputation mechanism. We run three treatments. In each treatment groups 6 players interact repeatedly and randomly and play a 30 periods investment game (Berg & al. 1995). In the first treatment players can’t identify each other, in the second one players can identify each other as trustee and in the third one players identify each other both as trustee and trustor. We show that, according to the expectation, trustees’ identification has a positive effect on reciprocity. However it doesn’t affect the average trust in the population. Trust is significantly higher than in the complete anonymous treatment only when players identify each other in both roles. We show that this enhance of trust is the result of mutual trust-reciprocity relationships formation.