Reputation formation in economic transactions

• We investigate information transmission, transfer rates, and return rates in repeated trust games where agents privately or publicly exchange reputational information.
• Information transmission is heavily used if it is costless but only moderately if it has positive cost.
• Objective information increases transfer and return rates if transmission is costless or the information is publicly observable.
• Subjective information (ratings) increase transfer and return rates only if the rating becomes publicly observable.
• Information transmission is more likely if a seller received high transfers and in case of exploitation.

We investigate patterns of information transmission and the evolution of behaviour in standard trust games where reputational information can be exchanged between individual agents within a population, but does not become public. We vary the nature of information that could be transmitted (either a subjective rating or the objective details of a transaction), as well as the cost of information transmission (either zero or positive). We find that information transmission is heavily used if it is costless but only moderately if it has positive cost. Objective information effectively increases the average amount sent in the trust game (and thus, efficiency and average profits) as well as the return rate only if transmission is costless and therefore is heavily used. Subjective information (a rating) only increases transfer and return rates in a control treatment where it becomes publicly available, but not for private communication. A detailed analysis of the determinants of information transmission shows a positive influence of (a) the transfer made to a receiver, (b) experience of exploitation, and (c) points to reciprocity in information transmission.