Trust: a game of co-evolving beliefs and preferences

Trust is often described as an important variable in social interactions, allowing exchanges and potentially influencing economic performance. In the last decades, the understanding of trust antecedents and consequences has deepened. While trust has often been defined as an expectation about a counterpart’s trustworthiness in a situation involving risk, experimental studies point to the social nature of the risk affecting trust-based interaction, and to the relevance of so-called social preferences. In this paper we present a model integrating some recent insights from experimental research. We use a definition of trust based on economic primitives, such as beliefs and preferences. The model extends previous research in that we assume endogeneous preferences and we focus on trust dynamics in a web of social interactions, and on learning as the essential mean for trust evolution and diffusion. We define various settings depending on how buyers update their beliefs and preferences: on one side beliefs are updated either only through direct, or thought direct and indirect experience; on the other side, preferences are updated either considering beliefs evolution, or based on outcomes evaluation. We investigate how the co-evolution of beliefs and preferences influences buyers’ performance in terms of the quality of the selected interactions and of aggregate welfare. We define performance measures in terms of social risk, social loss, welfare, and a synthetic measure of aggregate performance and efficiency. We show that the assumption of preferences endogeneity can have both negative and positive impact on trust dynamics, and on social welfare, depending on the average level of agents’ beliefs, as well as on learning dynamics. In particular, we show that if social learning is fully developed, 1Organization: University of Luxembourg CREA Email: [email protected] This research is made possible under the financial support of the FNR fellowship project AFR PHD-08-017