Social learning with subjective communication and self-selection
(with Gergely Ujhelyi)

Abstract

We study social learning about a new product in a model with subjective communication (influenced by agents’ beliefs) and self-selection of social contacts. Self-selection implies that consumers will be disappointed on average. Subjective communication implies that the messages sent will tend to be negative. In this model, unless the quality advantage of the new product is large enough, learning causes the set of consumers choosing the new product to shrink over time. Thus, learning may help efficiency by driving out inferior new products, but can hurt efficiency by reducing the consumption of new products that are superior. We show that learning in more ”diverse” neighborhoods implies more consumption of the new product. However, these neighborhoods are more receptive to new products regardless of their quality. A number of extensions are studied, including more accurate communication, communication with subjective thresholds, and less accurate consumption experience. The results provide a learning-based explanation of why some communities seem resistant to new products or ideas.