Mechanism design and intentions

We introduce intention-based social preferences into mechanism design. We explore information structures that differ with respect to what is commonly known about the weight that agents attach to reciprocal kindness. When the designer has no information on reciprocity types, implementability of an incentive-compatible social choice function is guaranteed if it satisfies an additional insurance property. By contrast, precise information on reciprocity types may imply that all efficient social choice functions are implementable. We show how these results extend to a two-dimensional mechanism design setting where the agents have private information about their material payoff types and their reciprocity types. We also provide a systematic account of the welfare implications of intentionality.