CIRJE-F-304. Matsushima, Hitoshi, "Non-Consequential Moral Preferences, Detail-Free Implementation, and Representative Systems", October 2004.

We investigate implementation of social choice functions where the central planner has no knowledge about the detail of model specifications, and only a few individuals participate in the mechanism. In contrast with the standard model of implementation, each agent has non-consequential moral preference in that she prefers truth-telling to lying whenever the resulting consequence is unchanged. We show that with complete information, there exists a single, detail-free mechanism that can implement any social choice function whenever agents regard its value as being socially desirable. This result holds even if psychological cost for lying is close to zero. Non-consequential moral preferences play a very powerful role in eliminating unwanted equilibria in detail-free mechanism design with representative systems. We extend this result to incomplete information.