Inferring Social Preferences over Income Distributions through Axioms

Numerous prior experimental studies have attempted to elicit people’s preferences over income distributions through appropriately incentivated questions asking subjects to choose between distributions. Instead, we follow the theoretical literature and start with the principles underlying these preferences. Such principles include, for example, the Rawlsian principle and the Lorenz principle. We implement possibly the first incentivated experiment concentrating solely and directly on these underlying principles. In essence, the experiment asks subjects to state their preferred principles, with the appropriate incentive being provided by the experimenter using the stated axioms to choose a preferred distribution from a randomly generated set, and this preferred distribution then being implemented on the participants in the experiment. Thus one of the subjects becomes the Social Planner. (We have two treatments, one in which the Social Planner is part of society and the second in which the Social Planner is outside society.) We solve problems implied when the chosen set of principle is either mutually contradictory or incomplete (by allowing sequential choice). We observe that the implied social preferences are different from those inferred indirectly through choice over distributions as in the earlier experimental studies, and that elicited preferences are different between the two treatments, suggesting that the disinterested Social Planner is more Equality-preferring than the Social Planner with self-interest.