Taking, Giving and Taking to Give: Experimental Evidence of Preferences over Actions

Abstract

The desirability of an outcome often depends on the action that must be taken to bring it about. Using a within-subject design that permits the identification of individual preferences, I report experimental evidence from disinterested dictator games suggesting that preferences over income distributions depend on whether implementing those distributions requires the allocation of gains, the imposition of losses or the redistribution of pre-existing endowments. Subjects exhibit a stronger preference for equality when the action required to implement it involves the unequal allocation of income or redistribution of wealth than when it requires the unequal imposition of losses, an effect that increases in the amount of initial inequality. The evidence is inconsistent with any model of preferences defined solely over outcomes, but can be explained by existing models of narrow choice bracketing.

Keywords: social preferences, dictator game, attribution, experimental economics

JEL Classification: A12, A13, C70, C91, D63