Testing motives for charitable giving: A revealed-preference methodology with experimental evidence

A large economics literature seeks to understand the reasons why individuals make charitable contributions. Fundamental features of most models of charitable giving are the inclusion of externalities induced by other agents and the Lancasterian characteristics approach to specifying utility functions. This paper develops a general, revealed-preference methodology for testing a variety of preference structures that allow for both externalities and characteristics. The tests are simple linear programs that are transparent, computationally efficient, and straightforward to implement. We show how the technique applies to standard models of privately provided public goods and novel models that account for other-regarding preferences based on relative consumption and donations among individuals. We also conduct an original experiment that enables testing and comparing many models on a single data set. Our experiment design allows us to focus on intrinsic motivations which are often hard to disentangle from other extrinsic or image effects in field data. The results provide the first revealed-preference evidence on the importance of social comparisons when individuals make charitable contributions. Models that include preferences for either relative consumption or donations yield significantly greater explanatory power than the standard model of impure altruism.