Testing the framework of other-regarding preferences

We assess the empirical validity of the overall theoretical framework of other-regarding preferences by focusing on those preference axioms that are common to all the prominent theories of outcome-based other-regarding preferences. This common set of preference axioms leads to a testable implication: the strict preference ranking of self over a finite number of alternatives lying on any straight line in the space of material payoffs to self and other will be single-peaked. The extent of single-peakedness varies from a high of 79% to a low of 54% across our treatments that are based on dictator and trust games. Positively and/or negatively other-regarding subjects are significantly less likely to report single-peaked rankings relative to self-regarding subjects. We delineate the potential reasons for violations of single-peakedness and discuss the implications of our findings for theoretical modeling of other-regarding preferences.