Preferences for honesty can support cooperation

Abstract

Many collective action problems are inherently linked to honesty. By deciding to behave honestly, people contribute to solving the collective action problem. We use a laboratory experiment from two sites (n = 331 and n = 319) to test whether honest preferences can drive cooperation and whether these preferences can be differentially activated by framing. Subjects participate in an asymmetric information variant of the public goods game in one of two treatments that vary only in their wording: The Contribution Frame uses a standard public good game framing, while in the Honesty Frame, words aimed to trigger honesty are used. We measure subjects' honesty in three ways using the (i) sender–receiver task, (ii) the die-roll task, and (iii) self-reported honesty levels and account for other-regarding preferences and social norms to disentangle key alternative motives. We find that all three measures of honesty preferences robustly predict contributions, as do other-regarding preferences and empirical expectations but not normative expectations. Additionally, honesty preferences predict contributions in the Honesty Frame but not in the Contribution Frame, although the difference between these is not consistently significant. Finally, we find no differences in average cooperation across the treatments.