Mental Accounts and the Marginal Propensity to Give
Journal of Economic Science Association, vol.
This study uses a framed field experiment to investigate whether the neoclassical model or<br>mental accounting better describes how income from different sources are treated in sharing<br>decisions. Participants play a dictator game after earning income in real-effort task and/or receiving<br>a windfall. I find that dictators treat marginal earned and windfall income as partially infungible, which supports mental accounting. Two-step estimates show that sharers shared<br>15% of a marginal windfall token and 7% of a marginal earned token. Strikingly, sharers who had income from both sources were sharply less generous with both earned and windfall income than those who had only a single source. This is consistent with other instances of<br>complex decision frames inducing selfish behavior by providing a cover of ambiguity. This aspect of mental accounts has thus far received little attention. A follow-up experiment shows that two accounts must qualitatively different, not just multiple in number, to induce more selfishness.