What makes an allocation fair? Some experimental evidence

We explore three two-person public goods games with similar equilibrium predictions, but with different rules of the game, different payoff possibilities, and, as we show, different choices by subjects. Comparisons among games allow inferences of what may or may not determine when the equilibrium prediction is a good approximation of actual behavior. We find that the equilibrium prediction can fail even when incentives off the equilibrium enforce it. Our result suggests that the selfish prediction is prone to deviations when the equilibrium results in unequal distributions of payoffs, and there are alternative outcomes that increase both equality and the payoff of the disadvantaged party. Furthermore, fairness is a function of more than just the final allocations to subjects; it depends on the actions not chosen as well as those that are. (C) 2002 Elsevier Science (USA).