Evolution and the Ultimatum Game: Why Do People Reject Unfair Offers?

Abstract

In this paper we review, upgrade, and synthesize existing models from evolutionary game theory that aim at explaining behaviour in the ultimatum game, and we compare their predictions with the existing experimental evidence. We find that the results in Gale et al. (1995) and Rand et al. (2013) are primarily driven by bias in the mutations. We make versions with local instead of global mutations for both. This minimizes the bias, and changes the results. We also consider Quantal Response Equilibria in combination with the assumption that individuals are selfish after all. The Quantal Response Equilibrium is the noisy twin of the Nash equilibrium, and looking at this combination we explore an alternative explanation for what we observe in the lab, namely noise instead of deviations from selfishness. Finally, we provide a refurbished version of the model of commitment in Nowak et al. (2000). The de-biased version of the model in Rand et al. (2013) becomes a special case of this more general model (with the possibility for commitment muted).

We find that the experimental evidence does not align with the models in Gale et al. (1995), Rand et al. (2013), or our de-biased versions of them, and that it also rejects the combination of selfishness and the Quantal Response Equilibrium. All of these models predict that the distribution of minimal ac- ceptable offers should start with high frequencies at 0, end with low frequencies at 1, and have decreasing frequencies in between, which is not what is found in lab experiments.

Keywords: Ultimatum game, fairness, mutation-selection equilibrium, Quantal Response Equilibrium, commitment, experimental evidence

JEL Classification: C73