Merit norms in the ultimatum game: an experimental study of the effect of merit on individual behavior and aggregate outcomes

Abstract

The paper reports the results of an ultimatum game experiment designed to test the effects of meritocratic norms on individual behavior and aggregate outcomes. In one treatment the roles of proposer and responder were assigned randomly. In the other treatment the roles were earned in a general knowledge quiz. The results show that proposers offer significantly less when they have earned their roles and responders have a significantly lower acceptance threshold. Rejection rates are lower for offers lower than the equal split when positions are allocated based on merit: Proposers earn significantly more in this setting. Responders suffer some loss in this treatment. This leads to an increase in overall inequality of payoffs measured by the Gini index when positions are allocated based on merit.

Appendix

1.1 Computation of the alpha-parameter

For cases where player $i$ has a lower payoff $x_i$ than the payoff $x_j$ of player $j$, Fehr and Schmidt (1999) put forward the following utility function:

$$\begin{aligned} U_i(x)=x_i-\alpha _i \cdot (x_j-x_i) \end{aligned}$$

(1)

The parameter $\alpha _i$ captures the aversion of player $i$ against disadvantageous inequality. The lowest offer accepted by a responder allows us to derive his $\alpha$-value since somewhere between the lowest offer $o_i$ a responder is willing to accept and the offer $o_i-1$ which he rejects he is indifferent between the rejection payoff and accepting the offer. We assume that this indifference point lies at $o_i-0.5=x_i$. We set $x_i-\alpha _i \cdot (x_j-x_i)=0$ and $x_j=20-x_i$ and this simplifies to

$$\begin{aligned} \alpha _i=\frac{x_i}{2 \cdot (10-x_i)} \end{aligned}$$

(2)

used to estimate $\alpha _i$. Since the maximum value of $o_i$ is 10, the maximum value of $x_i=o_i-0.5=9.5$ and no division by zero can occur. This procedure is identical to the one used by Blanco et al. (2011).