Reward the lucky? An experimental investigation of the impact of agency and luck on bonuses

We examine the impact of agency and luck on bonuses in a two player, two stage controlled laboratory experiment. In the first stage, Player A makes an investment decision on behalf of Player B. In the second stage, Player B makes a dictator allocation for each possible outcome from the investment. We compare dictator giving (bonuses) across outcomes and with a control treatment in which the stage 1 outcomes are determined randomly. We do not find that luck is rewarded with higher bonuses. However, we do find a general tendency to respond to agency with reduced bonuses and, in particular, a significant tendency to reduce bonuses to agents who are unlucky. Additionally, we find that those who are more risk tolerant are less likely to give no bonus under agency but not in the control.