The timing of discretionary bonuses – effort, signals, and reciprocity

In a real-effort experiment, we investigate the relationship between reciprocity and the timing of discretionary bonuses in a two-period principal-agent (manager-worker) setting. We vary the timing of the manager's bonus decision in order to examine two main channels, reward and trust, through which discretionary bonuses may operate. Average worker performance improves when bonus decisions are made between the two periods, since both channels are simultaneously active. First-period output significantly increases as workers attempt to signal their trustworthiness to managers. When the bonus decision is made upfront or at the end, average output is no different than in a baseline setting without the bonus mechanism. Furthermore, output after a bonus is not paid decreases substantially, consistent with negative reciprocity. Our main findings are reinforced by using time spent on the available real leisure activity as an alternative measure of subjects' effort.