Abstract: We conduct an experimental test of the long-standing conjecture that autonomy increases motivation and job performance. Subjects face a menu consisting of two projects: one risky and one safe. The probability that the risky project succeeds depends on the subject's effort. In one treatment, subjects choose a project from the menu; in the other treatment, they are assigned a project from the menu. Using a difference-in-difference approach that controls for various forms of selection, we show that autonomy (the right to choose a project) has a significant pure motivation effect on effort. The effect is consistent with aversion to anticipated regret (but not with standard expected-utility maximization): if the agent chooses the risky project and fails, he will regret not having chosen the safe project, and this motivates him to work hard to avoid a failure. Regret theory makes further predictions that are also supported by the data.

First, that the pure motivation effect is greater if the menu of feasible projects is diverse, generating a more meaningful choice among projects.

Second, that the effort on the risky project is greater, the greater is the return to the safe project, because this (foregone) return determines the amount of regret.

Finally, we find a significant negative relationship between the strength of the pure motivation effect and the subjects' expected earnings.