Abstract

THEORIES OF BEHAVIOR IN PRINCIPAL-AGENT RELATIONSHIPS WITH HIDDEN ACTION

Claudia Keser^a and Marc Willinger^b

^aUniversität Karlsruhe, Germany and CIRANO, Montréal (Québec), Canada
^bUniversité Louis Pasteur, Strasbourg, France

We present a series of principal-agent experiments with hidden action. The agent has to choose between high and low effort. Through his effort, a gain is produced which accrues to the principal. The gain can be high or low. The high gain is more likely with high effort than with low effort. The effort is not observable to the principal. Thus, the principal can pay the agent only contingent on the realized gain but not on his effort. In the subgame perfect equilibrium, assuming risk-neutrality, the principal offers payment schemes that induce the agent to choose high effort with an expected profit just above zero.

In the experiments, we vary effort costs. We observe that the smaller the difference between effort gains and effort costs, the better the success of the game-theoretic prediction. This can be explained by a distributive aspect inherent in the agency problem, which is ignored in the game-theoretic solution. We compare the predictive success of the game-theoretic solution to the one of the fair-offers theory presented by Keser and Willinger (IJIO forthcoming) and of threshold theories, similar to those presented by Ochs and Roth (AER 1989) for bargaining games.