Relative performance of incentive mechanisms: Computational modeling and simulation of delegated investment decisions

This paper evaluates the relative performances of several well-known and widely-used incentive mechanisms under controlled experimental conditions. The scenario utilized is a delegated investment setting where effort and risk aversions contribute to moral hazard among fund managers. Analytical intractability of the problem requires a computational modeling approach to simulate comparative solutions for specific contracts under different parametric settings. Through a simulation exercise, we consider multiple agents who decide their investment strategy over several consecutive periods. Agents learn about estimation and market uncertainty through repeated realizations of investment returns. In each sequence of periods, a number of different incentive mechanisms based on the agent's communication and/or outcome are considered. Results of the computational experiments are presented. Our results overwhelmingly show the efficacy of the incentive contracts in improving the welfare of the investors. In the presence of an estimation risk, when agents learn from their past performances, the market volatility interacts with the estimation risk that makes risk-sharing arrangements such as limited liability overly important. Paying the agent to assume the risk may no longer lead to the best performance incentives.