Cheat or perish? A theory of scientific customs

Performance incentives have become increasingly important in academia. Evidence suggest that they increase not only researchers' effort but also their cheating. We develop a principal-agent framework in which a research institution (principal) aims to reward scientific productivity considering the social benefit of research and the remuneration costs. Researchers (agents) take into account what others are doing. As a consequence, the introduction of performance incentives not only leads to fraudulent behavior in the short run, but may also deteriorate the scientific norm in the long run. In the face of these dynamics, the principal pursues two goals: to increase productivity (effort incentive policy) and to deal with fraud (anti-fraud policy). Several factors determine the optimal mix of policies: economic factors (researchers' cost of providing high or low effort, the societal value of scientific knowledge and the societal costs of scientific misbehavior) and psychological factors (the evolution of norms over time). An important insight of our theory is that when research institutions seek to encourage the production of high quality research, they should also raise the salary for normal research to avoid misconduct.