Contracting with researchers

We study a setting in which one or two agents conduct research on behalf of a principal. The agents' success depends on effort and the choice of a research technology that is uncertain with respect to its quality. A single agent has no incentive to deviate from the principal's preferred technology choice. In the multiagent-setting researchers pursue individual rather than overall success which yields a preference for the most promising technology. We show that a mechanism that deters this bias towards mainstream research always entails an efficiency loss if researchers are risk-averse. Our results suggest that there is too little diversity in delegated research.