Not Invented Here

We consider the problem of inducing agents who are concerned with their careers to reveal their private information about a project which has originated with one of them. A successful project raises the inventor's chance of promotion, at his peer's expense. Thus, the peer has an incentive to promote the inventor's bad project to see him fail, but to denigrate his most promising projects. Moreover, there is an incentive for junior workers to push their own work no matter what the perceived quality is, but an incentive for senior workers to suppress their own ideas in order not to have a big failure that ruins their career. In case of disagreement among agents, the optimal policy is to promote the agent who is more likely to have been truthful, not necessarily the one most suitable for promotion. This policy is not renegotiation proof. Within the class of renegotiation-proof mechanisms, self assessment (where no peer report is submitted) is always optimal. Exaggeration is a less serious problem than denigration in this model. It is risky strategy to exaggerate, since at best you can convince the principal to implement an un-promising project which is likely to fail. It is safer to denigrate, since if a promising project is stopped due to an unfair peer report, the principal will never learn the project's true quality(This abstract was borrowed from another version of this item.)