Rewarding mediocrity? Optimal regulation of R&D markets with reputation concerns

In this paper, we consider a dynamic signaling model of an R&D market in which a researcher can choose either a safe project (exploitation) or a risky project (exploration) at each instance. We argue that there are substantial efficiency gains from rewarding minor innovations above their social value and further that it is indeed superior to rewarding major innovations directly, even when those minor innovations are intrinsically valueless in themselves. When only major innovations are rewarded, the R&D market eventually shuts down due to a version of the lemons problem. Rewarding minor innovations is actually conducive to major innovations as it induces self-sorting among researchers, which is essential in providing time and resources necessary for more productive ones to take riskier but more ambitious approaches. This result draws clear contrast to the static counterpart where such a scheme can never be optimal. Our model also exhibits reputation dynamics which capture a pervasive view in academia that "no publications are better than a few mediocre publications" at an early stage of one's career.