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...

We develop a general model, with the exponential bandit as a special case, in which high-ability agents are more likely to achieve early success but also learn faster that their project is not promising. These counteracting effects give rise to a signaling model in which the single-crossing...

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...