Based on a simple 'contest' model of product innovation, we find that,
in accordance with conventional wisdom, and contrary to much of the p
revious theoretical literature, market competition implies an equilibr
ium level of risk which is too low from society's standpoint. The intu
ition is that the divergence between private and social marginal benef
its from R&D is greater the greater the probability of discovery by th
e rival firms; and this probability is higher for lower risk projects.