Producing Innovations: Determinants of Innovativity and Efficiency

In this paper we estimate, using stochastic frontier estimation techniques, the relationship between R&D inputs a innovative output in a sample of Dutch firms. We find that over 63% of between firm variation in observed "innovativeness" can be attributed to inefficiency in the innovation process. The remainder is due to differences in the innovation production process itself. We derive our results including the usual controls and find in addition that large firms tend to look more innovative. But when considered more carefully large firms turn out to be less efficient. With standard estimation techniques this inefficiency is masked by a more productive innovation technology. We thus find evidence of economies of scale in line with the Schumpeter mark II hypothesis (large firms are more innovative), but also show that large firms tend to operate at lower levels of efficiency.