This paper addresses a debated issue in the economics innovation literature, namely the existence of increasing return to R&D expenditures and sirm size on innovation output. It further explores how structural characteristics of the sirm as well as contextual factors affect the dynamics of product innovation over a relatively long period of time. Taking advantage of an original and unique database comprising innovation data recorded on a monthly base we show that: (i) a negative binomial distribution model is able to predict with great accuracy the probability of having a given number of product announcement sent out in a month; (ii) constant returns to size and R&D expenditure may reasonably characterize the innovation production function of sampled sirms; (iii) vertically integrated manufacturers as well as producers operating a larger product portfolio exhibit a higher propensity to introduce new products than their specialized competitors.
