There is an extensive literature on innovation, investigating both its determinants and its contribution to firm performance (measured as productivity growth or as market value). Since the seminal papers by Griliches (1979) a widespread approach is to frame the relationship between innovation and its determinants in a knowledge production function and the contribution of innovation to productivity growth in an output production function. A first order measurement problem that economists have had to face is how to measure `new knowledge'. A similar measurement problem arises in the attempt to capture knowledge flows or knowledge spillovers that, because of the public good aspect of knowledge, contribute to the production of innovative output.

This paper uses matched firm level data from the Community Innovation Survey with production data to overcome some of these measurement issues when modelling the links between innovation inputs, external knowledge flows, innovation outputs and productivity (TFP) growth with innovation. The paper develops a framework that explains the different stages of the development of both product and process innovation and links them with the productivity growth of the firm. The results show the following. Firstly, the firm's decision on whether and how much to invest in innovation is positively correlated with international competition, the availability of skilled workers and with the availability of methods of protection of innovations. Secondly, internal innovation expenditure is only one of the factors associated with successful innovation. The results show that cooperation and knowledge from other firms in the enterprise group and from suppliers and customers have a positive correlation with successful innovation. Thirdly, my results show that it is important to distinguish between product and process innovation, as well as novel and incremental (non-novel) innovations. Some of the self-reported estimates of innovation are correlated with TFP growth and some are not: in particular the negative correlation between novel process innovation and productivity growth might appear counterintuitive. This might reflect reality, since process innovations take time to feed through to productivity growth, or the fact that firms that face a decrease in demand or are going through financial difficulties are more likely to implement process and that this improvements will be reflected in increased productivity with some delay but it might reflect the weakness of the measures used. I investigate this issue further: I control for the delay in the effect for process innovation using information on the success of process innovation from the self-reported information from the innovation survey. I find that process innovations that firms report as being successful in improving production flexibility are correlated with positive productivity growth. I also show that organisational and managerial change is positively correlated with productivity growth conditional on technological process and product innovations.