Skills required for innovation: A review of the literature

Research teams from 18 OECD countries used the methodology introduced by Crepon-Dugay and Mairesse (CDM) to analyze the impact of innovation on labour productivity using firm data from national innovation and administrative surveys. To ensure international comparability, the OECD ‘core’ CDM model did not include variables for which data were missing in some countries. In spite of this shortcoming, the results are broadly in line with theoretical hypotheses and previous studies and show a surprising degree of similarity between countries. This paper builds on the Canadian application of the ‘core’ model used for the OECD project. It uses to the full extent all information available on manufacturing establishments from the Canadian Survey of innovation 2005 linked with the Annual Survey of Manufactures and Logging (ASML). The estimated econometric model controls for selection bias, simultaneity, size of firm and industry effects. The main findings suggest that (1) export outside of the US market, size of the firm and use of direct or indirect government support are factors increasing the probability to innovate and having positive innovation sales. (2) Exports (both to the US and outside of the US market), cooperation with other firms and organizations, and high share of the firms’ revenue coming from sales to its most important client are all factors correlated with higher innovation expenditures per employees. Moreover, firms with a higher market share at the beginning of the period are spending more on innovation by the end of the period. (3) Firms with higher innovation expenditures per employee generate more innovation sales per employee. Other factors increasing innovation sales are human and physical capital and introduction of process innovations. (4) Finally, the firms generating more innovation sales per employees achieve higher labour productivity, even when the size of firms, the intensity of human and physical capital and labour productivity at the beginning are taken into account. The results add valuable further information to and are in line with the simpler model applied to 18 other OECD countries. The paper concludes with discussion of policy implications.