Productivity Effects of Innovation Activities

Part of the ZEW Economic Studies book series (ZEW, volume 38)


Understanding and quantifying the driving factors behind productivity and productivity growth, and in particular the role of innovation activities in this context, has been of major interest in the field of empirical economics for several decades.76 This can be explained by the fact that innovation is widely believed to be a key long-term driving force for competitiveness and growth of firms and national economies as a whole. Recent years have even seen a surge of studies on productivity, in particular at the firm level. This is in part due to new theoretical underpinnings from the endogenous growth theory, which emphasises that economic growth is positively correlated with investments in research (see Romer, 1986; 1990) and human capital (Lucas, 1988). Another reason is the increasing availability of comprehensive micro databases. However, as was set out in section 1.2, quantifying the importance of innovation for productivity is a challenging task and, despite a large number of empirical studies, innovation research has only been partly successful (see Griliches, 1995; Bartelsman and Doms, 2000).


Labour Productivity Total Factor Produc Process Innovation Product Innovation Innovation Activity 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 76.
    Furthermore, there is a related group of studies focussing on the description of cross-sectional distributions of productivity across firms in an industry and its evolution over time (see, e.g., Nelson, 1981; Baily, Hulten, and Campbell, 1992; Bartelsman and Dhrymes, 1998).Google Scholar
  2. 78.
    Llorca Vivero (2002) examined the impact of process innovations on productivity growth using the number of product and process innovations. However, he did not analyse the whole innovation input, output, and productivity relationship.Google Scholar
  3. 80.
    Schumpeter himself only pointed to the qualitative difference between small and large firms. Nonetheless, the empirical literature has interpreted his claim as a more than proportionate relationship (Cohen, 1995).Google Scholar
  4. 81.
    But then, excessive bureaucratic control may impede innovation activities in large firms (Cohen and Levin, 1989).Google Scholar
  5. 83.
    Note that the causality may also run from innovation to export. The technology gap trade theory by Krugman (1979) or the life-cycle theory by Vernon (1966) states that innovation is the driving force behind export activities.Google Scholar
  6. 84.
    However, in open economies domestic firms also face competition via imports from foreign companies (see Bernard and Wagner, 1997).Google Scholar
  7. 85.
    Kleinknecht and Poot (1992) have linked this argument into a product life cycle approach. They argue that early stages of a cycle are associated with considerable R&D activities which are, therefore, carried out close to the headquarters while less R&D activities are necessary in later stages for incremental product or process modifications and can, hence, be decentralised.Google Scholar
  8. 86.
    An exception is the study of Lööf, Heshmati, Asplund, and Naas (2003) which finds a positive impact for Swedish firms but not for Finnish and Norwegian firms.Google Scholar
  9. 87.
    One of the main incentives to collaborate on innovation projects is to get access to external knowledge (Cassiman and Veugelers, 2002).Google Scholar
  10. 88.
    The majority of previous empirical studies confirm a positive correlation between productivity and exports at the firm level. Recent studies mainly focus on the direction of causality. The “learning-by-exporting” hypothesis states that exporting firms may profit from technological knowledge and expertise available on foreign markets, resulting in a positive productivity effect (see Evenson and Westphal, 1995). However, recent studies by and large confirm the hypothesis of self-selection, that is best-performing domestic firms self-select into export markets (see, e.g., Bernard and Jensen, 1999; Arnold and Hussinger, 2005; De Loecker, 2004). This implies that there is a causal link from productivity to exporting whereas no evidence of the opposite direction was found.Google Scholar
  11. 90.
    Instead of R&D investments knowledge capital has also been approximated by R&D employment, then labour is measured as non-R&D personnel (see Hall and Mairesse, 1995).Google Scholar
  12. 91.
    Results further show that basic research has a greater effect on productivity than applied research. The same is true of company-financed compared to publicly-funded R&D (Griliches, 1986).Google Scholar
  13. 96.
    Crépon et al. (1998) used additional information from the French R&D survey to build up an R&D stock. However, they ascertained that in cross-sectional analyses the results did not significantly change when using the flow instead of the stock measure. The explanation is that in cross-sections the flow variable is a good proxy for the stock variable. See Griliches and Mairesse (1984) or Hall (1990) for the construction of a knowledge capital stock using R&D expenditure and patents.Google Scholar
  14. 107.
    Under certain assumptions the pseudo maximum likelihood method leads to consistent estimates even in case of misspecification, see, e.g., Ruud (1986).Google Scholar
  15. 110.
    Scherer (1967) found evidence for an inverse U-shaped relationship. I tested for this hypothesis by running an additional regression including the square of HERFIN as well. However, this hypothesis was not confirmed in this analysis.Google Scholar
  16. 113.
    Lagged values are valid instruments as long as the error terms are not correlated over time. Blundell and Bond (2000) used a system GMM estimator. Recently, Olley and Pakes (1996) and Levinsohn and Petrin (2003) have suggested nonparametric estimation methods to control for endogeneity in total factor productivity (TFP) regressions.Google Scholar

Copyright information

© Physica-Verlag Heidelberg New York 2008

Personalised recommendations