Innovation at the firm level when the level of technological development of the home country differs: Evidence for 19 EU countries

This paper presents a comparative analysis of innovation behaviour at the firm level across countries with different levels of technological development. The literature has argued mostly from a macroeconomic perspective (e.g. Silverberg and Verspagen 1995, Acemoglu – Aghion – Zillibotti 2002, Aghion and Howitt 2006, Lazonick 2002, 2008) that with the proximity to the technological frontier the importance of cutting edge innovation activities increases for the aggregate growth path of countries. Only investment in advanced knowledge generation and basic research will allow countries to forge ahead. On the other hand, farther off the technological frontier technology absorption is necessary and sufficient to generate growth and catch up with more advanced economies in terms of GDP per capita. Based on this logic several authors have proposed that there is a need for institutions that are commensurate to the level of technological development of a country in order to support economic development and growth adequately.Whereas analyses based on aggregate or sectoral data seem to support this view, comprehensive analyses on the firm level have not been available to this date. This contribution tries to fill this gap. The paper presents a simple stylised model on the innovation decision at the firm level that is used to develop the principal hypotheses that are then tested econometrically using data from the Community Innovation Survey for 19 European countries. The analysis focuses on three aspects: the relationship between activities related to technology transfer and R&D, the relative importance of different types of knowledge sources and the degree of technological complexity of production in an industry across countries. Drawing on results of a prior paper (Reinstaller and Unterlass 2011) we group firm level data into larger country groups that have been identified through a statistical cluster analysis using aggregate data on direct and indirect R&D data from an input-output analysis. These groups capture differences in the level of technological development across countries. For each of these groups we fit the identical econometric models and test the hypothesis that the estimated coefficients are identical.The results show that the marginal effects of both the level of total innovation expenditures and the share of R&D investment in total innovation investment increase monotonically across country groups with the level of technological development of the country. The same strictly monotonic relationship is observed for indicators capturing the importance of firm specific capabilities. On the other hand, no such relationship can be observed for knowledge stemming from market sources or suppliers. These factors are of similar importance (in terms of the level, direction and significance of the marginal effects) across country groups. Surprisingly knowledge from academic institutions and government labs or indicators capturing the closeness to basic research of the knowledge sources upon which the firm draws are both statistically and economically not significant. The reasons for this finding are discussed in the paper. Finally, the results also show that the claim often made by academics and policy makers that R&D activities are necessary to promote innovation also in technologically less advanced countries is not confirmed by our econometric results. The interaction effects between the share of R&D in total innovation investments and total innovation investments is positive and significant only for the country group of the technologically most advanced countries while it is statistically not significant for all other country groups suggesting that a positive complementary relationship between technology transfer and R&D arises only at relatively high levels of technological development. All these results are very robust across different specifications.