ABSTRACT
This article examines the links between firm innovation and firm productivity performance across a range of European economies, and in particular, we explore the differences between countries which are in transition from efficiency-driven to innovation-driven with those which are primarily innovation-driven economies. We employ an endogenous-switching technique to explore micro-economic survey-based data from both innovating and noninnovating firms. The model allows us to construct counterfactual scenarios which overcome problems of self-selection in the data. Some of the findings provide support for the traditional patterns previously found in the innovation literature, in which innovation efforts and investments in physical and human capital are found to be important for product and process innovations in manufacturing and service firms and across economy types. Our counterfactual analysis also allows us to outline a rationale for policy intervention towards noninnovating firms as well as innovating firms depending on where the transitional heterogeneity effects are greatest.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Firm linkages and entrepreneurship are defined as the percentage of small- and medium-sized enterprises (SMEs) innovating in-house, collaborating with others, firm renewal (SME entries plus exits) and public–private co-publications per million population (European Commission Citation2009).
2 Innovation outputs are defined as the percentage of SME’s introducing product or process innovations and marketing or organization innovations and share of innovators where innovation has significantly reduced labour and the use of materials and energy (European Commission Citation2009).
3 The data sets in Business Environment and Enterprise Performance Survey (BEEPS) are larger than those employed in this study as firms categorized as construction and mining and where there were missing values recorded are excluded from the analysis.
4 The categorization of firms according to the Global Competitiveness Report (GCR)’s definition can pre-determine the answers that are being reported. If an alternative method for categorization was selected, the results may be different, hence the failure to categorize firms appropriately could easily introduce a form of bias. However, we believe it reasonable that the categorization of firms by country into innovation-driven and transition-driven economies according to the GCR is reasonable and does provide the best rationale for the separate categorization of firms by economy type.
5 R&D effort is one of the variables included in (in Equations (1) and (2)) and this measure depicts an innovation effort instrument estimated using observed R&D activity by firms in the BEEPS data set. A discrete (Probit) model is used to examine the knowledge sourcing step of the CDM model where R&D activity is assumed to be explained by the firms’ size, international competition (proxied by whether the firm exports or not) and country location. The analysis of R&D expenditure is not a central feature of this article, but in order to account for the endogeneity concerns between R&D activities and innovation outputs, this step is employed in the empirical structural CDM model. The individual firm’s innovation effort is proxied by the predicted values generated from the discrete model of R&D activity and these are then employed in the knowledge production step as depicted by
in Equations (1) and (2). This is estimated separately for manufacturing and service firms to account for sector differences. This estimation strategy is similar to that employed by Griffith et al. (Citation2008). The results from the knowledge sourcing steps are presented in of the Appendix. Larger firms and exporting firms are significantly more likely to spend on R&D.
6 Our choice of endogenous switching model is justified as the Wald tests for independent equations rejected the null hypothesis of ρ = 0 for all models except for the product innovation model for the manufacturing industry in transition economies. Therefore, the selection equation and outcome equations in the majority of the models are dependent.
7 The interpretation of firm size needs to be taken with caution as the number of products/services or innovations introduced are not controlled for.