A NEW SECTORAL TAXONOMY BASED ON PECUNIARY KNOWLEDGE EXTERNALITIES: KNOWLEDGE INTERACTIONS IN A VERTICALLY INTEGRATED SYSTEM

Abstract

The paper presents a new sectoral taxonomy that focuses on the existence of non-negligible external effects that derive from user–producer knowledge interactions. These are coupled with intermediate goods transactions, in a system of vertically integrated manufacturing and services sectors. These externalities, the so-called pecuniary knowledge externalities, are the main source of changing technological conditions experienced by downstream producers. A distinguishing feature of the taxonomy lies in its derivation from a particularly dynamic context of changing production functions. The taxonomy is empirically derived, examining effects generated by technological knowledge in a system of intermediate goods transactions and taking into account peculiar characteristics of sectors in European economies. The results allow for a classification of sectors in five groups. An analysis of these classes confirms previous evidence that technological characteristics of sectors across classes differ.

Keywords:

• Pecuniary knowledge externalities
• Sectoral patterns
• Vertical linkages

Notes

¹ Tidd et al. (1997), Evangelista (1999), Marsili and Verspagen (2002) and Castellacci (2006, 2007) proposed refinements of the original taxonomy. Using the same statistical source as Pavitt, i.e. the data collected at the SPRU by Townsend et al. (1981), Freeman (1982) offered a taxonomy of innovations. Observing technological characteristics of innovations, Freeman distinguished three major categories, namely, incremental, radical and pervasive. Aiming at finding common features relating both the sectoral taxonomy by Pavitt and the technological taxonomy by Freeman, Soete (1988) classifies sectors into six groups according to their characteristics as suppliers and/or users of innovations. Most importantly, he stresses on a great role played by pervasive innovative sectors and on a crucial influence that vertical interactions between users and suppliers of innovations exercises in designing technological trajectories.

² In analyzing interactions between users and producers, von Hippel (1988) argues that in certain areas users may appear particularly innovative, especially when they search for solutions satisfying their own needs. This, consequently, may produce welfare-upgrading effects, with active users initiating feed-back mechanisms towards producers.

³ There remains, however, the question of an appropriate interpretation of empirical results, where it is difficult to separate the effect due to PKEs from the effect due exclusively to rent spillovers (see Los and Verspagen, 2007).

⁴ Costs of adoption of knowledge are inclusive of costs necessary to learn and absorb it (Cohen and Levinthal, 1990) as well as of costs of imitation (Mansfield, 1961).

⁵ Given a high degree of trade openness, there is no reason to exclude that influences coming from international commercial relations might play some role in explaining technological dynamics. However, the evidence concerning the subject is quite mixed, with studies confirming the link between imports and the sectoral growth (Wixted, 2009) and other analysis finding difficulties in delivering such an evidence (Laursen and Meliciani, 2000b). This might be due to the fact that national industrial structures are determined by long-term country-specific technological and institutional developments that prevail over influences coming from cross-border relations.

⁶ See Schumpeter (1942).

⁷ We use the tables with entries expressing the sums of domestically produced and imported intermediate inputs. These tables reflect technological relations between industries, since trade coefficients do not play a role. For Austria (2005), the Czech Republic (2005), Norway (1995–2000 and 2005), Spain (2002, 2003 and 2005) and the UK (2004, 2005), we used data from national statistical institutes to fill gaps in Eurostat data.

⁸ The countries included are Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Italy, the Netherlands, Norway, Spain, Sweden and the UK.

⁹ All variables are in constant values, taken from OECD's STAN database.

¹⁰ Some attempts have been made in this direction (Parisi et al., 2006) producing, nevertheless, very similar results to the original computation based on revenues shares.

¹¹ Solow (1992) concludes that goods and services are not much different concerning measurement problems, and attention should be paid in productivity analysis in both branches.

¹² Additionally, we tested the fixed effect model against the alternative more efficient random effect model, but the Hausman test led to the adoption of the former.

¹³ According to Düring and Schnabl (2000) the chemicals sector did not manage to maintain its leading position and, consequently, was replaced over time by metal products.

¹⁴ Differences with the results obtained from the knowledge-weighted forward linkages matrix by Cerulli and Potì (2009) derive from the fact that we link the expenditure coefficients from IO tables to TFP growth rates and not to R&D expenditures. All in all, the former measure of technological change is supposed to encompass ‘broader’ information about intersectoral technological linkages than R&D expenditures do (see Antonelli and Scellato, 2007).

¹⁵ Many previous studies confirmed that finance plays an important role in the process of growth of the whole economic system and in transmitting the most relevant positive influence as provider of technological impact (Benhabib and Spiegel, 2000; Levine et al., 2000).