Abstract
This paper explores the role of internal and external knowledge in the generation of new technological knowledge. It implements the notion of recombinant knowledge generation function with the appreciation of: (i) the complementary—as opposed to supplementary—role of external knowledge and (ii) the role of the size and composition of the internal stock of knowledge. The empirical section is based upon a panel of companies listed on the main European financial markets for the period 1995–2006. The econometric analysis is based on simultaneous equations. The results confirm that R&D efforts and external knowledge are indispensable inputs into the generation of new technological knowledge.
JEL Classification::
Acknowledgements
A preliminary version has been presented to the conference “The governance of a complex world” held in Nice, 1–3 November 2012: the comments and criticisms of many participants are gratefully acknowledged. The institutional support of the Collegio Carlo Alberto, within the IPER project, is acknowledged. The authors acknowledge the useful comments of PierpaoloPatrucco, Mario Alexandre Silva, Claudia Vittori, the referees and the editor.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Email: [email protected]
2 The knowledge generation function has very often been called knowledge production function (KPF). Former use of this latter terminology can be found in Pakes and Griliches (Citation1984). Jaffe (Citation1986) instead calls it patent equation. However, there is also much literature that used to label KPF the technology production function, engendering some confusion. We use therefore the label “knowledge generation function” so as to rule out any possible ambiguity.
3 See Griliches (Citation1979, 95) note 3: “An alternative approach would complicate this model further by adding an annual knowledge production function of the form dK = H(R,K) and defining K accordingly.” (where K stands for knowledge and R for R&D activities). His subsequent contributions (1990 and 1992) fully confirm that the main interest of Griliches was the assessment of the statistical merits of patents as an indicator of R&D activities rather than the exploration of the relationship between patents and R&D expenditures as a knowledge generation function.
4 See Jaffe (Citation1986, 988): “I begin by assuming that the new knowledge produced by the firm in any period is related to its RaD in that period according to a modified Cobb–Douglas technology ….: where k, is the new knowledge generated by firm i, r, is its R&D spending, and s, is the potential ‘spillover pool’ whose construction is described above.”
5 The implementation of the IPER database has been financed by the Collegio Carlo Alberto, under the IPER project.
6 This approach reflects the assumption that all firms carry out innovative activities, although some of them do not report any innovative investment.
7 As in our model we do include the variable in logarithmic form, we obtain negative values in the descriptive statistics of lnR&D and lnCD.
8 In the calculations 4-digit technological classes have been used.
9 It must be stressed that to compensate for intrinsic volatility of patenting behaviour, each patent application is made last five years in order to reduce the noise induced by changes in technological strategy.