Indirect and feedback effects as measure of knowledge spillovers in French regions

Abstract

The aim of this article is to provide a precise measure of the role of geographical proximity in the innovation process on the French metropolitan NUTS (Nomenclature of Territorial Units for Statistics) 3 regions over the period 1995 to 2008. We study the relationship between patents applications and internal R&D, and we propose a spatial decomposition coefficient of the independent variables to measure more explicitly the spatial extent of knowledge spillovers (LeSage and Pace, 2009). Our estimation result shows that the internal R&D expenditures have a positive direct and indirect effect on the patents applications, but only for the regions with a strong innovation activity. For these regions, the spillover effect is observed in the first-order neighbourhood, but the spatial lag coefficient is not significant enough to get a positive feedback effect.

Keywords:

• knowledge spillovers
• spatial econometrics
• innovation
• indirect effect

JEL Classification:

• R11
• R12
• R15

Notes

¹ Bottazzi and Peri (2003) used the concentric circles but it is based on cross-sectional data and do not take into account the effect of heterogeneity between the regions. More, it is based on the centroids of regions (latitude and longitude) which determine the geographical localization of each department. However, if a large part of the surface of a department is outside the radius of 100 km and its centroid is inside, the entire R&D will be allocated to 100 km radius. Thus, it is difficult to determine whether it is really a knowledge spillovers or simple diffusion knowledge.

² For a review of the patent literature see Griliches (1990).

³ Descriptive analyses of all the variables may be provided if necessary.

⁴ We also test specifications wherein R&D variable is averaged over years t–1 and t–2; R&D variable becomes insignificant if it is lagged 2 years or more.