Abstract
This paper employs individual firm data in order to check the existence of industry-spatial effects alongside other microeconomic determinants of R&D investment. Spatial proximity is defined by a measure of firms’ industry distance based on trade intensity between sectors. The spatial model specified here refers to the combined spatial-autoregressive model with autoregressive disturbances. In modelling the outcome for each location as dependent on a weighted average of the outcomes of other locations, outcomes are determined simultaneously. The results of the spatial estimation suggest that in their R&D decision firms benefit from spillovers originating from neighbouring industries.
Acknowledgments
I am indebted to David M. Drukker, Maurizio Pisati, and Rafal Raciborski for helpful suggestions for the spatial analysis. I thank Juan de Dios Tena-Horrillo, Giuseppe Medda, and Claudio Detotto for valuable comments on the matrix construction. I am also grateful to two anonymous referees whose comments greatly improved the quality of the paper. All the usual disclaimers apply.
Notes
1. Further discussions of spatial-weighting matrices and the parameter space for the SAR parameter can be found in Kelejian and Prucha (Citation2010) and Drukker et al. (Citation2011).
2. Some studies measure distance between firms by considering inter-sectorial flows of intermediate goods. Other works employ patents of innovations to construct technology spaces. Among others, Adams and Jaffe (Citation1996), Orlando (Citation2004) and Aldieri (Citation2011) employ a measure of geographical distance between firms, while Macdissi and Negassi (Citation2002) model the external technological spillover on the basis of firms’ resources devoted to cooperation and capital flows.
3. The data available from ISTAT (the national statistical institute) for the period considered supply information about the amount of investment in research in each industry for a sample of roughly 12,000 firms.