Regional Impact of Research and Development on Productivity

Abstract

Lehto E. (2007) Regional impact of research and development on productivity, Regional Studies 41, 623–638. This paper considers how a plant's own research and development (R&D) and the geographical and technological proximity of external R&D affect a plant's productivity. According to the results, a plant's own R&D and other firms' past R&D which are located in the same subregion and in the same industry have a positive impact on productivity. Also, other firms' R&D in the same subregion but in other industries or in other areas and in the same industry positively contributes to productivity. If the number of those units from which R&D spillovers are received increases, given the amount of external R&D, the spillover effect on productivity weakens.

Lehto E. (2007) Les retombées régionales de la R et D sur la technologie, Regional Studies 41, 623–638. Cet article cherche à examiner comment la R et D d'un établissement et la proximité géographique et technologique de la R et D externe influent sur la productivité d'un établissement. Il en résulte que la productivité d'un établissement et la productivité antérieure des autres établissements situés dans la même sous-région et dans le même secteur industriel influent sur la productivité de façon positive. En outre, la R et D des autres établissements situés dans la même sous-région mais dans d'autres secteurs industriels, ou bien situés dans d'autres régions et dans le même secteur industriel, influe sur la productivité de façon positive. Etant donné l'importance de la R et D externe, si le nombre d'établissements d'où proviennent des retombées de R et D augmente, l'impact des retombées sur la productivité s'atténue.

Retombées de R et D Proximité géographique et industrielle Productivité

Lehto E. (2007) Die regionale Auswirkung von F&E auf die Produktivität, Regional Studies 41, 623–638. Wir untersuchen, wie sich die eigene F&E eines Betriebes und die geografische und technologische Nähe von externer F&E auf die Produktivität eines Betriebes auswirken. Den Ergebnissen zufolge wirken sich die eigene F&E eines Betriebes und die frühere F&E anderer Betriebe in derselben Subregion und Branche positiv auf die Produktivität aus. Ebenso leisten die F&E von anderen Betrieben in derselben Subregion, aber einer anderen Branche, oder von Betrieben in derselben Branche, aber einem anderen Gebiet, einen positiven Beitrag zur Produktivität. Wenn die Anzahl der Einheiten steigt, von denen F&E übertragen wird, schwächt sich der U¨bertragungseffekt auf die Produktivität aufgrund der Menge der externen F&E ab.

U¨bertragung von F&E Geografische und Branchennähe Produktivität

Lehto E. (2007) Impacto regional de I + D en la productividad, Regional Studies 41, 623–638. En este ensayo consideramos qué repercusión tiene en la productividad industrial disponer de una unidad propia de I + D en la planta y de la proximidad geográfica y tecnológica de las unidades externas de I + D. Los resultados indican que obtienen mayor productividad las empresas que disponen de su propio departamento de I + D y de los antiguos resultados de I + D de otras empresas que están ubicadas en la misma subregión y en la misma industria. Y también contribuyen positivamente a la productividad la unidad de I + D de otras empresas ubicadas en la misma subregión pero en otras industrias, o en otras áreas pero en el mismo sector. Si aumenta el número de las unidades de las que se reciben los desbordamientos de I + D, se debilita el efecto de desbordamiento sobre la productividad debido al tamaño de las unidades externas de I + D.

Desbordamientos de I + D Proximidad geográfica e industrial Productividad

Keywords:

• Research and development (R&D) spillovers
• Geographical and industrial proximity
• Productivity

Notes

1. Audretsch and Feldman (1996) turn the question the other way around and explain the concentration of production and innovative activity.

2. R&D of the present data is used to develop both processes and new products. Scherer (1982), while analysing the productivity impact of R&D, instead placed R&D into two classes: the first covers R&D for internal process and imported through product purchases; the second includes R&D for one's own products sold to others.

3. For a more detailed description of data characteristics, see Husso et al. (1996).

4. In this procedure the missing values are approximated by means of observations in other years. The missing observations are calculated according to the following principles: if a firm's R&D investments in a certain municipality are not observed at some date but are observed before and after the date in question, the missing observation is approximated by the use of the observed figures. The same procedure is used if an observation is missing at the beginning or end of such a period in which a firm has been alive and active in the municipality concerned (during two or more dates). If there is only one observation of the firm's R&D but it is known from the information from the Business Register that a firm has been alive during several periods (before or after the observation), the missing value is approximated by using the level of the observed value and the average R&D growth rate in the same industry (using one-digit NACE classes).

5. There is no straightforward theoretical or empirical basis for setting this value. Customarily, δ is assumed to be around 0.1–0.2. Husso et al. (1996) assume that it is 0.1. Dilling-Hansen et al. (1999) set it at the level of 0.2. The present paper follows Orlando (1999) by assuming that δ = 0.15.

6. The fact that the plants of the largest firms in Finland – such as Nokia and firms in the paper industry – belong to the same three-digit NACE level and that multi-industry conglomerates in the metal industry and machinery are divided into many firms which each report independently to Statistics Finland also means that the simplifying practice of specifying a plant's NACE code does not cause remarkable errors.

7. In fact, in a recent empirical study that analyses the importance of distance between acquired and acquiring firms with the Finnish data (Böckerman and Lehto, 2003), the chosen way to define closeness, either location in the same subregion or distance in kilometres, does not have a considerable impact on the results obtained.

8. The procedure to obtain real values for a value added (using the two-digit industry price deflator) and to approximate capital input is described more closely by Maliranta (1997).

9. If one instead had current values of R&D variables, it would be necessary to endogenize a firm's own R&D (Mairesse and Mohnen, 2003).

10. Husso et al. (1996), in a study that uses Finnish data (a 1987–93 panel of 74 firms), report that the estimate for R&D capital elasticity is 0.08. Dilling-Hansen et al. (1999), who use Danish data, report the estimates for R&D capital elasticity as being roughly 0.08 when so-called double-counting (a consequence of the simultaneous inclusion of R&D factors in labour and R&D stock) is corrected.

11. Actually, a countrywide variable was constructed that describes academic research (in technology and data and information processing) in various locations and it was tested whether the research in universities in the same subregion had any impact on the firms' productivity. An academic research variable was added to model (1) of Table 2. It turned out that this variable and its variant, also including research in the commercial field, had no impact on a plant's productivity. (The results are not reported.) To some extent these results are somewhat opposite to the findings of Adams (2002) who considered the impact of academic research and the firms' own research on the firms' learning activities. Adams discovered that academic spillovers are more localized than industrial spillovers.

12. The restriction of the time period is explained by the technical necessity of the statistical procedure that was available at Statistics Finland when the analysis was conducted. Owing to the confidentiality of the information one is allowed to analyse, the data only at Statistics Finland.