Does Social Capital Reinforce Technological Inputs in the Creation of Knowledge? Evidence from the Spanish Regions

Abstract

Miguélez E., Moreno R. and Artís M. Does social capital reinforce technological inputs in the creation of knowledge? Evidence from the Spanish regions, Regional Studies. This paper seeks to verify the hypothesis that trust and cooperation between individuals, and between them and public institutions, can encourage technological innovation and the adoption of knowledge. Additionally, the paper tests the extent to which the interaction of social capital with human capital and research and development expenditures improve their effect on a region's ability to innovate. The empirical evidence is taken from the Spanish regions and employs a knowledge production function and longitudinal count data models. The results suggest that social capital correlates positively with innovation. Further, the analysis reveals a powerful interaction between human and social capital in the production of knowledge, whilst the complementarity with research and development efforts would seem less clear.

Miguélez E., Moreno R. et Artís M. Le capital social, est-ce qu'il renforce les facteurs de production nécessaires à la création de la connaissance? Des preuves provenant de l'Espagne, Regional Studies. Cet article cherche à vérifier l'hypothèse suivant: la confiance et la coopération des individus, et entre les individus et les institutions à caractère publique, peuvent encourager l'innovation technologique et l'adoption de la connaissance. Qui plus est, l'article cherche à tester jusqu’à quel point l'interaction entre les dépenses pour le capital social et pour le capital humain et pour la recherche et développement améliorent leur impact sur la propension à innover d'une région. Les preuves empiriques proviennent des régions d'Espagne et emploient une fonction de production de la connaissance et des modèles de données chiffrées longitudinales. Les résultats laissent supposer une corrélation étroite entre le capital social et l'innovation. En outre, l'analyse laisse voir une forte interaction entre le capital humain et le capital social dans la production de la connaissance, tandis qu'une corrélation étroite avec la recherche et développement s'avère moins évidente.

Capital social Capital humain Innovation Corrélations étroites

Miguélez E., Moreno R. und Artís M. Verstärkt sich durch Sozialkapital der technologische Input bei der Erzeugung von Wissen? Belege aus den spanischen Regionen, Regional Studies. In diesem Beitrag versuchen wir die Hypothese zu verifizieren, dass Vertrauen und Zusammenarbeit zwischen Einzelpersonen sowie zwischen ihnen und öffentlichen Institutionen technologische Innovation und die Übernahme von Wissen fördern können. Zusätzlich wird in dem Beitrag überprüft, inwieweit die Wechselwirkungen zwischen Sozialkapital und Humankapital sowie den Ausgaben für Forschung und Entwicklung deren Auswirkung auf die Innovationsfähigkeit einer Region verbessern. Hierfür arbeiten wir mit empirischen Belegen aus den spanischen Regionen und unter Einsatz einer Wissensproduktionsfunktion sowie longitudinaler Zähldatenmodelle. Aus den Ergebnissen geht hervor, dass das Sozialkapital positiv mit der Innovation korreliert. Darüber hinaus zeigt sich bei der Analyse eine starke Wechselwirkung zwischen Human- und Sozialkapital bei der Erzeugung von Wissen, während die Komplementarität mit den Aufwendungen für Forschung und Entwicklung weniger deutlich erscheint.

Sozialkapital Humankapital Innovation Komplementaritäten

Miguélez E., Moreno R. y Artís M. ¿Refuerza el capital social los insumos tecnológicas en la creación de conocimiento? Evidencia de las regiones españolas, Regional Studies. En este ensayo pretendemos verificar la hipótesis de que la confianza y la cooperación entre personas y entre ellas y las instituciones públicas pueden fomentar la innovación tecnológica y la adquisición de conocimientos. Asimismo en este artículo comprobamos en qué medida la interacción del capital social con el capital humano y los gastos en investigación y desarrollo mejora su efecto en la capacidad para innovar de la región. La evidencia empírica se obtiene de regiones españolas y emplea una función de producción de conocimiento y de modelos de recuento con datos de panel. Los resultados indican que el capital social tiene un efecto positivo y significativo con la innovación. Además, el análisis indica una poderosa interacción entre el capital humano y el capital social en lo que atañe a la producción de conocimiento, mientras que la complementariedad con los esfuerzos en investigación y desarrollo parece menos clara.

Capital social Capital humano Innovación Complementariedades

Keywords:

• Social capital
• Human capital
• Innovation
• Complementarities

JEL classifications:

• C23
• J24
• O33
• R11

Acknowledgements

The authors would like to thank Josep-Lluís Carrión-i-Silvestre, Enrique López-Bazo, Fabio Manca, Jaume Puig-Junoy, Vicente Royuela, Fernando Sánchez-Losada, Esther Vayá, and two anonymous referees for their helpful suggestions and comments on previous drafts of this paper. Ernest Miguélez acknowledges financial support from the Universities and Research Commissioner of Innovation, Universities and Firm, Department of the Catalan Autonomous Government, and from the European Social Fund (Grant Number 2008FI—B 00737). Rosina Moreno and Manuel Artís acknowledge financial support from the Ministerio de Ciencia y Tecnología, Programa Nacional de I+D+I (Grant Numbers SEJ2005-07814/ECON and SEJ2005-04348/ECON, respectively).

Notes

See http://ec.europa.eu/growthandjobs/faqs/background/index_en.htm/.

At this point, the study by Coleman (1988) should be mentioned, which was a pioneering study in analysing the relationship between social and human capital. However, Coleman's work was focused on the impact of different dimensions of social capital in the creation of human capital in the next generation, and not the effect of social capital on the returns on innovation of human capital, and vice versa. Besides, another interesting study from an historical viewpoint is that by Goldin and Katz (1999), which explores, among other things, why social capital was important in the expansion of secondary school education in the United States at the beginning of the twentieth century.

Contrary to previous studies (Griliches, 1979; Hausman et al., 1984), the studies by Griliches (1991a) or Jaffe (1989) shifted the unit of analysis of the knowledge production function from the firm level to the aggregate level (region, country). This shift-focus was especially interesting for estimation purposes, since it better took into account the effects of spillovers and externalities between firms and their effect on further innovation.

The World/European Values Survey was designed to enable a cross-national comparison of values and norms on a wide variety of topics and to monitor changes in values and attitudes across the world. This data collection contains the survey data from the four waves of the World Values Survey and European Values Survey, carried out in 1981–1984, 1990–1993, 1995–1997, and 1999–2004. Broad topics covered in the integrated file include perception of life, family, work, traditional values, personal finances, religion and morals, the economy, politics and society, the environment, the allocation of resources, contemporary social issues, national identity, and technology and its impact on society. The European coordination centre is located in Tilburg University in Tilburg, the Netherlands, whilst the Survey was extended globally by Ronald Inglehart from the University of Michigan in Ann Arbor, Michigan, United States.

For a brief description of the computation of the measure of social capital used in this paper, see the Appendix. For a detailed explanation of the data set and its modelling, see Pérez et al. (2005, 2006), and for interesting empirical implementations of it, see Pérez et al. (2006) and Pastor and Tortosa-Ausina (2007).

As Baltagi (2005) summarizes, panel data allow one to control for individual heterogeneity, whilst cross-section and time-series studies do not, taking into account regional time-invariant characteristics within the considered period. Moreover, longitudinal studies ‘give more information data, more variability, less collinearity among the variables, more degrees of freedom and more efficiency’ (p. 5), providing less biased and more consistent estimations of the relationships between innovative inputs and their output, which might be considered an important gap in the previous literature on this topic.

According to some scholars, the difference between patent applications and granted patents is not noteworthy – neither in absolute quantities nor in terms of weights between regions and economic branches.

Although the preferred aggregation level of analysis would have been the Spanish province (corresponding to NUTS-3), given that it is both an appropriate unit of analysis within innovation and the knowledge diffusion literature (Rondé and Hussler, 2005; Anselin et al., 1997; Acs et al., 2002), the present analysis will be performed at the NUTS-2 level due to a lack of available data on R&D expenditure.

The BDMORES database is provided by the Spanish Ministry of Economy (available at: http://www.sgpg.pap.meh.es/SGPG/Cln_Principal/Presupuestos/Documentacion/Basesdatosestudiosregionales.htm).

At this point, it should be borne in mind that spatial autocorrelation (SAC) might well arise when treating with regional aggregated data. When this possible SAC exists and is not controlled for, it would lead to an omitted variables problem, which would make the estimations inconsistent. Thus, Moran's I-test and Geary's C-test were performed for all the years of the sample (as far as the authors know, no univariate test for longitudinal data exists). Using different definitions of contiguity, it is concluded that the null hypothesis of no SAC cannot be rejected. (Results can be obtained from the authors upon request.) Moreover, for the case of Table 2, Lagrange multiplier tests were performed for SAC for longitudinal estimation residuals in the presence of significant random effects (Baltagi et al., 2003). Again, the null hypothesis of no spatial autocorrelation cannot be rejected, so it can be concluded that if any spatial process was in the data, then the explanatory variables included in the model have taken it into account. However, it should be noted that Moran's I-test, Geary's C-test and the Lagrange multiplier test are developed on the basis of a large N, while the present data set contains only seventeen spatial units. Thus, it might well be the case that these tests are failing to detect the possible presence of SAC. Additionally, the authors checked for any kind of spatial heterogeneity in the data by including a dummy for those regions located in the north-east of the country (The Basque Country, Navarre, Aragon, Catalonia, La Rioja, Community of Valencia, Murcia, and The Balearic Islands). This dummy is strongly significant, so the data may well show a separate spatial pattern. However, this dummy is not significant when the sample is split into high-income and low-income regions, so it should be assumed that the methodology already takes account of this spatial heterogeneity. (Results can be obtained from the authors upon request.) The authors would like to thank two anonymous referees for pointing out the possible existence of misspecification due to SAC or spatial heterogeneity.

The Hausman test pointed to the fixed-effects model as being the most accurate in only two regressions out of the sixteen in the tables. The authors did not perform the fixed-effects model for these two regressions in order to allow for comparability between them, which is the main aim of this study.

The authors would like to thank the referees for pointing out this identification problem between formal and informal institutions in the empirical analysis.

The delta method is just a numerical procedure, widely applied in econometrics, to approach the value of the standard error for a non-linear combination of the parameters. Basically, it expands a differentiable function of a random variable about its mean, with a first-order Taylor approximation, and then takes the variance.

The authors would like to thank the referees for pointing out this issue.

The high-income region sample includes Aragon, The Balearic Islands, Catalonia, Madrid, Navarre, The Basque Country, and La Rioja, whilst the low-income region sample includes Andalusia, Asturias, The Canary Islands, Cantabria, Castile and Leon, Castile-La Mancha, Community of Valencia, Extremadura, Galicia, and Murcia.