![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
Abstract
Berlemann M. and Jahn V. Regional importance of Mittelstand firms and innovation performance, Regional Studies. Despite the deeply rooted belief of politicians all over the world about the important role of Mittelstand firms, there has yet been surprisingly little empirical research on this issue. This article contributes to the literature by studying whether the relative regional importance of Mittelstand firms has an effect on regional innovation performance. Using a cross-section of German NUTS-3 regions, a significantly positive relation between the relative importance of owner-managed small and medium-sized enterprises and patent applications is identified. This finding is highly robust when controlling for spatial correlations as they often occur in highly disaggregated regional analyses.
Berlemann M. and Jahn V. 中间阶层厂商的区域重要性与创新表现，区域研究。儘管世界各地的政治人物，皆深信中间阶层厂商具有重要角色，但出乎意料的是，此一议题却鲜少有经验研究。本文透过研究中间阶层厂商的相对区域重要性，是否对于区域创新表现具有影响，对此文献作出贡献。本研究运用德国欧盟第三层级（NUTS-3）区域的横断面，指出业主自营的中小型企业的相对重要性和专利申请之间的显着正向关係。当控制经常在高度分散的区域分析中发生的空间相关性时，此一研究发现便相当强健。
Berlemann M. et Jahn V. L'importance régionale des entreprises Mittelstand et la performance issue de l'innovation, Regional Studies. Malgré la conviction profondément enracinée des hommes politiques partout dans le monde à propos du rôle important des entreprises Mittelstand, il est étonnant que très rares soient jusqu'ici les recherches empiriques à ce sujet. Cet article contribue à la documentation en étudiant si, oui ou non, l'importance régionale relative des entreprises Mittelstand influe sur la performance du système d'innovation régionale. Employant un échantillon représentatif des régions allemandes NUTS-3, on identifie une corrélation très étroite entre l'importance relative des petites et moyennes entreprises, dont la gestion est en nom propre, et la demande de brevets. Ce résultat s'avère très solide quand on tient compte des corrélations spatiales qui se produisent souvent dans les analyses régionales faites à un niveau très désagrégé.
Berlemann M. und Jahn V. Regionale Bedeutung mittelständischer Unternehmen und Innovationskraft, Regional Studies. Trotz des tief verwurzelten Glaubens von Politikern in die Leistungsfähigkeit mittelständischer Unternehmen wurde dieses Thema bislang nur wenig empirisch erforscht. Dieser Artikel trägt zur Literatur bei, indem er analysiert, ob die relative regionale Bedeutung mittelständischer Unternehmen einen Effekt auf die regionale Innovationskraft hat. Auf Basis einer Querschnittsanalyse der deutschen NUTS-3-Regionen (Kreise) wird eine signifikant positive Beziehung zwischen der relativen Bedeutung eigentümergeführter KMUs und Patentanmeldungen identifiziert. Dieses Ergebnis bleibt auch erhalten, wenn für räumliche Korrelationen korrigiert wird, wie sie oft in stark disaggregierten regionalen Analysen auftreten.
Berlemann M. y Jahn V. Importancia regional de las pymes y el desempeño de innovación, Regional Studies. Pese a la creencia bien arraigada de los políticos de todo el mundo sobre la importante función de las pymes, se han realizado sorprendentemente pocos estudios empíricos sobre esta cuestión. Este artículo supone una contribución a la bibliografía al analizar si la importancia regional relativa de las pymes tiene un efecto en el desempeño de la innovación regional. Mediante un análisis transversal de regiones alemanas de nivel NUTS-3, identificamos una relación significativamente positiva entre la importancia relativa de pequeñas y medianas empresas gestionadas por sus propietarios y las aplicaciones de patentes. Este resultado se mantiene incluso al controlar las correlaciones espaciales que muchas veces ocurren en análisis regionales muy disgregados.
Acknowledgements
The authors acknowledge the helpful comments and recommendations made by Ralf Dewenter, Linus Holtermann and Annekatrin Niebuhr; the participants at the CESifo Regional Economics Workshop 2013 in Dresden, Germany; at the postgraduate workshop at Helmut Schmidt University and Hamburg Institute of International Economics 2013, Germany; at the Annual Scottish Economic Society Conference 2014 in Perth, UK; at the Annual Conference of the Verein für Socialpolitik 2014 in Hamburg, Germany; and at the research colloquium at RWTH Aachen, Germany. The authors also thank Tomasz Mickiewicz, Editor of Regional Studies, and two anonymous reviewers for their valuable suggestions, which helped greatly to improve the quality of the paper. Of course, all remaining errors are the authors’ alone. All results not shown explicitly in this paper are available from the authors on request.
Disclosure statement
No potential conflict of interest was reported by the authors.
Supplemental data
Supplemental data for this article can be accessed at http://10.1080/00343404.2015.1058923
Notes
1. For an international comparison of the importance of owner-managed SMEs, see Tchouvakhina and Schwartz (Citation2013).
2. The term ‘Mittelstand’ is not always used consistently in the literature. In Germany, the term goes well beyond the dimension of firm size. While there is some discussion on the most adequate definition of Mittelstand firms even in Germany (e.g., Berghoff, Citation2006), there is consensus that the most remarkable characteristic of Mittelstand firms is unity of ownership and management of an enterprise (IfM Bonn, Citation2015). Outside Germany, the term is often used as a synonym for SMEs. However, as the analysis in this paper is interested in the performance of the German specialty of small and medium-sized owner-managed firms, it sticks to the German definition.
3. Similar views are expressed by the German Institute for Economic Research (Citation2012), the Federation of German Industries (Citation2012) and Paust (Citation2014).
4. Other examples for this procedure are, for example, Audretsch and Elston (Citation1997), Icks (Citation2006), and Maaß and Führmann (Citation2012).
5. While the microdata in the REGPAT database were fully accessible, this does not hold true for the Creditreform database. While this database contains data at the micro-level, the authors were not given full access to the dataset. Instead, Creditreform assembled the necessary data for the authors at the regional level. As a consequence, matching at the micro-level was impossible. The present paper, therefore, has to rely on a macro-approach at the regional level.
6. Fehl (Citation1988) also argues in favour of a healthy mix of large and small market players in economies, however without referring explicitly to innovation.
7. As no obvious instrument variable is available for owner-managed SMEs, there are few possibilities to control for endogeneity. However, there is neither a credible theoretical argument for reverse causality nor empirical evidence pointing in this direction (e.g., Czarnitzki and Kraft, Citation2004).
8. This methodology is also used by Block and Spiegel (Citation2013) when analysing the relationship between the relative regional importance of medium- to large-scale family firms and regional innovation output.
9. Instead of controlling for industrial structure by transforming the left-hand variable, one might also think of adding variables measuring regional sector importance on the right-hand side of the estimation equation. However, as detailed data on almost 90 sectors are employed in the analysis, doing so would lead to a large loss in degrees of freedom for the estimation. The situation becomes even worse when studying spatial lags of the independent variables. The dependent variable is therefore corrected for sectoral structure.
10. For a small number of firms, no information on the location was available. These observations were dropped from the sample.
11. For a more detailed elaboration, see Berlemann and Jahn (Citation2013).
12. Note that the applied procedure has some similarities to the well-established shift–share analysis (Dunn, Citation1960; Esteban-Marquillas, Citation1972). As in shift–share analysis, this paper assumes that the regional industrial structure has an impact on regional (innovation) performance. To make the regions comparable, the measure of regional performance is corrected for industrial structure.
13. Obviously, firms with one owner-manager are clearly owner-managed firms. However, sticking to this very narrow definition would rule out many family firms, which have more than one but only a few owner-managers and thus still follow the model of the classic Mittelstand firm. The same holds true for many start-ups in which a few owner-managers founded a firm, thereby exploiting their differing job background. However, it seems not to be useful to extend the number of owner-managers by too much as then the typical advantage of owner-managed firms – flat hierarchies, little organizational slack and quick decisions – are typically lost.
14. When applying this definition, most owner-managed firms are classified as Mittelstand firms. However, a significant share of SMEs turns out not to be owner-managed.
15. As the Creditreform database contains information on turnover and employment, one might think of measuring the regional importance by the turnover (employment) share of Mittelstand firms. However, for many small but also for a considerable number of medium-sized firms, the turnover and employment data are missing, while for large firms the dataset contains no missing data at all. While owner-managed SMEs are thus classified correctly (based on the above-described definition), there is no adequate information on the turnover or employment share of Mittelstand firms. The regional importance of Mittelstand firms is therefore measured by the share of owner-managed SMEs in all regionally operating firms.
16. Alternatively, the authors experimented with using the logarithm of R&D expenditures in order to test for potentially diminishing marginal returns. However, since the regression results remained broadly unaffected by doing so, they decided to stick to the quadratic polynomial for R&D expenditures.
17. Stifterverband is a community initiative of the German economy supporting academic institutions in Germany.
18. Alternatively, the authors experimented with firms’ internal investments in R&D as a share of absolute turnover by NUTS-3 regions as a control variable. Again, this variable was also included as a quadratic polynomial in the regression equation. However, both coefficients turned out to be insignificant. Therefore, sticking to absolute investments in R&D per enterprise as a control variable seems reasonable.
19. According to the territorial boundaries of 31 December 2008, Germany consisted of 413 NUTS-3 regions.
20. does not report standardized coefficients for R&D investments as this variable enters the estimation equation in a linear-quadratic fashion. When estimating the model only with a linear R&D variable, the standardized coefficient is in a similar range but slightly below the GDP coefficient.
21. For a more detailed discussion of penalized spline smoothing, see Wong and Kohn (Citation1996) and Wood (Citation2000). For the analysis, the R-package mgcv (Wood, Citation2011) is used, which allows for a computationally stable and reliable estimation.
22. Those parts of the estimated function, which are based on relatively few observations, have to be interpreted with great care, which is already indicated by the quickly widening confidence bands in these areas.
23. Moran's I is positive (0.1851) and highly significant (0.0000).
24. The presented spatial models are based on a contiguity matrix including only regions with direct borders. Additionally, spatial models based on a contiguity matrix of second order were estimated as a robustness check. Again, the empirical findings are similar to those reported.
25. While the model fits of all spatial models estimated in this paper cannot be compared, the results of an F-test indicate that the model with spatially lagged independent variables is not a significant improvement over the simple OLS model. Thus, while spatial effects seem to play a role in this context, the effects seem to be comparatively unimportant.