Does Successful Innovation Require Large Urban Areas? Germany as a Counterexample

abstract

Popular theories claim that innovation activities should be located in large cities because of more favorable environmental conditions that are absent in smaller cities or peripheral areas. Germany provides a counterexample to such theories. We argue that a major reason behind the geography of innovation in Germany is the country’s pronounced legacy of political fragmentation that created a decentralized settlement structure, shaped the geographic distribution of universities and public research institutions, and brought about a rather uniform and local access to finance. We show how political fragmentation influenced the emergence of historic centers of knowledge production and impacts the positioning of innovation activities today. We conclude that institutional factors should play a more prominent role in theories that aim at explaining the spatial distribution of innovation activities.

Key words:

• innovation
• patents
• agglomeration economies
• cities
• Germany

JEL codes:

• 031
• R11
• L26

Acknowledgments

We are indebted to Christian Dienes, Jutta Günther, Maria Greve, Thilo Lang, Frank Neffke, Andrés Rodríguez–Pose, Rolf Sternberg, Mirko Titze, Friederike Welter as well as two anonymous reviewers for helpful comments on earlier versions of this article.

Notes

¹ A frequently made argument in favor of large agglomerations is their higher productivity that is reflected in higher wages, the urban wage premium (Ciccone and Hall 1996; Glaeser and Maré 2001; Carlino and Kerr 2015). This may, however, be of limited relevance because higher productivity is a static phenomenon while innovation is an inherently dynamic process. Hence, for successful innovation, it is important that places are able to manage and adapt to change.

² For differences across US metropolitan areas, see Chatterji, Glaeser, and Kerr (2014). Rodríguez‐Pose and Wilkie (2019), for example, find huge differences in the capacity of lagging regions to transform different types of R&D activity into innovation across North America and Europe.

³ In its original form the argument of borrowed size or agglomeration shadows refers to advantages of smaller cities when integrated within a larger metropolitan complex (Alonso 1973). This can, however, hardly explain the success of many smaller German cities embedded in rural regions.

⁴ Doloreux and Dionne (2008) present a case study of a peripheral region in Canada where they demonstrate that peripheral regions can have a long historic tradition of innovation activity.

⁵ Data are from the German Employment Statistics, which covers all employees subject to compulsory social insurance contributions (for details, see Spengler 2008). R&D employees are defined as those with tertiary degrees working as engineers or natural scientists.

⁶ Start-up data are from the Mannheim Enterprise Panel. For details, see Bersch et al. (2014). We use the common OECD-based classification to identify innovative industries that is based on their share of R&D inputs (Gehrke et al. 2010).

⁷ We followed the classification where Hamburg and Bremen are merged with their respective adjacent regions.

⁸ If a patent has more than one inventor, the count is divided by the number of inventors, and each inventor is assigned his/her share of that patent.

⁹ At the level of planning regions, 798 (47.2 percent) of the hidden champions are located in planning regions that are classified as agglomerations, 262 (15.5 percent) are in rural areas and 629 (37.2 percent) have a location in moderately congested regions that is the intermediate category.

¹⁰ There is a wide consensus in the literature that the typical success factors of the German hidden champions are technological leadership in a narrowly defined niche market, high-quality of products, and particularly a highly skilled workforce that plays an important role in developing new ideas and engaging in innovation (see, e.g., Audretsch and Lehmann 2016; Audretsch, Lehmann, and Schenkenhofer 2018; Pahnke and Welter 2018).

¹¹ We refer to the term political fragmentation in a broad sense, and use it interchangeably with decentralization and federalism. For a discussion of different definitions of these terms, see Rodden (2004).

¹² This persistence in decentralized structures is in line with the prediction of a political science model by Congleton, Kyriacou, and Bacaria (2003) who demonstrate that states with an initially (historically) low level of centralization tend to be systematically less centralized over time than states that were initially more centralized.

¹³ The first credit unions originated in the mid-nineteenth century and focused on traders, shop owners, and artisans, as well as farmers.

¹⁴ According to Audretsch and Lehmann (2016) these financial institutions provide about two-thirds of all financing to small- and medium-size enterprises (SMEs).

¹⁵ We have to exclude the Saarland because we lack historic data.

¹⁶ An exception is that there is a significant relationship between current population density and start-up activity. Results are available upon request.

¹⁷ The results for the robustness checks can be obtained upon request.

¹⁸ Note, that we are not dealing with the application of important process technologies in fields, such as transport and telecommunication, that may have significantly changed geographic structures.