Spatial and organizational peculiarities of biotechnology in Germany

Biotechnology in Germany has an extensive history. Beginning in the 1990s several incentives offered by policymakers promoted the development of the industry around the country in general, as well as in particular regions. The BioRegio competition led to four regions receiving financial support that extended through the period from 1997-2005. Other initiatives, BioChance (1999-2003) and BioChancePlus (2004-2007) were exclusively focused on funding start-ups.

As a result, the number of SMEs in Germany that focus on biotechnology is relatively high. The data from BIOCOM AG⁷⁹ indicates that in 2005, 88% of dedicated biotechnology firms (DBFs) had less than 50 employees. In 2015 this number fell only to about 85%.⁸⁰ Thus, SMEs play an important role in the innovative landscape of German biotechnology. This notwithstanding, these SMEs must compete with a number of large corporations from the biotechnology, chemical or pharmaceutical industry. Beginning in the mid 2000s, many SMEs entered and exited the market. Some SMEs encountered financial difficulties, while others were acquired by larger corporations. This situation, caused by competitive pressure, motivated biotech SMEs to innovate radically in order to stand out and remain in the market. Another possible consequence was the fostering of communication between the most innovative SMEs and SMEs still trying to find their market niche. As not every contact, however important, may lead to successfully funded projects, spillovers influenced by social proximity may be important.

The BioRegio competition and frequent industry/university cooperation (Zucker 1998), has tended to concentrate the biotechnology industry (Belenzon and Schankerman 2013). Fornahl et al. (2011) have identified seven clusters with the highest biotechnology patent activity:

Berlin, Göttingen, Hamburg, Munich, Rhine-Main, the Rhineland and Rhine-Neckar. Most of these regional clusters correspond to the areas with highest number of dedicated biotechnology firms. Because of biotech SMEs activities there, the Jena region, also funded by the BioRegio competition,⁸¹ can also be included in this list.

79 A firm, that specializes in gathering statistics on enterprises working in the field of life sciences, particularly biotechnology.

80 This number includes both independent SMEs and subsidiaries.

81 BioRegio Initiative winning regions: Rheinland (“BioRiver”), Heidelberg (Rhein-Neckar-Dreieck), Munich and Jena

179

DBFs location, 1996 DBFs location, 2016

Fig. 1 Location of dedicated biotechnology firms in Germany

Figure 1 shows the locations of DBFs in Germany in 1996, and the most recent year with available data, 2016. As can be seen, the main centers of excellence in western and southern Germany, as well as in Hamburg and Berlin, remained stable over time. An increase in biotechnology firms can be observed in eastern Germany, Saxony-Anhalt, Brandenburg and Western Pomerania, as well as a small firm cluster in Jena. In Mecklenburg-Western Pomerania a small cluster of firms appears around Rostock, probably due to the presence of a university and Leibniz Society institutions.

Spin-offs are quite common in the industry. Scientists who develop a commercial idea may work with a firm operating on the university campus or close by. Such constellations are especially favorable for SMEs, who profit from the synergetic effects of external knowledge (Simmie 2002).

The biotechnology field in Germany is rather complex, with a wide classification of subfields (e.g., McCormick and Kautto 2013; Richardson 2012) and high interdependencies with pharmaceutical and chemical industries. In addition, biotechnology innovation in Germany tends to be based on the recombination of already existing knowledge bases (Nesta and Dibiaggio 2003). Thus, the level of cognitive proximity among biotechnology SMEs may vary greatly and influence their innovative performance.

180

3.2 Data

In order to assess the effect of different proximity measures on the knowledge diffusion of radical innovation, we must identify the number of firms involved in this process. Because patent citations are a good indicator of knowledge diffusion (Jaffe et al., 1993) and can be used to track diffusion trajectories (e.g. Bednarz and Broekel 2019; Breschi and Lissoni 2003), the focal sample of radical innovators is compiled from data used in a previous research article (Shkolnykova and Kudic 2020). In that article, radical patents were identified based on new combinations of technology classes, new to the field and associated with a high number of forward citations. Relying on patent data does have certain limitations as not everything can be patentable. However, in the case of biotechnology patents can serve as a good proxy for innovation (e.g. Pajunen and Järvinen 2018). We then identify the SMEs that have patented radical innovations, and differentiate these SMEs based on the number of employees and turnover.⁸²

A sample of radical innovators (29 firm⁸³) and their patents was identified. A total of 9082 patents from 3470 patent families were identified.⁸⁴ A second sample of citing firms and any patents held by these firms was also identified. To ensure that firms citing patents of radical innovators were actually using the knowledge and subsequent inventions that stem from the patents, a time limit was set, starting with one year from the application of the original patent.⁸⁵ The average time between the focal patent application and a citation is approximately 5.6 years.

We identified 399 firms that cited patents and also held patents of their own.⁸⁶ Out of this group of firms, SMEs were distinguished by checking the number of employees and revenue of each firm using data from BIOCOM AG and Orbis. After filtering, only 78 firms were left, including spin-offs and subsidiaries. This sample of 78 firms is hereafter referred to as ‘citing firms. The list of the firms is presented in Appendix A.

4 Methodology 4.1 Research design

In order to check the stated hypotheses, the following steps need to be conducted on the sample of 78 SMEs (citing firms) in our sample:

82 According to EU recommendation 2003/361 less than 250 employees and 50 m Euro turnover.

83 One of the firms appears in dataset under two different names

84 With the help of PATSTAT 2017b Autumn edition.

85 This stems from the idea that radical innovations change the future of the field, as well as the innovative portfolio of radical innovators.

86 Private applications were excluded.

181 Innovative performance of the firm is the dependent variable, and must be identified. We measure innovative performance by compiling the number of patents a firm applies for each year, starting from the year of the radical innovator’s patent citation and ending in 2016.

Each of the four proximity dimensions discussed in the paper must be presented as a variable.

Relying on existing literature, the values of each variable are calculated per firm per year, and used to support or reject our hypotheses.

Control variables, reflecting firm and regional dimensions were calculated for each firm for each year. An unbalanced panel was created and the estimation of the negative binomial regression was performed.

In order to prove that the proximity measures actually reflect the diffusion of radical innovators’ knowledge, we evaluated the innovative performance of a control group of firms that haven’t cited radical innovators’ patents. Thus, for every firm in the citing sample a non-citing match was created using propensity score matching. Table 1 presents the criteria used to determine the match. The criteria used are expected to influence the innovative performance of each firm (e.g. Acs et al. 2002; Beugelsdijk 2009):

Tab. 1 Criteria for matching firms

Level Description Data source

Firm level Whether the firm is stock-exchange quoted Orbis Number of employees in the firm (last available

year) Orbis

Number of firms in corporate group Orbis Regional

level R&D expenditure by NUTS 2 region,

Euro/inhabitant Eurostat

Persons with tertiary education (ISCED) and / or with a scientific and technical career by NUTS2

region, thd

Eurostat

Whether the region, where the firm is located, was

a BioRegio winner BMBF, Orbis

In our matching sample, we only included non-citing firms that were assigned to the same NACE categories as the sample of citing firms. We also identified citing firms that belong to the same corporate group as the radical innovator, and which matching non-citing firms belong to the same corporate group. After identifying the non-citing matches, we calculated the values of our proximity dimensions, and the model was evaluated with respect to our hypotheses. We expect that firms that do not cite radical innovations (our matching sample of non-citing firms) will not be involved in knowledge diffusion. Thus, we anticipate that none of the coefficients for our proximity variables will be significant for our non-citing firms, and

182 proximity to a radical innovator will not impact innovative performance.⁸⁷ The list of the matching firms is presented in Appendix A, and the results of the analysis are presented in Appendix B.