Regional Specializations in Green Incumbents and Green Start‐ups in the German Transport Sector

Hansmeier, Hendrik; Losacker, Sebastian; Bersch, Johannes; Kroll, Henning Article — Published Version Regional Specializations in Green Incumbents and Green Start‐ups in the German Transport Sector Growth and Change Provided in Cooperation with: John Wiley & Sons Suggested Citation: Hansmeier, Hendrik; Losacker, Sebastian; Bersch, Johannes; Kroll, Henning (2025) : Regional Specializations in Green Incumbents and Green Start‐ups in the German Transport Sector, Growth and Change, ISSN 1468-2257, Wiley, Hoboken, NJ, Vol. 56, Iss. 1, https://doi.org/10.1111/grow.70025 This Version is available at: https://hdl.handle.net/10419/319367 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Growth and Change - ORIGINAL ARTICLE OPEN ACCESS Regional Specializations in Green Incumbents and Green Start‐ups in the German Transport Sector Hendrik Hansmeier 1,2 | Sebastian Losacker 3,4 | Johannes Bersch 5 | Henning Kroll 1,2 1 Fraunhofer Institute for Systems and Innovation Research ISI, Karlsruhe, Germany | 2 Institute of Economic and Cultural Geography, Leibniz University, Hannover, Germany | 3 Department of Geography, Justus Liebig University, Hessen, Germany | 4 CIRCLE–Center for Innovation Research, Lund University, Lund, Sweden | 5 ZEW ‐ Leibniz Centre for European Economic Research, Mannheim, Germany Correspondence: Hendrik Hansmeier ([email protected]) Received: 1 September 2023 | Revised: 23 September 2024 | Accepted: 3 January 2025 Keywords: actors | eco‐innovation | green regional development | incumbents | regional specialization | start‐ups ABSTRACT The regional variety of actors is considered a key determinant in the last decade's rich literature on the geography of eco‐ innovation and green regional development. However, little is known about the extent to which regions differ in their specialization in new and established eco‐innovation actors. In this article, we propose a regional typology based on green specializations concerning both incumbents and start‐ups in the German transport sector. While many regions show green specializations in either start‐ups or incumbents, only some regions manage to specialize in both. We find that the above‐average regional specialization in eco‐innovation does not seem to be primarily a phenomenon of urban areas, but rather depends on regions' human capital endowments and technological capabilities. The observed heterogeneity in eco‐innovation specializations, both in innovation centers and lagging regions, calls for regional policies that are more sensitive to these differences. 1 | Introduction The greening of the economy is decisive to achieve transformative change toward environmental sustainability. The transport sector is under particular pressure to transform; besides negative environmental impacts resulting from land sealing, local pollution and microplastic waste (Baensch‐Baltruschat et al. 2020), it is largely based on fossil fuels, accounting for approximately one quarter of energy‐related and 15% of global greenhouse gas (GHG) emissions (IPCC 2022). As more goods and passengers are transported by road, rail, air and water, sector emissions in countries of the Global North are even higher (20%–30%), with no significant decline observed in recent years (ClimateWatch 2022; Fransen et al. 2019). Countries and regions differ not only in terms of causing global environmental challenges, but also in addressing and solving them. The latter is the focus of research on the geography of eco‐ innovation and sustainability transitions that has emerged in the past decade (Hansen and Coenen 2015; Hansmeier and Kroll 2024; Losacker et al. 2023), pointing to the importance of place‐specific and regional factors for eco‐innovation processes. Geographical and related forms of proximity facilitate the emergence, diffusion and application of knowledge and innovation through access to networks and resources. As analyzed by previous work, the place‐specific nature of green technology, industry and regional development is mainly due to (in‐) formal institutions including policies (e.g., Bækkelund 2022; Bugge, Andersen, and Steen 2022), technological capabilities (e.g., Corradini 2019; Santoalha and Boschma 2021) as well as local actors (e.g., Binz, Truffer, and Coenen 2016; MacKinnon et al. 2019). In the literature, numerous studies focus on the specific roles of different groups of regional eco‐innovators as regional This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Author(s). Growth and Change published by Wiley Periodicals LLC. Growth and Change, 2025; 56:e70025 1 of 15 https://doi.org/10.1111/grow.70025 conditions for eco‐innovation development. These include primarily actors from industry, politics and academia, but also intermediaries and societal actors (e.g., Gibbs and Jensen 2022; Trippl et al. 2020). However, the relationship between established and new actors is also crucial to increase our understanding of regional sources of transformative change (van den Berge, Weterings, and Alkemade 2020). Although innovation studies have analyzed the influence of regional contexts on the emergence of eco‐innovations in incumbent firms (e.g., Horbach and Rammer 2018; Klitkou and Coenen 2013) as well as in green start‐ups (e.g., Abdesselam, Kedjar, and Renou‐Maissant 2024; Corradini 2019), both aspects have hardly been considered together in order to better grasp the regionally prevalent balance of different eco‐innovation actors. On the other hand, transition studies' interest in stability and change is reflected both in the focus on the forces of inertia and in novel developments depending on experimental alignment (Boschma et al. 2017; Gibbs and Jensen 2022). While overarching studies on different combinations of regional actors are missing, recent studies argue, however, that incumbents operate quite heterogeneously and do not only preserve the status quo, as often assumed. Accordingly, change can emerge from both incumbents and new entrants (Steen and Weaver 2017; Strambach and Pflitsch 2020; Turnheim and Sovacool 2020). The aim of this paper is to gain a better understanding of regions' green specialization, that is the degree to which innovation activities in the transport sector have become greened in the domains of (economic) incumbents and start‐ups. Eco‐ innovative activities in the German transport sector are of crucial importance not only from an ecological point of view, but also from an economic one. The automotive industry alone comprises around 1000 companies (manufacturers and suppliers) with around 830,000 employees, with medium‐sized to large companies dominating (Destatis 2023). Transport‐related industries are of considerable economic importance across regions, which makes the sector an ideal subject for investigation, given that this study focuses on regional particularities of eco‐ innovation emergence. Following Hockerts and Wüstenhagen (2010), we conceive green incumbents as organizations, usually firms, that are well established and engage in eco‐innovation activities, while green start‐ups are recently founded companies with an innovative and environmentally friendly business model or product. By focusing on regional specializations, we intentionally focus less on how the joint existence of established and newly founded companies develops (which would be a biased comparison), but rather on the extent to which the two types of actors in a region contribute to the process of greening. In addition to the general question of how regions differ in terms of relative actor specializations, we are particularly seeking to analyze the joint regional specialization of green transport incumbents and start‐ ups. Finally, we test which regional conditions explain the observed spatial patterns. Empirically, we estimate the greening of incumbents based on environmentally oriented patents, which are almost exclusively filed by large firms in the transport sector (Eurostat 2014), and green start‐ups based on data from the Mannheim Enterprise Panel (MUP) for the period 2009–2018 in the 96 labor market regions in Germany. As we analyze not only technological change via patents but also business model or service innovations, the combined analysis allows us to examine the broader eco‐innovative change necessary for sustainability transitions in the transport sector (Tödtling, Trippl, and Desch 2022). The outline of this paper is as follows. The next section presents the theoretical and empirical status quo of geographical research on eco‐innovators with a particular focus on the transport sector. Based on this, research questions are derived. Subsequent sections describe the data and methods of this study before we present key findings. This is followed by a discussion of the results and a conclusion, pointing to further research possibilities and policy implications. 2 | Regional Perspectives on Eco‐Innovation Incumbents and Start‐Ups in the Transport Sector Innovations are the result of new combinations of goods, skills, knowledge, techniques and resources. This implies the need for collaborating and interactive relations, which led to the systemic understanding, that is sectoral, technological, regional/national, of innovation emergence and application. From the perspective of geographical research on eco‐innovations and sustainability transitions, questions arise not only about spatial differences in technology and industrial development (Capasso et al. 2019; Tödtling, Trippl, and Frangenheim 2020), but also about the relevant actors. More precisely, whether, according to the Schumpeterian dichotomy, the necessary change is primarily carried out by established actors or whether new entrants are required and how regions differ in this respect (Trippl et al. 2020; Turnheim and Geels 2019). 2.1 | The Role of Incumbents in Eco‐Innovation Activities Given the heterogeneity of actors that constitute complex innovation and socio‐technical systems, a multiplicity of incumbents can be identified. Hence, incumbents exist and act to varying degrees at different (spatial) levels and social spheres, with incumbent behavior particularly visible among powerful firms and government actors (Ansari and Krop 2012; Turnheim and Geels 2019). Incumbent actors are closely linked to the regime concept of transitions research, according to which routine agency of established actors reproduces (in‐)formal institutions and rules, which in turn stabilize socio‐technical systems that lead to path dependencies and incremental changes (Miörner and Binz 2021). These path‐dependent developments are the result, for example, of incumbent firms predominantly adhering to existing and potentially successful business logics, which slows down or prevents transitions toward sustainability (Bohnsack, Pinkse, and Kolk 2014; Steen and Weaver 2017). With regard to the transport sector, these patterns are particularly evident for the globally dominant automotive regime that has evolved for more than a century, leading to stable 2 of 15 Growth and Change, 2025 production and consumption practices (Sengers and Raven 2015). This regime has been manifested through technical, institutional and social adaptations with little system‐changing innovation activities over the last decades (Sovacool, Noel, and Orsato 2017). Although established firms are active in regional and national research efforts and production, they are also strongly integrated into international markets and collaborations, promoting similar transport modes around the world (Nilsson, Hillman, and Magnusson 2012). Despite these characteristics, however, innovation and transition studies point to a more diverse picture of incumbents (e.g., Ansari and Krop 2012; Steen and Weaver 2017) and pluralizing perspectives on incumbencies across regions and sectors (Turnheim and Sovacool 2020). These views are based on the findings that established firms and other incumbents may pursue eco‐innovation development. Although incumbent firms seem less likely to be initial leaders of environmentally oriented innovation, differences emerge depending on the industry environment and firm properties, the mode of innovation, as well as challenges from new entrants (Ansari and Krop 2012; Tsouri, Hanson, and Normann 2021; Turnheim and Sovacool 2020). In addition, there are varying institutional field logics, such as differences in standardization (Dewald and Achternbosch 2016), which are likely to translate into spatial differences (Miörner and Binz 2021). For the transport sector, incumbents are significantly involved in the development of environmentally friendly transport technologies and business models in regions, for example, shared or autonomous mobility (Bohnsack, Pinkse, and Kolk 2014; Meelen, Frenken, and Hobrink 2019; Nilsson and Nykvist 2016). At the same time, it was found that incumbents can also influence the institutional framework in a way that promotes transforming the transport sector at the regional level (Bugge, Andersen, and Steen 2022; Miörner and Trippl 2019). Against this background, the geography of eco‐innovation has received increasing scientific attention over the past decade (Hansmeier and Kroll 2024; Losacker et al. 2023). While the complexity of green transitions and transformative change is characterized by multi‐scalar interdependencies in innovation processes, networks of actors and knowledge dynamics (Capasso et al. 2019; MacKinnon et al. 2019; Tsouri, Hanson, and Normann 2021), numerous geographical studies point to the importance of diversified capabilities and knowledge bases at the regional level. In the latter case, the relatedness to pre‐ existing capabilities is of particular relevance for green regional diversification (Montresor and Quatraro 2020; Santoalha and Boschma 2021). As such, a narrow regional specialization seems insufficient (Coenen 2015). Since the green transition will affect the transport sector in particular, regions with incumbent structures will be called upon to address change processes at an early stage and to develop alternative diversification pathways. Otherwise, these regions will be confronted with particularly negative developments (Rodríguez‐Pose and Bartalucci 2023). In this context, Miörner and Trippl (2019) point to the importance of establishing new networks with transport stakeholders to ensure access to and co‐create knowledge about more sustainable transport solutions in the future. Although green innovation activities of firms in urban environments are not necessarily influenced by spatial externalities (Galliano, Nadel, and Triboulet 2023), agglomeration effects, clustering, co‐location, infrastructure density and linkages between different actors, which are fundamental to many urban areas, are found to positively impact overall eco‐ innovation output (Del Río, Peñasco, and Romero‐Jordán 2016; Grillitsch and Hansen 2019; Horbach 2020) and green innovation activities in the transport sector in particular (Carvalho, Mingardo, and van Haaren 2012; Krauss, Krail, and Axhausen 2022). This is in line with the frequently confirmed finding that innovation activities are concentrated in more densely populated regions (e.g., Balland et al. 2020; Mewes and Broekel 2022). 2.2 | The Role of Start‐Ups in Eco‐Innovation Activities Eco‐innovative start‐ups are crucial in transforming production and consumption patterns, both within and beyond the region. They introduce environmentally friendly product innovations and green technologies, but also new business models and services (Abdesselam, Kedjar, and Renou‐Maissant 2024; Colombelli and Quatraro 2019). New entrants seem to be particularly influential in the early stages of an industry transformation (Hockerts and Wüstenhagen 2010), as they bundle new products and services in a unique way or generate new knowledge and capabilities that notably deviate from previous knowledge bases. Start‐ups can therefore boost eco‐innovation activities and influence established players (Bohnsack, Pinkse, and Kolk 2014; Dewald and Achternbosch 2016; Nilsson and Nykvist 2016). This is also true for the transport sector, in which start‐ups play a leading role in the development of eco‐innovation despite the above‐mentioned strong regime structures. In the case of electric mobility, for example, start‐ups seem to have developed new products and business models around the same time as incumbents (Bohnsack, Pinkse, and Kolk 2014; Sovacool, Noel, and Orsato 2017; Tödtling, Trippl, and Desch 2022). Against this background, conceptual and empirical research on knowledge spillovers shows that start‐ups benefit greatly from unused and non‐commercialized ideas, knowledge and network resources of established actors (e.g., Cojoianu et al. 2020; Colombelli and Quatraro 2019; Doblinger, Surana, and Anadon 2019; Klitkou and Coenen 2013). As such, both extra‐ and intra‐regional linkages are seen to positively affect green start‐ up emergence. Although the importance of existing structures varies greatly between sectors (Dewald and Achternbosch 2016), geographers in the field of regional studies and transitions research argue that entrepreneurial activities of start‐ups and spin‐offs are an essential mechanism of regional branching into related and new industries (Boschma et al. 2017; MacKinnon et al. 2019). Green industrial development in particular benefits from emerging innovation actors, as they are especially capable of developing skills and knowledge combinations beyond region‐internal path dependencies. The process of green path creation is therefore more likely to occur in regions where the support structures and the existing skill base are well‐developed (Gibbs and Jensen 2022; Tödtling, Trippl, and Frangenheim 2020; Trippl et al. 2020). Based on the above, we derive the first research question as follows: 3 of 15 RQ1: Are those regions that display a specialization in green start‐ups also more specialized in green incumbents? Start‐ups are particularly dependent on the external availability of knowledge and networks, not only due to the greater complexity of environmental innovations but also because of limited internal resources (Horbach 2020). Against this background, a large number of recent studies have examined regional determinants of green start‐up emergence. First, it has been shown that the knowledge accumulated in a region is positively associated with the emergence of green start‐ups (e.g., Coll‐Martínez, Malia, and Renou‐Maissant 2022b; Colombelli and Quatraro 2019; Horbach 2020). As new actors extensively build on pre‐existing structures, a differentiated but complementary knowledge pool is particularly crucial (Giudici, Guerini, and Rossi‐Lamastra 2019; Vedula, York, and Corbett 2019). Likewise, Cojoianu et al. (2020) and Coll‐Martínez, Jové‐Llopis, and Teruel (2022a,2022b) report a positive correlation between regional human capital endowment as well as the presence of research institutions and the emergence of new eco‐innovation entrants. Looking at the technological dimension, it is interesting to note that spillovers from both green and non‐green knowledge creation seem to influence green regional start‐up activities (Cojoianu et al. 2020; Colombelli and Quatraro 2019). In addition to skills and technological capabilities, regulation and (regional) policies influence green start‐up activities. For example, by directly promoting eco‐innovative and environmentally friendly start‐ups, negatively impacting non‐green start‐ups or creating demand incentives, thereby strengthening market certainty for green start‐ups (Giudici, Guerini, and Rossi‐Lamastra 2019; Hoogendoorn, van der Zwan, and Thurik 2020; Horbach 2020). In line with these findings, Bioret, Dechezleprêtre, and Fadic (2021) show that the majority of venture capital has been invested in green transport start‐ups in recent years. Moreover, numerous studies show that regions' informal institutional framework conditions are crucial. As such, social norms, shared meanings, behavioral patterns and environmental awareness influence firm formations and green entrepreneurship at the regional level. On the one hand, these can lend additional legitimacy to environmentally friendly technologies and practices, and on the other hand, they can simplify opportunity recognition. Both factors increase regional eco‐innovative firm entry (e.g., Cojoianu et al. 2020; Tödtling, Trippl, and Frangenheim 2020; Vedula, York, and Corbett 2019). Given the variety of regional and sectoral specificities that influence eco‐innovation activities, we formulate the second research question as follows: RQ2: Which external determinants help to explain (different types of) regional specializations in green incumbents and green start‐ups? 3 | Methodology To identify and explain activities of eco‐innovation incumbents and start‐ups at the regional level, we use two different data sources. We use patent data to proxy eco‐innovation activities of incumbents. Of the company‐related patents filed at the European Patent Office (EPO), the majority are held by large corporate entities; in the field of transport even more than 95% (Eurostat 2014). Accordingly, we can use patent applications as a very direct proxy of inventive activities driven by incumbents. In contrast, activities of new entrepreneurial market entrants can be measured directly by the number of start‐ups. Given the novelty of activities in the domain of green transport, a large share of new firms' activities can thus be considered innovative (e.g., Cojoianu et al. 2020; Vedula, York, and Corbett 2019). Using both regionalized patent and start‐up data, we then derive methodological approaches to analyze the regional prevalence of green incumbents and green start‐ups. 3.1 | Patent Data Patents are a key measure in innovation research and related fields to study the emergence and diffusion of knowledge and technological inventions. The use of patent data is accompanied by limitations, such as the prevailing focus on technologies, the low informative value with regard to the quality and impact of inventions and the possible non‐patentability thereof. However, patents are particularly valuable for geographical analyses because they allow to trace the production of (technological) knowledge in a comprehensive and spatially nuanced way (Griliches 1990; van den Berge, Weterings, and Alkemade 2020). This study draws on the PATSTAT database of the European Patent Office (EPO). In a first step, we collect all patent applications with priority dates from 2009 to 2018 and assign them regionally using the inventors' addresses. Like van den Berge, Weterings, and Alkemade (2020), we use whole counts in the case of multiple inventors, hence we consider knowledge to be a non‐divisible good. Since an overly small‐scale approach would risk that many regions have no or only a small volume of patents, we resort to the 96 German labor market regions. These represent functional territorial units 1 and are generally composed of several districts (“Kreise”/NUTS3 regions). At the same time, this regional setting allows annual socioeconomic data to be aggregated from the municipal or district level, whereas for smaller territorial units this data is often missing, particularly sectoral information to adequately estimate innovation activities. In a second step, we aim to identify all patents from the transport sector as well as those classified as green transport technologies. For the former, we draw on the WIPO technology concordance, which links codes of the International Patent Classification (IPC) to technology fields (Schmoch 2008). To identify green transport technologies, we rely on the Y02 class of the Cooperative Patent Classification (CPC), an extension of the IPC (see Table A1 for a detailed overview of data sources and search strategies). The subgroup Y02 T contains patents related to climate change mitigation technologies in the transport sector, covering road, rail, air and maritime/waterways transport technologies as well as enabling technologies such as charging of electric vehicles, and fuel cells (Coll‐Martínez, Malia, and Renou‐Maissant 2022b; Haščič and Migotto 2015; van den Berge, Weterings, and Alkemade 2020). Thus, for the 10‐year observation period, we identified 46,228 patents in the 4 of 15 Growth and Change, 2025 technology field of transport and around 15,000 patents are classified as green transport technologies. 3.2 | Start‐Up Data The data to identify (green) start‐up are taken from the Mannheim Enterprise Panel (MUP) which is generated by the ZEW– Center for European Economic Research since 1992. By cooperating with the largest German credit rating agency “Creditreform e.V.,” which contributes data twice a year on the total German corporate landscape, the MUP provides a comprehensive micro database on legally independent companies. As a result, parent companies and subsidiaries are reported separately. In total, it contains detailed information on around 9 million economically active and closed companies (Bersch et al. 2014). Unlike research that identifies and examines green start‐up activities via survey data (e.g., Hoogendoorn, van der Zwan, and Thurik 2020; Horbach 2020), the MUP, with a coverage of about 90% of the full stock of firms, 2 allows for more representative studies of business dynamics over longer observation periods and small‐scale disaggregation (Bersch et al. 2014). Although the mere number of start‐ups does not necessarily reflect their quality or impact (Cojoianu et al. 2020), the measure in fact offers a suitable way of covering entrepreneurship activities of a wide range of economic actors at the regional level. Of particular importance for the analyses in this paper is the information on the company address for the spatial allocation, the date of the foundation for the temporal limitation, and the classification of economic activities (NACE Rev. 2). The latter is used to exclude service‐oriented and predominantly non‐ innovative sectors and to delineate the transport sector (see Table A1). Again, we use the years 2009–2018 as the observation period and the 96 German labor market regions as the units of analysis (cf. chapter 3.1). Green start‐ups in the transport sector cannot be identified by sector assignments. Instead, we screen the company descriptions contained in the MUP using a keyword‐based search strategy, following previous studies with a similar approach (Hansmeier and Losacker 2024; Shapira et al. 2014). In order to limit the bias in the identification of start‐ups, which is also due to the fact that many newly born (micro) enterprises and start‐ ups have imprecise company descriptions, we follow a broad search approach. We derive the technology and sector‐specific search terms for green transport start‐ups from various databases and sources: the Y02 T patent class, transport start‐ups listed in the “StartGreen” network 3 and the work by Cojoianu et al. (2020) (Table A 1). To check the validity of the identified companies, we screened their brief description and conducted internet searches where necessary. This step eliminated those that are not applicable (e.g., companies primarily active in the energy sector). On the other hand, we add those green transport related start‐ups that were not initially found, but appear in the “StartGreen” network or received funding by the German Federal Environmental Foundation (DBU). 4 Overall, of the 42,000 start‐ups related to the transport sector, about 900 can be classified as eco‐innovative. 3.3 | Regional Patent and Start‐Up Specializations Comparing regions according to their relative progress in green incumbents and green start‐ups is the main objective of this paper. Technically, however, pure shares are often difficult to compare as their averages and variance differ between domains. It will also be important for the analysis to determine whether eco‐innovation activities of incumbents and start‐ups in regions are above or below the national average. An established and common practice in (economic) geography research to achieve this double objective is to measure specialization based on the location quotient (LQ). The LQ allows to determine whether a region is technologically or economically specialized compared to all other regions over a given period of time. Usually, LQ >1 indicates above‐average specialization, LQ =1 average specialization and LQ <1 below average specialization. To minimize distortions by outliers, we normalize the specialization indicators by means of a log transformation between þ100 and −100, with positive values indicating above‐average regional specializations (Losacker and Liefner 2020). Similar to the seminal work of Balland and Rigby (2017) who draw on relative technology advantage measures, we will refer to the normalized LQs as the relative patent advantage (RPA) and relative start‐up advantage (RSA) (see Equations 1a and 1b). For each region rin the green transport domain gthe indicators are given by: RPArg =100 ×tanhln [prg/∑rpg prt/∑rpt](1a) RSArg =100 ×tanhln [srg/∑rsg srt/∑rst](1b) Both specialization measures are calculated as the relation of green transport activities to patent por start‐up activities sin the transport sector t. This allows to control for appropriate baselines, as regional specializations in environmentally friendly innovation activities might depend on strong activities in the overarching transport sector. 4 | Regional Specializations in Green Transport Incumbents and Start‐Ups Against the background of this paper's core empirical prospect, we are interested in identifying regional specializations in novel and established eco‐innovators. Irrespective of the absolute innovation capacity, we seek to determine whether the current share of green actor activity in a region's transport sector is above or below national average. By analyzing this situation separately for both incumbents and start‐ups, it is possible to distinguish between different types of regional eco‐innovation actor specializations. 5 of 15 4.1 | Typology of Regions The comparison of the relative patent and start‐up advantage enables us to create a typology of the 96 German spatial planning regions. The RPA and RSA are determined for each region for the entire period from 2009 to 2018, allowing to depict four different types of regional eco‐innovation actor specialization in the transport sector (Figure 1). 1. Hotspot: Region with both above‐average green start‐up and incumbent specialization. 2. Start‐up‐driven: Region with above‐average green start‐up specialization but below‐average green incumbent specialization. 3. Incumbent‐driven: Region with above‐average green incumbent specialization but below‐average green start‐up specialization. 4. Laggard: Region with both below‐average green start‐up and incumbent specialization. When looking at the distribution of regions by type, it is noticeable that 35 of the 96 regions have neither a start‐up nor an incumbent specialization. These so‐called laggard regions represent the largest share (approximately 37%), followed by 29 regions in which the greening in the transport sector is driven to an above‐average extent by start‐ups but less by established actors. Conversely, 18 regions are incumbent‐driven and at the same time have a below‐average green start‐up specialization. Only 14 regions, that is a minority of roughly 15%, can be categorized as green transport hotspots, featuring specialization in both green start‐ups and green incumbents. Overall, this typology reveals a relatively heterogeneous and ambiguous picture of regional eco‐innovation actor specializations in the transport sector in Germany. Answering research question 1, this initial finding is remarkable in that it neither suggests that green transport start‐ups and incumbents mainly emerge in spatial proximity due to the importance of knowledge spillovers and technological specialization, nor does it seem to support that eco‐innovations primarily emerge outside established structures, as conceptualized in transition studies. In the FIGURE 1 |Regional typology according to eco‐innovation actor specializations in the transport sector in Germany (2009–2018). 6 of 15 Growth and Change, 2025 latter case, we would expect a more negative correlation between incumbent and start‐up specialization, while a co‐occurrence of start‐ups and incumbents would show a positive correlation between RPA and RSA, that is that regions fall primarily into either the laggard or the hotspot category. Rather, our descriptive results corroborate the findings of recent work. Not only does the importance of specific eco‐innovators vary between regions (e.g., Bugge, Andersen, and Steen 2022; Trippl et al. 2020), but it is also evident that start‐up and incumbent activities take place where the respective other is less dominant, indicating the influence of other/non‐green regional determinants (e.g., Colombelli and Quatraro 2019; van den Berge, Weterings, and Alkemade 2020). We do not observe that green transport start‐ups are more likelyto emerge in regions that are also specialized in eco‐innovation transport incumbents. Only a few regions feature a specialization in both green incumbents and green start‐ups. Furthermore, Figure 2shows that the different regional specializations of eco‐innovation actors can be found across Germany. At the same time, our results indicate that certain spatial clusters of the specific types of regions emerge. Green transport hotspots and start‐up‐driven regions predominate in the southern and eastern parts of the country. Consistently, lagging regions cluster in the center and (north) west of Germany, while incumbent‐driven regions seem to be evenly distributed. Interestingly, important locations of the automotive industry, such as Wolfsburg, Munich, Ingolstadt or Stuttgart, are mainly start‐up driven or even regional hotspots of eco‐innovative transport activities. This finding suggests that sectoral greening is also taking place in those regions that are characterized by strong industrial regime structures. The fact that hotspots rarely appear in isolation from start‐up or incumbent regions could also indicate inter‐regional spillover effects (Guo et al. 2020). Although some sparsely populated regions are considered actor hotspots, the geographical distribution seems to correlate with the population density in general. Accordingly, none of the depicted large cities has a below‐average specialization with both groups of actors. In order to better compare the regions, Table 1provides descriptive statistics for several socio‐economic indicators. In line with previous studies, we refer to key regional factors that have been shown to influence (green) innovation activities, particularly in the transport sector. These include, for example, the capital stock (GDP per capita), the population density to measure agglomeration externalities and the availability of high‐ skilled individuals measured as the share of employees with an academic background, that is tertiary education (e.g. Cojoianu et al. 2020; Coll‐Martínez, Malia, and Renou‐Maissant 2022b; Corradini 2019). As a sector variable, we also consider the regional motorization rate, which is calculated from the number of cars per 1000 inhabitants. In fact, the aforementioned finding seems to be confirmed, according to which regional hotspots of eco‐innovators related to transport tend to be more urban, that is have on average a higher population density. Moreover, these regions have the comparatively highest GDP per capita, the largest share of FIGURE 2 |Region types of eco‐innovation actor specializations in the transport sector in Germany (2009–2018). 7 of 15 highly qualified people and the fewest cars per inhabitant. These correlations are almost consistent across the different types of regions, with the exception of start‐up driven regions that have on average fewer highly qualified people than incumbent regions and both the lowest GDP per capita and population density. While these findings differ to some extent from previous studies in which start‐ups appear to be particularly dependent on qualified personnel and profit from agglomeration effects (e.g. Giudici, Guerini, and Rossi‐Lamastra 2019; Hoogendoorn, van der Zwan, and Thurik 2020; Horbach 2020), it should be noted that in this study we consider all firm births as start‐ups without explicitly controlling for their innovativeness. 4.2 | Explanation of Regional Differences To validate the descriptive results and address our second research question, we make use of econometric models. To this end, we predict the regional classification into the proposed typology (see Figure 1). In a first step, the focus is on estimating the probability that a region is an eco‐innovation hotspot in the transport sector. For this objective, a logistic regression model with robust standard errors by regions is estimated, where the dependent variable is dichotomous and takes the value 1 if a region is specialized into both eco‐innovation incumbents and start‐ups (RPA & RSA >0), otherwise it takes the value 0. In order to avoid biases driven by little innovation activities and outliers, we have summed the start‐up and patent counts for the corresponding year with the two previous years (moving window), resulting in a panel data set for the dependent variable covering the period 2011–2018. In addition, we use a one‐year time lag 5 for the independent variables, which thus cover the years 2010–2017. The logistic regression model is given by: logit(Typer,t)=α+β1Population densityr,t−1 +β2GDP per capitar,t−1 +β3High ‐ skilled individualsr,t−1 +β4Car industry siter,t−1 +β5Large enterprisesr,t−1 +β6R&D ‐ intensive industriesr,t−1 +β7Public transportr,2020 +β8Green votesr,t−1+θt+εr (2) where the subscripts rand tdenote the region and time period, θ t is the dummy variable for each time window (time fixed effect) and εis the error term. We include a number of common regional‐level factors (see also Table 1): population density, GDP per capita and share of high‐skilled individuals. Moreover, we include a region's share of large enterprises to proxy the influence of multi‐national companies and a dummy variable indicating main locations of the automotive industry (headquarter of car manufacturers with more than 10,000 employees). The regional share of employees in research‐intensive industries reflects technological capabilities, while a further dummy variable indicates whether the distance to the nearest public transport stop is above or below German average (values for 2020). The regional shares of votes for the political party “the Greens” in the 2009, 2013 and 2017 federal elections serve as a proxy for ecological ideals and lifestyles (e.g. Coll‐Martínez, Malia, and Renou‐Maissant 2022b; Horbach 2020). 6 The detailed descriptions of the variables are included in the annex (Table A2). In a second step, we employ a multinomial logit model, considering all four types of regions in the dependent variable. That is, the multinomial model uses the group of laggard regions as a baseline and estimates the probability that a region falls in one of the other groups as opposed to the reference group. This regression model takes a similar form as the logistic model described above, except that the probability distribution of the outcome variable is multinomial not binomial. Table 2presents the regression results, where model (1) is the binomial logistic model and models (2), (3) and (4) belong to the multinomial logistic regression. Likelihood ratio tests indicate that all models are significantly better at explaining differences among the groups than intercept‐only models. The models do not suffer from multicollinearity as indicated by correlations (see Table A2) and variance inflation factors (<5). We are aware that there might be minor problems with endogeneity for some variables, which are addressed by the use of time‐lags and the implemented panel structure. However, despite their robustness to the use of different time lags, the regression results cannot be seen as reflecting causality in the strict sense. In terms of the research question, however, they can be seen as contributions and explanations rather than mere correlations. In doing so, we follow the common practice in the geographical literature of using analytical statistics to partially explain effects that may be complex and difficult to disentangle (see e.g. Horbach, Oltra, and Belin 2013). The main results presented below are robust to several changes in the econometric strategy. That is, results hold when using subsets of the original data set (e.g. excluding years/ TABLE 1 |Mean values of socio‐economic and sectoral characteristics by type of region. Type of region N Population per km 2 GDP per capita (EUR) High‐skilled individuals (%) Cars per 1000 inhabitants Hotspot 14 325 37,941 17.2 496 Start‐up‐driven 18 180 32,146 11.5 543 Incumbent‐ driven 29 285 35,194 13.5 514 Laggard 35 201 33,467 11.0 551 Total 96 228 34,510 13.0 529 Source: Calculations based on data provided by Federal Institute for Research on Building, Urban Affairs and Spatial Development (2022). 8 of 15 Growth and Change, 2025 TABLE A2 |Descriptive statistics and correlations of variables. Variable Description Mean SD Min. Max. 1 2 3 4 5 6 7 8 1 Population density Population per km 2 329 502 42 4055 1 2 GDP per capita Gross domestic product (GDP) per capita (in EUR) 32,676 8220 17,935 66,429 0.35*** 1 3 High‐ skilled individuals Share of employees with an academic degree 11.3 3.8 5.0 28.3 0.49*** 0.56*** 1 4 Car industry site Location of a car manufacturer with more than 10,000 employees (1 if yes) 0.1 0 1 0.10*** 0.37*** 0.31*** 1 5 Large enterprises Share of companies with more than 250 employees 0.3 0.1 0.1 0.6 0.32*** 0.55*** 0.35*** 0.15*** 1 6 R&D‐ intensive industries Share of employees working in knowledge‐ or research‐ intensive industries 10.2 5.5 0.8 28.3 −0.11*** 0.43*** −0.06** 0.25*** 0.23*** 1 7 Public transport Distance to the nearest public transport stop with at least 20 departures per day in 2020 (1 if below median value) 0.5 0 1 0.40*** 0.32*** 0.53*** 0.24*** 0.38*** 0.10*** 1 8 Green votes Share of votes for “the greens” in the 2009, 2013, 2017 federal elections 8.4 3.1 2.8 17.4 0.38*** 0.49*** 0.31*** 0.19*** 0.16*** 0.25*** 0.37*** 1 **p<0.05; ***p<0.01. 15 of 15