Spatial distribution of bioeconomy R&D funding: opportunities for rural and lagging regions?

ABSTRACT The promotion of R&D carries the risk that economically strong regions will benefit to a greater extent and that regional disparities will increase by focusing the economy on high technology. However, this is expected to depend on the specific sectoral and technological composition of R&D funding. R&D funding for a bio-based economy in Germany is particularly illustrative for a shift in funding focus from a narrow concept of biotechnology to a broader concept of bioeconomy. Along with this shift, an impact on the spatial distribution of R&D funding is assumed. Against the background of inclusive innovation policy, this study examines the potential of the bioeconomy for a reduction of regional disparities in public R&D funding. Based on a database containing publicly-funded R&D projects in Germany and further regional data, comparative regressions are conducted in order to identify spatial patterns. The results demonstrate distinct funding mechanisms in the different areas of the bioeconomy. The broadening of R&D funding for bio-based activities from the biotechnology vision towards bioresources and bioecology leads to a greater participation of rural and lagging regions which is expected to be the result of the inclusion of more traditional industries as recipients of R&D funding.


Introduction
The promotion of R&D is of central importance for overall economic growth (e.g.Romer 1990).At the regional level, however, there is a risk that regions that are already economically strong will benefit to a greater extent and that the extent of regional disparities will increase by focusing the economy on high technology as a result of dynamic advantages from the concentration of R&D (Capello 1999;Audretsch and Feldman 2004).More recently, policies focussing on innovation-based growth have gained importance also in lagging regions (OECD 2011).Classical innovation policy has successively moved away from sectoral and geographical growth pole approaches and, thus, increasingly stressed the value of the spatial dimension as well as the significance of place-based approaches (Koschatzky 2005).Therefore, one-size-fits-all strategies have been renounced as a means to allocate limited resources most efficiently and to take varying regional endowments and capacities into account (Tödtling and Trippl 2005;Capello and Cerisola 2021).In addition, trade-offs between the overall efficiency of the promotion of high-tech industries and spatial equity have been strongly debated for a long time and suggestions have been made how to reconcile equity and efficiency with a suitable innovation policy design (Iammarino, Rodríguez-Pose, and Storper 2017).
The bioeconomy initially had a strong focus on biotechnology (mainly biopharmaceuticals) which clustered in a small number of science-driven bioregions (Cooke 2007), while the bioeconomy concept itself is much broader and comprises activities that offer new potentials for a broader set of regions (e.g.environmental protection, bio-energy, agrofood).R&D support for the bioeconomy in Germany represents a particular example for an innovation policy that shifted from a narrow focus on the biotechnology to supporting more diverse visions of the bioeconomy (National Academies of Sciences, Engineering, and Medicine 2020).According to the official Research Strategy on Bioeconomy in Germany (NFSB) published in 2010, innovation and diffusion should not only be driven by urban centres, but should rather develop collectively in all regions.The bioeconomy strategy itself evolved out of previous biotechnology funding, merged with regional and agricultural innovation approaches, and replaced them over time (Prochaska and Schiller 2021).While past competitions for public funding culminating in the systematic subsidization of biotechnology locations predominantly led to support mainly in leading agglomerations (e.g.BioRegio), the bioeconomy concept is now also applied in rural and lagging regions.
Against this background it is the aim of this paper to shed light on the implications of the shift in bio-based innovation policy from a narrow biotechnology vision to the broader bioeconomy concept for regional disparities in public R&D funding.The main research question is whether rural and lagging regions with a more traditional industrial base and less knowledge-intensive activities can benefit from this shift.
This paper compares bio-based R&D funding by the German federal government in two time periods (1995-2001 and 2009-2015).These two periods represent the early phase of bio-based R&D funding with a narrower focus on the biotechnology vision and a more recent period in which a broader set of bio-based activities has been covered (Prochaska and Schiller 2021).Due to the focus of the bioresource and bioecology vision of the bioeconomy on the (sustainable) use of natural resources, new potentials for rural and lagging regions to participate are expected.In other words, it is analysed whether the shift that has occurred in bio-based innovation policies has led to a more balanced spatial distribution of publicly-funded R&D projects and thus has the potential to counteract regional disparities.
This paper contributes to the literature in two main respects.On the one hand, research in the context of innovation policy that goes beyond economic growth is still quite sparse.The spatial inclusiveness of innovation policy has often been a secondary priority.Awareness of potential approaches to reduce disparities will help to better understand which policy practices are suitable when facing polarization tendencies.On the other hand, the bioeconomy concept is still rather intangible and needs more in-depth analysis to get a grasp of its scope and depth.There has not been much elaboration of which components of the bioeconomy are situated in which geographical areas.
Particularly, the shift from the biotechnology vision to a broader set of activities provides the opportunity to gain valuable insights about the potential for different kinds of regions to benefit from this change.
The remainder of the paper is organized as follows: In section two, spatial implications of innovation policy are discussed as a conceptual background.In section three, idiosyncrasies of bioeconomy as well as their expected potential for regional development are pointed out.Section four describes the data and method used for the empirical analysis and section 5 displays the results based on descriptive analysis and regression models.Finally, a discussion of the results and methods follows before the paper ends with a brief conclusion.

Spatial implications of innovation policy
Politicians frequently pursue an approach of subsidizing high-tech industries as they have the reputation of consequently triggering innovation and economic development (Mazzucato and Semieniuk 2017).Complementary studies have shown that employment in those sectors enables the generation of further jobs and raises wage levels in the respective country or region (e.g.Echeverri-Carroll and Ayala 2009;Moretti and Wilson 2014;Goos, Konings, and Vandeweyer 2018).In consequence, policy tools concentrated mostly on agglomerations, since actors in high-tech industries tend to be located in an urban environment and that is where the overall effect is expected to be at a maximum due to the nature of interactive learning and spatially-bound knowledge spillovers (Capello 1999;Audretsch and Feldman 2004).With regards to biotechnology funding in Germany the BioRegio and BioPofile contests that were introduced in 1997 and 1999 with the intention of establishing biotechnology in seven selected urban regions are good examples for these strategies.
Theoretical deliberations and empirical evidence thus show that in most cases, on the one hand, high-tech industries possess the greatest efficacy in terms of job creation and wage increases and, on the other hand, this works best in agglomerations (Shearmur 2017).However, globally, but also within nations we find an increasing trend towards inequality (Wei 2015).Over time, criticism has been voiced that view the mechanisms of growth and innovation less euphorically (Breau, Kogler, and Bolton 2014;Lee and Rodríguez-Pose 2016;Biggi and Giuliani 2020).The 'dark side' of innovation is increasingly the subject of debate in academic literature as well as in governmental organizations.Next to environmental issues and established negative externalities (pollution, overburdened infrastructure, housing prices etc.) in agglomerations, the topic of diverging living standards within countries is widely discussed due to the increasing inequality of wealth distribution (Atkinson 2015).It is essential to address the structural and spatial distribution of the benefits generated from innovation-based growth in order to do justice to the goal of a balanced and healthy social structure (Turok 2011).Organizations such as the OECD (2014) and the EC (2010) acknowledge the fact that mere economic growth is not sufficient as an objective for policy measures.
Although some strategy papers recognized the value of policies that aim to reduce the disparities within and between regions, many innovation or industrial policy measures still concentrate on high-tech industries in agglomerations.To alter this narrow view, foundational economists call for increased focus on basic industries that are not at the centre of R&D funding, for instance the construction or the agri-food sector (Boeck, Bassens, and Ryckewaert 2019).Some argue the foundational economy, that employs 40% of the whole labour force, receives hardly any R&D funding (Bentham et al. 2013).Thus, they propose directing more R&D subsidies towards foundational industries in order to rebalance this status.With regard to the geographies of inequalities, lagging regions do not just need more financial support, but also an adequate strategy adjusted to their specific context in order to catch up (OECD 2014).A better and more efficient diffusion of innovation is a prominent approach to effectively generate opportunities for job creation (e.g.entrepreneurship) and economic prosperity (McCann and Ortega-Argilés 2013).In light of the unequal spatial distribution of the economy and living standards, as well as the growing discrepancy between high-tech and low-tech salaries, the question of whether a job created in rural areas has the same societal value as one in an agglomeration arises (Capello and Cerisola 2021).
That means a systemic readjustment of the focus of innovation and industrial policy is required if politicians genuinely wish to achieve a reduction of regional economic disparities.In the following, the potential of public investments into the bioeconomy in order to counter polarization tendencies will be addressed.

Bioeconomy potentials to reduce regional disparities
Innovation policy measures in Germany with the intention to foster the biotechnology sector began around 1970 (Warmuth 1991;BMBF 2011;Schüler 2016).In the sequence, the engagement increased and shifted in terms of policy approaches and scope.While there is evidence of a transformation from sectoral funding towards a regionalized approach, the thematic scope shifted from a narrow biotechnology vision towards a broader understanding of the bioeconomy (OECD 2018;Prochaska and Schiller 2021).This shift is in line with the notion that the bioeconomy also consists of a bioresource vision with a focus on the production of biomass and its conversion into bio-based materials, energy, and new products such as bioplastics and a bioecology vision with a focus on ecological sustainability, particularly regarding land-use and processing of biomasses (National Academies of Sciences, Engineering, and Medicine 2020).
This broader bioeconomy concept possesses the potential to involve rural and lagging regions due to its focus on new production mechanisms using biological resources or novel procedures in traditional industries such as food and feed, pulp and paper, or construction industry.Respective bioeconomy strategies also explicitly mention rural regions.While the EU strategy states that it is intended 'to support local bioeconomy development (rural, coastal, urban)' (EC 2018, 18), the German government emphasises 'securing and creating employment and added value, especially in rural areas' (BMEL 2014, 20).
The most apparent fact, initially, is that most of the upstream industries involved in the production of biological raw materials, such as agriculture and forestry, are in rural areas.The same applies to some downstream industries, notably food and feed, the chemical industry, textiles, as well as the production of bio-energy.These sectors are often characterized by their lower technological requirements.Modern policy has recognized that it is not feasible to create strong high-tech sectors in any location whatsoever and hence, underscores the need to transmit knowledge from the core into lagging regions, according to the precept of 'adoption, adaption and diffusion' (McCann and Ortega-Argilés 2015, 1299) of (external) knowledge.Thus, the aspiration is, as Balland et al. (2019, 1) phrase it, not 'to leverage existing strengths, [but instead] to identify hidden opportunities and to generate novel platforms upon which regions can build competitive advantage in high value-added activities'.In other words, instead of specializing in already existing dominant industries, endeavours should rather be made to diversify the prevailing structural conditions.
Current policies pursuing the bioeconomy concept aim at a diversification of incumbent trajectories and path renewal, respectively.McCann andOrtega-Argilés (2015, 1296) clarify 'that very few regions make fundamental structural or sectoral shifts in the short-to medium-term' and thereby illustrate the relevance of regional branching.In the context of bioeconomy this means that in rural and lagging regions existing endowments in low technology sectors possess the potential to enrich the local capabilities with exogenously developed general purpose technologies (GPT), particularly biotechnology.These GPT are based on an analytical knowledge base and are better suited for implementation in geographically distant regions (Asheim, Boschma, and Cooke 2011).The possibility to codify and formalize the biotechnological knowledge provides the opportunity for traditional branches to transfer extant expertise over long distances, to employ them in new ways and through this, renew existing or even create new regional trajectories.Especially for lagging regions, which are often characterized by small and medium-sized enterprises (SMEs) without or with only few own R&D establishments, external and public knowledge are particularly viable (Isaksen 2015).In the context of forest-related strategies, Albert (2007, 65f.) stresses the need for rural areas 'to perfect their 'outside-in' thinking skills, relating information about development in the external world to what is going on internally'.This underlines the beneficial nature of the complementarity between exogenous and endogenous from several other facets of bioeconomy (Bugge, Hansen, and Klitkou 2016).
Natural resources can also have a great significance for future path development (Klitkou, Capasso, and Hansen 2021).As decisive elements within the bioeconomy, rural regions produce most of the biomass to be processed.Government support for R&D has been identified as an important element to foster new path development in rural regions with externally generated knowledge adapted to regional capabilities being particularly relevant for improving low-tech industries (Garud and Karnøe 2003;Isaksen 2015).
In general, innovation opportunities for rural regions have successively improved due to the availability of external knowledge (via modern ICT), growing negative externalities in agglomerations as well as local agency, and internal knowledge about the respective sites (Grillitsch and Nilsson 2017).In other words, regions that are often perceived and labelled as providers of natural resources and as locations for land-intensive industries might have a better chance of moving beyond this stigma of inferiority and becoming stronger economic actors themselves within the bioeconomy.
The remainder of the paper will evaluate empirically whether the changing structure of the underlying innovation policy altered the spatial distribution of R&D funding in the bioeconomy and thereby benefits rural and lagging regions.

Data and method
Due to the lack of a uniform and tangible bioeconomy comprehension, it is necessary to find a coherent definition for the bioeconomy concept that is suitable for the empirical analysis.Therefore, we gathered data and conceptions from various actors involved with the bioeconomy concept and systemized their opinions within a breakdown of the bioeconomy along the value chain.Since the bioeconomy is 'largely driven by policy action and the contents of bioeconomy strategies worldwide' (Viaggi 2016, 105), the political vision has determined our definition to a large degree.Thus, the derived definition has a broad range, like the German strategy paper NFSB.We propose a breakdown into four pillars: input, processing, and output dimension as well as the socio-economic framework (see Table 1). 1  One objective of this study is to trace the implications of the shift from the biotechnology vision towards a broader understanding of the bioeconomy that includes bioresource and bioecological visions as outlined, for example, by the National Academies of Sciences, Engineering, and Medicine (2020).Therefore, it is necessary to distinguish between the initial funding focus, namely the biotechnology, and the additional dimensions of bio-based support policies, which were conceptualized in the bioeconomy strategy.This is why the components of the bioeconomy concept are henceforth categorized and designated as follows (Figure 1): . biotechnology nucleus (biotechnology vision): containing projects related to green, red, and white biotechnology as key enablers for the innovative use of biomasses .bioeconomy shell (bioresource and bioecology vision): containing projects related to input (e.g.production of biomass) and output dimensions (e.g.utilization of biomass) as well as the socio-economic framework The biotechnology nucleus and the bioeconomy shell jointly represent the bioeconomy concept.As outlined above, we expect that the biotechnology nucleus is spatially more concentrated in agglomerations and high-tech regions while the bioeconomy shell has the potential to emerge in rural and lagging regions.
In order to examine the scope and scale of the actual implementation of bio-based R&D projects, we built a distinct database containing all projects funded by the federal ministries, which are recorded in the German funding database called 'Förderkatalog'  The data is structured based on the applied funding measures.This internal BMBF classification is called 'Leistungsplansystematik (LPS) -Benefit plan systematics' and has superordinate topics such as 'A -Health research and health economy', 'D -Food, Agriculture and Consumer Protection' or 'E -Energy research and technologies'.This classification is refined by two further tiers.Eventually, bioeconomy R&D is aggregated in its own category: 'B -Bioeconomy'.However, there are two issues which need to be considered during the analysis.On the one hand, the segment 'B' includes projects that date from far before the official bioeconomy policy concept was formulated.On the other hand, it is apparent that numerous topics or projects within several other classes such as 'EB1920 -Energetic use of biomass', 'GC2060 -Organic electronics' or 'KA1210 -Nanobiotechnology' can clearly be assigned to bioeconomy, but are not covered by this class.For that reason, we considered it necessary to integrate all these projects that actually operate in the scope of the bioeconomy approach, including projects outside of the category 'B -Bioeconomy'.
The database 'Förderkatalog' is openly accessible and offers valid information about the temporal horizon, the monetary investment, the names of the grantees as well as the executing organization along with their respective locations and information about the collaboration partners in the case of joint projects.We diagnosed two types of information about each undertaking's topic which were most relevant for the identification process.In addition to the BMBF's internal classification (LPS) that gives explanations about the subject area, the title of the project provides genuine indications about the project's content.Given these circumstances and based on the BMBF classification, we first categorized the dataset into three divisions on the basis of comprised projects, namely.
[i] classes that were ascertained to belong to the bioeconomy, [ii] classes that only partially belong to the bioeconomy and [iii] further categories that are unlikely to contain bioeconomy projects.
Consequently, text-mining techniques have been applied in order to dissect the project titles for a suitable analysis.Considering the main principle of the bioeconomy, the involvement of biological materials and processes, it is, in our opinion, an appropriate measure to draw on this basic idea and hence to create a collection of biomass-connected terms and expressions.Based on the project titles in category [i], we identified 374 terms that are specific for the bioeconomy and applied them to the project titles in categories [ii] and [iii].If a sufficient number of specific bioeconomy terms appeared in the project titles (two terms in category [ii] projects, four terms in category [iii] projects), the project was classified as bioeconomy.Following a manual check of the project titles, 16,500 bioeconomy projects have been identified in total.For further details on the process of text mining and database creation please refer to Prochaska and Schiller (2021).
In order to categorize the bioeconomy projects into the dimensions mentioned in Table 1, we determined groups within BMBF's internal classification (LPS), which are clearly assignable to one of the previously determined bioeconomy sections along the value chain.A significant proportion, however, had to be attributed by hand, which also served as a result review and occasionally led to the identification of projects which did not fit and were subsequently eliminated from the database.In addition, random samples have been drawn multiple times to check the quality of the identification and categorizing indicating an accuracy of over 90%.A detailed list of all BMBF classifications and their classification for our database is provided in Prochaska and Schiller (2021).
From this unique subset of the BMBF subsidy database, we calculate the project count and project funding of each bioeconomy dimension at any given time on any regional level.For this study, we work on the level of labour market regions used by the BBSR (2017).There were two key reasons for this choice.First, labour market regions have the advantage that pronounced linkages between districts ('Kreise') are taken into account and thus they provide a better view of the economic reality (e.g.commuting flows, urban-rural-relations) than administrative borders based on history.Secondly, spatial autocorrelation becomes a problem with most models on a district level, which eventually leads to incorrect estimates.
Based on these data, we are able to conduct different analyses in order to get insights into the spatial implications of the policy shift from a sectoral biotechnology-centred funding to a holistic bioeconomy concept.
In order to compare the spatial distribution of the funded R&D projects, we conduct four comparative regressions with varying dependent variables, (i) overall project count (whole subsidy database), (ii) number of all bioeconomy projects (core and shell), (iii) biotechnology core projects and, (iv) count of projects within the bioeconomy shell.
For the purpose of tracking the transformation of applied regional innovation policies, analyses are performed taking into account different time intervals, i.e. first we estimate which regional and structural parameters were vital for the acquisition of projects from 1995 to 2001, corresponding to the era of a predominant biotechnology vision in biobased funding as identified by Prochaska and Schiller (2021).Second, based on the same variables, comparative examinations are performed with data from the most recent time interval from 2009 to 2015, which mainly falls within the period after the introduction of the bioeconomy strategy with a broader understanding of bio-based funding.Since the significance of previous biotechnology knowledge and the specialization of the region is an integral component for the evaluation of path dependency, the share of biotechnology projects obtained as a proportion of the total number of projects in the preceding period of seven years is included into the model (BT t-1 ).Further regressors contain the regional data from the last year of each period, namely 2001 and 2015 respectively. 3 Various indicators from a regional database on Germany ('Regionaldatenbank Deutschland') provided by Statistische Ämter des Bundes und der Länder (2019) are employed as explanatory variables to capture characteristics of the observation units.We include the following independent variables at the regional level for the econometric model: .

Number of people employed (EMP)
. Gross domestic product per employee (GDP) .Unemployed people per capita (UNEMP) .Population density (POPDENS) .Employees in knowledge-intensive industries (KNOW) 4   The number of people employed (EMP) serves as a control for the absolute size of the workforce in a region, representing its economic potential to apply and conduct projects.
To capture the factors related to productivity, efficiency, and absorptive capacities, gross domestic product per employee (GDP) is included as an indicator.The proportion of employees in knowledge-intensive industries (KNOW) serves as a proxy for the availability and application of high-tech knowledge and education within the region.We assume that GDP per employee and the proportion of employees in knowledge-intensive industries are positively connected to a greater presence of biotechnology funding as they indicate modern economic sectors in a region and are, therefore, potential attractors of high-tech R&D projects.
The unemployment rate (UNEMP) is used as a control variable to assess the economic situation of a region, reflecting its structural strength or weakness.Productivity (GDP per employee) and UNEMP are also the main indicators used by German regional policy to identify lagging regions.
Rurality is approximated in the model by population density (POPDENS).It is expected that a higher degree of rurality could offer greater potential for R&D activities within the scope of the bioeconomy shell due to the access to biological resources and improvements in existing rural industries, e.g.agri-food production.Additionally, the BMEL (2014, 20) aims at 'securing and creating employment and added value, especially in rural areas' leading to the assumption that bioeconomy (shell) projects are targeted in less densely populated areas.Nevertheless, it is acknowledged that population density is only one aspect of rurality, but the core criterion used by the Federal Institute for Research on Building, Urban Affairs and Spatial Development (BBSR) to identify rural regions in Germany.
The inclusion of these factors is essential as they are expected to significantly influence the political efforts aimed at implementing a bioeconomy.This is attributed to their inherent capability to contribute to the production, refinement, and utilization of biological resources.Consequently, considering these variables helps capture the relevance and potential of regions in shaping the trajectory of bioeconomic development and policy implementation.We decided against dummy variables for rural and lagging regions to better capture the non-dichotomous character of rurality and structural weakness.
Furthermore, we add a dichotomous variable called EAST to control for the bias in funding which favours Eastern German regions due to the intention to accelerate the catching-up process and also to prevent spatial autocorrelation that can be caused by prevailing structural dissimilarities of Eastern and Western German regions.
In all regression models employed in this research paper, the variance inflation factors (VIF) of each model do not provide evidence of the presence of multicollinearity among the independent variables.
As is apparent, any dependent variable whatsoever represents count data.This implies that the error term of the regression will not be normally distributed.For that reason, we employ generalized linear models.Since overdispersion occurs in any model, we neglect the Poisson Regression Model and employ Negative Binomial Models for all estimations (Zeileis, Kleiber, and Jackman 2008).Negative binomial regression provides a flexible and robust method for analysing count data with overdispersion, allowing to account for the variability inherent in the data and obtain reliable estimates of the effects of the independent variables on the count outcome (Hilbe 2011).
The decision to employ cross-sectional analyses instead of panel analysis was based on the limitations of the available data, which did not support a reliable panel structure.For example, the dependent variable exhibits considerable annual fluctuations in many regions due to the inherent nature of the project selection process.Consequently, the volatility of the dependent variable undermines the stability required for panel analysis.Additionally, most regions do not attract or undertake a substantial number of projects in a single year.Even if we were to cluster multiple years into a single panel unit, the resulting dataset would not yield meaningful or robust results.Furthermore, the research question primarily focuses on examining whether the long-term shift in public bioeconomy funding had an impact on the spatial distribution of R&D projects.Hence, a cross-sectional approach was deemed more appropriate and expedient in addressing the research objective.Finally, not all independent variables are available for each region and year.
In order to gain a comprehensive understanding of the data and detect potential concerns related to spatial correlation, we employed the calculation of local indicators of spatial association (LISA) for each regression model, as outlined by Anselin (1995).It became visible that a notable concentration of bioeconomy shell projects has developed in and around Berlin, and an increased involvement of regions in Bavaria over time.At the same time, the spatial patterns within the biotechnology research funding have been quite persistent. 52 indicate underlying differences between projects in the biotechnology core and the bioeconomy shell.While the average grant in the biotechnology core is more than 150,000 EUR higher than for projects in the bioeconomy shell, the gap between the median projects is nevertheless still quite large at 100,000 EUR.Furthermore, the number of joint projects is noticeably higher in the biotechnology core and in either of bioeconomy components it is also higher than for other projects in the database.One basic notion of bioeconomy is the implementation of biotechnological procedures into traditional branches as well as knowledge diffusion in general, which should most likely occur in the context of collaborations between various actors.Hence, a greater number of joint projects might indicate that this approach is actually being applied in the bioeconomy.

Several results in Table
If one looks at the spatial distribution of the projects in Figure 2, it is notable that, for one thing, regions in the north of Germany and, furthermore, the outskirts of some agglomerations (e.g.around Berlin, Hamburg, and Munich) have received more attention since the introduction of the NFSB, especially in the bioeconomy shell.The funding of the bioeconomy shell seems to be more evenly distributed in general than that of the biotechnology core.
Complementary to the visual differences, the Gini coefficient, which measures the inequality of any distribution, shows some distinctions between the bioeconomy components (Table 3).While the distribution of all projects funded by the German government is more even than the projects in the bioeconomy, we find that the biotechnology core, both in the beginning as well as at the end of the observation period, is highly localized.The bioeconomy shell, however, developed differently, and the Gini coefficient decreased over time and depicts a spatially more evenly distributed nature.
These simple comparisons of different datasets reveal some structural dissimilarities, which require econometric analyses to identify and verify the underlying regional implications induced by the policy transition from sectoral to systemic innovation policies.Table 4 summarizes descriptive statistics about the variables used in the models.
The regression models contain all the key data for all 257 labour market regions in Germany.Table 5 shows the estimations for the first period (1995)(1996)(1997)(1998)(1999)(2000)(2001), which comprise the years of the beginnings of the BioRegio contest and following years.The coefficients were standardized due to their differing scales of measurement.As mentioned before, spatial autocorrelation is a severe issue that occurs in most estimated models on the district level.By aggregating the data to the labour market region level, the independence of the observation regions is given in all models except for the models that looked at all funded R&D projects ('Overall Projects') within the region.Nevertheless, we have also included those results in order to achieve a better comparability between the estimates.It is hardly surprising that there is evidence that the number of people employed in a region is positively related to the quantity of projects, irrespective of which kind.This illustrates the typical size effect, i.e. independent of the field of R&D, a workforce is required to execute any undertakings.
Interestingly, unemployment does not play any role in most models, yet projects in the bioeconomy shell seem to be funded more often in regions with a higher unemployment rate.The wealth of a region, measured by the GDP per employed person, is not connected to the number of projects at any point in the period between 1995 and 2001.
When it comes to the relevance of the population density, we find no determining indication of a connection, neither within the bioeconomy concept as a whole, nor in the biotechnology nucleus.Yet, it is striking that there is a significant negative correlation in the bioeconomy shell model.That means, the assumption that less densely populated regions are more often recipients of R&D projects within the broader dimensions of the bioeconomy shell connected to biomass production, bio-based products, bio-based energy as well as ecological and socio-economic aspects is affirmed, even despite the  absence of a dedicated bioeconomy strategy in the first observation period.Together with the positive link between unemployment and the number of projects, this may point to a spatially more inclusive policy by the definition of a bioeconomy that has more relevance for rural and lagging regions than biotechnology R&D funding (even if it was not formalized at this point).
Nevertheless, the impact of the share of people employed in knowledge-intensive sectors is positive and significant in all calculations.This result highlights the pertinence of knowledge as a resource and main driver for economic activities.The expectation that low-tech industries take part in R&D projects in the bioeconomy shell without any preexisting knowledge base more often than in the biotechnology sector is not met in the observations here.Regarding this finding, universities and public research institutes are potentially of particular importance in rural and lagging regions to activate bioeconomy potentials in traditional sectors.vations 1995-2001 2009-2015 1995-2001 2009-2015 1995-2001 2009-2015 1995-2001 2009-2015 Dependent  The share of previous projects in the biotechnology nucleus is crucial for the development of a region that is pursuing a strategy linked to research in the bioeconomy shell.There is a clear positive and significant effect in all models, except for the model with the entire database.In general, in the first observation period the estimations of the bioeconomy as a whole and the mere biotechnology nucleus do not vary greatly.This can be explained by the fact that the share of biotechnology represents up to 68% of the entire bioeconomy in this period.However, the bioeconomy shell displays some minor contrasts to the other models, even at this early stage.
The estimates of the second observation period, which comprises all projects from 2009 to 2015 and therefore represents mainly the time after the implementation of the German bioeconomy strategy, show the actual change of the policy transition (see Table 6).The decisive regional characteristics in the model with all projects as well as the model comprising projects of the biotechnology nucleus have mostly not changed.The variable EAST becomes significant and positive, but the remaining significant regressors KNOW and BT t-1 match with the first observation period with somewhat smaller coefficients.
However, since the share of the bioeconomy shell rose to a level of 50% of the entire bioeconomy (Prochaska and Schiller 2021), we find varying estimates when analysing the entire bioeconomy concept.In line with the results from the bioeconomy shell in the first period, the population density is negatively and significantly related to the number of all bioeconomy projects.The same applies to the bioeconomy shell and confirms the initial finding that bioeconomy projects are more frequently located in rural regions.Unlike the regression analyses from the first period, both projects in the bioeconomy as a whole and in the bioeconomy shell are negatively associated with the GDP per employed person.That is a compelling outcome that endorses the fundamental differences between biotechnology nucleus and the bioeconomy shell and shows which influence the policy change has.Hence, not only more rural regions, but also less affluent lagging areas are more frequently involved in projects of the bioeconomy and bioeconomy shell respectively.This might indicate a shift in the policy design from one that favours a certain sector and follows a 'strengthen the strong' approach towards a regionalized innovation policy that aims at balancing the regional economic structure.Nevertheless, it must be stated that the size of nearly all coefficients declined most likely because of the massive Absence of spatial autocorrelation not declinable Source: own calculations based on Förderkatalog database and additional regional data gain in terms of the project count, namely from 2,727 projects between 1995 and 2001 to 7,755 from 2009 to 2015 (the same applies to the total number of projects in the subsidy database which increased from 22,732 to 54,452).
As a robustness check, we also used the distribution of the funding amount instead of the number of projects.The findings are largely similar, but not as pronounced as in the negative binomial regressions utilizing the count data.

Discussion and critical appraisal
The research question as to whether the shift from biotechnology to bioeconomy might contribute to a reduction of regional disparities in public R&D funding has been tested with multiple approaches and methods.Evidence has been found that rural and lagging regions can benefit from the shift in bio-based innovation policy from a narrow biotechnology vision to the broader bioeconomy concept.The findings demonstrate in which ways the endeavour to establish a bio-based economy could concurrently lead to a more comprehensive spatial distribution of innovative activities.
We found that particularly the increasing project numbers of the broader bioeconomy shell are driving a more even distribution of R&D project allocations, which is reflected by the Gini coefficient.By conducting negative binomial regressions over two periods of time, the different parameters that are decisive for the acquisition of projects reveal structural differences between the biotechnology nucleus and the broader bioeconomy shell which comprise bioresource and bioecology visions of the bio-based economy.In particular, they show that the new dimensions of the bioeconomy shell have been gaining more political attention at the latest since the introduction of the German bioeconomy strategy and they involve sparsely populated and less wealthy regions to a greater extent.In consequence, this underlines the significance of this shift as a starting point for spatially more inclusive development of the whole bioeconomy concept.As studies on foundational economies suggest, more R&D in so-called 'low-tech' branches is necessary in order to support lagging regions.Subsequently, this commitment might induce a catching-up process and ultimately lead to more equal living conditions.The close connection to cross-sectional biotechnology might be useful as an instigator for future schemes.Overall, the existence of a knowledge base is a critical prerequisite for acquiring R&D projects in all kinds of regions.Therefore, developing a knowledge-based infrastructure in rural and lagging regions (e.g.universities, public research institutes or knowledge intermediaries) and linking it with existing economic activities is an important policy recommendation.
Notwithstanding all these results, we find some very initial evidence for an innovation policy that is somewhat less focussed on core regions.For instance, over time the Gini coefficient for all funded R&D projects has decreased.However, this finding would require more in-depth research to substantiate it in the future.It could be that innovation policy in Germany is evolving in a direction, in which more regions can participate in publicly funded R&D projects.This would be in line with the EU Smart Specialization approach that aims to build on existing place-based capabilities of all kinds of regions.
All in all, the findings highlight the potential to contribute to a reduction in regional disparities by distributing R&D funding more comprehensively in sectoral and spatial terms.New high-tech branches are vital for the purposes of competitiveness, but especially new combinations of knowledge from different, but related work fields, are evidently also fruitful for regional development.
Although we find some convincing results that support our deliberations, some shortcomings have to be taken into account.The subsidy database, on which all our findings are based, only provides information on the general research field and the topic of the project through the project title and the internal classification system of the BMBF.Further data, such as a more detailed project description or an abstract, would have offered the opportunity to employ a more detailed procedure for the identification of projects, e.g. more sophisticated text-mining methods complemented with machine learning techniques.Nevertheless, the project titles are in most cases quite specific and serve as a sufficient definition of the undertaking.Furthermore, since the categorization into our system (input, processing, output, socio-economic framework) of each of the 16,500 projects is too time-intensive, we partially rely on the BMBF'S classification system, which differentiates between several main subjects and subclasses, i.e. we translated some of the BMBF's subclasses to our logic where they were unambiguous.Whether, in reality, all projects correspond to our definition and understanding of the distinct bioeconomy components is dependent on the quality of BMBF's classification.Another issue with the database is that all departments must register their funding projects themselves.Ministries that do not enter the projects they are funding are underrepresented in this study.Nevertheless, BMBF is the main provider of funding for R&D in the bioeconomy and their projects are covered most completely compared to other ministries.Overall, although there are some limitations with the data, we still assume they give a representative image of the bioeconomy R&D funding landscape in Germany.
The variables that are most decisive for our estimations are derived from the funding database.That means, we model how much and what kind of funding a region receives.However, we cannot quantify the actual impact of public funding on regional development.Whether more projects or funding for a region trigger innovation in the bioeconomy would be a starting point for another study.Nevertheless, the notion of a spatially more even distribution of publicly-funded R&D projects that benefits rural and lagging regions in particular, is interpreted as an important foundation for regional development opportunities in a broader set of regions.
Moreover, contrasting different effectiveness numbers between bioeconomy and nonbioeconomy regions or urban and rural regions might yield new insights for policymakers.In our analyses, we primarily focus on the distribution of project numbers.Calculations with the funding amount were used as a robustness check for our models.They support the core findings.
It is important to acknowledge that when employing cross-sectional analyses, the potential issue of endogeneity should be taken into consideration.However, it is unlikely that the specific dependent variable utilized in this study exerts a significant influence on the socio-economic factors employed as independent variables.Furthermore, the robustness of the models is supported by the implementation of various checks using different independent variables, such as funding volume.
We also use data from the national level and neglect other public strategies to foster the bioeconomy, e.g. by the EU or at the level of the federal states.In order to compare the region's involvement in the bioeconomy, it would be helpful to integrate all R&D funding into the model.However, to gather and process all the data in a comparable way is, of course, time-consuming and does not contribute to the question of the implications of the strategy of a particular level of the government.It also has to be acknowledged that defining rural and lagging regions by population density and low productivity/high unemployment respectively is only one possible approach.With the availability of more detailed regional data, a further differentiation of these types of regions would be possible.

Conclusion
This article explored the implications of the shift from the narrow biotechnology vision to a broader understanding of the bioeconomy (bioresource and bioecology visions) for the spatial distribution of bio-based R&D projects in Germany.Bioeconomy strategies emphasize the potential of rural regions to play a major role in the implementation of the bioeconomy and the necessity for this to be the case in order to achieve more balanced spatial development.One policy approach to improve the regional development in rural areas is the support of local actors.In other words, a distribution of publiclyfunded R&D projects to sectors that are less concentrated in urban areas can provide an opportunity to revitalize branches that are more often situated in less densely populated areas.The former focus on the biotechnology core favoured a limited number of leading agglomerations and represented a mainly efficiency-oriented policy.The broader concept of the bioeconomy, however, has by its definition a more comprehensive and potentially a spatially more inclusive character that is more relevant for rural and lagging regions.However, the question as to whether and to what extent the emerging bioeconomy scheme is able to contribute to its intended objectives is highly controversial (Birch 2017;Frenken 2017) and requires research on micro and macro levels.In order to estimate the implications of the illustrated policy change, we investigated whether the spatial distribution of publicly funded R&D projects in the bioeconomy has led to a more equitable picture over time.
Therefore, we proposed a disaggregation of the bioeconomy into two main pillars, which represent different parts of a bio-based economy: a biotechnology nucleus and a bioeconomy shell which represents the broader bioresource and bioecology visions.By means of text-mining methods, we were able to detect a consistent database that includes segments that were neglected in preceding studies, e.g. the socio-economic framework.However, the data does not cover the actual bioeconomy in practice and only serves as an indicator for regional R&D activities.
The implementation of exogenous knowledge sources into bio-based upstream and downstream industries promises innovative solutions for diversifying the existing economy either into related or unrelated branches.Thus, this study showed that R&D projects in the bioeconomy shell, namely the input (biomass production) and output dimension (material utilization of biomass) as well as the socio-economic framework, were located in less densely populated (rural) regions and in lagging regions with a higher unemployment rate when compared to biotechnology core activities.In the most recent observation period, those results were affirmed and intensified.It also shows a negative relation between regional productivity (GDP per employee) and R&D projects in the bioeconomy.Moreover, the bioeconomy concept as a whole shows similar estimates because of the policy shift.The results of the paper have provided empirical evidence that the bioeconomy has the potential to counteract some aspects related to the 'dark side' of innovation in terms of spatial disparities (Biggi and Giuliani 2020).
Regarding structural change within the bioeconomy the increasing attention paid to the bioeconomy shell potentially has an impact on regional development at the expense of biotechnology core funding.The opportunity to innovate off the beaten track, i.e. outside core regions as well as without a focus on high-tech sectors, corresponds with recent theoretical considerations in the regional innovation policy literature (Capello and Cerisola 2021;Iammarino, Rodríguez-Pose, and Storper 2017).These very policies stress the importance of diversifying the economic status quo, which aims at viable and long-lasting solutions by fostering linkages to create synergies between actors and sectors to prevent lock-ins and to counter regional disparities by placebased innovation policies.
The role of the cross-cutting biotechnology core, however, should not be underestimated, because it is supposed to be the initiator of innovations in all the industries concerned.Hence, the diffusion and knowledge transfer into both geographically and technologically distant sectors is essential for the utilization of available bio-related innovation potential in any region.Because of its formalized and codified character, biotechnology is particularly suited to meet this requirement.It should also not be forgotten that although the share of biotechnology in the total number of projects has declined, the monetary distribution is still dominated by biotechnology projects.The estimations presented validate the crucial function of biotechnology to attract further public R&D support and emphasize path-and place-dependency.Whether and to what extent collaboration between actors in the biotechnology nucleus and the bioeconomy shell exist and are fruitful, is a starting point for further research.
Beyond that, it is vital and necessary to estimate the impact of the policy interventions undertaken.In this paper, only the input dimension in the form of publicly funded R&D projects was considered.Studies that aim to determine the actual and quantifiable significance of mission-oriented public efforts to create new markets and paths are crucial for the evaluation of policy measures and the rationale to go beyond market-fixing approaches.Therefore, a possible approach for research to be conducted at this level would be to include an output dimension, such as patent data.
Notes 1.A more detailed overview about the structure of the bioeconomy can be found in Prochaska and Schiller 2021.2. According to a statement by the BMBF, the database contains approximately 95 percent of all R&D projects funded by their ministry (with an increasing tendency).However, it is the responsibility of the other departments (e.g.Federal Ministry of Food and Agriculture, Federal Ministry for Economic Affairs and Energy) to record their projects and the data suggests that only a fraction of the ministries' projects have been entered into the database.However, the BMBF is not only in charge of implementing the biotechnology and bioeconomy strategies, but also accounts for approximately 58% of total R&D expenditure in Germany (BMBF 2017b) and therefore, is responsible for the lion's share of all funding.Thus, this database is sufficient in order to make empirical statements about the knowledge-driven bioeconomy funding landscape.3. Due to the lack of data regarding the working population in knowledge-intensive industries, we used the earliest available data from 2009.

(
BMBF 2017a). 2 By April 2017, the dataset comprised 191,347 projects with valid information.Even though the database covers projects dating back to 1968, the analysis is limited to a time frame from 1995 to 2015.In Germany, the change within biothemed R&D funding towards a broader understanding of the bioeconomy officially begun with the launch of the bioeconomy strategy in 2010.We, therefore, compare the period from 1995 to 2001 and from 2009 to 2015 in many analyses to capture this shift.While the historical development of biotechnology funding before 1995 could offer interesting insights, the focus of this study is on more recent trends.

Figure 1 .
Figure 1.Illustration of the bioeconomy components and structure.

Figure 2 .
Figure 2. Development of the funding within the bioeconomy shell (left) and the biotechnology nucleus (right).Source: own calculations based on Förderkatalog database.

4.
This variable is constructed based on the definition of the INKAR database (BBSR), suggesting that the industrial sectors 62-64, 66, 69 & 70-74 in the WZ 2008 classification of the Federal Statistical Office of Germany are characterized as knowledge-intensive industries and services.5. Respective data and calculations are available upon request.

Table 1 .
Structure of the bioeconomy.

Table 3 .
Regional distribution of projects; Gini coefficient.

Table 4 .
Descriptive statistics of the variables.

Table 5 .
Results of the negative binomial regression, 1995-2001labour market regions.Absence of spatial autocorrelation not declinable Source: own calculations based on Förderkatalog database and additional regional data

Table 6 .
Results of the negative binomial regression, 2009-2015labour market regions.