Evolution and co-evolution of regional innovation processes

ABSTRACT

Focusing on ‘evolution and co-evolution of regional innovation processes’, this special issue addresses a highly topical subject. The contributions cover a number of distinct but complementary issues of innovation processes at the regional level and provide rich insights for the improvement of regional policy designs. This editorial outlines the general theme of the special issue and briefly summarizes the main results of the contributions. It ends with some critical considerations and an outlook on future research needs.

KEYWORDS:

• innovation
• regional growth
• co-evolution
• entrepreneurship
• knowledge
• complex systems

JEL:

• B15
• O3
• O18
• R1

CO-EVOLUTION OF REGIONAL INNOVATION PROCESSES

Over the past three decades the emergence of a highly visible and interdisciplinary stream of literature has emerged and significantly improved our knowledge about the characteristic embeddedness of innovation processes in regions. One conclusion from this research is that there is no single representative region. Regions are characterized by a heterogeneous array of factors that profoundly influence the intensity, type and success of their innovation activities. These characteristics also effect the regional generation of income and long-term economic development. Because of the great variety of regional factors that drive innovation, any analysis of how innovative activities impact regional development is extremely complex.

It has been convincingly shown that innovation processes are deeply rooted in a variety of contextual environments: national, industrial, technological and regional. Hence, a comprehensive analysis of innovation activities should not only focus on the behaviours of single actors but also must consider specific contexts that influence these actors. It is, therefore, common practice to analyse innovation processes within the framework of innovation systems that are defined by certain institutions, interactions and the division of innovative labour (Chaminade, Lundvall, & Haneef, 2018; Freeman, 1987; Lundvall, 1988).

Today, the concept of a regional innovation system (RIS) is the most common approach in analysing the role played by regional factors in innovation activities and their effects on regional development. An RIS consists of a large number of actors who engage in complex relationships with each other, as well as with actors and institutions in other regions (Asheim, Isaksen, & Trippl, 2019; Cooke, 2001). The performance of the entire system is determined by how these actors and their respective relationships stimulate the generation, availability, commercialization and application of knowledge.

The common ground of the innovation systems approach is that innovation systems are considered to be highly complex and adaptive entities (Hanusch & Pyka, 2007). They are dynamic in the sense that not only the actors of the system and their connections but also the existing institutions are subject to continuous change. Accordingly, the concept of innovation systems provides a solid basis for analysing and understanding the complex and dynamic nature of innovation processes. The systems approach is well suited for analysing knowledge creation and learning, knowledge diffusion, division of innovative labour, as well as diverse feedback mechanisms that exist between different actors at different levels of the system. It can account for actors’ heterogeneity and the dynamic nature of innovation processes by allowing for the continuous change of actors (e.g., through entries or exits), their attributes (e.g., through learning) and the connections among them (e.g., through new, changing and dissolving relationships). By distinguishing different levels of an innovation system, one can analyse the relationships between macro-level growth, meso-level structural change and micro-level individual behaviour, as well as feedbacks between these levels such as the effect of macro-level conditions on new business formation and innovation activities of firms (Dopfer, Foster, & Potts, 2004; Gong & Hassink, 2019, in this issue).

The interdependencies within innovation systems are often described as co-evolutionary processes, a notion borrowed from evolutionary biology (e.g., Gong & Hassink, 2019, in this issue; Nuismer, 2017). In its most general sense, the concept of co-evolution refers to the idea that two or more dimensions (e.g., industries, networks, technologies, etc.) change simultaneously and affect each other while they evolve. The implications of such co-evolutionary processes for innovation policy are, however, rather unclear and we still face many questions in this research field. A main reason for our incomplete knowledge is that a comprehensive empirical analysis of co-evolutionary processes is a demanding task that requires fine-grained longitudinal data and methodological approaches that are capable of accounting for mutually interdependent processes that shape the behaviour and performance of the actors involved. In fact, even sophisticated econometric techniques may reach their limits here.

In complex systems it may not be possible to improve one system’s performance by simply adding an element from another system that is thought to be responsible for the good performance of that system. Owing to the given interdependencies, it is not unlikely that such a strategy may lead to undesired outcomes. Well-established traditional approaches for evaluating innovation systems, such as indicator score cards and benchmark comparisons, are examples of overlooking the underlying complexity and running the risk of implementing policies that are doomed to failure. Innovation researchers and policy-makers are, therefore, faced with the difficult task of choosing appropriate methodologies to help them understand the complex relationships that characterize innovation systems in order to avoid unintended consequences of their supposedly favourable interventions.

Despite the considerable progress research has made in recent years, we still face more questions than answers when it comes to the evolutionary and co-evolutionary nature of regional innovation processes. Evolutionary economics (Nelson, 2011; Nelson et al., 2018) and evolutionary economic geography (Boschma & Frenken, 2018; Boschma & Lambooy, 1999) have contributed important new insights and considerably broadened our understanding of innovation processes. However, the highly abstract and often vague definitions of evolutionary and co-evolutionary processes has led to the emergence of a wide range of concepts and empirical research designs. McKelvey (1997) and Lewin, Long, and Carroll (1999) emphasized the role of co-evolutionary concepts in gaining an in-depth understanding of organizational change. Others have focused on co-evolutionary processes in the context of technologies and networks (Gilsing, 2005; Koza & Lewin, 1998; Rosenkopf & Tushman, 1998). More recently, researchers have addressed the role of co-evolutionary processes at the intersection of industries, regions and networks (Ter Wal & Boschma, 2011), analysed the regional co-evolution of new industries and public funding (Blankenberg & Buenstorf, 2016), and made efforts in improving the conceptual clarity of the concept (Gong & Hassink, 2019, in this issue).

The stream of literature dealing with RISs (Asheim et al., 2019) and system failures (Dodgson, Hughes, Foster, & Metcalfe, 2011; Schmidt, 2018) shows the same conceptual richness and versatile potential for advancements in our knowledge base. As does literature that focuses on innovation policy design (Borrás & Edquist, 2019; Toedtling & Trippl, 2005), most notably the ‘Smart Specialisation’ concepts outlined by Foray (2014, 2018). However, we still have rather limited knowledge about the nature of the main co-evolutionary interdependencies in regional development and related policy implications. Hence, we face a number of highly relevant questions to which we currently have only rudimentary answers. What role do innovation networks play in the dynamics of an RIS, and what characteristics of these networks are conducive to co-evolutionary processes? How can cross-fertilizations between basic and applied research in an RIS be better organized? If the traditional market failure reasoning is insufficient, can the systems failure approach that goes beyond the idea of market failure contribute a more comprehensive solution? How can innovation research support regional policy-making in a world shaped by complex co-evolutionary dynamics?

OVERVIEW OF THE CONTRIBUTIONS

This special issue deals with recent developments in regional innovation research that addresses the complexity raised by co-evolutionary processes. It is based on a two-day international workshop held in Heilbronn (Germany) in late June 2017. Based on discussions at this workshop, all the papers included in this special issue have been considerably revised and were subject to the rigorous review process of the journal.

The contributions collected here focus on the co-evolutionary processes between innovative performance at the micro-level (i.e., firm or individual organization), the dynamics to be observed at the meso-level (i.e., regional sub-structures such as sectors, clusters, networks, firm agglomerations, industrial districts, etc.), and the development at the macro-level (i.e., the competitiveness and income generation of the entire region). Many of these contributions introduce new methods and modelling tools that allow for an explicit focus on co-evolution and complexity.

The first paper by Roberta Capello and Camilla Lenzi lays the groundwork for the contributions that follow by turning our attention to the causes and long-term implications of the dynamics of regional innovation processes (Capello & Lenzi, 2019, in this issue). Based on a conceptual framework of regional innovation patterns, the authors estimate a regional growth model for the period 2014–16. Results indicate a positive relationship between structural change patterns at the regional level and regional economic performance. Most interestingly, the policy implications drawn from this study advocate the well-established ‘Smart Specialisation Strategy’ that is currently applied in many European regions. Despite its strengths and merits, however, the Smart Specialisation policy framework bears some open questions, particularly when it comes to implementation. One of these open questions pertains to the choice of alternative ways by which regional diversification can be achieved.

Ron Boschma, Pierre-Alexandre Balland and David Rigby contribute to this debate by proposing a framework that explicitly integrates technological relatedness and knowledge complexity at the regional level (Boschma, Balland, & Rigby, 2019, in this issue). The authors break new ground in a variety of ways. They use patent data and employ network-based techniques to come up with novel relatedness and knowledge complexity indicators. The integration and combination of these measures then provides the basis for a Smart Specialisation tool that allows for the evaluation of the costs and benefits of alternative technological development paths in different regions.

In a similar vein, Artur Santoalha adds a missing puzzle piece to the Smart Specialisation debate by emphasizing the role of intra- and interregional cooperation with respect to a region’s technological diversification (Santoalha, 2019, in this issue). The underlying argument is that the Smart Specialisation concept is strongly associated with the idea of diversification in regions. Following this logic, he proposes a regional diversification index for measuring technological diversification and employs this indicator to 226 European Union regions. The results of his econometric analysis suggest that the main explanatory variables (intra- and interregional cooperation) qualitatively affect each other (as well as regional diversification) differently depending on whether the European region is more and less developed. The study raises awareness for the need of well-designed policies targeting research and development (R&D) cooperation.

Uwe Cantner, Eva Dettmann, Alexander Giebler, Jutta Günther and Mariia Kristalova access the long-term regional impact of innovation subsidies by combining panel models and time series characteristics for the period 1980–2014 (Cantner, Dettmann, Giebler, Günther, & Kristalova, 2019, in this issue). The authors compile a comprehensive data set and focus on gross domestic product (GDP) and employment change as dependent variables in their analysis of 75 West German planning regions. Employing dynamic panel estimation techniques, the authors show that innovation subsidies co-evolve with other factors influencing the economic performance of the regions. A main result is that innovation activities and innovation subsidies have a positive impact on long-run economic development of West German regions. The study also reveals that innovation activities in technologically advanced regions may have a negative long-term effect on employment.

Michael Fritsch, Martin Obschonka and Michael Wyrwich analyse how historical levels of self-employment in a region shape the attitudes of today’s population towards entrepreneurship and current rates of new business formation (Fritsch, Obschonka, & Wyrwich, 2019, in this issue). In their analysis for Germany, they find that high levels of self-employment in the early 20th century can explain correspondingly high shares of people with an entrepreneurial personality profile more than 80 years later, despite the diverse disruptive political and social shocks that the country experienced during this period. The authors regard their results as an indication that high levels of historical self-employment may create a regional culture of entrepreneurship that has long-lasting positive effects on new business formation and economic performance.

The distinction between quantitative and qualitative entrepreneurship provides the conceptual basis for the study of László Szerb, Esteban Lafuente, Zoltán Acs and Balázs Páger (Szerb, Lafuente, Acs, & Páger, 2019, in this issue). The authors analyse how these types of entrepreneurship affect regional performance in 121 European Union regions between 2012 and 2014 in terms of gross value added per worker and employment growth. They explicitly consider the characteristics and the moderating role of the entrepreneurial ecosystem in the respective region. The econometric analyses show that the creation of innovative businesses is associated with superior regional performance. However, when measuring only the quantity of new businesses but not accounting for their quality, a significantly negative relationship with regional performance emerges. The effect of the quantity of new businesses is, however, considerably shaped by the quality of the entrepreneurial ecosystem. The effect of high numbers of new businesses on regional performance can be positive if the region’s entrepreneurial ecosystem is well functioning.

The final two studies have a lot in common, particularly with regards to the method applied. Both studies introduce an agent-based model, a method that is designed to identify, analyse and deepen our understanding of complex socioeconomic systems (cf. Gilbert & Troitzsch, 2005). In its most general sense, agent-based modelling allows for the analysis of emerging structural phenomena, such as mutually interdependent and self-enforcing processes. In complex adaptive systems, these phenomena often remain undetected when using conventional analytic tools. This methodology, however, requires a high degree of abstraction and extremely demanding empirical calibrations and validations. From a conceptual point of view, the modelling approaches in both studies draw on the so-called SKIN approach (simulating knowledge dynamics in innovation networks), originally proposed by Gilbert, Pyka, and Ahrweiler (2001).

Andreas Pyka, Muhamed Kudic and Matthias Müller introduce an agent-based model that provides a virtual simulation environment for ex-ante evaluation of policy intervention in RISs (Pyka, Kudic, & Müller, 2019, in this issue). Drawing on the concept of RISs, the authors distinguish three categories of system failures: institutional infrastructure, organizational landscape and structural connectedness. For each of these categories, they specify an exemplary simulation scenario that represents one specific policy intervention. Their findings show that regional learning and knowledge-exchange processes tend to be accompanied by pronounced non-linearities. They demonstrate that systemic interventions designed to stimulate one group of actors also affect other groups of actors, and that policy interventions dealing with complex adaptive systems should not be evaluated in isolation. The insight that different policy interventions may affect each other in complex and often unexpected ways has far-reaching implications for policy-makers.

The concluding paper by Tamás Sebestyén and Attlia Varga also presents an agent-based model and its application (Sebestyén & Varga, 2019, in this issue). The authors focus on interventions that aim at increasing the level of interregional networking. Their analysis demonstrates that the effects of such interventions highly depend on the specificities of the respective regional context.

OPEN QUESTIONS AND AVENUES FOR FURTHER RESEARCH

The contributions to this special issue highlight the importance of considering co-evolutionary dynamics in the analysis of RISs, and foster both our quantitative as well as qualitative understanding of regional development processes. Path dependencies, complex network structures, non-linearities and feedback effects between the different levels of regional economic systems, as well as their institutional, historical and cultural contexts are all relevant factors that must be considered when attempting to analyse and explain the extraordinary variety of regional development paths. Accordingly, it is still unclear how a Smart Specialisation policy can best stimulate industry diversification in different regional contexts such as large agglomerations and less densely populated regions.

The contributions to this special issue clearly indicate the need for future research in at least five areas.

First, our knowledge about how regional conditions impact innovation processes is still rather limited (Boschma, Balland, & Rigby, 2019, in this issue; Capello & Lenzi, 2019, in this issue; Santoalha, 2019, in this issue). The refinements of the Smart Specialisation Strategy proposed by some contributions in this special issue are a first step of a long process towards more appropriate and effective policy frameworks that account for the specificities of a region.

Second, it is rather unclear how regional conditions shape the emergence of different types of new businesses (Szerb et al., 2019, in this issue). In particular, the effect of informal institutions that may vary considerably across regions and how they, directly and indirectly, shape individual behaviour and regional development (Fritsch et al., 2019, in this issue) is largely unknown.

Third, it is important to learn more about the co-evolution of policy measures and regional innovation processes (Cantner, Dettmann, Giebler, Günther, & Kristalova, 2019, in this issue).

Fourth, although the conventional analytical apparatus has been significantly improved and expanded in recent years, econometric methods still have limitations when it comes to the analysis of evolutionary and co-evolutionary innovation processes. In a similar vein, there is significant room for improvements in the quality and completeness of data sets, particularly of longitudinal data sets. Agent-based modelling shows great promise as a complement to conventional empirical research (Pyka et al., 2019, in this issue; Sebestyén & Varga, 2019, in this issue). Hence, the development and use of sophisticated estimation techniques and qualitatively improved longitudinal data, as well as advancements in terms of new modelling approaches, validation and calibration strategies for agent-based models, may lead to important new insights.

Finally, the knowledge we have already gained about the evolutionary and co-evolutionary nature of regional innovation processes needs to be more stringently reflected in policy designs. The methodological approach of agent-based modelling suggests the creation of policy laboratories with the express purpose of facilitating the understanding of complex relationships for policy makers as well as other affected stakeholders. With the focus on processes instead on equilibria, such models can explicitly address the dynamics and structural changes triggered by different policy measures.

The contributions to this special issue provide valuable suggestions and an inspiring starting point for further research that recognizes and takes into account the profound complexity of innovation systems. However, a lot remains to be done from a conceptual and an analytical point of view. Certainly, interdisciplinary research, particularly at the intersection of economics, geography, economics and management, is well suited to generate significant improvements our still incomplete understanding of the evolution and co-evolution of regional innovation processes.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

ORCID

Michael Fritsch http://orcid.org/0000-0003-0337-4182

Muhamed Kudic http://orcid.org/0000-0003-3677-8300

Andreas Pyka http://orcid.org/0000-0001-6207-6690

Additional information

Funding

The editors gratefully acknowledge the friendly support and generous funding by the Dieter Schwarz Foundation.