Innovation Networks for Regional Development

In this chapter we study the diffusion of automotive knowledge created in the Region of Stuttgart. Patents leave a paper trail in the form of citations that we analyze in order to show which regions learn from e-mobility and fuel cell knowledge created in the region of Stuttgart. We show that citations of the patents of both technologies tend to localize in Germany, Japan and the US. However, in case of e-mobility, while aggregating the countries by groups, the largest number of citations is made by European, more precisely Western European countries. The fact, that domestic knowledge flows in fuel cell technology are not the most intensive can be explained by the early stage of the technology and that there is no hard fundament, on which the new knowledge can be built.

This contribution reviews three major perspectives on local transition processes from the economic, social, and political lens. We discuss the major dimensions, concepts, and complementarities of the three approaches—the regional innovation systems’ approach, the sustainable communities perspective, and the local governance concept. Based on the discussion, we suggest three guiding questions that should be addressed in future research on local transition and innovation processes. Using the case of the local energy transition, we demonstrate the applicability of our approach suggested. We find that local innovation is much more bond to social processes of the community and strongly depends on existing, multilevel governance patterns. The social perspective adds substantial insights to the regional innovation systems’ approach taking into consideration a larger variety of actors, institutions, infrastructures, and interactions.

We propose a complexity-theoretic model of how invention is an iterative process of design conceptualization, component decomposition, overcoming technical challenges, and absorbing and recombining knowledge. Using this model, we study the technology development over time and space of two historic aerospace inventions (heavier-than-air aircraft and the jet engine), hereby discussing contributions of individual inventors, knowledge flows of various sorts, government interventions, role of institutes, and the moderating role of geographical distance. We find corroboration for iterative, decentralized search among different design paradigms, with inventors engaged in experiments with (configurations of) component technology. We also find evidence for flows across national borders of an accumulating body of technical knowledge ‘shelved’ in books and articles, embodied in inventions, and by public and private communications. Specific institutions played an important role in absorbing and diffusing knowledge, funding research tools, and establishing credibility to the field. Both invention processes feature substantial dynamic inefficiencies because of overlooked ‘shelved’ technological knowledge, late selection of design paradigms, and a lack of an integrated system perspective. We find that national governments did not support fundamental nor experimental research in the early stage, but invested in concrete projects and coordination at a later stage.

The findings reported in literature on spatio-temporal patterns in knowledge flows in inventive activities are mixed. We discern two basic theories. Firstly, breakthrough inventions require acquiring alien knowledge outside the region. Externalities then stimulate co-location of subsequent incremental innovation activities. Secondly, breakthrough inventions require cross-fertilization of tacit knowledge from different industries, which requires co-location. After this, progressive codification and technological crystallization facilitates diffusion and collaboration over greater distances. We formulate several additional hypotheses on spatio-temporal phenomena in knowledge flows. We then conduct a descriptive statistics exploration of forward citation graphs of breakthrough inventions of an originator in the pharmaceutical industry. We find indications for several distinct spatio-temporal phenomena following a breakthrough. We find progressive globalization in collaboration within groups of inventors and provide several potential causes. In addition, we also find indications for increasing spatial dispersal of groups of inventors collaborating locally on follow-up technology. Moreover, we find increasingly local follow-up, i.e. that the distance between groups of inventors of cited and citing patent becomes smaller. We provide several suggestions to extend this study.

There is an increasing demand for ex-ante impact assessment of policy measures in the field of research. Existing methods to explore the effects of policy interventions in innovation systems often lack transparency or just extrapolate current trends, neglecting real-world complexities. Therefore, we propose a simulation approach and develop an empirical agent-based model (ABM) of knowledge creation in a localized system of researching firms in a science-based industry. With its strong emphasis on empirical calibration, the model represents the Austrian biotechnology industry. In our simulations, effects of different public research policies on the knowledge output—measured by the patent portfolio—are under scrutiny. By this, the study contributes to the development of ABMs in two main aspects: (1) Building on an existing concept of knowledge representation, we advance the model of individual and collective knowledge creation in firms by conceptualizing policy intervention and corresponding output indicators. (2) We go beyond symbolic ABMs of knowledge creation by using patent data as knowledge representations, adopting an elaborate empirical initialisation and calibration strategy using company data. We utilise econometric techniques to generate an industry-specific fitness function that determines the model output. The model allows for analysing the effect of different public research funding schemes on the technology profile of the Austrian biotechnology innovation system. The results demonstrate that an empirically calibrated and transparent model design increases credibility and robustness of the ABM approach in the context of ex-ante impact assessment of public research policy in an industry-specific and national context.

The chapter presents a computational model for the development of a self-sustaining Regional Innovation System (RIS). The computational agent-based model is the core of a virtual laboratory, called CARIS (Complex Adaptive Regional Innovation System) aiming at (1) introducing the CAS (Complex Adaptive System) approach in the analysis of RISs; (2) enabling the development of effective innovation policies able to foster the growth and innovativeness of regions. This topic is particularly relevant for the so-called lagging regions, which, despite conspicuous policy interventions, have been unable to develop a significant capability to innovate. According to the European Union, lagging regions are those regions which show a GDP per capita less than 75 % of the European average. In this chapter, the methodological approach to verify the internal coherence of the model, as well as the simulation outputs are thoroughly discussed. Results show that the code is free of evident bugs, that it works coherently with the meta-model and that the agent-based computational model is able to reproduce some stylized representations characterizing the system under investigation. Finally, the first steps of the calibration activities and some preliminary results are described. Once fully validated, the CARIS laboratory should help researchers and practitioners to better investigate what critical mass of local resources and competencies are necessary to sustain the growth of RISs and, how effective current innovation policies are and what are the most effective measures to improve the current pattern.

Societies are addressing increasingly complex governance challenges that necessitate collaboration between many organizations. Harnessing the emergent abilities of these collective efforts requires new administrative strategies and techniques, but if done well also provides promise for addressing important social challenges. In Maricopa County Arizona the Department of Public Health reports 632 confirmed heat-associated deaths from 2006 to 2013. In response, public health and other organizations coordinate across the County with a collection of public and private organizations and non-profit groups to provide services for heat relief as cooling centers during the summer. Here we show how participatory modeling can be used as a tool to enable this ad-hoc collaborative network to self-organize to provide more efficient service. The voluntary nature of the network imposes a structure on cooling service provision as the locations and open hours of centers are largely based on other ongoing operations. There are consequently both gaps and redundancies in spatial and temporal cooling center availability that exist when the network is examined from a system perspective. Over the last year, we engaged members of the heat relief community in central Arizona in a participatory modeling effort to help improve a simple prototype agent-based model that visualizes relevant components of the regional Heat Relief Network’s function. Through this process, the members developed systemic awareness of both the challenges and opportunities of coordination across the network. This effort helped network members begin to see cooling centers from a systems perspective, leverage their ability to see dynamic cooling center availability spatially and temporally and thus increase opportunities to align services along both dimensions. Our collaboration with the Heat Relief Network in central Arizona highlights participatory modeling as an innovative means for translating evidence to practice and facilitating knowledge dissemination, two important elements for successful applications on complexity governance.

This study aims at identifying the effects of agents’ specialization in research fields on their research performance by means of an agent-based model of the Vienna life sciences, which builds upon the SKIN model. Specialization of agents, e.g. research organizations, firms or universities, is found to play a crucial role in the innovative performance of an industry or a research area. Also in the policy arena, specialization of regions and sectors attained renascent importance through the concept of smart specialization. In order to contribute to the crucial discussion whether specialization or rather diversification is more likely to promote innovative activities, we run simulation scenarios with varying degrees of specialization. Findings provide evidence for both aspects; whereas a higher degree of specialization is found to be favourable for the creation of patent applications and high-tech jobs, diversification is found to be favourable for the creation of scientific publications.

Liberalizing retail energy markets has become a tool for policy makers worldwide to introduce competition into a sector historically characterized by regional monopoly. Opening up the products offered by power retailers to free competition, irrespective of region or distribution network, is expected by policy makers to lead to lower markups and thus lower prices for end customers. We observe that this empirically holds true for industrial customers in Germany, but that markups in the price paid by households have not decreased as a result of increased competition. We apply a methodology of combining simulation modeling with insights obtained from survey data to develop an agent-based simulation of the liberalization of a retail electricity market. In the model, firms adjust prices by adjusting their markups to increase profits. Firms also expand by installing and selling capacity in regions outside of their own. Households are heterogeneous in their preferences and in their geographic position in the simulation. We show that for a wide range of realistic parameter settings, firm markups do not converge to zero in the long run, but flatten out to values possibly even higher than the firms’ initial markups before liberalization. Markups also initially rise before falling and/or stabilizing. This, and the non-linear path of average markups over time, indicate that liberalized markets need not leave end customers better off. Our results imply, however, also that the stability of markups is largely dependent on households’ preferences for their own regional public utility which has implications for new retail business models and investments on the regional level. The results on average markups and household preferences are corroborated by empirical data on the German market.

Over the last couple of decades, firms increasingly acquire locally unavailable inputs in other regions, and are increasingly engaged in research collaboration with firms across the world. In this chapter, we propose and use a spatial agent-based model to study the significance of supraregional relationships on technological progress, in general, and on the emergence of core-periphery structures in particular. We propose a novel ‘artifact-transformation’ model for technology development and have agents (1) construct artifacts using inputs possibly acquired elsewhere and (2) search for transformations to produce these artifacts, possibly in collaboration with other agents. We find that core-periphery structures emerge mostly for certain spatial layouts of regions and if relationships are not completely global while there are many technological cross-links. Moreover, we find that if there are few technological cross-links, supraregional relationships hardly contribute to technological progress and only a weak core-periphery structure emerges at best. We also find that technological progress ultimately levels off in all circumstances.

Over the last decades, research and development of technology has become a collaborative activity of firms creating new by combining existing knowledge. In stimulating technology development, the European Commission pursues ‘smart specialization of regions’ and thus creates a patchwork of regions in which firms collaboratively extend local, technologically specialized knowledge bases. However, creating genuinely path-breaking technological knowledge often requires combining knowledge from different sectors, possibly found in different regions. In this chapter, a fundamental spatial agent-based model is used to study which network structures are conducive to technological knowledge development given a particular geographical distribution and structure of technological knowledge. Unlike the technology discovery models found in literature, which predominantly use a highly simplified technology structure being searched, the model in this chapter is empirically calibrated to the structural features of the OECD patent database. Ultimately, it is concluded that technological knowledge progresses faster and becomes more advanced under regional diversification and does so for a wider variety of network structures. Smart specialization requires a smart or complete network with a high number of ties to attain a similar level of technological knowledge progress.