Bioeconomics and Innovation Networks

1. Bioeconomics and Innovation Networks Andreas Pyka June 2014

2. The Hohenheim Understanding of the Bioeconomy • Definition – Bioeconomy: • “Bioeconomy …covers all manufacturing sectors and associated service areas that develop, produce, process, handle, or utilise any form of biological resources, such as plants, animals, and microorganisms. This spans numerous sectors, such as agriculture, forestry, horticulture, fisheries and aquaculture, plant and animal breeding, the food and beverage industries, as well as the wood, paper, leather, textile, chemicals and pharmaceutical industries, and aspects of the energy sector.” (BMBF: National Research Strategy BioEconomy 2030) • Bioeconomy is associated with a knowledge based, innovative and sustainable way of economy. 1 - SEP der Universität Hohenheim (2013), S. 5f

3. Bioeconomy Expertise of Hohenheim Nutrition Natural Sciences System analysisofbiobasedvaluenets Biobased Products Bioenergy Environment and Climate Agricultural Sciences Sustainable land use Resource management Markets and Policy Economics and Social Sciences Society Services Innovation & Economic Development

4. The Economics of the Bio-Based Economy – A Research Agenda

5. The Challenge: Transforming an Economy towards a Biobased Economy • Economy 1900 • Shop floor based • Labor intensive • carbon based • … • Economy 2013 • Knowledge intensive • Capital intensive • Sustainable? • …. • A historical fact: Economies continuously have developed and transformed their structures.

6. The Challenge: Transforming an Economy towards a Biobased Economy • Textbook economics assumes stable structures in order to apply the famous optimization toolkit. • This is necessary because the uncertainty of open development processes (driven by innovation) is at odds with the idea of well-framed decision problems subject to optimization. • In economics, economic development is conceptualized as economic growth which is measured only quantitatively, i.e. as changing incomes per head. • → Qualitative changes are beyond the analytical framework and are completely ignored!

7. The quantitativ dimension: GDP per Capita in Germany

8. But what about qualitative changes? • Does four times the income per capita in 2013 compared to 1950 mean that we consume for times the goods and services our grandparents have consumed? That we exploit four times of the same energy sources? And buy four times the same typewriters? … • Obviously we consume and produce very different goods and services with our higher income as we have acquired different competences, hold down different professions and are embedded in a completely different institutional environment.

9. The qualitative dimension: Structural changes on the German labor market Changing composition of labor markets Employment in agriculture almost vanishes, employment in industrial sectors peaked in the 1970s; employment in services increased over the last 200 years.

10. Qualitative Changes – from micro to meso to macro • In a historical perspective the transformation towards a biobased economy is nothing exceptional. • However, in order to analyze these qualitative economic developments, one has to apply new tools which focus on change. • Economic development best can be observed on an aggregate macroeconomic level, but it can be neither explained nor understood without analyzing the sectoral transformations (new industries emerge, mature industries disappear, previously successful technologies and competences become obsolete while new competences become to dominate and economically very dynamic regions might stagnate with other geographical regions taking over the lead). • All sectorial changes are triggered on a microeconomic level by innovative and entrepreneurial activities – „creative destruction“ and „destroying the circular flow“ (J.A. Schumpeter, 1912) • And all sectorial changes ask for severe changes in the behavior of consumers.

11. Eric Hobsbawm

12. So what qualitative changes are to expected in a biobased economy? • The transformation will be driven by inventive and innovative activities. How does new knowledge emerge and spread within an economy? • New products and services based on renewable resources will replace CO2-intensive products. This substitution triggers several changes: • Will these products and services be produced by the same actors who have produced predecessor products and services? • At which part of the value chain the high incomes will be earned? • Which regions will be successfully implementing the new industries? • How is international competitiveness affected? • Do we have to expect so-called rebound effects?

13. So what qualitative changes are to expected in a biobased economy? • Can subsidies and taxes influence the transformation towards a bio-based economy? • Where does the new knowledge come from? • What role is play by university research, firm research and cooperations between different actors? • How will labor markets adapt? What role is played by high-skilled migration? … • How will the institutions co-evolve? (legal frameworks, physical infrastructures …) • How is daily life affected by the new technologies? • What role will consumers play? No innovation will be successful without consumers adopting it. • Pressure of demand side on the supply side to innovatively transform? • How can policy influence the transformation? • ….

14. Varieties of Bioeconomies Figure 2 The seven clusters according to the global analysis.

15. Innovation Networks: Modeling and Analyzing Complex Innovation Processes Andreas Pyka

16. From technological spillovers to innovation networks Modeling innovation networks in an agent-based framework Analyzing innovation networks – policy experiments in-silico Structure

17. From technological spillovers to innovation networks • The Public Good Features of Knowledge • The renaissance of economic growth theory in the 1980s was heralded by a positive interpretation of technological spillovers. • With the help of these positive feedback effects (stemming from the public good nature of technological knowledge) the bottleneck of diminishing growth rates has been put away. from old to new growth theory with technological spillovers

18. From technological spillovers to innovation networks • This positive interpretation, however, was at odds with the negative interpretation of technological spillovers prevailing in industrial economics. Industrial economics (from the 1950s to the 1980s): • The particularities of innovation processes were reduced to the standard set of modeling techniques: • Agents optimize. • The focus is on equilibrium. • True uncertainty doesn‘t matter. • Perfect competition as benchmark.

19. From technological spillovers to innovation networks • Innovation processes are treated as investment processes supposed to risk. • The linear view on innovation is dominant – new ideas are exogenous. • „It is as though mother nature has a patent on all techniques of production with unit costs c(x), (x>0) and that society has to pay x to purchase the right to use the technique of production with unit costs c(x).“ (Dasgupta/Stiglitz, 1980, footnote on p. 272) • Knowledge is not considered at all; innovation processes are treated as the accumulation of R&D expenditures. • The conclusion is: Invest in R&D as long as the marginal return of R&D is above R&D costs.

20. From technological spillovers to innovation networks • The incentive reducing nature of technological spillovers: Third parties can benefit from the new knowledge without contributing to its costs. This kind of knowledge transfer happens involuntarily. yi := Output of firm i xi := Inputs in the production of firm i ri := R&D-expenditures of firm i Zi := Spillover-Pool of firm i β := Spillover-Parameter; β between 0 (private good) and 1 (public good) rj := R&D-expenditures of another firm j

21. From technological spillovers to innovation networks • With public good features we have  = 1  free-rider problems • This leads to a negative interpretation of technological spillovers: The private incentives to invest in R&D are smaller than the social optimal incentives. • Market-failure: • - State as knowledge producer (e.g. basic research) • - Implementation of intellectual property rights • - Subsidizing of industry research

22. From technological spillovers to innovation networks • Is new technological know-how a pure public good? Additional features of knowledge: • firm-specific (tacit) • technology-specific (local) • cumulative and complex (absorptive capacities) • From i) and ii) follows that  usually is smaller 1. • From iii) follows , that even if  = 1, it is not clear whether third parties may benefit from spillovers. They have to provide absorptive capacities  to integrate the spillover.

23. From technological spillovers to innovation networks • Besides appropriability conditions also technological opportunities play an important role. • In industries with rich opportunities we find stronger R&D activities compared to industries with low technological opportunities (e.g. compare biotechnology and textiles). • Technological opportunities of different technologies may influence each other. So called cross-fertilization effects (or technological complementarities) create rich new technological opportunities (e.g. bio-informatics, CNC-machine tools, optoelectronics …). • Technological advances in some technologies can have an impact on the overall technological development (e.g. computer technologies, dynamo, mass production …). For this deep effect the term general purpose technologies is used.

24. From technological spillovers to innovation networks Winston Churchill listened to an economic policy session in which his economic advisor went on at length, "On the one hand this, but on the other hand that." After the session he turned to his special assistant and said, "The next time I appoint an economic advisor, remind me to find someone who’s one-handed!" • In the case of technological complementarities so-called idea-creating effects of technological spillovers are much more important than incentive-reducing effects. • (Ti := technological opportunities) in the case of technological complementarities in the case of not related technologies in the case of technological substitutes

25. From technological spillovers to innovation networks • The detailed view on technological knowledge and its features leads to a different interpretation of technological spillover effects: • Incentive-reducing effects change to idea-creating effects. Spillovers are now interpreted positively. • Conclusions for technology policy: • Create technological spillovers • e.g. technology transfer • e.g. collaborative R&D

26. From technological spillovers to innovation networks • Instead of R&D expenditures as an approximation of innovation the knowledge base of companies and firm learning moved into the center of interest: • „Firms cannot produce and use innovations by dipping freely into a general ‚stock‘ or ‚pool‘ of technological knowledge.“ (Dosi, 1988, S. 225) • „In particular, what often appears to be an involuntary flow of knowledge between firms may be nothing more than a pair of draws from a narrow but common pool shared by a group of agents within a common set of problems.“ (Geroski, 1995, S. 85)

27. From technological spillovers to innovation networks • „The dominant mode of innovation is systemic. Systemic innovation is brought about through the fission and fusion of technologies; it triggers a series of chain reactions in a total system. … The interactive process of information creation and learning is crucial for systemic innovation … The characteristic trait of the new industrial society is that of continuous interactive innovation generated by the linkages across the borders of specific sectors and specific scientific disciplines.“ (Imai/Baba, 1991, S. 389) • In modern innovation research innovation networks are considered as an advantageous organizational device for complex innovation processes.

28. 2. Modeling innovation networks in an agent-based framework • The recent literature focuses on knowledge creation and knowledge diffusion. • To model realistically processes in networks one cannot draw on R&D expenditures (e.g. large companies would always have an advantage over smaller companies). • For the analysis of innovation networks therefore the structure as well as the dynamics of firms‘ knowledge bases matters. How can one approach the knowledge base of a company?

29. 2. Modeling innovation networks in an agent-based framework Agent-based Modelling is a computational methodology that allows the analyst to create, analyse, and experiment with artificial worlds populated by agents (computer programs) that interact in non-trivial ways. • Agents are units that have behaviour • They act within a (simulated) environment • Agents can • react to other agents, • pursue goals, • communicate with other agents, • move around within the environment • Macro-level features can emerge from the interaction of agents Schelling model of residential segregation

30. Complex self-organizing processes Part II

31. 2. Modeling innovation networks in an agent-based framework Nigel Gilbert University of Surrey, UK Petra Ahrweiler University College, Dublin, Ireland

33. 2. Modeling innovation networks in an agent-based framework • SKIN (Simulation of Knowledge and Innovation Networks) by Petra Ahrweiler, Nigel Gilbert und Andreas Pyka • The knowledge of a firm is modeled as a Kene: A Kene is a set of knowledge elements of variable size. • Each unit (quadrupel) is composed of a research direction RD, the competences C, the corresponding abilities A as well as the expertise E.

34. 2. Modeling innovation networks in an agent-based framework • The Kene • RD (research direction)  {0, …, 9} 0:= pure basic research; 9 := pure applied research • C (capabilities)  {0, …, 1000} competence in a scientific domain e.g. organic chemistry ~ 3-digit IPC code C07 (organic chemistry) • A (ability)  {0, …,10 } specific ability e.g. preparation of peptides ~ 4-digit IPC code C07K (preparation of peptides) • E (expertise)  {0, …, 40} 0:= not available

35. 2. Modeling innovation networks in an agent-based framework Lexicographic Analysis of the knowledge base of Rhône-Poulenc: • Data: around 6000 Patents from 1985 to 2003 • Frequency and co-frequency of IPC codes (4-digit) • Knowledge network (only technological knowledge), which allows the analysis of the integration and dispersion of new biotechnological knowledge into the firm’s knowledge base.

36. 2. Modeling innovation networks in an agent-based framework 1995 - 1999 • 117 different nodes (IPC subclasses) • 376 connections • 1266 patents Frequency of IPC subclasses “BIOTECH”

37. 2. Modeling innovation networks in an agent-based framework • R&D • Radical Research -> new knowledge fields (C) are tested • Incremental Research -> new ability (A) is tested • Application of knowledge –> expertise (E) is growing

38. 2. Modeling innovation networks in an agent-based framework • Learning in Networks → Combination of Kenes

39. 2. Modeling innovation networks in an agent-based framework • KnowledgeMetrics • The integration of external knowledge as well as the success probability of its application in joint research project depends on the „knowledge distance“. • Very different knowledge (large distance) -> success probability is low; potential impact of the innovation is large. • Very similar knowledge (small distance) -> success probability is high; potential impact of the innovation is low.

40. 3.1 Network architectures and dynamics

41. Small Worlds [Watts und Strogatz 1998]„The importance of understanding small world structures stems from their great efficiency in moving information, innovations, routines, experience and other resources that enable organizational learning, adaptation and competitive advantage.“ [J. Baum et al. 2003] Scale free Networks [Barabási und Albert 1999] „The formalization of a scale-free class of networks is elegantly simple: the probability that a new entrant will choose to link with an incumbent node is proportional to the number of links the chosen node already has. … The well connected nodes, that newcomers attach to, become hubs that create short paths between any two nodes in the network.“ 3.1 Network architectures and dynamics network architectures

42. regular Graph Small World random Graph Smallworlds are characterized by a high effectiveness for knowledge transfer (the probability that all actors are affected by the new knowledge is high). Additionally small worlds show a high speed for knowledge diffusion. Cliquishness := how many of my friends are friends to each other L := shortest path length between two actors network architectures

43. Scale-free networks are characterized by a specific degree distribution. Some actors are very active (so called hubs). Scale-free networks are dynamically advantageous in generating structures with high degree of connectedness which allows for a fast diffusion of knowledge. Preference Attachment: The probability of new connections for an actor is proportional to the number of pre-existing linkages. Scale free networks Random Network Scale-free Network network architectures

44. Small Worlds: [Watts 1999] Empirical innovation networks: 2 Scale-free Networks: Degree-distribution is a Power-Law-distribution: P(k)  k- [Seyed-Allaei, Bianconi, Marsili 2006] network architectures Part II

45. Artificial innovation networks: network architectures

46. 4. Conclusions • Innovation networks cannot be understood without understanding the knowledge dynamics. • Recurring on R&D expenditures without explicitly taking into account knowledge and learning is not sufficient. • Agent-based models allow for wide applications on cases where mutually knowledge exchange and joint learning matters (e.g. Biopharmaceuticals (Triulzi and Pyka 2011), innovation networks in Vienna (Korber, Paier, Fischer 2011), Norwegian defense industry (Castellacci et al. 2011), Irish Nanotech industry (Schrempf 2012), Biotech and ICT (Ahrweiler, Gilbert, Pyka 2004), university industry networks (Ahrweiler, Gilbert, Pyka 2011) ... • Methodological challenges concerning empirical calibration and validation of results.

47. Thank you!

48. 5. Literature • Pyka, Gilbert, Ahrweiler (2001), Innovation Networks - A Simulation Approach, Journal of Artificial Societies and Social Simulation, Vol. 4, Issue 3, 2001 • Pyka, Gilbert, Ahrweiler (2002), Simulating Innovation Networks, in: Pyka, A. und Küppers, G. (Hrsg.), Innovation Networks – Theory and Practice, Edward Elgar, Cheltenham, 2002, 169-196  • Pyka, Gilbert, Ahrweiler (2004), Simulating Knowledge Dynamics in Innovation Networks (SKIN), in: R. Leombruni, und M. Richiardi (Hrsg.), The Agent-Based Computational Approach, World Scientific Press, Singapore, 2004 • Pyka, Gilbert, Ahrweiler (2006), Learning in Innovation Networks: Some Simulation Experiments, in Ahrweiler, P. (Hrsg.), Innovation in Complex Social Systems, Routledge Studies in Global Competition, 2006, 235-249 • Pyka, Gilbert, Ahrweiler (2007), Simulating Knowledge-Generation and – Distribution Processes in Innovation Collaborations and Networks, Cybernetics and Systems, Vol. 38, 667-693, 2007  • Pyka, Gilbert, Ahrweiler (2007), Learning in Innovation Networks – Some Simulation Experiments, Physica A: Statistical Mechanics and its Applications, Vol. 374(1), 100-109, 2007 • Pyka, Gilbert, Ahrweiler (2009), Agent-Based Modelling of Innovation Networks – The Fairytale of Spillover, in: Pyka, A. und Scharnhorst, A. (Hrsg.), Innovation Networks – New Approaches in Modelling and Analyzing, Springer: Complexity, 2009, 101-126 • Pyka, Gilbert, Ahrweiler (2009), The agent-based Nemo-model (SKEIN): Simulating European Framework Programmes, in: Ahrweiler, P. (Hrsg.), Innovation in Complex Social Systems, Routledge Studies in Global Competition, 2010, 300-314 • Pyka, Gilbert, Ahrweiler (2011), Agency and Structure. A social Simulation of knowledge-intensive Industries', in: Computational & Mathematical Organization Theory, Vol. 17 (1), 59-76, 2011  • Pyka, Gilbert, Ahrweiler (2011), A new model for university-industry links in knowledge-based economies, in: Journal of Product Innovation Management, Vol.27 (2), 218-235

49. 3.2. Analyzing innovation networks – Policy experiments in-silico Testing policy options for Horizon 2020 ICT • R&D is all about new knowledge, and networks are central to how it is produced and generated. • Since collaborative research has become the dominant and most promising way to produce high-quality output, these collaboration structures should be the target for policy formation and evaluation. • European Commission Workshop Report “Using Network Analysis to Assess Systemic Impacts of Research”, March 2009

50. 3.2. Analyzing innovation networks – Policy experiments in-silico • How do the new instruments in Horizon 2020 shape the European research landscape (= networks created in previous framework programs)? • Empirically calibrated simulation model of research networks created in FP6 and FP 7 in Information and Communication Technologies (DG InfSoc). • Ex-ante evaluation of policy instruments.