Applied Evolutionary Economics: New Empirical Methods and Simulation Techniques

Order this book

Reviewed by Marco Valente
Faculty of Economics, University of L'Aquila, Italy

From a book collecting conference papers the reader is ready to tolerate a certain degree of variety in the different chapters. The more so for papers collected not because of a common subject or methodology, but for the somewhat elusive character of applied evolutionary economics. In fact, however praised for its theoretical soundness and sensible perspective of economic facts, this relatively young approach is widely reputed as being methodologically weak, lacking clear indications on how an applied evolutionary study should be conducted and evaluated (witness the lively methodological debate, that this journal actively promotes). One of the reasons is that evolutionary economics throws its net over a rather wide range of topics: technological innovation, organizational and behavioural phenomena, income distribution, industrial dynamics, etc. Correspondingly, researchers adopt disparate (and sometimes relatively poorly developed) analytical tools, from case studies to simulations, from "appreciative theorizing" to formal models borrowed from advanced mathematical biology or network analysis. Such diversity, as useful as it is to adapt the evolutionary principles to different topics, generate much confusion when one attempts to provide a unified picture of what evolutionary studies can contribute to real world problems, both in absolute terms and compared to other approaches.

The book reflects the expected diversity in its chapters: case studies, simulation models, historical accounts and statistical analysis are variously used to study specific historical events, industries, or general properties of abstract models. These chapters form, overall, a good reading (more on these in the following), making these collection a representative sample of the different threads composing evolutionary economics. Given this premise, the reader would then expect the introductory section to summarize the chapters content with only a short, probably heroic, effort to delineate some shared characteristics among all subjects touched in the book. Instead, the editor's introduction avoids the simple route, and aims directly at the very core of most relevant problem concerning evolutionary economics, devoting a large part of the readers' attention to consider the reasons for evolutionary economics' methodological (apparent) weakness. The introductory chapter is an articulate analysis of current condition of the evolutionary movement, justifying its methodological limitations and even proposing a bold program for future development on this issue. For this reason, this review devotes a disproportionate comment to the book's introduction, and relative smaller space to the chapters' content.

Saviotti's text begins with a brief and convincing argument sustaining that evolutionary economics amounts not much to a novel approach to deal with a given set of phenomena already identified and (more or less convincingly) accounted for by an established, preceding approach. Rather, evolutionary thinking of social phenomena is a novel perspective, generating a brand new discipline put forward to deal with a new set of phenomena that can be conceived only in its new light. That is, the very phenomena object of evolutionary studies simply cannot be defined in other approaches, not to mention be analysed. From this premise, the introduction contains a lengthy and highly provocative discussion meant to individuate the epistemological and methodological problems concerning "young" disciplines in general, and derives sensible conclusions to help evolutionary economics to become a "grown-up" discipline, with a sound and rigorous methodological protocol.

The whole articulated argument is an exercise of recursive thinking: the methodological analysis of evolutionary economics as a discipline is developed along the same line evolutionary economics itself treats the very phenomena it is concerned with. Therefore, the parallel consideration of, on the one hand, methodological issues and, on the other hand, the very results produced by their application, allows the cross-breeding of both arguments, method and applications, within a single, simplified, cost-effective analysis. After all, both scientific methods and socio-economic phenomena concern the behaviour of human beings, and therefore how human beings make sense of their own behaviour must necessarily be a recursive process.

The starting point of the editor's argument is a definition of knowledge, which is of interest both because we deal with how scientists develop knowledge of observed phenomena, and because the phenomena objects of evolutionary economics are considered primarily from the creation and modification of knowledge (technological, productive, users', etc.).

The proposed definition of knowledge consists of two properties:
1) Knowledge is a correlation structure.
2) Knowledge is a retrieval and interpretative structure.

This proposal, convincingly supported, brings to the conclusion that a body of knowledge defines its own variables representing the phenomena of interest and consists of the correlation among these variables, providing the explanatory and normative function a discipline is expected to provide. Applying the usual recursive approach, Saviotti presses on by viewing the emergence and evolution of a new discipline in the same way evolutionary economics analyses the emergence and evolution of novelty in a system, say a new market. Initially, a new discipline identifies a new, unexplored portion of reality, borrowing from neighbouring disciplines the instruments and variables approximating the phenomena of interests. Since "knowledge is local" to specific discipline, this phase is likely to provide relatively uncertain results, appreciated by a relatively small niche of researchers, likely the ones most interested in those facts least explained by existing bodies of knowledge. The reason is that instruments and variables are not specifically designed for the new discipline, and reflect the content of their original disciplines. While time passes by, new tools and variables are proposed, tested and refined, initially likely to be rather imprecise and rough, and therefore generating limited, but increasingly convincing, "correlation" among its own variables. Eventually, we can expect that specific tools and measurements are generated on purpose for the new discipline, and this provides more reliable correlations, and hence greater explanatory power and increasing the status of the discipline. The editor provides an extensive argumentation of how evolutionary economics initiated by borrowing from neoclassical economics variables and from other disciplines (e.g. biology) methods that, though not perfectly suited for its specific characters, suggested that something can be said, after all, about phenomena like industrial dynamics, technological innovation, organization changes, etc. that were poorly accounted for, or simply ignored, by existing economic theory. Concerning future developments, Saviotti invites to insists in exploring all possibilities to identify specific measures and provide detailed analytical methods to obtain more and more reliable assertion concerning the (more and better identified) phenomena objects of evolutionary economic studies.

I find the Saviotti's thesis fundamentally correct, and expressed in such way to provide a robust framework to organize evolutionary inspired works and, therefore, contributing to improve its methodological base. Under this perspective, one can read the book's chapters for their own, explicit content, but also for their implicit contributions to a (badly needed) definition of the methodological open issues concerning evolutionary economics.

The contributions of the book are divided in two parts, with respectively seven chapters devoted to empirical studies and four concerning models. The two initial chapters on empirical studies adopt a network perspective. The first analyses the innovative behaviour in the pharmaceutical industry considering the networks' structures of the industry ("Technological Paradigms and the Evolution of Networks: Lessons from the Pharmaceutical Industry", by F. Pammolli and M. Riccaboni), and the second by studying the market for jet engines ("Increasing Returns and Network Structures in the Evolutionary Dynamics of Industries", by A. Bonaccorsi and P. Giuri). Though not adopting an explicit network approach, also the third chapter of this first part concerns a form of the interaction, focusing on the relevance of publicly funded research in the chemical and pharmaceutical industries ("The Evolution of Specialization: Public Research in the Chemical and Pharmaceutical Industries", by A. Geuna). Following, an in-depth analysis describes the time evolution and current competitive and technological conditions on the market of electric motors ("Evolutionary Patterns of Innovation and Product Life Cycle: Empirical Evidence from the Electric Motors Industy", by E. de Almeida). The sixth chapter reports in detail the peculiar character of the technological research for batteries to be used in electric vehicles, discussing the roles of different actors (researchers, car producers, public organizations, regulators)("Coping Collectively with the Exploration-Exploitation Trade-Off in Research Consortia: The Case of Advanced Batteries for Electric Vehicles", by P. Laurrue). The last two chapters focus on the persistence of innovation, in the petroleum refining industry ("Innovation Direction and Persistence within an Industry: The Refining Process Case", by F. Bel and B. Bourgeois), and in the French industrial system respectively ("An Evolutionary View on Persistence in Innovation: An Empirical Application of Duration Models", by C. Le Bas, A. Cabagnols, and C. Gay). These empirical chapters focus on a variety of phenomena concerning innovative behaviour; all of them represent attempts to identify, with more or less formal tools, general explanations of observed events that may provide insights on the future development.

Besides the specific relevance of these empirical chapters (which, overall, make a good reading), it is interesting to evaluate them from the Saviotti's perspective presented in the introduction. In effect, the studies build their own indicators (or variables, or observations) and manipulate them by various means to test intuitions or just to provide sensible explanations and likely generalizations. In this sense, the editor's core proposal of knowledge can be slightly corrected. The references to variables (being part of the interpretative structure) and correlations may be replaced with softer equivalent concepts, not necessarily quantitative, like "evidence" and "explanation". When the evidence is quantitative and the research question may be answered by means of data reproduction, then we fall into the category of knowledge suggested by Saviotti. But the real world contains much more than numbers, and entities and events not suited to measurement can anyway be formally treated to increase our knowledge. Institutional structures (represented by explicit networks or less formal means), technological constraints, competitive conditions etc. are multi-faceted concepts that cannot be represented exhaustively by quantitative means, and their effect is much more variegated than simply providing correlation among variables. Still, these concepts can be adequately described and managed verbally and logically; insisting on quantifying everything for the sake of respecting a scientific method risks removing the very relevant aspects they deserve attention.

Such opinion is corroborated by the chapters in the second part of the book, devoted to simulation modelling. The first studies the dynamics of income distribution, as resulting from a knowledge-based approach ("Twin Peaks: What the Knowledge-Based Approach Can Say About the Dynamics of the World-Income Distribution", by A. Pyka, J. Kruger and U. Cantner). The second and the third are both concerned with the uneven character of innovation dynamics. The first of these two uses a general model for industrial dynamics to study how routinized behaviour can overcome the difficulties from a rough environment producing differentiated rates of innovation ("Leaping Across the Mountains, Bounding Over the Hills: Punctualism and Gradualism in Economic Development", by W. Kwasnicki). The second chapter is limited to the description of the core elements of a future model, underlining how the "fitness" of an innovation depends on the states of surrounding environment, that is, the set of existing technologies and preferences ("Unlocking a Lock-In: Towards a Model of Technological Succession", by P. Windrum). Finally, the last chapter studies how conflicting short- and long-term selection criteria can be reconciled to allow for (presently) costly choices that need to survive to exploit their long-term advantage ("Selection and the Learning Curve", by P. Geroski and M. Mazzuccato).

These four chapters on modelling are devoted to study the non-obvious connection between assumptions and results. All these works can be considered as attempts to provide explanations of how given assumptions (typically, sensible representations for the characters of real-world actors and environmental conditions) manage to produce aggregate or dynamic properties frequently observed. The diverse methodologies of these works confirm that correlation and quantitative analysis are not the only, and even the central, aspects of a scientific reasoning. It is the explanations, the convincing power of the sequence of logical steps, that embody the knowledge of a discipline.

Concluding, this book aims at a double objective: contributing to the specific issues dealt with in the chapters, and proposing an agenda for the development of an evolutionary economics methodology, which the chapters support but as examples and as tests. On both these respects the book can be considered a success. The individual contributions constitute a good sample of the very frontier of their respective fields. Moreover, considering the methodological perspective, they allow to appreciate how diverse research techniques can be considered as members of a single and coherent methodology.

Return to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, 2006