Localised Technological Change
The extent to which firms can react creatively to rather than adjust passively against ...
8 downloads
712 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Localised Technological Change
The extent to which firms can react creatively to rather than adjust passively against new techniques and practices is dependent on their command of technological knowledge and relative competence. This book explores the characteristics of the path-dependent dynamics of localised technological change, demonstrating how the economics of complexity can inform our understanding of the economics of innovation and vice versa. The book is structured in three parts: Part I focuses on the ingredients of the economics of localised technological change, focusing on the legacies of the key economists and a critical assessment. Part II explores the governance of the generation, dissemination, use and exploitation of localised technological knowledge. Part III elaborates on the basic dynamic mechanisms of localised technological change, combining theory with specific empirical models. The final perspectives articulate the relations between the economics of localised technological change, the economics of path dependence and the challenge of the emerging economics of complexity. This book will be of great interest to students and researchers engaged with the economics of innovation and technological change, industrial organisation and technology management. It will also be of interest to professional economists working in corporations and public government. Cristiano Antonelli holds the chair of Political Economy of the University of Torino, where he is also the Director of the Department of Economics Cognetti de Martiis and of BRICK (Bureau of Research on Innovation, Complexity and Knowledge) at the Collegio Carlo Alberto. He is the managing editor of Economics of Innovation and New Technology. His previous book The Economics of Innovation, New Technologies and Structural Change (2003) is also available from Routledge.
Routledge studies in global competition Edited by John Cantwell Rutgers, The State University of New Jersey and David Mowery University of California, Berkeley, USA 1 Japanese Firms in Europe Edited by Frédérique Sachwald 2 Technological Innovation, Multinational Corporations and New International Competitiveness The case of intermediate countries Edited by José Molero
11 Multinational Firms The global–local dilemma Edited by John H. Dunning and Jean-Louis Mucchielli 12 MIT and the Rise of Entrepreneurial Science Henry Etzkowitz
3 Global Competition and the Labour Market Nigel Driffield
13 Technological Resources and the Logic of Corporate Diversification Brian Silverman
4 The Source of Capital Goods Innovation The role of user firms in Japan and Korea Kong-Rae Lee
14 The Economics of Innovation, New Technologies and Structural Change Cristiano Antonelli
5 Climates of Global Competition Maria Bengtsson 6 Multinational Enterprises and Technological Spillovers Tommaso Perez 7 Governance of International Strategic Alliances Technology and transaction costs Joanne E. Oxley 8 Strategy in Emerging Markets Telecommunications establish ments in Europe Anders Pehrsson 9 Going Multinational The Korean experience of direct investment Edited by Frédérique Sachwald 10 Multinational Firms and Impacts on Employment, Trade and Technology New perspectives for a new century Edited by Robert E. Lipsey and Jean-Louis Mucchielli
15 European Union Direct Investment in China Characteristics, challenges and perspectives Daniel Van Den Bulcke, Haiyan Zhang and Maria do Céu Esteves 16 Biotechnology in Comparative Perspective Edited by Gerhard Fuchs 17 Technological Change and Economic Performance Albert L. Link and Donald S. Siegel 18 Multinational Corporations and European Regional Systems of Innovation John Cantwell and Simona Iammarino 19 Knowledge and Innovation in Regional Industry An entrepreneurial coalition Roel Rutten
20 Local Industrial Clusters Existence, emergence and evolution Thomas Brenner
30 Mergers and Acquisitions in Asia A global perspective Roger Y. W. Tang and Ali M. Metwalli
21 The Emerging Industrial Structure of the Wider Europe Edited by Francis McGowen, Slavo Radosevic and Nick VonTunzelmann
31 Competitiveness of New Industries Institutional framework and learning in information technology in Japan, the US and Germany Edited by Cornelia Storz and Andreas Moerke
22 Entrepreneurship A new perspective Thomas Grebel 23 Evaluating Public Research Institutions The US advanced technology program’s intramural research initiative Albert N. Link and John T. Scott 24 Location and Competition Edited by Steven Brakman and Harry Garretsen 25 Entrepreneurship and Dynamics in the Knowledge Economy Edited by Charlie Karlsson, Börje Johansson and Roger R. Stough 26 Evolution and Design of Institutions Edited by Christian Schubert and Georg von Wangenheim 27 The Changing Economic Geography of Globalization Reinventing space Edited by Giovanna Vertova 28 Economics of the Firm Analysis, evolution and history Edited by Michael Dietrich 29 Innovation, Technology and Hypercompetition Hans Gottinger
32 Entry and Post-Entry Performance of Newborn Firms Marco Vivarelli 33 Changes in Regional Firm Founding Activities A theoretical explanation and empirical evidence Dirk Fornahl 34 Risk Appraisal and Venture Capital in High Technology New Ventures Gavin C. Reid and Julia A. Smith 35 Competing for Knowledge Creating, connecting and growing Robert Huggins and Hiro Izushi 36 Corporate Governance, Finance and the Technological Advantage of Nations Andrew Tylecote and Francesca Visintin 37 Dynamic Capabilities between Firm Organisation and Local Systems of Production Edited by Riccardo Leoncini and Sandro Montresor 38 Localised Technological Change Towards the economics of complexity Cristiano Antonelli
Localised Technological Change Towards the economics of complexity
Cristiano Antonelli
First published 2008 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Avenue, New York, NY 10016 Routledge is an imprint of the Taylor & Francis Group, an informa business
This edition published in the Taylor & Francis e-Library, 2008. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2008 Cristiano Antonelli All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Antonelli, Cristiano. Localised technological change : towards the economics of complexity / Cristiano Antonelli. p. cm. “Simultaneously published in the USA and Canada by Routledge.” Includes bibliographical references and index. 1. Technological innovations–Economic aspects. 2. Technological innovations–Management. 3. Corporations–Growth. 4. Diffusion of innovations. I. Title. II. Title: Localized technological change. HC79.T4A586 2008 338′.064–dc22 2007031816
ISBN 0-203-93200-5 Master e-book ISBN
ISBN10: 0–415–42683–9 (hbk) ISBN10: 0–203–93200–5 (ebk) ISBN13: 978–0–415–42683–1 (hbk) ISBN13: 978–0–203–93200–1 (hbk)
To Anna, the person to whom I owe most
Contents
1
List of figures List of tables
xv xvii
Introduction
1
PART I
The ingredients
15
2
The dynamic legacies: Smith, Marx, Marshall and Schumpeter 1 Introduction 17 2 Adam Smith: the division of labour and the extent of the market 17 3 Alfred Marshall: variety and complementarity 22 4 Karl Marx and induced technological change 29 5 Joseph Schumpeter and the competitive innovation 38 6 Towards a full endogeneity of technological change 42 7 Conclusion 44
17
3
Localised technological change: a critical assessment 1 Introduction 47 2 The process 48 3 The basic ingredients 49 4 The multidimensional scope 57 5 Complex system dynamics and path dependence 60 6 Conclusion 62
47
x Contents PART II
The governance of localised technological knowledge
63
4
Information economics for the economics of localised technological knowledge 65 1 Introduction 65 2 Localised technological knowledge as an interactive and collective activity 69 3 An information economics framework for the economics of localised technological knowledge 75 4 Conclusion 83
5
Models of knowledge and systems of governance 1 Introduction 84 2 Knowledge as a quasi-public good 85 3 Knowledge as a quasi-proprietary good 89 4 Knowledge as a collective process 95 5 The path-dependent complex dynamics of collective knowledge 103 6 Conclusion: knowledge as a path-dependent emergent property 106
6
The new dimensions of knowledge indivisibility: fungibility, cumulability, compositeness and stickiness 108 1 Introduction 108 2 The unfolding of knowledge indivisibility 108 3 Knowledge fungibility, cumulability and compositeness as a source of external increasing returns 111 4 Multiple equilibria, instability and the governance of knowledge commons 116 5 Conclusion 118
7
Knowledge and the theory of the firm: the interdependence among transaction, coordination and production 1 Introduction 120 2 Competing theories of the firm 121 3 The interdependence between production, transaction and coordination 124 4 Implications 130 5 Conclusion 133
84
120
Contents xi 8
9
The localised generation and appropriation of technological knowledge 1 Introduction 135 2 The new economics of knowledge indivisibility and localised appropriability 136 3 The emergence of the direction of technological change 140 4 Implications for empirical analysis and technology policy 146 5 Conclusion 148 To use or to sell technological knowledge 1 Introduction 150 2 Supply-side transaction costs 151 3 A new knowledge trade-off: to use or to sell 154 4 Implications for knowledge exploitation strategies 162 5 Conclusion 164
10 The governance of localised knowledge in the business sector 1 Introduction 167 2 Hybrid forms of knowledge governance 167 3 Types of knowledge and governance mechanisms 171 4 Conclusion 191 11 The new economics of the university: a knowledge governance approach 1 Introduction 193 2 From open science to the academic management of creativity 194 3 The comparative economics of agency costs: the university as an effective institution to solve principal–agent problems in creative work 198 4 The professional university 201 5 Conclusion and policy implications 211 12 Towards non-exclusive property rights: knowledge as an essential facility 1 Introduction 215 2 The knowledge trade-off reconsidered 215 3 An essential facility: economies of density and mandated interconnection – from the economics of telecommunications to the economics of knowledge 220 4 The liability rule as a mechanism for the governance of technological knowledge 223
135
150
167
193
215
xii Contents 5 Implications for knowledge governance 237 6 Conclusion 239 PART III
The introduction of localised technological change
243
13 Localised technological change: the benchmark 1 Introduction 245 2 The basic model with strong irreversibility 247 3 The basic model with weak irreversibility 254 4 Conclusion 259
245
14 The system dynamics of collective knowledge: from gradualism and saltationism to punctuated change 1 Introduction 263 2 Smooth vs discontinuous change 264 3 The complex dynamics of collective knowledge: networking and spillovers 268 4 The dynamics of collective knowledge and localised technological change: anecdotal evidence in Piedmont 277 5 Conclusion 279 15 Factor markets: constraints and inducements to innovation 1 Introduction 282 2 The induced technological change approach 282 3 Composition effects: the interaction between relative prices and the direction of technological change 283 4 General and contingent technological change 285 5 The inducement of general and contingent localised technological change 288 6 Applications and implications 294 7 Conclusion 299 16 Localised product innovation: the role of proximity in the Lancastrian product space 1 Introduction 301 2 Proximity and product innovation 301 3 A Lancastrian model of localised product innovation 304 4 Conclusion 313
263
282
301
Contents xiii 17 Diffusion as a process of creative adoption 316 1 Introduction 316 2 Adoption and diffusion in the business sector 317 3 The role of external knowledge and supply of new technologies 323 4 Creative adoption and the diffusion of innovations 328 5 Conclusion 331 18 Path dependence and the quest for complexity 1 Introduction 333 2 Path-dependent dynamics 337 3 The engine of growth: collective knowledge, localised technological change and path dependence 346 4 Conclusion 353
333
19 Conclusions: hysteresis and creativity
356
Notes References Index
364 379 397
Figures
4.1 The knowledge generation process 6.1 Appropriability conditions and knowledge externalities in the generation of new knowledge 8.1 The direction of the generation of knowledge 8.2 Output- and revenue-maximising incentives to make intensive use of idiosyncratic innovations 9.1 The frontier of possible markets 11.1 The D–D flow chart of the academic system 11.2 The three-party compensation of academic pay 11.3 Alternative institutions for knowledge generation and dissemination 12.1 The effects of innovation 12.2 Knowledge as an essential facility 12.3 The knowledge appropriability trade-off 12.4 The identification of the rents for knowledge 13.1 Change in output with irreversible fixed capital 13.2 Changes in factor price with irreversible fixed capital 13.3 Decision making for the introduction of localised technological changes with strong irreversibility of production factors 13.4 The trade-off between technical change and technological change 13.5 The frontier of possible adjustments 14.1 The cumulative inducement mechanism 14.2 The changing slope and position of the frontier of possible adjustments 14.3 The dynamics of knowledge and pecuniary externalities for research activities 15.1 The nested frontiers of possible adjustments and shift/bias technological change 16.1 Equilibrium in a Lancastrian product space 16.2 Product and process innovation in a Lancastrian product space 16.3 The production of technological change, technical change, product innovation and process innovation with a given amount of resources
72 114 141 146 161 196 204 209 228 231 233 235 249 250 251 255 257 270 274 275 293 306 308
309
xvi Figures 16.4 The nested frontiers of possible adjustments and possible innovations 17.1 The production of technological change, original innovation, passive adoption and technical change with a given amount of resources 17.2 The nested frontiers of possible adjustments and creative adoptions 17.3 The dynamics of gross profitability of adoption, adoption costs and net profitability of adoption and the S-shaped diffusion process 18.1 Typology of dynamic processes 18.2 The evolving interaction between individual path dependence and systemic path dependence
311
324 325
329 338 351
Tables
5.1 5.2 5.3 8.1 10.1 10.2 10.3 18.1 18.2 18.3
Knowledge as a quasi-public good Knowledge as a quasi-private property Localised knowledge as a collective activity Modes of production of new technological knowledge Characteristics of knowledge and the conditions of governance Governance mechanisms for knowledge generation and types of knowledge Governance mechanisms for knowledge exploitation and types of knowledge Path and past dependence Types of path dependence The factors of individual and systemic path dependence
86 90 99 137 170 182 186 340 343 350
1
Introduction
The aim of this book is to demonstrate that innovation is a path-dependent, collective process that takes place in a localised context, if, when and where a sufficient number of creative reactions are made in a coherent, complementary and consistent way. As such, innovation is one of the key emergent properties of an economic system viewed as a dynamic complex system. The book elaborates the foundations and explores the characteristics of the pathdependent complexity of localised technological change. Bounded rationality limits the ability of firms to foresee all the possible events in the marketplace, including new technologies being introduced. The necessary choices about long-term commitments are based upon sequential decision making based upon procedural rationality and satisfying criteria. Quasi-irreversibility of both tangible and intangible production factors limits traditional adjustments to the systematic mismatch between plans and actual product and market conditions by means of price and quantity adjustments in the space of existing techniques. Learning provides opportunities to react creatively. Localised technological change is the endogenous outcome of the induced creative reaction of firms exposed to unexpected changes but able to change their technology. The extent to which firms can react creatively, as opposed to adjust passively within the existing map of techniques, depends upon the amount of technological knowledge and competence they command. In turn, their creative command of technological knowledge depends upon the structure of social interactions that shape the conditions to access and dissemination of knowledge in the economic system. The structural characteristics of the system – not only in terms of the types and forms of competition, but also and mainly with respect to the architectures of knowledge flows – play a key role in assessing the dynamic capabilities of each agent and of the system at large. The direction and the rate of technological change reflect the historical conditions into which firms and the system are embedded, yet at each point in time the interaction between path-dependent agents and path-dependent structural conditions of the system generates unexpected outcomes at both the aggregate and the disaggregate levels. Hence, the determinants, characteristics and effects of technological and organisational change, viewed as the product of interacting and creative agents, affected by irreversibility and yet able to react, are at the origin of the complex
2 Introduction dynamics of the system. From this viewpoint the economics of innovation, and specifically the economics of localised technological change and its implications in terms of path-dependent dynamics, can provide important contributions to the emerging complexity economics, bringing together, in a single framework, elements and aspects that have been explored in separate contexts, so far (Rosser, 1999; Durlauf, 2005). The book is structured in three parts: Part I focuses the ingredients of the economics of localised technological change. Part II explores the governance of the generation, dissemination, use and exploitation of localised technological knowledge. Part III elaborates the basic dynamic mechanisms of localised technological change. The final perspectives articulate the relations between the economics of localised technological change, the economics of path dependence and the challenge of the emerging economics of complexity. The economics of localised technological change builds upon the path-breaking contributions of Anthony Atkinson and Joseph Stiglitz (1969) and the rich analysis of Paul David (1975). The economics of localised technological change has been explored and articulated in previous contributions (Antonelli, 1995, 1999a, 2001, 2003a). It remains a fertile and challenging area of investigation, which makes it possible to manage the transition from the static context of much economics towards the analysis of dynamic efficiency based upon the new economics of complexity. The basic apparatus of microeconomics is retained, and yet the appreciation of the complexity of decision making in a world characterised by irreversibility, localised knowledge, creativity and social interactions leads to understanding of the dynamic variety of outcomes that are possible at both the aggregate and the disaggregate levels, where learning agents are supposed to be embedded in a collective and localised context of action and able to try to change intentionally their technologies. The features of their localised context of action and reaction shape the results of such interactions, including the creative reaction to unexpected events. This book contributes to the economics of localised technological change in four distinctive ways. In Part I it identifies the many sources and ingredients upon which the analytical core is based. In Part II it explores the emerging field of the economics of knowledge governance and integrates the notion of localised technological knowledge into the broader context of the economics of localised technological change. This part examines the systemic characters of the process of generation and dissemination of knowledge and provides a unique analysis of a key attribute of the economic system such as the governance of knowledge. Part III provides a number of expansions and applications of the original framework. Part III builds upon the systematic widening of the choice set into which the innovative behaviour is set. From a methodological viewpoint the novelty consists in two elements: 1) the analysis of the changing context of the innovative choice, at the system level, as the endogenous product of the interplay between the dynamics of knowledge complementarity and knowledge communication costs, among heterogeneous learning agents, able to change their technology, embedded in a given local structure of social interactions that shape knowledge communications networks and collective knowledge commons; 2) the articulation of a twin decision mechanism
Introduction 3 that makes it possible to model, at the same time, the decision to change locally the technology and yet assess its factor intensity, according to the local endowments, and to choose whether to introduce product or process innovations, adopt existing technologies or introduce new ones. The final chapter of Part III provides the economics of localised technological change with a broader and explicit perspective about the long-term dynamic process, articulating the relationships with the economics of path dependence within the context of complexity economics. The conclusions summarise the main results and make explicit the reference to complex dynamics. While unfolding of the foundations of the economics of localised technological change, special attention will be given to stressing the many crossroads and the overlapping with the building blocks of the emerging theory of complexity. The theory of complexity and the economics of localised technological change share many basic ingredients and the dynamic mechanisms at work. The theory of complexity builds upon six basic elements: 1) individual and hence heterogeneous agents with specific characteristics are elements of a system and their action is characterised by systemic interdependence; 2) the distribution of agents in a multidimensional space is essential to understanding the dynamic behaviour of the system; 3) each agent has access only to local information and local knowledge, i.e. no agent knows what every other agent knows; 4) agents are embedded within multidimensional topological spaces shaped by networks of relations, ranging across a variety of interactions, transactions and communication channels that affect locally their behaviour; 5) agents are creative, i.e. agents can change the rules of their behaviour; 6) the architecture of the system into which firms are embedded, as well as its performance, changes as a consequence of the action of agents (Anderson et al., 1988; Arthur et al., 1997; Arthur, 1999; Rosser, 1999; Taylor, 2001; Barabasi, 2002; Wolfram, 2002).1 Complex systems are characterised by non-ergodicity, social interactions, phase transition and emergent properties. When non-ergodicity applies a little shock at one point in time it affects the long-run dynamics of a system. Phase transitions consist in qualitative changes that can be determined by small changes in the parameters of the system. Emergent properties are properties of a system that apply at a specific level of aggregation of the system. In the theory of complexity, feedback and interactions play a key role in assessing the conduct of agents and specifically the chances of changing their behaviour (Durlauf, 2005). The merging of the theory of complexity and economics contributes to the building of an economic theory of complexity based upon non-ergodicity, social interactions, phase transition and emerging properties. The notion of path dependence as the specific form of complex dynamics applied to understanding economic systems as evolving systems makes it possible to integrate into a single and coherent framework a number of relevant and complementary contributions (David, 1988). The implementation of the overlapping of the economics of localised technological change and the theory of complexity makes it possible to elaborate an economic theory of path-dependent complexity. In so doing much traditional
4 Introduction microeconomics can be used and recombined so as to build a generalised localised economic approach to understanding the laws of the conditions and the processes that generate endogenous change and transformation into economic systems that are necessarily characterised by the action of agents that are constrained by bounded rationality and hence procedural rationality that consists in sequential decision making based upon satisfying criteria. In this context the effects of irreversibility are not negligible. Firms however are able to learn and to react creatively; hence, within a proper system of social interactions, they can change their knowledge, their technology and ultimately their actual location in the structure of the system by means of limited mobility in the relevant spaces.2 Path dependence is the specific form of complex system dynamics most apt to understand the process and the outcomes of the interactions among myopic agents embedded in their own context and constrained by their past decisions and yet endowed with creativity and able to generate new knowledge and new technologies by means of both learning and intentional innovative strategies and by means of structural changes. In our approach, in fact, firms are always on the move as they are supposed to be able to elaborate strategies finalised to change their current technology, as well as the context of their action. Such changes are continual because agents are myopic and cannot foresee all the changes introduced by each other agent in the system. Such changes concern both the technology and the architecture of the system: 1) They may lead to the introduction of actual total factor productivity improving technological innovations only when appropriate interaction contexts, conducive to the generation of new knowledge, are available. Hence actual technological innovations, i.e. efficiency-augmenting technical changes, can be considered the result of the emergent positive properties of the system defined by the characteristics of the interactions of the components. 2) They may change the architecture of the system. Firms are able to change their technologies as well as their context of action: firms can change their boundaries by means of strategies of inclusion and exclusion of activities; firms can diversify or integrate downstream or upstream; firms can specialise in core business. Firms can change their location in regional space by means of actual spatial mobility. Firms can change their connections with customers, suppliers, rivals and research institutions. As a consequence, the architecture or topology of the system is exposed to structural change: industrial structures change their shape; new industries emerge and others decline; the division of labour is altered both within and among economic systems. The structure of relative prices of inputs is also affected by new technologies according to their bias and the derived demand.3 In an economic approach to understanding the behaviour of agents and the results of their interactions, the topology of the space into which agents are embedded is at the same time relevant and endogenous. In an economic approach to complexity, agents are credited with three relevant characteristics: 1) creative reaction: agents are able to change their knowledge and hence their technology in response to unexpected events; 2) creative reaction is localised and context dependent: the structure of non-market interactions, connections and relations in place at each point in time affects the extent to which the creative reaction takes place
Introduction 5 successfully; 3) limited mobility: agents are able to enter and to exit, to create communication channels and new relations, and to change their location in the space of technologies, in the space of products and factors, and in the space of technological knowledge, albeit in a small context, limited by switching costs (Gleaser and Scheinkman, 2000). In the economics of complexity, agents are supposed to be able to learn and to rely upon procedural rationality and hence intentional action albeit based upon sequential decision making (Simon, 1982). The intentional action of a myriad of heterogeneous and myopic agents cum systemic interdependence generates structural changes in both product and factor markets and in the knowledge space. The interaction between individual conduct, constrained by irreversibility and the procedural rationality of a myriad of heterogeneous agents, and structural change is continual. In such a context the sequential decision making based upon procedural rationality is the single possibility of intentional action. The understanding of the recursive effects of the intentional creative reaction of agents helps in grafting the tradition of complex adaptive dynamics into economics which, traditionally, builds upon agents credited to limit their conduct to passive adaptation. While creative reaction is intentional and can affect both the location of each agent in the relevant space, by means of mobility and its performances, and the generation of new knowledge and the introduction of new technologies, the actual results are very much influenced by the systems of social interactions in place at each point in time. So far intentional creative reaction at the agent level and stochastic outcomes both at the micro- and the macro-level do coexist. Productivity-enhancing technological change is a special case of complex dynamics; its emergence takes place when a number of specific conditions apply. Dynamics is assumed to be intrinsic to the system, as soon as agents are supposed to be at the same time myopic – and as such induced to react to unexpected changes – and yet creative, that is able to change their laws of conduct and specifically their own technologies: its outcome however depends upon a number of key conditions. The successful accumulation of new technological knowledge, the eventual introduction of new and more productive technologies and their fast diffusion are likely to take place in a self-propelling and spiralling process and at a faster pace within economic systems characterised by fast rates of growth where interaction, feedbacks and communication are swifter. In such special circumstances, the system can undergo a phase transition leading to the introduction of a new radical technological system (Casti, 1998). Such dynamics can be represented today by the merging of economics and complex system dynamics, but it was already well articulated in many writings of the classical school. Part I recalls the basic ingredients of this approach and retrieves its classical foundations. The analysis of the contributions of Adam Smith and Karl Marx in understanding the basic inducement mechanisms and the key role of learning as the prime source of technological knowledge open Part I. The classical legacy consists in fact in the appreciation of the underlying process of learning in its different facets and in the grasping of the key role of the inducement provided either by the dynamics of the demand, as is the case in Adam Smith, or by the
6 Introduction changes in the relative prices of production factors, as is the case in the writings of Karl Marx. Learning per se is not a sufficient condition for technological change to take place. Alfred Marshall highlights the key role of the variety of firms and the complementarity among learning agents in the generation of technological knowledge viewed as a collective process that is favoured by proximity within industrial districts. Joseph Schumpeter has highlighted the analysis of the role of competitive mechanisms based upon the continual introduction of innovations, in products, processes, organisation methods and intermediary inputs, as tools of the market rivalry between firms in the marketplace. The appreciation of the classical legacy pushes towards the integration of the three mechanisms into a single dynamic model where the many fragile conditions for the actual introduction of successful innovations are well articulated. In the economics of localised technological change, the introduction of technological innovations is the emergent outcome of a complex mechanism. The introduction of new technologies is induced by unexpected changes and can eventually lead to the actual growth of productivity. Changes in product and factor markets, to which firms cannot adjust by means of changes in a given technical space, because of limited information, localised knowledge and irreversibility of tangible and intangible production factors, act as pushing factors. Firms can counteract the decline in their performance and the increase in actual costs by changing their technologies, with the introduction of process and product innovations, provided a conducive context is available. The characters of the system from the viewpoint of the governance of the generation and dissemination of knowledge are key to assessing the scope of creative reactions. Here the key role of path dependence both for each agent and for the system at large becomes evident. The introduction of new technologies is constrained and shaped by the role of historical factors such as quasi-irreversibility, learning, and the structure of interactions at work. New knowledge and new technologies however do change both the specific conditions of each firm and the structure of the system. At each point in time new directions and new challenges appear and characterise the path of change. In this approach, the generation of technological knowledge is viewed as a collective process made possible by the continual efforts of accumulation of competence and technological knowledge based upon the localised learning processes and the access to external knowledge by means of knowledge communication processes of a myriad of creative agents rooted in a well-defined set of scientific, technical, geographic, economic and commercial circumstances. This systemic dynamics paves the way to the eventual introduction of innovations by agents constrained by substantial irreversibility of both tangible and intangible production factors, including reputation and communication channels with customers and suppliers, and yet able to react with creativity to the emerging mismatch between expected and actual market conditions. Part II focuses on the analysis of the generation and the governance of knowledge. The distinction introduced by Joseph Stiglitz between information economics and economics of knowledge provides basic guidance (Stiglitz, 2000, 2002).4 The tools
Introduction 7 elaborated by information economics, namely the analysis of the characteristics of the economic system from the viewpoint of the quantity, quality, symmetry among agents, distribution, access and transparency of information, and their effects on the conduct of agents are presented in Chapter 4. The basic argument elaborated here is that information economics provides relevant methodological tools to build an original framework useful to achieve a better understanding of the economics of knowledge and implement the notion of localised technological knowledge. The rest of Part II is dedicated to applying and implementing the basic guidelines presented in this chapter and to showing their analytical fertility. Chapter 5 shows the evolution of the economic understanding of knowledge as an economic good. The path-breaking analysis of Kenneth Arrow and Richard Nelson has long shaped the economics of knowledge. Knowledge has been consequently regarded as a public good because of the key features of indivisibility, non-appropriability, non-tradability and non-exhaustivity. In a second step, however, the actual conditions of appropriability and indivisibility have been questioned, with important consequences. The role of alternative governance mechanisms has been praised. In a first stage the markets for knowledge have been considered as reliable mechanisms to organise the generation and dissemination of technological knowledge. Later on, the role of externalities and spillovers has received a deeper attention in the context of a systemic understanding of the processes by means of which technological change is generated, exchanged and disseminated. In this context, the notion of localised technological knowledge can be implemented so as to better appreciate the complementarity between external knowledge and internal research and learning and to grasp the character of localised knowledge as a collective process. Chapter 6 elaborates the notion of knowledge indivisibility and its implications for grasping the relevance of path dependence in the economics of knowledge. Cumulability, compositeness, fungibility and stickiness are in fact specific aspects of knowledge indivisibility. Cumulability concerns the relations among current knowledge and the previous vintages of the same knowledge. Knowledge is composite when its generation of new knowledge requires the combination of diverse and yet complementary bits of knowledge. Fungibility is found when some units of knowledge can apply in a variety of different contexts, to different knowledge items, different products and different processes. Knowledge stickiness matters when the separability of knowledge with respect to the human capital and the organisation into which knowledge has been generated is difficult and limited. Knowledge cumulability, compositeness and fungibility are the cause of increasing returns and path dependence. The analysis of the implications of the new understanding of knowledge divisibility, knowledge path dependence and knowledge appropriability requires a dedicated analytical framework. Chapters 7, 8 and 9 are dedicated to elaborating a theory of the firm which combines the intuitions of Edith Penrose with the basic notions of information economics to build the economics of knowledge governance. An economics of knowledge governance is necessary to elaborate a comprehensive approach able to understand the complementarity and systemic role of the variety of knowledge
8 Introduction governance mechanisms that can be identified. The economics of governance provides the basic reference. The governance approach elaborated by Oliver Williamson can be further elaborated and applied to the analysis of knowledge exploration and exploitation. The characteristics of knowledge and the details of its generation and dissemination process can be appreciated from the viewpoint of the economics of governance especially when the basic ingredients of the resourcebased theory of the firm are taken into account and properly integrated into a single interpretative frame. Chapter 7 presents an integrated framework for a broader institutional theory of the firm. In so doing the research programmes of the resource-based theory of the firm and of the economics of transaction costs are made complementary. Transaction costs economics pays little attention to organisational knowledge. The resource-based theory is not able to appreciate the role of organisational constraints in shaping the rate and the direction of the growth of the firm. The economics of knowledge governance provides a framework able to integrate the research programmes of the resource-based theory of the firm and of the economics of transaction costs and overcome their own limits. In the context provided by the economics of knowledge governance, the overlapping between the theory of production, the theory of the markets, the economics of innovation and knowledge and transaction costs economics can be better appreciated. The interdependence between transaction, coordination, production and generation of knowledge is modelled as a micro-system and as such treated with the apparatus of both the corporate function that includes a production function and the transformation curve, traditionally applied to handle the working of economic systems at large. Chapter 8 explores the embedded conditions for the generation of knowledge. In this chapter firms are viewed as learning agents that, in order to generate new knowledge, elaborate and implement intentionally strategies of knowledge generation. These strategies include the identification of the sources of external technological knowledge. Following Edith Penrose it is true that the growth of the firm depends, in fact, for given values of the other factors affecting transaction, coordination and production costs, upon the capability to generate and use new knowledge. Internal learning however is a necessary, but not sufficient, condition for the generation of new knowledge. The understanding of the external conditions for the intentional generation, exploration and exploitation of technological and organisational knowledge becomes crucial. Contextual spillovers and localised appropriability have a strong effect on the directional generation of new technological knowledge. Chapter 9 analyses the strategies of technological exploitation as a part of the valorisation of the technological knowledge generated. This chapter presents a closer analysis of knowledge appropriability from the viewpoint of the costs of using the markets either for disembodied knowledge or for different layers of products embodying the new knowledge as appropriate mechanisms to exploit technological knowledge. The debate on the notion of appropriability has shown that knowledge can be often appropriated more than it is assumed in the public good tradition of analysis. Imitation is costly: the not-invented-here syndrome is
Introduction 9 much more effective than has been recognised. As a matter of fact, knowledge can be appropriated partly by the inventor, to varying extents, which depend upon a number of variables. Vertical integration of inventors in downstream manufacturing activities to embody the proprietary knowledge can be seen as a radical remedy to inappropriability. In this context, the extension of analysis of the trade-off between to make and to buy, an essential tool of the theory of the firm, to the choice between to make and to sell is fruitful. This chapter shows that knowledge transaction costs matter on both the demand and the supply side. Knowledge transaction costs on the supply side play a major role in making operational the notion of imperfect appropriability. This approach applies to the economics of knowledge. Technological knowledge, in fact, can be conceived as an intermediary product, which can be either sold in the markets for disembodied knowledge or used as an internal input for other products. Chapter 10 explores the range of choices in terms of governance mechanisms for the generation, exploration and exploitation of technological knowledge at the firm and the aggregate level. The boundaries of the corporation, as a learning agent, are analysed and understood with respect to the characteristics of the processes of knowledge generation and usage. Different viable knowledge governance mechanisms and governance choices emerge according to the characteristics of technological knowledge and to the related levels of knowledge transaction and agency costs. The integration of the analysis of agency and transaction costs in the resourcebased theory of the firm shows that firms select their strategies of inclusion and exclusion, and specifically knowledge exploitation and exploration, with respect to the joint assessment of coordination, transaction and production costs for a given product and a given item of technological knowledge. In all circumstances external knowledge remains an essential and non-disposable input, vital for ensuring appropriate levels of creativity in an economic system. In this context, hybrid forms of governance such as coordinated transactions, quasi-hierarchies and constructed interactions emerge as the key mechanisms of knowledge governance. Knowledge transactions in quasi-markets are implemented and enforced by a myriad of coordination mechanisms that try to reduce the inefficiencies of pure market transactions. Knowledge interactions, as distinct from transactions, play a key role in this context, and yet they are not spontaneous, but the product of intentional action directed to make knowledge communication possible. Knowledge coordination within hierarchical organisations is implemented by means of quasi-market mechanisms, often based upon non-exclusivity in employment contracts. Chapter 11 and Chapter 12 explore the basic tools of the public governance, as distinct from the business governance, of technological knowledge. Specifically, Chapter 11 develops the analysis of quasi-hierarchies with an application to the academic system as a viable institutional mechanism that makes it possible to address the knowledge trade-off and to reduce agency costs in the management of creative talents. Chapter 12 presents the implications of the economics of localised technological change about the role of technological knowledge as an essential facility on the intellectual property right regime, with the shift towards reduced
10 Introduction levels of exclusivity of patents associated with the introduction of the liability rule for imitators. Part III addresses the analysis of the path-dependent dynamics of localised technological change and the working of a number of processes: the role of collective knowledge in shaping the decision making of the introduction of new technologies, the inducement dynamics exerted by factor markets, product competition, and creative adoption. This part presents the results of the work done to extend the basic model of localised technological change to a broader array of interrelated issues. The basic argument is developed as a trade-off between the costs of adjusting passively to a new and unexpected economic environment and the costs of changing the context in a creative way. Such creativity takes different forms: the introduction of product or process innovations, or the adoption of innovations together with the introduction of new technologies. Two analytical devices are systematically applied. The first is a nested transformation curve, which compares switching activities with product or process innovations and either innovation or adoption. The second is the analysis of the dynamic interplay between positive and negative externalities and its effects on the temporal distribution of events. The analysis of the interplay between positive and negative externalities makes it possible to model the effects of entry and exit dynamics on the general process of accumulation of new knowledge and to grasp the dynamics of localised increasing returns. The economics of localised technological change provides an original framework to model the path-dependent dynamics of the introduction of new technologies as the result of the interaction between the inducement to change the technology generated by the mismatch between plans and expectations. Because of irreversibilities, limited knowledge and local learning, the introduction of new technologies is induced by the disequilibrium conditions brought about in each system by all changes in relative factor prices. Such a framework can be articulated and expanded so as to include the analysis of the dynamics of collective knowledge on the one hand and the variety of forms which localised technological change can take, including its direction and whether it consists of (more of) product or process innovation, adoption or original innovation. The contributions of the economics of knowledge and the related emphasis about the role of external knowledge and knowledge communication and more generally about the characteristics of the network system into which creative reaction takes place, elaborated in Part II, push to articulate a closer relationship between the analysis of innovation and the analysis of diffusion. In the traditional divide the economics of diffusion focuses the process of adoption of a given innovation, while the economics of innovation addresses the analysis of the generation of a new technology. When the role of external knowledge in the generation of new knowledge is appreciated together with the pervasive role of knowledge cumulability, complementarity and fungibility, each innovation can be regarded as the result of a creative adoption and usage of existing bits of knowledge and in turn each adoption can be considered as the result of a creative action of selection and assessment. The notion of creative adoption and the analysis of the interplay
Introduction 11 between the costs of knowledge communications and the benefits stemming from the access to external knowledge make it possible to extend the application of the diffusion process to understanding, at the system level, the path-dependent process of generation and dissemination of new knowledge. More specifically, Chapter 13 provides the basic model of localised technological change in its traditional and stylised structure. This chapter presents the benchmark upon which a number of inclusions and applications are conducted in the following chapters. Chapter 14 shows how collective knowledge emerges when knowledge widening leads to knowledge deepening. The interplay between networking costs and knowledge super-modularity can accommodate both saltationist and gradual change. The divide between Schumpeterian theories of saltationism and Marshallian gradual cumulative growth can be reconciled when the essence of the Schumpeterian and Marshallian approach is properly combined. Smooth, Marshallian dynamics can easily generate major Schumpeterian discontinuities. Small variations in the parameters can generate either gradual or discontinuous changes, provided their cumulative dynamics is appreciated. Punctuated technological change is likely to take place when, within communication networks, the interplay between positive and negative knowledge externalities leads to the creation of commons of collective knowledge and hence new technological systems. The correct appreciation of the structure of knowledge networks and hence of the interactions between individual action and the characteristics of the environment makes room for complex dynamics and provides an integrated framework able to explain in a single context both Marshallian gradualism and Schumpeterian saltationism. The notion of local externalities emerges as the main contribution of this chapter: local externalities define the scope and extent, hence the boundaries, of the local system, within which net positive knowledge externalities are at work. The rest of Part III is dedicated to applying and implementing the basic guidelines presented in this chapter and to showing their analytical fertility. In Chapter 15 it is argued that the direction of technological change in terms of its specific form of bias and how it is introduced and adopted reflects the specific conditions of local factor markets. Well-defined long-term technological paths emerge in each region and they depend on the selection process in product markets. The more rigid and idiosyncratic the endowment of production factors and the system of relative prices are, the more specific the technological path of each region is likely to be. The divide between the microeconomic and the macroeconomic models of induced technological change is reconciled. Chapter 16 explores the role of proximity in the Lancastrian product space. Product proximity matters when relevant knowledge is acquired and localised by learning by doing current products, learning by using the techniques in place and learning by interacting with current customers and rivals. The rate of technological change and the mix between product and process innovations are endogenous and localised by the key role of irreversibility and by the competence accumulated by means of learning processes. In Chapter 17 the traditional divide between innovation, adoption and imitation is reconciled in the context of the economics of localised technological change.
12 Introduction Firms are induced to change their technology when product and factor market conditions do not meet their expectations and irreversible choices make adjustments expensive. Technological change is the result of the combination of research and search activities that lead both to the introduction of new technologies and to imitative adoptions. Both command resources and engender specific revenues. Localised technological change consists of creative adoption where external knowledge and embodied technologies are implemented, with internal competence and idiosyncratic knowledge acquired by means of learning processes. The identification of the net profitability of adoption as defined by the gross profitability of adoption minus adoption costs contributes the economics of technological change. The analysis of the evolution of the net profitability of adoption in the context of the economics of localised technological change shows that the dynamics of creative adoption is able to generate an S-shaped diffusion path at the aggregate level. Chapter 18 shows how and why the complex dynamics of localised technological change leads to path dependence when applied to the analysis of social interactions. Localised technological change is path dependent because of the role of the interplay between irreversibility and hence past dependence, learning conditions and the local characteristics of knowledge externalities and positive feedbacks within knowledge networks. This definition makes it possible to articulate the radical differences between past dependence and path dependence and consequently between path-dependent innovation and path-dependent adoption. These different forms of historicity are analysed, and the key role of path-dependent innovation and structural change as a form of complex system dynamics is appreciated: in this context the distinction between systemic path dependence and individual path dependence is articulated. More specifically this chapter makes clear that path dependence emerges as the form of complex dynamics most apt to analyse the systemic interdependence of creative agents able to change their technology, their knowledge and their location in the topological map of spaces into which their action is embedded. Path dependence provides the analytical framework into which it is possible to analyse the results of complex system dynamics when applied to the intentional action of a myriad of heterogeneous and interdependent agents that are indeed rooted in the relevant spaces but are credited with the capability to elaborate strategic choices albeit based upon procedural rationality. In standard complex dynamic system methodology the topology of the system into which firms are embedded can change but only as the result of stochastic processes. In our approach structural changes instead do take place, as they are one of the results of such systemic interdependence among strategic and creative players. The architecture of the networks into which agents are embedded, next to the generation of new localised technological knowledge and the introduction of new localised technologies, both in the product and the factor markets, is the endogenous result of the collective action of agents. In turn the changes at the system level and specifically in the architecture of the networks of relations that shape the pools of common knowledge generate new mismatches for individual agents and new opportunities for the generation of new knowledge with a continual process.
Introduction 13 Chapters 18 and 19 provide a broader perspective to the analysis as they argue that the process of generation and dissemination of technological knowledge and introduction of technological innovations that lead to economic change is path dependent because at each point in time irreversibility constrains the decision making of actors. Yet their creative reaction and strategic action, based upon a procedural rationality, can engender structural changes that cannot be fully anticipated from their past nor fully foreseen by each individual actor.5 The context into which the reaction takes place is most important. A conducive economic environment, one where collective knowledge is properly implemented by means of effective knowledge governance mechanisms and combines effectively the incentives to the generation of new knowledge and its dissemination and nonexclusive access, facilitates creativity and hence the introduction of innovations. Passive, adaptive reactions are more likely to emerge where poor knowledge networks and knowledge governance mechanisms are in place. The notion of pathdependent complexity here plays a central role: the creative performances of each agent are embedded into a systemic and ever-changing context characterised by the dynamic interplay of a myriad of individuals.6
Part I
The ingredients
2
The dynamic legacies Smith, Marx, Marshall and Schumpeter
1 Introduction The understanding of the endogeneity of technological change is the primary element of the classical legacy. Adam Smith and Karl Marx analyse technological change and technological knowledge as the cause and the consequence of the economic process. Technological change is generated within the economic system and is determined by economic action. Technological change is one of the main types of economic action. In turn technological change has effects on the economic process. The interaction between all changes in product and factor markets in terms of quantities and prices and all changes about the state of technological knowledge and the introduction of new technologies is the prime engine of a restless dynamics (Day, 1983). The classical legacy provides three different and yet consistent models about the working of the endogenous change of technology. Adam Smith puts the emphasis on the role of demand, as the pulling factor of the division of labour and hence of the generation of new technological knowledge and the eventual introduction of new technologies. Karl Marx stresses the role of factor prices and more broadly of substitution mechanisms in inducing technological change. The classical legacy is further enriched by the contributions of Joseph Schumpeter and Alfred Marshall so as to form a dynamic legacy. Alfred Marshall develops the approach of Adam Smith in the analysis of technological change as an endogenous process that is the cause and the consequence of the division of labour, with two important contributions: the role of knowledge in determining the variety of firms that confront each other in the marketplace; and their intrinsic complementarity with respect to the generation and exploitation of technological knowledge. Joseph Schumpeter calls attention to the dual relationship between innovation and competition, stressing both the role of innovation in the working of competition among heterogeneous firms in the markets for products and the role of competition in inducing a continual flow of innovations. These mechanisms deserve careful examination.
2 Adam Smith: the division of labour and the extent of the market Adam Smith provides the most impressive and clear account of the essential role of technological knowledge and technological change as endogenous factors in
18 The ingredients explaining the dynamic character of the economic process (Rosenberg, 1965). The first four books of the founding stone of economics are fully devoted to exploring the economic process and its determinants. The very first lines of An inquiry into the nature and causes of the wealth of nations are devoted to the relationship between productivity and division of labour. According to Adam Smith, the growth in productivity is a consequence of the division of labour: ‘The greatest improvement in the productive powers of labour, and the greater part of the skills, dexterity, and judgment with which it is any where directed, or applied, seem to have been the effects of the division of labour’ (Smith, 1776: 13). The division of labour has a clear causal role in Adam Smith’s view of the origins of the accumulation of competence and knowledge: The division of labour, however, so far as it can be introduced, occasions, in every art, a proportionable increase of the productive powers of labour. The separation of different trades and employment from one another seems to have taken place, in consequence of this advantage. This separation too is generally carried furthest in those countries which enjoy the highest degree of industry and improvement; what is the work of one man, in a rude state of society, being generally that of several in an improved one. (Smith, 1776: 13) More specifically, Adam Smith elaborates a sequence according to which the division of labour is the cause of an increase in competence. The generation of new knowledge builds upon the increase in competence. Technological innovations are the final result of the process: This great increase of the quantity of work, which, in consequence of the division of labour, the same number of people are capable of performing, is owing to three different circumstances: first, to the increase in dexterity in every particular workman; secondly, to the saving of the time which is commonly lost in passing from one species of work to another; and lastly, to the invention of a great number of machines which facilitate and abridge labour, and enable one man to do the work of many. (Smith, 1776: 17) A bottom-up theory of technological knowledge is fully articulated by Adam Smith. Learning by doing and learning by using are at the origin of inventions which eventually make possible the introduction of new and improved machineries: Thirdly and lastly, everybody must be sensible how much labour is facilitated and abridged by the application of the proper machinery. It is unnecessary to give any example. I shall only observe, therefore, that the invention of all those machines by which labour is so much facilitated and abridged seems to have been originally owing to the division of labour. Men are much more likely to discover easier and readier methods of attaining any object, when the whole
The dynamic legacies 19 attention of their minds is directed towards the single object, than when it is dissipated among a great variety of things. But in consequence of the division of labour, the whole of every man’s attention comes naturally to be directed towards some one very simple object. It is naturally to be expected therefore, that some one or other of those who are employed in each particular branch of labour should soon find out easier and readier methods of performing their own particular work, wherever the nature of it admits of such improvements. A great part of the machines ‘made use of’ in those manufactures in which labour is most subdivided, were originally the inventions of common workmen, who, being each of them employed in some very simple operation, naturally turned their thoughts towards finding out easier and readier methods of performing it. Whoever has been much accustomed to visit such manufactures, must frequently have been shewn very pretty machines, which were the invention of ‘such’ workmen, in order to facilitate and quicken their own part of the work. In the first fire-engines, a boy was constantly employed to open and shut alternatively the communication between the boiler and the cylinder, according to the piston either ascended or descended. One of those boys, who loved to play with his companions, observed that, by tying a string from the handle of the valve, which opened this communication, to another part of the machine, the valve would open and shut without his assistance, and leave him at liberty to divert himself with his play-fellows. One of the greatest improvements that has been made upon this machine, since it was first invented, was in this manner the discovery of a boy who wanted to save his own labour. (Smith, 1776: 20–1) Learning by doing and by using processes, internal to each firm, is not the sole factor in the accumulation of new knowledge. An important role is played by the producers of machines and also by scientists: All the improvements in machinery, however, have by no means been the inventions of those who had occasion to use these machines. Many improvements have been made by the ingenuity of the makers of the machines, when to make them became the business of a peculiar trade; and some by that of those who are called philosophers or men of speculation, whose trade it is, not to do anything, but to observe everything; and who, upon that account, are often capable of combining together the powers of the most distant and dissimilar objects. In the progress of society, philosophy or speculation becomes, like every other employment, the principal or sole trade and occupation of a particular class of citizen. Like every other employment too, it is subdivided into a great number of different branches, each of which afford occupation to a peculiar tribe or class of philosophers; and this subdivision of employment in philosophy, as well as in every other business, improves dexterity, and saves time. Each individual becomes more expert in his own peculiar branch, more work is done upon the whole, and the quantity of science is considerably increased by it. (Smith, 1776: 21–2)
20 The ingredients According to Adam Smith the professional competence acquired by means of learning processes and ultimately because of the division of labour is the cause of the skills of workers: The difference of natural talents in different men is, in reality, much less than we are aware of; and the very different genius which appears to distinguish men of different professions, when grown up to maturity, is not upon many occasions so much the cause, as the effect of the division of labour. (Smith, 1776: 28) The division of labour, in conclusion, enters the working of science and becomes a powerful factor in the organisation and efficiency of scientific progress. The reading of Adam Smith confirms the key role of the economics of knowledge in the understanding of the economic process shaped by continual development based upon the introduction of new technologies. One finds in Adam Smith the early foundations of the economic understanding of the mechanisms at work in the generation of technological knowledge. Adam Smith provides a comprehensive analysis where technological knowledge is regarded as the eventual result of at least three processes: 1) a bottom-up process by means of learning by doing and learning by using; 2) the specialised activity of ‘philosophers’ in a top-down process; and finally 3) the interactions with suppliers of machinery and intermediary inputs. After this impressive analysis of the consequences of the division of labour, the attention of Adam Smith concentrates upon the analysis of its causes. The human propensity to exchange and trade is considered an important underlying factor: This division of labour, from which so many advantages are derived, is not originally the effect of any human wisdom, which foresees and intends that general opulence to which it gives occasion. It is the necessary, though very slow and gradual consequence of a certain propensity in human nature which has in view no such extensive utility; the propensity to truck, barter, and exchange one thing for another. (Smith, 1776: 25) Such human inclination however is fuelled and pulled by demand. Adam Smith gives a clear title to Chapter III of the first book, ‘That the division of labour is limited by the extent of the market’. The first lines of the chapter are clear enough: As it is the power of exchanging that gives occasion to the division of labour, so the extent of this division must always be limited by the extent of that power, or, in other words, by the extent of the market. When the market is very small, no person can have any encouragement to dedicate himself entirely to one employment, for want of the power to exchange all that surplus part of the produce of his own labour, which is over and above his own consumption, for such parts of the produce of other men’s labour as he has occasion for. (Smith, 1776: 31)
The dynamic legacies 21 The dynamic engine of Adam Smith is in place. The division of labour is the consequence of the extent of the market and is the cause of the increase of technological knowledge, hence of inventions and eventually technological innovations. Technological innovations in turn lead to the increase in productivity. The increase in productivity leads to the increase in the demand and hence of the extent of the market. The analysis of Adam Smith comes to full circle. The economic process is inherently dynamic and the market forces are likely to generate a self-sustained process of growth. Adam Smith is very clear about the market failure intrinsic about the provision of appropriate levels of support of scientific research and general education: Were there no public institutions for education, no system, no science would be taught for which there was not some demand; or which the circumstances of the times did not render it, either necessary, or convenient, or at least fashionable to learn. A private teacher could never find his account in teaching, either an exploded and antiquated system of science acknowledged to be useful, or a science universally believed to be a mere useless and pedantick heap of a sophistry and nonsense. Such systems, such sciences, can subsist no where, but in those incorporated societies for education whose prosperity and revenue are in great measure independent of their industry. (Smith, 1776: 781) The virtuous mechanism of the division of labour can be trapped by the progressive decline of the sources of creativity: In the progress of the division of labour, the employment of the greater part of those who live by labour, that is, of the great body of the people, comes to be confined to a few very simple operations; frequently one or two. But the understandings of the greater part of men are necessarily formed by their ordinary employments. The man whose whole life is spent in performing a few simple operations, of which the effects too are, perhaps, always the same, or very nearly the same, has no occasion to exert his understanding, or to exercise his invention in finding out expedients for removing difficulties which never occur. He naturally loses, therefore, the habit of such exertion, and generally becomes as stupid and ignorant as possible for a human creature to become. The torpor of his mind renders him not only incapable of relishing or bearing a part in any rational conversation, but of conceiving any generous, noble, or tender sentiment, and consequently of forming any just judgement concerning many even of the ordinary duties of private life . . . But in every improved and civilized society this is the state into which the labouring poor, that is, the great body of the people, must necessarily fall, unless government takes some pains to prevent it. (Smith, 1776: 781–2) In conclusion, the State has the primary responsibility of the provision of basic education and in the support of scientific activities, which could not be fully supported by the spontaneous demand of the marketplace.
22 The ingredients The inquiry of Adam Smith on the wealth of the nations did not finish with the first three chapters of Book I. The rest of his work in fact is dedicated to the theory of value. That theory has received much attention, far more attention than the dynamic theory articulated in the first four books.
3 Alfred Marshall: variety and complementarity Alfred Marshall follows the line of inquiry elaborated by Adam Smith and acknowledges the dual relationship between the division of labour and the introduction of new technologies. Technological change and specialisation are two sides of the same process: Thus the two movements of the improvement of machinery and the growing subdivision of labour have gone together and are in some measure connected. But the connection is not so close as is generally supposed. It is the largeness of markets, the increased demand for great numbers of things of the same kind, and in some cases of things made with great accuracy, that leads to subdivision of labour; the chief effect of the improvement of machinery is to cheapen and make more accurate the work which would anyhow have been subdivided. (Marshall, 1920: Book IV, Chapter IX, §10) Alfred Marshall elaborates the legacy of Adam Smith, grasping the dynamic complexity of structural change, as articulated in the interaction between specialisation and technological change leading to a growing heterogeneity of firms in a context characterised by variety and complementarity: The development of the organism, whether social or physical, involves an increasing subdivision of functions between its separate parts on the one hand, and on the other a more intimate connection between them. Each part gets to be less and less self-sufficient, to depend for its wellbeing more and more on other parts, so that any disorder in any part of a highly-developed organism will affect other parts also. This increased subdivision of functions, or ‘differentiation’, as it is called, manifests itself with regard to industry in such forms as the division of labour, and the development of specialized skill, knowledge and machinery: while ‘integration’, that is, a growing intimacy and firmness of the connections between the separate parts of the industrial organism, shows itself in such forms as the increase of security of commercial credit, and of the means and habits of communication by sea and road, by railway and telegraph, by post and printing-press. (Marshall, 1920: Book IV, Chapter I, §§3–4) Alfred Marshall’s dictum about biology as the Mecca of economics is well known, and indeed Marshall can be considered the first economist to look to biology as a source of inspiration to elaborate a dynamic approach to economics (Foster,
The dynamic legacies 23 1993). Alfred Marshall however is well aware of the traps that the application of a full-fledged Darwinism to the analysis of economic change can engender: The doctrine that those organisms which are the most highly developed, in the sense in which we have just used the phrase, are those which are most likely to survive in the struggle for existence, is itself in process of development. It is not yet completely thought out either in its biological or its economic relations. But we may pass to consider the main bearings in economics of the law that the struggle for existence causes those organisms to multiply which are best fitted to derive benefit from their environment. The law requires to be interpreted carefully: for the fact that a thing is beneficial to its environment will not by itself secure its survival either in the physical or in the moral world. The law of ‘survival of the fittest’ states that those organisms tend to survive which are best fitted to utilise the environment for their own purposes. Those that utilise the environment most, often turn out to be those that benefit those around them most; but sometimes they are injurious. Conversely, the struggle for survival may fail to bring into existence organisms that would be highly beneficial: and in the economic world the demand for any industrial arrangement is not certain to call forth a supply, unless it is something more than a mere desire for the arrangement, or a need for it. It must be an efficient demand; that is, it must take effect by offering adequate payment or some other benefit to those who supply. A mere desire on the part of employees for a share in the management and the profits of the factory in which they work, or the need on the part of clever youths for a good technical education, is not a demand in the sense in which the term is used when it is said that supply naturally and surely follows demand. This seems a hard truth: but some of its harshest features are softened down by the fact that those races, whose members render services to one another without exacting direct recompense are not only the most likely to flourish for the time, but most likely to rear a large number of descendants who inherit their beneficial habits. (Marshall, 1920: Book VIII, Chapter I, §§5–6) Structural change, in Marshall’s view, takes place slowly and gradually: For though, institutions may be changed rapidly; yet if they are to endure they must be appropriate to man: they cannot retain their stability if they change very much faster than he does. Thus progress itself increases the urgency of the warning that in the economic world, Natura non facit saltum. Progress must be slow; but even from the merely material point of view it is to be remembered that changes, which add only a little to the immediate efficiency of production, may be worth having if they make mankind ready and fit for an organization, which will be more effective in the production of wealth and more equal in its distribution; and that every system, which allows the higher faculties of the lower grades of industry to go to waste, is open to grave suspicion. (Marshall, 1920: Book IV, Chapter VIII, §§19–20)
24 The ingredients Alfred Marshall further elaborates the dynamic approach along the lines of the analysis paved by Adam Smith with a few important contributions that make it possible to embed the analysis of technological knowledge at a microeconomic level. First, Alfred Marshall makes explicit that knowledge is a key component of capital and itself a production factor: Capital consists in a great part of knowledge and organization: and of this some part is private property and other part is not. Knowledge is our most powerful engine of production; it enables us to subdue Nature and force her to satisfy our wants. Organization aids knowledge; it has many forms, e.g. that of a single business, that of various businesses in the same trade, that of various trades relatively to one another, and that of the State providing security for all and help for many. The distinction between public and private property in knowledge and organization is of great and growing importance: in some respects of more importance than that between public and private property in material things; and partly for that reason it seems best sometimes to reckon Organization apart as a distinct agent of production. It cannot be fully examined till a much later stage in our inquiry; but something has to be said of it in the present Book. (Marshall, 1920: Book IV, Chapter I, §2) Second, Alfred Marshall identifies the collective character of technological knowledge as a process where a variety of agents, co-localised within the industrial districts, contribute complementary bits of knowledge: When an industry has thus chosen a locality for itself, it is likely to stay there long: so great are the advantages which people following the same skilled trade get from near neighbourhood to one another. The mysteries of the trade become no mysteries; but are as it were in the air, and children learn many of them unconsciously. Good work is rightly appreciated, inventions and improvements in machinery, in processes and the general organization of the business have their merits promptly discussed: if one man starts a new idea, it is taken up by others and combined with suggestions of their own; and thus it becomes the source of further new ideas. (Marshall, 1920: Book IV, Chapter X, §7) Knowledge externalities play a key role in providing firms essential inputs for the generation of new knowledge. The notion of knowledge as both an input and an output can be traced in the Marshallian legacy: On the other hand the small employer has advantages of his own. The master’s eye is everywhere; there is no shirking by his foremen or workmen, no divided responsibility, no sending half-understood messages backwards and forwards from one department to another. He saves much of the book-keeping, and nearly all of the cumbrous system of checks that are necessary in the business
The dynamic legacies 25 of a large firm; and the gain from this source is of very great importance in trades which use the more valuable metals and other expensive materials. And though he must always remain at a great disadvantage in getting information and in making experiments, yet in this matter the general course of progress is on his side. For External economies are constantly growing in importance relatively to Internal in all matters of Trade-knowledge: newspapers, and trade and technical publications of all kinds are perpetually scouting for him and bringing him much of the knowledge he wants – knowledge which a little while ago would have been beyond the reach of anyone who could not afford to have well-paid agents in many distant places. Again, it is to his interest also that the secrecy of business is on the whole diminishing, and that the most important improvements in method seldom remain secret for long after they have passed from the experimental stage. It is to his advantage that changes in manufacture depend less on mere rules of thumb and more on broad developments of scientific principle; and that many of these are made by students in the pursuit of knowledge for its own sake, and are promptly published in the general interest. Although therefore the small manufacturer can seldom be in the front of the race of progress, he need not be far from it, if he has the time and the ability for availing himself of the modern facilities for obtaining knowledge. But it is true that he must be exceptionally strong if he can do this without neglecting the minor but necessary details of the business. (Marshall, 1920: Book IV, Chapter XI, §§14–15) Alfred Marshall accommodates, within competitive markets, the heterogeneity of firms and explains it in terms of the different levels of knowledge and competence possessed by each firm. Quasi-rents are the direct remuneration of the stock of knowledge and competence that each firm has been able to accumulate and valorise: Indeed, in some cases and for some purposes, nearly the whole income of a business may be regarded as a quasi-rent, that is an income determined for the time by the state of the market for its wares, with but little reference to the cost of preparing for their work the various things and persons engaged in it. In other words it is a composite quasi-rent divisible among the different persons in the business by bargaining, supplemented by custom and by notions of fairness – results which are brought about by causes, that bear some analogy to those that, in early forms of civilization, have put the producer’s surplus from the land almost permanently into the hands not of single individuals, but of cultivating firms. Thus the head clerk in a business has an acquaintance with men and things, the use of which he could in some cases sell at a high price to rival firms. But in other cases it is of a kind to be of no value save to the business in which he already is; and then his departure would perhaps injure it by several times the value of his salary, while probably he could not get half that salary elsewhere. (Marshall, 1920: Book VI, Chapter VIII, §35)
26 The ingredients Finally, Alfred Marshall introduces the notion of quasi-irreversibility in economics. Marshall in fact recognises that capital and competence cannot be changed easily. On the contrary, only in the long term can they be adjusted to the changing conditions of the factor and product markets. In the short term, that is as long as capital and competence cannot be adjusted, firms need to cope with the effects of irreversibility: To sum up then as regards short periods. The supply of specialized skill and ability, of suitable machinery and other material capital, and of the appropriate industrial organisation has not time to be fully adapted to demand; but the producers have to adjust their supply to the demand as best they can with the appliances already at their disposal. On the one hand there is not time materially to increase those appliances if the supply of them is deficient; and on the other, if the supply is excessive, some of them must remain imperfectly employed, since there is not time for the supply to be much reduced by gradual decay, and by conversion to other uses. Variations in the particular income derived from them do not for the time affect perceptibly the supply; and do not directly affect the price of the commodities produced by them. The income is a surplus of total receipts over prime cost [that is, it has something of the nature of a rent, as will be seen more clearly in Chapter 8]. But unless it is sufficient to cover in the long run a fair share of the general costs of the business, production will gradually fall off. In this way a controlling influence over the relatively quick movements of supply price during short periods is exercised by causes in the background which range over a long period; and the fear of ‘spoiling the market’ often makes those causes act more promptly than they otherwise would . . . In long periods on the other hand all investments of capital and effort in providing the material plant and the organization of a business, and in acquiring trade knowledge and specialized ability, have time to be adjusted to the incomes which are expected to be earned by them: and the estimates of those incomes therefore directly govern supply, and are the true long-period normal supply price of the commodities produced. (Marshall, 1920: Book V, Chapter V, §§6–7) In conclusion, Alfred Marshall provides a central contribution to the elaboration of a dynamic approach with the understanding of technological knowledge as: 1) a key component of capital, 2) the product of a collective process, 3) the cause and the consequence of the division of labour, 4) a production factor that differentiates firms, and 5) something that commands a specific remuneration called ‘quasi-rents’. Alfred Marshall has enriched considerably the analysis of Adam Smith, grasping the dynamic complexity generated by the interaction between technological change and specialisation, and has provided the first elements in understanding the emerging tension between the growing specialisation and differentiation, on the one hand, and the need for integration and coherence in such a process, on the other. Moreover, Alfred Marshall has introduced the notion of irreversibility in economics,
The dynamic legacies 27 although no attempt has been made to elaborate the relationship between it and technological change. Allyn Young and Nicholas Kaldor have further elaborated the legacy of Adam Smith and Alfred Marshall. Allyn Young is probably the author who contributed more to focus attention on the key role of endogenous dynamics in the work of Adam Smith: Modified, then, in the light of this broader conception of the market, Adam Smith’s dictum amounts to the theorem that the division of labour depends in large part upon the division of labour. This is more than a tautology. It means, if I read its significance rightly, that the counter forces which are continually defeating the forces, which make for economic equilibrium are more pervasive and more deeply rooted in the constitution of the modern economic system than we commonly realize. Not only new or adventitious elements, coming in from the outside, but elements, which are permanent characteristics of the ways in which goods are produced, make continuously for change. Every important advance in the organization of production, regardless of whether it is based upon anything which, in a narrow or technical sense, would be called a new ‘invention’, or involves the fresh application of the fruits of scientific progress to industry, alters the conditions of industrial activity and initiates responses elsewhere in the industrial structure which in turn have a further unsettling effect. Thus change becomes progressive and propagates itself in a cumulative way. (Young, 1928: 533) In so doing, Young captures the critical role of technological change, as both the product and the cause of increasing functional differentiation and complementarity within the economic system, in economic growth. Young lays down the first elements of a system-dynamic approach to understanding economic growth. Economic systems in fact are viewed as complex and dynamic adaptive organisations composed by autonomous and yet interrelated and interdependent units that change over time. Nicholas Kaldor digs even deeper and fully recognises the essential contribution of Adam Smith to building a dynamic theory of the economic process where technological change and technological knowledge pulled by the interplay between the beneficial effects of the division of labour and the extent of the market are the cause and the product of cumulative causation (Kaldor, 1981). As Kaldor (1972) clarifies: The difficulty with a new start is to pinpoint the critical area where economic theory went astray. In my own view, it happened when the theory of value took over the centre of the stage – which meant focusing attention on the allocative functions of markets to the exclusion of their creative functions – as an instrument for transmitting impulses to economic change. To locate the source of error with more precision, I would put it in the middle of the fourth chapter
28 The ingredients of Vol. I of the Wealth of Nations. The first three chapters are devoted to the principle of the Division of Labour. These explain that the larger the production, the more efficient the modes of production that can be employed: the greater the specialisation and the subdivision into different processes. In the first chapter Smith gave numerous reasons for the basic law, beautifully illustrated by the example of pin-making. In the second chapter he explains the peculiarly human characteristic of the propensity to truck, barter and exchange one thing for another – ‘nobody ever saw a dog make a fair exchange of the one bone for another with another dog’ – which alone makes it possible to develop the division of labour through social co-operation. Indeed for Smith the existence of a ‘social economy’ and the existence of increasing returns were closely related phenomena. And the third chapter, perhaps the most significant of them all, is devoted to the proposition that ‘the division of labour is limited by the extent of the market’ a theorem which Allyn Young, writing 150 years later, (in a paper to which I shall refer more extensively presently) regarded as ‘one of the most illuminating and fruitful generalisations which can be found anywhere in the whole literature of economics’. But in the following chapter, after discussing the need for money in a social economy, Smith suddenly gets fascinated by the distinction between money price, real price and exchange value, and from then on, hey presto, his interests get bogged down into the question of how values and prices for products and factors are determined. One can trace a more or less continuous development of price theory from the subsequent chapters of Smith through Ricardo, Walras, Marshall, right up to Debreu and the most sophisticated of present-day Americans. (Kaldor, 1972: 1240–1) If we put aside the theory of value, the dynamic legacy of Adam Smith can be fully appreciated: Yet on an empirical level, nobody doubts that in any economic activity which involves the processing or transformation of basic material – in other words, in industry – increasing returns dominate the picture for the very reasons given by Adam Smith in the first chapters of the Wealth of Nations: reasons that are fundamental to the nature of technological processes and not to any particular technology . . . Finally, there are inventions and innovations induced by experience to which Adam Smith paid the main emphasis – what we now call ‘learning by doing’ or ‘dynamic economies of scale’. The advance in scientific knowledge in physics or in the science of engineering in the laboratory cannot by itself secure the innumerable design improvements that result from the repeated application of particular engineering principles. The optimum design for the steam engine or the diesel engine or the sewing machine has only been achieved after any years or decades of experience: that for the nuclear plant is still far away. The gain in design through experience is even more important in the making of plant and equipment; hence the annual gain of productivity
The dynamic legacies 29 due to ‘embodied technical progress’ will tend to be all the greater the larger the number of plants constructed per year. (Kaldor, 1972: 1243) Kaldor recognises also that such a dynamic and virtuous mechanism can easily fall apart, stressing the fragility of the self-sustained, demand-driven interpretation elaborated by Adam Smith: it is evident from our analysis that the ‘self-sustained growth’ of decentralised economic systems, largely directed, not by exogenous factors, but by the growth and the constellation of demand, is a fragile thing which will only proceed in a satisfactory manner if a number of favourable factors are present simultaneously: such as merchants who are ready to absorb stocks in the short run rather than allow prices to fall too far – because experience has taught them that market prices have some long-run stability – and the manufacturers who respond to the stimulus of growing sales with an expansion of productive capacity, because experience has taught them that over a period markets are growing and not stable. (Kaldor, 1972: 1252) Building on these foundations the post-Keynesian literature has provided much empirical evidence about the key role of the demand pull to explain both the rate and the direction of technological change (Schmookler, 1966).
4 Karl Marx and induced technological change The analysis of technological change plays a key role in the work of Karl Marx. Technological change in fact is the basic tool by means of which capitalists increase profits together with the extraction of surplus value from the production process. Karl Marx stresses the dual role of technological change. On the one hand technological change makes it possible to reduce the price of goods in the marketplace. On the other, technological change makes it possible to increase the extraction of surplus value (Rosenberg, 1992). The competitive process among capitalists feeds the former. The exploitation of labour by capitalists as a class is the ultimate result of the latter. The close intertwining of these mechanisms characterises Marx’s analysis: John Stuart Mill says in his Principles of Political Economy: ‘It is questionable if all the mechanical inventions yet made have lightened the day’s toil of any human being.’ . . . That is, however, by no means the aim of the capitalistic application of machinery. Like every other increase in the productiveness of labour, machinery is intended to cheapen commodities and, by shortening that portion of the working day, in which the labourer works for himself, to lengthen the other portion that he gives without an equivalent to the capitalist. In short, it is a means for producing surplus value. (Marx, 1976: 492)
30 The ingredients Karl Marx makes two founding contributions to the economics of knowledge. First, the collective character of technological knowledge is clearly grasped. Technological knowledge consists in a complex system of machines, skills and workers, all characterised by distinctive elements of complementarity, interoperability and necessary compatibility. Second, technological and scientific knowledge are characterised by the strong elements of non-exhaustibility and limited appropriability (Rosenberg, 1974). Their application however requires dedicated competence and resources, which have a strong idiosyncratic character. As Marx notes: The implements of labour, in the form of machinery, necessitate the substitution of natural forces for human force, and the conscious application of science, instead of rule of thumb. In Manufacture, the organisation of the social labourprocess is purely subjective; it is a combination of detail labourers; in its machinery system, Modern Industry has a productive organism that is purely objective, in which the labourer becomes a mere appendage to an already existing material condition of production. In simple co-operation, and even in that founded on division of labour, the suppression of the isolated, by the collective, workman still appears to be more or less accidental. Machinery, with a few exceptions to be mentioned later, operates only by means of associated labour, or labour in common. Hence the co-operative character of the labour-process is, in the latter case, a technical necessity dictated by the instrument of labour itself . . . We saw that the productive forces resulting from co-operation and division of labour cost capital nothing. They are natural forces of social labour. So also physical forces, like steam, water, etc., when appropriated to productive processes, cost nothing. But just as a man requires lungs to breathe with, so he requires something that is work of man’s hand, in order to consume physical forces productively. A water-wheel is necessary to exploit the force of water, and a steam-engine to exploit the elasticity of steam. Once discovered, the law of the deviation of the magnetic needle in the field of an electric current, or the law of the magnetisation of iron, around which an electric current circulates, cost never a penny. But the exploitation of these laws for the purposes of telegraphy, etc., necessitates a costly and extensive apparatus. The tool, as we have seen, is not exterminated by the machine. From being a dwarf implement of the human organism, it expands and multiplies into the implement of a mechanism created by man. Capital now sets the labourer to work, not with a manual tool, but with a machine which itself handles the tools. Although, therefore, it is clear at the first glance that, by incorporating both stupendous physical forces, and the natural sciences, with the process of production, Modern Industry raises the productiveness of labour to an extraordinary degree, it is by no means equally clear, that this increased productive force is not, on the other hand, purchased by an increased expenditure of labour. Machinery, like every other component of constant capital, creates no new value, but yields up its own value to the product that it serves to beget. In so far as the machine has value, and, in consequence, parts
The dynamic legacies 31 with value to the product, it forms an element in the value of that product. Instead of being cheapened, the product is made dearer in proportion to the value of the machine. And it is clear as noon-day, that machines and systems of machinery, the characteristic instruments of labour of Modern Industry, are incomparably more loaded with value than the implements used in handicrafts and manufactures. (Marx, 1976: 508) The introduction of new technologies cannot be separated from the introduction of new machines in the production process and more generally the substitution of machines to labour. Once again the analysis elaborated by Karl Marx rests on the integration of aspects that are traditionally treated in different analytical layers: the introduction of technological innovations, their embodiment in machines and their substitution to labour: It is evident that whenever it costs as much labour to produce a machine as is saved by the employment of that machine, there is nothing but a transposition of labour; consequently the total labour required to produce a commodity is not lessened or the productiveness of labour is not increased. It is clear, however, that the difference between the labour a machine costs, and the labour it saves, in other words, that the degree of its productiveness does not depend on the difference between its own value and the value of the implement it replaces. As long as the labour spent on a machine, and consequently the portion of its value added to the product, remains smaller than the value added by the workman to the product with his tool, there is always a difference of labour saved in favour of the machine. The productiveness of a machine is therefore measured by the human labour-power it replaces. (Marx, 1976: 513) The relationship between the levels of wages and the actual profitability of introduction and adoption of the new machines is so clear and direct that Marx can understand the relative differences in the profitability of adoption of the same machine in different countries, characterised by different relative factor costs: The use of machinery for the exclusive purpose of cheapening the product is limited in this way, that less labour must be expended in producing the machinery than is displaced by the employment of that machinery. For the capitalist, however, this use is still more limited. Instead of paying for the labour, he only pays the value of the labour power employed; therefore, the limit to his using a machine is fixed by the difference between the value of the machine and the value of the labour power replaced by it. Since the division of the day’s work into necessary and surplus labour differs in different countries, and even in the same country at different periods, or in different branches of industry; and, further, since the actual wage of the labourer at one time sinks below the value of his labour power, at another rises above it, it is
32 The ingredients possible for the difference between the price of the machinery and the price of the labour power replaced by that machinery to vary very much, although the difference between the quantity of labour requisite to produce the machine and the total quantity replaced by it, remain constant . . . But it is the former difference alone that determines the cost, to the capitalist, of producing a commodity and, through the pressure of competition, influences his action. Hence the invention nowadays of machines in England that are in the employed only in North America; just as sixteenth and seventeenth centuries, machines were invented in Germany to be used only in Holland, and just as many a French invention of the eighteenth century was exploited in England alone. In the older countries, machinery, when employed in some branches of industry, creates such a redundancy of labour in other branches that in these latter the fall of wages below the value of labour power impedes the use of machinery, and, from the standpoint of the capitalist, whose profit comes, not from a diminution of the labour employed, but of the labour paid for, renders that use superfluous and often impossible. In some branches of the woollen manufacture in England the employment of children has, during recent years, been considerably diminished, and in some cases has been entirely abolished. Why? Because the Factory Acts made two sets of children necessary, one working six hours, the other four, or each working five hours. But the parents refused to sell the ‘half-timers’ cheaper than the ‘full-timers’. Hence the substitution of machinery for the ‘half-timers’ . . . Before the labour of women and of children under 10 years of age was forbidden in mines, capitalists considered the employment of naked women and girls, often in company with men, so far sanctioned by their moral code, and especially by their ledgers, that it was only after the passing of the Act that they had recourse to machinery. The Yankees have invented a stone-breaking machine. The English do not make use of it, because the ‘wretch’ . . . who does this work gets paid for such a small portion of his labour, that machinery would increase the cost of production to the capitalist . . . In England women are still occasionally used instead of horses for hauling canal boats . . . because the labour required to produce horses and machines is an accurately known quantity, while that required to maintain the women of the surplus population is below all calculation. Hence nowhere do we find a more shameful squandering of human labour-power for the most despicable purposes than in England, the land of machinery. (Marx, 1976: 515–17) The increase in the profitability of each firm is the direct incentive to the action of each capitalist introduction of technological innovation embodied in new machines. Its introduction at the system level has the direct effect of substituting capital to labour: The instrument of labour, when it takes the form of a machine, immediately becomes a competitor of the workman himself . . . The self-expansion of capital
The dynamic legacies 33 by means of machinery is thenceforward directly proportional to the number of the work-people, whose means of livelihood have been destroyed by that machinery. The whole system of capitalist production is based on the fact that the workman sells his labour-power as a commodity. Division of labour specialises this labour-power, by reducing it to skill in handling a particular tool. So soon as the handling of this tool becomes the work of a machine, then, with the use-value, the exchange-value too, of the workman’s labourpower vanishes; the workman becomes unsaleable, like paper money thrown out of currency by legal enactment. That portion of the working class, thus by machinery rendered superfluous, i.e., no longer immediately necessary for the self-expansion of capital, either goes to the wall in the unequal contest of the old handicrafts and manufactures with machinery, or else floods all the more easily accessible branches of industry, swamps the labour market, and sinks the price of labour-power below its value. (Marx, 1976: 556–7) The competitive pressure of the capitalists however fuels the dynamics of the process. The decline of the number of workers is relative, but not absolute: Nevertheless, in spite of the mass of hands actually displaced and virtually replaced by machinery, we can understand how the factory operative, through the building of more mills and the extension of old ones in a given industry, may become more numerous than the manufacturing workmen and handicraftsmen that have been displaced. Suppose, for example, that in the old mode of production, a capital of £500 is employed weekly, two-fifths being constant and three-fifths variable capital, i.e., £200 being laid out in means of production, and £300, say £1 per man, in labour-power. On the introduction of machinery the composition of this capital becomes altered. We will suppose it to consist of four-fifths constant and one-fifth variable, which means that only £100 is now laid out in labour-power. Consequently, two-thirds of the workmen are discharged. If now the business extends, and the total capital employed grows to £1500 under unchanged conditions, the number of operatives employed will increase to 300, just as many as before the introduction of the machinery. If the capital further grows to £2000, 400 men will be employed, or one-third more than under the old system. Their numbers have, in point of fact, increased by 100, but relatively, i.e., in proportion to the total capital advanced, they have diminished by 800, for the £2000 capital would, in the old state of things, have employed 1200 instead of 400 men. Hence, a relative decrease in the number of hands is consistent with an actual increase. (Marx, 1976: 577) The systematic application of scientific knowledge to the production process becomes, in Marx, the distinctive feature of capitalism. In order to keep the process in motion, technological knowledge is constantly reproduced and expanded:
34 The ingredients As with the division of labour in the interior of the manufacturing workshops, so it is with the division of labour in the interior of society. So long as handicraft and manufacture form the general groundwork of social production, the subjection of the producer to one branch exclusively, the breaking up of the multifariousness of his employment . . . is a necessary step in the development. On that ground-work each separate branch of production acquires empirically the form that is technically suited to it, slowly perfects it, and, so soon as a given degree of maturity has been reached, rapidly crystallises that form. The only thing, that here and there causes a change, besides new raw material supplied by commerce, is the gradual alteration of the instruments of labour. But their form, too, once definitely settled by experience, petrifies, as is proved by their being in many cases handed down in the same form by one generation to another during thousands of years. A characteristic feature is that, even down into the eighteenth century, the different trades were called ‘mysteries’ (mysteres) . . . [and] into their secrets none but those duly initiated could penetrate. Modern industry rent the veil that concealed from men their own social process of production, and that turned the various spontaneously divided branches of production into so many riddles, not only to outsiders, but even to the initiated. The principle which it pursued of resolving each process into its constituent movements, without any regard to their possible execution by the hand of man, created the new modern science of technology. The varied, apparently unconnected, and petrified forms of the industrial processes now resolved themselves into so many conscious and systematic applications of natural science to the attainment of given useful effects. Technology also discovered the few main fundamental forms of motion, which, despite the diversity of the instruments used, are necessarily taken by every productive action of the human body; just as the science of mechanics sees in the most complicated machinery nothing but the continual repetition of the simple mechanical powers. Modern industry never looks upon and treats the existing form of a process as final. The technical basis of that industry is therefore revolutionary, while all earlier modes of production were essentially conservative . . . By means of machinery, chemical processes, and other methods, it is continually causing changes not only in the technical basis of production, but also in the functions of the labourer, and in the social combinations of the labour process. At the same time, it thereby also revolutionises the division of labour within the society, and incessantly launches masses of capital and of workpeople from one branch of production to another. But if modern industry, by its very nature, therefore, necessitates variation of labour, fluency of function, universal mobility of the labourer, on the other hand, in its capitalistic form it reproduces the old division of labour with its ossified particularisations. We have seen how this absolute contradiction between the technical necessities of modern industry, and the social character inherent in its capitalistic form, dispels all fixity and security in the situation of the labourer; how it constantly threatens, by taking away the instruments of labour, to snatch from his hands
The dynamic legacies 35 his means of subsistence . . . and, by suppressing his detail-function, to make him superfluous. (Marx, 1976: 615–16) In the Grundrisse (economic manuscripts of 1857–58, first version of Capital), the analysis of Marx of the central role of science in the capitalist process reaches extraordinary levels of clarity and insight. Technological change is fully endogenous to the economic system. More specifically, Marx argues that the levels of endogeneity of technological change are themselves an indicator of the advance of an economic system: The full development of capital, therefore, takes place – or capital has posited the mode of production corresponding to it – only when the means of labour has not only taken the economic form of fixed capital, but has also been suspended in its immediate form, and when fixed capital appears as a machine within the production process, opposite labour; and the entire production process appears as not subsumed under the direct skilfulness of the worker, but rather as the technological application of science. [It is,] hence, the tendency of capital to give production a scientific character; direct labour [is] reduced to a mere moment of this process. As with the transformation of value into capital, so does it appear in the further development of capital, that it presupposes a certain given historical development of the productive forces on one side – science too [is] among these productive forces – and, on the other, drives and forces them further onwards. (Marx, 1857, Notebook VI) In the Grundrisse, Marx introduces the notion that invention becomes a business: In machinery, the appropriation of living labour by capital achieves a direct reality in this respect as well: It is, firstly, the analysis and application of mechanical and chemical laws, arising directly out of science, which enables the machine to perform the same labour as that previously performed by the worker. However, the development of machinery along this path occurs only when large industry has already reached a higher stage, and all the sciences have been pressed into the service of capital; and when, secondly, the available machinery itself already provides great capabilities. Invention then becomes a business, and the application of science to direct production itself becomes a prospect which determines and solicits it. But this is not the road along which machinery, by and large, arose, and even less the road on which it progresses in detail. This road is, rather, dissection [Analyse] – through the division of labour, which gradually transforms the workers’ operations into more and more mechanical ones, so that at a certain point a mechanism can step into their places. (See under economy of power.) Thus, the specific mode of working here appears directly as becoming transferred from the worker to capital in the form of the machine, and his own labour capacity devalued thereby. Hence
36 The ingredients the workers’ struggle against machinery. What was the living worker’s activity becomes the activity of the machine. Thus the appropriation of labour by capital confronts the worker in a coarsely sensuous form; capital absorbs labour into itself – ‘as though its body were by love possessed’. (Marx, 1858, Notebook VII) The notion of knowledge as an endogenous productive force is clearly identified by Marx. The levels of endogeneity of knowledge, as a distributed economic force shared by a myriad of agents, fed by the combination of learning processes that lead to the accumulation of competence and tacit knowledge with scientific processes of deduction, and transformed into a means of accumulation of capital pushed by profit maximisation (the general intellect), become a measure of the advance of capitalism as a social system: Nature builds no machines, no locomotives, railways, electric telegraphs, selfacting mules etc. These are products of human industry; natural material transformed into organs of the human will over nature, or of human participation in nature. They are organs of the human brain, created by the human hand; the power of knowledge, objectified. The development of fixed capital indicates to what degree general social knowledge has become a direct force of production, and to what degree, hence, the conditions of the process of social life itself have come under the control of the general intellect and been transformed in accordance with it. To what degree the powers of social production have been produced, not only in the form of knowledge, but also as immediate organs of social practice, of the real life process. (Marx, 1858, Notebook VII, italics added) The strength of Marx’s analysis lies in the close intertwining of the dynamics at the system level with the understanding of the competitive pressure at the disaggregate level. The Marxian legacy about the economic analysis of technological change and technological knowledge at large has been often blurred, as in the case of Adam Smith, with the fate of his theory of value. Yet it deserves to be identified and appreciated. The role of the price inducement mechanism, at the aggregate level, was retained eventually by John Hicks: A change in the relative prices of the factors of production is itself a spur to invention, and to invention of a particular kind – directed to economising the use of the factor which has become relatively expensive. The general tendency to a more rapid increase of capital than labour which has marked European history during the last few centuries has naturally provided a stimulus to laboursaving invention. If, therefore, we properly appreciate the place of invention in economic progress, we need to distinguish between two sorts of inventions. We must put on one side those inventions which are the result of a change in the relative prices of the factors; let us call these ‘induced’ inventions. We shall expect, in practice, all or nearly all induced inventions to be labour-saving;
The dynamic legacies 37 but there is no reason why autonomous invention should be predominantly labour-saving. (Hicks, 1932: 125) The strength of Marx’s analysis is clearly the implementation of an inducement mechanism which can be considered an augmented substitution process. The weaknesses are all related to the specific theory of value into which his theory of technological change is embedded. When his theory of value is dropped, there is no reason to assume that technological change is necessarily directed towards the reduction of labour, and a broader hypothesis about the role of factor intensity and factor costs in determining the direction of technological change can be elaborated (Marquetti, 2003). Specifically a distinction has to be made between models of induced technological change, which focus the changes in factor prices, and models of induced technological change, which stress the static conditions of factor markets. In the first approach, following Hicks and Marx, firms are induced to change their technology when the price of a production factor increase (Hicks, 1932). The change in factor prices acts as a powerful inducement mechanism, which explains both the rate and the direction of introduction of new technologies. The change in factor prices in fact induces firms to introduce new technologies, specifically directed to save on the factor, which has become more expensive. The introduction of new technologies complements the standard substitution process, i.e. the technical change consisting in the selection of new techniques, defined in terms of factor intensities, on the existing isoquants. In this case technological change is considered an augmented form of substitution: technological change complements technical change. This approach to the induced technological change differs from the static version, elaborated by Kennedy (1964) and Samuelson (1965), according to which firms introduce new technologies in order to save on the production factors that are relatively more expensive. In this second approach the levels of factor price matter instead of the rates of change. This approach has shown a major limitation of the former. From simple algebraic calculation it is in fact clear that firms have an incentive to introduce labour-intensive technologies, in labour-abundant and capital-scarce regions and countries, even after an increase in wages. The Kennedy–von Weiszacker–Samuelson approach however is severely limited from the dynamic viewpoint. It is no longer clear when and why firms should innovate. Consistently only the direction of technological change can be induced, rather than the rate (Binswanger and Ruttan, 1978; Ruttan, 1997, 2001). Both approaches, as is well known, have been often criticised using Salter’s argument, according to which firms should be equally eager to introduce any kind of technological change, either labour or capital intensive, provided it makes it possible to reduce production costs and increase efficiency. It is interesting to note that the analysis of the role of relative factor endowments in explaining the direction of technological change has been recently revived to explain the bias of new information and communication technologies in terms of skill intensity (Acemoglu, 2002).
38 The ingredients The analysis of the role of capitalistic rivalry, as the essential dynamic component of the process, has been taken over by Joseph Schumpeter.
5 Joseph Schumpeter and the competitive innovation Every student of economics of innovation is aware of the path-breaking contribution of Joseph Schumpeter to the economics of innovation and technological change (Rosenberg, 1994). An extensive and comprehensive analysis of his many contributions to this approach would easily risk being repetitive. Much attention has been called to the evolution of the thinking of Schumpeter upon the role of technological change in economic development. A divide between the ‘first’ and the ‘second’ Schumpeter has been identified. The ‘first’ Schumpeter, that is the tradition based upon the Theorie der Wirtschaftlichen Entwicklung originally published in German in 1911, pays attention to entrepreneurship as the driving mechanism. Entrepreneurs who create new firms to enter the markets would be the primary source of technological innovations. The key role attributed to entrepreneurship has raised some problems about the endogeneity of innovation: entrepreneurs are outsiders. Little analysis is made of innovations introduced by incumbents. The ‘second’ Schumpeter is based upon the 1942 book Capitalism, socialism and democracy, where the driving role of the large corporation as the engine for the introduction of innovations is highlighted. The well-known Schumpeterian hypothesis is based on this second book: monopolistic power and the large size of corporations favour the allocation of resources to generating and appropriating new technologies. A divide between static and dynamic efficiency arises. The ‘second’ Schumpeter however expresses some concern about the longterm viability of the competitive mechanisms based upon innovations, because of the increasing routinisation of the activities leading to the introduction of innovations within the large corporation. Innovation here is fully endogenous, but ‘routinised’: it is expected to disappear. While much attention has been paid to these books, two key journal articles, ‘The instability of capitalism’ of 1928 and ‘The creative response in economic history’ of 1947, deserve new attention. In ‘The instability of capitalism’, published in the Economic Journal in 1928, the ‘two Schumpeters’ are well integrated and coexist consistently. Here the theoretical distance between the dynamic analysis of the economic process, based upon the understanding of the central role of technological change in the market competition, and the Walrasian analysis of the general equilibrium is especially clear. Schumpeter stresses that equilibrium analysis cannot consider the role of technological change: Innovations in productive and commercial methods, in the widest sense of the term – including specialization and the introduction of production on a scale different from the one which ruled before – obviously alter the data [italics in the text] of the static system and constitute, whether or not they have to do with ‘invention’, another body of facts and problems. (Schumpeter, 1928: 18)
The dynamic legacies 39 The divide between the equilibrium analysis and the Schumpeterian approach becomes clearer and clearer: What we, unscientifically, call economic progress means essentially putting productive resources to uses hitherto untried in practice [italics in the text], and withdrawing them from the uses they have served so far. This is what we call ‘innovation’. What matters for the study of this subject is merely the essentially discontinuous character of this process, which does not lend itself to description in terms of a theory of equilibrium. (Schumpeter, 1928: 32) Innovation, as distinct from invention, is not only endogenous but the intrinsic element of the capitalistic economy. Innovation cannot be regarded as an external economy, because this is the distinctive feature of the competitive process: The firm begins to undersell the others, part of which are thereby definitely pushed into the background to linger there on accumulated reserves and quasirents, whilst another part copies the methods of the disturber of peace. That [italics in the text] this is so we can see every day by looking at industrial life; it is precisely what goes on, what is missing in the static apparatus and what accounts both for dissatisfaction with it and for the attempts to force such phenomena into its cracking frame – instead of, as we think it natural to do, recognizing and explaining this as a distinct process going on along with the one handled by the static theory. (Schumpeter, 1928: 33) The Schumpeterian approach to innovation as an essential component of the competitive process is quite consistent with the Marshallian interpretation of the competitive process. Variety and selection are essential elements of the Marshallian notion of competition. Firms, diverse in terms of size, location and efficiency, confront each other in the product marketplace and are sorted out by the working of the competitive process. Entry and exit feed the dynamics of the process. In this context each firm is confronted with a continual redefinition of its relative market context and has to face the competitive threat brought about by firms that are able to produce at lower costs because of either the access to cheaper production factors or more effective production technologies. In the Marshallian competition, the duration of the adjustment process to an eventual equilibrium is endless, and firms experience prolonged out-of-equilibrium conditions in which they can earn transient and yet heterogeneous levels of profits. Once more, after Adam Smith and Karl Marx, it seems clear that innovation is the distinctive element of a dynamic process which cannot be analysed with the equilibrium approach: If, then, the putting to new uses of existing resources is what ‘progress’ fundamentally consists in; if it is the nature of the entrepreneur’s function to
40 The ingredients act as the propelling force of the process; if the entrepreneur’s profit, credits, and the cycle prove to be essential parts of this mechanism – the writer even believes this to be true of interest – then industrial expansion per se [italics in the text] is better described as a consequence than a cause; and we should be inclined to turn the other way around what we have termed the received chain of causation. In this case, as those phenomena link up so as to form a coherent and self-contained logical whole, it is obviously conducive to clearness to bring them out boldly; to relegate the one distinct body of doctrine the concept of equilibrium, the continuous curves and small marginal variations, all of which, in their turn, link up with the circuit flow of economic routine under constant data; and to build, alongside of this, and before [italics in the text] taking into account the full complexity of the ‘real’ phenomenon – secondary waves, change occurrences, ‘growth’ and so on – a theory of capitalist change, assuming, in so doing, that non-economic conditions or data are constant and automatic and gradual change in economic conditions is absent. But there is no difficulty in inserting all this. (Schumpeter, 1928: 38) With the analysis of the role of creative response in economic history Schumpeter (1947) fully elaborates the view that firms and agents at large are induced to innovate as a form of creative reaction. Firms react to the changing conditions of both product and factor markets in a creative way, with the introduction of innovations, in both technologies and organisations and changing their products and processes. In the Schumpeterian approach, as articulated with clarity in 1947, firms do more than adjusting prices to quantities and vice versa: firms innovate. At the same time innovation is no longer viewed as the result of the ingenuity of outsiders which enter the marketplace by means of new products or new processes: incumbents innovate in order to face unexpected changes in the economic environment. With the notion of creative response, innovation becomes fully endogenous to the economic system: What has not been adequately appreciated among theorists is the distinction between different kinds of reaction to changes in ‘condition’. Whenever an economy or a sector of an economy adapts itself to a change in its data in the way that traditional theory describes, whenever, that is, an economy reacts to an increase in population by simply adding the new brains and hands to the working force in the existing employment, or an industry reacts to a protective duty by the expansion within its existing practice, we may speak of the development as an adaptive response. And whenever the economy or an industry or some firms in an industry do something else, something that is outside of the range of existing practice, we may speak of creative response. Creative response has at least three essential characteristics. First, from the standpoint of the observer who is in full possession of all relevant facts, it can always be understood ex post; but it can practically never be understood ex
The dynamic legacies 41 ante; that is to say, it cannot be predicted by applying the ordinary rules of inference from the pre-existing facts. This is why the ‘how’ in what has been called the ‘mechanisms’ must be investigated in each case. Secondly, creative response shapes the whole course of subsequent events and their ‘long-run’ outcome. It is not true that both types of responses dominate only what the economist loves to call ‘transitions’, leaving the ultimate outcome to be determined by the initial data. Creative response changes social and economic situations for good, or, to put it differently, it creates situations from which there is no bridge to those situations that might have emerged in the absence. This is why creative response is an essential element in the historical process; no deterministic credo avails against this. Thirdly, creative response – the frequency of its occurrence in a group, its intensity and success or failure – has obviously something, be that much or little, to do (a) with quality of the personnel available in a society, (b) with relative quality of personnel, that is, with quality available to a particular field of activity relative to the quality available, at the same time, to others, and (c) with individual decisions, actions, and patterns of behavior. (Schumpeter, 1947: 149–50) The Schumpeterian legacy has been especially fertile in articulating the key relationship between rivalry and intentional innovation. The notion of creative destruction plays a key role in this line of inquiry and has made it possible to explore the causal relations between barriers to entry, levels of mark-ups, market structure and the incentives to introduce new technologies (Sylos Labini, 1956, 1984). The structure–conduct–performance paradigm has been elaborated and it has made it possible to gather much empirical evidence able to confirm that the size distribution of firms, the levels of concentration and the forms of competition among firms do affect the rates of introduction of innovations and their characteristics (Scherer, 1984, 1992; Aghion and Howitt, 1992; Audretsch, 1995). The circular relationship between market structure and innovation has been subsequently identified: the conduct and the performances of firms are indeed influenced by the market structure as it stands at time t, but in turn they exert strong influences upon the characteristics of the market structure at time t+1, with the introduction of innovations. A new market structure is determined and firms, in order to readjust to it, elaborate new strategies that include the introduction of further innovations. The understanding of this recursive relationship has paved the way to grasping the basic elements of the continual and dynamic system of feedbacks between the conduct and the performance of firms and the rate and direction of technological change and structural change, with a growing awareness of its evolving and historical characteristics (Momigliano, 1975; Antonelli, 1982, 1995). The dynamic efficiency of a system, as measured by its capability to increase the overall efficiency by means of the introduction of technological and organisational innovations, becomes the new key parameter to assess the actual welfare.
42 The ingredients
6 Towards a full endogeneity of technological change The integration of the three models provided by the classical legacy offers the base upon which a fully endogenous approach to explaining the causes of both the generation of technological knowledge and the rate and direction of the eventual introduction and selective diffusion of new technologies can be elaborated. It is important here to draw a clear distinction between four levels of endogeneity of technological change that can be found in the literature. Weak endogeneity recognises that the generation of new knowledge is endogenous to the institutional set-up of an economic system as well as to the amount of resources that are invested in research activities. For a given supply of talented and creative agents, the social organisation of the education and research system and the amount of resources available influence the actual amount of knowledge generated and hence the eventual supply of new technologies. In this approach the rate of technological change is considered as partly endogenous as it is set to depend – also – upon the amount of resources invested in research activities. The direction of technological change remains exogenous, and no assumptions are made upon the rationale of incentives and causes for the actual introduction of new technologies at the microeconomic level. No explanation of the interplay between the availability of funds, the generation of new knowledge and the introduction of new technologies is provided in this context. A benevolent and foresighted dictator is requested in order to assess the correct amount of resources and their distribution in the many possible complementary activities that contribute the successful introduction of new technologies (Arrow, 1962a, 1969). Baumol (2002) has recently contributed to this line of inquiry, highlighting the role of the social organisation of economic, institutional and social mechanisms of identification and valorisation of the ‘given’ supply of creative talents distributed at random in any economic system. The larger the number of creative talents each system is able to identify and valorise, the larger is the dynamics of growth in output and efficiency in the economic system. Here creative talents are an exogenous characteristic distributed at random, but the filtering mechanisms elaborated within the economic system are endogenous. The analysis of Audretsch (2006) on the relationship between knowledge and entrepreneurship provides another interesting approach to the elaboration of a partial endogeneity with a strong Schumpeterian flavour. According to Audretsch, entrepreneurship, very much along the lines of the line of inquiry framed by Schumpeter (1911), is a key factor of introduction of technological change into economic systems. In turn however entrepreneurship is not the exogenous product of societal or anthropological or institutional forces, but rather a direct expression of the levels of knowledge available in an economic system. The flow of innovative entrepreneurs in other words is larger in economic systems where and when the sources of knowledge are wider and deeper. The determinants of the generation of knowledge remain an explanandum. Much recent economics, influenced by the Arrovian legacy, has made significant efforts to provide a framework of soft endogeneity, building upon the role of learning and spillover as the key base for the generation of knowledge. Learning
The dynamic legacies 43 is the basic ingredient that leads to the generation of knowledge and eventually to the growth of total factor productivity. In this approach the rate of introduction of new technologies should be ubiquitous, spontaneous and automatic. One again, the great heterogeneity of the rates of introduction of innovations with respect to firms, regions, industries and times remains unexplained. The new theory of growth has contributed to this approach, building upon a very preliminary economics of knowledge. In the new growth theory knowledge is considered as a quasi-private good, with the standard characteristics of diminishing returns. Two rough categories of knowledge are considered: generic knowledge and specific knowledge. Only the latter can be fully appropriated by the inventor. A strong complementarity between generic and specific knowledge is assumed. Appropriability of specific knowledge engenders the incentives to private investments to research and development. Firms invest in research and development attracted by the benefits of high appropriability regimes of specific knowledge; in so doing however they also fund the accumulation of generic knowledge that spills freely in the air and benefits the system at large. The acquisition of external knowledge can be done with no cost and in any institutional environment. The relationship between the specific knowledge generated by means of research and development activities and the generic knowledge spilling in the atmosphere is not investigated. No assumptions are made about its variance over time and in different contexts. Moreover it is clear that in this approach the system is unable to reach an equilibrium point: private investment in research and development activities, in fact, is funded only taking into account the private profitability stemming from the appropriability conditions of specific knowledge. The strong diachronic and synchronic variance, i.e. within each system, in the levels of research and development activities, in the rates of introduction of innovations, in the directions of technological change, far away from the implicit assumptions about its neutrality, in the rates of growth of total factor productivity all documented by much empirical literature, across countries and regions within countries, across industries and firms and most important across periods of time cast many doubts on the relevance of this approach (Romer, 1986, 1990, 1994). Partial endogeneity assumes that the economic process plays a role, although limited to the selective diffusion of the technologies. In this approach the generation of new technological knowledge remains exogenous, as well as the introduction of new technologies. Many new technologies are being introduced for reasons that remain beyond the boundaries of economic analysis. They eventually become available on a shelf that closely resembles the textbook shelf of techniques. Occasionally a subset of such an unlimited and unconstrained supply of new knowledge and new technologies is sorted out; some fail to diffuse while others are adopted and effectively applied. A theory able to explain the determinants of such a selective diffusion process can be elaborated on such a base, but this is not a theory about the determinants of the introduction of innovations (Lane et al., 2006). Evolutionary approaches have contributed much to this approach, but they have not been able to overcome the limitations of Schumpeter mark one, according to
44 The ingredients whom innovations are either the result of random processes or the product of the supply of entrepreneurs. Much progress has been made in understanding the morphology of technological change with the identification of many relevant characteristics, but little analysis has been done to understand why some firms are able to generate more innovations than others, why some firms are more willing to innovate than others and why technological change is faster in some regions and in some periods than in others. The evolutionary approach, consistently with its Darwinian foundations, has a strong theory about the diffusion and the selection of innovations, but lacks a theory of the determinants of innovation. Innovations, like mutations, seem to be introduced at random, with no clear insight about the causes (Dosi and Nelson, 1994). Yet much empirical evidence confirms that the rates of introduction of innovation differ widely across regions, firms and time. Yet evolutionary approaches provide the building blocks upon which an economic theory of endogenous change can be implemented. Finally strong endogeneity can be fully articulated only when the generation of new technological knowledge, the direction and the rate of introduction of innovations, and their actual adoption in the system are analysed as the result of the intentional conduct of agents and of the results of their interactions. Strong endogeneity can be built merging the classical and the Schumpeterian legacies in an evolutionary framework implemented by the notion of path-dependent system dynamics where the result of the aggregate dynamics is indeed the product of the decision making at the individual level of a myriad of creative and interdependent agents, embedded within networks of interactions, constrained by irreversibility but able to change their technologies, but does not correspond to the project of any of them.
7 Conclusion The dynamic legacies provide economics of innovation and new technology with the founding stones for an economic theory of technological change as an intrinsic and fully endogenous form of economic action. There is a clear and strong common point: technological change is not exogenous, but rather the endogenous, intrinsic character of the economic process. Two dynamic models have been articulated in the classical tradition. Adam Smith has laid down the basic elements of a theory of economic development based upon the continual, albeit fragile, increase of productivity explained by the specialisation and effective learning made possible by the division of labour and ultimately by the increasing extent of the market. The productivity growth is generated by the division of labour and is the cause of the growth of the demand. In turn the growth of the demand makes possible increased levels of division of labour and hence of productivity growth. Karl Marx has understood, better than any other, the dialectic role of technological change. Technological change makes it possible to save on the factors that are becoming more expensive. At the same time however it engenders further shift in its demand and hence feeds a self-sustained process of change and development. Technological change is considered in this model as an
The dynamic legacies 45 expanded and augmented form of technical substitution. The increase of wages pushes the firm not only to substitute labour with capital, but to introduce new technologies that, embodied in new machines, drastically reduce the demand for labour. The decline in the prices for products manufactured with the new machines, however, pushes an extensive growth of the system, unemployment falls, wages increase again and again, and capitalists rely on intensive mechanisms: they take advantage of new technological knowledge and introduce new labour-saving technologies. Alfred Marshall implements the classical legacy, stressing the collective character of technological knowledge within a process of continual structural change articulated in a twin process of specialisation and integration. Finally Joseph Schumpeter relates technological change to the competitive mechanism of the market process. Competition is based upon the introduction of new products, new processes and new uses of productive resources, at large. Thus innovation is at the heart of capitalistic development. Innovation is not an externality and as such cannot be reconciled with the equilibrium approach. The models elaborated in the classical tradition are complementary and can be easily regarded as components of a broader and more comprehensive Schumpeterian approach that includes the classical legacy (Marchionatti, 1988). Joseph Schumpeter provides the microeconomic ingredient. Firms introduce technological changes in the context of a Marshallian market rivalry as a form of creative reaction to unexpected changes to (re)gain market share and increase profitability. Marshallian knowledge externalities play a key role in this context as they provide firms with essential inputs for the generation of new knowledge. Profits last as long as imitators enter the market and draw prices closer and closer to marginal costs. Imitative entry pushes the growth of demand. Monopolistic competition characterises the industrial dynamics in product markets much more than price adjustments in competitive markets. At the aggregate level the model elaborated by Adam Smith enters into the picture. The growth of demand leads to a wider and deeper division of labour. The new levels of the division of labour stimulate the learning processes and the generation of new technological knowledge and hence offer new opportunities for the introduction of new technologies by profit-seeking entrepreneurs. The demand-pull mechanism works at the aggregate level and provides a clue to understanding the rates of introduction of new technologies. Here the relationship between the division of labour and the extent of the market is the founding stone of a process of creation of opportunities. The growth in demand and the increasing levels of division of labour help in understanding the rates of technological change. The inducement mechanism elaborated by Karl Marx helps in understanding the direction of technological change. The dynamic process progressively exhausts the relative availability of production factors. The introduction of technological changes appears to be more profitable if directed towards the reduction in usage of the resources that have become scarcer and towards an increase of the resources that are relatively more abundant. Technological change makes it possible to substitute the factors that have become relatively more expensive and to make a more productive use of the resources that are relatively less expensive. The inducement mechanisms
46 The ingredients engendered by the changing relative prices helps in explaining the direction of technological change as well as the rates. The model of demand pull put forward by Adam Smith and the factor price inducement mechanism grasped by Karl Marx complement each other at the aggregate level and together they rely upon the disaggregate dynamics articulated by the analysis of the competitive pressure elaborated by Joseph Schumpeter (Elster, 1983, 1986). The classical legacy implemented by the dynamic analysis of Alfred Marshall and substantiated by the key Schumpeterian notion of creative reaction provides a solid point of departure to elaborate a more articulated evolutionary understanding of the endogenous generation of new technological knowledge, the eventual introduction of technological innovations and the structural changes that characterises the complex dynamics of economic development and growth.
3
Localised technological change A critical assessment
1 Introduction The localised approach to technological change makes it possible to fully articulate the hypothesis of strong endogeneity of technological change. In this approach technological change is endogenous to the economic process and it is the prime factor of continual change as it is the result of the pressure of economic forces on both the demand and the supply side, at the aggregate level and the agent level. The localised approach makes it possible to overcome the traditional dichotomy. Clearly technological change cannot be treated like the exogenous fall of manna from heaven. Neither is it possible to treat technological change like the customary result of routine activities: total factor productivity growth measures confirm that technological change yields results that are far greater than any rational calculations based upon marginal productivity might account for. The need to combine into a homogeneous framework the endogenous understanding of the dynamics by means of which technological – and organisational – change is introduced in the economic system, with the elements of surprise and the unknown that necessarily characterise it, has always proven challenging for economic analysis. The localised technological progress approach provides an attempt to solve the puzzle by building upon different traditions of analysis: the bounded rationality and limited knowledge framework for understanding sequential decision making based upon procedural rationality that now includes learning capabilities and creativity, the inducement approach, the economics of learning and the economic analysis of irreversibility. The key point is that firms are induced to change their routines and their technologies when a mismatch between plans and actual conditions emerges. Such an innovative reaction is made necessary and shaped by the burden of irreversibility. At the same time it is made possible, and yet constrained, by the dynamics of learning and the effects of limited knowledge and procedural rationality. The localised approach provides a framework to analysing technological change as the endogenous and induced outcome of an out-of-equilibrium self-sustaining dynamics that takes place in a set of highly specific and contextual circumstances. To do this it integrates different strands of literature in order to overcome the
48 The ingredients criticisms and shortcomings of each of them. The rest of the chapter is organised as follows. A simple exposition of the process mechanism is presented in section 2. The basic ingredients of the localised approach are identified in section 3. Section 4 presents the multidimensional analysis of the localised approach. Section 5 highlights the implications of the analysis in terms of complex dynamics and path dependence. The conclusion summarises the main results and puts them in perspective.
2 The process The introduction of technological innovations is the result of the creative reaction of firms, induced by changes in product and factor markets that firms are not able to cope with, by means of movements in the given technical space. The creative reaction of firms is possible especially when an appropriate environment favours it, although constrained in a limited multidimensional space by the effects of irreversibility, limited information and learning processes that reduce their mobility (Schumpeter, 1947). In this approach, technological change is the outcome of the creative reaction to the failure to meet the expected levels of aspiration and the mismatch between expectations and actual facts. It is made possible by the continual efforts of accumulation of competence and technological knowledge and the eventual introduction of innovations by existing agents rooted into a well-defined set of scientific, technical, geographic, economic and commercial circumstances. When product and factor market conditions change in unexpected ways, sheer resilience in any given condition engenders actual losses or results below subjective expectations. The constraints imposed by irreversibility and limited knowledge about alternative techniques in the existing range of options reduce the scope for traditional substitution and make it expensive and resource consuming. The search for new routines and new technologies is now activated. Technological change, however, cannot be introduced without appropriate levels of competence and technological knowledge. Technological change is primarily the result of the valorisation and implementation of underlying learning processes, in doing as well as in using and in interacting, that are localised into the specific context of action of each economic agent. Technological change moreover is also influenced by the strategic decision making of agents, which try to maximise their profits and necessarily take into account the product and factor markets into which they are based. Finally and consistently, the rate and the direction of technological change are influenced by the specific set of circumstances, as decision makers, at each point in time, perceive them. The efforts and the outcomes of the introduction of new routines and new technologies are confronted with the opportunity costs of resilience and the costs of switching, i.e. the costs of facing the constraints caused by irreversibility and limited knowledge. The firm will implement its adjustment by means of a mix of technical changes, consisting in movements in the existing space of techniques
Localised technological change 49 and products, and technological and organisational innovations, consisting in the actual modification of the space of techniques and products. The composition of the mix will depend upon the relative costs of technical changes with respect to technological (and organisational) ones. Firms introduce technological change as a creative response to the mismatch between expectations and plain facts: hence technological change is generated in out-of-equilibrium conditions. The larger are the discrepancies between the expectations of each agent and its actual conditions, the faster are the rates of introduction of new technologies. The introduction of new technologies by each agent in turn however engenders new discrepancies between the expectations of any other agent and their actual market conditions. Hence technological change feeds technological change, and out-of-equilibrium conditions further reproduce out-of-equilibrium outcomes. Firms can react to the mismatch between expectations and actual conditions by means of the introduction of localised technological changes only if the specific context of action provides appropriate opportunities for the introduction of new technologies. Localised technological change in turn can engender an increase of total factor productivity levels, with respect to any other technique that belongs to the existing space of techniques, or simply make it possible for firms to be as productive as in the equilibrium technique. The specific contextual conditions, internal to each firm, each region, each industry, each institutional context and each scientific and technological field, play a major role in assessing the actual technological opportunities for each firm. The costs of innovative activities as well as their outcome are highly contextual and contingent on the specific set of circumstances into which the action of firms is embedded.
3 The basic ingredients The notion of localised technological change is the result of the selective merging of quite distinct strands of literature: the notion of bounded and procedural rationality, the induced technological change approach, the economics of learning, the economics of knowledge and the economics of irreversibility. Two characterisations of Homo economicus emerged in the economic literature in the second part of the twentieth century and have an important role in this analysis: the notion of bounded rationality and the prospect theory. The analysis of bounded rationality, the notion of satisfying behaviour and especially the distinction between substantive and procedural rationality introduced by Herbert Simon provide the basic context for the analysis.1 The knowledge of economic agents is characterised by relevant search and information costs: agents do not control all the information about all the techniques available at each point in time on the existing maps of isoquants, all the preferences of consumers and all the strategies implemented by rivals on product markets and suppliers in the factor markets. Agents are localised in a limited space, both synchronically and diachronically, of the wide cognitive map: the notion of cognitive distance plays a major role here. Agents are knowledgeable about a limited portion of the cognitive
50 The ingredients map and experience major limitations in acquiring and processing information about problems and conditions that are far away from their own original location. Agents are able to elaborate appropriate cognitive frameworks about a circumscribed set of conditions that is defined in terms of close proximity to their usual range of actions, routines and problem-solving activities currently carried out. Moreover and quite obviously agents are myopic, for they are unable to foresee all the possible consequences of their actions and cannot anticipate correctly all the possible technologies that any other agent is trying to introduce at each point in time, all the possible changes in the preferences of consumers and all the possible strategic changes of rivals in product markets and suppliers in factor markets. Agents however are able to organise rationally, albeit within a limited number of possible alternatives that do not include all the possible solutions both with respect to the existing conditions and obviously with respect to all the possible future configurations, the sequence of actions when facing changes and alterations in their plans. In these conditions agents do their best to maximise profits and minimise costs: the outcomes of their optimisation efforts however are constrained by their huge cognitive limitations. Optimisation is intrinsically subjective; as such it is localised and leads to local optima and to sequential decision making (Simon, 1982; Rizzello, 1999, 2003; Foster, 2005).2 Finally, agents behave on the basis of their own subjective perceptions of the environment and are especially sensitive to the subjective definition of internal satisfaction. When the levels of aspirations are not realised, agents take into consideration the introduction of innovations. The introduction of innovation is the result of the deception and dissatisfaction of agents that can overcome their reluctance to innovate only in a specific and favourable context of complementary circumstances. This approach highlights the notion of creative reaction as the qualifying aspect of the behaviour of innovative agents: agents innovate when their expectations are deceived and their performances fall below subjective levels of aspiration. Creative reaction is a part of the satisfying behaviour of economic agents afflicted by bounded rationality but able to learn, to generate new knowledge and to modify their conditions. The notion of failure inducement elaborated in the behavioural approach complements and integrates the induced technological change approach (March and Simon, 1958; Antonelli, 1989, 1990). In the localised technological change approach, however, the introduction of innovations is not regarded as the automatic product of learning routines. The introduction of innovation in fact is risky, and agents are reluctant to innovate. Innovative behaviour is solicited and induced by emerging discrepancies between plans and reality when performances fall below the expected levels of satisfying thresholds: the notion of creative reaction is the second basic ingredient. The notion of creative reaction plays a central role in this analysis. Firms are induced to try to change their routines and their technologies when a mismatch takes place between their plans and expectations and their performance: innovation is induced by the failure to achieve the goals of the firm and by the need to cope with unexpected events. Firms facing unexpected events do more that adjusting prices to quantities: firms try to react by means of the intentional introduction of changes in their routines
Localised technological change 51 and their technologies. This approach was first articulated by Nelson and Winter (1973), who contend that a period of losses is required to unseat a firm from looking backwards and taking on the ‘intentional’ exploration of new technologies. This approach has been familiar throughout the management and organisation literatures, but little considered by the economic literature and even by evolutionary economics. The notion of reaction is key to understanding the grafting of complex dynamics into economics of innovation and economics at large. As Brian Arthur notes, ‘common to all studies on complexity are systems with multiple elements adapting or reacting to the patterns these elements create’ (Arthur, 1999: 107, bold added). In complex dynamics, agents are allowed to react to the structural characteristics of the system and in so doing contribute the emergence of new structures, which in turn lead to new changes. Innovation is one of the main forms of such reaction. It can be introduced and lead to successful technological change when a sufficient number of complementary actions and conducive conditions apply (Nelson and Winter, 1982: Ch. 9; Antonelli, 1989). Prospect theory as developed by Kahneman and Tversky (1979, 1992) provides important elements to understanding the process by means of which firms are pushed to innovate. Prospect theory assigns an important role to gains and losses, rather than revenue levels, in assessing the decision making. The reflection effect, a key component of the prospect theory, suggests that risk aversion is strong when gains are considered. Risk-seeking behaviour however emerges and can lead to action when losses are taking place: the more agents are exposed to frustration, the less risk adverse they are and they become more and more ready to accept higher levels of uncertainty. The search for new technologies and the introduction of product, process and organisational innovations are likely to take place, with accelerated rates, when frustration and actual losses are encountered by agents, dissatisfied with the current state of their business. The analysis of Paul David on the effects of irreversibility and of physical and human capital, as well as of reputation and market relations, contributes the understanding of the factors of localisation. Following Paul David, a distinction between irreversibility and quasi-irreversibility can be made. When irreversibility applies, agents cannot overcome the effects and are bound to keep using the production factors in place; they cannot change their location, their customers or their suppliers. Irreversibility takes place when no change can be made to a given context. Irreversibility lies at the heart of traditional, textbook microeconomics: the theory of costs, the theory of the firm and the theory of the market are all shortterm theories. As is well known, the very definition of ‘short-term’ in microeconomics impinges upon the notion of irreversibility. When an amount of capital has been chosen it cannot be changed swiftly. Only in the long term does capital become a flexible input. As a consequence, firms can change the levels of output, but incur a reduction of technical efficiency and an increase in average costs. In turn the positive slope of the supply curve is a direct consequence of the shortterm ‘U’ shape of the average and marginal costs. In turn the theory of the market is fully based upon short-term analysis and hence upon the assumption of ‘rigid’ irreversibility (David, 1975).
52 The ingredients When quasi-irreversibility applies, instead, agents are rooted by an irreversibility which limits their mobility and requires dedicated resources to handle it: constrained adjustment is possible. The introduction of new technologies is induced to overcome the limits of irreversibility. The case of quasi-irreversibility emerges when a given constraint caused by the irreversible elements can be modified by means of creative reaction and the introduction of localised technological changes. The localised technological change approach elaborates an augmented induced innovation mechanism by integrating the economics of learning and the economics of irreversibility. According to the traditional induced technological change approach, the introduction of new technologies is determined by the conditions of factor markets (Antonelli, 2003a). Changes in product markets however also induce technological changes. As a matter of fact all changes in aggregate demand, as in the post-Keynesian tradition, and in the specific product markets of each firm, brought about by Schumpeterian rivalry among firms, induce the creative reaction of firms constrained by irreversibility but endowed with creativity. The separation of these levels of analysis has caused much detriment to the actual understanding of the dynamics of technological change (Momigliano, 1975). The separate appreciation of the role of capitalistic rivalry, or demand pull or factor market inducement, is not sufficient to grasp the actual factors of the continual growth of productivity. The analysis of the inducement mechanism, elaborated in the localised technological change approach, is expanded so as to integrate the changes in both the factor and the product markets. Not only do the changes in the factor markets induce the innovative reaction of firms, but also all the changes in the expected levels of the demand. Firms have made plans and built irreversible production capacities for expected levels of output. When the demand for the products of the firm changes, the firm once again is exposed to switching and information costs. Elaborating on this argument, both the demand-pull analysis and Schumpeterian rivalry become part of the augmented inducement mechanism. Innovative reaction in fact is now induced not only by changes in factor markets, but also by the macroeconomic pressure of aggregate demand, as in the demand-pull tradition of analysis (Schmookler, 1966), and by changes in the demand curve of each firm, determined by the rivalry among firms within each industry, as in the Schumpeterian tradition (Scherer, 1984, 1992). An important ingredient of the localised technological change approach is provided by the critical assessment of the microeconomic implications of the debate upon the classification of technological change and the early economics of growth, at the aggregate level, that took place in the late 1930s (Robinson, 1937; Asimakopulos and Weldon, 1963; Besomi, 1999). The traditional classification of technological change, whether neutral, labour saving or capital saving, had been elaborated in the analysis of economic growth within a single and homogeneous economic system. In a different context, one where many different firms compete in global product markets and have access to different factor markets, the application of that classification of technological change yields surprising results. When technological change is either Harrod neutral or Hicks neutral, i.e. it can be
Localised technological change 53 expressed by a generalised shift of the map of isoquants towards the origin, the new technology is always better than the previous one: each technique of the new map is in fact more efficient than the previous. Not casually, neutrality is assumed as the standard hypothesis in macroeconomic modelling (David, 2004c: 41). It is clear moreover that the traditional assumptions about the neutrality of technological change, elaborated at the macroeconomic level, have little, if any, meaning at all at the microeconomic level. At the firm and industry level it seems in fact quite obvious that technological change is strongly biased. When technological change is Hicks biased, so that its geometric representation consists in a new map of isoquants with a change in their slopes, the new isoquants are likely to intersect the old ones. This is especially clear when technological change is incremental. The intersection is most relevant when the new technology consists only in the introduction of a bias, with no shift effects. When an isoquant that belongs to the new technology intersects the equivalent isoquant that belongs to the old one, technological change no longer consists of new techniques that are ‘always’ superior to the old ones. It becomes clear, instead, that the new technology is only locally superior: it is superior for some techniques and inferior for others: technological change is local.3 Firms are viewed as learning agents, which do not limit the scope of their action to adjusting prices to quantities and vice versa. They are also able to try to change intentionally and purposely their technology, as well as their strategies, building upon the competence acquired by means of learning processes. From this viewpoint the resource-based theory of the firm, as elaborated upon the basic intuition of Edith Penrose, provides basic guidance (Penrose, 1959). A clear line of continuity can be found when the contributions of Edith Penrose (1959), Kenneth Arrow (1962b) and Anthony Atkinson and Joseph Stiglitz (1969) are considered together as sequential steps in a process of specification and incremental enrichment of the identification and analysis of the implications of the economics of learning. The economics of learning makes a major contribution to understanding the dynamics of localised technological change. Here the basic building block is provided by the Arrovian analysis of learning as the key factor in the increase of efficiency. New technologies are, mainly, the result of learning processes that consist in the accumulation of experience and tacit knowledge and are strictly defined and circumscribed by the technical context of activity. Agents learn by doing for well-defined products and by using well-defined machines. Learning is inherently localised in a narrow technical context (Arrow, 1962b). Edith Penrose (1959) had anticipated the Arrovian notion of learning and qualified the firm, its organisation and its routines, as the privileged actor in the learning process. The firm precedes the production function, as its primary activity consists in the generation of new technological knowledge. Each firm, as is well known, learns and builds up new capabilities and eventually discovers new possible applications for production factors and competences that are found within its own boundaries. According to Edith Penrose, in other words, innovative firms are successful when they try to make the most effective use of production factors that are not only locally abundant, but also internally – within their own boundaries –
54 The ingredients abundant. The bottom-up approach to understanding the dynamics of knowledge finds here the first input and in so doing it stresses the role of technological knowledge, acquired by means of localised learning processes, as the primary input in the generation of new knowledge at large, together with the scientific advances made available by the scientific community and the acquisition of external knowledge spilling in the atmosphere. The analysis of learning has been subsequently stretched and sharpened by the insight of Anthony Atkinson and Joseph Stiglitz (1969), who elaborated further upon the key role of learning in the generation of new technologies and introduced the strong hypothesis that technological change can take place only in a limited technical space, defined in terms of factor intensity. Technological change is localised because it has limited externalities and affects only a limited span of the techniques, contained by a given isoquant, that are identified by the actual context of learning. In other words, technological change can only take place where firms have been able to learn: the localisation here is strictly defined in terms of factor intensity and with respect to the techniques in place at each point in time. In the analyses of Penrose and Atkinson and Stiglitz, technological change is localised and constrained by organisational routines, but it is the automatic result of learning without any intentional and explicit effort. The inducement context, characteristic of the localised technological change approach, makes it possible to overcome this major limitation. The understanding of the specific context in which learning takes place is necessary to grasp the foundations of the intentional action of the generation of new technological knowledge and the introduction of new technologies. The economics of learning contributes to the economics of knowledge and paves the way to understanding a broader notion of localised technological knowledge that includes also the effects of proximity in geographic and technological space. The notion of learning and localised technological knowledge in fact makes it possible to stress the role of knowledge as a joint product of the economic and production activity. Agents learn how, when, where and what, mainly from their experience, accumulated in daily routines. The introduction of new technologies is heavily constrained by the amount of competence and experience accumulated by means of learning processes in specific technical and contextual procedures. Agents, in this approach, can generate new knowledge only in limited domains and fields where they have accumulated sufficient levels of competence and experience. A strong complementarity must be assumed between learning as a knowledge input and other knowledge inputs, such as R&D laboratories, within each firm. Technological knowledge is the result of the intentional valorisation of potential competence based upon learning: such valorisation requires other knowledge inputs, not only internal R&D but also external knowledge (Arrow, 1962b, 1969; Lamberton, 1971; Loasby, 1999). A second and most important complementarity takes place in the localised technological knowledge approach between internal and external knowledge. Firms can generate new knowledge and hence eventually introduce new technologies only when and if they are able to take advantage of external knowledge. No firm can
Localised technological change 55 rely exclusively on its own internal knowledge, either tacit or codified, whether it is the result of learning processes or of formal research and development activities. The notion of knowledge cumulability and complementarity between external and internal knowledge is central in the understanding of the localised technological knowledge (Antonelli, 1999a, 2001, 2003a). The relationship between external and internal knowledge becomes a key issue. Neither can firms generate new knowledge relying only on external or internal knowledge as input. With an appropriate ratio of internal to external knowledge, internal knowledge and external knowledge inputs enter into a multiplicative production function. Both below and above the threshold of the appropriate combination of the complementary inputs the firm cannot innovate. Innovation is not the automatic result of the spontaneous accumulation of learning and competence, but the result of intentional decision making that takes place when firms are faced with unexpected events. Innovation is the possible result of the creative reaction of firms that can take place only when the surrounding environment is conducive to favouring the introduction of new technologies (Schumpeter, 1947). The legacy of Hayek plays here a key role and makes it possible to articulate the analysis of the contribution of external knowledge in the generation of new knowledge. The retrieval of two distinctive features of the Austrian tradition, such as the notions of dispersed and fragmented knowledge as a key characteristic of economic systems, even in static conditions where technological change is not considered, and the intrinsic complementarity of the bits of dispersed and yet indivisible knowledge possessed by each agent contribute to this line of analysis. Following Hayek (1937, 1945, 1952) it seems clear that no firm can command all the knowledge that is necessary in order to generate new knowledge. Technological knowledge is now viewed as dispersed and fragmented into a variety of complementary and yet specific and idiosyncratic applications and contexts. The localisation into technological commons, such as technological districts, professional communities, industrial sectors and filieres, where other firms carry on complementary innovation and share substantial portions of the relevant technological knowledge, is a key factor in assessing the actual innovative capability of firms. Relevant non-market interactions take place and affect the access to the knowledge commons (Gleaser and Scheinkman, 2000). Proximity in geographical and knowledge space among learning agents able to react to the failure of their aspiration levels, by means of the shared access to and participation in a collective knowledge base and the eventual introduction of complementary innovations, is an essential condition to activate the failure inducement mechanism of technological change, to overcome the reluctance to change and to convert the isolated reaction of dispersed agents to adverse market conditions into the collective introduction of systemic innovations. In less favourable contexts, firms are obliged to face the discrepancy between expectations and actual conditions just by means of technical adjustments, bearing all the costs engendered by irreversibility and limited knowledge. Technological change is slower, and the gaps between expectations and actual conditions are smaller: the system can converge towards more stable and static equilibria with lower opportunities for growth.
56 The ingredients Innovation is a highly contextual outcome, conditional on the occurrence of a large number of necessary conditions. The economics of knowledge and more specifically the economics of knowledge governance here provide basic insights about the systemic conditions that favour the actual accumulation of knowledge, its role as a condition for creative reaction to take place and its use for the eventual introduction of new technologies. The conditions for the generation, use and dissemination of technological knowledge can be grasped only at the system level. Technological knowledge in fact is not a standard economic good for the pervasive role of indivisibility, cumulability and inappropriability. A variety of governance mechanisms however can be implemented so as to combine at the same time the incentives to its generation and the conditions for its dissemination. Localised technological change combines the inducement mechanism with the economic implications of learning and irreversibility in a unique analytical system based upon the Schumpeterian notion of creative reaction. Firms are characterised both by learning capabilities and by bounded rationality and limited knowledge. Necessarily, firms make plans and consequently decide actions, whose effects are partly irreversible. All discrepancies between expected market conditions, now including both factor and product markets, and planned decisions should engender a creative reaction that consists in the introduction of technical changes, that is changes in the existing space of techniques, consisting in either substitutions on a given isoquant or changes from one isoquant to another, or both (Schumpeter, 1947). All changes in the existing space of techniques, however, engender specific costs, owing to the irreversibility of the production factors as well as to the information costs that are necessary in order to operate the new desired techniques. Switching costs prevent standard adjustments realised by means of substitution or sheer change in input levels. This is the context into which the creative reaction takes place. Dedicated and intentional efforts to valorise localised learning and external knowledge locally available provide the opportunity to react to the emerging mismatches between plans and expectations and actual economic facts and to introduce technological changes, albeit in a narrow technical space. Firms exposed to the discrepancy between plans and actual market conditions, bounded in their mobility by limited knowledge and irreversibility, but able to rely upon procedural rationality, are induced to take advantage of the localised knowledge accumulated by means of learning processes, relying upon the local pool of external knowledge made available by appropriate knowledge governance mechanisms and good information channels, and to introduce technological innovations in a limited technical space, so as to overcome the constraints raised by factor irreversibility. Localised technological change is endless and fully endogenous. Firms cannot anticipate all the possible innovations introduced by any other firm in the economic system. And yet any discrepancy between plans and actual market conditions is likely to induce the localised introduction of new technologies, which in turn are the cause of new discrepancy between expectations and actual market conditions for other firms.
Localised technological change 57
4 The multidimensional scope Localised technological change reflects the pervasive role of irreversibility, externalities, information asymmetries and bounded rationality and interdependence, as well as the amount of creativity each agent is able to express as a way to solve specific and contextual problems arising in the daily management of its business. Hence technological change is necessarily localised in a multidimensional space that is deeply rooted into the context of characteristics which define the activity of each agent. Technological change is localised in historical time, in technical space, in the knowledge space, in technological systems, in the structural conditions of each economic system, in geographic space and in the space of product characteristics; technological change is localised in firms. The analysis has investigated the variety of processes by means of which technological change is localised in the historical, technical, technological, structural, institutional, regional, knowledge and organisational spaces, highlighting how and why the introduction of innovation is conditional on the effects of proximity. Localised technological change is primarily localised into historical time. Each technological innovation and each element of technological knowledge and competence can be understood only as a step in a historical sequence of the cumulative introduction of technological innovations and other bits of technological knowledge. Technological change is characterised by path dependence in that it can be analysed effectively only when the effects of cumulability and irreversibility are put in context. Cumulability plays a key role in the production of knowledge and in the integration of the new production factors into the existing production process. Irreversibility is an essential characteristic of fixed capital, both tangible and intangible. The fixed and irreversible capital can be changed only at a cost, and this affects the scope of any further innovative choice. The introduction of new technologies that are complementary to the existing ones becomes a clear constraint and incentive (Antonelli, 1999a, 2001). Technological change is localised into technical space, that is the space defined in terms of factor intensity, by the essential role of learning in building the competence and the technological knowledge that is necessary to introduce new technology and increase the efficiency of the production process (Atkinson and Stiglitz, 1969; David, 1975; Stiglitz, 1987). Learning is essentially localised in a limited technical space and as such it cannot be applied easily elsewhere. Antonelli (1992, 1995) has further developed this notion of localised technological change, emphasising the role of irreversibilities in fixed and immaterial capital stocks and the related switching costs and coupling its effects with the local dimension of learning originally highlighted by Atkinson and Stiglitz (1969) and further stressed by Stiglitz (1987). As a result it seems clear that technological innovations are introduced within technical corridors identified by the original technical localisation of innovating agents and defined by barriers to mobility originated by switching costs and learning opportunities. Technological change is localised into the gales of technological systems, activated by technological indivisibilities, complementarities and interdependencies
58 The ingredients among technologies. The efficiency of each technology is greatly enhanced by the availability of the other complementary and interdependent technologies. Firms induced to innovate are pushed to direct their innovative efforts towards the introduction of new technologies that are complementary to others so as to take advantage of typical network externalities in their dynamic form (David, 1987; David and Bunn, 1988; Antonelli, 2001). Technological change is localised into knowledge space. High levels of vertical and horizontal indivisibility characterise knowledge space. Each unit of knowledge has high levels of complementarity and cumulability with other units that are ordered vertically across historical time. Vertical upstream complementarity across different fields of origin is also relevant and can be defined in terms of compositeness. Finally, each unit of technological knowledge can be used in different fields of application: the notion of fungibility is important in this context. The production of knowledge relies on the continual recombination of existing bits of knowledge. The characteristics of such indivisibilities are most relevant and make it possible to identify the commons of knowledge. The boundaries of such commons of specific knowledge and competence become a powerful factor of specialisation into well-defined technological fields. Good access conditions to the knowledge commons and good communication channels among learning agents make easier for firms the introduction of innovations as a response to unexpected declines in performances (Weitzman, 1996; Antonelli, 2003b). Technological change is localised into geographic space by the effects of the proximity between learning agents, because of the pervasive role of scientific communication and technological spillover. Proximity in geographic space interacts with proximity in knowledge space. Regional proximity favours the generation of new knowledge on three counts: 1) it helps in reducing knowledge transaction costs; 2) it facilitates the division of scientific and technological labour and hence increases the efficiency of the generation of new knowledge; and 3) it makes it possible to accelerate the circulation of knowledge and hence to capitalise on knowledge externalities. Proximity in geographic space helps in increasing trust among trade partners and reduces the scope for opportunistic behaviour. Consequently proximity favours the emergence of local markets for technological knowledge, qualified systems of knowledge interactions, higher levels of specialisation and division of labour in the generation of new knowledge, and hence higher levels of general efficiency in the generation of new knowledge and in the introduction of technological innovations. The localisation into a technological district and membership of professional communities make easier the access to local knowledge commons and hence increase the effects of local knowledge externalities and increase the probability of the introduction of successful innovations. The quality of the local scientific infrastructure and the connectivity of the communication channels in place between the academic and the business community add strong elements to understanding the key role of technological districts in localising technological change from a geographical viewpoint (Antonelli, 2001).
Localised technological change 59 Technological change is localised in the economic structure of each economic system by local endowments, intermediary markets and hence factor costs. The structure of relative prices reflects the local endowments and the vertical organisation of the economic system. It reflects the vertical relations among industries along the ‘filieres’ within the input–output matrix. The characters and types of market structures in each given layer have powerful effects downstream in terms of relative factor prices. The effects of technological innovations vary according to the interplay between the direction of technological change, defined in terms of the marginal efficiency of each production factor, and the local structure of relative factor prices. The composition effect, that is the consequences of relative factor prices for each possible direction of technological change, has powerful consequences in terms of total factor productivity growth. The endowments of each region and the structure of relative prices within each industrial system become a powerful factor in explaining the differentiated effects of the introduction of the same technology across economic systems. Composition effects may account for the delays in adoption of incremental and biased technologies. By the same token, however, composition effects are a powerful inducement mechanism to selecting and focusing the factor intensity of the new technologies. Firms located in a labourabundant region have a clear incentive to introduce labour-intensive technologies, for they make it possible to make the most intensive use of the more abundant and hence cheaper local production factors. The analysis of localised technological change has shown that changes in the levels of factor price can be considered the prime mechanisms of inducement, and hence the determinant of the rate of change. The direction of the new technologies being introduced however is determined by the relative abundance of the production factors in the local factor markets (Antonelli, 2003a). The analytical framework introduced by Lancaster (1971) proves to be especially fertile and productive in accommodating the analysis of the innovation process in the space of product characteristics. The Lancastrian approach in fact can be easily used as a tool to stretch the localised technological change approach to analyse the role of proximity in the space of product characteristics and assess the choice between product and process innovations. It seems clear that product innovation is more localised the more specific and localised the process of accumulation of competence is and the more relevant the latter is in the generation of technological knowledge and the more dispersed the distribution of firms in the product space (see Chapter 16). Finally, it is clear that technological change is localised within firms. Firms differ in many relevant ways: the vintage of irreversible factors, the competitive context, the factor markets, the location and the communication channels in place with the external environment, the organisation and the structure of decision making, the learning procedures, the portfolio of products and the knowledge fields where competence is based, and the composition of human capital. Each firm as a consequence follows its own path in reacting to the mismatches and introducing technological innovations. Elaborating on the legacy of Edith Penrose and Ronald Coase, it seems clear that not only does the dynamics of accumulation of
60 The ingredients technological knowledge matter, but also the dynamics of competence in managing the governance mechanisms, including both transaction and coordination activities, as determined by the generation of organisational knowledge and the introduction of organisational innovations, play an important role. Governance mechanisms are no longer viewed as the result of the static combination defined by given levels of coordination and transaction costs, but as the result of dedicated activities affected by the rates of accumulation of competence and organisational knowledge. Next to the production function a corporate function where alternative governance mechanisms are considered and assessed needs to be used to analyse the firm. The interplay between technological and organisational innovations, based upon technological and organisational learning respectively, shapes the growth path of the firm and defines the sequence of technological innovations that each firm can generate and introduce successfully (Antonelli, 1999b, 2001: 111–45). The logic of localised technological change applies to understanding the diffusion and the selection of new technologies as well. Each new technology is localised and as such reflects the specific conditions of innovators, and each firm is localised in its specific context of action. The factors of localisation in fact help in explaining why firms may delay the adoption of some technologies. Firms will select the new technologies that fit better with their own highly specific and idiosyncratic context of operation. New technologies may happen to be superior for some firms and in some circumstances and inferior for others (David, 1985).
5 Complex system dynamics and path dependence Localised technological change, as determined and shaped by irreversibility, is inherently path dependent. New technologies must be introduced in order to cope with the discrepancy between plans and actual market conditions to which irreversibility prevents adjustment, by means of standard substitution. Yet they can change the course of actions, modifying the effects of irreversibility, although within a narrow and limited space of alternatives, defined by the effects of switching costs and learning. The distinction between past dependence and path dependence is crucial in this context. Irreversibility is a source of past dependence if no action may modify and integrate the effects of irreversibility. Irreversibility engenders path dependence when and if specific and intentional actions take place and modify the course of sequential events, albeit in a narrow and limited region. Irreversibility is the cause of technological change because switching costs limit the possibility of firms reacting to changes in their markets by means of traditional substitution on existing isoquants. The introduction of innovation is necessary in order to adjust to the new market conditions and yet save on switching costs. Irreversibility is also at the origin of the localised introduction of innovations because it shapes the corridors of introduction of the new technologies and prevents the radical – chaotic – change of technical coefficients. The single attractor of the general equilibrium analysis no longer applies to studying a system where agents are able to change endogenously their production
Localised technological change 61 and utility functions. Evolutionary biology provides important inputs to understanding the working of system dynamics when agents can change their capability, behaviour and conduct in response to the unexpected changes in the conditions of their habitat, building upon the competence and the irreversibility conditions defined at each point in time by their past. The analysis of the working of cumulative selection, as opposed to one-shot selection, articulated by Richard Dawkins (1986), seems especially attractive and consistent with the analytical framework emerging from the analysis of localised technological change and its system implications in terms of path dependence. The contributions of the economics of knowledge and the new understanding about the role of external knowledge and knowledge communication in the generation of new knowledge favour a closer relationship between the analysis of innovation and the analysis of diffusion. The adoption of a new technology and the generation of new knowledge that builds upon the access to external knowledge share many elements. The pervasive role of knowledge externalities pushes to articulate the notion of the diffusion of innovative activity. The implications of the economics of localised technological change for the dynamics of the system are most important. The system is inherently complex and dynamic. There is not such a thing as a stable attractor towards which the elements of the system are eventually induced to converge. Endogenous technological change in fact is the vector of continual change and development. When the dynamics of technological change is based upon ubiquitous and automatic processes of learning, as is the case in the new growth theory, a clear sequence of attractor points can be identified and even predicted. When the variety of agents is assumed, creativity matters, and its effects are not based upon automatic and ubiquitous learning processes; however, the dynamics of the system risks becoming unpredictable and unintelligible: system dynamics acquires all the characteristics of a chaotic process. The mix of quasi-irreversibility, local externalities and creativity, upon which the economics of localised technological change is based, is crucial in this context. Localised technological change provides to system dynamics elements of intelligibility and even limited predictability. Irreversibility and local externalities in fact characterise the likelihood that creativity will take place and shape its effects. Irreversibility pushes creativity, as an extreme form of reactivity. Local externalities and in general the local context of learning and interaction shape the actual scope of creativity and qualify its effects. Local externalities assess the effects of creativity, whether it consists of total factor productivity effects or simply of technical changes that do not increase overall efficiency levels. Irreversibility and local externalities define the context into which creativity takes place and exerts its effects. The economics of localised technological change and path dependence provides the economics of complex dynamics with new tools to understand, if not predict, economic development (Taylor, 2001). At the system levels the dynamics exerted by the mix of irreversibility, local externalities and creativity acquires all the characteristics of path dependence. Path dependence is the relevant analytical result, at the aggregate level, of the dynamics generated by localised technological change at the microeconomic level. Creativity
62 The ingredients and endogenous change in fact take place because of the effects of irreversibility and bear the characteristics of the local context of learning and interacting into which the action of a variety of agents takes place. Path dependence is a form of constrained chaotic behaviour where the past defines at each point in time the basic ingredients of change and hence its direction.
6 Conclusion Localised technological change integrates a variety of complementary approaches so as to provide a powerful analytical framework able to accommodate into a rational procedural explanation the introduction of the unknown and of the surprise. Firms are reluctant to change their behaviour and their routines: innovation is difficult and risky. Firms are induced to introduce new technologies when a number of conditions are met: when satisfying thresholds are not met or the levels of aspiration are not realised, because of a mismatch between plans and product and factor markets; when limited knowledge and irreversibility matter and limit the scope of substitution and technical change within the existing space of techniques; when learning and accumulated competence provide the opportunity to generate new technological knowledge, albeit in a limited technological space; when knowledge interactions and knowledge spillovers, within limited regional spaces, make external knowledge available and hence innovative activities possible and more productive; when technological complementarities with other parallel new technologies increase the stream of benefits stemming from the introduction of a new technology within a technological system; and when the distribution of rival products and customers’ preferences favour the introduction of product innovations, albeit in the proximity of existing product lines. Technological innovations are introduced when a number of forces are at play and a highly qualified set of sequential conditions favour the positive outcome. Thus technological change is the conditional and unpredictable result of an evolving systemic context of opportunities and constraints shaped by the intentional capability of agents to change their location and the architecture of their relations. Technological change is endogenous to the working of the economic system, as it is induced by all changes in factors and product markets that myopic firms are unable to foresee, and localised by the effects of irreversibility and limited knowledge and learning. Localised technological change is the result of out-ofequilibrium conditions and is the cause of out-of-equilibrium conditions. In such conditions, convergence towards a stable equilibrium point may take place, but it is likely to be quite a contingent outcome. Opposed to this, an endless and pathdependent process of endogenous change that consists in both the rate and the direction of both technological and structural change takes place, as is the result of the interplay filtered and shaped by the contingent architecture of the relations of intentional, creative and yet myopic agents.
Part II
The governance of localised technological knowledge
4
Information economics for the economics of localised technological knowledge
1 Introduction In the last 50 years the economic role of knowledge has emerged and progressively gained a central place in the arena of the economic debate. Knowledge is indeed the primary resource into all human activity. Its identification as an economic good, however, has needed a long time and enduring efforts. The economics of knowledge has gradually emerged as a discipline in a context characterised by a sharp evolution of the analysis as well as of the basic foundations. Such an evolution has made it possible to increase substantially our understanding of the economic characteristics of the generation and use of knowledge in the economic systems. Its evolution is the consequence and the cause of much change in the economic understanding of technological change and more generally in the new thinking about economic growth. The great ingenuity of economics rests in the asserted coherence between profit maximisation, at the agent level and social welfare, at the system level. When markets are competitive, returns are constant and all products are economic goods, profit-seeking and profit-maximising agents, interacting in the marketplace exclusively by means of fully fledged transactions, are able to identify the best combination and hence to generate the maximum amount of social welfare. Markets can perform such functions successfully only in a context where profit maximisation and maximum social welfare are jointly achieved. In turn, as is well known, this can be achieved only if appropriate information is available regarding the technological competence of the respective partners and the future paths of technological change and technological knowledge are perfectly known to everybody. As is well known, only future prices make it possible to solve the problems of dynamic coordination. When a vector of future prices is available for all products, agents can identify the correct amount of resources to invest in each activity, and the effects of the trajectories of demand and future entry and exit can be assessed. When future markets and future prices are available, the market is able to perform properly its basic function of dynamically coordinating the expectations and behaviour of a variety of agents; hence it is unable to provide the indispensable consistency necessary for the long-term allocation of the resources. The production and circulation of knowledge do not match these conditions. High levels of uncertainty characterise the generation of knowledge. It is very
66 Governance of localised technological knowledge difficult to anticipate the outcome of a research process, its duration, its actual economic value and even its specific content. Serendipity plays a crucial role: it is difficult to organise the division of creative labour both within and among firms. It is clear that social support is necessary to sustain the early phases of the generation of new scientific activities and the provision of general knowledge that is fungible, i.e. relevant to a wide variety of knowledge-generating activities. In this context, it becomes evident that agents differ greatly in their capability to generate, use and understand technological knowledge: knowledge asymmetries are intrinsic. Information regarding knowledge is not only asymmetric, but also intrinsically limited: opportunistic behaviour and bounded rationality are not exceptions but rules. Moreover, different levels of uncertainty can be found, and hence there are different levels of complexity in the solution of the typical problems raised by information and knowledge asymmetries, owing to knowledge transaction costs and principal–agent relationships in the generation of new knowledge. There is a strong divide between ex post and ex ante mechanisms of compensation. The identification of each bit of complementary and useful knowledge as well as the interaction with the agents possessing specific bits of knowledge and the assessment of whether they are complementary or not, with respect to their present and future needs and opportunities, the correct definition of the flows of entry into new knowledge modules and the exit from declining ones, the proper combination of the incentives to invest in the generation of new knowledge, and the incentives to disseminate and use external knowledge are all essential functions that the spontaneous governance mechanisms in place at present perform poorly. When increasing returns associated with knowledge indivisibility, as articulated in terms of cumulability, compositeness and fungibility, are at play, especially within the modules that characterise technological systems, the market mechanism does not necessarily lead to achieving dynamic efficiency. When there are increasing returns, the case for dynamic market failure emerges. The dual role of knowledge as an output and an input both in the production of other goods and in the generation of new knowledge amplifies such problems. Because it becomes clear that nobody can command the full range of knowledge that is necessary to generate new knowledge, internal and external knowledge are now complementary inputs in the production of new knowledge. Especially if the latter is specified in terms of a multiplicative relationship, the aggregate outcome of both market transactions and interactions is unstable and sensitive to interactions and subjective decision making. When both demand and supply schedules are influenced by externalities, multiple equilibria exist. The amount of knowledge each firm can generate depends upon the amount of external knowledge available, that is to say, upon the amount of knowledge that other firms, especially when involved in complementary research projects, have generated and cannot appropriate or are willing to share. An iterative dynamic process is at work with no stable attractors: both negative and positive self-reinforcing mechanisms can take place. The availability of external knowledge is crucial. Inclusion needs to be coordinated and managed. Free riding can take place, although reciprocity and mutuality in interactions based upon knowledge barters, implemented by repeated
Information economics 67 and long-lasting exchanges, can help to reduce its extent and its effect. Exclusion is dangerous because of the risks of missing the relevant complementary input, which characterises the generation of new technologies. Multiple equilibria and micro–macro feedback affect the workings of bureaucratic coordination, networking interactions and transactions in the markets for technological knowledge and their outcome. The dynamic coordination of agents plays a central role in this context. There is a gap between the results of the maximisation of profits and the conditions for the maximum social welfare. The governance mechanisms in place appear to provide a set of incentives that may or may not lead the system towards stable and fair solutions. Tradability is a necessary, but not sufficient, condition for dynamic efficiency to be achieved; bureaucratic coordination and networking do not ensure that profit maximisation coincides with social welfare. The aggregate outcomes of the governance mechanisms at the firm level are far from being attracted to a single equilibrium point (Nelson, 2005). In such a context, where the organisation of the generation and use of knowledge are affected by a variety of economic problems, such as transaction costs, agency costs and networking and communication costs, the identification, creation and adaptation of appropriate institutions for the governance of the generation and dissemination of knowledge are more and more necessary to solve the key problems related to correctly identifying the appropriate incentive mechanisms and selecting the areas in which to invest new resources, so as to increase the amount of knowledge available in an economic system. A new step along this line of inquiry can be made with the full appreciation of the localised character of technological knowledge and the understanding of the implications of the key role that localised learning processes play in this context. The notion of localised technological knowledge in fact makes it possible to stress the role of knowledge as a joint product of the economic and production activity. Agents learn how, when, where and what, mainly out of their experience, accumulated in daily routines. The introduction of new technologies is heavily constrained by the amount of competence and experience accumulated by means of learning processes in specific technical and contextual procedures (Antonelli, 1999a). Agents, in this approach, can generate new knowledge only in limited domains and fields where they have accumulated sufficient levels of competence and experience. Because of substantial indivisibility of knowledge, however, no agent can generate new knowledge in isolation: external knowledge, invented elsewhere, is strictly necessary. Once again a strict complementarity must be assumed between learning, as a knowledge input, and other knowledge inputs, either internal, such as R&D laboratories, or external. In the approach to localised technological change as a process of creative reaction elaborated in this book, technological knowledge is the primary condition for creative reaction to take place successfully. When firms have access to technological knowledge and can generate it easily, the emerging mismatches, between plans and expectations on one hand and actual performance on the other, can be matched by means of the effective introduction of new technological innovations. The cost of
68 Governance of localised technological knowledge technological innovation is lower than the costs of switching. When instead the access to new technological knowledge is difficult and hence the generation of new knowledge costly or impossible, firms rely on more traditional switching activities and cannot feed successfully appropriate processes of creative reaction. Technological knowledge takes different forms, is the result of different processes, exhibits its powerful effects in a wide variety of contexts and is highly sensitive to a number of key conditions. The identification of its different forms and characteristics and their systemic assessment into a single frame is the primary task of the economics of knowledge as a discipline and competence, within the broader context of economics. A variety of knowledge governance mechanisms has been spreading in the last few decades. The boundaries between the public research system and the rest of the economic systems, and between markets and organisations, changed substantially in the last decades of the twentieth century. The interpenetrating of academic and private research has grown substantially, often reversing the traditional relations. More and more universities and other public research centres patent the results of their research activities and provide knowledge-intensive services to private firms on a contractual basis. They have entered the market for knowledge outsourcing. Corporations keep funding R&D activities but perform a decreasing share of research activities in intra muros laboratories and outsource them to academic research centres. Public research laboratories, including academic ones, become more and more the locus for research activities, although the public funds a much smaller part of it. Within the business sector a variety of new hybrid forms of knowledge governance far away from pure market transactions and pure hierarchical coordination has been emerging. A wide range of alternative mechanisms has been elaborated as tools to organise the acquisition, the generation, the economic exploitation and the use of technological knowledge. Traditional dichotomies between the State, the corporation and the market are less and less useful. Within the business sector a variety of new forms of coordinated transactions in interdependent types of markets and types of organisations has been emerging, with important consequences in the organisation of the division of creative labour. The empirical literature in the economics of knowledge has identified a variety of forms of knowledge governance based upon hybrid forms of coordinated transactions, ranging from long-term contracts to quasi-vertical integration, from epistemic communities to technological districts, and from patent thicketing to venture capital. New evidence about the close interdependence between the different modes of knowledge governance has grown. Technological knowledge is the object of a variety of forms of interactions and transactions in different contexts and different yet related markets, from the markets for products using new knowledge to the markets for property rights, from barter relations to trust-based interactions where reciprocity matters, and to dedicated financial markets. The notion of knowledge governance, defined as the set of institutions, corporate strategies, types of transactions and forms of interactions that characterise and shape
Information economics 69 the organisation of knowledge production, exchange and usage in the business sector, is the result of this inquiry. The rest of this chapter provides the basic inputs to elaborate such a framework. Section 2 introduces the new understanding about the characteristics of knowledge, as a localised and collective activity. Section 3 explores the scope for the application of information economics to the economics of knowledge and provides the basic tools. The conclusion summarises the main results and puts them in perspective.
2 Localised technological knowledge as an interactive and collective activity The notion of localised technological knowledge stresses the role of the dynamic complementarity of agents embedded in their specific and idiosyncratic conditions of learning and interaction in the generation and exploitation of knowledge as a collective activity. The generation of localised technological knowledge is viewed as the result of the emerging complementarity of a myriad of learning agents, each possessing a bit of knowledge. In turn high levels of indivisibility, which takes the forms of knowledge complementarity, knowledge cumulability and knowledge compositeness, characterise such knowledge. The access to external knowledge is conditional on the successful generation of new knowledge for each agent. At each point in time agents organise their knowledge exploration strategies in order to increase their chances of accessing the pools of external knowledge: agents build communication channels and elaborate interaction strategies that are finalised to increase their chances of acquiring external knowledge. Accordingly, learning agents try to focus the direction of their research activities so as to enhance the compatibility, interactivity and hence complementarity between accessible external knowledge and internal knowledge. At the same time agents try to increase the appropriability of the knowledge they have helped to generate so as to increase the valorisation of their knowledge. The new understanding about technological knowledge as a localised, interactive, collective and heterogeneous process calls for a specific effort to identify and frame into an integrated approach the broad variety of knowledge governance mechanisms that contribute the coordination in the production and usage of knowledge. The notion of localised technological knowledge provides the basic elements to elaborate an integrated framework, able to articulate a single logic underneath such a variety of modes of organising the generation and dissemination of new technological knowledge. The new approach to technological knowledge, as a localised process, stresses the dynamic complementarity among a variety of agents participating in the generation of knowledge as a collective activity: such an intertwined procedural action of a myriad of heterogeneous agents requires substantial levels of coordination in order to succeed. Learning indeed is one of the basic sources of new technological knowledge. As such it exerts a strong and clear effect in terms of a definition of the cognitive space into which each firm can expand its current technological base. As a
70 Governance of localised technological knowledge consequence the new technological knowledge generated by each firm is constrained within the proximity of its current activities. In other words, learning exerts a powerful localising effect, which limits the spectrum of possible discoveries. At the same time however the generation of new knowledge can take a wide variety of possible directions, leading towards the introduction of either product or process innovations, impinging upon the specific form of learning that is actively implemented. Learning, by definition, takes place while performing the current activities, which in turn are highly differentiated according to whether firms rely more on learning by doing, learning by using or learning by interacting with customers or suppliers, and in each of the many specific sub-activities which are contained within the current boundaries of each firm at each point in time. Here the complementarity between the theory of the firm developed by Edith Penrose and the notion of localised technological change originally articulated by Joseph Stiglitz and Anthony Atkinson and Paul David respectively in the context of the theory of production and the early economics of innovation and technological change is most clear. A number of important qualifications however are necessary in order to build upon the overlapping between these two strands of economic theory. Learning in fact is a necessary but not sufficient condition for the generation of new knowledge. The resource-based theory of the firm moreover presents learning as an automatic process, the joint product of current activities, and assimilates knowledge to learning. Little attention is given to the context in which learning takes place and especially to the analysis of the conditions and motivations: learning seems to coincide with new knowledge. The appreciation of the role of intentional decision making in the generation of new knowledge, and specifically the identification of the inducement mechanism that obliges firms to actually generate new knowledge, has been already much debated. This is the first major point of departure from the notion of knowledge as the automatic and spontaneous outcome of learning, put forward by Edith Penrose. Firms are reluctant to change their routines, their production processes, the networks of suppliers and their marketing activities as much as their goals and their understanding of the product and factor markets. Firms can overcome their intrinsic inertia and resistance to change only when a powerful failure mechanism is at work. Firms are pushed to take advantage of the tacit knowledge acquired by means of learning processes by emerging mismatches between their own beliefs, based upon perceptions, and related plans and the actual conditions of the markets for products and production factors.1 Only when such a mismatch takes place and the quasiirreversibility of decisions implemented impedes simple adjustments are firms pushed, by emerging losses and performances below expected levels, to react creatively by means of the introduction of innovations. To do so, the intentional and explicit generation of new technological and organisational knowledge becomes necessary (Ruttan, 1997, 2001; Antonelli, 1999a, 2001, 2003a). Recent advances in cognitive economics confirm the role of intentional creativity in the generation of new knowledge and the specific behavioural context into which discoveries take place (Rizzello, 2003). As Nooteboom (2003: 225) properly articulates:
Information economics 71 discovery is guided by motive, opportunity and means. One needs an accumulation of unsatisfactory performance to generate motive; to overcome one’s own inertia or that of others in organization. In markets, one also needs an opportunity of demand and/or technology. And one needs insights into what source and how to incorporate them in present competence.2 The transformation of the competence based upon learning processes into new, actual technological knowledge requires specific and dedicated efforts. The generation of new technological knowledge can be considered the specific activity of the firm and its distinctive function within the economic system: the firm is indeed the locus of technological discovery. Yet discovery and creativity are not an automatic, incremental, past-dependent and hence deterministic activity guided by the sheer accumulation of competence based upon tacit learning, but rather the result of a complex path-dependent process where at each point in time firms make explicit and intentional efforts to generate new technological knowledge. Such efforts are most likely to be successful when a number of contextual conditions apply. The generation of new technological knowledge is seen as the result of four specific activities: learning, socialisation, recombination and research and development, where both the general process and each activity are localised by the effects of bounded rationality and proximity. These activities are complementary and none can be disposed of. A limited substitution can take place among them. In the localised technological knowledge approach, firms generate new knowledge relying on both external and internal knowledge as complementary inputs. Essential inputs of tacit, codified, internal and external knowledge all enter into a multiplicative relationship within the knowledge production function. External knowledge is strictly necessary to generate new knowledge both when it concerns the same module of knowledge and when it belongs to other knowledge modules: intratechnological and inter-technological flows of knowledge are both necessary, in varying levels according to the character of the knowledge being generated, whether mainly analytical or synthetic (Laestadius, 1998). In some cases, substitution between the different knowledge inputs is possible, but only above minimum thresholds. As Figure 4.1 shows, firms can rely on different combinations of the four basic activities that lead to the generation of new knowledge. All combinations include minimum levels of the four basic activities: respectively OA for learning, OB for recombination, OC for R&D and OD for socialisation. Learning is the process of accumulation of tacit knowledge. R&D leads to the generation of codified knowledge. Both learning and R&D are the sources of internal knowledge, as they are conducted within the borders of the firm. External knowledge consists of activities of socialisation and recombination. The former makes possible the acquisition of external knowledge by means of constructed interactions with other learning institutions, including other firms and research institutions such as universities. The latter consists in the formal acquisition of knowledge generated by other agents by means of transactions in the markets for knowledge. The basic
72 Governance of localised technological knowledge
Recombination
B 0 Learning
A
C
R&D
D
Socialisation Figure 4.1 The knowledge generation process
complementarity among the minimum levels provides the basic mix, upon which firms implement their own specific knowledge generation choice. A great variety of knowledge-generating processes can be identified according to the relative weight, beyond the minimum levels, of the activities. The knowledge generation can be more or less intensive of internal (see in Figure 4.1 the area within the solid line) or external activities (see in Figure 4.1 the area within the dotted line) and in turn may privilege learning activities and R&D activities, or socialisation and recombination.3 The economic and institutional conditions of the economic systems into which firms are embedded play a key role in the selection of the mix of knowledge generation activities. The characteristics of knowledge per se, and specifically its levels of fungibility, cumulability and compositeness, also affect the choice of firms. According to the varying combinations of contextual conditions that characterise each agent participating in the process, the system of interactions and the supply of scientific opportunities, technological knowledge is seen as an emergent property of the complex and dynamic system in which agents are embedded. As such it acquires a strong idiosyncratic and heterogeneous character, specific to the circumstances that have characterised its generation and usage. Consequently and consistently, technological knowledge is localised in the cognitive space of agents and in the technological space that defines the proximity between techniques, in the geographical space that shapes the interactions among agents, in the industrial space that characterises the transactions among firms, in the organisational space
Information economics 73 that shapes the learning processes internal to each firm, in the institutional space that defines the relations among firms and between firms and the scientific community, and in the space of product characteristics that qualify the choices of consumers and their learning procedures. As soon as knowledge is no longer viewed as a homogeneous product, generated elsewhere, but rather as an intricate and complex web of heterogeneous and endogenous activities, the issue of knowledge governance becomes central. Many key questions emerge: how firms can manage all the transactions and interactions about knowledge, why knowledge is the object of such a variety of interactions and transactions, and what the relationship is among such a variety of governance mechanisms. Recent advances in the economics of knowledge have made it possible to articulate the traditional analysis of its main characteristics and to identify a wider range of types of knowledge. Knowledge is now viewed as a heterogeneous bundle that can be qualified with respect to its levels of tacitness, indivisibility and appropriability. Let us consider these features in detail: •
•
Knowledge tacitness. Different levels of knowledge tacitness can be identified: knowledge can be highly tacit or sticky when it is fully embodied in persons and organisations and cannot be transmitted or communicated to third parties. Part of the knowledge can be articulated when dedicated efforts have been made to identify the basic elements and to express them in a basic system of codes. Finally knowledge can be codified when it has been fully translated in a consistent code and an appropriate language to express it has been found (Cowan and Foray, 1997; Cowan, David and Foray, 2000; Cooke, 2002). Some levels of tacitness however characterise even codified knowledge. A fully codified knowledge that can be easily transmitted and communicated does not exist. Relevant absorption and assimilation activities are necessary even for codified knowledge if it is to be transferred among individuals and organisations (Cohen and Levinthal, 1990). The notion of cognitive distance among agents plays a key role in this context: the degree of commonality among agents, in terms of shared codes and languages, interacts with the levels of knowledge tacitness (Nooteboom, 2000). Knowledge indivisibility. The analysis of the notion of knowledge indivisibility has led to the identification of different forms according to the structure of the relationship among the basic knowledge modules. A weak indivisibility can be identified when each knowledge module is more self-contained, as opposed to strong indivisibility when high levels of interdependence characterise modules. In turn the notions of knowledge cumulability, knowledge fungibility and knowledge compositeness have been identified with respect to the functional form of the relationship among knowledge modules. Knowledge cumulability is found when different vintages of knowledge are necessary in order for new knowledge to be both acquired and enriched. Knowledge is fungible when a basic module of knowledge finds an array of possible applications in a wide range of scientific fields and products. Knowledge is
74 Governance of localised technological knowledge
•
•
composite when it is itself the result of the synthesis of many different elementary knowledge modules (Laestadius, 1998; Antonelli, 2000, 2001). Knowledge complementarity. The notion of knowledge complementarity can be identified when attention is focused upon the agents involved in the generation and use of new technological knowledge. In the localised technological knowledge approach, the generation of technological knowledge is viewed as the result of the emerging complementarity of a myriad of learning agents, each possessing a bit of knowledge. The successful generation of new knowledge for each agent is conditional on the access conditions to external knowledge. No agent can command the full variety of knowledge modules that are necessary to generate new knowledge. At each point in time agents organise their knowledge exploration strategies in order to increase their chances of accessing the pools of external knowledge. Agents build communication channels and elaborate interaction strategies that are finalised to increase the chance of acquiring external knowledge. Accordingly, learning agents try to focus the direction of their research activities so as to enhance the compatibility, interactivity and hence complementarity between accessible external knowledge and internal knowledge. The amount and the characteristics of new technological knowledge being generated are shaped by the conditions into which the basic complementarity between internal and external knowledge takes place and is articulated. Knowledge appropriability. Although it is difficult for inventors to fully appropriate the stream of economic advantages stemming from the generation of new knowledge, a variety of institutional and market conditions do have an impact on knowledge appropriability. Moreover the actual appropriability of knowledge differs with respect to the characteristics of knowledge, whether it is tacit or codified, cumulative, fungible or composite. Different levels of appropriability can be identified. Finally, the strategic conduct of agents needs to be considered. Agents try to increase the appropriability of the knowledge they have helped to generate so as to increase the valorisation of their proprietary knowledge.
In sum, knowledge, as an activity, is characterised by high levels of uncertainty. It is very difficult to anticipate the outcome of a research process, its duration, its actual economic value and even its specific content. Serendipity plays a crucial role. The organisation of the generation and use of knowledge is afflicted by a variety of economic problems such as transaction costs, agency costs and networking and communication costs. Appropriate governance mechanisms however have gradually emerged and make it possible to coordinate some levels of division of labour and exchange. A wide gulf between the two extreme cases of knowledge as a public good and knowledge as a quasi-proprietary good exists. For the same reasons a variety of hybrid forms of knowledge governance based upon coordinated transactions and quasi-hierarchies can be found in between the two extremes of pure market transaction and pure hierarchical command. Here the characteristics of knowledge matter and the application of the basic tools of
Information economics 75 information economics provides major opportunities to grasp the rationale of knowledge governance mechanisms within the economic system. From the governance viewpoint it is most relevant to know whether knowledge is mainly tacit, codified or articulable and what are its appropriability conditions, together with such characteristics, stemming from knowledge indivisibility, as cumulability, compositeness and fungibility, which have a major role in assessing the appropriate governance mechanisms and modes for their specific effects on knowledge transaction, coordination, communication and interaction costs (Antonelli, 1999a, 2001).
3 An information economics framework for the economics of localised technological knowledge The distinction introduced by Joseph Stiglitz between information economics and the economics of knowledge provides basic guidance to implement the economics of knowledge (Stiglitz, 2000, 2002).4 As Kenneth Arrow (1969) had anticipated, the typical fields of investigation of information economics, namely the analysis of the characteristics of the economic system from the viewpoint of the quantity, quality, symmetry among agents, distribution, access and transparency of information and communication and their effects on the conduct of agents and the tools that have been elaborated since then are most important to understanding how the generation, dissemination and use of knowledge are organised in the economic system. The application of the basic tools of the economics of information, and specifically the analysis of transaction costs, agency theory, asymmetric information, communication and networks, to the economics of knowledge makes it possible to elaborate and operationalise the debates between the Arrovian notion of knowledge as a public good (Arrow, 1962) and knowledge as a quasi-proprietary good (Jones, 1998). This approach makes it possible to articulate the analysis of both the limitations and the conditions for which knowledge can be generated, exchanged and traded among individuals within an organisation and among firms and organisations, across the full range of intertwined product and factor markets, by means of dedicated and innovative forms of hybrid governance based on quasihierarchies, coordinated transactions and constructed interactions. The aim of this section is to show that the tools provided by information economics are useful and fertile inputs to implementing the analysis of technological knowledge as a localised and collective activity. Let us introduce a preliminary and tentative application of the basic tools of information economics to the economics of knowledge in more detail, so as to pave the way to the analysis carried out in the following chapters of Part II: Principal–agent problems When attention is focused on the activities that lead to new knowledge, as opposed to knowledge as a good, a new distinctive characteristic emerges: its unpredictability. High levels of uncertainty characterise the generation of knowledge: serendipity and creativity play a crucial role. The generation of new knowledge is characterised by substantial unpredictability about
76 Governance of localised technological knowledge many different facets: its results, its timing, its content, its scope of application and hence its economic value. Unpredictability affects especially the generation of new knowledge with high levels of codification and scientific content. In this context it is very difficult to organise and manage employment relations. Principals have major problems in assessing the actual levels of creativity and effort of their agents and to value their output. The costs of hierarchical coordination, articulated in agency and organisation costs, limit severely the size and the span of knowledge-intensive activities conducted within the boundaries of a single unit (Arrow, 1974). Agency costs limit the use of hierarchical command of the activities that are necessary to generate and use technological knowledge within the boundaries of the firm for two classes of reasons. Knowledge asymmetries do play a major role within organisations as well. Because of the key role of serendipity and creativity in the generation of new knowledge it is difficult for principals to control the actual content of the operations that lead to the generation of a given amount of standardised knowledge. Agents can try to take opportunistic advantages of the basic information asymmetries with respect to principals about: 1) the perspective value of the knowledge produced; and 2) the actual effort and work that have been necessary to generate it. Agency costs in the generation of knowledge within complex organisations are consequently very high also because of the limitations in anticipating the outcome of a research-in progress not only in terms of rates but also and mainly in terms of directions. The outcome of a given research project can be relevant but in fields of application that differ from the expected ones. The traditional organisation of labour in knowledge-intensive activities characterised by high levels of craftsmanship and self-employment with strong professional content is clearly explained by the high levels of agency costs in monitoring efforts, outputs and applications in the generation of knowledge (Holmstrom, 1989; Garicano, 2000). Internal organisation costs as well limit the number of complementary activities that can be internalised by each firm and hence the amount of knowledge that can be generated, implemented and exploited internally. Unit organisation costs are elastic not only to the size of activities but also and mainly to the variety of activities that need to be internalised. The greater the rate of increase, with respect to the number of activities, of unit organisation costs, the greater is the number of complementary activities that cannot be retained within the boundaries of the firm. Because of hierarchical coordination costs, incumbents miss important opportunities. Large corporations are unable to implement all the opportunities they help to create. Coordination costs in fact apply both to the specific activities that are required to generate new knowledge and to the production processes that are necessary in order to use and exploit the knowledge generated (Arrow, 1974). Static knowledge transaction costs Knowledge transaction costs are found both on the demand and the supply side. Knowledge transaction costs on the supply side define all the costs that agents bear to use the markets for knowledge as a product per se. Knowledge transaction costs on the supply side consist primarily of all the activities that are necessary to make sure that, while attempting to exploit proprietary knowledge, it does not leak out, depriving the legitimate holder of
Information economics 77 part, if not the whole, of the revenue. Such knowledge transaction costs can also be quantified by the sum of the costs of the activities that are carried on to prevent disclosure and to secure the possession of proprietary knowledge plus the missing portions of revenue stemming from unintentional disclosure and the following leakage. Knowledge asymmetries here are most relevant, as they provide the basis for opportunistic behaviour: the understanding about the economic value of new modules of knowledge differs greatly among agents, according to their competence and specific conditions (Oerlemans and Meeus, 2001). Next to the problems determined by imperfect appropriability, the costs of using the markets for knowledge include more traditional activities such as marketing, advertising, technical assistance and in general all the activities that are necessary to identify prospective customers and to strike appropriate contracts with them. The provision of technical assistance to the users of the technological knowledge is at the same time a cause of considerable costs and an effective mechanism to prevent uncontrolled leakage and the opportunistic behaviour of users. Technical assistance is the base on which to implement pricing strategies that take into account the effective amount of economic benefits stemming from the downstream use of the knowledge. Knowledge transaction costs on the demand side define all the costs associated with the exploration activities in the markets for disembodied knowledge such as search, screening, processing and contracting. Knowledge exploration strategies take into account knowledge transaction costs on the demand side in the context of the choice between ‘make’ internal knowledge and ‘buy’ external knowledge. As is well known, the assessment of the actual quality of the knowledge can be difficult when the vendor bears the risks of opportunistic behaviour and dangerous disclosure. Hence the prospective buyer bears the risks of buying a ‘lemon’ (Akerlof, 1970). Knowledge screening costs are especially relevant in this context: the identification of the complementary modules of existing knowledge and the assessment of their conditions of access and use require dedicated resources and qualified competence. A close interaction takes place between knowledge transaction costs on the demand side and knowledge transaction costs on the supply side. Dynamic knowledge transaction costs A relevant distinction is found between static knowledge transaction costs and dynamic knowledge transaction costs. The latter are defined by the instantaneous costs of using the market to trade knowledge at each point in time, with no appreciation of the stream of long-term consequences engendered by the use of the markets. Knowledge asymmetries are most relevant when the prospects for future developments matter. Dynamic transaction and coordination costs are defined in terms of the expected costs of the governance of the stock of knowledge with respect to the stream of generation of new knowledge. Inclusion now yields the opportunity to appropriate the eventual benefits stemming from the accumulation of knowledge in terms of higher opportunities for the introduction of additional units of knowledge. Exclusion and transaction instead yields new costs in terms of the missing opportunities to benefit from the
78 Governance of localised technological knowledge cumulative learning processes associated with the production process itself (Langlois, 1992). Dynamic knowledge transaction costs are relevant on both the demand and the supply side. On the demand side, search and screening costs include the resources to evaluate the scope for incremental advance on the supply side; dynamic knowledge transaction costs arise mainly because of the high risks of opportunistic behaviour of the customers with respect to derivative knowledge. When derivative knowledge matters, the vendor of the knowledge bears the risks of non-appropriation of the results of the scope of implementation of the knowledge, which has been sold. Uncontrolled appropriation of the stream of rents associated to use of the stock of proprietary knowledge, by means of small incremental research costs, can take place with evident damage for the vendor. The working of the markets for knowledge is greatly favoured by the extent to which patents and copyrights can be enforced in the marketplace and licensing is an effective tool to trade specific items of knowledge and competence. The enforcement of the markets for patents is a primary condition for the reduction of knowledge transaction costs and hence the creation of markets for knowledge. The role of the judiciary system in this context is extremely important. Economies of scope in transactions The bundling of two products and more specifically the bundling of services and products has often the direct and clear effect of saving in transaction costs. This case leads to the identification of the notion of economies of scope in transactions. Economies of scope in transactions differ radically from economies of scope traditionally referred to production costs. Economies of scope have first been applied to describe the case where the joint production of two goods costs less than their separate production. Economies of scope in transactions do not apply to the production process: they apply instead to the costs of using the market. There is a case for economies of scope in transactions when and if, production costs being otherwise equal, the joint transaction of two goods, combined into one single new product, makes it possible to bear a lower level of transaction costs than is the case for the use of the markets for each of them. It is important to stress here that such economies of scope in transaction costs take place both on the demand and on the supply side. Economies of scope on the demand side Economies of scope have been traditionally applied to understanding the reduction in production costs that stems from the combined production of two goods into a single production process. Relevant economies of scope on the demand side may take place when the preference of users for a bundle of two or more goods is larger than the sum of the preferences of each of the goods considered as single products. In this case, clearly the demand for the bundle of goods is larger than the sum of the two separated demand schedules. The demand for the bundle of two goods can be larger than the sum of the demand for the single goods when the costs of using the market on the demand side are considered. If the levels of the preferences for the single goods are otherwise equal, compared to the preference for the bundle of goods, consumers may prefer to
Information economics 79 purchase a bundle, instead of the single goods, because of the reduction in search costs that the provision of a bundle as a combined product can yield. This is also the case for a variety of cases of joint consumption: when the use of a product requires or facilitates the use of other products. When economies of scope on the demand side matter, producers have an incentive to bundle two or more products into a single one because of higher revenues, not because of lower costs. The application of the notions of economies of scope in transactions and economies of scope on the demand side to the economics of knowledge is most useful. Nested transactions Information economics has little explored the intricate realm of nested transactions. As a matter of fact many transactions in a single market do take place because they are nested with transactions in other related markets. Transactions in upstream markets can take place, even with high transaction costs, if and when transaction costs in downstream markets are much lower. This is also true when the relationship between markets and related transaction costs takes place horizontally and diagonally between markets that share some basic inputs, customers or providers. Such nested transactions need to be explored when a variety of forms of indivisibility and direct interdependence among products takes place either in the markets for inputs or in the markets for outputs. Joint production of different goods and related economies of scope are often found. Finally, nested transactions can take place when indirect network externalities apply: when the number of users of either a product or a production factor affects the conditions of production and use of other products. When a direct relationship takes place between the levels of transaction costs in the first market and the levels of transaction costs in the second market, with the levels of production costs otherwise equal, the analysis of the market transaction cannot be applied to each single market but to the complete set of interrelated markets. Three specific cases can be identified: 1) interdependent transactions; 2) economies of scope in transactions; and 3) economies of scope on the demand side. Interdependent transactions can be considered as externalities in transactions, when transactions costs in a market have effects on transactions costs in other, related markets. Nested transactions, in the form of interdependent transactions and economies of scope in transactions and on the demand side, are most relevant in the economics of knowledge for the high levels of knowledge indivisibility. The same knowledge module can apply to a variety of products and be embodied in a variety of activities that, because of that, become strongly interdependent. Agents need to assess the full set of transactions and can accept the use of markets with high transaction costs if and when a clear and direct compensation can be found in other related markets. Economies of scope in transactions and on the demand side are most important in understanding the working of active intermediaries such as venture capital firms. Knowledge signalling costs Information about knowledge is especially confused. The mapping of the modules of knowledge in the economic and scientific space and the identification of the competence of agents possessing modules are most difficult. Both the agents involved in the generation of new knowledge and the
80 Governance of localised technological knowledge agents involved in its use have a great incentive to rely on systematic signalling respectively about their competence and their needs. Prospective suppliers of new knowledge have a clear interest in disseminating information about their capabilities in order to attract the interest of prospective buyers. Prospective users need to disseminate information about the specific details of the complementary modules of knowledge that are necessary in order to attract potential suppliers. Knowledge signalling helps the emergence of the markets for knowledge. Yet it has clear drawbacks. On the supply side it risks increasing the non-appropriability of knowledge, favouring uncontrolled leakage. On the demand side it risks revealing to rivals sensitive information about the missing internal competences and the research projects that are being implemented, with negative implications in a strategic setting. Knowledge signalling can take different forms. Scientific publications can be regarded as a form of knowledge signalling by researchers, working both in the academic system and in corporations. The increasing number of scientific publications authored by scientists working in corporations can be regarded as a form of knowledge signalling directly solicited by the firm. Patents play a major role in this context, as they can be considered both a form of knowledge appropriation and a form of ‘protected’ signalling of the competence of assignees. Networking and communication costs The application of information economics analysis to understanding the governance of localised knowledge makes possible further progress. The important notion of communication costs and specifically knowledge networking costs can be introduced here. Knowledge does not spill freely and automatically in the atmosphere: dedicated efforts are necessary to create the institutional context into which external knowledge can be acquired and to reduce its uncontrolled leakage. The capability of agents both to retain some proprietary control and to communicate and hence to access external technological knowledge depends on the fabric of institutional relations and shared codes of understanding which help in reducing information asymmetries and the scope for opportunistic behaviour and in building a context into which reciprocity, constructed trust and generative relationships can be implemented. The economics of communication, an emerging branch of the economics of information, is necessary in order to grasp the working of knowledge externalities and technological transfer. Knowledge networking is necessary when knowledge is dispersed and fragmented, retained by a myriad of heterogeneous agents, and yet characterised by high levels of indivisibility, with important potential benefits in terms of externalities stemming from its integration and recombination. Yet knowledge communication is not automatic and neither is it obvious: dedicated reception efforts are necessary, together with the appropriate emission of relevant signals. The identification and appropriation of relevant codes, protocols and communication channels and specific languages require substantial efforts. An actual translation is necessary in order to understand the content of existing knowledge and its scope of application. Absorption costs are relevant and make difficult the utilisation of existing knowledge and its necessary recombination with the internal sources of knowledge (Cohen and Levinthal, 1990).
Information economics 81 Knowledge communication is the result of much intentional activity designed to create a context conducive to combining variety and complementarity. Systematic networking is necessary to establish knowledge communication flows. The network structure of the system plays a key role in shaping the flows of knowledge communication and hence the availability of external knowledge. Specific, dedicated networking activities are necessary in order to manage the flows of knowledge that are not internal to each firm and yet cannot be reduced to arm’s length transactions. Networking activities make knowledge interactions, as distinct from knowledge transactions, possible. Networking activities are a quite specific – indispensable – ingredient of the basic governance of knowledge (Antonelli, 2001, 2003a). Firms often rely on networking with other independent parties, to increase the proprietary control of their knowledge, to acquire external knowledge and to better exploit it. External knowledge can be acquired by taking advantage of the spillovers from the academic activities, and from localisation in the proximity of other firms. Qualified user–producer interactions are the source of new knowledge both for upstream suppliers and for downstream customers. Knowledge dissemination is better controlled within networks of interactions based upon constructed and repeated interactions, qualified by contractual relations (Antonelli and Quéré, 2002). The notion of nested networking interactions is parallel to the notion of nested transactions. Once again, because of high levels of intrinsic knowledge indivisibility, a variety of interactions in different and yet related contexts can take place. In these circumstances the analysis should embrace the full set of related contexts and assess the levels of interaction costs in each of them and for the full set. A specific case of nested networking interactions and transactions emerges. Here interactions take place because of the effects in terms of transactions in related markets. Such related markets could not work without the set of interactions in the related market. Networking in a market can even help in creating a market for a related product or, in the other way round, transactions in a given market can help in relying on interactions in the related set (Patrucco, 2003, 2005). Knowledge rationing Because of the effects of indivisibility and complementarity, technological knowledge shares the intrinsic features of an essential facility. Its availability is necessary for the generation of further knowledge, both horizontally, that is across different knowledge fields at the same time, and vertically, that is among different vintages of knowledge. The case for a knowledge trade-off emerges: with a weak appropriability regime the incentives to the generation of technological knowledge are put at risk, yet it is clear that with strong appropriability regimes the efficiency in the generation of new knowledge is hampered by the reduction in the effective access to knowledge complementarity. At the system level the risks of knowledge rationing are very high: prospective users of existing and complementary knowledge, even if active in different product markets, are blocked by the limitations to the dissemination of knowledge that arise both from intellectual property right regimes and from the transaction and communication
82 Governance of localised technological knowledge costs that stem from the specific and idiosyncratic character of knowledge embedded in routines and tacit codes. Prospective users are excluded from access to complementary technological knowledge, even if they are willing to pay. The risks of knowledge rationing are augmented by the specific effects of credit rationing for innovative projects. Credit rationing in fact limits access to financial markets for innovative projects for the high risks associated with the radical uncertainty that characterise both the generation and the exploitation of new knowledge. Prospective lenders and investors are worried by the combined high levels of risk: 1) that the activities that have been funded with their own money will not succeed; and 2) that the new knowledge, occasionally generated, will not be appropriated by the inventor, at least to an extent that makes it possible to repay the credits and remunerate the capital invested. Even in the case of a successful generation, lenders have good reasons to worry about dissipation stemming from uncontrolled leakages of proprietary knowledge. As a consequence, worthy inventive activities and innovative projects risk being sorted out in the marketplace (Stiglitz and Weiss, 1981; Guiso, 1998). Credit rationing conflicts with knowledge rationing: the greater the unrestricted access to new knowledge modules and the lower the chances for inventors and innovators to be remunerated, the greater are the risks of credit rationing for inventors and innovators. Here the workings of intellectual property rights, knowledge appropriability and financial markets are deeply intertwined. As a result the generation of knowledge is twice rationed: the first time it is rationed by the reluctance of lenders to commit their resources to a risky activity, and the second time because access to the necessary inputs of external complementary knowledge are constrained by proprietary regimes. In both cases inventors do not have access to essential inputs even when they are willing to pay. The solutions that a system is able to elaborate in order to reduce the risks of knowledge rationing, in terms of both exclusion of prospective users from access to essential knowledge modules and exclusion of prospective inventors and innovators from essential financial resources, play a key role in assessing its rate of technological advance. Innovations in organisations and institutions New institutional arrangements, with respect to both new forms of contracts and new types of intellectual property rights, emerge. Procedural contracts, where the parties agree upon procedural rather than content obligations, emerge, with positive effects upon both coordination among partners in research ventures, once the latter have been established, and transaction costs in the building phase of new consortia and clubs. Procedural contracts make it possible to articulate the behavioural obligations of the partners about the sequence of operations and their timing, even when their content cannot be fully specified. From this viewpoint procedural contracts are an important institutional innovation that helps the better definition and hence the management of incomplete contracts, with clear effects on knowledge governance, typically afflicted by serendipity, bounded rationality and limited knowledge about the full set of future events (see Brousseau and Fares, 2000; Brousseau and Glachant, 2002). The emergence of new specialised intermediaries such as venture capitalists able to
Information economics 83 perform essential functions such as knowledge screening, assessment, evaluation, and bundling new knowledge together with managerial skills and equity into new companies that can be eventually traded in the form of knowledge-intensive property rights in new specialised financial markets plays a crucial role in this context (Avnimelech and Teubal 2004; Antonelli and Teubal, 2007). New forms of quasi-hierarchical command have been explored and implemented with the introduction of non-exclusivity in employment contracts within research organisations and the exposure of internal units to increasing levels of competitive selection in the provision of knowledge inputs by external sources. Institutional innovations are being introduced in the field of intellectual property rights, such as the introduction of the General Public Licence, where the erosion of the exclusivity of proprietary and patented knowledge is balanced by the obligation of users to notify their use to the assignee, to make explicit reference to prior knowledge and to make it available to third parties, that are paving the way to new hybrid forms of intellectual property rights influenced by original applications of the liability rule (see Samuelson et al., 1994; Lessig, 1999; Reichman, 2000).
4 Conclusion Information economics provides important tools to articulate an economics analysis of the governance mechanisms for the generation and exploitation of localised technological knowledge. A variety of hybrid forms of knowledge governance, ranging from coordinated transactions and constructed interactions to quasihierarchies, can be found between the two unrealistic extremes of pure markets and pure organisations. The notion of localised technological knowledge as a highly heterogeneous dynamic process characterised by varying levels of appropriability, tacitness, unpredictability and indivisibility, which take the forms of complementarity and modularity, cumulability, compositeness and fungibility, helps in grasping the logic behind the variety of knowledge governance mechanisms at work. The analysis of transaction, agency and communication costs provides basic guidance to elaborate an integrated framework able to understand the matching between types of knowledge and modes and mechanisms of knowledge governance in both generation and exploitation. The rest of Part II is dedicated to applying and implementing the basic guidelines presented in this chapter and to showing their analytical fertility.
5
Models of knowledge and systems of governance
1 Introduction The notion of localised technological knowledge is the result of a long process of analysis and investigation. This chapter provides an analysis of the changing foundations of the economics of knowledge and of their effects upon the assessment of the design, the characteristics and the performances of the institutions and processes that shape the generation and distribution of technological knowledge. Major changes have occurred in the economic understanding of knowledge in the second part of the twentieth century. Knowledge has first been regarded as a typical public good that markets and profit-seeking agents could not produce in the appropriate quantities and with the appropriate characteristics. These theoretical ingredients paved the way to the build-up of the infrastructure for the public provision of knowledge. Consensus on the analysis of the public good character of knowledge was first contrasted and eventually substituted by the new argument about the quasi-private nature of technological knowledge. The privatisation of the knowledge commons and the new reliance on the markets for knowledge, as a private good, emerges as the second model of governance. Appropriate institutions need to be elaborated and enforced to make this model viable. Intellectual property rights and the new organisation of financial markets based on the direct access of new high-tech companies to the stock exchange become important tools for the governance of technological knowledge. The identification of the central role of external knowledge in the production of new knowledge marks the third step. The ‘discovery’ of the knowledge trade-off stressed the key role of its dissemination and the limitations of the intellectual property rights. Eventually a view based upon a deeper analysis of the interaction between the generation and distribution of knowledge as a collective process based upon the appreciation of its localised character and the key role of the participation of an array of complementary activities has been elaborated. The appreciation of the different forms and characteristics of knowledge makes possible a closer analysis of the role of knowledge interactions and transactions as aspects of a broader governance problem. This debate has had important consequences on the analysis, the design and the implementation of the institutional architectures best suited to favour the production and distribution of knowledge. The shifting focus from knowledge as a public good to knowledge as a proprietary
Models of knowledge and systems of governance 85 good and eventually to localised knowledge as a collective process should not be regarded as a linear evolution, but rather as a process of enrichment of the analytical tools of the economics of knowledge. Today economics of knowledge is characterised by an increasing attention to the variety of characteristics of knowledge and the related implications from a governance and policy viewpoint.
2 Knowledge as a quasi-public good The seminal contributions of Kenneth Arrow and Richard Nelson marked for a long time the economics of knowledge. The Arrovian frame shaped the debate about the economic organisation for the supply of knowledge and provided the theoretical foundations for the build-up of the public knowledge commons. In the Arrovian approach technological knowledge was seen as a public good for the high levels of indivisibility, non-excludability, non-exhaustibility, nonappropriability and hence non-tradability. In this context markets fail to provide the necessary coordination and the case for undersupply takes place. Markets are not able to provide the appropriate levels of knowledge because of the lack of incentives, and the opportunities for implementing the division of labour and hence achieving adequate levels of specialisation (see Table 5.1 for a synthesis of the main arguments). The public provision of technological knowledge, and especially scientific knowledge, has long been regarded as the basic remedy to under-provision. This led to the build-up and the systematic implementation of public knowledge commons. The legacy of patronage, such as universities and other public research centres, received new endorsement and support (Nelson, 1959; Arrow, 1962a). The key role of the public knowledge commons, based upon the public funding of universities and other public research centres, was also consistent with the topdown view about the generation of technological knowledge. Universities and public research centres were better equipped to perform scientific research. The eventual application of scientific discoveries for the generation of technological knowledge and the introduction of technological innovations was instead assigned to corporations. The provision of public subsidies to firms undertaking research and development activities was regarded as a necessary condition to remedy the low appropriability conditions and hence the lack of incentives. Public procurement was the third basic tool to increase the production of knowledge. The demand for weapons especially was regarded as a major instrument to focus resources and identify research direction and objectives with a broader and general scope for derivative technological applications at the system level and relevance from the viewpoint of the general production of new scientific and technological knowledge. The natural leakage of technological knowledge from the military sector – often within the same corporations – was expected to feed the levels of technological opportunity for the rest of the system. The spillover from the high-tech military activities provides unique opportunities for the introduction of product and process innovations in all the other sectors of the economy.
86 Governance of localised technological knowledge Table 5.1 Knowledge as a quasi-public good Characteristics
Processes
Corporate governance Policy
Domain Limitations
Non-appropriability Non-excludability Non-rivalry in use Non-divisibility Non-exhaustibility High replicability Non-tradability Non-fundability Radical uncertainty Serendipity Deductive approach Linear system Top-down Research and development Large corporation Ex ante barriers to entry Internal financial markets Knowledge commons Public procurement Public subsidies Narrow-scope patents Nation state Poor incentives Low transparency Poor allocation Poor monitoring Poor dissemination International free riding
The Arrovian approach was consistent with the neoclassical views about the exogeneity of technological change. New scientific discoveries and eventual advances in technological knowledge could not be regarded as the endogenous product of economic decision making, but as the result of a sphere of human activity that could not be approached with the tools and the instruments of economic analysis. The Arrovian approach provided good economic foundations to the neoclassical assumptions about exogeneity. The limitations of knowledge as an economic good were such that economics could not provide basic assistance in understanding the process by means of which scientific knowledge was produced. The Arrovian approach was also consistent with the so-called ‘linear model’ of the production of science. Scientific knowledge was regarded as the product of deductive analysis fed by acts of insight. Technological knowledge was eventually implemented, taking advantage of the new scientific discoveries. In this context scientific discoveries would generate ‘technological opportunities’ that firms might eventually take advantage of. A science-push model is articulated. Studies of particular innovations in the 1960s provide support to the hypothesis that technology draws directly on science that has been established for some time.
Models of knowledge and systems of governance 87 The notion of technological opportunities easily integrated also into the Schumpeterian legacy according to which the large corporation with substantial market power was the appropriate institution to accelerate the rate of introduction of technological change stemming from new technological opportunities. Because of the low levels of natural appropriability, financial markets perform poorly in providing the necessary amount of external funds to firms undertaking research activities. Financial markets and specifically banks are most likely to ration the credit to innovation because of both radical uncertainty and low appropriability. Even when a research project finally generates a new bit of knowledge, the likelihood that the inventor, and hence the banker who provided the funds, is able to reap the benefits is put at stake by non-appropriability. Non-appropriability leads to non-fundability (Stiglitz and Weiss, 1981). The role of the State is central in this context. A division of labour between public universities and firms was deemed necessary and had strong implications for the architecture, articulation and implementation of the knowledge governance regimes. In this division of labour, universities would specialise in the production of scientific knowledge, i.e. knowledge mainly generated by means of deductive reasoning and applicable to a great variety of specific circumstances. Firms would specialise in technological knowledge, i.e. bits of knowledge directly applied to solve specific and narrow problems. Public subsidies and public direct participation in the production of knowledge, with an emerging infrastructure of dedicated public research laboratories and a strong public demand for knowledge-intensive products, are regarded as the basic instruments to push the introduction of radical technological innovations (Machlup and Penrose, 1950; Alchian and Demsetz, 1973). 2.1 The first knowledge trade-off In this context the scope and breadth of patents were limited. Patents were thought to increase the appropriability of highly specific bits of technological knowledge with a limited scope of applicability. Scientific knowledge could not be patented and in general patents were not granted for large bodies of knowledge (Kingston, 2001). At this time intellectual property rights are not considered as the major tool to improve the static and dynamic efficiency of the economic system in the production of knowledge. Patents are mainly viewed as an instrument designed to increase the incentives of firms to introduce minor technological innovations. Only large incumbents in product markets, characterised by barriers to entry, could fund internally research and development activities with their own money. Ex ante monopolistic market power based upon barriers to entry in existing product markets would provide extra profits and hence secure the financial resources to fund research and development expenditures and, most importantly, reduce the risks of uncontrolled leakage and imitation. Competitors have yet to enter and entry is barred by substantial cost disadvantages. Appropriability is provided by barriers to entry rather than by barriers to imitation. The large corporation is also considered the
88 Governance of localised technological knowledge appropriate tool to increase the rate of introduction of innovations, as it provides internal markets for financial resources and competence: because of low appropriability regimes, arm’s length transactions in external markets cannot be used to coordinate either the allocation of financial resources into research activities and their selection or the necessary division of scientific and technological labour. The foundations of the well-known Schumpeterian trade-off between static and dynamic efficiency are laid down in this context. Monopolistic market power may provide sufficient incentives to innovators to undertake risky activities finalised to the introduction of innovations. Hence monopoly reduces static efficiency, but makes it possible to achieve, via extra profits and increased ex post appropriability based on barriers to entry and hence imitation, dynamic efficiency. Dynamic efficiency stems from the increased amount of resources and incentives made possible by barriers to entry to generate new knowledge and to augment the flow of innovations. In this context monopolistic power in the markets for products makes it possible to reduce the need for intellectual property rights. Arrow (1962a) contrasts the Schumpeterian argument that monopolistic power provides more incentives to introduce innovations and shows neatly that competition provides the larger amount of incentives. Yet Nelson (1959) stresses the positive role of large corporations for their higher scope of action across product markets. High levels of unpredictability of the results of research activities and even lower levels of predictability about the specific applications of the new knowledge characterise the production of knowledge. Large firms can better exploit the results of scientific and technological knowledge by means of diversification strategies in markets where the knowledge generated happens to command larger rewards. From this viewpoint the notion of barriers to mobility, introduced by Caves and Porter (1977), is especially relevant. Large corporations, as incumbents in adjacent markets, have much lower barriers to mobility across product markets. The Schumpeterian trade-off is more and more regarded as a transient problem. Monopolistic market power in the markets for products based upon proprietary technological knowledge and the technological innovations stemming from its implementation was deemed to be temporary. Hence the welfare losses generated by the divergence between marginal and average costs were assumed to be short lived. The short-term duration of monopolistic power in the markets for goods manufactured with the new knowledge seemed to be a solution to the Schumpeterian trade-off between dynamic and static efficiency. In a context where monopolistic market power was regarded as transient, the creation of intellectual property rights emerged as a complementary institutional set-up, parallel to the public provision of scientific knowledge. The demise of the Schumpeterian trade-off leads to the identification of a knowledge trade-off between the increased dynamics efficiency provided by patents, by means of increased appropriability and hence larger incentives to fund the production of knowledge, and the loss in static efficiency determined by patents in the markets for goods. The understanding of the increased monopolistic market power engendered by intellectual property rights suggests the limiting of the scope for patents and their duration, but relying more and more on intellectual property rights.
Models of knowledge and systems of governance 89 In a proper institutional design, intellectual property rights may favour tradability and hence lead to higher levels of specialisation and division of labour in the technological applications of new scientific discoveries, made possible by public support. Intellectual property rights can help in increasing the incentives for the production of incremental technological knowledge, but only in a broader context shaped by the role of the State.
3 Knowledge as a quasi-proprietary good The first major shift in the economics of knowledge takes place when the notion of knowledge as a public good is challenged and knowledge is regarded as a quasiprivate good with higher levels of natural appropriability and exclusivity. Romer (1990) distinguishes between generic technological knowledge, germane to a variety of uses, and specific technological knowledge embodied in products and as such with strong idiosyncratic features. Specific knowledge can be appropriated; generic knowledge instead retains the typical features of the Arrovian public good. Innovators generate generic knowledge while engaged in the introduction of new specific knowledge embodied in new products and new processes. The production of specific knowledge takes advantage of the collective availability of generic knowledge. The spillover of generic knowledge helps the generation of new specific knowledge by third parties and yet does not reduce the incentives to the generation of new knowledge for the strong appropriability of the specific applications. In this new approach, the distinctions between science and technology are blurred and the traditional sequence is reversed. Scientific knowledge can be considered as the ultimate result of an inductive process of articulation and codification of knowledge originated in a tacit form and acquired by means of learning processes. Here the work of Polanyi (1958, 1966) becomes a basic reference. The distinction between tacit and codified knowledge provides the foundation to the new approach to technological knowledge (see Table 5.2 for a synthesis of the main arguments). Tacit knowledge is the result of learning processes: it is not easy to articulate it and to make it explicit. It cannot be shared and applied outside its original ‘locus’ of generation. The eventual articulation makes generalisation possible. The final end outcome of a bottom-up process of systematisation and generalisation is the full codification and hence the generation of scientific knowledge. Scientific knowledge follows technological knowledge rather than preceding it. Knowledge emerges out of the inductive process of abstraction and generalisation rather than from the deductive process of application of general ideas to specific circumstances. Much empirical analysis has explored the role of learning in the accumulation of competence and the eventual generation of new knowledge. Interesting distinctions have emerged between learning by doing and learning by using. Imitation is hampered by major information and adaptation costs; appropriability is de facto secured by high levels of stickiness in routines and procedures: the notinvented-here syndrome is much more effective than assumed in the public good tradition (Mansfield et al., 1981; Harabi, 1995).
90 Governance of localised technological knowledge Table 5.2 Knowledge as a quasi-private property Characteristics
Processes
Business governance
Policy Domain Limitations
Tacit vs codified Stickiness Limited appropriability Modular divisibility Fundability by private equity Limited tradability Inductive approach Bottom-up Learning Spillover as a free good Global companies Financial equity markets: venture capital + IPO + M&A Knowledge outsourcing Markets for knowledge University as a knowledge services supplier Privatisation of the knowledge commons Strong and broad IPR Global markets Knowledge trade-off in the markets for goods and in the markets for ideas Exclusion Concentration
The theory of the firm is deeply affected by the new approach. Within the evolutionary approach, the resource-based theory of the firm consolidated around the assumption that the generation of technological knowledge is the distinctive feature of the firm. The accumulation of competence, technological and organisational knowledge and the eventual introduction of technological and organisational innovations are now considered the essential role of the firm. The firm does not coincide with the production function and cannot be reduced to a production function. From this viewpoint the firm precedes the production function: the technology is in fact the result of the accumulation of knowledge and its application to a specific economic activity. Edith Penrose marks an important contribution to implementing this approach with the identification of the sequence of developing new knowledge as a resource, conceiving of new services which it could deliver and imagining new productive opportunities. In so doing Penrose provides a direct link between the notion of learning and the emergence of novelty within firms (Penrose, 1959; Foss, 1997). The resource-based theory of the firm has grown as a development and an application of the economics of learning. The inquiry about the dynamics and the characteristics of learning processes, such as learning by doing and learning by using, and their relevance in explaining technological change has led to the identification of the firm as the primary locus of the generation and valorisation of knowledge immediately relevant for the economic action, at least in market economies (Loasby, 1999).
Models of knowledge and systems of governance 91 The resource-based theory of the firm focuses attention on the characteristics of the process of accumulation of competence, the generation of technological knowledge and the introduction of technological and organisational innovations, not only as key factors to understanding the firm, but also as the relevant characteristics in the production of technological knowledge at the system level. In this context the firm is the primary, if not the single, actor in the production of knowledge for the whole economic system. The firm is viewed as the privileged locus where technological and organisational knowledge is generated by means of the integration of learning processes and formal research and development activities. The firm is considered in this approach primarily as a depository and a generator of competence and eventually knowledge (Foss, 1997; Nooteboom, 2000). Large firms are regarded as the key actors in the generation of knowledge. The size of firms is considered to be a key condition to generate successfully new knowledge. Increasing returns are considered to play an important role with major threshold effects. Below a given size, the effective performance of research activities and the implementation of learning processes would not be considered possible. The focus on the role of large corporations of this approach is clearly consistent with the Schumpeterian legacy based upon Capitalism, socialism and democracy (1942), itself far away from the Schumpeterian legacy drawn from The theory of economic development (1936 [1911]). Because technological knowledge is now viewed as the sticky joint product of internal learning, it cannot spill freely in the air. Relevant absorption costs for potential users should be taken into account and qualified interactions between producers and users of new knowledge are necessary for technological knowledge to be transferred from one organisation to another. The explicit and intentional assistance of original knowledge holders to prospective users is relevant, if not necessary (Von Hippel, 1988). More recently much empirical evidence and theoretical research have shown that appropriability is de facto much higher than assumed. Knowledge is contextual and specific to the original conditions of accumulation and generation: as such, natural appropriability conditions are far better than assumed. Imitation costs seem high, as do the costs of receptivity and re-engineering necessary to make use of nonproprietary knowledge. The costs of the not-invented-here syndrome are appreciated. The notion of non-appropriability has been the object of systematic redefinition and new understanding (Demsetz, 1967; Levin et al., 1987). The new growth theory builds upon the new appreciation of de facto appropriability, arguing that the economic rents of knowledge can be substantially appropriated, at least to such an extent that firms can fund correct levels of research and development expenditures. According to much new theorising, the characteristics of knowledge are no longer regarded as conducive to market failure. Markets are able to allocate viable amounts of resources to fund the generation of knowledge. The optimum amount however is not properly identified by market forces (Romer, 1990, 1994; Aghion and Tirole, 1994). On a parallel and complementary ground, evolutionary theory stresses not only the role of the firms as the locus of accumulation and generation of technological
92 Governance of localised technological knowledge knowledge, but also that of the marketplace as the proper selection mechanism. The selection in the marketplace among competing bits of knowledge and related innovations that impinge upon them makes it possible to fully endogenise the rate and the direction of technological knowledge. The supply of knowledge and its selection can be considered as two different and yet related steps. In this context, intellectual property rights play an important role in creating the institutional conditions to secure appropriability and hence to increase the levels of incentives to fund research activities by firms. Intellectual property rights, if properly designed, may also favour tradability and hence lead to higher levels of specialisation and division of labour. Intellectual property rights can help in increasing not only the incentives to the production of scientific and technological knowledge, but also its tradability and hence the efficiency of the generation process. With a significant analytical effort, the new conditions for the appropriability of knowledge are considered sufficient to claim that tradability is also possible. Trade in knowledge is now regarded as a major opportunity to achieving the well-known advantages of market coordination in terms of division of labour, specialisation and selection. A huge literature explores the evidence about the growing flows of trade in patents and licences in both international and domestic markets (Geroski, 1995a; Arora et al., 2001). The new approach is also fed by growing concern about the limitations in the governance of the model based upon knowledge as a public good. In that context, in fact, such issues as the criteria for the allocation of resources, the methodology for the correct monitoring of their use and the assessment of their results were missing. The case for ‘government failure’ is now applied to science policy as well: the risks of government failure in selecting, conducting and funding research activities are now regarded as non-trivial (Buchanan, 1965). Fundamental questions were left without an answer and even worse without a methodological clue to address them. How should the levels of public funds to scientific activities be fixed? How should they be distributed among universities and scientific disciplines? How should they be delivered? How should the results of public funding be assessed (Kealey, 1996)? The dissemination of the results of the scientific research were left to publications in scientific journals and the hiring of Ph.D.s from the business community with little understanding of the problems of scientific dissemination and of the serious limitations to the spontaneous communication of the new scientific discoveries to the community of potential users. The perverse effects of international free riding have been much emphasised. The international uncontrolled leakage and spillover of technological knowledge would benefit the technological strategy of laggard countries based upon fast imitation and re-engineering. This new approach paved the way to significant steps towards the privatisation of public knowledge commons. The public provision of subsidies to firms undertaking research and development activities and the direct role of the State in the production of knowledge come under a closer scrutiny. The role of the university as the single provider of externalities to the economic system is questioned (Henderson et al., 1998). Specifically, empirical evidence is found according to
Models of knowledge and systems of governance 93 which public universities are far less efficient than private research institutions both in the generation of scientific knowledge and in its dissemination (Lach and Schankerman, 2003). A closer look at the working of the public commons and the need to put under scrutiny the productivity of the resources invested in the public knowledge commons, at both the system and the single unit level, is advocated (Jaffe and Lerner, 2001). The general reversal of policy emphasis – which gained consensus in the 1990s – from the role of public policy to the privatisation of public provision inspired by the swing from the presumption that governments can do everything that markets do and more besides to the new presumption of market efficiency, associated with rational expectations, together with the new theory about knowledge as a proprietary quasi-private good, provided theoretical support to a new – problematic – understanding about the role of public research. As a consequence, a wave of privatisations has been taking place: universities have been pushed to enter the markets for knowledge and knowledge outsourcing. The new enclosures substitute for the knowledge commons. Public research centres and universities were solicited to patent their discoveries and often forced to enter the markets for the technological outsourcing of large corporations. The conditions for the effective appropriation of knowledge are enforced both at the firm level and in public organisations: the mobility of human capital is more and more regarded as a sensitive issue (May, 2000; Mowery et al., 2001). Academic patenting and scientific entrepreneurship have been praised as new effective tools to stimulate the distribution of knowledge and to increase the incentives to its production. Much analysis has been carried out on the regional aspects of the interplay between the research system and the business community: geographical distance has proved a relevant factor in this context (Feldman, 1993, 1994, 1999; Audretsch and Feldman, 1996; Audretsch and Stephan, 1996; Geuna, 1999). The multinational firm and its evolution into the global corporation emerge as the most appropriate tool to foster the levels of access to scientific knowledge and the rate of accumulation of technological knowledge by means of the capability to combine the selection of a few sites worldwide for the location of research facilities with a portfolio of production and commercial activities scattered throughout the global economy. The global corporation is able, at the same time, to generate and disseminate technological knowledge by means of the powerful internal markets, into which the international transfer of technological knowledge can take place with reduced risks of dissipation and opportunistic behaviour. Enhanced internal tradability favours the division of labour and makes it possible to reap the benefit of specialisation (Cantwell and Iammarino, 2003). At the same time, the role of intellectual property rights is reconsidered. Intellectual property rights can complement and integrate the appropriability of technological knowledge, so that markets for knowledge, now much closer to traditional economic goods, can be developed. Intellectual property rights are now regarded as a complementary condition to increase the tradability and consequently to achieve the standard conditions for equilibrium supply of knowledge in the
94 Governance of localised technological knowledge economic system. The extension of patent protection to new forms of knowledge such as software, algorithms and genetic entities finds here its foundations (Merges and Nelson, 1994; Sakahibara and Bransletter, 2001). In this context, the role of financial markets as an important component of the mechanism design for the governance of technological knowledge is appreciated (Florida and Kenney, 1988). Special attention is paid to the opportunities provided by financial markets as an institution for the exchange of property rights of new innovative companies. The role of IPOs (initial public offerings) as a way to convey financial resources into new ventures and at the same time to assess and select the choices of venture capital receives much attention (Gompers and Lerner, 2003). Venture capital in fact can work only if financial markets are ready to offer an opportunity to dispose of the shares of new companies after incorporation and eventual market success (Lerner, 1995; Kortum and Lerner, 2000). Pension fund capitalism here plays a key role, as it provides the ‘perfect matching’ between financial institutions that work with a long-term horizon and low levels of exposure to the short-term mass exit of disappointed investors and the long-term business of sorting out profitable new ventures from a class of financial investments with high rates of variance and short-term risk (Clark, 2000). This evolution of financial markets marks a major shift with respect to the traditional emphasis on the limits of the provision of funds for innovative activities by banks. In the previous approach the stock exchange did not play a role and only credit was considered. Bankers were supposed to be reluctant to provide credit to risky ventures based on research activities and innovation. The case for credit rationing emerged as a major problem in funding innovation activities (Stiglitz and Weiss, 1981; Guiso, 1998). In turn, targeted credit rationing in financial markets stressed the role of internal financial markets and extra profits to fund internally research activities that bankers could not properly fund because of the lack of instruments able to generate and assess the necessary information on the risks and the eventual pay-off of debt-based resources in research activities (Saint Paul, 1992; Gompers and Lerner, 1999). The new thinking stresses the radical asymmetry between debt and equity in providing the funds for research activities. Research activities are characterised by high pay-offs and high risks: lenders bear the risks of failure and do not participate in the benefits of successful ventures, while equity investors instead can appropriate the returns of the small but highly profitable share of research activities that hit commercial and economic success (Stiglitz, 1985).1 The new emphasis on the working of equity markets as an effective and efficient tool to provide funds to innovative undertakings leads to a new assessment of the role of scientific entrepreneurship. Scientific entrepreneurship grows into a fully fledged new viable mechanism to incentivise, generate and disseminate technological knowledge in economic systems. The role of new science-based firms as converters of technological opportunities, available in science-rich environments such as universities and large corporations, into market experiments is now better recognised. The dynamics of spin-off from research laboratories and the supply of funds and managerial competence by venture capitalism are now regarded as
Models of knowledge and systems of governance 95 complementary tools for the eventual creation of new science-based firms. Next to the size of firms, new attention is given to their age. The complementarity between new and small and large firms is now advocated: small start-ups are seen as the best way for new technologies to enter the marketplace. Eventual mergers and takeovers by larger firms will lead to the integration and dissemination of the new technologies to larger corporations (Audretsch, 1995; Audretsch and Feldman, 1996; Audretsch and Stephan, 1996). Scientific capitalism is based upon scientific entrepreneurship, effective intellectual property rights systems, academic patenting, venture capitalism, initial public offerings and financial institutions, including dedicated stock exchange systems (NASDAQ), where the new ventures can be assessed and possibly recombined with existing companies by means of mergers and acquisitions. Technological knowledge can flow within the economic system embedded in new companies. But, once again, the trade-off between static and dynamic efficiency emerges. The determinants of the partial excludability and appropriability of the new knowledge and more specifically the ratio of the generic component of the new knowledge with respect to the idiosyncratic one are not investigated. Hence the definition of the incentives to the generation of new knowledge on the one hand and the contribution of the spillovers of generic knowledge to the generation of new knowledge remains unclear.
4 Knowledge as a collective process 4.1 The discovery of the second knowledge trade-off The second major swing takes place when a closer analysis of knowledge appropriability made it possible to understand, next to its negative effects in terms of missing incentives and hence undersupply, the positive effects of technological spillover and the key role of technological externalities (Nelson, 1982; Griliches, 1992; David, 1993; Cooke, 2002). The new approach is based upon the discovery of external knowledge as an essential intermediary input in the production process of new knowledge. The discovery of external knowledge, available not only by means of transactions in the markets for knowledge, but also by means of technological interactions, marks a new, important step in the debate. External knowledge is an important input in the production process of new knowledge. Thus major progress is made when the special character of knowledge as a non-exhaustible good that is at the same time an output and an input into the production of other knowledge is grasped and retained at the core of the analysis. Here the derivation from the Arrovian notions of the non-excludability and non-divisibility of knowledge is clear. Static efficiency provided by the articulated mechanism design of so-called ‘scientific capitalism’ is now confronted with the negative effects in terms of dynamic efficiency by the excess control and exclusivity of knowledge as a quasiprivate good. The dissemination of knowledge is put at risk by its very stickiness,
96 Governance of localised technological knowledge the proprietary character of much technological knowledge, the dwindling role of public knowledge commons, and the strong intellectual property rights regimes that have been enforced. Poor dissemination and exclusivity put at risk the access to external knowledge for each agent and hence the working of cumulability and complementarity. This reduces the future flow of additional units of new knowledge. The core of the new analysis is now centred upon the exploration and identification of the conditions into which external knowledge, as an essential input in the production of new knowledge and new technologies, is effectively disseminated in the economic system. This line of inquiry contributes to the founding of the systems of innovation approach, where the production of knowledge is viewed as the result of both knowledge transactions and the cooperative interactions, mainly rooted in regional space, of agents undertaking complementary research activities.2 The focus is now more and more centred upon the analysis of the mechanisms of governance of the broad array of knowledge interactions among agents, including coordinated division of labour and market transactions, and their effects in terms of generation of and dissemination of new knowledge. Regional economics contributes significantly to the new approach, highlighting effectively the role of geographic space in the distribution and circulation of knowledge, and at the same time regional analysis is deeply affected by the new understanding of knowledge as a way to understand the role of geographic space (Feldman and Audretsch, 1999; Antonelli, 2001, 2003a; Feldman and Massard, 2002). Here in the economics of technological knowledge the issues of externalities on the demand side become relevant and evident. The generation of technological knowledge is now considered to be characterised by relevant and necessary externalities, both technical and pecuniary. The notion of user interdependence makes its foray into the scene when agents value the levels of usage of other agents of certain goods. As far as scientific and technological knowledge is concerned, interdependence among users, hence on the demand side, is very strong. The chances of generating a new relevant bit of knowledge for each agent depend upon the levels of accumulation of skills and competence, education and access to information of the other agents in the community. The evidence especially in new information and communication technologies confirms that proximity matters in assessing the rates of introduction of innovation. An important contribution to understanding the role of external knowledge is provided by an extension and implementation of the approach based upon learning processes. The competence and experience that are necessary to innovate are acquired not only in the repeated usage of a given set of capital goods and intermediary products and in the production of well-identified products. The experience accumulated in marketing and interacting with a well-defined set of consumers and competitors in a limited range of products is necessary in order to generate new knowledge and eventually introduce new products. Interactions with customers are a primary source of tacit knowledge about their tastes and needs (Lundvall, 1985). No successful product innovation can be effectively and successfully introduced without some dedicated competence about the marketplace. The distance, in the product space, from the products being traditionally delivered to the marketplace
Models of knowledge and systems of governance 97 can be considered a strong factor in increasing innovation costs and decreasing efficiency in the generation of innovations. The amount of external technological knowledge, available in a given context, industrial, technological or regional, and its conditions of accessibility and proximity, becomes an important endowment, as well as the conditions of access to it and the characteristics of the relational set-up. The issues of the distribution of knowledge become central in the debate, and the notion of a knowledge tradeoff is articulated. Uncontrolled leakage and low appropriability regimes reduce incentives, but may not necessarily lead to under-provision. Low appropriability engenders technological externalities and spillovers that are the prime factor in increasing the efficiency of the generation of new knowledge, at the system level: the growth of efficiency can compensate for lower inputs (Griliches, 1992). The advantages of the intellectual property rights regime, in terms of increased incentives to the market provision of technological knowledge, are now balanced by the costs in terms of delayed usage and incremental enrichment. The vertical and horizontal effects of indivisibility display their powerful effects in terms of cumulability. Indivisibility of knowledge translates into the basic cumulative complementarity among bits of knowledge. Complementarity and cumulability in turn imply that new bits of knowledge can be better introduced, building upon other bits already acquired, both in the same specific context and in other adjacent ones. The access exclusion from the knowledge already acquired reduces the prospect for new acquisitions and in any event has a strong social cost in terms of duplication expenses (O’Donoghue, 2001). The costs of exclusion associated with intellectual property rights, as a consequence, should be taken into account. Monopolistic control of relevant bits of knowledge, provided both ex ante and ex post by patents and barriers to entry in the product markets respectively, can prevent not only its uncontrolled leakage and hence its dissemination but also further recombination, at least for a relevant stretch of time (Arrow, 1969; Dasgupta and David, 1987, 1994; David, 1993; Shavell and Van Ypersele, 2001). A new knowledge trade-off emerges. The ‘old’ knowledge trade-off questioned the negative effects of intellectual property rights in terms of monopolistic consequences in the markets for products. The new knowledge trade-off questions the negative effects of intellectual property rights in terms of the monopolistic effects on knowledge itself. In the ‘old’ knowledge trade-off, the positive advantages of intellectual property rights in terms of increased incentives to generate new knowledge and hence in terms of dynamic efficiency were balanced by the negative effects in terms of a reduction in static efficiency associated with the extra profits of patent holders. In the ‘new’ knowledge trade-off, intellectual property rights are questioned also in terms of dynamic efficiency losses. It seems evident that, while too strong a regime of protection may have positive effects in terms of increased incentives to the generation of knowledge, it has clearly negative effects also in terms of delayed and slower circulation and distribution of the new knowledge available (Mazzoleni and Nelson, 1998). The new knowledge tradeoff adds to the old knowledge trade-off. Murmann (2003) provides a fascinating
98 Governance of localised technological knowledge account of the perverse effects of the patenting strategy pursued by Bayer on the technological development of competitors in terms of the negative effects of the increased monopoly both in the markets for goods and in the markets for ideas. The duration of exclusive property rights assigned by patents and the conditions for their renewal become a central issue not only from an antitrust perspective, but also for the possible negative drawbacks in slowing the rate of generation of new knowledge, especially when general-purpose knowledge with a wide scope of applications is concerned (Scotchmer, 2001; Shankerman and Scotchmer, 2001). The breadth of patents is also questioned: when the breadth is large, the protection is not specific and the negative effects in terms of foreclosure can easily exceed the advantages in terms of increased incentives. A narrow definition of the scope of application of intellectual property rights is thus recommended (Klemperer, 1990; Merges and Nelson, 1994; Hopenhayn and Mitchell, 2001). The understanding of the twin aspects of the old and the new knowledge tradeoff contributes to the parallel development of a systemic approach to the understanding of the economics of technological change, an approach in which the characteristics of the regional, industrial, professional and national systems play a major role in determining the rate and the direction of technological change. Technological change is endogenous to the system in which each agent is rooted. 4.2 The emergence of localised technological knowledge Complex system dynamics provides important methodological contributions to implement a new systemic and dynamic approach to the economics of knowledge. Technological knowledge is now viewed as the outcome of the localised interactions of a variety of learning agents, embedded in a network of qualified relations, able to learn and to accumulate experience and competence, although in a limited range of activities, localised in the technical space and in the product space where each firm has accumulated competence by means of processes of learning by doing and learning by using (see Table 5.3 for a synthesis of the main arguments). In this approach technological knowledge is seen as the ongoing result of a collective process activity made possible by the continual efforts of accumulation of highly differentiated competence and technological knowledge, based upon localised learning processes and the eventual introduction of innovations, of a myriad of heterogeneous and interacting agents rooted in a well-defined set of scientific, technical, geographic, economic and commercial circumstances. Localised technological knowledge is primarily the result of the valorisation and implementation of underlying learning processes, in doing as well as in using and in interacting, which are localised in the specific context of action of each economic agent. The capability and the competence acquired by means of learning processes are heavily localised in a limited technical and product space for many reasons. Agents are characterised by bounded rationality and yet are able to learn. The mix of bounded rationality and learning capabilities makes sure that the generation of technological knowledge is possible only in the proximity of the specific learning context with respect to both the product and the process.
Models of knowledge and systems of governance 99 Table 5.3 Localised knowledge as a collective activity Characteristics and forms
Processes
Business governance (see Chapters 6–10) Policy (see Chapters 11 and 12)
Domain Limitations
Articulable Dispersed and fragmented Modular complementarity Cumulability Compositeness Fungibility Knowledge path dependence Unpredictability Interaction of induction and deduction Complementarity of external, internal, tacit, codified Knowledge Research Learning Exploration Communication Absorption Recombination Socialisation Quasi-hierarchical command of tacit and sticky knowledge Constructed interactions for articulable knowledge Coordinated transactions for codified knowledge Academic reputation and the new markets for knowledge Fungible knowledge commons Interface agencies Patents as signals Knowledge as an essential facility Non-exclusive IPR Scan globally Implement locally Knowledge transaction and communication costs
Firms are viewed as creative agents, which are not limited to adjusting prices to quantities and vice versa. They are also able to learn and change their technology, as well as their strategies. When market conditions do not match the plans and irreversibility and limited knowledge make sheer technical substitution expensive, firms are induced to introduce technological innovations. In order to do so, the generation of new technological knowledge is activated. The new understanding of knowledge indivisibility contributes to this approach. Knowledge indivisibility triggers two important implications: 1) The generation of new knowledge within firms relies upon the stocks of knowledge accumulated. The issue of knowledge cumulability emerges here with great emphasis. 2) Firms can generate new knowledge and hence eventually introduce new technologies only when and if they are able to take advantage of external knowledge. No firm can rely exclusively on its own internal knowledge, either tacit or codified, whether it is the result of learning processes or formal research and development activities. External knowledge is provided by suppliers, customers and competitors, as well
100 Governance of localised technological knowledge as by universities and public research centres. Thus the localised knowledge approach recognises the role of universities and other research centres as suppliers of external knowledge to the production of knowledge conducted by firms. The flows of both tacit and codified knowledge from universities to firms by means of the supply of general scientific knowledge, graduate students, knowledge outsourcing, informal contacts and occasionally academic spin-offs are fully acknowledged (Geuna, 1999; Geuna, Salter and Steinmueller, 2003). The flows of external knowledge within the business sector, among suppliers, customers and competitors, take place by means of a variety of mechanisms, ranging from transactions in the markets for knowledge to interactions based upon socialisation processes. The appreciation of the role of external knowledge owes much to the analysis of Hayek (1952) about the dispersion of knowledge as a key characteristic of economic systems. Technological knowledge is dispersed and fragmented into a variety of specific and idiosyncratic applications and contexts. New knowledge can be generated only if and when such dispersed and fragmented bits are brought together and their potential complementarity is discovered and valorised. This view contrasts sharply with the top-down understanding of knowledge built around the Arrovian tradition and contributes to the implementation of a bottom-up approach where knowledge is viewed as the product of a collective process of discovery and understanding (Rizzello, 1999, 2004). Learning agents are scattered in space and each possesses a bit of knowledge. In turn high levels of potential complementarity characterise such bits of knowledge. The localised context of learning and action affects the chances of generating new technological knowledge because of the key role of external knowledge as an indispensable input. By the same token the similarity and alignment of individual learning processes play an important role in the collective undertaking that leads to the generation of new knowledge: dynamic coordination becomes a central issue (Richardson, 1972; Amendola and Gaffard, 1988, 1998). In this approach, trade in knowledge remains indeed an important opportunity for the acquisition of external knowledge and to enhance the division of labour in the production of knowledge. The crafting of an array of institutional arrangements however is now viewed as a necessary process in handling the difficulties in understanding and using the differentiated set of new knowledge modules supplied in the marketplace: the relevance of dynamic complementarity emerges (Menard, 2000; Guilhon, 2001, 2004). Proximity matters in many different ways. Proximity matters in regional space, as well as in technical, professional and industrial space. Proximity of firms to large research laboratories and academic centres is now regarded as a vital condition for the successful introduction of new technologies. Proximity in product space matters as the prime source of information about the tastes of customers and their potential interests. Proximity matters in the product space as a factor which makes the acquisition of information and eventually knowledge easier with respect not only to the habits and preferences of customers but also to the capabilities of competitors and their strategic attitude. The introduction of product innovations in market niches that are far away from the source of the experience of each firm is put at risk by
Models of knowledge and systems of governance 101 the lack of specific competence, and relevant, additional costs should be recognised (Boschma, 2005). Variety is the second key factor in enhancing the chances of the generation of new knowledge. Too much cognitive proximity among learning agents can affect the capability of a system to engender successful innovation. Variety of complementary competence is key in assessing the innovation capabilities at both the firm and the system level (Nooteboom, 2000, 2002). The key distinction between receptivity and absorptive capabilities as distinct from the strength and intensity of the message plays a key role in this context. The structure of economic systems from the viewpoint of the knowledge communication flows receives much attention: the structure of the communication channels is analysed and the organisation of communication flows within the networks of relations appreciated (Cohen and Levinthal, 1989, 1990). The role of communication and transmission of knowledge is more and more appreciated. Communication theory is applied successfully to the analysis of knowledge communication processes. The density of communication channels and their duration are considered relevant structural elements of an economic system. The role of business interactions is now appreciated from the viewpoint of their communication role. Prices, of course, are no longer viewed as the single vectors of all relevant information for economic decision making. Next to prices, vital information is transferred and contributes to the generation of knowledge by each economic agent. The communication of bits of knowledge, in other words, is not considered as obvious and spontaneous, but on the contrary is viewed as the result of intentional efforts in terms of both connectivity and receptivity (Gibbons et al., 1994; Antonelli, 2001). Systems differ with respect to the speed and capillarity of the flows of knowledge communication. In turn the rate of generation of new knowledge and introduction of new technologies is clearly influenced by the permeability of the system. Percolation analysis borrowed from physics and communication theory is introduced in the economics of knowledge so as to provide tools to appreciate the distinctive role of receptivity and connectivity in knowledge communication processes (Antonelli, 1999a; Solomon et al., 2000). According to the acquisitions of the localised approach, the firm cannot be seen as the single actor in the process of the generation of new knowledge. The assumptions about the complementarity between internal and external knowledge play a key role in this context. The variety of firms and learning institutions is most important in the generation and circulation of knowledge when the latter is viewed as a collective good, with varying degrees of appropriability, dispersed and fragmented in the economic system, the result of both top-down and bottomup processes, where learning by doing, learning by using and learning by interacting with suppliers, customers and rivals play an essential role, together with research and development activities. Knowledge is now, more and more, viewed as a collective process. The notion of a collective process differs sharply with respect to both the Arrovian tradition of knowledge as a public good and the approach to knowledge as a quasi-private
102 Governance of localised technological knowledge good. Collective processes in fact are characterised not only by partial appropriability and shared property rights but also by the role of the intentional effort, participation and contribution of each agent. Collective knowledge is more than a club good for two important reasons: knowledge is an activity, rather than a good, first, for the continual efforts of acquisition and implementation it requires and, second, because of the increasing returns which can take place when and if dynamics coordination and stochastic processes make it possible to take advantage of potential knowledge complementarities (Buchanan, 1965). Collective knowledge in other words is a shared activity that can be implemented only by interactive agents that belong to a community of action and understanding. Collective knowledge pays attention to the consequences of knowledge indivisibility and the role of the complementarity among the localised bits of knowledge possessed by each agent that characterise both the generation and the dissemination of knowledge in the system and value the contribution of external knowledge in the production of new knowledge. In this approach the role of technological communication among learning agents is stressed as a major systemic character affecting the capability of each agent to implement its internal knowledge (Guiso and Schivardi, 2007). New hybrid governance structures are identified as viable solutions according to the characteristics of knowledge. Forms of quasi-hierarchical command are found when knowledge is mainly tacit and sticky. When knowledge is articulable, constructed interactions among learning agents appear as the typical form of governance. Finally, when knowledge is more codified, coordinated transactions are often used in quasi-markets (see more in Chapter 10). The network structure of knowledge communication networks affects deeply the flows of knowledge communication and hence the actual availability of external knowledge. There is an array of possible network architectures. In geodesic networks, i.e. networks where each agent has a direct link to each other agent, communication costs are very high: the dissemination of new knowledge is hampered by relevant communication costs and by the decay of knowledge spillovers associated with distance and heterogeneity among agents. Within centred networks based upon many interconnected and hence competitive hubs, knowledge is disseminated far better than in fragmented networks, where only a few links connect scattered clusters or in networks based upon monopolistic hubs able to exert a control upon knowledge flows and to extract rents out of it. Scale-free networks organised upon hubs can better favour the dissemination and eventual recombination of knowledge bits imperfectly appropriated by their first users (Barabasi, 2002). In this context it is clear that the issue of dynamic coordination of the interdependent activities of the myriad of complementary learning agents involved and the related design of network architecture that shapes the flows of communications and the interaction modes, including transactions in the marketplace, becomes most relevant (Richardson, 1960, 1972; Amendola and Gaffard, 1988, 1998). Key questions need to be addressed in this context: whether economic systems are able to generate and implement the perfect network architecture; and whether
Models of knowledge and systems of governance 103 spontaneous multi-task interaction of heterogeneous agents active in a variety of markets and embedded in a variety of contexts can actually lead to the ‘perfect’ design. The difference in the time scale of the flows of knowledge among the agents within the networks of communication channels existing at each point in time and their duration and the time required for their implementation and incremental construction becomes a relevant factor in assessing the emergence of appropriate network structures. Within knowledge networks, localised technological knowledge can be understood as a collective activity characterised by the complementarity between heterogeneous and yet complementary items. Such complementarity takes place especially between external and internal knowledge and the stock of existing knowledge and the flows of new knowledge. The implications of the indivisibility are reconsidered. The generation of knowledge exhibits the distinctive features of a path-dependent process. New knowledge in fact is now viewed as the result of the creativity of single learning agents, which are localised in an array of technological, industrial, economic and epistemic spaces, able to rely upon the cumulated stock of previous knowledge and yet highly sensitive to the dynamics of localised interactions, knowledge externalities and feedbacks, structured by the architecture of quasiirreversible communication channels. This approach contrasts sharply with the pastdependent analysis of the evolutionary understanding of technological knowledge whereby firms generate new knowledge along trajectories defined at the onset and characterised in terms of irreversibilities by the role of cumulability.
5 The path-dependent complex dynamics of collective knowledge Path-dependent complex dynamics provides a major opportunity to articulate the indeterminacies of the interplay among supply and demand externalities into which traditional microeconomics stumbles. In the approach developed so far, five characteristics of knowledge matter: • •
•
•
•
At each point in time knowledge is dispersed and fragmented, scattered among a myriad of learning agents. The complementarity among modules of knowledge possessed by each agent is relevant: its valorisation provides the opportunity to generate new radical advances. No agent can possess and control all the knowledge available at each point in time; hence the complementarity between internal and external knowledge plays a key role in the generation of new knowledge. Proximity among agents is relevant for the complementarity among external and internal knowledge, as well as among the learning efforts of each agent and the modules of knowledge possessed by each agent, to be implemented. Finally, and most importantly, agents can learn and make intentional efforts to generate new knowledge, including intentional strategies, based upon
104 Governance of localised technological knowledge procedural rationality, to modify albeit locally – because of bounded rationality and switching costs – their position within the knowledge networks. Such efforts take place when the dynamics of localised technological change exerts its effects and agents need to change their current techniques. In this context, the conditions which characterise the access to external knowledge play a key role in explaining the innovation capability of agents within a system. At each point in time the topology of agents in the space of knowledge, and hence their relative distance and the structure of their relations and interactions, are key features of the system. Accessibility of external knowledge and hence the scope to implement the potential complementarities are influenced by the channels and flows of communication in place at each point in time. So far the typical frame for complex dynamics is set. Occasional stimulations – such as the mismatch between plans and actual market conditions – push agents to seek for new knowledge. When they do so, the links in place provide access to external knowledge. According to an array of parameters such as the existing topology, the distance among learning agents, the stock of communication channels in place, their connectivity, the receptivity of co-localised agents and hence the complementarities of their internal modules of knowledge and the structure of the stochastic effects of such relations, some agents in some regions are more successful than others. Eventually the process can spread across regions and the full map is finally covered. Complex system dynamics provides an analytical framework into which our analysis can be accommodated with some qualification. Kauffman (1993) provides a path-breaking definition of complexity based upon the interaction of size and interdependence, which fits very well in this approach to the economics of knowledge. Kauffman elaborates a model of complexity based on two parameters, N, the number of components making up a whole, and K, the degree of interdependence among these components. In this context, complexity is defined as the number of components of a certain piece of knowledge and the degree of interaction between them. In this way, indeed, complexity provides a general context in which the generation of knowledge can be viewed as a collective process undertaken by a myriad of interacting and complementary agents. It is worth recalling that the complex dynamic system approach builds upon four basic elements: 1) systems are constituted by heterogeneous agents with specific characteristics distributed in a relevant multidimensional space; 2) location in the multidimensional space is relevant: each agent has access only to local information and local knowledge, i.e. no agent knows what every other agent knows; 3) agents are embedded within multidimensional topological spaces shaped by networks of relations, ranging across a variety of interactions, transactions and communication channels that affect locally their behaviour; 4) agents are creative, i.e. agents are able to learn and to change the rules of their behaviour. Creative agents can also change their location in the multidimensional space so as to increase their proximity to other agents and change the density of knowledge networks: in so doing creative agents can change their creative capabilities and the topological features of the system at large.
Models of knowledge and systems of governance 105 With respect to standard complex dynamics, however, economic analysis cannot forget two major aspects that are intrinsic to economics and often bypassed by superficial applications of models drawn from physics or biology. First, the scope for intentional action of agents matters. Agents are indeed affected by bounded rationality and myopic capabilities; nevertheless they can act with procedural rationality and elaborate plans so as to build connections, create communication channels, implement their connectivity, align their receptivity and especially direct their learning efforts towards fields of knowledge that are localised and yet richer in terms of potential integrations with external pools of knowledge. In so doing agents are expected to try to maximise the expected, although myopically, profitability of their intentional activities. Barabasi (2002) provides a fascinating analysis of the emergence of scale-free networks based upon hubs with an array of communication channels within network systems: heterogeneity within networks may be the result of micro-mobility in the knowledge space of strategic although bounded agents able to understand the relevance of local commons of collective knowledge. Second, and most importantly, market selection is at work. Within a given topology of agents and hence a given map of distances among agents and related structure of communication channels and communication flows, some agents can happen to align their learning capabilities and direct their communication efforts better than others, or simply in a more effective way than others, even by chance, if not by mistake. Such agents will nevertheless experience faster rates of generation of new technological knowledge and hence faster rates of introduction of localised innovations. Market selection will appraise such events. In both cases the effects of profit maximisation do change the course of events as anticipated by the sheer adaptive dynamics of complex systems where blind agents are not able to anticipate the course of their actions and the market is never able to select the direction of their actions. It is clear that the metaphor of stochastic action within a symmetric lattice is not appropriate in accounting for the intentional activity of economic agents within an economic system (Kirman, 2004).3 Path dependence seems in this context the single approach which can help in understanding the dynamics of the process. The approach to path dependence considers creative agents that at each point in time are both learning and capable of intentional action and yet under the constraints of the effects of irreversibility and local externalities. The topology of the space exists as much as the array of its characteristics such as the structure of communication channels in place, their connectivity and receptivity, the structure of relations, and interactions which build upon the connections in place. Agents however, within such a context, are able both to generate new knowledge and also to change the topology at each point in time. Path dependence provides the framework for understanding such a twin dynamics, the dynamics of the generation of new knowledge and the dynamics of the change of the topology within which agents operate. The issues of dynamic coordination are clearly central in this context and hence the related key notion of governance. Systems where agents are better able to achieve dynamic coordination are likely to experience faster rates of generation of
106 Governance of localised technological knowledge new technological knowledge and hence faster rates of introduction of technological innovations. By means of knowledge governance, missing links among key complementary modules of knowledge can be elaborated and effective alignment of agents towards a common design able to enhance the potential complementarities among the learning agents can be practised. In such a perspective the notion of dynamic coordination plays a key role (Amendola and Gaffard, 1988, 1998).
6 Conclusion: knowledge as a path-dependent emergent property A variety of rival frameworks for understanding knowledge has been elaborated. A long process has been taking place since the old days when knowledge as a public good was the single available framework. A better understanding has been elaborated of the dynamics of knowledge accumulation and dissemination. The specific context in which appropriability and indivisibility apply seems now more and more relevant. Demand and network externalities play a much stronger role now. Transactions in the markets for knowledge do take place, along with technological interactions based upon barter and reciprocity. A variety of governance mechanisms has been designed and implemented. The notion of knowledge as a path-dependent activity emerges as a key acquisition. According to the different characteristics of knowledge, different governance and policy mechanisms emerge as viable solutions. The understanding of knowledge governance mechanisms at the microeconomic level is useful in order to assess both the conduct of agents and the policy of public institutions. Their selective integration into an organised framework is the new necessary step for both economics and economic policy. Firms and public institutions, by means of market transactions or bureaucratic coordination and networking or institutional interventions respectively, can provide the necessary coordination among the undertakings of a myriad of learning agents rooted in their own specific context according to the specific characteristics of knowledge. A systemic approach to understanding the path-dependent mechanisms of the institutional set-up that are most conducive to fostering the rate of accumulation of technological knowledge, as a heterogeneous good, and its distribution and hence to increasing the rate of introduction of technological innovations proves to be the appropriate analytical framework. From the systemic analysis of the interdependent and complementary conditions of access and exclusion to the flows of technological interactions, transactions, coordination and communication that are specifically designed to organise the generation and the distribution of technological knowledge emerges a specific area of investigation and inquiry. Each mechanism and specifically intellectual property rights need to be assessed and considered in this broader framework. The notion of localised technological knowledge and its key ingredients such as the complementarity between external and internal knowledge as inputs in the production of new knowledge, the role of learning, the emphasis on knowledge indivisibility, articulated in knowledge fungibility, cumulability and complementarity,
Models of knowledge and systems of governance 107 the vision of knowledge as a complex product combining a variety of heterogeneous, dispersed and fragmented, yet complementary, components paves the way to understanding the contribution of the notion of complexity and complex dynamics. It is now clear that knowledge can be considered an emergent property of a complex system generated by a path-dependent process where the capability of each agent to generate new knowledge depends both on internal factors and upon the characteristics of the system into which it is embedded, and specifically it is influenced by the number of active connections in place. At the same time, however, it is also clear that the effort to generate new knowledge modifies the structure of the network of interactions in place. Thus the generation of knowledge changes the context. Here complex dynamics leads directly to path dependence. The quality of the interactions in place in a system and the degree of connectivity within the system emerge as key characteristics of a system. Economics of knowledge provides a clear and pertinent field of application of complex dynamics. In the approach to knowledge as a collective process, in the broader context of the model of localised technological knowledge and innovation elaborated so far, it is immediately clear that: 1) agents are supposed to be able to learn in their specific context of action; 2) they are induced by changes in factor and product markets; 3) they can try to make localised efforts to change their current technical and organisational characteristics; 4) such efforts are conducive to the introduction of novel and relevant technological knowledge; 5) the eventual introduction of total factor productivity increases technological innovations only when and if a number of contextual conditions apply; and 6) such a positive outcome in fact takes place only when and where the local context of interaction and its social governance is appropriate to enhance the quality of individual learning and to activate the scope for complementarity and both internal and external economies of scope among the bits of knowledge possessed by each learning agent. The founding blocks of complex dynamics are consistent with such a representation of knowledge dynamics, provided that the path-dependent action of agents endowed with procedural rationality and creativity and hence topological micro-mobility in product and factor markets as well as in the space of knowledge networks and interactions is accommodated. It is now time to turn our attention to the basic tools of the economics of knowledge governance.
6
The new dimensions of knowledge indivisibility Fungibility, cumulability, compositeness and stickiness
1 Introduction Technological knowledge is an interactive and collective activity because it is the result of the intentional interaction and participation of a variety of learning agents, embedded into geographic and professional pools of knowledge commons and because of the intrinsic indivisibility of technological knowledge. Interaction is requested in order to acquire and implement external knowledge, an essential input for the generation of new knowledge. The analysis of the indivisibility of technological knowledge in this context makes it possible to achieve significant progress. The new attention to and analysis of the notion of indivisibility provides a relevant contribution to understanding the complementarity and interactivity of knowledge generation and knowledge dissemination and hence the key role of the mechanisms of knowledge governance at play in an economic system. This chapter provides in section 2 a synthetic account of the new understanding of the non-divisibility of knowledge and identifies four interrelated and yet idiosyncratic aspects, namely knowledge cumulability, knowledge compositeness, knowledge fungibility and knowledge stickiness. The implications of such characteristics of knowledge in terms of increasing returns and knowledge governance are analysed in section 3. Section 4 explores the consequences in terms of multiple equilibria, instability, the dynamic coordination of knowledge commons, and path dependence.
2 The unfolding of knowledge indivisibility Indivisibility has long been considered one of the most problematic attributes of knowledge as an economic good. According to Kenneth Arrow, non-divisibility, together with non-appropriability and non-rivalry in use, had contributed to the understanding of knowledge as a public good (Arrow, 1962a). Our understanding of the non-divisibility of knowledge has made great progress more recently. The analysis of the specific dynamic characteristics of the production processes that characterise the generation and the usage of new knowledge has made it possible to appreciate the differences in the key role of the indivisibility of knowledge in its own generation from the role of indivisibility in the usage of new knowledge. Knowledge indivisibility is defined in terms of complementarity
The new dimensions of knowledge indivisibility 109 of bits of knowledge. Knowledge indivisibility is defined in terms of diachronic and synchronic complementarity of bits of knowledge. Diachronic complementarity leads to cumulability. New knowledge is generated mainly by means of the recombination of both pre-existing and parallel units of knowledge. Such recombination is both synchronic and diachronic. Diachronic, vertical, recombination consists of the reorganisation of elements of knowledge acquired in the past with new bits and insights recently elaborated. Here the Newtonian understanding of the production of science as ‘standing on giants’ shoulders’ identifies a key attribute of knowledge such as cumulability, i.e. the cumulative complementarity between different vintages of knowledge (Scotchmer, 1991, 1996, 2001). When attention is focused on synchronic complementarity, the traditional notion of knowledge indivisibility is articulated here in the more specific notion of knowledge compositeness. The chances of generating new knowledge are conditional on the identification and integration of the diverse and dispersed bits of complementary knowledge that are inputs into the knowledge generation process. The understanding of the notion of ‘modularity’ contributes to this field of investigation. The map of knowledge can be organised in terms of modules. Each module is associated by weak and strong ties of complementarity to others, according to the specific direction of the research process (Loasby, 1999; Nooteboom, 2000; Brusoni and Prencipe, 2001). When knowledge is composite, recombination plays a key role in the generation of new knowledge. New knowledge is generated mainly by means of the combination of both pre-existing and contemporary units of diverse types of knowledge. The composition of technological knowledge varies with respect to the variety of different bits that feed the new composite: there are varying levels of knowledge compositeness. A comparative cross-industrial approach shows that different levels of compositeness characterise the technological knowledge necessary to introduce technological innovations and to run the current business effectively. Technological knowledge is more composite in some industries more than in others. The sources of the knowledge currently used are diverse and yet all need to be kept under control.1 The automobile industry is a clear example of an industry with high levels of technological compositeness. The effective production of competitive cars requires the command and composition of an impressive range of different technologies, including mechanical engineering, electronics, chemistry, electrical engineering, plastics technology, informatics, telecommunications and robotics. The introduction of new technologies in the automobile industry requires the full understanding of the compatibilities and complementarities of each and between each of these technologies. New information and communication technologies themselves are the result of the complementarity among a wide variety of scientific fields, including electronics, telecommunications, space technology, physics, chemistry, plastics and rubber. The new information and communication technological system is the result of the sequential introduction of a variety of complementary and interdependent technological innovations.
110 Governance of localised technological knowledge General systemic technologies emerge when a variety of specific bits of knowledge are drawn together and organised and combined in a new system of understanding. New information and communication technologies provide today a clear example of a new technological system, which emerges on the basis of the identification and valorisation of both synchronic and diachronic complementarities among units of knowledge possessed by a myriad of actors and as such dispersed and fragmented. New technological systems emerge around new organising principles, which make it possible to recombine different bits of knowledge and integrate them into a new single framework (Bresnahan and Trajtenberg, 1995; Lypsey et al., 1998). The chances of generating new knowledge are conditional on the capability of drawing together bits of knowledge that are actually diverse and yet complementary. In turn, with high levels of knowledge compositeness the visible hand (Chandler, 1977) is likely to be more effective in the form of either the hierarchy of large corporations or the provision of public knowledge by public research agencies. When attention is concentrated upon the use of new technological knowledge, a third and quite distinct specification of the notion of indivisibility emerges: fungibility. Fungibility defines the downstream complementarity of any bit of knowledge. Some elements of technological knowledge may apply to a narrow and specific range of activities, either new products or new processes. Other bits of new knowledge can have important applications to a great array of new products and processes. Fungibility is defined and measured by the scope of application of a new bit of knowledge. The same core of technological knowledge and competence can be applied to the production of a wide range of products. This relevant aspect characterises new information and communication technologies, as it did previous general-purpose technologies. New information and communication technologies also have high levels of fungibility, as they apply to a great variety of products and processes. No product and process can be manufactured without the substantial application of new information and communication technologies or without substantial effects of the application of new information and communication technologies (Antonelli, 1992). Biotechnology provides clear evidence about the pervasive role of knowledge fungibility and yet low levels of knowledge compositeness. Biotechnologies apply to a wide range of industries and activities, including pharmaceuticals, food and beverages, pesticides and agricultural chemical products at large. Advances in biotechnology stem from a rather limited range of scientific fields and technological competences. A large part of the twentieth century has been characterised by the high levels of fungibility of mechanical engineering in internal combustion technologies. The same core of technological knowledge and competence has been sequentially applied to the production of a wide range of products, including cars, trucks, buses, armoured vehicles, agricultural machinery, construction machinery, ships and planes. Compositeness feeds the generation of new technological knowledge. If the new composite knowledge has high levels of fungibility a self-reinforcing process takes
The new dimensions of knowledge indivisibility 111 place. New fungible technological knowledge in turn feeds new recombination and hence new steps forward. Finally, a fourth dimension of knowledge indivisibility has emerged. Knowledge stickiness is found when it is difficult to separate knowledge, mainly because of its heavy tacit content, from the human capital and the organisational routines of the unit where learning activities have been taking place and the knowledge has been generated. When knowledge stickiness is high, knowledge is embodied in personal relations. Dissemination and communication of sticky knowledge are strongly determined by the mobility of personnel and organisations. Appropriability of sticky knowledge is easier, because unintended leakage is more difficult, although high levels of inter-company mobility of modules of skilled personnel can compromise its actual levels of control. Knowledge stickiness is defined in terms of embeddedness of knowledge in human capital and routines. When knowledge is sticky it cannot be sold as a product per se. The property rights of the organisations which embody it, however, can become the objects of trade and exchange. In sum, four distinct and well-specified dimensions of the traditional notion of knowledge indivisibility have emerged: knowledge cumulability, knowledge compositeness, knowledge fungibility and knowledge stickiness. All specifications have a direct and clear-cut empirical dimension. The extent to which knowledge is a composite varies and this can be identified and measured with respect to the variety of bits of knowledge that it is necessary to recombine in order to generate a new bit of knowledge. Knowledge fungibility is measured with respect to the number of units of knowledge and products it applies to. The number of vintages of technological knowledge that are required in order to generate and command new technological knowledge easily measures knowledge cumulability. Knowledge stickiness can be proxied by the extent to which knowledge is embodied and embedded in organisational routines and human competence.
3 Knowledge fungibility, cumulability and compositeness as a source of external increasing returns Knowledge indivisibility, when appropriate governance mechanisms are in place, may be the cause of increasing returns. Much progress has been made in economic analysis in understanding the different forms of increasing returns. Different types of increasing returns have different implications for both economic analysis and economic policy. The variety of different types of knowledge indivisibility has strong consequences for the forms of increasing returns stemming from the collective use of technological knowledge. Knowledge compositeness, knowledge cumulability and knowledge fungibility have different impacts on both the partial excludability and the appropriability of the new knowledge and more specifically affect the ratio of the generic component of the new knowledge with respect to the idiosyncratic one. Hence the forms of technological knowledge have a direct bearing upon the definition of the incentives to the generation of new knowledge on the one hand and the contribution of the spillovers of generic knowledge to the generation of new knowledge. When specific conditions of knowledge governance
112 Governance of localised technological knowledge are met, knowledge compositeness, knowledge cumulability and knowledge fungibility are the cause of different forms of increasing returns that have significant and yet different effects on the organisation of economic activity. This distinction has relevant implications both for the theory of the firm and for the economics of knowledge. When and if the access to knowledge is possible, increasing returns take place. As far as knowledge cumulability is concerned it is immediately clear that the powerful notion of economies of density applies. The greater the number of units of knowledge and applications of the same stock of technological knowledge, the lower are its average costs and the greater are the opportunities for knowledge holders to generate new knowledge with small incremental costs. Knowledge cumulability has obvious consequences in terms of increasing returns. The distinction introduced by Baumol and Sydak (1994) between incremental and marginal costs is relevant here. Old vintages of knowledge, like sunk costs, exert a long-lived effect because of their extended contribution to the production activities of new knowledge. The efficiency of the new resources invested in new research ventures can be properly assessed only if all the stock of knowledge available were accounted for. Alternatively it is clear that the new costs are only incremental costs, in that they do not cover the full range of knowledge inputs that are used in any new activity. The stocks of knowledge exert long-term effects, and hence average costs keep decreasing with the extension of the period and scope of usage, with relatively small additions of new incremental research activities. Knowledge cumulability has important effects not only in terms of economies of density but also as a powerful focusing mechanism that selects the directions of the knowledgegenerating activities towards paths that are influenced by the opportunities to take advantage of the complementarity between the new research activities and the characteristics of the existing stock. When knowledge is composite, it is clear that the number of bits of knowledge, which can be recombined, is greater and the chances of generating new relevant knowledge are greater. Here the effects are found on the input side. More specifically the effects run from the variety of inputs towards the levels of output. A special kind of increasing returns where the relationship between inputs and outputs is shaped by the variety of inputs that can be engaged in the process matters here. Traditional forms of increasing returns associated with the quantitative scale of the production are substituted here by increasing returns associated with the variety of inputs. When knowledge is composite, there is a greater variety of carriers of different bits of knowledge that are able to interact and a greater result in terms of the amounts of new knowledge which can be generated. The efficiency of the production is affected by the variety of the specific activities that are brought together. The new understanding of supermodularity can provide many advances in this context. According to Milgrom and Roberts: supermodularity provides a way to formalise the intuitive idea of synergies and system effects – the idea that ‘the whole is more than the sum of its parts’ . . . Supermodularity is mathematically equivalent to the statement that for
The new dimensions of knowledge indivisibility 113 every such x and y, the gains from increasing every component to y1 and x1, is more than the sum of the gains from the individual increases. (Milgrom and Roberts, 1995: 184) Knowledge supermodularity applies to the generation of new knowledge when the positive effects of the increasing number of complementary kinds of knowledge on the efficiency of the generation process are considered. In the case of knowledge fungibility, the effects are found on the other side. Here for a given amount of new knowledge the economic effects are greater as the number of activities to which the new knowledge can be applied is greater. Low costs of replicability play a key role. When knowledge fungibility matters, the greater the variety of the activities which can share the same pool of knowledge, the greater are the possibilities of implementing new technologies and hence the lower are the unit costs. In this case the notion of joint use seems relevant and hence the dynamics of economies of scope. The knowledge pool can be assimilated to a quasi-fixed production factor whose applications to the diverse specific contexts engender low variable and incremental2 costs and almost no wear costs. The greater the number of activities, the greater is the opportunity to spread the quasi-fixed costs. Economies of scope are found when, with a given fixed or quasi-fixed fungible input, costs decline when the variety of outputs increases, because of the opportunity to use it repeatedly. Economies of scope emerge typically when relevant excess capacity is caused by imperfect divisibility and a high threshold of investment and when there is little exclusivity and decay in usage. The same knowledge can be applied to an increasing number of different activities with no or little duplication and wear costs. Both supermodularity in the production of knowledge stemming from the composite character of technological knowledge and economies of scope stemming from the joint use of knowledge are forms of increasing returns where unit costs decline with the variety of activities involved. The number of units engaged in the use or generation of knowledge characterises supermodularity associated with knowledge compositeness, economies of density associated with knowledge cumulability, and economies of scope associated with knowledge fungibility. In all cases, the broader the range of learning agents and activities, the greater are the effects of increasing returns in the generation of new knowledge. Such effects of increasing returns however are partly offset by the costs of the networking and searching activities that are necessary to access and coordinate them. The interplay between the positive effects and the negative effects defines the maximum size of the knowledge commons of collective knowledge. This leads to the very interesting case of bounded increasing returns: increasing returns that apply only within commons of a limited size. 3.1 The working of external increasing returns When knowledge is considered as both an output and an input into the generation of new technological knowledge, knowledge indivisibility has powerful
114 Governance of localised technological knowledge consequences. The actual amount of knowledge generated in the system, even with imperfect appropriability, depends now upon the conditions of access of external knowledge and its role in the generation of new knowledge. Figure 6.1 presents a synthetic account of the issue. Let us assume as a benchmark that, if knowledge were a perfect economic activity, the equilibrium point A, at the crossing between the derived demand and the supply curve, would identify the correct amount of quantities and the correct levels of prices as the result of a standard market process. Imperfect appropriability however reduces the value of the output because of a relevant portion of the output of research activities that cannot be appropriated by each researcher. So far external knowledge coincides with a loss and specifically with a missing revenue from the viewpoint of the agent that produces the knowledge. Consistently with the Arrovian approach this leads to point B, where the system is deemed, because of the lack of incentives, to systematically undersupply knowledge. As the new growth theory stresses, equilibrium conditions would be restored if the losses for producers stemming from non-appropriability match exactly the gains for the users stemming from the spillover of external knowledge. As long as the positive effects of external knowledge in the production of new knowledge are not significantly lower than the negative effects in terms of missing incentives because of non-appropriability, the case for market failure would no longer apply. In the new growth theory approach, the strong natural appropriability of knowledge
VK
C A
MPK″
MPK B KGA MPK′
KGA′
KGAB
KGAA
KGAC
KGA
Figure 6.1 Appropriability conditions and knowledge externalities in the generation of new knowledge
The new dimensions of knowledge indivisibility 115 reinforced by a strong intellectual property rights regime reduces the losses of non-appropriability. Moreover the positive ‘technological’ externalities stemming from knowledge spillovers would be able to offset almost completely the negative effects. Now researchers would have appropriate incentives to undertake the activities that are necessary to generate new knowledge. Users moreover can take advantage of free spillovers. Consistently with the new growth theory the system is able to generate ‘quasi-equilibrium’ levels of knowledge. In the localised technological knowledge approach, knowledge, both internal and external, is indeed viewed as an essential input to the production process of new knowledge: knowledge is both an input and an output. External knowledge however is not free. It can be acquired only at a cost: dedicated resources are necessary in order to make use of it. Now the costs for the production of new knowledge are influenced by the availability and the costs of external knowledge. Pecuniary knowledge externalities apply when the costs of external knowledge are lower than in equilibrium conditions. The effects of knowledge pecuniary and technological externalities moreover depend upon the working of effective knowledge governance mechanisms. The positive effects of both pecuniary and technological knowledge externalities take place only if and when articulated knowledge governance mechanisms are in place. Only when such knowledge governance mechanisms are in place can the system experience increasing returns and hence a net positive result. The missing portion of incentives to generate new knowledge, corresponding to the missing portion of revenue that cannot be appropriated, is more than compensated for by the positive shift effects of the inclusion of external knowledge in the production of new knowledge. The analysis of the role of external knowledge in the production process of new knowledge, both as an unpaid factor and as an input with costs that are lower than equilibrium levels, becomes, clearly, a central issue.3 In Figure 6.1 we see that, when knowledge appropriability is far from perfect, its private marginal product (MPK′) is lower that the social welfare levels (MPK), and hence the funding of knowledge generation activities is lower than is socially desirable (KGAB < KGAA). When, however, the positive effects of external knowledge, in terms of technological externalities, are accounted on the derived demand schedule and on the supply side, in terms of pecuniary knowledge externalities, it is clear that: 1) the costs of the generation of knowledge that combines internal activities (research and development and learning) with external knowledge are reduced for the positive effects of external knowledge used as an input for the production of new knowledge, and hence a new and lower supply curve has to be considered; 2) the marginal productivity of knowledge is far larger because of the repeated positive effects of knowledge applied to generate further knowledge, and hence a new derived demand curve has to be considered (MPK″). The system is able to move towards point C. In this third case we see not only that KGAC > KGAB but also that KGAC > KGAA. This highly circumstantiated event takes place when knowledge can be appropriated by ‘inventors’ for their original purposes and, in addition, it can be used as external knowledge that happens to be easily accessible to third parties and highly productive in the generation of new knowledge.
116 Governance of localised technological knowledge It is important to understand where, when and how external knowledge can be acquired, exchanged and accessed: in which conditions and with the assistance of which circumstances, in which localised context, firms can rely upon external knowledge in order to generate new knowledge. The governance of the production and dissemination of technological knowledge and the dynamic coordination of the efforts and activities of the variety of actors involved emerge here as the key issues to assess the chances and capability of an economic system to generate an amount of technological knowledge that makes growth possible in a competitive context.
4 Multiple equilibria, instability and the governance of knowledge commons Because technological knowledge is a localised and interactive process characterised by the complementarity between both external and internal knowledge and the stock of existing knowledge and the flows of new knowledge, the aggregate outcomes of the knowledge governance mechanisms at the system level are far from being attracted by a single equilibrium point. In these conditions the results of market interactions may or may not lead the system towards stable and fair solutions. The effects of the notion of knowledge as a localised and interactive process are most significant when we compare its implications with the previous approaches. In the Arrovian tradition of the analysis of knowledge as a perfect public good with low appropriability, there are low incentives to fund research and development activities: no agent would undertake the production of knowledge. Market failure is radical and knowledge can be generated only by means of public interventions: hence the creation of public commons. With imperfect appropriability, agents would produce an amount of knowledge that is smaller than the social welfare levels. There is still a market failure, although the welfare losses are now lower and some knowledge can be produced in the marketplace. New growth theory assumes that net positive effects are due to take place because, while externalities spill freely in the air, the appropriability conditions are sufficient to incentivise adequate levels of funds towards research and development expenditures so that the risks of market failure are negligible (Aghion and Howitt, 1998). The tradition of analysis based upon the localised knowledge approach, instead, stresses the specific and stochastic features of the circumstances that make it possible for the positive effects of external knowledge to compensate for the negative effects of non-appropriability (Antonelli, 2001). Proximity in geographic, organisational and technological space, communication conditions and knowledge transaction costs become central elements in the broader context of a knowledge governance approach to grasp the conditions for the dissemination and usage of external knowledge. The assumptions about the relationship between external and internal knowledge become a key issue. With perfect substitutability between external and internal knowledge, opportunistic behaviour can engender a major market failure: firms try to acquire the knowledge spilling in the atmosphere without actually performing
The new dimensions of knowledge indivisibility 117 any internal research. The amount of knowledge available declines and falls well below welfare levels. When the costs of external knowledge are accounted for and the role of internal learning and research and development activities as an essential input are properly considered, instead, firms cannot dismiss the activities leading to the generation of knowledge. The amount of knowledge generated by each firm, now, depends upon the constrained multiplicative relationship between internal and external knowledge inputs. The availability of external knowledge to each firm depends upon four factors: 1) the amount of knowledge that has been generated in the system by other firms; 2) the amount of knowledge generated by the public research system; 3) the levels of knowledge receptivity considered as the result of a specific activity of searching; and 4) the quality of knowledge governance. This system of dynamic interdependences can be modelled as follows: (1) YKi = a(IKi, rEKi) (2) EK = b((A), g((Σ YKt–1), (PRD))) where YK is the knowledge output for each firm, IK and EK are respectively internal knowledge and external knowledge inputs, r measures the receptivity of each firm to the knowledge communication flows at work within the system, and g measures the knowledge governance capability of the system in terms of communication networks and hence the connectivity of the network to provide appropriate dissemination of external knowledge within the system and to provide the necessary levels of dynamic coordination.4 A stands for the non-appropriability conditions of the knowledge generated within the system: it is clear that, if appropriability is very high, little external knowledge can spill and it can only be acquired in the markets for knowledge where vendors are willing to sell it. Intellectual property rights regimes play a key role here in assessing the levels of exclusivity of proprietary knowledge. If appropriability is low, instead, and firms can retain a limited control of the knowledge they have generated, spillovers are relevant and hence the amount of external knowledge available within the system is larger. PRD is the knowledge supplied by public research institutions and universities. The amount of knowledge generated by all the other firms at time t–1 is measured by Σ YKt–1. Equation (2) shows that the amount of external knowledge available at each point in time is directly influenced by the amount of knowledge generated in the system at time t–1. A recursive system is in place. An important result is now obtained. Because of the complementarity between internal and external knowledge, especially if it is specified in terms of a constrained multiplicative relationship, the aggregate outcome of both market transactions and interactions is clearly unstable and sensitive to interactions and subjective decision making. The output of knowledge at the system level depends upon the amount of external knowledge available, but the amount of external knowledge available depends upon the amount of knowledge generated, its appropriability and the quality of knowledge governance mechanisms in place.
118 Governance of localised technological knowledge When both demand and supply schedules are influenced by externalities, multiple equilibria exist (Marmolo, 1999). In these circumstances the effects of path dependence appear immediately clear, especially when the role of knowledge cumulability is considered. When the external knowledge available at each point in time includes the stock of knowledge already generated, the role of historical time becomes very strong. The amount of knowledge that a system can generate at each point in time depends on the amount of resources invested at each point in time and their efficiency and also on the stock that has been generated and its conditions of appropriability.
5 Conclusion The amount of knowledge each firm can generate depends upon the amount of external knowledge available, that is upon the amount of knowledge that other firms, especially when involved in complementary research projects, have generated and cannot appropriate or are willing to exchange. The effects of path dependence are clear. The past dependence exerts a strong influence, yet it is not the single factor at work: at each point in time firms can change the amount of resources invested in the generation of knowledge, new governance mechanisms can be introduced, and lower appropriability regimes can emerge. The amount of external knowledge available at any point in time and in regional and technological space is not determined, but strongly influenced, by the amount of technological knowledge generated and upon the conditions of technological communication within modules of complementary technological knowledge. The market provision of technological knowledge is possible, provided appropriate governance mechanisms are in place, but the levels are undetermined. A very interesting case now emerges: in the markets for knowledge; both demand and supply externalities as well as joint production apply and exert their effects. On the supply side, the amount of knowledge generated depends upon the innovative behaviours of the agents as well as on the general production levels of the economic system at each point in time and in the relevant past, because of the role of learning. On the demand side, as is quite clear, externalities among knowledge users play a central role. The position and the slope of the demand schedule depend on the position and the slope of the supply schedule and vice versa. The latter in turn are influenced by the aggregate conditions of the economic system: learning rates depend upon the amount of output. Needless to say, however, aggregate output is influenced by the amount of technological knowledge generated in the system, via the total factor productivity effects. At each point in time any solution can be found, but such a solution does not have the standard characteristics of stability and replicability. Each equilibrium point is erratic. Little shocks, at the aggregate and disaggregate levels, can push the system far away from any given values. No forces will act to push the system back towards the levels experienced in the previous phase. At the heart of the market system, the production and the distribution of technological knowledge are characterised by multiple equilibria as well as micro–macro feedbacks and thus are sensitive to small and unintended shocks. Macroeconomic or monetary policies
The new dimensions of knowledge indivisibility 119 can have long-lasting consequences if and when they affect the joint supply of experience and competence, and hence they have an impact on the supply of technological knowledge. The strategic decision of firms to increase either the demand or the production of technological knowledge can also have long-lasting effects, changing the parameters of the system. Entrepreneurial action hence may have direct consequences at the economic system level, changing the equilibrium conditions. Either failure or success however can be the result, depending on the outcomes of the chain of reactions which may take place. Economic systems may be trapped in a low knowledge generation regime, while others remain in a high knowledge generation regime. Path dependence, because of the role of learning and interdependence, deploys here its powerful effects. The stock of available knowledge and the systems of knowledge communication in place, at each point in time, catch the effects of past dependence. Small events, however, can push the system to oscillate from one regime to the other, with longlasting consequences. In this context the issues of dynamic coordination among agents and institutions becomes most relevant in order to assess the general outcome of each single action (Amendola and Gaffard, 1988, 1998). The understanding of the role of knowledge cumulability and complementarity and of the interplay between internal and external knowledge in the generation of new knowledge, and hence the analysis of the interdependence and interaction conditions between individual agents and the dynamics of the system, makes it possible to identify the past-dependent attributes of the process of generation and dissemination of knowledge. The generation and dissemination of technological knowledge however are inherently path dependent because they combine the elements of past dependence well articulated by the role of learning and knowledge cumulability with the pervasive role of local externalities at each point in time, well documented in the key role of external knowledge and in the very notion of collective knowledge with the understanding of the role of feedback and strategic decision making by firms. The grasping of the many new facets of knowledge indivisibility has led to the identifying of: 1) knowledge fungibility, defined by the scope of possible applications of a given unit of knowledge as measured by the variety of its possible uses and applications that can be obtained with few incremental and variable costs; 2) knowledge compositeness, defined by the variety of complementary units of knowledge that is necessary to generate a new element of knowledge by means of recombination; and 3) knowledge cumulability, defined by the vertical and diachronic complementarity between the stock of existing knowledge and the flow of new knowledge. All these specifications have important and quite distinct effects in terms of forms of increasing returns and hence mechanisms of governance. Supermodularity and hence increasing returns in the generation and dissemination of technological knowledge can take place only when effective knowledge governance mechanisms are put in place. In these especially favourable circumstances, the negative effects in terms of missing incentives stemming from nonappropriability are more than compensated for by the positive effects of external knowledge as an input in the production of new knowledge.
7
Knowledge and the theory of the firm The interdependence among transaction, coordination and production
1 Introduction Many different and rival theories of the firm have been elaborated and confront each other in the economic literature. Transaction costs economics, the resourcebased approach and the agency theory provide useful inputs to a broader approach. The attention to the role of learning and information impactedness is a common thread. The progress in the economics of knowledge provides an opportunity to elaborate a broader and integrative framework. Such an integrated framework in turn makes it easier to contribute the microeconomic foundations of the economics of localised knowledge. Transaction costs economics has made possible significant progress in the economic analysis of the firm, focusing on the limitations of the marketplace from the viewpoint of the costs of information. In transaction costs economics the firm is viewed as a bundle of activities selected according to the relative costs of transaction and coordination. Inclusion is decided when the costs of using the markets are higher than the costs of coordinating internally the production. The basic choice is whether to buy a given component or other intermediary inputs or to make them. The decision is taken in a static context where coordination and transaction costs are given and depend upon exogenous factors. The role of competence and knowledge is not considered. An alternative view of the firm has been elaborated by the resource-based theory of the firm. The resource-based theory of the firm has emerged as a consistent body of literature centred upon the key role of the learning firm in the accumulation and generation of technological knowledge and competence and its transformation into technological and organisational innovations (Penrose, 1959; Foss, 1997). In the resource-based theory of the firm little attention is paid to understanding the role of coordination costs in limiting the size of the firm and to the constraints and opportunities of the marketplace as an alternative mechanism of governance. In agency theory much attention is paid to the role of information asymmetries between agents both among firms and more specifically within firms. Coordination costs are viewed as specific consequences of relevant information asymmetries between the principal and the agent (Grossman and Hart, 1986). The analysis of coordination and transaction specific activities cannot be conducted in isolation with respect to the choices and the characteristics of the
Knowledge and the theory of the firm 121 production process, the markets for products and intermediary inputs and the specific aspects of the associated learning processes. The decisions of inclusion and exclusion of each specific segment of the production process can be assessed only when coordination and transaction are viewed as the result of well-specified forms of economic activity characterised by their own specific form of competence, technological and organisational knowledge. This makes it possible to move towards a broader economics of knowledge governance approach. The object of analysis in the economics of knowledge governance approach is the organisation of the firm, with a special emphasis upon the localised process of accumulation of technological and organisational knowledge and the introduction of both technological and organisational innovations. The rest of the chapter is structured as follows. In section 2 the comparative assessment of the elements of strength and weakness of transaction cost theory, incomplete contract theory and the resource-based theory of the firm is elaborated as a step towards an integrated economics of knowledge governance. Section 3 discusses the interdependence between production, transaction and coordination and provides an analytical model, which is subsequently applied, in section 4, to grasping the complexities of interdependence in a dynamic context. The conclusion summarises the argument and puts it in perspective.
2 Competing theories of the firm Three different approaches confront each other in the theory of the firm: transaction costs economics, agency theory and the resource-based theory.1 The first two approaches impinge upon the foundations of information economics and, to some extent, even anticipate it. The latter, instead, draws from the Arrovian legacy, as implemented by Edith Penrose. A comparative analysis makes it possible to stress their relative advantages as well as their weaknesses. In so doing it provides the elements to elaborate an integrated approach. 2.1 The contribution of transaction costs economics Transaction costs economics is the result of an incremental process of extension and implementation of the framework first elaborated by Ronald Coase. Oliver Williamson provided an operational context, which proved to be extremely fertile. The unit of analysis here is the transaction. The firm is viewed as a nexus of contracts and a portfolio of given production functions, which coexist within the same organisation according to the trade-off between coordination and transaction costs. The choice of whether to include or exclude a given production process within the boundaries of the firm depends upon the levels of coordination and transaction costs respectively. When the costs of internal coordination are higher than the costs of using the market, a transaction takes place and that production function remains outside the boundaries of the firm. Inclusion takes place when the costs of internal coordination are lower than the costs of using the market (Williamson, 1975, 1985, 1990, 1996).
122 Governance of localised technological knowledge Transaction costs depend upon given technological features such as the asset specificity and the frequency of exchanges, the characteristics of the marketplace in terms of transparency, common trust and actual enforcement conditions of obligations in contracts, hence institutional reliability. The levels of transaction costs mainly consist in the costs of the resources that are necessary to search for possible suppliers of specific components and activities, the assessment of their quality, price and delivery conditions, and the costs of designing effective contracts with the prospective suppliers and enforcing them. Transaction costs as well are an expression of bounded rationality and limited knowledge, but they concern the prospective external suppliers rather than internal agents. In transaction costs economics, neither transaction nor coordination is viewed as an activity, but solely as costs: there is no analysis of the efficiency of the activities, which are put in place in order to perform the required coordination and transaction. There is no analysis of the knowledge and the competence necessary to coordinate and use the markets respectively and hence little room is left for understanding the process of accumulation of new organisational knowledge and the introduction of organisational innovations. By the same token, the technology of the production process is considered as given and exogenous.2 In transaction costs economics the firm does not consider the issues of the choice among technologies, and even less attention is paid to the governance of the accumulation of new knowledge and the introduction of new technologies. The interdependence between technological choices and organisational ones is not considered. Transaction costs economics lacks two relevant aspects. First, the specific aspects of internal coordination activities, as distinct from transactions, are not appreciated. Here however agency theory provides basic guidance. Second, the poor attention paid by transaction costs economics to the conditions and the dynamics of the accumulation and generation of new knowledge and competence is a major weakness. Knowledge and competence applied to the manufacturing processes as well as to the management of the internal coordination and to the procedures and the skills that are necessary to use the markets are key to understanding the firm. A clear understanding of the role of technological and organisational knowledge in the theory of the firm is provided by the resource-based approach. 2.2 The contribution of the agency theory The coordination of diverse activities within the production process entails specific costs associated with the need to control the actual performance of the tasks assigned to the agents and to monitor their efficiency. Coordination costs are specific information costs stemming from the bounded rationality and limited knowledge of managers (Simon, 1947, 1982; Alchian and Demsetz, 1972). Agents and principals have access to different information and elaborate different knowledge. Moreover agents and principals have different goals and objectives. The costs of gathering information and creating information channels are relevant within organisations. The costs of monitoring the behaviour of agents are relevant and increase more than proportionately with the size of operation. The complexity
Knowledge and the theory of the firm 123 of multilayer coordination among the elementary production functions that might be retained within the firm needs to be analysed properly. The path-breaking contribution of Kenneth Arrow (1974) about the limits of organisations and his analysis of coordination as a specific activity within the firm provide a major complementary contribution to transaction costs economics so as to include the costs of internal transactions within the broader framework of analysis (Radner, 1996).3 An important dynamic aspect of agency theory concerns the assessment of the divergent and conflicting interests of principals and agents in appropriating and exploiting the benefits of learning activities and the related knowledge. Learning is intrinsically the joint product of current activities. As such, learning takes place while performing other activities. The appropriation of the benefits of learning is not immediately clear: whether it should be retained by the principal or by the agent. This ambiguity has important implications with respect to the design of organisations and the definition of the appropriate incentives for the intentional generation of knowledge.4 2.3 The contribution of the resource-based theory The resource-based theory provides a distinctive and yet complementary approach to analysing the firm. The emphasis here is put on the process by means of which the firm is able to introduce technological and organisational innovations (Penrose, 1959). The resource-based theory of the firm has grown as a development and an application of the economics of learning. The path-breaking contribution of Edith Penrose paved the way to the economics of learning: the founding contribution of Kenneth Arrow (1962a) follows the 1959 book by Edith Penrose. A wave of contributions has subsequently explored the dynamics and the characteristics of learning processes, such as learning by doing, learning by using and learning by interacting, and has led to the identification of the firm as the primary locus of the generation of new technological knowledge. The generation of technological knowledge can be considered as the distinctive feature of the firm: the firm exists, beyond and actually before the production function, as the institution able to generate technological knowledge and apply it to the production of goods (Penrose, 1959; Arrow, 1962a; Lamberton, 1971; Langlois and Foss, 1999; Loasby, 1999). The firm is viewed as the locus where technological and organisational knowledge is generated by means of the integration of learning processes and formal research and development activities. The firm is considered in this approach primarily as a depository and a generator of competence (Foss, 1997, 1998; Foss and Mahnke, 2000). In the resource-based theory of the firm, the generation of technological knowledge is regarded as the distinctive feature of the firm. The firm does not coincide with the production function and cannot be reduced to a production function because its essential role is the accumulation of competence, technological and organisational knowledge and the eventual introduction of technological and organisational innovations. From this viewpoint the firm precedes the production function: the
124 Governance of localised technological knowledge technology is in fact the result of the accumulation of knowledge and its application to a specific economic activity. Technological knowledge can be considered the primary output of the firm and in turn an intermediary input. The choice of whether to sell it or to use and make with it is especially relevant. In the resource-based theory, the firm cannot be viewed only as a nexus of contracts: the specificity of the production process and the characteristics of the products are a consequence of the process of generation of technological and organisational knowledge. Hence the firm, in the resource-based theory, is much more than a nexus of contracts: it is primarily a mechanism for the production of knowledge. The resource-based theory of the firm however has paid little attention to understanding the role of organisational factors in shaping the accumulation and generation of new knowledge. Specifically, the resource-based theory of the firm has not elaborated a full understanding of the constraints, in terms of both rate and direction, to the dynamics of learning that arise from the costs of using the hierarchies and the markets respectively. Organisational factors shape the valorisation of the knowledge accumulated by means of the learning processes and constrain the direction as well as the rate of learning. The blending of the resource-based theory of the firm, agency theory and transaction costs economics into a fully articulated economics of knowledge governance seems a necessary step to appreciate the key role of localised technological and organisational knowledge in shaping the growth of the firm.
3 The interdependence between production, transaction and coordination The integration of transaction costs economics, agency theory and the resourcebased theory of the firm provides major opportunities for implementing a broader economics of governance. Important complementarities are found when an effort is made to understand the role of competence and knowledge in the definition of the boundaries of the firm, under the constraint of the resources that are necessary to coordinate the diverse activities retained within its boundaries. The generation of knowledge is the primary role of the firm but under the constraint of governance costs. The integration of transaction costs economics, agency theory and the resourcebased theory is possible when attention is focused upon the interdependence between decision making in manufacturing activities and in coordination and transaction activities. In such an approach competence and knowledge are the basic factors necessary to performing the full range of activities. The understanding of the factors affecting the choice between inclusion and exclusion, including the costs of using respectively the markets and the internal hierarchies, is a basic ingredient in a theory of the firm which no longer coincides with the textbook production function. In the economics of knowledge governance the definition of the boundaries of the firm and the choice between exclusion and inclusion are the result of a broad
Knowledge and the theory of the firm 125 range of dynamic factors. The assessment of the inclusion/exclusion choice includes the efficiency of internal manufacturing of components with respect to their market prices, as well as the competence of the firm in performing transaction and coordination activities respectively. The characteristics of the process of accumulation of technological and organisational knowledge and of the endogenous introduction of new technologies and innovations in the governance activities that are necessary to perform transaction and coordination influence inclusion/exclusion decision making, as well as all innovations in the production process. Economics of knowledge governance benefits from the resource-based theory of the firm in expanding the scope of transaction costs economics and agency theory so as to include the analysis of: 1) the accumulation of competence and knowledge; 2) the introduction and selection of technological and organisational innovations; and 3) their effects on the design of the portfolio of activities which are sorted to be respectively included within the firm and assigned to transactions in the marketplace (Penrose, 1959; Chandler et al., 1999). The understanding of the overlapping between production theory, economics of innovation and economics of knowledge makes it possible to provide an integrated analytical framework which is able to study the broad range of factors that affect the governance of the firm viewed not only as a nexus of contracts but rather as a selective and selected combination of complementary activities based upon the capability to accumulate competence and knowledge. In the economics of knowledge governance, the firm is a bundle of activities selected under the constraint of technological, organisational and market factors. Neither factor can be isolated: the actual size of the firm and its structure can be understood only when the three classes of factors are analysed in close conjunction and an effort is made to appreciate their interdependence. Specifically within the boundaries of the firm we can identify production activities, a coordination activity and a transaction activity. The implementation of all activities implies appropriate levels of knowledge and competence and hence of efficiency. The introduction of organisational innovations in coordination and transaction activities and of technological innovations in production in turn leads to increasing their efficiency. The coordination activity provides the management, monitoring and assessment of the relations between the indivisible modules that are retained within the boundaries of the corporation. The transaction activity consists in the use of the markets for the provision of intermediary inputs. The boundaries of the firm are assessed according to the costs of intermediary products internally manufactured relatively to the costs of external inputs. The choice between the exclusion and the inclusion of each input is influenced by an array of factors that are strongly interdependent in assessing the size of the portfolio of activities performed within the boundaries of each firm. The understanding of such interdependence makes possible important progress in the theory of the firm. Firms select the mix of internal and external products and services according to the combined costs of production and coordination on the one hand and the combined costs of purchasing and using the markets on the other. Coordination activities cannot be separated from firms’ own internal manufacturing of the
126 Governance of localised technological knowledge products and services. By the same token transaction activities cannot be separated from the actual use of the market as an alternative means of procuring or selling some products. Thus some substitution takes place. Neither coordination nor transaction activities however can be cancelled. A notion of partial substitutability between coordination and transaction activities emerges. The choice between coordination and transaction, and hence between inclusion and exclusion, can take place, but only up to a point. The traditional analysis of complementary substitutability between production factors, familiar to the theory of production, applies also to the analysis of the governance of firms. This notion of partial substitutability between coordination and transaction activities makes it possible to explore a wide range of mixed governance structures where varying mixes of transaction, production and coordination activities are at work. In so doing the key role of localised technological and organisational knowledge can be fully appreciated. The cost of internal inputs depends upon the sheer cost of the production process of each activity and the costs of their coordination. The cost of external inputs depends upon their market price and the costs of their procurement in using the markets. These decisions however cannot be taken without a clear assessment of the costs associated with inclusion and exclusion respectively (North and Wallis, 1994; Antonelli, 1999b). Both coordination and transaction are resource-consuming activities. Dedicated inputs are necessary to perform the coordination and transaction activities. The usual relationship between inputs and outputs applies. The efficiency of the coordination and transaction activities is determined by the competence accumulated and the organisational knowledge available to each firm. Higher levels of organisational competence may eventually lead to the introduction of organisational innovations, which in turn make it possible to improve the efficiency of both the coordination and the transaction activities (Argyres, 1995; Morroni, 2006). Here the interdependence between the factors becomes evident. At each point in time, for given levels of competence in transacting, the adoption of a technology may be influenced by the levels of the transaction costs that are associated with the asset specificity and the frequency of the transactions that characterise it. With different levels of competence however the firm may select other rival technologies. In this approach the technology of each production process is the result of the innovative choice of the firm itself: the characteristics of each technology are not given and exogenous, but are the result of the innovation and the related accumulation of knowledge and competence within the firm itself. Here it seems clear that the conditions of the coordination and transaction activities affect directly the process of generation and use of technological knowledge and eventually the design and the specific introduction of the new technologies (Loasby, 1999; Nooteboom, 2000; Teece, 2000). The blending of transaction costs economics, agency theory and the resourcebased theory makes it possible to understand the constraints and the limitations that the costs of using the hierarchies and the markets respectively exert upon the accumulation and generation of new knowledge. The firm itself can be regarded as an island of coordination procedures that facilitate the accumulation of knowledge.
Knowledge and the theory of the firm 127 The Coase–Williamson argument, much applied to the choice between coordination and transaction in the organisation of the economic activity, can now be stretched and elaborated so as to understanding the characteristics and the effects of the firm as the fabric of localised technological knowledge (Antonelli, 1999b, 2001). 3.1 The corporate function In standard microeconomics the firm coincides with the production function. In transaction costs economics, coordination and transaction costs define the boundaries of the firm, but no analysis is provided of the activities that are necessary to perform these functions, the role of competence and knowledge, both in the organisation and in the production, and their interdependence. On the contrary, in the resourcebased theory of the firm, learning generates knowledge and knowledge makes growth possible, but little attention is paid to the constraints and limitations of organisational factors. In the economics of knowledge governance, the output of each firm is the result of the combination between internal and external inputs, respectively manufactured, managed, selected, monitored and purchased by means of dedicated activities. Activities in turn are shaped and characterised in terms of competence and dynamic efficiency. The firm is viewed as a micro-system where many interdependent learning activities are at work and influence each other. The governance choice is made according to the costs of external inputs and internal ones. These however are determined by the efficiency of the activities that are necessary to produce them. The governance of the firm can be viewed as the selection of the combination between bundles of production and organisational activities, rather than goods: the selective procurement of external inputs and the production and coordination of internal ones.5 A simple governance system of five equations accommodates the analysis elaborated so far. The working of the firm can be grasped by means of a corporate function and a production function where standard substitution takes place and a transaction and coordination activity characterised by fixed coefficients. Each is qualified by the key role of knowledge and competence modelled as a shift parameter. A standard cost function completes the set of constraints that make it possible to analyse the behaviour of the firm. Formally we see the following: (1) Y = A1(t) ((TRA)α, (CO)β) (2) TRA = (A2)(t) (EXTERNAL, R) (3) CO = (A3)(t) (INTERNAL, R) (4) INTERNAL = (A4)(t) (Ka, Lb) (5) C = p EXTERNAL + uR + rK + wL
128 Governance of localised technological knowledge where Y denotes the output levels that are obtained by means of a corporate governance function characterised by a general level of competence A1(t) that can increase in time and provide the combination of inputs that are either (EXTERNAL), purchased in the marketplace by means of transaction activities (TRA), or internal, i.e. manufactured internally – by means of a standard production function based upon capital (K) and labour (L) inputs as well as specific technological knowledge (A4) which increases over time because of learning processes and dedicated research activities – and managed by means of coordination activities (CO). Coordination activities are the product of the organisational resources (R) that perform the specific task of coordinating the inputs produced internally by means of the production function. Coordination activities moreover are characterised by some dedicated levels of competence and organisational knowledge (A3) that is allowed to change over time because of learning and dedicated research activities.6 Transaction activities also are the output of organisational resources (R) that perform all the clerical tasks that are necessary to purchase in the marketplace the external inputs. Transaction activities in turn are characterised by specific and dedicated levels of competence and organisational knowledge (A2) that changes over time because of learning processes and dedicated research activities. For both activities, a fixed coefficient between the amount of internal and external inputs respectively and the organisational resources (R) that are necessary to perform the coordination and transaction activities is given. It may change over time according to the value of the specific shift parameter that measures the rates of accumulation of dedicated knowledge in each activity and to the informational conditions of hierarchies and markets. The governance function can be characterised by returns to scale that can be increasing or decreasing according to the parameter α and β. The production function in turn can exhibit increasing or decreasing returns to scale according to the value of the parameters a and b. Next to the governance function there is a general cost function where the costs of the external inputs that enter the transaction activities (p) and the unit costs (u) of the organisational resources (R) that enter both the transaction and coordination activities respectively are considered together with the unit cost of capital (r) and labour (w). The working of the governance system is quite simple. For given market prices of the output, the firm will select not only the levels of output but also the portfolio of activities according to: 1) the efficiency of the production process; 2) the effects of increasing returns in production; 3) the efficiency of the corporate function; 4) the effects of increasing returns in the corporate function; 5) the efficiency in the transaction activities; and 6) the competence and hence efficiency in coordination activities. The firm will rely more on external rather than internal inputs when the production function is characterised by a relative inefficiency with respect to other suppliers or when decreasing returns affect its average manufacturing costs, when coordination activities are less effective than transaction activities and hence coordination costs are larger than transaction costs.
Knowledge and the theory of the firm 129 The details of the production process, such as the efficiency of the internal production process and the extent to which increasing and decreasing returns are at work, can be assessed with respect to the prices of the products in the markets. The levels of transaction costs, as determined by the dedicated competence of each firm in using the markets, interact both with the comparative costs of the products manufactured internally with respect to their market prices and with the levels of efficiency of the coordination function. The governance choices are made under the influence and the effects of all the factors that have been considered so far. The quality of the markets, from both an informational and a competitive viewpoint, the characteristics of the products and especially their novelty, the features of the production process with respect both to the levels of asset specificity and to the costs of production, the levels of technological advance in manufacturing with respect to competitors, and the levels of competence in performing respectively coordination and transaction activities are interdependent factors which influence each other and which cannot be separated and isolated in assessing the governance choice of the firms. All changes in the levels of competence and in the knowledge base of the firm are likely to affect not only its conduct but also its structure. The increase in the general knowledge base at the corporate level (A1) as well as the generation of new production knowledge (A4) above the average of competitors will favour the expansion of the boundaries of the firm with processes of vertical integration, diversification and multinational growth. This is also the case when the firm is able to increase its coordination knowledge base (A3). Conversely, the increase of transaction knowledge is likely to push towards the selection of activities retained within its boundaries. The dynamics of the knowledge base becomes the central issue in assessing the evolution of the corporation (Loasby, 2002). The analysis of the interdependence between the laws of accumulation of competence and knowledge, their effects on the production process and the organisation of the bundle of activities retained within the boundaries of the firm makes it clear that at each point in time an equilibrium point between contrasting forces can be identified. Yet the understanding of the dynamics of the learning processes which constitute the essence of the firm, and their effects in terms of the introduction of competence, knowledge and innovations, makes it clear that an equilibrium point is nothing more than a step into a path of continual transformation. All differences in the localised rates of learning and accumulation of knowledge and competence, across the different modules and activities retained within the firm, and with respect to other agents in the marketplace, are likely to change its boundaries and the architecture of the organisation. At the same time it is now clear how the rates of accumulation of localised knowledge in coordination and transaction have a direct bearing on the likelihood of each firm benefiting from the accumulation of technological knowledge and generating successful technological innovations. From this viewpoint, technological change is localised by the interplay between the dynamics of technological learning and the dynamics of organisational learning.
130 Governance of localised technological knowledge
4 Implications The model elaborated so far to handle the analysis of the interdependence between production, transaction and coordination activities is a first result of the attempt to merge the transaction costs economics approach with the agency theory and the resource-based theory of the firm, still in a static context, yet it has many important dynamic implications. The focus upon transaction and coordination primarily as activities which entail specific competencies and dedicated levels of organisational knowledge, rather than sheer costs, has in fact direct and relevant consequences in dynamic terms. Here the variety of firms and their localised endowment of competencies and experience, built by means of learning processes, matter. The firm is no longer viewed as a representative agent. The specific characteristics of each firm need to be investigated and assessed both with respect to the organisational processes and with respect to the production processes. The analysis of production and organisation cannot be separated. The corporation is a resource pool designed and managed so as to implement the opportunities for the accumulation of both new technological and new organisational knowledge. The rates of technological and organisational learning influence each other in shaping the dynamics of the firm, the evolving composition of the collection of activities that are retained within its boundaries and ultimately its growth (Chandler et al., 1999; Teece, 2000). The notions of localised technological knowledge and localised technological change stress the relevance of the learning processes circumscribed in the specific and idiosyncratic locations, within technical, organisational, product and geographical spaces, of each firm at each point in time. The learning processes in such locations are the basic conditions for the accumulation of experience and the eventual generation of both competence and tacit knowledge. On these bases in turn each firm is able to acquire other forms of knowledge, respectively external codified and tacit knowledge, that are contextual and specific to its location and history, and to implement the internal tacit knowledge, acquired by means of a highly idiosyncratic learning process with research and development activities. In this approach, the firm is primarily defined as a specific, contextual and idiosyncratic bundle of activities that are complementary with respect to the generation of knowledge and competence (Antonelli, 1999a, 2001). The characteristics of the process of accumulation of competence, of the generation of technological knowledge and of the introduction of technological and organisational innovations are key factors to understanding the firm. Parallel to knowledge, competence is a central ingredient. Competence is defined in terms of problem-solving capabilities and makes it possible for the firm not only to know how, but also to know where, to know when and to know what to produce, to sell and to buy. Competence and knowledge apply to the full set of activities: production activities, transaction activities and coordination activities (Nooteboom, 2000). The dynamics of the firm is shaped by the dynamic interdependence among the accumulation of localised knowledge and competence respectively in coordination, transaction and production (Chandler, 1962, 1977, 1990).
Knowledge and the theory of the firm 131 The production and direct coordination within the boundaries of a firm of a given product may be determined by the learning economies that are specific and idiosyncratic to the firm. The accumulation of experience and competence, out of learning processes, leads to more efficient production processes. The costs of the internal production are lower than the markets’ prices for the same goods even in competitive markets. The firm will internalise that production even if transaction costs are low and coordination costs are high: production costs matter and interact with the organisational decision making. By the same token, all learning in coordination, i.e. organisational learning, is likely to increase the stock of dedicated organisational knowledge and hence to increase the efficiency of the firm in performing coordination activities (Nooteboom, 2000, 2002). The greater the competence in coordination, the larger is the portfolio of activities which can be retained within the boundaries of the firm. Firms grow into large diversified, integrated and possibly multinational corporations when coordination competencies are good. Firms which have been able to elaborate a distinctive competence in dealing with market transactions, however, are likely to shrink the size of their portfolios of activities conducted internally and to reduce their boundaries. Learning in transaction increases the competence of the firm in using the markets and hence reduces the levels of transaction costs, with the ultimate effect, ceteris paribus, of pushing the firm to reduce the number of activities retained within its boundaries. Firms able to elaborate a distinctive competence in dealing with market transactions shrink the size of their portfolios of activities conducted internally but can extend the scope of their operation as intermediaries (Spulber, 1999). The introduction of innovations in coordination and transaction activities has the direct effect of changing the boundaries of the firm and the composition of the portfolio of production functions retained within the firm. Innovations in coordination lead to an increase in the portfolio of activities, while innovations in transaction activities have the opposite effect. The introduction of an array of innovations in coordination activities, such as the multidivisional form, the matrix structure and in-house outsourcing, have made it possible to reduce coordination costs. Decreasing returns in the corporate function can become a major obstacle for the firm in benefiting from the accumulation of technological knowledge and prevent the successful introduction of technological innovations. Organisation costs limit the growth of the firm, when it is based only upon the generation of technological knowledge – or increasing returns in manufacturing – that is not paralleled by the accumulation of organisational knowledge (Chandler, 1962, 1977, 1990; Arrow, 1974; Bonazzi and Antonelli, 2003). The closer the market prices are to the internal costs of manufacturing, the more relevant is the ratio of transaction costs to coordination costs in defining the boundaries of the firm and the size of the portfolio of activities which it is profitable to include within its boundaries. When the market prices differ from internal manufacturing costs there is a direct incentive to change the boundaries of the firm. Such a difference may depend upon a variety of factors. Two classes can be easily identified: external factors and internal ones. The former concern the
132 Governance of localised technological knowledge conditions of the marketplace, the latter the internal conditions of the manufacturing process. Imperfect market conditions in the supply of inputs and hence market prices that differ from the minimum average cost levels push either towards inclusion, when market prices are above minimum average costs, or towards exclusion, when market prices for complementary products, often in the case of barriers to exit, are below the minimum average cost levels. Internal factors in manufacturing matter as well in assessing the choice of whether to include or exclude. Internal manufacturing costs may differ from market prices for a variety of factors that belong to the idiosyncratic characteristics of the firm. Idiosyncratic increasing returns can explain internal manufacturing costs that are lower than market prices. This is clearly the case when factors of indivisibility and irreversibility, specific to the history of each firm, lead to economies of scale, economies of density, economies of scope and agglomeration externalities. When increasing returns apply to a product, the firm has a powerful incentive to include its production process. Clearly with decreasing returns in the manufacturing of a specific product the firm has a strong incentive towards its exclusion. Industrial dynamics is most relevant here.7 The reduction of barriers to entry and monopolistic power stemming from antitrust interventions or easier access to domestic markets for international competition is likely to push towards the reduction in the size of the portfolio of activities retained within the boundaries of the corporations. Barriers to entry engender a relevant difference between market prices and minimum average costs. Their reduction should push firms to reduce the levels of inclusion and use more effectively the markets for the provision of complementary inputs. High rates of entry and exit are likely to affect the levels of search costs and hence to have a positive effect on the size of the portfolio of activities within the boundaries of the corporation. The size of firms may also reflect the early phases of diffusion of major technological changes. The costs for the search and assessment for new products, especially when they are complex ones and the dominant design has not yet emerged, should push firms to increase their size. This force is especially effective when incremental innovations are still being introduced and the selection environment is characterised by high rates of volatility. When consolidation takes place, a dominant design has emerged and hence the assessment of the quality and performance of the products and their suppliers is easier; transaction costs decline, together with the forces in favour of inclusion. In this context, the quality of markets in terms of trust, information and public knowledge is an essential component of the endowment of social capital (Dasgupta and Seragekdin, 2000). The quality of the markets varies according to their thickness: the number of players on both the demand and the supply sides. Industrial dynamics, in terms of rates of entry and exit, may impose additional burdens in terms of transaction costs even if it has positive effects in terms of the reduction of market prices towards competitive levels. The better the quality of the markets is, from an informational viewpoint, the lower is the amount of search costs to identify the correct price and reliable partners in trade. The levels of opportunism socially accepted and hence the levels of trust that are necessary to stay in the marketplace are clearly relevant. Institutions and norms hence enter on to the scene.
Knowledge and the theory of the firm 133 All changes in the conditions in which transactions, interactions and coordination take place can be considered as changes in the endowment of social capital and in its structure. When the thickness of the markets increases, as well as their informational transparency, and the levels of trust increase, transaction costs decline and hence the use of the markets becomes more effective: the boundaries of the firm shrink. From a comparative viewpoint, the differences, across regions and industries, in the organisation and in the size of the firms can now be viewed as determined by the differences in the thickness, transparency and competitiveness of the markets. Countries and industries where the average size of firms, as well as the scope of their portfolio, is greater are likely to be characterised by lower levels of transparency. The size of firms is smaller, as are the levels of diversification, in industrial districts typically characterised by high levels of trust and transparency, mainly because of a historical tradition of repeated interactions. Large diversified holding companies can be considered the end result of lower levels of informational quality of the markets for intermediary inputs. Firms may also grow into large diversified companies however when coordination competence is very high and is rooted in the national and industrial traditions and institutions: hence it is difficult to swarm elsewhere (Nooteboom, 2002). The introduction of major technological innovations, such as new information and communication technologies, has important implications in terms of organisational innovations. Information and communication technologies have made it possible to reduce the information asymmetries and hence coordination costs. Similar effects however have been observed in transaction costs: e-commerce and especially e-markets seem to make possible relevant reductions in the costs of transactions (Antonelli, 1988). Finally, the complexity and the novelty of the products play a key role in determining the amount of information that is necessary to assess their quality and play a major role in the definition of transaction costs. At times of sweeping introduction of new radical technologies, transaction costs rise. They may eventually decline, together with the emergence of new dominant designs and the selection of new technological platforms. The general rate and direction of technological change may have deep effects on the evolution of general levels of transaction costs in the marketplace.
5 Conclusion Transaction costs economics, agency theory and the resource-based theory have contributed along parallel lines of inquiry on the nature of the firm. In the resourcebased theory, the firm is viewed as a bundle of activities defined by their complementarity with respect to the generation of new knowledge and competence. In transaction costs economics and in the agency theory the firm is a bundle of activities defined by given and exogenous costs of information and transaction. The merging of these research programmes into a broader economics of governance is fruitful from many viewpoints.
134 Governance of localised technological knowledge The integration of the dynamics of accumulation of localised technological knowledge and the dynamics of the introduction of technological and organisational innovations is a necessary step towards a more articulated theory of the firm. The essential understanding of the basic trade-off between inclusion and exclusion, elaborated along the lines set forth by Ronald Coase and developed systematically by Oliver Williamson, can be further implemented in a more dynamical context. The approach to the firm as a bundle of interdependent activities, where the generation of knowledge, production, coordination and transaction are complementary aspects of a broader process of governance, can be developed into a dynamical framework where the firm is viewed as a bundle of activities characterised by localised learning. Such an approach yields useful outcomes in terms of the systemic understanding of the interdependence and reciprocal feedback between different and yet complementary aspects of decision making within the firm. Following Edith Penrose and Oliver Williamson we can claim that the growth of the firm depends upon the capability to generate and use new knowledge, under the constraint of the other factors affecting transaction, coordination and production costs. The analysis of the conditions for the generation, exploration and exploitation of technological and organisational knowledge becomes crucial. Such conditions are not only internal to each firm but also and mainly external to the firm, but clearly internal to the system. Their analysis provides key inputs to understanding both the conditions for the growth of the firm and the growth of economic systems into which firms are embedded.
8
The localised generation and appropriation of technological knowledge
1 Introduction The new analysis of the characteristics of knowledge indivisibility and appropriability makes it possible to appreciate the key role of external idiosyncratic factors in shaping the intentional strategy of firms about the direction of technology strategies. These strategies lead to the introduction of directional technological change based upon the identification and combination of the external sources of complementary technological knowledge and of the idiosyncratic production factors that it is convenient to use intensively both to reduce production costs and to increase the mark-up. The generation of new knowledge is viewed as the result of intentional conduct induced by a specific process that can be successfully implemented only when a number of key conditions apply. Learning and research and development activities are a necessary, but not sufficient, condition for the generation of new knowledge. External factors play a key role in both the intentional generation and the exploitation of technological and organisational knowledge. The combined effect of internal learning, external knowledge and the conditions for exploitation associated with the intensive use of idiosyncratic factors by means of the introduction of biased technological change provides key inputs to understanding the path-dependent features of the knowledge generated by the firm properly embedded in the local context and of successful regional growth. In this chapter the role of knowledge indivisibility and appropriability, in the generation and exploitation respectively of new knowledge, is discussed in section 2. Specifically section 2.1 explores the implications of knowledge indivisibility for the generation of new knowledge. Section 2.2 elaborates the implications of knowledge appropriability in terms of the exploitation of a directional knowledge that makes possible an intensive use of idiosyncratic production factors and hence obtains both a reduction in production costs and an increase in the levels and duration of transient rents. Section 3 provides a simple model that brings together in a single framework the two lines of analysis and spells out the basic methodology for the identification of the focusing mechanisms that shape the direction of the inventive activity of firms. Section 4 explores some of the implications of the analysis in terms of the variety of possible kinds of technological knowledge according to the characteristics of the local contexts and stresses the path-dependent
136 Governance of localised technological knowledge aspects of this dynamic process. The conclusion highlights the role of the selective generation and exploitation of knowledge, as a strategy towards the creation and exploitation of the distinctive competence of firms and regions.
2 The new economics of knowledge indivisibility and localised appropriability The economics of knowledge has made substantial progress in recent years and has provided a new and better understanding of the characteristics and implications of knowledge as an economic activity. The appreciation of knowledge indivisibility has led to a better assessment of the role of external knowledge. The new analysis of knowledge appropriability, based upon the conditions of usage and application to downstream production processes, has shed new light upon the incentives to bias technological change towards the intensive use of idiosyncratic production factors. The combined appreciation of these two characteristics becomes a powerful tool in understanding the criteria by means of which firms select the direction of the generation of new technological knowledge. Let us analyse them in detail. 2.1 The role of knowledge indivisibility in the intentional generation of new knowledge Following the analytical track initiated by Nelson (1959) and Arrow (1962a), the analysis of knowledge indivisibility has been mainly articulated in terms of diachronic indivisibility: new vintages of knowledge build upon the previous advances. Recent advances in the economics of knowledge have made it possible to better appreciate the role of synchronic knowledge indivisibility. The notion of knowledge complementarity has been elaborated in terms of the interdependence between different modules of contemporary knowledge generated, at the same time, by different agents and possibly in different fields (Griliches, 1992). This new thinking about knowledge indivisibility has been reinforced by the understanding that knowledge is not only an output but also an input. Knowledge generated in a specific field and at a specific time for a dedicated purpose is an output. It is also, however, an essential input into the generation of other knowledge both synchronically, e.g. in other complementary fields at the same time, and diachronically, e.g. cumulatively either in the same or in other fields (David, 1993; Stiglitz, 1994). As a consequence it is now clear that no firm can command all the knowledge that is necessary to generate new knowledge. Knowledge external to the firm, at each point in time, is a necessary and relevant complement to knowledge internal to the firm, in order to generate new knowledge. Because of the intrinsic indivisibility of technological knowledge, the successful generation of new knowledge depends upon the access to external knowledge. The generation of new knowledge is the specific outcome of intentional conduct and requires four distinct and specific activities: internal learning, formal research and development activities, the acquisition of external tacit knowledge and the acquisition of external codified
Localised generation of technological knowledge 137 knowledge (see Table 8.1). Each of them is indispensable. Firms that have no access to external knowledge and cannot take advantage of essential complementary knowledge inputs can generate very little, if any, new knowledge, even if internal learning and systematic research and development activities provide major contributions. No firm, in fact, can innovate in isolation (Antonelli, 2001). External knowledge can be accessed by means of a variety of tools, ranging from transactions in the markets for knowledge to an array of interaction modes with public research centres, customers, suppliers and competitors, including the hiring of qualified personnel embodying the competence acquired by means of learning in other companies, the creation of technological clubs and technological platforms, the acquisition of new high-tech companies and mergers with other firms in order to acquire specific knowledge modules (Arora and Gambardella, 1994; Amin and Cohendet, 2005). The acquisition of external knowledge is made difficult by relevant transaction, networking and absorption costs. The cognitive distance among agents, the complementarity in competence and research agenda, the levels of trust, the institutional setting, and the quality of knowledge governance mechanisms put in place are key factors. When external knowledge is available, at low total costs that include the sheer costs of the purchase of knowledge and knowledge transaction and networking costs, the firm needs to manage the absorption and inclusion of external knowledge in the research process. External knowledge is only potentially useful: systematic efforts have to be done in order to take advantage of such possibilities. Only when a corporation is able to fully combine all the learning and research activities conducted within its boundaries with the relevant sources of external knowledge, both tacit and codified, can new knowledge be successfully generated (Cohen and Levinthal, 1990; Teece, 2000). According to the levels of knowledge transaction, networking and absorption costs, the sheer purchasing costs of external codified knowledge and the costs of internal research and learning activities, firms can select the ‘best’ mix of knowledge inputs. It is clear, for instance, that when and where external knowledge is cheap, because of both low purchasing costs in the markets for codified knowledge and Table 8.1 Modes of production of new technological knowledge Tacit knowledge
Codified knowledge
Internal knowledge
Learning
Research and development
External knowledge
Networking interactions with customers, rivals, academics and suppliers Hiring of qualified personnel
Knowledge transactions with knowledge-intensive businesses (KIBs) and universities Purchase of patents and licences in the markets for knowledge Mergers and acquisitions of high-tech start-ups
138 Governance of localised technological knowledge low knowledge transaction and networking costs, firms will rely less on internal learning and research activities. On the contrary, when and where the access to external knowledge is difficult, firms will rely more on internal research and learning activities (Pisano, 1990).1 This approach has two important implications about the costs and the characteristics of the technological knowledge being generated by the firm. First, and most important, for a given budget, firms that have access to cheaper external knowledge can generate a larger amount of knowledge. The unit costs of the new knowledge generated in a fertile knowledge environment are clearly lower than the unit costs of the knowledge generated in a ‘hostile’ context by a single firm that can rely almost exclusively on its own internal competence.2 Second, firms select the characteristics of the technological knowledge they can generate according to the characteristics of the context into which they are embedded. As a consequence, knowledge generated with a strong content of external localised inputs has a stronger idiosyncratic and contextual character (Nooteboom, 2003, 2004). When external knowledge is less available and absorption costs are high, firms generate less knowledge and rely more on internal research efforts. Occasionally, however, they can succeed in implementing a technological knowledge that has a stronger scientific content and applies to a larger variety of technical solutions. Generic knowledge has a wider scope of applicability and generality. 2.2 The role of localised appropriability in the exploitation of knowledge Following the approach elaborated by Nelson (1959) and Arrow (1962a) the analysis of knowledge appropriability has been mainly developed considering knowledge as an economic good per se. Disembodied knowledge can be appropriated both because of high levels of ‘natural’ appropriability and because of the effects of strong and effective intellectual property right regimes. In such circumstances firms have a strong incentive to produce knowledge as a product per se. Markets for knowledge can emerge and regular knowledge transactions can take place, with the well-known positive effects in terms of division of labour, specialisation and efficiency (Arora et al., 2001; Guilhon, 2004). When knowledge is characterised by low levels of natural appropriability that are poorly served by intellectual property rights, localised appropriability becomes relevant. Following a well-established line of analysis of technological change at the system level it is well known that the intensive use of more abundant and hence cheaper production factors leads to a larger increase in productivity (Kennedy, 1964; Samuelson, 1965; Ruttan, 1997; Acemoglu, 2002). Yet little attempt has been made, so far, to integrate this approach – centred upon the analysis of the aggregate direction of technological change – with the analysis of the conditions of usage of knowledge as an incentive towards the selection of knowledge generation strategies at the firm and regional level. The direction of technological change has a strong effect on the results in terms of performance both at the level of the economic system and at the level of the firm (Antonelli, 2003a).
Localised generation of technological knowledge 139 It has been well known for quite a long time that knowledge generation is a localised joint product of manufacturing where learning by doing and learning by using play a key role (Atkinson and Stiglitz, 1969; Rosenberg, 1976; Stiglitz, 1987; Antonelli, 1995). It is now more and more clear that knowledge exploitation is also a joint product of manufacturing activities. The conditions of knowledge usage sharply affect its appropriability: the notion of localised appropriability has important consequences (Antonelli, 2003a). The identification and valorisation of local and idiosyncratic resources that it is convenient to use intensively become a clear and strong focusing device along which firms can align their research activities. The conditions for the exploitation of knowledge feed back to the generation of new technological knowledge (March, 1991; Nooteboom, 2003). Knowledge can be exploited and better appropriated when its application and usage impinge upon selective production factors characterised by asymmetric access conditions. The productivity of new technological knowledge, when applied to the actual production process, and the appropriability of the economic value stemming from its use are much influenced by, respectively, the relative price and the conditions of access to the production factors being used. Firms that are able to identify idiosyncratic production factors upon which they exert a specific control that enables low purchasing costs can direct the introduction of new technologies so as to increase their role in their production process and make an intensive use. The local social value of directed technological change is higher. When such local factors are idiosyncratic and other firms cannot use them on the same conditions, innovators are able to extract much higher rents from their knowledge generation activities for a much longer period of time. In this case the innovator can increase the private value of the innovation.3 Schumpeterian market dynamics provides the basic elements to fully understand the mechanism at work. Since The theory of economic development by Joseph Schumpeter it has been well known that innovators can take advantage of a monopoly power that is, however, transient. Extra profits associated with the introduction of successful innovations stimulate the imitative entry of newcomers. Increased competition drives price–cost margins to minimum levels. In this context, the more specific the technology introduced by innovators is, i.e. the more it makes possible an intensive use of idiosyncratic production factors that are specific to innovators, the less likely is the possibility that newcomers, even when and if they succeed in grasping the new technological knowledge and imitate the new technology, will be able to match the production costs of innovators and hence reduce their competitive advantage. In such a marketplace the competitive advantage of innovators is based more on the biased mix of idiosyncratic production factors that have shaped the direction of technological change than on the exclusive command of proprietary technological knowledge. Even if new competitors can imitate the new idiosyncratic and localised technology, their production process will be less effective than that of innovators because of the differences in the costs of production factors.
140 Governance of localised technological knowledge Innovators relying on idiosyncratic production factors can command a cost advantage upon which long-lasting barriers to entry and to mobility can be built. Each innovator becomes the local monopolist in a well-defined market niche. The size of the niche depends upon the specification of the products with respect to the preferences of consumers and upon the cross-price elasticity with respect to other similar products. The latter in turn are built around the idiosyncratic competences of other competitors. Innovators will fix strategic prices in the niche according to the ease of mobility and entry of the competitors in the broader basket of niches competing for the demand of similar customers and the levels of cross-price elasticity, that is the mobility of customers across the different niches. In sum, when the generation of new knowledge is directed towards the introduction of new biased technologies that consist in the intensive use of locally abundant production factors so as to reduce production costs, the local social value of technological knowledge is greater. The private share of such a greater social value is larger when barriers to entry and imitation, based upon the intensive use of idiosyncratic production factors, prevent the dissipation of the economic rents stemming from their introduction and hence increase de facto knowledge appropriability. The new understanding of the role of localised knowledge appropriability leads to a new appreciation of the idiosyncratic character of production inputs and its productive and competitive effects. The search for new, more effective, uses of locally abundant production factors is a powerful alignment mechanism for the research strategies of innovators and a strong incentive to the generation of directed technological knowledge. The biased production technology that makes the most intensive use of locally abundant and hence cheaper production factors is more efficient and profitable, as it engenders systematic cost asymmetries that are long lasting, when competitors do not have access to the same factor markets.
3 The emergence of the direction of technological change The appreciation of the constraints and opportunities provided by knowledge indivisibility and localised appropriability makes it possible to frame in a single framework the analysis of the incentives that contextual and localised factors exert in shaping the direction and the characteristics of the new knowledge generated by firms. Although learning localises the cognitive base of firms in a limited spectrum from the original focal point of activity, there are still many possible directions along which the generation of new technological knowledge can be aligned. The choice among an actual array of possible knowledge modules becomes a crucial issue. At each point in time the firm has a variety of possible directions towards which the creative activities can be ordered. Each needs to be assessed and the relative profitability needs to be valued from the viewpoint both of the costs of introduction and of the revenue stemming from its application. As Figure 8.1 shows, at each point in time the learning firm has the opportunity to move in a Lancastrian (Lancaster, 1971) space of knowledge characteristics
Localised generation of technological knowledge 141 and related technological characteristics, branching out from the original point A to a variety of points, B, C and D. Each of the new points exhibits an improvement and a change and it is the result of the generation of new technological knowledge and of the introduction of new technologies. In the subsequent unit of time, t2, the learning firm has the opportunity to further branch out from the new points B, C and D: towards the points E and F if it had reached point B at time t1, the points G and H if it had reached point C at time t1, and points I and L if it had reached point D at time t1. The theory of the firm, so far, is able to explain retrospectively why and how the learning firm has moved from any of such points to the next and, indeed, each of the points generated sequentially is related to the previous vintages by clear elements of cumulability along a technological trajectory. The direction of the selected trajectory however can be identified only ex post. From an ex ante perspective the theory of the firm does not supply any strong analytical support to elaborate possible hypotheses about the direction of the future steps (Dosi et al., 2000). Here an important step forward can be made if the factors that constrain the selection of the direction of the sequential steps and act as focusing mechanisms are identified and analysed within a single framework. The new characteristics of knowledge indivisibility and appropriability, e.g. knowledge complementarity and localised appropriability, make it possible to identify such focusing mechanisms. Our basic argument can be spelled out as follows: firms have an incentive to direct the generation of new knowledge according to the contextual spillovers of complementary knowledge and the localised conditions of appropriability. The analysis of knowledge indivisibility has made clear how relevant the access to external knowledge is. The analysis of knowledge appropriability as a joint product of manufacturing has stressed the key role of the intensive usage, by means of the introduction of biased technological changes, of local and idiosyncratic
X1 E
F G
B C
H I
A D
L X2 Figure 8.1 The direction of the generation of knowledge
142 Governance of localised technological knowledge production factors. The combination of these two complementary arguments makes it possible to identify in the local context a powerful focusing mechanism of the direction of new technological knowledge. Let us start from a population of heterogeneous firms, distributed in different regions, that have access to different sources of knowledge and factor markets. Each firm exposed to a mismatch between beliefs and related plans, and the eventual factor and market conditions, is pushed to generate new knowledge and to introduce new technologies. This creative reaction requires dedicated activities: the valorisation of internal learning, the conduct of formal research and development activities, and the acquisition of external knowledge, both tacit and codified. Such activities entail the assessment of specific costs such as the costs of coordination of the valorisation of internal learning, the knowledge transaction costs necessary to purchase codified knowledge in the markets for knowledge, and the networking costs necessary for implementing the acquisition of external knowledge, both codified and tacit. Even tacit external knowledge does not spill freely in the air: its acquisition is itself the result of intentional activities. Relevant absorption costs add to the costs of external knowledge.4 The selection of the kind of technological knowledge is affected both by the conditions for its generation and by the conditions for its exploitation. Each firm has a clear incentive to direct the generation of new technological knowledge towards applications that make it possible to make an intensive use of locally and internally abundant factors that have a strong idiosyncratic character. In so doing, firms can generate more knowledge, produce at lower costs and take advantage of barriers to imitation based on idiosyncratic production conditions. In a heterogeneous population of firms engaged in the effort to generate new knowledge and appropriate its benefits in terms of extra profits, each firm can discover that the alignment of their internal research activities with complementarity research activities of other co-localised firms and the identification of the local idiosyncratic production factors are a powerful factor of competitive strength. The creation of distinctive competences requires the systematic exploitation of the characteristics of the local context. By the same token a consistent directionality of technological knowledge is the collective result of a process of convergence of co-localised firms able to integrate their own specific competence with the features of the local context. This analysis can be understood as a discovery process articulated upon the sequence of three steps. Localised learning is the basis for the generation of new knowledge. As such it consists of enriched techniques, typically found in the technical region where each firm is localised. Other basic sources of knowledge enter the knowledge production function: research and development, external codified knowledge and external tacit knowledge. The costs equation includes the costs of internal research and learning activities, including coordination costs, the total costs of external codified knowledge, including knowledge transaction, and networking costs for external tacit knowledge. We shall assume that each firm has a budget available to fund activities for the generation of new knowledge.5 Let us specify the following knowledge production function and knowledge cost equation:
Localised generation of technological knowledge 143 (1) KN = (R&D&La, EKb) (2) TCK = (jR&D&L + zEK) where KN is the knowledge generated, R&D&L is the internal research, development and learning activities to generate new knowledge, j is their unit costs, EK is the external knowledge and z is its unit cost, which includes the pecuniary costs of purchase, when possible, and the costs of knowledge transaction, networking and absorption activities that are necessary in order to acquire and use it. With standard maximisation procedures it is immediately clear that the lower the unit costs of external knowledge are the greater is the amount of knowledge that the firm is able to generate and the greater is its localisation in the specific context.6 A firm that is located within an effective knowledge network, and is able to identify and access the local pools of knowledge at low costs, is induced to take advantage of it and hence to root the generation of its new knowledge in the characteristics of the environment into which it is based (Nelson, 1982; Patrucco, 2008).7 It is clear that the amount of knowledge generated when z < j and firms are able to align their research strategies so as to take advantage of locally abundant knowledge is greater. Consistently in the downstream applications firms can rely upon a greater increase in efficiency with the same amount of budget available to fund the generation of new knowledge. The amount of knowledge generated has a direct effect upon the general efficiency of the production function: (3) A = l(KN) where A measures the general efficiency of the production function, i.e. the neutral shift towards the origins of the map of isoquants. It is clear that the larger KN is the larger is the shift effect A. Firms that are able to take advantage of the local pools of collective knowledge can produce at lower costs. The amount of external knowledge that has been used in the knowledge generation process has a direct bearing not only upon the amount of knowledge being generated and hence on the shift efficiency engendered in the production process but also on its characteristics. Firms that rely more upon internal knowledge are more likely to produce generic knowledge. Firms that rely more upon external knowledge are more likely to produce idiosyncratic knowledge. This amounts to saying that the generation of idiosyncratic technological knowledge leads to the introduction of idiosyncratic technological change that shapes the production function in such a way that the output elasticity of idiosyncratic production factors (I) is much higher than the output elasticity of generic production factors (G). This is convenient when, for the innovating firm, locally abundant production factors are available at a price (r) that is lower than the price of the other production factors (p), i.e. when r < p. Conversely the introduction of generic technological change has no effect on the ratio of output elasticities. In other words the generation of (more) generic knowledge leads to the introduction of a (more)
144 Governance of localised technological knowledge neutral technological change with no modifications in the output elasticity of the production factors G and I. To make this point clear let us consider a standard production function prior to the introduction of the new technology: (4) Y(t) = (IE GF) where I and G are respectively the idiosyncratic and generic inputs, and E and F measure their output elasticities. After the introduction of respectively generic and idiosyncratic, hence biased, technological changes, the new alternative production functions can be specified as follows: (5) Y(t+1)g = A (Iu Gv) (6) Y(t+1)l = A (Is Gt) (7) C = rI + pG where at time t+1 after the introduction of the new technology Yl is the production process that uses idiosyncratic technological knowledge and Yg is the production process that uses generic technological knowledge; u, v, s and t measure the different output elasticities, with u = E, s > E and, possibly, s > t when r < p. Let us now consider the effects of the alternative directions of technological knowledge in terms of knowledge exploitation. When factors are not equally abundant in each local factor market, it is clear that the unit costs of the goods manufactured by means of an intensive use of locally abundant factors are lower than the costs of the goods manufactured with inputs that are available to every firm at the same price. On top of this, we see that the efficiency of the two production processes differs because of the greater amount of knowledge that has been generated by the firms that have better access to external knowledge and are better able to take advantage of it with the introduction of a bias in the direction of their knowledge. The working of the two mechanisms is consistent, and clearly the average costs of the goods that have been manufactured with an idiosyncratic technology (CYl) are lower than the average costs of the goods that have been manufactured with a generic technology (CYG): (8) CYl < CYG Finally, we consider the price at which the goods that have been manufactured with the new technologies can be sold. The products manufactured with a more idiosyncratic technology that make a more intensive use of the locally abundant factors, including those internal to the firm, and that are not available on the same conditions to competitors enjoy systematic cost asymmetries with respect to imitators and hence can take advantage of substantial barriers to entry and to
Localised generation of technological knowledge 145 mobility. In product markets characterised by monopolistic competition, incumbents protected by barriers to entry and to mobility can fix high prices for their products, far higher than those of competitors. This is not the case when technological change is generic: every firm can use production factors that are not idiosyncratic. Hence new competitors can imitate the new technology and their entry drives the prices to competitive levels. Clearly the prices of products manufactured with a higher intensity of idiosyncratic inputs (PI) are higher than the prices of the products manufactured with a low intensity of idiosyncratic inputs (PG). In these circumstances, moreover, the mark-ups of idiosyncratic-intensive firms are likely to last longer. Equations (5) and (6) can be combined into the traditional frontier of possible production: (9) YG = e(YI) The solution to the optimisation problem is easily found with an isorevenue that defines the possible revenues that can be earned with the alternative production functions considered. The slope of the isorevenue measures the ratio of the prices of the products manufactured with a new generic technology (PG) to the prices of the products manufactured with a new localised technology (PI). The equilibrium is found where: (10) dYg / dYl = PI / PG Clearly there are stronger incentives to select the mix with more biased technologies than generic ones. A simple geometric exposition can help in grasping the point. As is shown in Figure 8.2, the shape of the frontier of production possibilities, which considers the trade-off between the levels of output Yl which can be attained with the introduction of a new technology that makes intensive use of locally abundant and idiosyncratic production factors and the levels of output YG that can be attained with the introduction of a new technology which uses generic production factors, is clearly asymmetric. Moreover the slope of the isorevenue, much smaller than 1, reflects the positive effects for innovators of the price asymmetry with respect to imitators, which do not have access to the same idiosyncratic production factors. Optimisation clearly favours the introduction of a mix of technologies based upon the intensive use of locally abundant and idiosyncratic production factors. Firms able to select their technological innovations so as to introduce a bias in favour of the creation and subsequent intensive use of idiosyncratic production factors have a larger mark-up because of four factors: 1) lower research costs; 2) lower production costs; 3) higher product prices; and 4) barriers to entry and imitation lasting for a longer stretch of time. In sum, the generation of technological knowledge and the eventual technological change is directed by: 1) the conditions of access to local pools of external knowledge and participation in commons of collective knowledge where
146 Governance of localised technological knowledge
Y1 PG P1
Y Figure 8.2 Output- and revenue-maximising incentives to make intensive use of idiosyncratic innovations
interactions and transactions are shaped by proximity; 2) the cost-reducing use of locally abundant production factors; and 3) the profit-increasing use of local idiosyncratic production factors. According to the value and weights of these parameters, the characteristics of new knowledge and the direction of technological change (see Table 8.1) can be assessed ex ante.
4 Implications for empirical analysis and technology policy The implications of the analysis are far reaching. The analytical frame suggests that the appreciation of the characteristics of the region in which firms are rooted plays a key role in the implementation of successful knowledge strategies. For the firm, this implies that the identification and exploitation of the sources of external knowledge and of the idiosyncratic production factors that it is more convenient to use intensively, contribute to shape the technology strategy. The analysis of localised spillovers and contextual appropriability becomes a key component in the creation and exploitation of the distinctive competence of the firm. For the region, this provides basic guidance for understanding the factors upon which a successful technology policy can be implemented. The appreciation of the key role of the regional context where technological knowledge is being generated opens new prospects of empirical inquiry about the variety of types of knowledge that different groups of firms, localised in different contexts, have an incentive to generate. The key role of the direction of technological change, as a fundamental aspect of the innovation process, makes it possible to understand that a variety of paths to innovation can persist successfully. According to the local endowments, articulated in material inputs, skills and mechanisms of knowledge governance, firms have clear incentives to identify a
Localised generation of technological knowledge 147 specific typology of technological knowledge and the ensuing technological innovations. In a heterogeneous system, where local endowments differ, firms do not compete on the same knowledge frontier but, on the contrary, have a strong incentive to identify the kind of technological knowledge that is better appropriated to their own specific conditions and traditions. Such specific conditions are not only internal to each firm, as the resource-based theory of the firm contends, but also external. As a consequence, a variety of paths to technological change is likely to emerge and consolidate. Firms based in countries and regions with a stronger scientific infrastructure have an advantage in the introduction of science-based technologies. This is not necessarily the case for firms based in countries where the endowment of human capital is lower or different. The distinction between skills acquired on the job and skills based upon formal education, for instance, has important consequences. The specific characteristics of the industrial structure also play a major role here. Firms based in countries and regions specialised in capital goods have a structure of incentives to align their knowledge-generating activities that differs from that of firms based in countries specialised in final goods. In the globalising learning economy, regions have a strong incentive to pursue dedicated and specialised knowledge strategies based upon their own endowments in terms of both knowledge generation and knowledge exploitation mechanisms (Archibugi and Lundvall, 2001). The well-known positive effects of the division of innovative labour in terms of specialisation, trade and increased efficiency that are possible at the firm level when knowledge has high levels of ‘natural’ appropriability and intellectual property rights are effective can be reaped, even when the ‘natural’ appropriability of knowledge is low, in a global economy, at the regional level. This is possible when and if the local commons of collective knowledge are implemented and enriched within regions by the consistent research strategy of firms able to understand the key role of contextual spillovers and localised appropriability so as to generate and introduce, with the support of local government, technological knowledge and hence technological innovations with a strong local content and a clear directionality. Both at the firm and at the regional level these processes are likely to take place with a strong non-ergodic and sequential stratification (David, 1994b). At each point in time, each firm, based in a region where knowledge transaction, networking and absorption costs are low, is induced to take advantage of available external knowledge and hence to select a direction for the generation of new knowledge that reflects the proximity to firms localised nearby in spatial, cognitive and technological space. Additional flows of external complementary knowledge become available in the same space and reinforce the process. At the same time, better knowledge governance mechanisms are likely to be implemented and stronger communication channels among firms are likely to emerge. Moreover the systematic implementation of complementary research strategies is likely to have positive effects on knowledge absorption costs for each firm. Larger and more accessible commons of collective knowledge consolidate and push innovating firms to rely more systematically upon external knowledge. The process is further
148 Governance of localised technological knowledge reinforced by the selective use of idiosyncratic production factors. A larger derived demand for local factors is likely to favour their additional supply with further benefits in terms of specialisation, especially when distinctive skills are a core, idiosyncratic input.8 This new progress in the analysis of the effects of the characteristics of technological knowledge helps in grasping its path-dependent dynamics (David, 2000a, 2000b). The new thinking on knowledge indivisibility stresses the role of the external environment in the form of the conditions of access to the local pools of collective knowledge. The new thinking on knowledge appropriability stresses the role of exploitation as a joint product of manufacturing and hence of biased technological change. The path-dependent dynamics stems from the interplay between past dependence and intentional action. Learning and the features of the local pools of knowledge and of the economic structure are the past-dependent components, as they are at each point in time the result of historical accumulation. The interaction of agents with the existing past-dependent features, in terms of the amount of knowledge and the direction of technological change being generated, provides the opportunities for intentional action to change the original path. At each point in time the intentional action of the embedded agents adds a new layer to the original structure: the original shape exerts an effect that the new layers can modify, however, according to their thickness and density. At each point in time the generation of new knowledge by each firm is influenced by the dynamics of internal learning, by the structure of local interactions that shape access to the knowledge generated within the system and by the structure of local endowments. Each firm, however, is able to interact with the system and change it. This takes place at three levels: by changing the amount of knowledge made available to the other firms, by changing the structural conditions of the systems in terms of knowledge governance mechanisms and topology of communication channels, and by changing the factor markets by means of changes in the derived demand for production factors.
5 Conclusion The new understanding of knowledge indivisibility and knowledge appropriability provides important clues to assessing the selection of the direction of technological knowledge. The generation of new knowledge is not the automatic and spontaneous product of internal learning processes. Internal learning is a necessary, but not sufficient, condition for the generation of new knowledge. Because of intrinsic knowledge indivisibility in terms of both knowledge cumulability and complementarity and its twin character of being both an input and an output, the acquisition of technological knowledge external to each firm is a necessary and indispensable activity in the generation of new knowledge. Hence firms are pushed to select the generation of new knowledge so as to make the best use of external knowledge that is available within the local knowledge networks. When ‘natural’ knowledge appropriability is low, innovating firms have a strong
Localised generation of technological knowledge 149 incentive towards the generation of technological knowledge that makes possible the introduction of an intentionally biased direction of technological change. The intensive use of local and possibly internal production factors that are highly idiosyncratic and hence cheaper for a limited number of users favours both the productivity of new biased technologies and their profitability, because it reduces the risks of imitation by rivals that do not have access to the same factor markets. Such inputs are made idiosyncratic to the innovating firm by both the selection of locational factors and their intentional creation. Thus firms build a localised appropriability. Strong positive effects, in terms of reduced knowledge generation costs stemming from knowledge complementarity, reduced production costs engendered by the ensuing technological innovations that make an intensive use of locally abundant factors, and increased knowledge appropriability based upon vertical integration in the downstream production process based upon the systematic and intensive use of idiosyncratic – either locally available or internally created – production factors, provide a clear incentive to direct the generation of new knowledge according to the local knowledge networks and endowments. This strategy can exert positive effects not only on the growth of firms but also upon regions able to implement the local pools of collective knowledge by means of effective knowledge governance mechanisms. As a consequence, technological knowledge acquires a strong idiosyncratic character that is indeed influenced by the internal learning processes, but also by the local structure of knowledge interactions and the characteristics of local factor and product markets.
9
To use or to sell technological knowledge
1 Introduction The grafting of the recent advances of the theory of the firm on to the economics of knowledge seems a promising field of investigation and cross-fertilisation. The result can be relevant both for the economics of knowledge and for the theory of the firm itself. Economics of governance has benefited from the resource-based theory of the firm so as to include the analysis of such processes as the accumulation of competence and knowledge and the introduction and selection of technological and organisational innovations, as well as the assessment of their effects on the design of the portfolio of activities which are sorted to be respectively included within the firm and assigned to transactions in the marketplace. The dynamics of the firm is shaped by the dynamic interdependence among the accumulation of localised knowledge and competence respectively in coordination, transaction and production. The characteristics of the process of accumulation of competence, of the generation of technological knowledge and of the introduction of technological and organisational innovations are key factors to understanding the firm. Parallel to knowledge, competence is a central ingredient. Competence is defined in terms of problem-solving capabilities and makes it possible for the firm not only to know how, but also to know where, to know when and to know what to produce, to sell and to buy. Competence and knowledge apply to the full set of activities: production activities, transaction activities and coordination activities. The fabric of the economics of governance however can be extended further so as to include in its weave the broader array of issues related to the alternative means of exploitation of technological knowledge. The attention of governance economics has been traditionally concentrated upon the ‘make or buy’ alternative. More and more attention is now being paid to the symmetric and yet complementary issue of the ‘use or sell’ alternative. The ‘use or sell’ alternative requires the assessment of the relative costs of using the markets to sell a product with respect to the costs of using that product in a further stage and eventually selling it embodied in a more elaborated product. The inclusion/exclusion choice concerns whether to sell the output at a given stage of the production process or to use it to make another product. The firm is seen here as an intermediary between production stages coordinated by the marketplace (Spulber, 1999).
To use or to sell technological knowledge 151 The ‘use or sell’ trade-off applies to a wide range of industries where the general production process is broken down into production units characterised by high levels of specialisation and technical indivisibility. The selective internalisation of modules highlights the choice between the direct sale of the goods of a given module into the intermediary markets or the use of its products to feed further sequential manufacturing modules. On a similar ground the extensive growth of the service economy and the decline of manufacturing industry at large seem characterised by the intensive externalisation of knowledge-intensive business services. Again firms decide whether to sell their knowledge as a service or embody it into goods, through internalised manufacturing activities. The ‘use or sell’ choice seems most relevant from the viewpoint of the integration of transaction costs economics and the resource-based theory of the firm into the broader context of the economics of governance in general. This approach becomes extremely fertile, for its analytical and normative implications, when attention is paid to technological knowledge as a good which can be either sold as such in the markets for technology or used as an intermediary internal input. The chapter explores the context of this trade-off in the analysis of modularity in manufacturing in section 2. Section 3 provides an assessment of the ‘use or sell’ trade-off to the governance of technological knowledge, and on this basis a model is developed. The implications for knowledge exploitation strategies are considered in section 4. The conclusion summarises the results of the analysis and puts it in perspective.
2 Supply-side transaction costs Modularity is swarming into economics and the economy. The architecture of the production process is more and more articulated in an array of modules of quasiindivisible production units. Modularity is the result of the intensive and systematic breakdown of complex production processes into self-contained production units where technical indivisibility is stretched to minimum levels (Baldwin and Clark, 2000). The governance of the transactions between modules is more and more complex. Modules can be included within corporations or excluded, according to the economics of governance as dictated by the assessment of the costs of using both the inputs and the product markets with respect to the costs of internal coordination. The coordination of the supply and the demand for the products of the modules can be either left to the marketplace or provided by means of bureaucratic organisation within the firm. This process is well documented in many industries. In the automobile industry, as well as in most engineering industries, the traditional vertical integration has been supplanted by a variety of interdependent players which contribute to the general production process with their specialised services and products. The general production process leading to automobiles, as well as appliances and machinery, has been broken down into a variety of complementary modules of specialised units which sell and buy in the marketplace their components and their products. The
152 Governance of localised technological knowledge inclusion and exclusion of these units may take place at different stages of the general production process and are frequently reassessed (Bonazzi and Antonelli, 2003). Traditional industries such as textiles, garments and the fashion industry generally have a long-standing tradition of modularised production processes. The evidence shows that the inclusion and exclusion of the modules in these industries is no longer organised along the sequential lines of the general production process but rather according to systematic and selective evaluation of the incentives to buy or make and to make or sell applied to each stage of the production process and each intermediary product. The design of the structure of each company gets closer and closer to a flexible jigsaw where each element enters and exits the company. The correspondence between the structure of the corporation and the layout of the general production process leading to the final products purchased by households is more and more fuzzy. The communication industry provides excellent evidence on these processes. Following Fransman (2002), in the communication industry six layers of specialised activity can be identified: equipment and software; networks; connectivity; navigation and middleware; applications; and content packaging. The specialisation of firms into each layer depends upon a variety of factors. The levels of transaction costs in each market differ widely, as well as barriers to entry and levels of markups. Sunk costs characterise each firm in different ways. The dynamics of the institutional factors affecting the definition of property rights such as the forms and levels of mandatory interconnection and the evolution of de facto and de jure standards plays a major role. The boundaries of the firms with respect to the layers vary according to their own specific and idiosyncratic characteristics. In the communication industries a great variety of firms can be found with respect to the mix of layers in which they are active. Fully vertically integrated firms coexist next to others that are fully specialised in just one layer. Many firms select specific combinations of layers and they systematically operate in the intermediary markets as sellers and/or buyers (Krafft, 2003). In this context the characteristics of intermediary markets play a key role. The product of each module is sold and bought in markets that are far away from the final consumers. In intermediary markets customers are firms. The same firms are also sellers. Often the same firm is at the same time a customer and a supplier. Because of modularity and inclusion/exclusion practices, intermediary markets are more and more thick, with an increasing number of players on both the demand and the supply side. The relative efficiency of intermediary markets from both an informational viewpoint and a competitive one now plays a key role in the design of the portfolio of activities that are retained within companies. The firm considers whether to make or buy a specific component or stage of the production process, and also whether to sell the products of each module in the intermediary markets or to use the output of the same module as an intermediary input for the following production process and eventually deliver the product to the final markets where households are the customers (Teubal et al., 1991). Such transaction activities must be considered not only on the demand side, as in the ‘make or buy’ tradition, but also on the supply side. Transaction costs on the
To use or to sell technological knowledge 153 supply side include an array of resource-consuming activities. The actual sale of a product requires appropriate levels of marketing, advertising, credit assessment and post-sale assistance. Customisation and versioning costs can be assigned to transaction costs on the supply side when monopolistic competition prevails and the entry into a market requires that dedicated investments be made in order to identify and implement a niche of loyal customers. Markets differ widely in terms of the quality of information available on the products and the dispersion of prices, and the characteristics of users and customers. The distribution of information among vendors as well as among customers has varying levels of asymmetry. Markets differ widely also in terms of their conditions of competition. Market power can be found with varying levels of intensity, either on the supply or on the demand side. These differences matter when a firm is considering whether to sell a product or use it as an intermediary input for the production of another downstream good. When transaction costs, on the demand side, are high, coordination costs are lower, and the market price for the product is higher than the internal production costs, the firm decides to make a component instead of buying it. Upward integration in the general production process takes place. A new module is added to the portfolio of activities retained within the boundaries of the firm, and the coordination between the production of the upstream module and the production process of the downstream module is provided internally by bureaucratic structures. The firm is not a customer in the market for that intermediary input. The firm instead enters the market for that input, on the demand side, when it chooses to buy. Here third parties provide the supply of the product and the coordination takes place externally in the market. Symmetrically, in the ‘use or sell’ approach the firm needs to assess whether to enter the market for each component or intermediary product or to use each product as a component of a downstream product. The role of transaction costs on the supply side becomes evident. When the cost of using the product markets is high and in any case higher than the cost of using the markets for downstream products, the firm has a clear incentive to integrate downstream and enter the market in the next stage of the production process. More specifically it is clear that the firm confronts the costs of using the market upstream with the costs of using downstream markets after discounting the costs of the internal coordination between the two stages of the production process, and the direct manufacturing costs of the second stage of the production process. In other words it is clear that the firm has an incentive to integrate downstream when the net revenue of the sale in the upstream markets is lower than that stemming from the entry into the downstream markets. The case is interesting when such difference is determined primarily by lower transaction costs. The choice between make or sell and make or buy, moreover, is most frequently partial rather than exclusive. Firms decide whether to sell in the intermediary markets varying shares of the production of upstream production modules. They rarely swing from the sale of the full output to its full inclusion. By the same token, firms rely upon intermediary markets for the provision of varying shares of the
154 Governance of localised technological knowledge intermediary inputs that are necessary for subsequent production stages: some production is retained within the boundaries of the firm. In this broader economics of governance context, transaction costs are defined as the costs of using the markets on both the supply and the demand side. The firm uses the markets also to sell its products, not only to buy the intermediary inputs to manufacture its products. In the governance economics context of analysis, a new area of analysis emerges, one where the governance choice also concerns the markets for outputs rather than solely the markets for inputs. The firm in fact considers not only the possibility of making or buying a specific component or stage of the production process, but also whether to sell its products in the intermediary markets or to the final ones. Needless to say the stages of the intermediary markets for selling are also a matter of choice and assessment. The firm can decide whether to integrate and diversify downward, as well as upward. In this context the firm can also make the choice to sell and eventually to buy again at a later stage of the production process. Here the firm selects the stages of complex and interdependent production processes which can be internalised, and the stages to externalise, but retains control of the overall production process articulated in sequential steps. The market and the organisation become interdependent. The firm can be at the same time the vendor of a product and the buyer at a later stage. The firm can buy back the full amount of the goods produced with its own original inputs or only a part. The boundaries between the firm and the markets become more and more flexible and subject to continual redefinition. The firm is more and more a system integrator, able to combine the subsystems that are included and those that are delivered by third parties (Bonazzi and Antonelli, 2003; Antonelli, 2004a).
3 A new knowledge trade-off: to use or to sell 3.1 The rationale The analysis developed so far has important applications to understanding the conduct of the innovative firm and more broadly the economics of knowledge governance. The stock of proprietary technological knowledge accumulated within each firm and the competence built by means of learning processes and formal research and development activities can be considered an output per se rather than exclusively and necessarily an input for the subsequent production of goods and services in the markets for technological knowledge. In this context the analysis of the factors affecting the choice between selling and making use of the knowledge as input makes sense. Specifically firms implement not only knowledge exploration strategies but also knowledge exploitation strategies. This means that firms need to assess not only whether to produce internally all the knowledge that is necessary for the introduction of new technology or to purchase it in the markets for external knowledge, but also whether to sell the knowledge in the markets for knowledge or to use it to make other products.
To use or to sell technological knowledge 155 The use of the marketplace to exchange technological knowledge is more and more common. The marketplace provides a wide variety of governance mechanisms for the trade of technological knowledge. Markets for technological knowledge are spreading in the economic systems. The use of the marketplace to exchange technological knowledge is more and more common. Technological knowledge can be traded in an array of ways. Firms rely on the marketplace both to sell knowledge and to buy it. Knowledge can be purchased, and sold, in the form of patents and licences, and as a knowledge-intensive service. Technological knowledge can be purchased, and sold, embodied in the financial markets. Knowledge here is traded in the form of the property rights of a company, typically after the IPO of a new high-tech start-up assisted by venture capitalists. Finally, knowledge can be acquired, and sold, in product markets, embodied in new intermediary inputs and capital goods (Geroski, 1995a; Arora et al., 2001). Markets for technological knowledge are spreading in the economic systems. The use of the marketplace to exchange technological knowledge is more and more common. Technological knowledge can be sold with varying levels of embodiment into other goods and services. Technological knowledge can be sold as an intangible good, more or less associated with other services such as the assistance of the vendors to the customers. Technological knowledge can be sold as a service, a knowledge-intensive business service. Technological knowledge can be sold incorporated in weightless products such as software. Technological knowledge can be sold embodied at an early stage of a broader production process, or embodied in products that are manufactured at other stages further down in the general production process within the same filiere or across different filieres leading to the products actually purchased by the final consumer: the household (Arora et al., 2001; Guilhon, 2001). The case of numerical control provides the full range of cases. The technology of numerical control can be sold as a patent or a licence. It can be sold embodied in software, in the numerical control itself or finally in a machine tool with numerical control. The machine tool in turn can be sold as such or it can be used as a capital good in the production of cars and trucks. The engineering industries and specifically the packaging and textile machinery industry provide similar evidence. Each of these industries differs widely in terms of transaction costs on both the supply and the demand side. The chemical industry is characterised by similar trends, with the identification of companies specialised in the supply of the design for chemical plants, as well as by companies that coordinate internally the competence in the design and the delivery of the plants. Finally, important companies in the chemical industries operate the full ‘filiere’ of activities, from the design of the plants, to their construction, to use for delivery of chemical products to the markets. The analysis of the ‘make, use or sell’ trade-off applied to knowledge as a product makes clear that the knowledge exploitation strategies of the firm will be influenced, for given levels of relative revenues in either market, by the relative levels of transaction costs on the supply side in the upstream markets for knowledge as a product per se, compared with the costs of coordinating internally the application
156 Governance of localised technological knowledge of the knowledge to the production of a new good, the costs of the sheer production and the costs of using the downstream markets to sell the products. Economies of scope on the demand side are relevant in this context. Users may prefer to buy products that incorporate knowledge at different stages, rather than purchasing single goods and combining them internally. Economies of scope on the demand side may reflect sheer preferences for downstream products or relevant transaction costs on the demand side. The assessment of risk may favour the preference for downstream applications. In such circumstances producers have a strong incentive to bundle their products and integrate vertically so as to be able to supply downstream applications. Here the incentive is not provided by reduction in costs but rather by an increase in the levels of demand. The profitability of downstream applications is clearly larger. Firms elaborate knowledge exploitation strategies to assess whether to exploit the knowledge available internally as a good in itself and sell it disembodied in the markets for technological knowledge, or to use it as an input in the production of other goods. Here knowledge transaction costs limit the use of the marketplace as an efficient governance mechanism. Knowledge transaction costs are relevant both in the acquisition of codified external knowledge and in the exploitation of new knowledge generated by each firm. Knowledge transaction costs are higher, the lower the appropriability, indivisibility and non-excludability of knowledge are. The exploitation of the knowledge generated can take a variety of forms: firms can use it to produce a new product or sell it as a product per se (Pisano, 1990). Dynamic knowledge transaction costs are relevant on both the demand and the supply side. Exclusion, by means of the use of transactions in the markets for knowledge, yields specific costs in terms of the missing opportunities to benefit from the cumulative learning processes associated with the production process itself. On the supply side, dynamic knowledge transaction costs include the resources to evaluate the scope for incremental advance; dynamic knowledge transaction costs arise mainly because of the high risks of opportunistic behaviour of the customers with respect to derivative knowledge. When derivative knowledge matters, the vendor of the knowledge bears the risks of non-appropriation of the results of the scope of implementation of the knowledge which has been sold. Uncontrolled appropriation of the stream of rents associated with use of the stock of proprietary knowledge, by means of small incremental research costs, can take place with evident damage for the vendor. The working of the markets for knowledge is greatly favoured by the extent to which patents and copyrights can be enforced in the marketplace and licensing is an effective tool to trade specific items of knowledge and competence. The enforcement of the markets for patents is a primary condition for the reduction of knowledge transaction costs and hence the creation of markets for knowledge. The role of the judiciary system in this context is extremely important (Scotchmer, 1996). The governance of technological knowledge is deeply affected by the comparative assessment of the costs of making, buying and selling each component of the knowledge that is required. With low coordination costs and high transaction costs in upstream markets the firm has a clear incentive to make internally all the
To use or to sell technological knowledge 157 knowledge that is necessary. Conversely, with low coordination costs and high transaction costs in downstream markets, the firm has a strong incentive to use the knowledge and apply it to manufacturing products and eventually sell them. Coordination costs apply both to the specific activities that are required to generate new knowledge and to the production processes that are necessary in order to use the knowledge generated. When transaction costs are low and coordination costs high, however, the firm has a strong incentive to act as a knowledge-intensive business service provider. It will acquire the bits of knowledge in the markets for knowledge, add its specific competence, and sell it as a disembodied piece of knowledge, a product per se1 (Holmstrom, 1989). In the context of analysis of the governance of knowledge a new area of analysis emerges, one where the governance choice concerns the markets for outputs rather than solely the markets for inputs. The failure of markets as the appropriate governance mechanisms for the organisation of the generation and dissemination of knowledge does not necessarily lead to undersupply but rather pushes the knowledge-creating firm to use it as an intermediary input for the sequential production of economic goods. Downstream vertical integration is the remedy to the problems raised by the non-appropriability and low tradability of knowledge as an economic good. Poor appropriability of proprietary technological knowledge can be considered a specific cause of knowledge transaction costs on the supply side. When knowledge appropriability is reduced to nil, firms will integrate downstream. Conversely, when knowledge appropriability is high, firms will specialise in the production of knowledge and will rely on the marketplace as an appropriate mechanism for its economic exploitation. With imperfect knowledge appropriability, firms will select the markets where proprietary knowledge can be sold. In other contexts firms will exploit their proprietary knowledge by means of vertical integration. This result is important as it contrasts the traditional argument about the failure of markets, as a coordination system, in the allocation of resources to the production of knowledge because of the lack of incentives stemming from low appropriability and the related ‘knowledge as a public good’ tradition of analysis (Antonelli, 2004a). The generation of appropriate quantities of knowledge can be stimulated by the opportunities in the markets for the products that are manufactured and delivered by means of the technological knowledge they embody. The analysis developed so far has important applications to understanding the conduct of the innovative firm when the stock of technological knowledge accumulated within each firm and the competence built by means of learning processes and formal research and development activities are considered an output per se, rather than an input for the subsequent production of goods and services in the markets for technological knowledge. Now the choice between making and buying is integrated by the choice between selling and making. Specifically firms assess both whether to produce internally all the knowledge that is necessary for the introduction of new technology or to purchase it in the markets for external knowledge, and whether to sell the knowledge in the markets for knowledge or to use it to make other products.
158 Governance of localised technological knowledge 3.2 A simple model At any point in time, the firm needs to assess whether to sell the stock of proprietary knowledge or to use it as an intermediary input for sequential stages. The levels of profitability attained with inclusion are compared with the profitability stemming from exclusion. The inclusion of a sequential step into the complex production process that takes place within the boundaries of the firm, and hence the choice to make instead of selling, depends upon three classes of factors: the quality of the markets from an informational and competitive viewpoint, the relative efficiency of the internal stage with respect to that performed by third parties, and the implications for the process of accumulation of technological knowledge. The sale of a knowledge product is a resource-consuming activity. In intermediary markets the number of potential customers is small and market power on the demand side is often found. The markets for intermediary knowledge products are often global with a scattered geographical distribution. Customers need to be identified and convinced about the quality of the knowledge product sold: dedicated marketing activities need to be carried on. Full disclosure cannot take place and yet prospective customers need to be convinced about the actual technological advance made possible in their specific context of application. The risks of opportunistic behaviour of customers and prospective customers are high, and the vendors need to secure, as much as possible, the actual possession of their proprietary knowledge. Knowledge transaction costs on the supply side consist mainly in longterm technical assistance. Technical assistance makes it possible to implement appropriate strategies to hold the risks of derivative knowledge: both parties have access to the stream of incremental knowledge made possible by the enrichment of the proprietary knowledge that has been the object of the transaction. This is the case especially when high levels of cumulability characterise knowledge. Price and revenue elasticity differ widely across markets as well as the sensitivity of users to advertising expenditures. The distribution of customers in geographical space and in the spaces of product characteristics is not necessarily homogeneous even within the same ‘filiere’. Customisation plays a relevant role, and the packaging of the details of the products is not a trivial activity. Major resources are necessary to apply the knowledge to the specific characteristics of the production process of prospective users. The efforts that are necessary to write the contracts are major for the complexity of timing, quality, and quantity contingencies (Hart, 1995). Intangible sunken assets play a major role in this context. The firm may enjoy advantages of a reputation in downstream markets which is not easy to extend upstream, and vice versa. By the same token, distribution channels and competence accumulated in dealing with some classes of buyers cannot be easily transferred to other upstream or downstream markets. The dedicated competence of the firm in dealing with the markets plays a major role, and its spectrum of application is limited. The competence of an organisation in dealing with the markets in selling a product cannot be easily transferred and applied to selling other products. When upstream markets are less transparent and opportunistic behaviour is more rooted in the business practice than in downstream markets, transaction costs are
To use or to sell technological knowledge 159 higher and hence firms will be induced to use the products of the upstream module rather than to sell them. When upstream markets are less competitive than downstream ones and hence price–cost margins are higher, firms have a clear incentive to sell rather than to make. The firm has a strong incentive to compare both the relative amounts of resources that are necessary to produce and to sell the product in a market with respect to another sequential one and the relative revenue of the sale downstream and upstream. Upon this basis a simple model can be set out. The revenue function is defined as the revenue obtained by the firm for the sale of its products. Two revenue functions (R) can be identified for, respectively, the proprietary knowledge K and the product Z which embodies the proprietary knowledge.2 The revenue functions for the proprietary knowledge K (RK) and Z respectively (RZ) are equal to the standard product of prices and quantities (PK QK) and (PZ QZ): (1) RK = PK QK (2) RZ = PZ QZ Equations (1) and (2) provide the basic ingredients to build a map of isorevenues. Their slope is measured by the ratio of the unit revenue of the proprietary knowledge K sold as a product with respect to the unit revenue of the good Z (PZ/PK). The map of isorevenues in turn provides the constraint to the ‘to make or to sell’ decision making. Economies of scope on the demand side may affect the revenue function, as the demand for the bundle of products is larger than the sum of the demand for the single products. The firm will decide whether to sell directly the proprietary knowledge K or to use it as an intermediary input for the sequential production of the good Z, also according to its comprehensive production costs. The firm is represented as the set of activities that are necessary to produce and deliver the proprietary knowledge K and the product Z. It includes the strict manufacturing process as well as the organisational activities necessary to coordinate the internal exchanges between the research and the production functions and to use the downstream markets. In the case of the specialised production of proprietary knowledge, no coordination activity is necessary: here all the production is contained within the single module specialised in research activities. The activities that are necessary to use the market to sell the proprietary knowledge K must be considered. The organisational inputs that are necessary in using the markets on the supply and the demand side and to coordinate the exchanges between modules within the firm are the product of well-identified activities with inputs, outputs and specific levels of efficiency. This confirms that the firm is a set of activities which goes beyond the production functions of the modules. Formally we have then the activities that are necessary in using the markets for knowledge on the supply side (TRSK), the activities that are necessary in using the markets for the products Z on the supply side (TRSZ), and the activities that are necessary to coordinate
160 Governance of localised technological knowledge internally the applications of the proprietary knowledge K to the production of the good Z. Hence: (3) TRSK = a(I) H(QK) where the supply transaction dedicated activities necessary for the sale of the proprietary knowledge K are the result of appropriate inputs (I) and specific efficiency levels (a) with a fixed coefficient H. (4) TRSZ = b(I) W(QZ) where the transaction dedicated inputs necessary for the sale of the quantities of the product Z (QZ) are the result of appropriate inputs (I) and specific efficiency levels (b) with a fixed coefficient W. (5) CO = c(I) T(QK) where the dedicated inputs necessary for the coordination of the production of proprietary knowledge K and the product Z are the result of appropriate inputs (I) and specific efficiency levels (b) with a fixed coefficient T. The production and sale of the proprietary knowledge K requires the combination of research and development activities and transaction cost activities on the supply side to operate in the markets for knowledge. Formally we have: (6) K = (R&D + TRSK) where R&D measures the unit of inputs specialised in research and development activities, and TRSK measures the units of inputs that are necessary to use the market to sell the proprietary knowledge K as a good per se. In the case of the product Z the resources that are necessary to perform the coordination between the modules K and Z are taken into account together with the inputs into the production function of the module Z and the resources that are necessary in using the markets on the supply side. Formally: (7) Z = (R&D + PRO + TRSZ + CO) where PRO measures the units of inputs that are necessary to manufacture the good Z, TRSZ measures the units of inputs that are necessary to use downstream markets for the product Z, and CO measures the activities that are necessary to coordinate internally the exchanges between the module R&D and the module Z. The total cost equation is determined by the unit costs of all the inputs: (8) TC = r(R&D) + e(PRO) + f(I) where, as usual, r stands for the unit costs of R&D activities (R&D), e stands for the unit manufacturing costs of the product Z, and f measures the unit costs of the organisational resources (I).
To use or to sell technological knowledge 161 The combination of the two sets of activities yields the transformation curve: (9) K = f(Z) According to standard optimisation procedures, the equilibrium conditions are easily found where the slope of the isorevenue equals the slope of the transformation curve (see Figure 9.1): (10) f´(Z) = PZ / PK The shape of the transformation curve reflects the relative convenience of the supply conditions in the two alternative markets. The slope of the isorevenue reflects the relative prices of the two products in their respective markets. The equilibrium point found in the tangency of the relevant isorevenue and transformation curve identifies the best mix of make–use and sell for the profit-maximising firm. The decision whether to sell or to make use of the proprietary knowledge is now framed in an analytical context where many variables matter: the relative prices of the goods delivered to the marketplace, the relative efficiency of production in the modules, the relative efficiency of the two transaction activities on the output markets, the efficiency of the internal coordination activities, and the effects of economies of scope on the demand side. Let us consider them in turn. The effects of transaction costs on the supply side are now fully accounted for. It is clear that, when transaction costs on the supply side, in the upstream markets, are too high, firms prefer to make and use rather than to sell. Conversely, efficient and transparent upstream markets favour specialisation. When the markets for intermediary products do not exist, transaction costs are very high on both the supply and the demand side. The prices of the proprietary knowledge PK can incorporate a relevant portion of transaction costs on the supply side. Prices are too low, well
Markets for knowledge as KK
E
Markets for goods Figure 9.1 The frontier of possible markets
162 Governance of localised technological knowledge below marginal costs, when for instance appropriability is low and uncontrolled leakage takes place beyond all possible efforts of vendors to retain some control of the knowledge. The case for loss of profits (lucrum cessans) due to imperfect inappropriability can be registered on this side of the equation as well, when it occurs even after that the holders of proprietary knowledge have taken all possible measures, and relative transaction costs on the supply side have been registered on the cost side. The degree of relative competitiveness of upstream and downstream markets respectively matters. If in upstream markets barriers to entry are high and large mark-ups prevail, while the downstream product market is closer to perfect competition, the firm operating the module R&D has a strong incentive to sell rather than make. Conversely if higher price–cost margins are found in downstream markets the firm will not sell but rather make. When the upstream activity is shaped by technologies which cannot be easily imitated, the sale of products both upstream and downstream can affect the competitive conditions of the markets. Here the firm will choose whether to be a monopolist upstream or downstream according to the differences in the revenue and price elasticity of the demand. Here economies of scope on the demand side may have powerful effects on the position of the demand curve. The rates of imitative entry downstream can play a role in non-myopic decision making. The equilibrium conditions can easily identify the convenience for the firm in selling or using as against manufacturing and selling downstream products. This result is consistent with much empirical evidence and confirms the heuristic strength of the analytical framework elaborated. The efficiency of the internal coordination of the production of the modules K and Z has a direct bearing on the make–use or sell trade-off. Firms may be forced to sell their proprietary knowledge simply because internal coordination is too expensive. This in turn may depend on the size of the firm. The slope of the curve of coordination costs may become steeper and steeper with the general size of the firm. Large firms may be obliged by coordination costs to be very selective with respect to the make–use option and sell their proprietary knowledge or simply let it ‘spill in the air’. Smaller firms may prefer, ceteris paribus, to make or use their proprietary knowledge and integrate downstream. Production costs exert similar effects; the conditions of the markets from the viewpoint of their informational efficiency being otherwise equal, firms are induced to make–use or sell by the slope of production costs and specifically by the levels of their production costs with respect to those of downstream competitors. Entry in downstream markets may be foreclosed by sharp differences in production costs that favour downstream incumbents. The sale of proprietary knowledge remains the single possibility for exploiting it.
4 Implications for knowledge exploitation strategies The implications of the analysis on knowledge transaction costs on the supply side and vertical integration, as a remedy to imperfect knowledge appropriability, are most important for understanding knowledge exploitation strategies.
To use or to sell technological knowledge 163 When an individual generates new technological knowledge, downstream vertical integration takes the form of entrepreneurship. Individual inventors face the clear alternative of selling their proprietary knowledge as a product per se or using it as an intermediary input. To do so, however, the inventor needs to create a new firm. The entry of new firms, hence, can be seen as the direct consequence of a flow of new technological knowledge modules which cannot be sold as products per se. The creation of a corporation can be the indirect form of the trade of the technological knowledge. An incumbent corporation can eventually acquire the new company. The inventor in this case sells the property rights to the company rather than the intellectual property rights. Here there is a direct relationship between patents and shares, and the markets for knowledge and the financial markets. When the inventor is an incumbent corporation, already existing and active at least in a given product market, knowledge exploitation strategies lead to the growth of the firm. The application of the technological knowledge to the current activities of the firm is expected to have positive effects in terms of performance and ultimately profitability. In turn the growth can be internal or external. Takeovers, mergers and acquisitions can be seen as the direct consequence of the internal use of new technological knowledge by the firm. The acquisition of new firms makes it possible to extend the scope of application of the new knowledge and hence the range of its exploitation. Such growth can take place within the same product market or in adjacent ones. When technological knowledge applies to products that differ from the current ones, diversification, vertical integration and multinational growth can be seen as remedies to the imperfect appropriability of proprietary technological knowledge. Vertical integration, for an incumbent, can be both downward and upward: the new proprietary knowledge can be used both as an input for the internal production of a new product and for the direct internal production of an intermediary input previously purchased in the market for components. The coherence in the growth strategies can be found with respect to the characteristics of the new technological knowledge rather than with respect to the portfolio of current activities and the characteristics of the stock of tangible capital. Strategies of internal exploitation, by means of downstream and upstream vertical integration and horizontal diversification, and strategies of external exploitation by means of the sale of the technological knowledge as a product in the markets for knowledge can coexist, especially when some barriers to mobility across markets can be found. The sale of the technological knowledge can coexist with the direct exploitation when geographic distance matters, when the new knowledge applies to different products and when barriers to international trade are found. Coexistence can be diachronic in that firms sell their technological knowledge but retain the right to implement it and to use the derivative knowledge. In this case the firm sells the proprietary knowledge that already exists but does not sell the rights to take advantage of the stream of new knowledge. Mixed strategies of direct and indirect exploitation take place within the boundaries provided by property rights. Low knowledge transaction costs on the supply side can be found within global corporations where the internal markets are made reliable by proprietary ties among affiliates that are at least partially owned by a central holding. In this case the central
164 Governance of localised technological knowledge laboratories can sell the knowledge to divisions and affiliates when appropriability is lower and use external markets for knowledge with higher levels of natural appropriability. Coordinated transactions in quasi-markets for technological knowledge are also frequent within technological districts, where trust is enforced by high risks of retaliation and localisation exposes firms to reciprocity. In these circumstances firms may specialise in the production of knowledge and in its trade as a product per se, with strong benefits in terms of specialisation and access to technological knowledge. The incorporation of the new knowledge into a new company that operates in downstream markets can provide the incumbent with the opportunity to reduce internal coordination costs and yet to retain the control of the proprietary knowledge. Moreover the equity can be partly sold in the stock exchange, and eventually become a spin-off, or it can be used as a tool for establishing alliances and joint ventures. Once more the incorporation of the proprietary knowledge into a new company provides opportunities for mixed exploitation strategies. In sum, the application of this analysis to the economics of knowledge is fruitful. The marketplace provides the opportunity for firms to sell their technological knowledge in the form of patents, licences and services, or embodied in products. The sale of technological knowledge can substitute its use as input into both the downstream and the upstream production of new goods or new processes. The sale of disembodied knowledge, however, can complement the sale of embodied knowledge. Substitution takes place when the profits stemming from its disembodied sale are larger than those provided by its embodied sale. This can take place when the costs of internal coordination are larger than the costs of coordination in the quasi-markets for knowledge, or when competition is stronger in downstream markets than in upstream ones. Complementarity between the sale in the markets for knowledge and its internal exploitation takes place when the customers of knowledge operate in different markets from the customers of the products (Baumol, 2002). Ex ante standardisation in this context emerges as a powerful knowledge exploitation mechanism. Firms that command proprietary knowledge can impose the standards of the manufacturing process and the design of the modules that are likely to contribute to the final product. Standards matter in this context as the codes of technological platforms that define the interfaces between modules and the role of each specific player. Standards are defined before the implementation of the manufacturing process. Ex ante standards precede and complement patents as appropriation mechanisms. Standards make it possible to select the downstream applications that are retained within the boundaries of corporations and the markets into which proprietary knowledge can be sold as a product per se.
5 Conclusion Transaction costs economics has paved the way to understanding the firm as a bundle of activities which coexist when the costs of internal coordination are lower
To use or to sell technological knowledge 165 than the costs of using the markets. The analysis, so far, has mainly focused on the costs of using the markets on the demand side. Upward integration has been regarded as the consequence of high costs of transaction in the markets for complementary products and intermediary inputs at large. The spreading of modularisation has brought to attention the key role of the costs of using the markets on the supply side. The relative levels of transaction costs in the usage of the markets on the supply side become a relevant factor in assessing the choice of the firm between the sale of products of upstream modules or their integration into the operation of downward modules as intermediary inputs. When the firm decides to use the products of a module as an intermediary product for the following module, the exchange takes place in the internal market, coordinated by means of bureaucratic procedures. The firm is no longer a vendor. The firm instead is a vendor of the product of a module when the relative transaction costs on the supply side are lower upstream. The application of this analytical framework is especially fertile in the economics of knowledge. The economics of knowledge has long been shaped by the seminal contributions of Kenneth Arrow and Richard Nelson about the public good character of technological knowledge. In this approach technological knowledge is regarded as a public good for the high levels of non-appropriability and hence non-tradability. Following our approach, however, even the public good nature of technological knowledge does not necessarily lead to undersupply but rather pushes the knowledge-creating firm to use it as an intermediary input for the sequential production of other economic goods. Vertical integration into downstream activities is an important alternative that the possessor of technological knowledge can assess, in order to exploit its economic rents. The incentives to the generation of appropriate quantities of knowledge can be found in the markets for products that are manufactured and delivered by means of the technological knowledge they embody. This analysis contrasts with the traditional argument according to which the market supply of technological knowledge is deemed to undersupply because of its public good nature. When knowledge cannot be sold as a good, there are still opportunities for its exploitation in the markets for products that can make use of it as an intermediary input. The strategies for knowledge exploitation include downstream vertical integration into the production of goods that incorporate the new knowledge and yet deliver it to the marketplace. It is clear that all factors increasing the absolute and relative tradability of technological knowledge have positive effects for two classes of reasons. First, better knowledge tradability leads to more effective incentive alignment and hence a better allocation of resources to generate new knowledge. Second, better knowledge tradability favours better division of labour and specialisation and hence higher efficiency. When and if the exploitation of technological knowledge as an intermediary input and its embodiment in downstream products is not the cause of limitations to its dissemination, the plurality of markets, rather than the single marketplace, can provide viable mechanisms for an efficient generation of technological knowledge. Even with low levels of natural appropriability, appropriate levels of incentives and division of labour are provided
166 Governance of localised technological knowledge by the opportunity to exploit the proprietary knowledge embodied in downstream products. From a welfare viewpoint the knowledge trade-off is at work again, although in a new context. The levels of fungibility of the new knowledge and the timing and the costs of the reverse engineering play a key role. Reverse engineering is the activities that are necessary to extract the basic knowledge from the downstream products in which it has been embodied. Reverse engineering costs and times are the main source of barriers to imitation and hence the main factor of appropriability for inventors. At the same time, however, with high costs of reverse engineering and long delays the social costs of the embodiment of the new knowledge in terms of impediment to other purposes increase. The wider the scope of application is, i.e. the higher the levels of fungibility, the greater the social costs are.
10 The governance of localised knowledge in the business sector
1 Introduction The application of the basic tools of the economics of information, and specifically the analysis of transaction costs, agency theory, asymmetric information, and communication and networking costs to the economics of knowledge makes it possible to implement the notion of localised technological change and operationalise the debate between the Arrovian notion of knowledge as a public good deemed to undersupply (Arrow, 1962a) and knowledge as a quasi-proprietary good that can be exchanged in markets for knowledge (Jones, 1998; Arora et al., 2001) or generate equilibrium levels of free spillovers (Romer, 1990, 1994). The application of information economics to the economics of knowledge makes it possible to identify a much broader variety of forms of knowledge governance and to provide an integrated framework able to understand the logic underneath their introduction (Arrow, 1969; Stiglitz, 2000). This approach in fact makes it possible to articulate the analysis of the specific conditions for which knowledge can be generated, exchanged and traded among individuals within organisations and among firms and organisations, across the full range of intertwined product and factor markets, by means of a variety of governance modes.
2 Hybrid forms of knowledge governance A flow of hybrid forms of knowledge governance based upon coordinated transactions, quasi-hierarchies and constructed interactions emerges as the result of the continual efforts of agents and organisations to introduce institutional and organisational innovations able to overcome the many limitations of knowledge as an economic activity. Knowledge transactions in quasi-markets are implemented and enforced by a myriad of coordination mechanisms that try to reduce the inefficiencies of pure market transactions. Knowledge interactions, as distinct from transactions, play a key role in this context and yet they are not spontaneous but the product of intentional action directed to make knowledge communication possible. Knowledge coordination within hierarchical organisations is implemented by means of quasimarket mechanisms, often based upon non-exclusivity in employment contracts.
168 Governance of localised technological knowledge When knowledge is mainly tacit, transaction costs in the marketplace are very high. It is difficult to assess its economic value: its tradability is hampered by major information and knowledge asymmetries between producers and users. The latter are not able to use it without the close assistance of the former. Without direct control the producers risk losing command of the economic rents stemming from its use. Intellectual property rights cannot be used to enforce appropriability without consistent preliminary codification efforts. Opportunistic behaviour is easy when no proprietary control can be exerted both within and among organisations. The hierarchical coordination of knowledge producers with downstream activities that embody technological knowledge is an effective way to extract economic value. Thus hierarchy is a remedy to the limitations of knowledge as an economic good per se and a tool to achieve, indirectly, its tradability and hence division of labour among knowledge producers. Agency costs and coordination costs however limit its internal exploitation and generation. Agency costs are relevant within organisations for the risk of opportunistic behaviour of employees, who can try to exploit directly the advantages of the knowledge embedded in their own brains: tacit knowledge with high levels of modular divisibility can be easily exploited on a personal basis. In this case the likelihood of principals appropriating the rents stemming from the research they have funded are low. Quasi-hierarchical forms of internal coordination can be successfully applied in this context. This is less true when knowledge indivisibility is high and knowledge is embedded in larger organisations and its implementation requires groups of experts with diverse competences. Coordination costs rise fast with the variety of activities retained within the borders of the firm. This is especially relevant when knowledge is either composite or fungible. In the first case, transaction costs are found on the supply side, while in the latter they are found on the demand side. When technological knowledge exhibits high levels of cumulability, dynamic transaction costs matter because of inter-temporal complementarity among different vintages of knowledge: vendors risk missing the opportunity for major future improvements, and customers may be unable to master the flow of sequential improvements because of a lack of embedded competence. When knowledge is composite and fungible, coordination costs exhibit fast rates of increase with the variety of activities and competences. Coordination costs increase when knowledge is composite, for it is necessary to manage the many different fields from which knowledge modules are drawn and synthesised. Coordination costs increase when knowledge is fungible; there is a strong incentive to diversify into the different fields of application of the same module of fungible knowledge. Networking costs are lower when knowledge is articulated and a basic rationale can be elaborated so as to manage the interactions among parties that rely on each other for access to external knowledge. Hybrid forms of governance emerge as appropriate mechanisms. Transactions are implemented and associated with complementary organisational devices. Trust and reciprocity help networking activities when the value of the knowledge modules is not yet fully assessed and articulated. Physical proximity among agents helps in reducing the scope for opportunistic behaviour for the higher levels of monitoring and repeated inter-
Governance of localised knowledge in business 169 actions. Procedural contracts among the parties can be articulated so as to make explicit the sequence of actions and commitments even with low levels of definition of the actual content. Contracts are incomplete with respect to the content, but specified with respect to the obligations of the parties through the process of knowledge generation and exploitation. Low levels of cognitive distance among the parties involved in the interaction help in the creation of epistemic communities based on shared knowledge. Reputation plays a major role to ease both transactions and interactions. The analysis of nested networking and transaction costs becomes relevant here. Finally, when knowledge is more codified transaction costs are still relevant and yet lower. Pure markets would fail. By means of coordinated transactions, however, the agents involved are better able to elaborate contractual relations which specify the content obligations of the parties involved: contracts are ‘less’ incomplete with respect to the terms of exchange. Intellectual property rights can be assigned and content contracts can be articulated. Hostages can be exchanged so as to increase the reliability of the parties involved. Lower transaction costs often match high coordination and agency costs, with clear incentives for firms to rely on the markets for knowledge both to purchase and to sell it. The selection of the appropriate kind of markets becomes a major issue. Nested transaction costs play a major role in this context. Technological knowledge is intrinsically heterogeneous: it is a basket of different activities and different processes characterised by significant variety. According to the types of technological knowledge and the related levels of knowledge transaction, communication, networking and coordination costs, firms select the modes of governance that make possible: 1) the effective generation of knowledge by combining with cheaper conditions the external and internal sources of knowledge;1 and 2) the best forms of knowledge exploitation. Rarely are the mechanisms identified exclusive: firms rely on articulated mixes of governance mechanisms according to the characteristics of their portfolios of knowledge activities. In this context coordinated transactions, quasi-hierarchies and constructed interactions emerge as the key mechanisms of knowledge governance that feed a variety of hybrid forms of governance. Knowledge transactions in quasi-markets are implemented and enforced by a myriad of coordination mechanisms that try to reduce the inefficiencies of pure market transactions. Knowledge interactions, as distinct from transactions, play a key role in this context and yet they are not spontaneous, but the product of intentional action directed to make knowledge communication possible. Knowledge coordination within hierarchical organisations is implemented by means of quasi-market mechanisms, often based upon nonexclusivity in employment contracts. Table 10.1 summarises the main results of the application of information economics to the economics of knowledge. Knowledge transaction, agency, coordination and networking costs are set to vary according to the characteristics and forms of knowledge. The following section of the chapter, section 3, applies the analytical framework elaborated so far to the ever-increasing variety of knowledge governance
Knowledge asymmetries Dynamic transaction costs Agency and organisation costs Transaction costs on the supply side
Knowledge asymmetries Exploration costs Transaction costs on the demand side Diseconomies of scope in coordination
Knowledge asymmetries Exploration costs High agency costs Transaction costs on the supply side Dynamic transaction costs Diseconomics of scope in coordination
Knowledge asymmetries High transaction costs High agency costs Nested transactions
Cumulative
Composite
Fungible
Partial divisibility
Tacit/sticky
Low networking costs Nested interactions Complementary competencies
Low networking costs Reputation Nested interactions and externalities in transactions Procedural contracts
Low networking costs Nested interactions and economies of scope in transactions Procedural contracts
Low networking costs Proximity Trust Reciprocity Procedural contracts
Articulable
Table 10.1 Characteristics of knowledge and the conditions of governance
Cognitive distance Unpredictability Agency costs Low transaction and networking costs Nested transactions
Unpredictability Agency costs Low transaction and networking costs Content contracts Hostages Intermediaries Nested transactions
Unpredictability Agency costs Low transaction and networking costs Intermediaries Hostages Content contracts
Unpredictability Agency costs Low transaction and networking costs Hostages
Codified/public
Governance of localised knowledge in business 171 mechanisms. Tables 10.2 and 10.3 summarise the analysis, filling Table 10.1 with the matching between the types of knowledge and the wide array of forms of knowledge governance identified by the literature of the economics of knowledge, with respect to governance mechanisms on the generation and exploitation side respectively (March, 1991).
3 Types of knowledge and governance mechanisms 3.1 The quasi-hierarchical command of tacit knowledge Tacit knowledge is the result of learning processes and cannot be separated from the individuals who have accumulated the relevant experience. In this case, internal coordination within the boundaries of the firm is more appropriate. Knowledge asymmetries are most relevant: with low levels of knowledge appropriability and hence high risks of opportunism and dissipation of the rents associated with knowledge, knowledge transaction and agency costs are very high and firms cannot rely on the marketplace to valorise their intangible outputs. Hybrid forms of quasihierarchical coordination however can be successfully applied. University The academic system is an effective institution for the governance of the generation and dissemination of new knowledge characterised by high levels of tacitness. Scientific knowledge, even when it takes the form of a highly codified expression, has high levels of tacitness and requires high levels of competence to be generated, transmitted and communicated. As Dasgupta and David (1987, 1994) have shown, open science works when an academic institution provides the necessary monetary and hierarchical rewards to scientists, according to their qualification and their reputation. The reputation of scientists is built upon publications scrutinised by peer review. In open science the production and dissemination of new knowledge signal the levels of competence and the skills of the scientist and hence disseminate new knowledge. Because of its effects in terms of reputation and hence ultimately inclusion in the academic system, however, the pursuit of publication is, at the same time, an incentive. This mechanism works properly as long as the costs borne by the system to fund the academic system are compensated by the externalities generated by the academic system. The application of the principal–agent viewpoint provides a complementary interpretation for the understanding of the working of the academic system. From this perspective the non-exclusivity that characterises the employment contract within universities, and the freedom to enter the markets for professional services traditionally recognised for academics, combined with the joint production of education and knowledge, plays a crucial role. The university can now be regarded as a unique form of quasi-hierarchical organisation where academics are free to choose their activities and to publish the result of their research. Publications signal their competence and are not only a means to build reputation. The publication is part of a dynamic process where the scientist has a direct incentive to publish as a way to attract resources in external professional
172 Governance of localised technological knowledge markets (Spence, 1973; Antonelli, 2007d). From this viewpoint the need for public funds is much less relevant. In the extreme case, the academic system comes closer to a special form of professional order: membership in the academic system provides the basic qualifying conditions to operate in the markets for high-quality knowledge-intensive professional services. Research and development Knowledge generation is conducted internally by means of research activities conducted mainly within research laboratories located near to production plants in order to enhance the interaction with learning processes. Academic consultants are often used in this context to access external knowledge: academics are hired as private consultants and are directly integrated into the internal production of knowledge. The allocation of resources to fund new research activities and the identification of prospective users of new knowledge generated are managed internally by means of internal exchanges among affiliates and operative units. The management of the internal markets for technological knowledge and its matching with competent and dedicated competencies in the allocation of financial assets are more and more a key element in understanding the working of multinational corporations and large holdings when tacit knowledge matters. Intra muros research and development laboratories however are more and more exposed to new forms of competitive pressure as their products are exposed to the comparative assessment of external sources such as academic laboratories and other knowledge-intensive business services providers. In this way, new forms of quasi-hierarchical command of the internal production of knowledge are being implemented. Scientific entrepreneurship The creation of new firms by new entrepreneurs with an academic background is often the direct result of the exploitation of tacit knowledge, which, as such, cannot be valorised by other governance modes. Scientific entrepreneurs are inventors, occasionally of academic origins, who cannot rely on the markets for disembodied knowledge and prefer to exploit the rents associated with their knowledge by means of the production and sale of the products that embody, as either a product or a process innovation, the new knowledge (Etzkowitz, 2002). Entry into new markets is often the consequence of serendipity in knowledge creation. Incumbents enter into new product markets in order to exploit new technological knowledge that has not been generated intentionally. Here the choice between selling and using applies. The creation of new firms and diversification and downstream vertical integration of incumbents can now be seen as a governance mechanism specifically implemented in order to increase the appropriability of new knowledge. The firm will choose to make, and hence to include within the boundaries of its portfolio of activities, the modules which use the knowledge as an intermediary input when the tradability and appropriability conditions are low.
Governance of localised knowledge in business 173 Corporate growth The embodiment of technological knowledge in new products and their eventual sale in the marketplace become necessary in order to exploit effectively new technological knowledge. Internalisation of knowledge exploitation and creation is necessary when knowledge appropriability is low. Such internalisation takes place at different levels: within the filiere of activities that use the same module of knowledge the firm selects the stages where integration is necessary and may rely on either networking or even the marketplace for others. Nested analysis of the bundle of activities comes into place. Broad, seemingly unrelated, diversification is often the result of exploitation of fungible knowledge. With given knowledge transaction costs, firms able to introduce technological knowledge with high levels of fungibility are likely to be larger and diversified into the variety of product markets where the same knowledge module can be successfully applied. Strong increasing returns take place in the usage of the same stock of technological knowledge and can counterbalance the increase in average coordination and manufacturing costs. Knowledge fungibility has a direct bearing on the choice of internalisation. When the generation of new knowledge in operating downstream modules is directly influenced by the competence and the knowledge acquired in operating the module upstream, the firm has an incentive to make rather than to sell. Conversely, from the viewpoint of knowledge generation strategies, when knowledge is composite and knowledge transaction costs are high the firm has an incentive to integrate vertically in upstream activities. Vertical flows of knowledge, from the peripheral units to the centre, contribute to the continual growth of corporations. The distinction between knowledge fungibility and knowledge compositeness helps in understanding the strategies of external growth of incumbents with respect both to strategies of knowledge exploration and to knowledge exploitation. When knowledge is composite and has low levels of tradability, external growth by means of takeovers, mergers and acquisitions is a powerful tool to internalise essential knowledge components that are embedded in the firms that are acquired. Conversely, when proprietary knowledge has both high levels of fungibility and low levels of tradability, external growth becomes an effective strategy of knowledge exploitation. The differences of the two strategies, in terms of their effects on the performance of firms, are sharp. When the external growth is the result of a strategy of knowledge exploitation, its direct effects concern primarily economic performance. The external growth makes possible the application of superior knowledge to pre-existing activities and has an effect in terms of an increase in profitability and market share. When the external growth is the result of a strategy of internalisation of external knowledge instead, its direct effect consists primarily in the increase in the output of knowledge and generally in the command of technological knowledge. Eventual effects in terms of economic performance also take place, albeit with some lags. Multinational growth often takes place when the scope of both profitable application and the sourcing of new knowledge is global and high levels of knowledge transaction costs afflict the international markets for
174 Governance of localised technological knowledge knowledge. Global and yet internal markets for knowledge substitute and often complement knowledge transactions on international markets (Fai and Von Tunzelmann, 2001a). Technological platforms Technological platforms are emerging forms of quasi-hierarchical command of tacit knowledge. Technological platforms in fact can be considered as an intermediate form of vertical integration where, however, the elements of the platform are independent companies that cooperate within a hierarchical architecture but are not fully coordinated ex ante by the centre. The units enjoy some degree of autonomy, and their inclusion in the platform is often assessed ex post. The creation of shared technological platforms is especially appropriate for the generation of technological knowledge that exhibits high levels of compositeness and cumulability when coordination costs are too high for a single company to control the full process. Large firms able to command the basic technology and to provide ex ante goals provide the rest of the system with the selective opportunity to contribute technological platforms where other specialised firms can integrate their own distinctive competencies. An array of industries based on complex systems, such as energy production, railways and aeroplane engineering systems, rely more and more on the centralised coordination of a variety of specialised, independent suppliers organised in a single frame. The platform is managed and designed by the firm, which retains the command of the basic knowledge and is able to play the role of hub company. In turn a hierarchy of such systems is often articulated. The hierarchy is based upon the levels of responsibility in the definition of the objectives and goals of the collective undertaking and in taking the risks for the final results. The automotive industry relies more and more on the notion and the methodology of the technological platform as an effective tool to stimulate the division of innovative labour and to coordinate the complementarity and consistency of the innovative activities of a myriad of firms. The hub company defines the basic goals of the project and invites other firms to contribute to the objectives, in a context of delegated flexibility. Each firm specialises in a narrow niche that contributes to the broad array of competences and skills required to introduce a new car. The introduction of a new car requires the command of high levels of systemic compositeness which spans from product to process innovations, and each is the result of hundreds of interdependent components. Technological platforms make it possible for hub companies to organise the creative contribution of a variety of firms by means of structured access to the general design of the new product. In this context it is clear that the implementation of an array of standards, including interface standards, makes easier the management of the flows of goods within the network of firms. The hub company plays a key role as the designer both of the new technology and of the architecture of standardised interfaces. The financial industry provides clear evidence about the key role of technological platforms as a way to generate and exploit technological knowledge (Consoli, 2005a, 2005b).
Governance of localised knowledge in business 175 When the dynamics of technological knowledge and the frequency of market change increase, technological platforms are exposed to the limitations of rigidity. In such circumstances, instead of ex post outsourcing of rigid components that have been already designed by the hub company, technological platforms often evolve into interactive co-design processes where specialised suppliers enter into a coengineering process where each new component is designed jointly by the specialised producer and the hub company into a flexible and yet organised system of distributed competence. Technological platforms are an effective tool to exploit fungible technological knowledge as well. The general quality of the services provided by the platforms is enriched by the collective endeavour, with positive externalities for all the parties involved. The mechanism of network externalities is fully exploited by means of the selective entry of competent players into a single integrated framework centrally organised and managed by the platform builder. The hub company, able to command fungible technological knowledge, can exploit its technological advance and retain the control of incremental value generated by the enrichment of the variety of services and products made available by a common platform. The evidence of the creation of such technological platforms in advanced mobile telecommunications has gradually diffused into other sectors. In-house outsourcing In-house outsourcing emerges as a new way to govern complementary modules of knowledge that must be eventually recombined into a composite final product, and it is built on the combination of under-the-same-roof outsourcing, interdependent outsourcing and continual re-contracting. According to the different levels of compositeness of knowledge modules and knowledge activities, different and ad hoc organisational and contractual solutions can be implemented in order to lower transaction costs and exploit effectively core competencies and operations, that is competencies and operations that can yield major returns. In other words, in-house outsourcing allows efficiency in the recombination of different modules of knowledge into a composite final bundle (i.e. a new product) and at the same time supports knowledge fungibility in that the single modules that are the outcomes of interdependent and yet autonomous activities can enter the innovation production function of prospective buyers external to the firm and even competitors (Bonazzi and Antonelli, 2003). Joint ventures Joint ventures among firms that are competent in complementary bits of knowledge appear as appropriate governance mechanisms for the generation of new knowledge. Joint ventures are especially useful when technological knowledge is composite and each founding partner commands a different type of knowledge. By the same token, joint ventures are a reliable form of exploitation of new knowledge, especially when the latter is fungible. The joint venture can be the result of the
176 Governance of localised technological knowledge combination of complementary assets owned by company A with fungible technological knowledge possessed by company B. In this case company B, unable to exploit internally the new knowledge because of diseconomies of scope and coordination, can rely upon a form of indirect command. So far joint ventures can be considered a form of quasi-hierarchy effective both in the generation and in the exploitation of new knowledge. The joint venture in fact is owned by both companies and yet enjoys substantial levels of autonomy. 3.2 Coordinated transactions for codified knowledge Codified knowledge consists of a body of consistent and explicit information that can be transmitted and applied, although substantial levels of competence and experience are necessary in order to understand and make use of it. Codified knowledge is often found in fields where technological opportunities are slowing down and the levels of knowledge cumulability are lower. When technological knowledge can be better appropriated by the innovator, either because of its high levels of natural appropriability or because the regime of intellectual property rights is highly effective and easily enforced, firms may prefer to sell the technological knowledge directly as a good per se in the markets for knowledge, or to buy it to generate new knowledge. Even when knowledge is fully codified and systematic efforts of articulation have been made, however, the intentional assistance of the original holder is necessary for its use. The markets for technological knowledge become an effective mechanism both to exploit and to acquire knowledge only if implemented with dedicated forms of coordination. Arm’s length markets perform poorly and are substituted by quasi-markets. When such markets for knowledge are available, the selection of knowledge activities that firms retain within their boundaries is much wider. The exploration for external sources of knowledge and knowledge outsourcing becomes common practice. Firms can rely on external providers for specific bits of complementary knowledge. Knowledge outsourcing on the demand side matches the supply of specialised knowledge-intensive business service firms. Universities and other public research centres can complement their top-down research activities, finalised to the production of scientific knowledge, with the provision of elements of technological knowledge to business firms. The exploitation of the knowledge generated can take a variety of forms: firms can use it to produce a new product or sell it as a product per se. Academic outsourcing Outsourcing of research activities to qualified academic laboratories is becoming common practice. Firms perform research activities with a high scientific content less and less within their own laboratories and rely upon the competence of universities. This is especially relevant when technological knowledge is codified and composite: in this case firms should command a wide array of scientific fields
Governance of localised knowledge in business 177 with little chance of achieving high levels of specialisation and competence in each. The systematic access to the wide range of competence provided by universities makes it possible to increase the chances for effective recombination and eventual generation of new knowledge at much lower costs. Universities can be selected according to their reputation and competence, and a variety of contingent contracts can be activated with highly specialised laboratories. When technological knowledge exhibits lower levels of codification, the relations between universities and firms are typically based upon long-term broad contracts within framework programmes that cover many different contracts and include funded chairs and bilateral transfer of personnel, as well as the systematic hiring of students who have finished a doctoral programme. The more structured the fabric of contractual relations is, the lower are the risks of leakage and premature disclosure by scientists seeking visibility and enhanced reputation. Firms try to exert strong control over the results of the research activities by means of intellectual property rights and specific contracts based upon timing and priority in dissemination. The academic ethos based upon open science is put at risk (Geuna, 1999).2 Trade in patents and licences A strong intellectual property rights regime clearly favours the reduction of knowledge transaction costs. The role of the judiciary system with respect to the enforcement conditions of the contracts for disembodied technological knowledge is also most relevant (Anand and Tarun, 2000; Kingston, 2001). With lower levels of knowledge transaction costs and high internal coordination costs, firms are induced to consider the marketplace both to exploit their knowledge and to explore for external sources of knowledge. Trade in licences and patents however can take place only within the context of tight relations between vendors and customers. The former want to control the use of their proprietary knowledge. The latter need the technical and managerial assistance of the innovators. Once more transactions do not take place alone, but are implemented by strong contractual agreements. Recent statistical work by OECD and other national statistical institutes has made available an interesting and reliable body of data about international transactions in disembodied technological knowledge. The technological balance of payments is built upon the records about international technological transactions in terms of technology payments and technology receipts among a large number of advanced countries. The evidence provided by the statistics of the technological balance of payments suggests that international markets for disembodied technology are growing very fast. Throughout the 1990s, international transactions in technological knowledge grew faster than domestic expenditures in R&D activities. Data show that the technology payments represent a significant share of total expenses in R&D activities in most countries. Technology payments have an order of magnitude very close to that of the research activities funded by the business sector in the main OECD countries.3
178 Governance of localised technological knowledge Knowledge-intensive business services The trade in disembodied knowledge in the marketplace is favoured by specialised business service firms which act as intermediaries. Specialised intermediaries act as go-between firms respectively searching for complementary bits of knowledge and/or possible fields of application of the technology already generated in order to test its fungibility. Specialised business services can help the parties to establish the direct relationship when they act as assistant to the exchanges and help the transactions to be performed. In this case knowledge intermediaries specialise in reducing the amount of search costs and provide basic assistance in assessing the reputation and reliability of the parties. They can also act as full intermediaries: they buy the licences and sell them to third parties. Knowledge-intensive business services emerge as effective intermediaries to trade knowledge, especially when fungibility is high. Specialised intermediaries play a major role as knowledge converters: they accumulate generic knowledge and specialise in the delivery of a variety of specific and contextual applications (Antonelli, 1999a; Spulber, 1999). When knowledge is composite, knowledge-intensive business services able to concentrate different knowledge modules and activate different knowledge communication flows from a variety of sources become system integrators and as such are able to command the relevant recombination processes so as to play a key role as central hubs of the knowledge generation process. In this way knowledgeintensive business services become the centre of the innovation process. Tight contractual relations qualify the transactions between KIBs and their customers: both parties want to keep clear control of the knowledge being exchanged and of their conditions of usage and access. Market cascades The notion of market cascades plays an important role in this context. The notion of nested transactions helps in clarifying this point. Firms can combine different strategies for knowledge exploitation and knowledge outsourcing because they act in different and yet related markets. Firms can choose the layers and the stages of the chain value leading to final products into which either to sell or to buy knowledge. Firms can sell their knowledge as a licence in upstream markets and yet manufacture the products that embody such knowledge and sell them in the downstream markets as well. Conversely firms can select the downstream markets for services, associated with a given knowledge module, and abandon the upstream product markets. Such decisions about the layer in which to enter the marketplace depend upon the levels of knowledge compositeness, cumulability and fungibility. With high levels of fungibility the firm can easily combine selected sales in upstream markets with downstream operation. With high levels of cumulability, firms have a strong incentive to exploit directly the layers in which not only higher mark-ups, but higher rates of incremental learning as well, are possible. The telecommunications industry provides much evidence on such dynamics. A broad digital knowledge base makes it possible for firms to select a variety of layers, from transmission to switching and distribution, in a wide range of product
Governance of localised knowledge in business 179 markets, including broadband or cellular technology, television and fixed telephony (Antonelli, 2001; Krafft, 2003). The case of numerical control provides the full range of cases. The technology of numerical control can be sold as a patent or a licence. It can be sold embodied in software, in the numerical control itself or finally in a machine tool with numerical control. The machine tool in turn can be sold as such or it can be used as a capital good in the production of cars and trucks. The engineering industries and specifically the packaging and textile machinery industry provide similar evidence. Each of these industries differs widely in terms of transaction costs on the supply side. On the demand side, it is clear that the purchase of external knowledge with high levels of complementarity with the internal knowledge base is especially attractive. This is the case also when knowledge is composite. In turn, the notion of market cascades applies: firms can select whether to buy patents, licences, knowledgeintensive business services and an array of knowledge-intensive products that belong to the same filiere and the selected stage of the production process either as capital goods or as intermediary goods incorporating high levels of technological knowledge. Transaction costs for the knowledge input here need to be assessed in each of the markets and valued at the overall level. Market transactions of knowledge are often characterised by the systematic use of long-term contracts and structured interactions among the parties involved. This is especially the case when knowledge cumulability is high. In such cases, knowledge transactions include post-assistance and the help of knowledge producers to knowledge users not only to increase the chance of more effective technology transfer but also to increase the chances of the vendor retaining some control over the flows of incremental knowledge that is likely to be generated by the users. The markets for the property rights of knowledge-intensive firms provide an additional layer of the market cascade that makes possible arm’s length knowledge transactions. Financial markets provide an alternative and a remedy to knowledge transaction costs. When knowledge cannot be sold as a disembodied commodity, or internal coordination and monitoring costs limit the opportunities for its embodiment into new products, the property rights on that knowledge can be traded. This is a viable solution when technological knowledge is sticky: financial transactions are better suited than knowledge transactions. Financial markets, in this case, provide effective governance mechanisms for both knowledge exploitation and knowledge exploration strategies. Financial markets The new understanding about the asymmetry between debt and equity in the provision of funds for research activities elaborated by Stiglitz (1985) paves the way to a revolution in financial markets. Equity finance has an important advantage over debt in the provision of funds to innovative undertakings because it can participate in the bottom tail of the highly skewed distribution of positive returns stemming from the generation of new knowledge and the introduction of new
180 Governance of localised technological knowledge technologies (Hall, 2002). This has important consequences in terms of reduction of both the risks of credit rationing and the costs of financial resources for research activities. Lenders need to charge high interest rates in order to compensate for the risks of failure and to sort out a large portion of the new research activities to avoid as many ‘lemons’ as possible. Equity investors instead find an equilibrium rate of return at much lower levels because they can participate in the huge profits of a small fraction of the new ventures. The fraction of lemons that equity can support is much larger than that of debt; hence, as a consequence, financial equity can provide a much larger amount of funding for research activities. Venture capitalism emerges as a useful and distinctive tool to support the birth and growth of new knowledge-intensive firms because it is able to combine the selective allocation of financial funds with the provision of competence and rare business skills. The goal of the creation of the new company, here, is not, as in scientific entrepreneurship, the foundation of a new firm and its eventual growth, but rather its listing on the stock market. Venture capitalism is an effective governance mode, especially when knowledge is sticky. Knowledge stickiness is found when it is difficult to separate the knowledge not only from the human capital but also from the routines and procedures of the organisation where learning activities have been taking place and the knowledge has been generated and articulated. In this case an issue of indivisibility emerges. According to the localised knowledge approach, venture capitalism is a distinctive and effective governance tool precisely because it makes it possible to combine several elements: 1) the assessment of the possible interfaces between scientific and technological knowledge; 2) the articulation of technological knowledge; 3) the selection of new ventures; and most importantly 4) the assistance to newcomers in terms of managerial competence and knowledge about the appropriate organisation, marketing and production based upon the new technological knowledge. A bundle of transactions and interactions are integrated into venture capitalism as a governance mechanism: the provision of managerial and technical assistance is as important as the provision of financial funds. Funding indeed plays a role, but it is comparatively quite marginal. The basic difference between venture capitalism and banks resides precisely in the role of the knowledge communication flows among the parties involved. Banks able to provide effective managerial assistance to new firms are performing the typical function of venture capitalists. From this viewpoint the combined emergence of venture capitalism and dedicated financial markets specialised in transactions of knowledge-intensive property rights, such as the NASDAQ, is providing a new, effective form of knowledge governance. The role of bundling aimed at minimising transaction costs plays a major role in grasping the working of knowledge governance mechanisms. Here the notion of economies of scope in transaction costs fits the evidence: the creation of a bundle of products makes it possible to save on transaction costs. The bundling of products and services into new high-tech start-ups moreover makes it possible to combine two distinct demand schedules: the demand for financial products expressed by asset managers and financial institutions dealing with the investment of financial resources and the demand for knowledge expressed by firms. The combination of
Governance of localised knowledge in business 181 these two demand schedules has a strong positive effect in terms of the provision of funds on the production of knowledge, its distribution within the system and its selection. The interaction between the competences of different categories of financial operators contributes to the increase of information transparency within the system. Financial markets, and more generally the markets for knowledge-intensive property rights, provide an opportunity for a market for knowledge to emerge. Venture capitalism can be regarded as a major institutional innovation. The incorporation of the knowledge-intensive organisation into a new company, either as a start-up or as a spin-off, and its sale in a dedicated financial market become a viable solution in trading knowledge, with clear, positive effects on both the demand and the supply side. The sale of knowledge embodied in knowledge-intensive property rights becomes a viable solution for new firms specialising in the production of knowledge, as well as for firms that cannot exploit the new knowledge directly because of steep organisation costs curves. Conversely, mergers and acquisitions are more and more a viable solution, especially when the object is a high-tech IPO, for integrating new reliable modules of knowledge into a broader corporate structure.4 Mergers and acquisitions of new small high-tech firms become an effective tool to increase both the acquisition and hence the effective dissemination of technological knowledge into the economic system. In information and communication technologies the Cisco model has emerged as a reference: Cisco has pioneered the acquisition of high-tech start-ups as the primary if not the exclusive tool for acquiring new technological knowledge (Chesbrough, 2003). The evidence provided by biotech small firms is especially convincing here. New small firms assisted by venture capitalism and often created by academic spin-offs have introduced all the major innovations in biotechnology. Eventually, however, the large established corporations, traditional incumbents in the pharmaceutical industry, have acquired most of them. The incumbents here had all the advantages of global marketing and production capabilities, high levels of visibility and reputation in the markets for final products. The incumbents however were far less successful than smaller and younger high-tech firms. Here technological knowledge is embodied in the corporate structure and could be appropriated by means of the organisation of routine-based firms (Gompers and Lerner, 1999). Financial markets perform an important role in the governance of knowledge not only as an effective tool for the provision of financial resources to new technological undertakings. Financial markets make it possible to implement and valorise the working of knowledge complementarities. Financial markets make it possible to manage a flow of mergers, initial public offerings and acquisitions by means of which firms are able to change their boundaries. The effects of knowledge fungibility and compositeness can be better managed by means of the continual selection of the units and activities that it seems appropriate to coordinate internally, within the boundaries of the bureaucratic organisation, and the units and activities with which knowledge transactions can take place in the markets (Avnimelech and Teubal, 2004; Antonelli and Teubal, 2007).
Learning Intra muros R&D Open science Academic consultants
Learning Upstream integration Intra muros R&D Open science Academic consultants Technological platforms Joint ventures
Learning Intra muros R&D Downstream integration Open science Academic consultants KIBs Joint ventures
Scientific entrepreneurship
Composite
Fungible
Modular divisibility
Quasi-hierarchies for tacit and sticky knowledge
Mechanisms
Cumulative
Modes
M&A
Technological platforms Procedural contracts Academic outsourcing Inter-industrial joint ventures KIBs Centred networks Reputation
Procedural contracts KIBs as hubs University–corporation liaison centres Multi-industrial and metropolitan technological districts
KIBs Procedural contracts Constructed trust within vertical clubs Technological districts Academic outsourcing
Constructed interactions for articulable knowledge
Table 10.2 Governance mechanisms for knowledge generation and types of knowledge
Markets for licences and markets for goods
Corporate R&D centres Content contracts Academic outsourcing Cross-licences KIBs Open source Acquisition of high-tech small firms Quasi-markets for patents and licences
Academic outsourcing Content contracts Technological clubs Standardisation committees Open source Acquisition of high-tech small firms KIBs Markets for patents Markets for licences
Patent thicketing Content contracts Vertical clubs Academic outsourcing Epistemic communities Open source
Coordinated transactions for codified knowledge
Governance of localised knowledge in business 183 3.3 Constructed interactions for articulable knowledge Articulable knowledge consists of a mix of tacit and codified knowledge and it can be considered a step in a process of codification. As such it exhibits intermediate conditions of appropriability. In such conditions, knowledge spillovers are possible but require substantial efforts if they are to be absorbed by prospective users. Firms may select external coordination strategies based upon networking to implement both the development of a research project with the acquisition of relevant external knowledge and its commercial exploitation. Technological communication can take place however only if the parties are able to manage the strong information asymmetries. For this reason the exchanges of articulable knowledge take place by means of constructed interactions within technological clubs and coalitions. Here knowledge networking activities are required and include high levels of monitoring and assessment of the conduct of the partners in the club. When technological knowledge is articulable, the contractual interaction among partners within technological clubs can be better implemented because of the reduction in information asymmetries among parties and the higher levels of observability of the efforts and related contributions of each member. Trust combined with cognitive and geographical proximity can help the division of labour and complement corporate hierarchies. The exchange of articulable scientific and technological knowledge is also practised within research communities based upon repeated interactions and close reciprocity in communication. The incentives for the creation of informal and yet intentional interaction procedures, often implemented by colocalisation within technological districts, are very strong in this case. Long-term cooperation contracts Knowledge interactions and networking based upon long-term bilateral contracts make it possible to valorise knowledge indivisibility and hence to access, generate and exploit technological knowledge that cannot be either coordinated internally or traded in the marketplace. External coordination among a small number of firms is formed around long-term cooperation contracts that specify the conditions of access and usage of the eventual findings. Knowledge contracted cooperation differs from knowledge transactions. The latter take place when exchanges take place and two parties agree to sell and buy respectively a piece of knowledge. Knowledge networking based upon contracts makes possible structured, cooperative interactions based upon flows of knowledge communication among parties, with strong elements of coordination and duration in time. Knowledge cooperation based upon contracts provides a barrier to the explosion of free-riding and opportunistic behaviour. Knowledge networking however is not a ‘free lunch’ but requires dedicated activities and receptivity-enhancing networking behaviours. Networking consists in the systematic and organised sharing of codes of conduct among independent firms which agree explicitly upon knowledge interactions qualified in terms of trust, reciprocity and repetition and based upon contracts to access the competence and the expertise of the other party in a context qualified by a clear identification of the parties (Menard, 2000; Cassier and Foray, 2002).
184 Governance of localised technological knowledge Technological clubs A specific form of knowledge cooperation can be identified when more than two or three firms cooperate. Technological alliances and research coalitions are formed with the specific task of creating and managing collective research pools. Such relations can be symmetrical when each partner owns a complementary bit of original knowledge and asymmetrical when the value and the relevance of the proprietary knowledge possessed by each party differ. The assessment of the worth of the knowledge controlled by each party of course is the first problematic issue to be solved. Each firm will try to secure the benefits stemming from its own specific bit of knowledge and will try to minimise the risks of leakage or uncontrolled dissemination, even within the club. The implementation of specific control rights is a typical solution. By means of a clear definition of the control rights of each partner in the club and the allocation of dedicated markets, defined in product and geographical terms, and timing of access to the sequential results, partners can solve the problems of information asymmetry and the risks of opportunistic behaviour (Lerner and Merges, 1998; Brousseau and Glachant, 2002). The distinction between procedural and content contracts is relevant here. Procedural contracts are incomplete contracts designed to specify the modality of the interaction, while content contracts focus on the characteristics of the transaction. It is possible to implement and eventually to enforce specific procedural contracts about the process of participation and timing of assignment of property rights, temporary and partial exclusivity, time lags and partial and discriminated domains of privilege to subsets of contributors, selected according to both the amount of inputs and the results. The reputation of the members of the club plays an important role in building technological clubs, as it increases trust: the higher the reputation of the members, the higher is the stability in cooperation (Attalah, 2003). The characteristics of knowledge matter in considering the kinds of technological clubs. When knowledge cumulability matters, vertical clubs are often found aligned along user–producer relationships. Horizontal clubs are more effective when knowledge is composite. Vertical technological clubs often complement the sale of patents and licences and are based upon close inspection of the activities of the customers and users of the patents. The relationship between the vendors and the customers takes place within long-term contracts, which include the assistance and active cooperation of the two parties. The major goal here is not only the reduction of transaction costs stemming from the prospects for future knowledge but also the coordination of the learning opportunities stemming from eventual and shared implementation of the original knowledge. The vendors participate in both the appropriation and the creation of the derivative knowledge stemming from its implementation and incremental accumulation (Johnson, 2002). Generative relationships are very effective within vertical technological clubs when new knowledge is generated within the context of user–producer interactions (Von Hippel, 1988). Vertical technological clubs differ from horizontal ones. In the latter all parties are involved in a shared research activity where each member contributes its own competence and nobody claims the role of knowledge originator (Foray and Steinmueller, 2003).
Governance of localised knowledge in business 185 Sponsored spin-off Firms practise sponsored spin-off more and more as a tool to valorise second-best technological opportunities. When coordination costs and specifically agency costs in the generation of new knowledge are too high and technological knowledge is sticky, the creation of a new enterprise by the team of researchers and experts with the help and assistance of the former employer is a viable solution. The sponsored spin-off is assisted by the parent company in terms of technical assistance, provision of funds and especially long-term purchasing contracts for the output. In turn the sponsored spin-off remains under the formal and informal control of the parent company in terms of incremental knowledge generated and definition of the standards and characteristics of the products. Sponsored spin-off is also a way to reduce agency costs and yet to increase the division of labour and the specialisation. The parent company may even rely for the production of a component or a dedicated input, formerly manufactured internally, on the new company (Patrucco, 2005). Patent thicketing Cross-licensing and patent thicketing are useful governance mechanisms when technological knowledge is composite and appropriability conditions exist. The chances to go ahead depend on the command of a variety of different bits of technological knowledge. The costs of internal coordination of the activities that is necessary for the accumulation and implementation of the full range of kinds of knowledge quickly become prohibitive. The distinctive specialisation and capabilities of each firm cover only a minor portion of the full range of complementarities. In these circumstances firms may find it profitable to create a pool of knowledge resources by means of cross-licensing. The access of each firm to the proprietary knowledge of the others depends upon the amount of proprietary knowledge each firm is able to contribute. Cross-licensing is an effective mechanism of governance especially when the range of applications of the knowledge generated is itself wide and barriers to mobility limit competition among firms. Each firm can benefit from the knowledge generated in quite distinct product markets. Crossbarriers to entry into national markets and transportation costs may favour such agreements (Reitzig, 2004). Standardisation committees When knowledge is both codified and composite, standardisation committees are useful governance mechanisms. Standardisation committees help the valorisation of the complementarities of knowledge modules possessed by different firms. Standards emerge in this case as the result of the intentional participation of firms in a process of collective implementation of a common knowledge base and are especially effective in managing its applications in specific technological results that exhibit high levels of interoperability and compatibility (Antonelli, 1999a).
Vertical integration Centred platforms In-house outsourcing Joint ventures
Upstream diversification Open platforms Joint ventures
Downstream diversification Open platforms Joint ventures In-house outsourcing
Scientific entrepreneurship
Composite
Fungible
Modular divisibility
Quasi-hierarchical command of tacit and sticky knowledge
Mechanisms
Cumulative
Forms
Venture capitalism M&A
Inter-industrial joint ventures Procedural contracts Sponsored spin-off KIBs Growth poles Reputation
KIBs as hubs Procedural contracts Centred networks
KIBs Procedural contracts Sponsored spin-off
Constructed interactions for articulable knowledge
Table 10.3 Governance mechanisms for knowledge exploitation and types of knowledge
Markets for licences and markets for goods
Quasi-markets for patents and licences Market cascades with vertical linkages Cross-licences Open source and liability regime Content contracts
Patent thicketing Quasi-markets for patents and licences Content contracts
Patent thicketing Quasi-markets for licences with technical assistance Long-term content contracts
Coordinated transactions for codified knowledge
Governance of localised knowledge in business 187 Networking within geographical and technological clusters Knowledge interactions based upon geographical and technological proximity differ from contractual networking and are quite distinct and specific with respect to knowledge transactions. Proximity substitutes for contracts for many reasons. First, proximity reduces the scope for opportunistic behaviour because of the exposure to repeated interactions and also reduces the costs of communication. Second, proximity favours the sharing of language and communication protocols. Third, proximity favours the connectivity of labour markets and hence the circulation of tacit knowledge embodied in the skills of personnel. Fourth, proximity favours the informal barter of know-how both in user–producer relationships and among competitors relying on tacit codes of reciprocity and repetition because of the frequency of mutual interactions. High levels of reputation for local trust and an effective tradition of mutuality in knowledge interactions qualify the attraction of regions for firms seeking to benefit from the advantages stemming from knowledge indivisibility (Feldman, 1999). Nested networking and transactions are most important in understanding the working of knowledge dissemination and generation within geographic clusters. Here the behaviour of firms in labour markets and the conduct in many markets for intermediary inputs, including professional services, are strongly intertwined. Transaction and networking costs need to be assessed in the full range of activities where direct interdependence among firms takes place. Reciprocity and assessment take place across many markets that are closely related both by proximity and by knowledge indivisibility. Spillovers do not flow freely in the atmosphere, but can be absorbed only when communication protocols have been established (Cohen and Levinthal, 1990). Constructed interactions are necessary to absorb knowledge spillovers for the high levels of knowledge communication costs. The distinction between knowledge cumulability, fungibility and compositeness makes it possible to identify four quite distinct mechanisms at work within geographical clustering: 1
2
When fungibility applies, coordination costs prevent firms, typically large corporations, from taking advantage of all possible applications of their proprietary knowledge. Firms are induced to select the technologies they want to develop internally and may allow the leakage of marginal technological knowledge. Interstitial opportunities for smaller firms are created. Small firms grow around the driving engines provided by large corporations. The flows of technological communication are vertical, as they are centred upon a central beam that provides the role of a switching system. Here the analysis of François Perroux on the ‘poles de croissance’ on the driving role of large corporations applies (Perroux, 1964). When complementarity among diverse and dispersed bits of knowledge matters and takes the form of knowledge cumulability instead, the spatial agglomeration of small firms that are active within the same filiere and command complementary bits of knowledge may favour the collective generation of new technologies. Here there is no beam, and the flow of technological knowledge
188 Governance of localised technological knowledge
3
4
is typically horizontal and is based upon reciprocal access. Sectoral technological districts specialised in a narrow range of products, typically characterised by high levels of sequential complementarity, emerge as effective nodes that favour knowledge communication and hence the generation of new technological knowledge by firms that rely upon external knowledge (Patrucco, 2003). When technological knowledge is composite, the variety and heterogeneity of the competence of firms become the key issue. The complementarity among a broad array of technological fields is key to favouring the recombination and eventual generation of new knowledge. Multi-industrial technological districts, with a strong urban character and effective access to high-quality academic infrastructures, become an effective mechanism of governance (Jacobs, 1969). Finally, within technological districts, vertical knowledge interactions between users and producers – along the industrial filieres that link industries specialising in final products to upstream industries specialised in advanced intermediary inputs and capital goods – play a key role. Such interactions stimulate and qualify the feedbacks between the demand for new, advanced technologies and their supply. Close relations between vendors and customers make easier the mutual understanding and bilateral flows of knowledge that can be better articulated, with huge positive effects in terms of rates of introduction of innovations (Von Hippel, 1988; Antonelli and Barbiellini Amidei, 2007).
These four models differ sharply, as they are based upon different characteristics of technological knowledge and grasp completely different modes of interaction among firms. In the first case, diseconomies of scope and coordination costs for large corporations are the key factor that leads, often by means of sponsored spinoff, to the creation of technological interstices. Growth poles are clearly a mechanism for the exploitation of technological knowledge. In the second case, proximity favours strategies of social generation of a given module of technological knowledge. Higher levels of total factor productivity increase can be directly expected from the localised exploitation of knowledge fungibility. In the third case, instead, proximity favours the implementation of strategies of explorations in a variety of technological modules. Broad exploration eventually leads to higher rates of generation of new knowledge and as a consequence to faster rates of introduction of innovation and ultimately to higher levels of productivity growth. When knowledge is composite, the variety of the knowledge modules available within a geographical cluster becomes a relevant issue. The greater such variety is, in fact, the higher are the chances that the recombination process is able to yield the generation of new technological knowledge that is composite. Here the distinction between inter-industrial and intra-industrial knowledge externalities is important. When knowledge is composite, the clustering of a wide variety of firms, active in different industries and with a different knowledge base, is far more conducive to the generation of new knowledge, than mono-industrial clusters, specialised in a narrow range of activities. Urban districts typically provide such multi-industrial
Governance of localised knowledge in business 189 opportunities. Finally, the fourth case differs for the key role of vertical interindustrial complementarities between firms and industries that belong to the same filiere: in most cases the filiere itself is the result of a localised process of constructed interactions that make possible increasing levels of division of labour and specialisation. Reputation within epistemic and professional communities Epistemic communities play an important role when the division of knowledge labour is based upon knowledge interactions, as distinct and specific with respect to knowledge transactions and knowledge contracts, qualified by a clear membership into a well-defined professional community. The generation of new knowledge and its free dissemination within epistemic communities are based upon the rewards stemming from the correlated production of professional reputation. Here the notion of nested interactions and transactions is most important and provides a major clue to assessing much-debated evidence. Members of professional communities are ready to contribute to the collective production of new knowledge provided not only that they can access the knowledge generated by others, but also and primarily that they can capitalise on the professional rewards based upon reputation that is associated with social visibility and recognition of the contributions to the common endeavour. Agents are clearly able to assess jointly the costs stemming from possible opportunistic behaviour in one set of interactions and the advantages stemming from transactions in the related markets for professional services. This analysis makes it possible to understand an important aspect of the working of open science in general and of one of its specific applications, open source. Open source A relevant application of the basic model, elaborated in open science and based upon professional reputation, has been spreading in the software industry, where knowledge is characterised by strong elements of cumulability and fungibility. The open source software can be considered the best case of an open technological platform. In the open source experience the source code of the GNU/Linux computer operating system has been made available to the public. Incremental enrichment of the basic code is then provided by the spontaneous contributions of a myriad of actors selected by an informal organisation of gatekeepers. Richard Stallman and Eben Moglen have elaborated the notion of General Public Licence (GPL) as the basic governance tool. According to the GPL each player can access the source code, modify it and distribute the results of his/her work to third parties at no cost, provided that three important conditions are respected: 1) the original assignee is notified by the new user and registered; 2) an explicit reference to prior proprietary – albeit non-exclusive – knowledge is made in the new knowledge; and 3) all the advances introduced are in turn made public and available to third parties.5 The introduction of the GPL has been most effective as a tool to secure
190 Governance of localised technological knowledge the systematic codification of the new software produced and the implementation of its complementarity and accessibility to third parties. Spontaneous entry in the GNU/Linux platform is induced by the free access to the code source and by the increasing size of the library of applications made available by previous users. Each additional project however is not planned in advance and it is rather the result of the idiosyncratic activity of each user. In turn, each user is expected to make the results of its activity accessible to third parties by means of the GPL. In the generation of this kind of software knowledge, learning as a joint product of current activity plays a key role. New enriched codes are the result of exploration activities and specific dedicated activities being carried out as a part of the professional activity of new users. The social recognition of their contributions plays an important role in this community as an indicator of professional expertise and contributes to the creation of professional reputation. In turn for each agent the exploitation of the reputation conveyed by the granting of the GPL takes place easily in the supply of an array of software services ranging from maintenance to specific customised applications and localised implementation. The successful working of the GPL seems to be based upon the strong notion of complementarity: complementarity between the elaboration of new applications and the implementation of the source code and complementarity between reputation and exploitation, at the agent level. At the system level it is clear that the GPL works as an affective signalling mechanism that makes public and easily accessible all the implementations to the source code. In a free software system the identification of all progress would be undermined and the working of the open knowledge platform would be hampered by major and recurrent information asymmetries. From this viewpoint the General Public Licence can be considered as a first step in the departure from the property rule and a first application of the liability rule in the governance of knowledge. According to the General Public Licence agreement the patent is assigned to the inventor, who is informed but cannot prevent the use of his/her proprietary knowledge by third parties. The latter, however, are obliged to register and to recognise, by means of explicit references, the contribution of the prior and proprietary knowledge patented and to provide the results to third parties. These obligations, together with the automatic granting of the authorisation to the user after notification, can be considered a form of application of the liability rule. This is especially clear if the crucial role of the citations is appreciated as a main factor in implementing the social recognition and professional reputation of the first inventor and hence in increasing the rents he/she can extract in the markets for professional services (see Samuelson et al., 1994; Reichman, 2000). The open source software seems to work successfully for the strong role of complementarity between learning and working on the one hand and between social recognition and professional reputation, with the attached effects in terms of fees and wages for the contributors, on the other hand. In such circumstances agents are willing to contribute freely to the social endeavour because of the lateral effects in terms of increased visibility and earnings (Lerner and Tirole, 2002, 2004). In the working of epistemic communities the complementarity between the generic knowledge embedded in the source code and the specific and idiosyncratic
Governance of localised knowledge in business 191 knowledge upon which applications can be introduced plays a major role. The appropriability of the specific knowledge upon which each application to a narrow and highly contextual case is based is high. Those making the applications can command a large share of the rewards stemming from them. They hence have a clear incentive to share the basic knowledge embedded in the source code and to implement collectively the code, which becomes the common pool from which each will eventually draw in order to apply the general principles to specific cases (Dalle and David, 2003; Dalle et al., 2004). Epistemic communities appear to be especially effective when the common undertaking has no specific and explicit ex ante tasks. In the open source case, the path of knowledge generation advances by means of the proliferation of incremental applications that build upon cumulability and compositeness, but without an explicit direction and final destination. Conversely, free contribution to the common undertaking can take place precisely because each agent contributes the specific results of its own activity, after it has been conceived and used according to the specific and idiosyncratic needs of that undertaking. Complementarity here is rather an ex post outcome, rather than a planned, ex ante, condition for the upgrading of the knowledge path. Thus the working of epistemic communities is characterised by substantial unpredictability in terms of both the rates and the direction of the collective activity (Von Hippel and Von Krogh, 2003).
4 Conclusion This chapter has provided an information economics analysis for the economics of knowledge and explored systematically the effects of knowledge characteristics upon the assessment of the design, characteristics and performance of the institutions and processes that shape the governance of the generation and distribution of technological knowledge. Economics of information provides important tools for understanding the details of knowledge governance mechanisms. Neither pure markets nor pure hierarchies can provide the necessary levels of coordination and division of labour. An array of knowledge governance mechanisms, ranging from coordinated transactions and constructed interactions to quasi-hierarchies, however, has progressively emerged according to the characteristics of knowledge and the costs of using markets and organisations. Both the forms and the characteristics of knowledge have a direct bearing on the costs related to knowledge transactions, knowledge interactions and the internal coordination of knowledge generation and hence on knowledge governance mechanisms and knowledge governance modes. The application of the basic tools of information economics to the economics of knowledge provides an interpretative frame able to appreciate the variety of constraints and incentives of the different governance mechanisms which shape the generation of technological knowledge in a market economy. This approach has made it possible to appreciate the constraints raised by organisational factors such as coordination, networking and transaction costs in shaping the process of accumulation and generation of new knowledge and to articulate a single analytical
192 Governance of localised technological knowledge framework that seems able to integrate the broad variety of modes of governance that a wide empirical literature in the economics of knowledge has identified. At the same time the approach elaborated so far has made it possible to highlight the key role of nested transactions and interactions and the need to go beyond the analysis of single transactions when indivisibility matters. Both economies of scope and externalities in transaction matter. This notion is most relevant in the analysis of knowledge governance but seems to have a wider scope of application. In this context, the dynamic coordination among such a myriad of agents, which keep changing their technologies, their knowledge, their competences, i.e. their location in the multidimensional spaces in which they are located, and their knowledge exploration and exploitation strategies, becomes the key issue. In the specific context of the economics of localised knowledge, dynamic coordination is especially relevant to reduce the costs of ‘multiple inventions’, that is the waste of resources invested in the generation of the same knowledge by different agents unaware of the parallel efforts of the others. Only when such a dynamic coordination takes place can external knowledge be timely, consistently available and complementary with internal learning and research strategies. Increasing returns in the generation of knowledge depend, of course, on the solution of the many facets of the knowledge trade-off but depend primarily upon the levels of dynamic coordination a system is able to express. Knowledge is the key emergent property of the complex system dynamics only when dynamic coordination is successfully implemented (Amendola and Gaffard, 1988, 1998). Business knowledge governance is not a sufficient condition for dynamic efficiency to be assured in the knowledge markets. When increasing returns matter, such as in the case of knowledge complementarity, cumulability, fungibility and compositeness, and the price mechanism is unable to convey all the relevant information, the markets are unable to set the right incentives and hence move in the right direction. In the present institutional context, knowledge governance mechanisms in place are not able to provide all the necessary coordination between the variety of agents that participate in the collective process of generation of new knowledge. The basic trade-offs between appropriation and dissemination, concentration and distribution, incentives to produce and incentives to use, variety and complementarity yet remain to be solved. Technological knowledge is such an imperfect good that spontaneous market coordination cannot provide the necessary consistency between private and public optima. Public policy interventions specifically designed to increase dynamic coordination by means of the defence of the working of knowledge commons and the increase of the informational efficiency of knowledge governance are necessary.
11 The new economics of the university A knowledge governance approach
1 Introduction The organisation of the production of knowledge in advanced economic systems is facing a rapid shift away from the corporate model established in the second part of the twentieth century in the US towards a new distributed model (Etzkowitz, 1998; Zeitlin and Herrigel, 1999; Etzkowitz and Leydesdorf, 2000). The old model was based upon the pivotal role of the large corporation and was articulated around the key role of direct public subsidies to firms investing in research and development activities, strong public demand for goods and services incorporating high levels of knowledge-intensive products, and the complementary role of the academic system supported by public funding. In the new model, which is still emerging, the generation of knowledge is the result of enhanced social interactions. Transactions in the new markets for knowledge are complemented by technology sharing among firms within research consortia and technological platforms and the new venture capitalism, with the emergence of new surrogate markets for knowledgeintensive property rights that is the result of the merging of financial markets and the markets for knowledge (Antonelli and Teubal, 2007). Corporations are performing a declining role in the performance of research and development activities, while they remain active in funding the generation of new knowledge and its eventual purchase, often in the form of mergers and acquisitions of new innovative small firms and research contracts assigned to the academic system. Small firms play a much bigger role in the process (Chesbrough, 2003; Chesbrough et al., 2006). In this new model the academic system plays a new pivotal role. Etzkowitz (2002) proposes the successful metaphor of a triple helix where government, universities and firms are the three elements of a dynamic process of interaction and interdependence. The academic system is emerging as a key player in the new model, as it appears to be an institution which is more adept at managing creative talents. Specifically, the university is now regarded as an institution which has elaborated a form of quasi-hierarchical coordination based upon a set of rules and routines, articulated in a unique mix of incentives, and contracts characterised by non-exclusivity in the terms of employment that are especially efficient in the organisation of the generation and dissemination of knowledge as an economic process. Such an assessment is the result of a closer analysis of the role of the academic system as
194 Governance of localised technological knowledge an institutional device that favours the management of creative talents from a principal–agent viewpoint. The new academic system can be regarded as a form of intermediate governance mechanism that has gradually emerged through the centuries with specific characteristics that, if properly identified and implemented, facilitate the coordination, within a quasi-hierarchy, of some levels of division of labour and exchange. So far the academic system seems to have been able to fill the wide gulf between the two extreme cases of the State, as the single provider of knowledge as a public good, and the corporation, as the appropriate institution for the provision of knowledge as a quasi-proprietary good. Here the characteristics of knowledge matter, and applying the basic tools of information economics provides major opportunities to grasp the rationale of knowledge governance mechanisms. In this context, applying the tools of information economics to understanding the economics of knowledge and the workings of economic institutions in the knowledge economy makes it possible to explore new facets of the reasons for the increasing role of the academic system as a viable institution (Stiglitz, 2000, 2002).
2 From open science to the academic management of creativity The university is a long-established institution. Since its origins, in Bologna in 1088, it has been able to survive and change, adding new facets and new aspects. The shifting identity of the academic system is the result of a process of introducing sequential and incremental steps when, at each point in time, the pressure of new forces pushed the institution to adapt to the changing context in an effort to reduce discontinuities. Thus continuity and innovation coexist (David, 2004a, 2004b). The university’s new emerging role as a cornerstone of the new organisation of the production and dissemination of knowledge leads us to try to understand the reasons for such an extraordinary story of success and adaptation from an economic viewpoint. It is, in fact, generally agreed that the academic system is an effective institution for the governance of the generation and dissemination of new knowledge, characterised by high levels of tacitness. Scientific knowledge, even when it is highly codified, has high levels of tacitness, and high levels of competence need to be generated, transmitted and communicated. It is possible to find many different reasons to explain why the academic system is an effective institution. Different interpretative frameworks seem useful both in understanding its vitality and in guiding its evolution. The work of Dasgupta and David (1987, 1994) has long been regarded as the most comprehensive analysis of the economic foundations of the academic system. Dasgupta and David have shown that the academic system provides a viable institutional set-up to combine the incentives to the dissemination and the generation of new knowledge. Universities facilitate the workings of open science, that is to say, the peculiar combination of incentives necessary to generate new knowledge and also to disseminate it into the economic system.
The new economics of the university 195 In their study, Dasgupta and David have provided a clue to the economic analysis of such a peculiar institution, where knowledge producers have a clear set of incentives to generate new knowledge and yet to give up their rights over it, via its rapid dissemination by means of publication. The open science mechanism works when an academic institution provides the necessary monetary and hierarchical rewards to scientists, according to their qualification and their reputation. A scientist’s reputation is built upon publications scrutinised by peers. In open science, the production and dissemination of new knowledge signals a scientist’s level of competence and skills and hence the need to make public new knowledge. However, the pursuit of publication is at the same time an incentive because of its effects in terms of reputation and hence ultimately inclusion in the academic system. Such a mechanism works properly as long as the costs borne by the system to fund the academic system are offset by the externalities generated by the academic system. Here both the amount of knowledge actually produced and the part of it which is effectively communicated to the rest of the system matter. If the levels of knowledge generation are high but the levels of effective knowledge communication are low, the amount of costs borne by the economic system can become higher than the return. Dasgupta and David’s analysis is based on the Arrovian tradition of analysis of the economics of knowledge and specifically it can be regarded as an insightful elaboration of the well-known knowledge trade-off. Along the lines of the Arrovian approach, knowledge is regarded as an economic good that has many limitations and drawbacks, namely non-appropriability, non-divisibility, and non-rivalry in use (Arrow, 1962a, 1962b, 1969). Moreover, knowledge is at the same time an output of a specific activity and yet an input, not only in the production of other goods, but also and mainly in the production of further knowledge. Hence, the knowledge trade-off between the incentives to increase appropriability takes place. There is an appropriate stream of economic benefits for inventors, and the contrasting need to increase access to existing knowledge so as to facilitate its use in the production of new knowledge. Working within this analytical context, Dasgupta and David have made it possible to understand the original and innovative combination of incentives that can be found in the open science system. The academic system enables us to find a solution to the knowledge trade-off. It works, however, if and only if: 1) the provision of public funds makes it possible to secure a reward for the inventor after publication – giving up the rights over it – with tenure and an appropriate salary; and 2) scientific publication is an effective medium for knowledge communication. Figure 11.1 provides a schematic account of the interpretative rationale elaborated by Dasgupta and David (D–D). The system is based upon a triangle where the insertion of the State makes the indirect relationship between the demand and the supply of knowledge possible. It is a relationship which the knowledge trade-off impedes. The business system accepts that it has to pay some taxes that are transferred by the State to the academic system. The latter in turn manages the open science system, providing incentives for the generation and eventual dissemination of knowledge by means of chairs assigned to creative scientists.
196 Governance of localised technological knowledge State
funds
taxes
publishing and teaching university
corporations
Figure 11.1 The D–D flow chart of the academic system
The creativity of scientists is measured by their publications. Tenured scientists are expected to teach and publish at the same time: in so doing they create and disseminate new knowledge. The interpretative approach elaborated by Dasgupta and David highlights the facets of academia that were most consistent and coherent with the organisation of the generation of knowledge in the corporate age. The changing organisation of the production of knowledge has called increasing attention to a number of limitations of this model. In so doing it has pushed towards the exploration of other facets of the academic institution and more careful examination of a number of aspects that Dasgupta and David had not considered, as they were less relevant in the corporate age. The debate has clustered upon two main points: 1) the conditions for technology transfer to take place; and 2) the poor allocation mechanisms of the academic system. Let us consider them in turn. The dissemination of the knowledge generated by academic institutions and hence the effective levels of knowledge externalities depend upon the levels of communication within the economic system. Knowledge communication depends upon the strength of the emission from the academic side and the levels of receptivity of the business system. In the corporate age the receptivity of large firms was very high for a variety of reasons. Corporations operated intra muros large research and development laboratories where high-quality scientists were employed: corporate scientists and academic scientists could communicate easily. Corporations could hire large flows of young Ph.D.s: knowledge communication was enhanced by close relationships between professors and their former students. Competent readers in the business community could easily read scientific publications outside academic circles. In the corporate model, knowledge communication was quite strong because of the central role of industrial research and development. The decline of the share of large corporations in the performance of research and development activities and in the introduction of technological innovations in recent years and the growing role of knowledge transactions and interactions among
The new economics of the university 197 a variety of small firms changed the context in which the Dasgupta–David model applied.1 Within economic systems characterised by distributed processes of knowledge generation based upon constructed transactions and interactions among firms with low levels of internal research and development activities, knowledge communication between the academic and the business systems is much less effective. The receptivity of firms to the advances of generic, scientific knowledge achieved by academics, and made available by means of the traditional communication channels, based upon publications, is lower. The need for an interface able to extract dedicated solutions, hence idiosyncratic knowledge, from such advances is much stronger. Interactions between the academic and the business community characterised by small firms require intentional efforts on both sides. The traditional mission of universities to generate knowledge as a public good becomes questionable: the risks that nobody is able to take advantage of the new advances of scientific knowledge increase. New mechanisms for the effective transmission of new knowledge to the business community including effective knowledge signalling devices are needed. The descent of academics from their ivory towers and into the new knowledge markets becomes necessary (Lawton Smith, 2006). A large part of the twentieth century was characterised by a steady rate of scientific advance in well-defined directions. An acceleration of the rate of scientific advance and several changes in direction took place towards the end of the century. This discontinuity put under stress the limitations of the previous system of resource allocation to and within the academic system. The Dasgupta–David model provides poor guidance to understanding the criteria that are necessary to identify the correct amount of public resources to be provided to the academic system. Secondly, there is no indication of how to identify the criteria for the distribution of public funds within the academic system among the different academic disciplines and scientific fields. Thirdly, it provides poor guidance on how to allocate a given amount of public funds in a given discipline among the different possible academic institutions. In a pure ‘open science’ system a chair should be given to the best scientist, with no prior identification of its discipline or location. Consequently, the amount of resources that should be transferred to the academic system should be based exclusively on the number of scientists who are actually able to raise the absolute level of scientific excellence, independently of their field of activity. This is clearly not the case. A variety of spurious mechanisms are at work. Rules of thumb are used to fix the general amount of resources that the State transfers to the university. Didactic factors play a strong role. The pressure of student numbers has a strong effect, even when scientific reasons would not suggest funding the growth of some schools and some universities instead of others. Here, the typical problems of the principal– agent relationship emerge. Academic institutions may have specific internal incentives to direct public funding towards field A instead of field B that do not necessarily coincide with the optimisation of social welfare. All attempts at technological and scientific forecasting have proved to be not very reliable. Here the risks are that hierarchical control, within the academic system, pushes towards the misallocation of funds, that is to say, away from fertile and productive new fields, in defence of tradition and established academic corporations.
198 Governance of localised technological knowledge The changes in the organisation of the production of knowledge have exposed the ‘open science’ model to increasing criticisms. Both internal and external forces push the academic system towards new configurations. New facets of the academic system are emerging, and some pre-existing features attract new attention and consideration. The application of the tools provided by the economics of information to analyse the economics of knowledge can be useful to understanding the new directions of the university as an economic institution (Antonelli, 2006c). The new academic system, characterised by the professional university, as it has been emerging in recent years, can be considered a viable institution for the governance of the generation and dissemination of new knowledge from another viewpoint: the principal–agent approach. From this perspective the typical nonexclusivity that characterises the terms of employment within universities and the freedom to enter the markets for professional services traditionally accepted for academics plays a crucial role. Professional rewards are now viewed as a sufficient incentive to generate and disseminate new knowledge. Academics publish to signal their competence and attract resources to fund their activities. A strong incentive compatibility between the generation of public science and research performed under contract takes place: academics cannot dismiss their reputation because it is the basic signalling mechanism to attract resources in the markets for research services. Incentive compatibility adds to efficiency compatibility stemming from qualified user–producer interactions between academics who produce knowledge and firms that use it. The quality of the research at the same time is likely to enhance the levels of the didactic activities both in terms of true content and in terms of reputation. Substantial economics of scope between research and teaching emerge. From this viewpoint the need of public funds is much less relevant. In the extreme case, the academic system comes closer to a special form of professional order: membership in the academic system provides the basic qualifying conditions to operate in the markets for high-quality knowledge-intensive professional services. The academic scientist is often also a member of the professional community for which the new knowledge is directly relevant in his daily professional work. In other words, publications are signals that are directly valuable in the adjacent professional community, overlapping with the scientific one. Hence the close overlap between recognition within the scientific community and professional reputation is a strict, necessary condition for the system of incentives to work. According to this approach, the academic system seems to be a much more viable institution than the corporation for the governance of the generation of knowledge characterised by high levels of uncertainty and serendipity, where the principal has little chance of properly assessing ex ante the creativity of the agents, the amount of effort made, or the outcome of the research process in terms of both timing and specific content.
3 The comparative economics of agency costs: the university as an effective institution to solve principal–agent problems in creative work A new interpretative model to understand the rationale of the working of the new type of university based upon the professional academic can be elaborated, drawing
The new economics of the university 199 upon the applications of the economics of information to the economics of knowledge. It can be useful both in understanding the economic mechanisms at work and as a tool to appreciate the implications for stakeholders and the needs for complementary changes in the institutional set-up. The generation of knowledge is characterised by high levels of uncertainty: serendipity and creativity play a crucial role. Even with the heroic assumption that the need for a specific module of knowledge can be identified, and consequently a hierarchy and a sequence of possible necessary modules of knowledge can be agreed upon and an amount of resources can be funded, it is clear that the process is still affected by basic uncertainty. Once the amount of resources has been fixed and an objective has been identified, basic unpredictability about many different aspects of the knowledge generation process emerge: if the new knowledge will be generated, when the new knowledge will become available, where, i.e. in which field, it will be generated and even how, i.e. by means of what modules of preexisting knowledge, it will be generated. It is very difficult to organise and manage employment relations in such a context. Principals have enormous problems in assessing the levels of their agents’ creativity and effort and evaluating their output. 3.1 Agency costs in corporations The costs of hierarchical coordination, articulated in agency and organisation costs, severely limit the size and the span of knowledge-intensive activities carried out within the boundaries of a single unit (Arrow, 1974). Agency costs limit the use of hierarchical control over the activities that are necessary to generate and use technological knowledge within the boundaries of the firm for two kinds of reasons. Knowledge asymmetries also play a major role within organisations. Because serendipity and creativity play a key role in the generation of new knowledge it is difficult for principals to check on the actual content of the operations that lead to the generation of a given amount of standardised knowledge. The management of research activities within corporations is hindered by problems associated with the identification of a correct system of incentives. This has negative consequences because of the characteristics of the distribution of creativity. Wide-ranging empirical evidence confirms that high levels of asymmetry – close to a typical Pareto distribution – characterise the distribution of scientific creativity among qualified and competent scientists. A small number of scientists are responsible for a large share of all publications and an even larger share of all references. Any mistake in identifying the creative minds and, worst of all, mistakes in implementing an appropriate incentive structure able to motivate the creative efforts of the few creative minds will have major negative consequences on the output of the research activities and hence on average costs. As is typical in such conditions, there are two sources of possible errors: 1
Failure to identify the true creative minds. The rare skills of the true creative minds are lost because of the lack of incentives and mistakes in their
200 Governance of localised technological knowledge 2
identification. Appointing agents who are not, in fact, truly creative. Non-creative agents can try to take opportunistic advantage of the basic information asymmetries with respect to principals regarding: a) the expected value of the knowledge produced; and b) the effort and work that were necessary to generate it.
Agency costs in the generation of knowledge within complex organisations are consequently very high also because of limitations in anticipating the outcome of a research in progress, not only in terms of rates, but also, and mainly, in terms of directions. The outcome of a given research project can be relevant but in fields of application different from the expected ones. The traditional organisation of labour in knowledge-intensive activities characterised by high levels of craftsmanship and self-employment, with strong professional content, is clearly explained by the high levels of agency costs incurred in monitoring effort, output and applications in the generation of knowledge (Holmstrom, 1989; Garicano, 2000). Internal organisational costs also limit the number of complementary activities that can be internalised by each firm and hence the amount of knowledge that can be generated, implemented and exploited internally. Unit organisational costs are elastic not only to the size of the activities but also, and mainly, to the variety of activities that need to be internalised. The larger the rate of increase of unit organisational costs is, with respect to the number of activities, then the larger is the number of complementary activities that cannot be kept within the boundaries of the firm. Incumbents miss important opportunities because of hierarchical coordination costs; large corporations are unable to implement all the opportunities they help to create. Coordination costs, in fact, apply both to the specific activities that are required to generate new knowledge and to the production processes that are necessary to use and exploit the knowledge generated (Arrow, 1974). In such conditions, it seems clear that the greater the uncertainty of the research projects is then the greater is the unpredictability of the outcome and the lower is the efficiency of traditional business systems to manage the generation of new knowledge. Firms have a strong incentive to rely more and more not only on the traditional dissemination tools of the knowledge generated by universities, i.e. publications and Ph.D.s, but directly on academic consultants, who can be hired on a professional basis, as intermediate knowledge-intensive inputs to perform a specific research activity. Intra muros research and development is substituted or strongly complemented by the services provided by the academic system. Academic patenting, very much like academic publishing, becomes an important signalling mechanism that helps firms to identify and locate dedicated competence and high levels of scientific creativity. Corporations act more and more as system integrators of large research programmes that are performed by a variety of academic centres. The corporation, of course, keeps control of the division of labour and manages the integration of the different modules of knowledge within an internal knowledge platform. The chief scientist in the corporation organises the general research project and elaborates its structure in complementary modules of knowledge. Part of the research is carried
The new economics of the university 201 out intra muros and part is outsourced to competent academics. The identification and selection of the academic individuals and academic centres able to provide the necessary modules are crucial. 3.2 Agency costs in the academic system Alternative institutions are necessary to manage the production of such a specific and idiosyncratic kind of good. Here the departure from the Arrovian tradition of analysis is clear. The emphasis of this analysis is no longer concentrated on the problems of allocation but rather on the problems of generation. By contrast, the organisation of creative work within the academic system can be appreciated for its unique combination of sophisticated ex post compensation and a two-party incentive system. From such a viewpoint, the academic system seems to be based on the certification of creative talents. The principal rewards the creative workers by issuing a certificate that testifies to their levels of creativity. The qualification, in fact, is based on the reputation acquired in the open science system. The qualification, however, provides more than tenure and the resulting salary. It enables the scientific worker to enter both the markets for knowledge services and the related markets for professional services. In the former the worker can sell his/her specific research capabilities to firms that are ready to hire competent researchers to carry out specific research tasks in the fields in which the scientific worker has accumulated competence and expertise. In the latter, the scientific worker can provide specific services directly where his/her competence is established. In order to obtain such a certificate, the scientific worker has a strong incentive to establish his/her own reputation by means of publications. The compensation scheme however is broader than the one considered by Dasgupta and David. Hence the mechanism identified by Dasgupta and David to solve the knowledge trade-off is at work, but in a broader system of incentives and agency costs. The principal–agent approach however makes it possible to explore these other aspects. In the open science system identified by Dasgupta and David, the principal bears all the agency costs arising from the need to check the efforts and the creativity of the tenured workers over time. The risks of opportunistic behaviour and declining creativity are high.
4 The professional university The professional university mechanism makes it possible to reduce the nonobservability of the efforts of academics and of the value of their output. Principal– agent problems are much lower in assessing teaching activities. Training services can be checked better: the amount of effort in teaching, the competence of teachers and, to some extent, the output of didactic activities can be assessed better than the activities leading to the generation of new knowledge. The accountability of the research output of academics is much more difficult. Scientific reputation earned in the epistemic communities helps in assessing the scientific value of the output of an experienced researcher in an established scientific field.
202 Governance of localised technological knowledge Relevant economies of scope in the generation and dissemination of knowledge and in monitoring costs provide the foundations for the viability of the joint production of training services and new knowledge. The university has a clear incentive to hire qualified scientists who have been able to build up a consistent reputation by means of publications because the quality of teaching and the quality of scientific competence are complementary. Moreover, the university has a strong incentive to hire qualified researchers because academic reputation based upon publications has a considerable effect on tuition fees. The levels of tuition fees are highly elastic to the reputation of the academic staff. On top of this, it may be argued that the quality of students is also sensitive to the reputation of the academic staff. In turn, the intrinsic quality of the students prior to enrolment has a strong correlation with the quality of the students after graduation, and this has a subsequent positive feedback on the ranking and reputation of the university itself. From this viewpoint the professional university can be considered to be a viable institution for the governance of the generation and dissemination of knowledge for two complementary and yet distinct economic reasons. First, the academic system brings together, in a single and unique institutional set, the solution to the knowledge trade-off. Secondly, the academic system, articulated in a combination of didactic activities, certification of the competence and skill of creative workers, and non-exclusivity of labour contracts, provides the institutional setting which engenders the creation of an efficient supply of certified knowledge workers to the rest of the economic system. In so doing, the academic system provides basic signals about the supply of creativity and competence and their distribution across fields. In other words, the university certifies the actual talent of the scientific worker and provides him/her with the opportunity to enter the markets for research and professional services. In the professional university the incentives for the generation and eventual dissemination of new knowledge are no longer provided exclusively by the academic system. The incentives to publication are now generated by two distinct mechanisms. The first is internal to the academic system: the scientific reputation acquired by means of scientific publications and certified by the academic system can engender some additional salaries paid by the academic system.2 The second component however is external and it can be identified in the indirect compensation scheme for certified creative talents arising from non-exclusivity. Scientific reputation now engenders monetary rewards that can be earned in the markets for research and professional services. Such rewards can be capitalised especially when the conversion of generic knowledge into highly specific and idiosyncratic applications is both necessary and not easy.3 The second important condition for such a system to work is the high level of knowledge fungibility. Fungible knowledge can be applied to a variety of specific cases. Idiosyncratic applications cannot be imitated and replicated easily. Finally, reputation plays an important role when the opportunity cost of choosing the wrong expert is high, owing to the wide gaps between ex ante and ex post conditions. Patients praise most the reputation of their doctors when their life is at risk. Heavy investment in irreversible industrial projects suggests using the best
The new economics of the university 203 experts available to minimise the technical and commercial risks of the undertaking and to avoid writing off huge amounts of brand new fixed capital. Spontaneous epistemic communities based on nested interactions and transactions are especially successful in academic communities and in the adjacent professional markets (David, 2004b; Antonelli, 2006c). The unique institutional set-up of the professional university makes it possible both to reduce the negative effects of principal–agent problems, when the productive process is characterised by uncertainty and basic information or knowledge asymmetries, and to create a supply for professional research services. A closer analysis of the workings of the professional university reveals three basic features that make it possible to reduce the agency costs and to bear a limited amount of the costs of scrutiny and assessment of the creative skills and efforts of its workers. The professional university is characterised by: 1) non-exclusivity in the terms of employment and the related dual ladder structure of the compensation scheme; 2) the total lack of hierarchical direction in the definition of research activities; and 3) its intrinsic division into two activities and two markets: the market for education activities and the market for knowledge. Scientific workers, after qualification and formal entry into the academic corporation, are expected to provide basic training services and to respect some generic indicators of scientific activity. The labour contract with the university includes basic teaching duties for the scientific worker, but there is almost no specification about the kind of research activity. Tenured researchers are expected to maintain some levels of research activity, documented by publications. The academic system, however, rarely expresses binding obligations about the content, the objectives and the methodologies of the research carried out by its members. Here the notion of the university as a quasi-hierarchical organisation is fully confirmed: it is difficult in fact to consider the case of a fully fledged principal that does not specify a relevant portion of the activity of its agents. Non-exclusivity in the labour contract, typical in the academic system, means that the certified creative talent can sell his/her competence and creativity in the markets for research and professional services. Here creative minds can find a return for the marginal productivity of their competence. In some extreme cases certified, creative talents, i.e. university professors, can buy back their teaching time, renounce all wages paid by the university and spend all their working time in the markets for research and professional services.4 Eventually, however, if and when creativity slows down, professors can go back to providing teaching services and earn the salaries paid by the university. An academic worker’s output is composite, as it is made up of the research output and the didactic output. The former is defined by the worker and, because of the non-exclusivity clause, can be partly compensated by third parties. The professional university specifies only the content of teaching activities and consequently pays a salary that is not expected to remunerate the full marginal productivity of academic labour, but only a part of it, which includes the component arising from the marginal productivity of teaching activities and the social value of knowledge that is not appreciated by the price mechanism.5
204 Governance of localised technological knowledge
v C TV B
FWS PF A AW
MPA MPR MPD
q Figure 11.2 The three-party compensation of academic pay
As Figure 11.2 shows, the total value of the marginal productivity of academic work (MPA) can be split into three components: 1) the marginal productivity of didactic activities (MPD); 2) the private marginal productivity of research (MPR); and 3) the total marginal productivity of research including the social effects that are not fully reflected by the price mechanism (SMPA). In this figure the distance between MPD and MPA measures MPR. Universities pay a salary (AW) that is defined by the marginal productivity of didactic activities. Professional fees (PF), earned in the markets for knowledge, either directly as private consultants or indirectly via contractual relations between the university and the firm, pay for the private marginal productivity of research activities and, added to academic salaries, complement the salaries of scientific workers (FWS). The distance between MPR and MPA measures SMPA, the social effects of new knowledge that are not accounted for by the price mechanism. The missing portion of the total value of knowledge (TV) stemming from the social benefits not accounted for by the price mechanism should be matched by public subsidies.6 In the open university system there is no price mechanism at work, and the full social value of the knowledge generated is matched by public subsidies. In the professional university the level of subsidies instead can be reduced and partly substituted by the fees earned by academic researchers in the markets for knowledge. 4.1 The incentive complementarity between public science and contract research A debate has been going on about the relations between public science and contract research. Some argue that in the professional university a substitution effect may
The new economics of the university 205 take place: scientists would be less and less active in the production of science and would dedicate most of their attention and efforts to securing and performing research for firms. Moreover scholars in the professional university would be less and less keen to share information with colleagues, who are now seen as dangerous competitors in the markets for research services. The academic ethos would be undermined (Nelson, 2004). Others stress the complementarity between the generation of public science and the performance of research under contract in terms of efficiency. The results of the investigations carried out by Mansfield (1995) suggest that the productivity of scientists was enhanced by their interactions with the business community. A specific case of user–producer interactions has been made: the producers of knowledge would take advantage of closer relations with their users. When the basic tools of the economics of information are applied, we see that an incentive complementarity between the generation of public science and the research under contract can be identified (Spence, 1973). It is clear that scientific reputation based upon publication is an essential ingredient of the working of the interaction mechanism between firms and the professional university. Academics need to publish so as to confirm and possibly increase their reputation. Reputation is an essential tool to signal their competence in the markets for research services and hence attract resources to fund their activities. The incentive complementarity moreover works both ways: publications are necessary to build up reputation and hence attract resources. In turn the resources provided by firms are an input into the generation of future reputation. Publication is part of a dynamic process where the scientist has a direct incentive to publish at time t1, as a way to build a reputation and hence to attract resources at time t2. The rational scientist should be reluctant to stop publishing and switch all his/her energies to performing contract research without any fallout in terms of new publications. It is clear that the scientist’s reputation would quickly decay and no more firms would be interested in hiring him/her to perform new research activities under contract. The reluctant scientist might accept substituting production of public science with the production of proprietary and dedicated knowledge with idiosyncratic applications only if the amount of the contract would match the longterm discounted value of his/her opportunity costs. This amounts to saying that the value of the contract should match the stream of the scientist’s wages for a considerable amount of time. It seems clear that such conditions could be rarely fulfilled, as they would be equivalent to an employment relation. 4.2 Non-exclusive intellectual property rights The argument in favour of the professional university based upon the nonexclusivity of the employment relation, and elaborated so far, holds more and better if non-exclusivity applies to intellectual property rights as well. It is clear that if and when the customer of the academic services has the right to retain exclusive property rights of the results of the research activities, the social marginal product of the professional university is seriously undermined. Because of non-divisibility,
206 Governance of localised technological knowledge articulated in cumulability, complementarity and fungibility, and the sharp difference between production and reproduction costs, knowledge exhibits all the characteristics of an essential facility. Relevant increasing returns in the production of knowledge stem from the ensuing economies of density. Marginal costs are always below average costs. In such circumstances, as is well known, non-exclusive property rights need to apply in order to prevent welfare losses. This condition is even stronger for knowledge, as it is at the same time an output and an input for the generation of further knowledge. Exclusive intellectual property rights would reduce the dynamic efficiency of the system (David, 2004a; Antonelli, 2007b). This is true especially for the product of academic research conducted within the frame of professional contracts between academics and firms. The access of third parties to the result of the research conducted by academics for business firms is necessary for the working of the new system. Exclusive intellectual property rights over the results of academic research would lead to the banning of dissemination of new knowledge by means of publications. Academic personnel impeded from publishing the results of their work by exclusive intellectual property rights would lose the opportunity to gain reputation. In the long run this would undermine their position in the market for professional research activities. In the short term, incumbent academics would take advantage of their existing reputation based upon previous work, conducted within the open science model. In the long run, however, newcomers could never build a scientific reputation. Clearly this process becomes quickly unsustainable: exclusive intellectual property rights over the results of research conducted on a contractual base would undermine the working of the full system, with clear disadvantages on both the demand and the supply side and for all the stakeholders at the system level. The elaboration of mechanisms that make the disclosure of new knowledge possible, even when it has been generated within the context of a contractual relationship between an academic and a firm, and favour publications and other dissemination devices rather than preventing them is a necessary complement of the new model of professional university. The fine-tuning of non-exclusivity becomes necessary: it is clear that, if intellectual property rights cannot command any privilege over the new knowledge, firms would have very low incentives to outsource research activities. On the other hand it is clear that, with the forms of exclusivity that are featured by the current intellectual property rights, such as in the case of the present patent legislation, not only could new cohorts of researchers never establish a reputation and hence enter in the markets for knowledge but the risks of knowledge exclusion and knowledge rationing would be too high. Knowledge externalities would be drastically diminished, even below the levels associated with the traditional forms of dissemination of knowledge as a public good typical of the open science model, and the new markets for knowledge would collapse. New forms of reduced intellectual property rights seem necessary both for the dynamic efficiency of the system at large and for the working of the professional university model. Publication and appropriation of the result of outsourced research must become complementary rather than mutually exclusive (Reichman, 2000; Antonelli, 2007b).
The new economics of the university 207 In a regime of non-exclusive property rights, academic patenting can become a useful signalling mechanism that helps firms to identify and locate the competence of scientific talents. The granting of a patent by qualified patent authorities, very much like the editorial assessment of high-ranking journals, certifies the quality of the research conducted by identifiable scholars in well-defined locations. A new complementarity between academic publications and non-exclusive intellectual property rights should take place. After a limited time window of restriction designed to secure the incentives to firms to outsource research activities, scholars can choose whether to publish or to patent according to: 1) the specific content of their discovery; and 2) the norms and rules practised in their specific community. In both cases the firms that have funded the research should exert a limited control in terms of reduced ownership for a short period of time. Only when non-exclusivity applies both to the employment relationship between academics and universities on the one hand and to the intellectual property rights over the results of the research can the model of the professional university substitute successfully for the open science system. 4.3 Markets for research services The institutional ingenuity of the professional University engenders the creation of a supply for professional research services. On the supply side of this special market there are academics, both as individuals and as academic centres that are certified and evaluated as to their levels of creativity. Moreover, the compensation schemes used in the academic system allow the supply side to operate on a variable cost basis. The fixed costs of the academics are covered by internal payment for their teaching activities within the academic system. The supply in this market is now characterised by high levels of signals about the quality of the supply and the reduced costs, as the total cost of academic supply can be shared between universities as institutions and the customers of the research services. The position of the supply curve is much lower than it would be without the academic system. On the demand side, moreover, transaction costs and specifically search and screening costs are much lower for two reasons. First, suppliers are evaluated and signalled by their academic career. This evaluation is based upon their reputation and ultimately on publications and other scientific scores based on references and the quality of journals. Secondly, suppliers are ready to work on a professional basis and accept compensation that is clearly dependent on the delivery of the knowledge module. In other words, academics are not looking for a permanent employment contract with firms, but operate on a professional basis. This in turn allows ex post payments dependent on the delivery of an output to be made. Thus the academic is paid by the job.7 In the markets for research and professional services a firm’s demand for knowledge inputs can be met by the supply of certified, part-time talented workers, with a significant reduction in costs by using such markets. Qualified and certified scientific workers can earn substantial rewards by supplying their creative talents in the markets for research and professional services. In this market, scientific workers can be paid by the job as professionals.
208 Governance of localised technological knowledge The demand side is made up of firms that are exploring external sources of knowledge, in the markets for research and professional services. Firms are ready to substitute internal research activities with scientific skills and competence that can be acquired in the marketplace. Outsourcing of research activities to qualified academic laboratories has become common practice. Firms are reducing their research activities with a high scientific content which they used to carry out in their own laboratories and they are increasingly turning to the competence of universities. This is especially relevant when technological knowledge is codified and composite: for in this case firms would have to be involved in a wide array of scientific fields, with little chance of achieving high levels of specialisation and competence in every one. The systematic access to the wide range of competence provided by universities makes it possible to increase the chances for effective recombination and eventual generation of new knowledge at a much lower cost. Firms can take advantage of the supply of scientific and creative competence of the academic system either directly, hiring individuals who operate as professionals, or indirectly, when the contractor is the university itself. The latter case is typically used when teamwork is necessary to carry out the research activities. Individuals in the latter case however do retain the right to share with the university the rewards arising from their professional services. In this case the university performs the functions of an associated partnership, usual practice in the legal services and other markets for professional services. The new emerging markets for research services provide important opportunities for small firms to take advantage of structured research activities. As is well known, research and development activities are characterised by economies of scale, at least until a minimum threshold. As long as research activities could only be conducted intra muros, small firms were not able to organise systematically the generation of knowledge and the implementation of internal learning processes. Increasing returns in the performance of research activities exerted major discriminating effects. The supply of research services by the academic system provides small firms with the opportunity to fund specific and dedicated research activities that are performed by large research centres. Universities can be selected according to their reputation and competence, and a variety of contingent contracts can be activated with highly specialised laboratories. When technological knowledge is at lower levels of codification, the relations between universities and firms are typically based upon long-term broad contracts within the framework of programmes that cover many different contracts and include funded chairs and bilateral transfers of personnel, as well as the systematic hiring of students who have finished a doctoral programme. The more structured the fabric of contractual relations is, the lower are the risks of leakage and premature disclosure by scientists seeking visibility and enhanced reputation. Firms try to exert strong control over the results of research activities by means of intellectual property rights and specific contracts based upon timing and priority in dissemination. Scientists however need to reconfirm their reputation and hence have a strong incentive to publish. The reputation-seeking behaviour of the scientists prevents dissemination from being reduced and hence favours the solution of the knowledge trade-off.
The new economics of the university 209 4.4 The dynamics of the demand: a simple graphic exposition The emergence of the professional university can be considered the result of the strong increase in the demand for research activities that has characterised the last few decades. The dynamic implications of the analysis conducted so far can be presented with the help of a simple figure. It is assumed that the principal in the economic system is able to identify the correct quantity of knowledge that is necessary to pursue the correct level of economic growth. A clear minimisation problem can be set: the economic system has a strong incentive to try to minimise the institutional mechanism that makes it possible to reduce the costs of the necessary knowledge. The correct amount of knowledge that is necessary is identified in Figure 11.3 as Q* on the horizontal axis. Three alternative institutional solutions for the provision of that quantity of knowledge are now considered. They are respectively open science (OS), the corporation (CO) and the professional university (PU). The cost function of knowledge (KC) in the corporation mode is characterised by decreasing returns to scale because of the sharp effects of monitoring and screening activities. Principals are limited in their ability to evaluate the creative skills of their scientific workers, the levels of their efforts and even the value and the timing of their output. Formally it can be seen that the cost function of the corporate mode has a positive slope with a positive second derivative: (1) KC (CO) = a(q) with a′ > 0 and a″ > 0
p
CO
PU
OS
P(CO)
A
P(OS)
B
T P(PU) Z
C Q*
Figure 11.3 Alternative institutions for knowledge generation and dissemination
q
210 Governance of localised technological knowledge The cost function of knowledge in the open science mode (OS) is characterised by a two-party scheme. Total costs are made up of fixed and variable costs. Fixed costs are anticipated by the principal, i.e. the public sector that acts as an intermediary between taxpayers (ultimately firms) and the salaries of scientific workers. Variable costs account for the activities of dissemination and absorption that are necessary for the knowledge produced within the academic system to be effectively passed on to the rest of the economy. Formally we see a positive slope of the variable costs: (2) KC (OS) = T + b(q) with b′ > 0 and b″ = 0 Finally, the cost function of knowledge of the professional university (PU) is characterised by some fixed costs that are necessary to pay part-time scientific workers and variable costs for the generation of knowledge intra muros. There is the derived demand of firms which play an active role in the markets for research and professional services, firms which act as customers for intermediary inputs in the form of the production of new knowledge provided by certified creative minds who are part-time academics. This latter component has decreasing returns to scale, but to a lesser degree than in the case of the corporate mode. The basic activity of signalling provided by the certification of the academic system and the professional type of relationship that is established between firms and academics helps to reduce screening and assessment costs and hence to minimise agency costs. Formally we get: (3) KC (PU) = Z + c(q) with c′ > 0, c″ > 0, c″ < a″, T > Z Figure 11.3 provides a graphic expression of the basic argument. The vertical axis shows the difference in the costs of the given quantity of knowledge that a system requires. The costs of providing the desired quantity of knowledge (Q*) are greater with the corporate mode, especially if and when Q* is large. With low levels of Q* the corporate mode, the case in which the generation and dissemination of new knowledge rests upon the central role of in-house research and development laboratories funded and operated by large corporations, is more effective. The pure academic mode is more effective than the corporate mode as long as the slope of the communication costs variable is not too steep. The solution provided by the academic outsourcing mode is clearly the lowest. This reveals the competitive advantage of combining the academic provision of qualified and certified personnel who enter the markets for the provision of research and professional services on the supply side. On the demand side, firms are ready to purchase such services, on a case-by-case basis, with compensation schemes that are tied to specific performance and tasks within a freer organisation of the division of scientific labour, which is still managed by corporations. The latter however have lower levels of agency costs because there is less need to check and assess the performance, effort and creativity of their professional inputs. We can explore how this situation changes when the quantities of knowledge that are considered necessary for the system change. If we consider the region before
The new economics of the university 211 Q*, it seems clear that, the lower the quantity of knowledge is, the larger is the competitive advantage of the corporate solution. This can be regarded as a schematic representation of the period between 1950 and 1990. Research and development activities funded and performed by large corporations were the main source of knowledge. The university was assigned an ancillary role. Corporations tried to reduce the amount of public resources allocated to the academic system on the grounds that research and development activities carried out intra muros were far more efficient in terms of selecting the goals and objectives, performance and close interaction with users. The region identified by a quantity of knowledge slightly to the left of Q* can be considered a reliable approximation of the transition towards the knowledge economy at the end of the twentieth century. The corporate mode was becoming increasingly less effective. The traditional academic system where research was mainly carried out within universities and knowledge communication was expected to take place via the combination of scientific publications and graduates hired by corporations gained momentum. The limits of this traditional system of knowledge communication, however, quickly emerged as the main constraint and source of inefficiency. The region to the far right of Q* can be considered as the schematic representation of the new emerging knowledge economy where knowledge becomes an essential input for economic development, especially for advanced countries specialising in providing knowledge-intensive business services and high-tech products to the rest of the world economy in the twenty-first century. Now the limits of the corporate-based model become evident. The fast-increasing slope of the costs of knowledge produced mainly by corporations reveals all the drawbacks of the limits of organisations in managing principal–agent problems. Ageing scientific personnel with declining creativity employed by corporations become a burden, as there is little scope for useful job rotation. Opportunistic behaviour spreads. Firms are increasingly reluctant to fund large internal research laboratories as they have major problems in coping with low levels of predictability in the timing and content of scientific output while costs are rapidly increasing. Here the advantages of the renaissance of the university-based model become apparent. Provided that there is effective interaction between the academic and the business community, the new university mode of knowledge governance is clearly superior, especially if it is implemented effectively with: 1) the systematic application of long-standing traditional practices such as non-exclusivity of labour contracts; b) active participation of academics in the markets for knowledge; c) close interaction between research and didactic activity, particularly above college levels in graduate schools; and 4) non-exclusive intellectual property rights.
5 Conclusion and policy implications This new assessment of the role of knowledge indivisibility and external knowledge provides new arguments in defence of universities as knowledge commons. Now, however, the argument is reversed with respect to tradition, which was based upon the notion of knowledge as a public good.
212 Governance of localised technological knowledge The application of the basic tools of information economics to the economics of knowledge provides an interpretative framework that is able to evaluate and highlight one aspect of the institutional economics of the university that, so far, has attracted little attention. The academic system, because of its traditional characteristics, emerged through a historical process that has now lasted for over 900 years, since its origins in Bologna. Universities appear to possess a unique mix of incentives and rewards that makes them especially suitable to handle the deep and complex principal–agent problems that characterise the employment of creative talents at large. A shift in the role and the organisation of the university within the economic system has been taking place in the last 20 years. New aspects and facets of the academic institution emerge as key factors. The interaction between the academic and the business system is changing, and new mechanisms have been implemented. The open science model articulated by Partha Dasgupta and Paul David encapsulated the key characteristics of the economics of the university at the time of the corporate economy. Since then the general organisation of the production, dissemination and use of knowledge has been changing: a new model of the university seems to be emerging. An analysis of the academic system in the context of the principal–agent approach makes it possible to identify the factors that favour the evolution of the academic system towards a new model. More specifically, the application of the principal–agent approach provides a clue to understanding why the shift in the governance of knowledge generation and dissemination is taking place, away from the corporation-based model and towards the renaissance of an academically based model that encourages the active participation of academic workers in the markets for knowledge and the joint provision of educational and research activities. Clearly, adopting the principal–agent approach to understanding the advantages of the academic system provides important policy guidelines when it comes to implementing its positive aspects. Joint production of research and didactic activities and establishing non-exclusivity both in employment relations and in intellectual property rights over knowledge generated under contract are necessary to increase the viability of the academic system as a cornerstone of an effective organisation of the generation and dissemination of scientific and technological knowledge, based upon enhanced systemic knowledge transactions and interactions. An important implication of this approach and a strong reason for the implementation of the academic outsourcing mode of knowledge governance can be seen when the basic issue of the allocation of public funds among scientific disciplines is considered. When the principal–agent approach is applied and the academic outsourcing mechanism is implemented, the feedback signals from the markets for research and professional services towards the academic system can be better appreciated and measured. The static and dynamic characteristics of the demand for research and professional services can be considered an important input in the identification of the scientific fields where public funds should be allocated. Although it is in a limited time frame, the directions of the demand for research services can be considered reliable signals of the relevance of some scientific fields with respect to others. The provision of public funds can now be directed, taking into account such signals about the relative importance of some fields with respect
The new economics of the university 213 to others. Of course, the demand for research services provides direct funding itself. Hence the public bodies responsible for decision making regarding the allocation of public funds for academic research can assess whether such public funds should be used to defend minimum levels of knowledge creation in some fields and/or to further encourage the specialisation of the academic system in new, emerging fields where many firms are willing to purchase the professional services of certified creative scientists. Universities and public research centres play a central role in providing minimum levels of accumulation, generation and dissemination of general knowledge. The academic system turns out to be a viable institution not only to solve the knowledge trade-off between appropriation and dissemination but also, and mainly, because it is an effective institution for the management of creative talents. The unique blend of non-exclusivity in labour relations and joint production of educational and research services seems especially appropriate in implementing a two-ladder system of incentives and compensation. The university has the key role of a standardisation committee that certifies the quality of the scientific worker. It remunerates the didactic activities and the production of basic knowledge. Non-exclusivity in labour contracts, implemented by creating partnerships when teamwork is necessary, helps to create a supply of research and professional services. The matching of the demand for research and professional services by the business sector provides ample opportunities for second-tier compensation of the creative skills of certified scientific workers. The basic function of public funding to the knowledge commons is to defend efficiency thresholds in entertaining and implementing the stocks of knowledge across the board. All eventual progress depends on the multiplicative relationship between bits of knowledge and the key role of the stock of knowledge, and it is clear that a fall in the competence and expertise in a few knowledge modules can have dramatic consequences for the whole system. Minimum levels of efficiency have to be identified and presidia have to be created. Scientific presidia have to be kept both across scientific fields and across regional space. A public university system can be funded on the solid grounds of public funds, allocated with a clear methodology based upon the notion of knowledge fungibility. The wider the fungibility of each bit of knowledge is, the wider its relevance in terms of indivisibility is likely to be and hence its multiplier role for the whole system. The workings of the knowledge commons can be used to take advantage of the opportunities offered by firms’ demand for scientific and technological outsourcing in such a way as to use the available budget in selected areas, as long as there is a non-exclusivity clause regarding the knowledge generated and its dissemination is not prevented by obstacles created by proprietary assignment of the results of the research undertaken when there is private funding. The interaction between the public academic system and the market for knowledge-intensive services should be increased also as a way of exploiting the relevant economies of scale associated with the sheer size of some research facilities and the ubiquitous economies of density arising from the relevant fixed costs associated with the creation of dedicated skills with high levels of specialisation (David, 2004c).
214 Governance of localised technological knowledge In this context, universities and public research centres at large are pushed to enter the markets for knowledge on the supply side. Academic departments become suppliers of knowledge-intensive business services to firms that rely more and more on the outsourcing of research-intensive activities formerly carried out in their own laboratories. Knowledge generated by academic departments within the context of specific contracts with firms risks becoming subject to proprietary agreements with clear limitations on its dissemination. At the same time, however, much of information economics argues that the workings of competition in a market characterised by radical knowledge asymmetries provide an important counterbalancing effect when the role of signalling is appreciated. Academic departments have a strong incentive to signal to potential customers the quality of the research in progress and to disseminate information about the scientific scores. Academic publication, no longer viewed as the distinctive mission of publicly funded researchers, is now pursued as a signal to attract new potential customers for their services. There is far less spontaneous knowledge communication than there should be. When knowledge communication takes place at appropriate levels, it does so accidentally and occasionally in a few regional and institutional settings. Knowledge communication between the academic and the business community seems to be especially poor. Publications are not very effective as vectors of information about new scientific discoveries, which should be seen as possible areas of development and implementation of technological knowledge. The relationship between the top-down process of deductive ‘scientific’ work and the bottom-up generation of technological knowledge is often characterised as an ‘uneasy alliance’. The direct association and participation of scientists and technologists in common ventures seem to be able to reduce the gaps. Public policy can be the key component of a dynamic process which brings together universities and firms yet respects their basic mission: respectively the production and dissemination of generic knowledge with high levels of fungibility and its application in specific and idiosyncratic contexts. In conclusion, a closer analysis of the workings of the academic system reveals one more peculiar aspect of this old and yet evolving institution that has shaped and characterised the European economy for centuries. The university is indeed an efficient institution in solving the knowledge trade-off, that is to say, the contrasting need to increase the incentives to produce knowledge but also to disseminate it as much as possible. The university however is also an efficient institution in managing the generation of a highly unpredictable activity such as the generation of knowledge. The rapid transition of advanced economies towards a knowledge economy suggests that the advantages of the academic institution as a quasi-hierarchical system that makes it possible to select, provide incentives for and reward creative talents should be studied carefully. Its foundations might be imitated and applied to the rest of the economic system and extended to other institutional contexts. The workings of many professional communities seem to be very similar to that of the academic system in which a process of mutual interaction between the evolution of professional orders and academic systems seems to be at work.
12 Towards non-exclusive property rights Knowledge as an essential facility
1 Introduction Intellectual property rights have a twin effect on the economic system. On the one hand they favour the introduction of new technological knowledge. On the other they reduce competition and eventually may limit the rate of introduction of new knowledge. A trade-off takes place between such positive and negative effects. The application to the economics of knowledge of the notions of an essential facility and the liability rule can correct the balance of the trade-off and contribute to the rate of advance of technological knowledge and its effective use in the economic system. The tuning of exclusive property rights makes it possible to minimise knowledge rents and favour the dissemination and use of knowledge in the economic system, taking advantage of its intrinsic cumulability and complementarity. The rest of the chapter is structured as follows. Section 2 reviews the emerging evidence about the many facets of the knowledge trade-off. Section 3 introduces the notion of an essential facility, as it was first elaborated in telecommunication economics, and shows how it can be applied to the economics of knowledge. Section 4 elaborates the application of the liability rule to fix ex post the proper rewards for innovations that have been used. Section 5 elaborates the implications for knowledge governance. The conclusion summarises the main findings.
2 The knowledge trade-off reconsidered According to the basic foundations of the economics of knowledge laid down by Kenneth Arrow and Richard Nelson, technological knowledge, as an economic good, is characterised by a few relevant features: non-appropriability, non-rivalry in use and non-divisibility. Because of these features, tradability is hampered; hence arm’s length transactions in fully fledged markets cannot be used to coordinate in an efficient way the allocation of financial resources into research activities, their selection of the most rewarding directions and the necessary division of scientific and technological labour in their generation. This leads to low levels of funding, specialisation and efficiency and hence systematic undersupply. In this context intellectual property rights and specifically patents were thought to increase the rates of generation of technological knowledge.
216 Governance of localised technological knowledge Intellectual property rights and patents are institutional instruments designed to increase the incentives of firms to generate new technological knowledge and introduce technological innovations and to increase the viability of market coordination. So far, intellectual property rights are institutions designed to create markets and hence make possible all the advantages of unplanned and spontaneous coordination among agents, in terms of dissemination of information, signalling of new opportunities, division of labour and specialisation. By means of intellectual property rights impersonal transactions can take place, and the traditional coordination among agents within markets and among markets can take place without further public intervention. Intellectual property rights can be considered a market-creating activity: property rights on inventions make it possible to trade them, with all the well-known advantages in terms of division of labour and hence specialisation and productivity. Resources can be allocated in the factor markets, and knowledge can be exchanged both in the product markets and in the markets for intermediary production factors. Agents and firms can specialise in the generation of knowledge modules where each has a comparative advantage. Markets for knowledge both as an intermediary input for the production of new knowledge and as an input for the introduction of new technologies can flourish. Intellectual property rights granted to inventors lead to monopolistic market power in the markets for products that use the new knowledge. Such monopolistic power provides incentives to innovators to undertake risky activities finalised in the introduction of innovations. Monopoly makes it possible to increase both incentives and resources for the generation of new knowledge via the increased appropriability based on legal barriers to imitation. Thus intellectual property rights make it possible to grasp the advantages of the dynamic efficiency engendered by the increased amount of knowledge generated and hence the augmented flow of innovations. Now inventors are less scared by the risks of uncontrolled leakage of their knowledge and have an incentive to invest resources in research activities. Moreover, the rents stemming from the now proprietary knowledge can be used to fund additional research and hence the creation of further knowledge. Intellectual property rights have an important role from an informational viewpoint and so exert relevant consequences. In order to obtain a patent, as is well known, the inventor is obliged to disclose the original knowledge being patented. As a consequence, patents are the carriers of key information about scientific and technological advances. Patents moreover help to identify the inventors and to qualify the field in which the knowledge has been generated. From this viewpoint the role of patents as signalling mechanisms, which provide information about new inventions and relevant technological applications, seems at least as relevant as their traditional role of appropriability mechanisms based upon the enforcement of excludability. With a weak intellectual property right regime, the holders of each bit of knowledge have a much stronger incentive to rely upon industrial secrets as a way to reduce the informational leakage, with the radical reduction of the dissemination of the relevant bits of disembodied knowledge. Secrecy, the alternative to intellectual property rights, to secure exclusive ownership can have dramatic effects
Towards non-exclusive property rights 217 generally in terms of networking costs and specifically in the form of technological communication costs and hence upon the amount of knowledge complementarities which can be effectively activated. Intellectual property rights moreover are a remedy to tight vertical integration between the generation of new technological knowledge and its application to the production of new goods or to new production processes. The public good nature of technological knowledge pushes the knowledge-creating firm to use it as an intermediary input for the sequential production of economic goods. Vertical integration and direct embodiment of technological knowledge – within the borders of a single company – in the production of goods limit severely the emergence of markets for knowledge as a good per se, with negative consequences in terms of reduced scope of application of technological knowledge. When technological knowledge has high levels of fungibility, i.e. has a wide scope of application, vertical integration has strong negative effects, as it impedes the valorisation of the broader array of possible applications. In conclusion, intellectual property rights perform many positive functions in the economic system. First, they favour appropriability and hence secure rewards to inventors. In so doing intellectual property rights help to increase the incentives for the creation of technological knowledge and provide resources for its generation. Second, they favour the dissemination of knowledge as they make publicly available information about new technological advances. In so doing patents act as powerful signalling devices that may favour the distribution of resources among a variety of possible directions in the activities geared towards the generation of new technological knowledge. Third, they reduce the incentives to embody directly, by means of downstream vertical integration in the production of goods that use the new knowledge, and hence they limit the negative effects in terms of reduced scope of application of knowledge with high levels of fungibility. Fourth, patents play a major role as signalling devices, which helps the identification of the available bits of complementary knowledge and their owners so as to reduce search costs. Finally, they improve the viability of the markets for knowledge and facilitate interactions among holders of bits of complementary knowledge. Patents can help in reducing knowledge transaction costs in the markets for knowledge because they reduce information asymmetries and the risks of opportunistic behaviour and make it easier for demand and supply to meet by means of impersonal transactions in the marketplace where a large number of customers and sellers interact. Hence effective property right systems favour the creation of specialised and dedicated markets for disembodied technological knowledge where the firms can specialise in the production of knowledge as a good per se (Arora et al., 2001). Intellectual property rights however have many shortcomings and undesired effects. The literature has been adding new analytical evidence about many such negative effects (Boldrin and Levine, 2002; Jaffe and Lerner, 2004). 2.1 The first knowledge trade-off The foundations of the first trade-off between dynamic efficiency and static inefficiency are laid down in the context of competitive analysis. The first trade-off
218 Governance of localised technological knowledge consists in the identification by means of a classical cost–benefit analysis of the balance between the increased dynamics efficiency provided by patents, by means of increased appropriability and hence larger incentives to fund the production of knowledge, and the loss in static efficiency determined by patents, as ingredients for the creation of monopolistic market power in the downstream markets for goods. Monopolistic power reduces static efficiency. Firms can charge monopolistic prices and hence appropriate a large share of the total surplus stemming from the introduction and application of new knowledge. The understanding of the increased monopolistic market power engendered by intellectual property rights suggests limiting the scope for patents and their duration. The first trade-off has been traditionally regarded as a transient problem. The monopolistic market power in the markets for products based upon proprietary technological knowledge and the technological innovations stemming from its implementation was deemed to be temporary because of the Schumpeterian assumptions about the irreversible flows of entry of new competitors attracted by extra profits and able to invent around and imitate the original technological knowledge of the early incumbent. Hence the welfare losses generated by the divergence between marginal and average costs were assumed to be short lived. The short-term duration of monopolistic power in the markets for goods manufactured with the new knowledge seemed to be a solution to the trade-off between dynamic and static efficiency. 2.2 The second knowledge trade-off The second knowledge trade-off is identified as a result of a closer analysis of the implications of the notion of knowledge indivisibility. The new approach is based upon the discovery of knowledge cumulability, i.e. the diachronic complementarity between different vintages of knowledge. Following Newton, much emphasis is now given to a famous sentence of the English scientist: ‘To make science means standing on giants’ shoulders’. Intellectual property rights limit access to the new vintages of knowledge, at least for a considerable period of time: in so doing they delay the possibility for new generations of dwarfs to climb upon the shoulders of previous giants. Intellectual property rights now are seen not only as the cause of the static efficiency associated with monopolistic market power stemming from patents, but as a source of dynamic inefficiency as well. Intellectual property rights increase the incentives to generate new knowledge, but risk reducing dramatically the efficiency of the generation activity. Intellectual property rights limit the vertical or diachronic dissemination of knowledge: the access and use of prior vintages of knowledge are put at risk. The efficiency of the generation of new technological knowledge is now reduced by the delays in access to the last vintage of knowledge. New generations of inventors cannot rely upon previous progress made. Hence additional resources are necessary to rediscover what has been already invented. Duplication of efforts can take place. In the extreme case the generation of new knowledge can be inhibited by the duration of the life of the exclusive property
Towards non-exclusive property rights 219 rights assigned by patents to inventors. Intellectual property rights limit the working of knowledge cumulability. 2.3 The third knowledge trade-off The third knowledge trade-off stems directly from the second. It becomes evident when the effects of knowledge cumulability upon the duration of monopolistic power are grasped. Because of knowledge cumulability old inventors can build up new knowledge, taking advantage of lead times secured by exclusive intellectual property right regimes. Old inventors can retain for themselves and for a long stretch of time larger portions of the competitive advantages stemming from the repeated use and non-exhaustibility of the knowledge generated at time t–1 and appropriate all the stream of additional bits of knowledge which build upon the previous ones. Monopolistic rents are no longer temporary. Hence exclusive intellectual property rights engender the risk of contributing to the building of long-lasting static and dynamic inefficiency. 2.4 The fourth knowledge trade-off The fourth knowledge trade-off has been finally identified when the analysis of the indivisibility of knowledge has made it possible to appreciate the role of external knowledge as an essential intermediary input in the production process of new knowledge. Here in the economics of technological knowledge the issues of externalities on both the supply and the demand side become relevant and evident. The generation of technological knowledge is now considered to be characterised by relevant and necessary externalities, both technical and pecuniary. The notion of user interdependence makes its foray into the scene when agents value the levels of usage of other agents of certain goods. As far as scientific and technological knowledge is concerned, interdependence among users, hence on the demand side, is very strong. The chances of generating a new relevant bit of knowledge for each agent depend upon the levels of accumulation of skills and competence, education and access to information of the other agents in the community. The evidence, especially in new information and communication technologies, confirms that complementarity matters in assessing the rates of introduction of innovation. At each point in time the modules of technological knowledge possessed by each agent have high levels of complementarity with other modules of technological knowledge possessed by other firms. No firm can claim to be able to command all the relevant knowledge. External knowledge is an important input in the production process of new knowledge. This major progress is made when the special character of knowledge as a non-exhaustible good that is at the same time an output and an input into the production of other knowledge is grasped and retained at the core of the analysis. Here the derivation from the Arrovian notions of the non-excludability and non-divisibility of knowledge is clear. The horizontal or synchronic dissemination of knowledge is put at risk by strong intellectual property rights regimes. Poor dissemination and exclusivity put at risk
220 Governance of localised technological knowledge the access to external knowledge for each agent and hence the working of knowledge complementarity. Hence additional resources are necessary to rediscover what has been already invented elsewhere at the same time. Duplication of effort can take place. This reduces the future flow of additional units of new knowledge. In sum, intellectual property rights have clear advantages as market-creating institutions that favour the identification of the correct levels of incentives, the allocation of resources, the exchange in the marketplace of knowledge modules and hence higher levels of specialisation and efficiency. Intellectual property rights however engender at least three classes of negative effects. First, monopolistic power in the markets for the products embodying the new knowledge are less efficient for the monopolistic market power assigned to inventors, with clear losses in terms of static efficiency. Second, the efficiency in the generation of new knowledge is reduced by delays in the dissemination of prior knowledge, associated with the duration of patents. Such delays last as long as knowledge is made proprietary by patents. Third, the efficiency in the generation of new knowledge is hampered by the limitations in access to complementary sources of knowledge being generated in parallel at each point in time. Intellectual property rights play a key positive role in the provision of technological knowledge. Yet the many facets explored by economic analysis about their many negative and unintended consequences push towards an effort to reconsider the basic elements of the knowledge trade-off. The new understanding about the second and the third knowledge trade-off is crucial in this context for the new light shed on the dynamic inefficiency of the present intellectual property rights regimes in terms of missing opportunities to exploit latent increasing returns. Intellectual property rights are a necessary institution for the enhancement of the social capability to generate new technological knowledge. The present intellectual property rights however need to be improved in order to take advantage of the potential for increasing returns in the generation of new knowledge stemming from knowledge cumulability and knowledge complementarity. It seems clear that the present intellectual property rights regime impedes the working of such increasing returns, as there is a strong contradiction between the need to remunerate the generation of knowledge and the need to increase its dissemination and recombination.
3 An essential facility: economies of density and mandated interconnection – from the economics of telecommunications to the economics of knowledge The notion of an essential facility has been elaborated in the economics of telecommunications to regulate the problems raised by complementarity and cumulability. A production factor is an essential facility when its use in the production process is characterised by substantial indivisibility. Relevant economies of density take place when the repeated use of the same input is possible. Increasing returns take place because output increases with the increase of variable factors only. When long-term duration, excess capacity and little wear and tear characterise fixed inputs,
Towards non-exclusive property rights 221 marginal costs remain below average costs. Moreover incremental costs, i.e. the costs of additional production units, display low average costs, lower than total average costs. The social use of essential facilities requires strong regulation, as the rights of exclusive use have significant asymmetric effects on competition in the marketplace, which favour the exclusive users. When a piece of property acquires the characteristics of an essential facility, the rights to use, access and interconnection cannot be exclusive. A separation between the rights of ownership and the rights of use is necessary in order for actual and workable competition to be implemented and eventually made possible (Baumol and Sydak, 1994). As is well known, privatisation of networks and competition in the telecommunications industry has been made possible by mandated interconnection. Mandated interconnection has been a major factor of change and evolution in the definition of property rights. Ownership rights on the one hand and the rights of exclusive use on the other, traditionally associated in one single right, have been separated, and rights of use of the network have been separated from the ownership rights. Firms do and can own telecommunication networks and can claim their property on all the segments of the network, but they cannot claim any longer the right to exclusive usage. Other firms have the right to access the network and make a selective use of it. Dedicated authorities have been established since the late 1980s in most advanced countries in order to implement the right to interconnection, to regulate it and to fix the prices of interconnection (Fransman, 2002). Communication authorities have been established to monitor the effective separation between the right of ownership and the rights of usage of telecommunication networks. Their activity here is most necessary because of the everchanging conditions of the technology and hence the ever-changing conditions of the separation between ownership and usage. Second and most importantly, communications authorities have been established in order to fix ex ante the levels of interconnection tariffs. Interconnection tariffs must reflect properly the costs of the network and must make possible both appropriate returns on the investments for the owners and viable conditions of entry to new competitors. In order to avoid suboptimal provision of communication infrastructure, investors need to receive appropriate rewards and hence incentives for future investments. At the same time, however, newcomers must be put in conditions of actual cost symmetry in downstream markets with respect to incumbents and other competitors in the telecommunications industry (Madden, 2003). The evolution of property rights in the telecommunications industries has been the result of the understanding of the role of sunk costs and complementarities and their effects in terms of economies of density and incremental costs on the actual costs of both incumbents and new competitors in the industry. Mandated interconnection is indeed a significant departure from a full-fledged and traditional definition of property rights. A process of widespread generalisation of the application of the notion of an essential facility has been taking place since the last decade of the twentieth century. The separation between ownership and rights of exclusive use and the introduction of mandated interconnection are now regarded as a necessary regulation within economic and physical systems where and when complementarities and
222 Governance of localised technological knowledge indivisibilities matter, in order to restore and enforce the conditions for the viability of competitive markets. Such evolution of the property rights regime has been spreading from the original application in the telecommunications industry to all the network industries from electricity to gas and railways. There are today strong reasons to believe that the notion of an essential facility and mandated interconnection is directly relevant for the governance of technological knowledge. According to the results of much economics of knowledge, knowledge shares all the relevant characteristics of an essential facility. Knowledge is characterised by intrinsic indivisibility and yet it is dispersed and fragmented in a variety of uses and possessed by a variety of owners. Each bit of knowledge is complementary to the others along chains of weak and strong indivisibilities, which act both synchronically and diachronically. The exclusive access to each bit of knowledge can prevent others from cumulative undertakings (Antonelli, 2001, 2003a). Since the path-breaking contribution of Kenneth Arrow (1962a), the economics of knowledge builds upon the analysis of knowledge as a good per se and explores all the limitations to its production and dissemination in the marketplace that are engendered by its limited appropriability, non-divisibility and non-rivalrous use. Intellectual property rights regimes have been built mainly to increase the appropriability of the benefits generated by the introduction of new technological knowledge. The basic claim has been that the natural appropriability of knowledge is too low to induce investors to fund appropriate levels of research activities. Following the resource-based theory of the firm, however, technological knowledge cannot be separated from the firm. Technological knowledge can be considered both the primary input of the firm and its basic output. The firm exists because it is the institution that, by means of the valorisation and direction of learning processes, makes possible the accumulation of technological knowledge. At the same time the firm can be considered the basic tool of exploitation of new knowledge that cannot be sold as a good itself. The choice of whether to sell technological knowledge or to use it and make with it is especially relevant for the analysis of the firm.1 This approach can contribute to the debate on the governance of knowledge. A new appreciation of the role of intellectual property rights can now be found in the assessment of their positive effects from an informational viewpoint in terms of higher levels of specialisation and division of labour, rather than in terms of the assignment of exclusive property rights. From this viewpoint the so-called knowledge trade-off, that is the balanced assessment of both the positive effects of the monopolistic control of patents in terms of increased incentive to the supply of knowledge and the negative effects in terms of the reduced distribution of knowledge, needs to be reconsidered (Machlup and Penrose, 1950; David, 1993). From a welfare point of view, the rents associated with intellectual property rights could be regarded as a cost: a cost that society is ready to pay in order to increase the amount of goods that firms can produce and consume. Intellectual property rights and the related knowledge rents are a tool, an incentive and an allocative mechanism, designed to stimulate the efficient generation of new knowledge, hence
Towards non-exclusive property rights 223 new technological innovations, and by that means to increase the general efficiency of the economic system. From a welfare viewpoint neither the generation of knowledge nor the levels of the knowledge rents that are necessary to stir and fund the activities geared towards the generation of new knowledge are to be considered as a goal per se. The single, effective goal is in fact the increase in the general efficiency of the economic system and hence the amount of goods that can be produced with a given level of production factors. Along these lines a new framework can be articulated, one where intellectual property rights can be regarded as an institutional device designed to pay a fee to inventors and innovators in order to push their production of technological knowledge and the related introduction of new technologies towards the levels of dynamic efficiency that the system is not able to reach spontaneously. From this perspective it becomes clear that the knowledge rents associated with intellectual property rights should be minimised, under the clear constraint that such rents are necessary in order to fund the efficient production of new knowledge that engenders positive effects at the system level, by means of the introduction of new technological innovations, in terms of increased efficiency. The positive outcome of the new knowledge being generated can be easily measured by the consumer surplus stemming from any reduction in costs. Consistently it follows that the design of intellectual property rights should be modelled according to the results of the combined maximisation of the consumer surplus generated by new technological knowledge and the complementary minimisation of the levels of knowledge rents that need to be granted to inventors.
4 The liability rule as a mechanism for the governance of technological knowledge The separation between ownership and usage conditions and the extension of the notion of an essential facility to technological knowledge experienced in the case of the telecommunications industry can apply with success to intellectual property rights. The application of the notion of an essential facility and mandated interconnection to the governance of technological knowledge can be implemented by the adoption of the liability rule and the parallel reduction in the exclusivity of patents. Because knowledge is at the same time an output and an input in the production of new knowledge, exclusivity, traditionally associated with patents, is the cause of knowledge rationing, with major drawbacks in terms of both static and dynamic efficiency. This institutional innovation can improve the governance of technological knowledge and increase its rates of both dissemination and generation. The present intellectual property rights regime, based upon exclusive rights, suffers from at least four main problems: 1) Exclusive intellectual rights delivered to inventors reduce the allocative and technical efficiency in the product markets and favour their strategic use in oligopolistic rivalry (McDonald, 2004). 2) Exclusive intellectual rights delivered to inventors reduce dissemination of proprietary
224 Governance of localised technological knowledge knowledge and hence limit the dynamic efficiency of the system. Such effects are especially negative when knowledge complementarities apply and bits of knowledge can have important effects for the production of other knowledge in other fields of application, often remote from those of the original invention and introduction (Nelson, 2004). 3) Litigation costs and generally transaction costs, typically associated with the delivery and defence of exclusive intellectual property rights, have been growing at a dangerous pace, faster than investment in R&D (Barton, 2000). 4) Duplication of efforts and major coordination problems limit the general efficiency of the system in the production and use of knowledge. In this context, the present intellectual property rights regime can be improved substantially with the reduction of levels of exclusivity. Here the guidance provided by the analysis of the evolution of property rights, experienced with the application of the notion of an essential facility to technological knowledge and the related institutional innovations introduced in telecommunications, is central. The separation between the ownership of intellectual property and the right of exclusive use, already experienced with success in the telecommunications industry with the notion of mandated interconnection, can apply in this central and strategic area as well. Such a reduction of the exclusivity of intellectual property rights can be realised by means of the application of a compensatory liability regime. The extension of the liability rule in this field seems to be able to provide important positive effects. The application of a compensatory liability regime in intellectual property rights can be considered a useful device to implement ‘mandated interconnection’ in intellectual property rights. The application of a compensatory liability regime consists in the elimination of the exclusive rights of use of an intellectual property by the owner – the assignee of a patent – who is compensated by the right to claim for appropriate payments for the usage of his/her rights. In this context, the right of exclusive use is no longer associated with the rights of ownership of any intellectual property. Inventors have a clear incentive to hold such a patent: it entitles them to command a claim upon the rents stemming from the use of the proprietary knowledge by third parties. 4.1 The proposed regime and its antecedents Reichman (2000) deserves the credit for the first explicit and articulated analysis of the application of the compensatory liability regime to knowledge governance. The notion of a compensatory liability regime deserves, however, a generalised and extensive application to large-scale innovations in the broad markets for knowledge, well beyond the limitations of its original idea, which was applied only to small-grain-sized innovations within the boundaries of monopolistic competition. For this purpose it seems useful to trace the antecedents to the new proposed regime for intellectual property rights. A first important antecedent to the application of a compensatory liability regime, as already noted, is provided by the evolution of property rights in network industries. Mandated and compensated interconnection, introduced in telecommuni-
Towards non-exclusive property rights 225 cations networks and eventually in many other network industries, can be considered as a new form of non-exclusive ownership of a capital good that combines compensation to the owners of the infrastructure for all users with the right of third parties to access it. The separation of ownership rights from exclusivity rights means that the right of ownership to oblige users to comply with some obligations including the payment for the use of a proprietary asset is recognised as well as the right of other parties to take advantage of it. Compulsory licensing of course provides a useful reference. Compulsory licensing has often been advocated and debated. The generalised application of compulsory licensing however seems problematic because of the low protection recognised for the assignees of intellectual property rights. Too low levels of compensations have a twin negative effect: reducing the incentive to generate new knowledge and, even more threatening, reducing the incentive to patent new knowledge and, hence, risking favouring the recourse to secrecy. Ultimately compulsory and non-compensatory licensing is likely to exert negative effects in terms of the amount of new knowledge being both generated and disseminated. An important antecedent to the articulated application of a compensatory liability rule to managing intellectual property rights can be traced in the notion of the General Public Licence elaborated by Richard Stallman and Eben Moglen to implement the governance of open source software platforms (Stallman, 1998; Raymond, 1999). According to the General Public Licence, all agents can access prior and proprietary software provided that the owners of the original software are notified, new users make a clear and explicit reference to the original knowledge used and such users make available to third parties the results of their elaboration. In this context, the right of exclusive use is no longer associated with the rights of ownership of any intellectual property: users have free access but are obliged to meet some basic requirements. Public registration of the use and quotes to prior, proprietary software has a strong economic value as a main factor in building and increasing the reputation of their owners – the ‘inventors’ – with positive effects in terms of the levels of professional fees they can claim in the markets for professional services (Lessig, 1999).2 The generalised application of the compensatory liability rule to managing intellectual property rights can take advantage of the experience gathered with the elaboration of the General Public Licence in the open source software community. Three elements can be retained: 1) Users notify the assignee that the proprietary knowledge is being accessed and used, either for the production of additional knowledge or for direct use in the production of goods. 2) Users acknowledge the role of the original knowledge in the production of new knowledge, if this is the case. 3) Users agree that their use engenders a specific obligation in favour of the assignee of the patent being used. In this case the obligation takes the form of a payment of eventual fees to the assignee of the patent. Thus the ‘inventor’ still has a clear incentive to patent the new knowledge. The patent becomes the legitimate foundation of a claim upon the total surplus stemming from the direct or indirect economic application of the new knowledge. The drawbacks of secrecy are avoided. The advantages of patents as a powerful mechanism to enhance the dissemination
226 Governance of localised technological knowledge of information about new advances in scientific and technological knowledge become even stronger. The definition of a compensation mechanism able to engender the best mix of both static and dynamic efficiency becomes the corner point for an effective application of the new regime. 4.2 Implementation: cost-based versus revenue-based compensation schemes The definition of the ‘correct’ compensation mechanism is not simple. A major problem is whether the compensation for free access to the patented knowledge should be based upon the costs of the new knowledge or the revenue stemming from its application. Here the analogy between the new institutional arrangements put in place in network industries and intellectual property rights does not apply. Communication authorities have been successfully put in place and have performed their task of fixing the fair value of interconnection quite well. The creation of an independent authority charged with the task of assessing the ex ante value of a new piece of technological knowledge seems far less credible for the amount of ingenuity and Olympian, far-sighted rationality required. In the case of intellectual property rights, the definition of the cost of a successful patent and hence of the contributions due to the ‘inventor’, the equivalent of interconnection tariffs, seems difficult and problematic on many counts: the basic requirements of dynamic efficiency, i.e. the need to increase the allocation of resource to the generation of new knowledge, need to be properly assessed. First of all, research activities are characterised by high levels of risk and intrinsic uncertainty, in terms of the chances of generating an output, so that the allocation to each novelty of the effective costs is most difficult. Second, because of the role of cumulability in the production of knowledge, the identification of the actual levels of incremental costs, that is the identification of the specific costs for the last bit of knowledge as distinct from the sunk costs, is problematic. Third, technological knowledge is very much the result of the valorisation of bottom-up processes built upon learning processes and accumulated competence: it is difficult if not impossible to disentangle the specific cost items that can be charged: a major issue of indivisibility applies. Fourth and consequently, it seems that sheer costpricing cannot apply when technological knowledge is concerned.3 The role of creativity and ingenuity here is a key factor, and its costs impossible to assess. Moreover creative talent is a scarce and rare resource, which needs to be valorised and used in social applications which are more valuable.4 From the viewpoint of dynamic efficiency the correct reward for technological knowledge should be influenced by its value in terms of the total surplus stemming from its application and hence it is necessarily influenced by both supply and demand forces. A divide between the fair price for knowledge and its cost emerges. Relevant technologies are characterised not only by a larger derived demand curve, but also by lower levels of price elasticity, with higher levels of total surplus stemming from their industrial application.
Towards non-exclusive property rights 227 In the effort to define the compensation for the inventor stemming from the application of the liability rule to the governance of knowledge, it seems clear that the correct compensation for the inventor who no longer commands exclusive rights to the successful technological knowledge generated should be generated from a mimic of the working of Schumpeterian markets where the incentives for risky research activities are provided to successful innovators by the appropriation of the surplus generated by its application: the value of the technological knowledge is influenced by the demand, rather than by the sheer costs of the single research programme that is at the origin of the rare success. The issue then is the identification of criteria for the distribution of the total surplus between the inventor and the user(s) that attract new risky research activities and scarce creative talents without preventing the ‘free to use’ but ‘not free of charge’ access to the new relevant technological knowledge.5 The substitution of the property rule with the liability rule and its application to revenue-based compensation schemes based upon actual and measured – hence necessarily ex post – total surplus stemming from the application of the new technology can overcome the traditional problems associated with compulsory licensing. With the liability rule the definition of payments for the owner of the intellectual property rights used by third parties can be defined by taking into account both the demand and the supply side, that is after its use has been experienced. The identification of the economic value stemming from the application of a given specific piece of new knowledge requires that both demand and supply of knowledge be considered. The definition of a value for the unrestricted use of proprietary technological knowledge by third parties, based upon evidence about the full range of its economic effects and hence with the basic reference to the derived demand for the knowledge considered, seems the proper solution. By means of negotiations the parties involved can try to find an agreement about the share of the value stemming from the use of the technological knowledge that the owner should receive from the user. As soon as a patent holder realises that his/her proprietary, albeit no longer exclusive, knowledge has been put in place by a third party, a litigation procedure can be activated. The patent holder will claim a share of the actual economic value generated by the applications of the proprietary knowledge. The evidence about the effects of the use of the new knowledge can be gathered and an economic assessment elaborated. Formal litigation will be the extreme context in which the parties involved will solve the problem (Reichman, 2000). The judiciary system will enforce the procedure and define the correct share of the revenue stemming from the use of the new knowledge which should be paid by the user to the original owner. 4.3 Welfare effects A simple application of welfare analysis can help to identify the correct level of the share of the revenue stemming from the use of proprietary knowledge. From a system viewpoint the allocation of shares between the owner and the user(s) is most important.
228 Governance of localised technological knowledge The traditional Arrovian setting (Arrow, 1962a) can be used to assess the definition of the correct level of royalties and generally the criteria for the allocation of the revenue stemming from the use of the new knowledge. With the present intellectual property rights regime, the monopolistic owner has the right to take all the benefits stemming from the use of the new knowledge. In the context elaborated here the owners of technological knowledge should receive only a portion of the total benefits stemming from the application of their proprietary knowledge: as we have seen, knowledge dissemination has a key role in securing appropriate levels of generation of new knowledge. On the other hand, fair incentives to the generation of new knowledge are necessary to avoid the well-known risks of undersupply of such a key input for economic activity. The size of the royalty paid to the knowledge owner by the knowledge user is influenced by the size of the total surplus generated by the use of the knowledge. The size of the total surplus engendered by the introduction of an innovation of course is very much influenced by the kind of market forms, before and after the introduction of the innovation. Let us consider first the extreme case that competitive pricing applies before and after. Following Arrow (1962a), Figure 12.1 shows how the introduction of an innovation measured by a reduction in production costs from the levels of the costs before innovation (CBI) to the level of the costs after innovation (CAI) yields positive effects in terms of total surplus (TS) measured by the size of the difference between the area of the triangle DPB and the triangle DPA. This case can be confronted with the possibility that monopolistic pricing applies before and after the introduction of the innovation: in this latter case the amount of total surplus shrinks. Finally and consistently with the results of the analysis on the
P D d
PA
CBI
A
CAIR
C
PC PB
B
QA Figure 12.1 The effects of innovation
QC QB
CAI
Q
Towards non-exclusive property rights 229 dynamics of localised technological change we shall also assume that the downstream markets for the products that use knowledge as a production factor are characterised by monopolistic competition, with relevant barriers to entry and mobility. For this reason relevant extra profits are likely to persist in the long term.6 Specifically the relative mark-up, that is the difference between the amount of the costs incurred in the generation of the new knowledge (CK) and the royalties (R) paid to the inventor, weighted by the knowledge costs, can be set to be a function of the total surplus stemming from the introduction of the new knowledge in the markets for the products that use it: (1) (R – CK / CK) = f(TS) A maximum level of the relative mark-up needs to be identified. The rationale behind such a ceiling is clear: excess profitability for inventors would easily become an incentive to duplication efforts and inventing around activities with clear costs in terms of social welfare. Such a maximum level can be empirically defined with specific reference to the industrial context of application. The royalties paid to the knowledge owner will enter the costs of the firm and increase the market cost for the product and hence the price: see in Figure 12.1 the new costs after the introduction of the innovation now including the royalties (CAIR). In a competitive market, firms should account for the royalties to be paid. In a market characterised by monopolistic competition with barriers to entry and to mobility, royalties can be charged on the profits of the firm. Let us now try to show how the proposed extension of the liability rule can be implemented, with the identification of the optimum level of rents that it is necessary to pay to inventors in order to speed up the generation of knowledge in an economic system. 4.4 The identification of the optimum levels of knowledge rents: a simple geometric analysis The identification of the optimum level of rents that it is necessary to pay to inventors, i.e. to knowledge producers, is crucial. The analysis builds upon the notion of knowledge indivisibility elaborated by Kenneth Arrow (1962a, 1969). Knowledge indivisibility can be articulated in terms of cumulability and complementarity: the former is found when new knowledge builds upon previous knowledge. The latter is identified when knowledge in field A benefits from the contemporary advances in field B. In both cases existing knowledge is an input with low levels of exhaustibility and low costs of replication that can be used in the generation of new knowledge. The Arrovian analysis of knowledge cumulability and complementarity hence can be easily formalised in terms of economies of density. There are economies of density when low levels of exhaustibility characterise one or more production factors, among others. The larger the output is that can be produced with the repeated use of the same, given amount of non-exhaustible production factors, the lower are average costs.
230 Governance of localised technological knowledge 4.4.1 The analysis on the demand and the supply side Unlike Arrow (1962a), however, we frame the analysis in a derived demand context, where firms decide, on the base of the marginal productivity of knowledge, how much knowledge to use to introduce innovated products in downstream markets and how much knowledge to generate new knowledge. The demand for knowledge can be considered a derived demand expressed by two kinds of users: innovators and inventors. Innovators are firms that are engaged in the introduction of innovations in downstream product markets. Inventors are firms or agents that use existing knowledge as an input in the generation of new knowledge. The horizontal summation of these two quite distinct derived demand curves generates the aggregate derived demand for knowledge in the economic system. We shall assume that all the firms can generate the same knowledge with the traditional dynamics of multiple inventions. If firms can access external existing knowledge at costs that are below internal generation ones, they will use it rather than duplicate it. Let us consider first the case where both the demand and the supply of knowledge are elastic to knowledge rents. As Figure 12.2 presents, D1 identifies the derived demand for knowledge and the average costs of knowledge (ACK). The latter decline significantly if and when all potential users can take advantage of it: the working of the economics of density makes it possible to share the same fixed costs, necessary to generate a given piece of knowledge, among all eventual users. If, instead, exclusive intellectual property rights apply, all eventual users need to reinvent the piece of knowledge and average costs remain at a constant level. From the viewpoint of sheer costs, non-exclusivity of intellectual property rights seems to yield a clear social benefit. When the reduction of the incentives to generate new knowledge stemming from such a reduction in exclusivity is considered, however, the picture is worsened. Now agents are reluctant to fund research activities when the knowledge generated cannot be appropriated. A reduction in exclusivity engenders a reduction in appropriability. A reduction of knowledge appropriability reduces the profitability of its downstream applications, i.e. of the transient monopolistic rents stemming from the ensuing introduction of innovations. The lower the appropriability of knowledge and the duration of monopolistic rents are, the lower are the incentives to innovate and to engage in the generation of new technological knowledge. The schedule of the derived demand for knowledge is likely to shift to the left, because of the exit of a number of prospective investors and the general reduction in the levels of knowledge-generating activities. With exclusive intellectual property rights the traditional monopolistic equilibrium is found in A. In A the profitability for inventors is very high and hence there are incentives to fund R&D activities. When the equilibrium is in A, however, the costs of R&D activities are much higher as well. If non-exclusivity applies and the ACK′ is relevant, the new equilibrium, on the same demand curve, would be found in C. Clearly if the derived demand for knowledge were not affected by the reduction in exclusivity the system would benefit from the increased level of research activity R&DC.
Towards non-exclusive property rights 231
MR1
A
ACK D1 D2
D3 ACK
B F
C ACK′
R&DA
R&DF R&DC
R&D
Figure 12.2 Knowledge as an essential facility
The reduction in exclusivity however has a negative effect on the demand side. As the knowledge trade-off teaches, the reduction in exclusivity is likely to reduce the profitability of the knowledge generated: hence the derived demand for knowledge is expected to bunch back. If and when such a leftward movement from D1 towards D2 goes beyond the point B, it is clear that the amount of knowledge generated in the system shrinks, yet its costs also decrease. At the same time, however, the reduction in the costs for knowledge should engender a positive shift in the overall demand for research activities fuelled by both output and substitution effects. The goods manufactured with cheaper knowledge cost less in the final markets and their demand is larger. Moreover firms are now induced to substitute more knowledge for other production factors. In sum the new position of the derived demand for knowledge is likely to be affected by both a negative shift due to a reduction in monopolistic rents and a positive one, stemming from output and substitution dynamics. The actual position of the demand curve after taking into account both negative and positive effects should be D3; hence the equilibrium level of R&D expenditures should be found in R&DF. At the system level it is clear that the area between the ACK′ (with a negative slope because of the effects of economies of scope), the ACK parallel to the
232 Governance of localised technological knowledge horizontal axis (when exclusive property rights apply) and the actual demand curve defines the social benefit of non-exclusive access to proprietary knowledge from the supply side.7 The notion of knowledge as an essential facility makes it possible to implement an analysis of the introduction of the right of usage of proprietary knowledge by third parties, provided that the compensatory liability regime applies. Hence the users of proprietary knowledge generated by third parties being forced to pay a fee to the original inventors is likely to increase the amount of knowledge a system can generate. Figure 12.2 exhibits the working of the twin effects of the knowledge trade-off and shows with clarity that all reductions in exclusive intellectual property rights affect both the supply and the demand side with contradictory effects. Only the fine-tuning of both effects can yield positive aggregate effects in terms of increased levels of knowledge generated at the system level. Figure 12.2 shows how the definition of the levels of the appropriate levels of the mark-up for inventors must take into account both the effects on the supply of knowledge and the effects for its derived demand. The crucial issue is the elasticity of the position of the derived demand, together with the price elasticity on each of the schedules of the derived demand. When the price of proprietary knowledge is larger than its costs, and hence rents are granted to inventors and the total surplus is shared between producers and users, the derived demand shifts towards the right. When instead the mark-up is low and the price for proprietary knowledge is close to its costs, the derived demand shifts towards the left. Crucial points are: 1) the extent to which the derived demand is elastic to the levels of the mark-up in defining the equilibrium levels of the research activities, in terms of both slope and position; and 2) the extent to which the supply curve reflects the effects of the economies of density. Building upon this argument we can elaborate a methodology to identify the optimum level for the knowledge rents. The lack of a reward for the generation of new knowledge risks having dramatic effects on the position of the derived demand for technological knowledge in value terms. If the price for new technological knowledge is driven to zero by legislation in favour of free, unrestrained and unregulated access and use of new knowledge, and no reward is taken into account for undertaking the risky production of knowledge and subsequent introduction of innovations, no agent would invest in innovation activities and the position of the derived demand for technological knowledge would be pushed dramatically towards the origin. The levels of the prices and rents directly paid for the use of new technological knowledge influence the position of the derived demand for technological knowledge. At the same time, as already noted, the levels of knowledge rents are influenced by the shape of the supply curve for knowledge. If the rents are high, prospective users have a strong incentive to invent around and hence to use internal rather than external knowledge. The slope of the supply curve is no longer influenced by the working of the economies of density. If, instead, the costs of using existing and proprietary knowledge with no exclusive property rights attached are low, firms would rely more upon existing external knowledge, with clear effects in terms of
Towards non-exclusive property rights 233 a negative slope of the aggregate supply curve. The supply curve rotates upward according to the levels of the rents paid to use external, proprietary knowledge: the negative slope will be larger, the lower the levels of the knowledge rents. Now we can combine the two arguments. The levels of knowledge rents have a twin effect: 1) On the one hand the higher such rents are, the lower are the positive effects of the economies of density stemming from knowledge non-exhaustibility, as measured by the negative slope of the average cost curve with economies of density. 2) On the other hand, the higher such rents are, the higher is the position of the derived demand for technological knowledge. As a consequence, the higher the position of the demand curve for technological knowledge is, the greater are the benefits that stem from the negative slope of the average cost curve for technological knowledge. All the elements of the traditional trade-off are now encapsulated into our analytical frame, and the identification of an optimum level for knowledge rents can be easily calculated. The size of the total surplus generated in a market system by the interplay between a long-term supply curve for knowledge, characterised by a negative slope stemming from the economics of density, and a derived demand for knowledge depends upon the identification of the level of knowledge rents. The level is fair when it makes it possible to maximise the total surplus under the constraint that both the absolute value of the negative slope of the long-term supply for knowledge, as shaped by the effects of the economies of density in the production of knowledge, and the position of the derived demand are affected by its level. This point can easily be made clear with the help of the simple maximisation exercise represented in Figure 12.3. We have shown how knowledge appropriability has a twin effect. On the one hand it increases the revenue (RA) of the inventor. It
CA
RA
RA
CA
LA Figure 12.3 The knowledge appropriability trade-off
LA
234 Governance of localised technological knowledge is clear that the higher the levels of appropriability (LA) are, the greater the prices are for the knowledge generated by the inventor. On the other hand, because knowledge is also an input in the generation of new knowledge, it is clear that, the higher the levels of knowledge appropriability are, the greater the costs of new knowledge (CA) are. If we assume that the positive and negative effects of appropriability can be sketched in terms of a functional relationship with a positive first derivative and a negative second derivative, we see immediately that too much appropriability, as well as too little appropriability, exerts negative effects. There is a region of ‘optimum’ appropriability that is placed between the two extremes. The specific conditions at which appropriability levels affect the prices for knowledge as an output and its costs when it is considered as an input define the actual levels of ‘optimum’ appropriability. 4.4.2 The analysis on the supply side In this section we shall focus the analysis of alternative knowledge proprietary regimes upon the supply curve. We assume in other words that the demand curve for knowledge is given and we do not consider the negative effects of alternative proprietary regimes about knowledge upon the position of its derived demand. In this section we are considering, in other words, a derived demand where there are no effects of the reduction in prices of the products manufactured with knowledge as an input. This case applies when the demand for the products that use the knowledge is directly or indirectly influenced by the State. The demand for weapons is a clear example. Other examples might come from the public demand for drugs and medical equipment. Public regulation, for instance in the case of safety requirements or ecological constraints, has powerful effects in the definition of the levels of demand for specific products. In both cases we assume that public authorities are able to control the levels of competition and of the prices of the goods that use the knowledge. The unconditioned access to previous vintages of technological knowledge can yield substantial welfare gains from the strict viewpoint of the positive effects of the working of economies of density. When such economies of density in the production of knowledge are fully exploited, the system can experience an increase in the total surplus. When the access to previous vintages of technological knowledge is barred by either sheer legal impediments or high costs, the rate of generation of new knowledge is reduced. Moreover prospective users try to invent around, that is try to generate surrogate technological knowledge that can be used to generate additional knowledge. Inventing around is socially expensive, as it consists in the replication and duplication of efforts to access knowledge that exists already and would be easily accessible at reproduction costs that are a small fraction of the regeneration costs. It seems clear that the higher are the levels of royalties paid to knowledge producers, or the impediment to the free access to existing knowledge, the lower are the positive effects of the economies of density. The higher the rents for knowledge are, the lower is the negative slope of the supply curve.
Towards non-exclusive property rights 235 Along these lines it becomes clear that both exclusive intellectual property rights and high levels of royalties risk spoiling the advantages of the economies of density stemming from the non-exhaustibility of technological knowledge as an input into the production of new knowledge. At the same time, however, it is clear that the higher the price paid to inventors, the larger the number of prospective knowledge producers and hence the lower the position of the supply curve. Now we can combine the two arguments. The levels of rents paid to access existing proprietary knowledge have a clear twin effect: 1) on the one hand, the higher such rents are, the lower are the positive effects of the economies of density stemming from knowledge non-exhaustibility, but the larger the number of suppliers and hence the further on the right the position of the supply curve; 2) but, on the other, the lower the rents paid to inventors are, the smaller their number and hence the higher the position of the supply curve, but the greater the positive effects of economies of density, as measured by the negative slope of the average cost curve. Two quite distinct supply regimes can now be identified. In the first regime each inventor must rely primarily upon internal knowledge and can access only a limited portion of the existing, contemporary, knowledge. In the second regime, external knowledge can be freely accessed, although a rent has to be paid to the legitimate owner after a patent has been granted. In the first case the supply curve exhibits a positive slope but stays on the lower right of the quadrant. In the second case, the supply curve for knowledge has a negative slope but stays on the higher left of the quadrant. Figure 12.4 shows that the total surplus stemming from supply curve 1 is larger than the total surplus stemming from supply curve 2. Both supply curves incorporate the rents paid to inventors, calculated as a share of the costs. In supply curve 2 the
P S2
S1
D Q
Figure 12.4 The identification of the rents for knowledge
236 Governance of localised technological knowledge rents are higher and inventors benefit from exclusive property rights. In supply curve 1 inventors command a smaller rent, incorporated as before in their cost schedule, and property rights are not exclusive. As a consequence now, the basic conditions for the production of new knowledge are changed by the free access to external knowledge, paid a low level of rents, and the benefits of economies of density are fully exploited. New knowledge is effectively generated standing on the shoulders of giants: no limitations bar their access. The consumer surplus stemming from supply curve 1 is clearly larger than the consumer surplus stemming from supply curve 2. Figure 12.4 provides clear evidence that supply curve 2 would provide a larger consumer surplus if and when it happens to be beyond the point of intersection of the demand curve with supply curve 1. 4.4.3 Implementation All the elements of the traditional trade-off are now encapsulated into our analytical frame, and the identification of an optimum level for knowledge royalties can be tentatively calculated. The size of the consumer surplus generated in a market system by the interplay between the supply curve for knowledge, characterised by economics of density and sensitive to the levels of rents paid to inventors, and the demand for knowledge depends upon the identification of the level of the rents paid to knowledge producers. The rents define the share of the surplus that is paid to knowledge producers and the complementary share of consumer surplus. The rent is fair when, assuming that it is possible to guess the position of the derived demand, it makes it possible to maximise the surplus under the constraint that the absolute value of the negative slope of the long-term supply for knowledge, shaped by the effects of the economies of density in the production of knowledge, and its position, shaped by the number of firms active in the generation of knowledge, are both affected by its level. The ‘optimum rent’ for technological knowledge is influenced by: 1) the extent to which the supply curve reflects the effects of the economies of density; 2) the elasticity of the position of the supply curve, as shaped by the elasticity of entry; and 3) the extent to which the derived demand is elastic to the levels of the markup in defining the equilibrium levels of the research activities, in terms of both slope and position. The identification of the optimum level cannot be done by means of simple mathematical calculus, but rather by means of societal processes of trial and error. Ex post controls upon the effective social relevance of the knowledge used and generated are necessary. On an ex ante basis, different types of knowledge can be identified and different classes of knowledge rents can be created according to rules of thumb. It is clear that both knowledge users and knowledge producers need to know ex ante what kind of knowledge rents are applied to the kind of knowledge they want to generate. A more articulated legal framework should be implemented to make effective the social usage of the economic framework proposed. The definition of an ex ante temporary price is in any case necessary to allow the choice of the prospective user, whether and to what extent to rely upon internal
Towards non-exclusive property rights 237 or external knowledge as inputs for the generation of new knowledge. Hence a temporary price for the access to proprietary knowledge can be defined ex ante on the basis of the precautionary estimations of the expected demand. If the actual demand is larger, the actual price for the proprietary knowledge used by third parties can decrease. The ex ante price can only be greater than the actual ex post price. The ex post assessment about the actual position of the demand curve for knowledge is in any case necessary. When a new knowledge module has a radical and unexpected effect on a wide range of downstream activities, the position of the demand curve would be clearly far more to the right than anticipated, and hence the fair price for it far smaller. With such a framework, the tuning of the intellectual property rights regime and of the royalties that knowledge users should pay to knowledge producers so as to make possible at the same time the minimisation of the rents paid to inventors, the maximisation of the net social surplus stemming from the introduction of new technological knowledge and the maximum level of efficiency in the production of technological knowledge becomes possible.8 The application of the notion of an essential facility to the economics of knowledge and the implementation of the liability rule in the design of intellectual property rights help in this direction. When the derived demand for knowledge is triggered by a final demand for goods that is directly or indirectly influenced by the State, and it seems clear that the tradeoff is consequently reduced, the levels of the royalties may be kept to a minimum.
5 Implications for knowledge governance Building upon the Arrovian tradition of analysis, knowledge as a private good suffers from many and relevant limitations stemming from its well-known characteristics of non-appropriability, indivisibility and non-rivalry in use. Intellectual property rights are a necessary and yet amendable economic institution. It seems useful to try to assess the application of the notion of knowledge as an essential facility and of a compensatory liability regime based upon the participation of inventors in a share of the total surplus generated by the application of their proprietary but no longer exclusive intellectual property rights with the classic categories of the Arrovian analysis. It seems clear that the stronger are the effects of the indivisibility of technological knowledge in terms of cumulability, fungibility and compositeness, the stronger are the incentives to remove the exclusivity of property rights (Antonelli, 2006c). This is true for many reasons. First, from a general ‘technical’ efficiency viewpoint, when technological knowledge is characterised by high levels of cumulability, fungibility and complementarity the long-term cost curve exhibits a negative slope (see ACK′ in Figure 12.2) and the cost elasticity is steeper the greater are the levels of divisibility: hence the greater are the effects of knowledge economies of density and the greater the social waste stemming from exclusive intellectual property rights. Second, from an allocative viewpoint, exclusive intellectual property rights provide to initial inventors control over the sequence of additional bits of knowledge
238 Governance of localised technological knowledge that build on the previous bits, with clear asymmetric effects. With high levels of knowledge cumulability and knowledge fungibility, strong inventors are likely to be primarily sequential inventors. Third, relevant effects on the demand side can take place when knowledge compositeness matters. Knowledge compositeness is defined by the variety of specific knowledge units that are necessary to generate new knowledge in any specific field (Antonelli, 2005b). With high levels of knowledge compositeness, increased levels of access to a given bit of knowledge can exert strong widespread effects across the board, with typical network externalities. The larger is the number of users of a given piece of technological knowledge, the wider the incentive to increase its usage. In such a case the position of the derived demand for knowledge would not be changed towards the left by the introduction of non-exclusive intellectual property rights, but towards the right, with significant incremental benefits in terms of the levels of research and development activities undertaken in a given economic system. The implementation of the liability rule seems most promising in the presence of patent pools, i.e. when there are many patents for one innovation (Cohen et al, 2000; Reitzig, 2004). The literature refers in such cases to two quite distinct notions of patent thickets and patent fences. Groups of patents are ‘fences’ when they protect substitute technologies. Patent fences are used to exclude potential competitors from adjacent markets. Typically, in the case of patent fences, there are a large number of patents and one assignee. Groups of patents are thickets when knowledge complementarities are at work and many different original technologies, implemented by many ‘inventors’, are complementary in the generation of a new composite technology. In the case of patent thickets, there are often a large number of assignees. The liability rule should have a clear impact on ‘fences’ that would be no longer useful, because of the end of exclusive property rights, with strong positive welfare effects. The implementation of a compensatory liability rule should also have positive effects on the complex definition of both procedural and content contracts among the parties involved when patents as thickets are concerned. The holders of complementary units of knowledge retain the same incentive to participate in the collective undertaking based upon the royalties that stem from the successful introduction of a new technology. The value of the collective undertaking and hence the distribution of the benefits can now be made on an ex post basis, after the economic effects of the introduction of the new knowledge have been quantified in the marketplace. The objective definition of the outcome should reduce the coordination costs in building and managing the technological club. By the same token the suggested implementation of the intellectual property rights regime with the application of a compensatory liability regime seems especially effective when non-rivalry in use, another key feature of knowledge, is considered.9 The distinction between user value and exchange value rooted in the classical legacy comes in useful in this context. Knowledge is clearly a non-rivalrous in use, hence a public, good from the user-value viewpoint. But knowledge is quite a rivalrous, hence a private, good from the exchange-value viewpoint. In fact, the
Towards non-exclusive property rights 239 larger is the number of users of a new knowledge, the lower are the rents that can be extracted from its application. When everybody has unlimited access to the same unit of knowledge and everybody can apply it, no rents can be extracted from it. The contradiction between these two notions is key to understanding the need for an effective incentive system that guides the allocation of resources to the generation of new knowledge. The definition of a royalty-based liability rule to compensate for the loss of exclusive intellectual property rights seems an effective solution, able to save a portion of the rivalrous character of knowledge as a private good from the exchange-value viewpoint and yet to valorise the public good nature of knowledge from the viewpoint of user value. It is clear that the implementation of a revenue-based compensatory liability regime to knowledge governance can engender significant litigation costs that might diminish the positive impact of the suggested regime in terms of social welfare. The levels of possible litigation costs stemming from the suggested procedure need, however, to be confronted with the huge levels of actual litigation costs. According to Hall (2002), average, per patent, legal costs for litigation vary in the range of the astonishing level of US$400,000–2,500,000 in the US and US$50,000–500,000 in the European Union.10
6 Conclusion Technological knowledge is a collective, highly imperfect and heterogeneous activity. Moreover it is not only an output, but also an input, an essential intermediary production factor that is relevant both in the generation of new technological knowledge and in the generation of other goods. The dynamic efficiency of each firm and of the system at large depends upon the factors affecting the dissemination and the conditions of access to existing knowledge, as a basic essential facility. This analysis has clear consequences in terms of allocation of the knowledge total surplus. In the allocation of the total surplus stemming from knowledge indivisibility, a larger portion, in terms of consumer surplus, should be granted mainly to users, rather than to producers. Lower levels of exclusivity and lower rents for such technological knowledge seem useful also for competitive advantage. The larger are the effects of the economies of density in knowledge-generating activities, the stronger are the asymmetric advantages for incumbents who are old inventors. Such a new governance of intellectual property rights, based upon the notion of knowledge as an essential facility and hence the extension of the notion of mandated interconnection with the application of a compensatory liability regime, can balance the defence of intellectual property rights and the rewards stemming from the introduction of an original piece of knowledge with the need to increase the dissemination of relevant knowledge so as to favour its cumulative and competitive applications. The evolution of the intellectual property rights regime towards the separation between ownership and the exclusive right of access to knowledge can provide
240 Governance of localised technological knowledge important opportunities for the systematic valorisation of both the markets for technology and the interactions among holders of complementary bits of knowledge. The mandated right of interconnection to bits of knowledge owned by third parties can take place with the implementation of the compensatory liability regime and the ex post payment of royalties without the preliminary consensus of the patent holders. The reduction of the rights of exclusive use of intellectual property, and the introduction of the mandated right to access intellectual property for third parties, combined with the eventual enforcement of the compensatory liability regime such that the judiciary system can help in securing ex post the payment of fair levels of royalties to the effective owners, can become an effective institutional innovation. Intellectual property and hence patents can play a strong role in increasing the quality of the knowledge interactions. Full visibility of intellectual ownership can help in locating bits of complementary knowledge and hence in reducing the costs of technological communication and networking activities at large, especially when the parties can agree eventually upon the payment of appropriate royalties. By means of non-exclusive property rights, implemented by liability rules, knowledge interactions come closer to market transactions and hence increase the scope for the valorisation of knowledge complementarities. According to the localised technological change approach, technological change is the emergent property of an economic system if, when and where the latent complementarities among the fragmented bits of indivisible knowledge possessed by a myriad of agents dispersed and isolated are valorised and exploited. In this context, an open innovation system and hence non-exclusive intellectual property rights are far more productive (Chesbrough et al., 2006). The informational role of patents as carriers of relevant information about the actual levels of technological competence of agents and the availability of new bits of knowledge in this context is crucial. Technological signalling becomes relevant as a device to reduce knowledge transaction and networking costs. The appreciation of the informational role of patents has significant implications for their characteristics. With respect to the automatic granting of intellectual property rights, as in the case of copyrights, the selective and discretionary assignment of patents seems even more appropriate. The scrutiny of an authority is in fact most useful as a screening device which makes it possible to sort out the bits of new knowledge that are relevant and useful. For this very same reason, patents assigned following the first-to-invent procedure seem more useful than patents assigned with the first-to-file approach: the latter procedure better qualifies the content of the patent in terms of novelty and ingenuity. Second, it seems also clear that a narrow definition of the scope of a patent is more useful, from an informational viewpoint, than a wide one. The identification and location of the relevant bits in the great map of knowledge become easier for each prospective user. In such a context of governance of intellectual property rights, it seems clear that the granting of patents should be made easier and the fees charged for renewal should be lowered also so as to increase the role of patents as signals: patentees are now charged with far higher knowledge transaction costs in the form of litigation and judiciary activities.
Towards non-exclusive property rights 241 The costs of identification of imitations and the activation of the liability rule are now fully shifted to the undertaking of the original inventors. Patents are essential tools to signal the levels and the characteristics of the knowledge embodied in each organisation. A new chapter in the economics of intellectual property rights emerges here. Patents are no longer regarded only as tools to increase appropriability, but also as devices to increase transparency in the knowledge markets and hence facilitate market transactions. The new assessment of the informational role of intellectual property rights in terms of increased incentives to the production and trade of knowledge and hence a remedy to undersupply needs however to be reconsidered, because of the perverse effects of exclusion on the efficiency of the generation of new knowledge, especially when radical innovations are under question. The notion of knowledge as an essential facility becomes relevant. The extension and generalisation of the notion of an essential facility, elaborated in the telecommunications industry in the last decades of the twentieth century, are fruitful in the economics of knowledge and hence in the governance of knowledge commons. Let us spell out the main points through which the intellectual property regime should be redesigned, as follows: 1
2
3
4
5
Intellectual property rights should be granted to inventors in order to secure appropriability, disseminate information, prevent secrecy, reduce the incentives to vertical integration and favour the working of financial markets and of the markets for knowledge as an intermediary input for the production of new knowledge and as an input for the generation of technological innovations. Intellectual property rights do not include any longer exclusive property rights. Mandated, or compulsory, licensing applies: users can use the proprietary knowledge protected by intellectual property rights. Users should notify to inventors the actual use of proprietary knowledge. The application of the liability rule makes it possible to increase the general efficiency in the generation of new knowledge, as the working of knowledge complementarity and knowledge cumulability is no longer impeded by exclusive property rights. The use of proprietary knowledge is not free of charge. Patent assignees have the right to claim a royalty for the use of the proprietary knowledge. Royalties make sure that inventors receive a compensation for the risks and the costs associated with the activities that have been put in place in order to generate new knowledge. It is clear that mandated licensing with no royalties should expose inventors to the well-known negative effects of knowledge nonappropriability. The amount of the royalties is defined taking into account the benefits of the new technological knowledge assessed in terms of the social surplus that has been made possible by its use. The role of the derived demand for technological knowledge is central in this context. All attempts to guess a fair price for technological knowledge based only upon its costs seem doomed to failure. The methodology laid down with the maximisation of the net social surplus and the related minimisation of knowledge rents makes possible the definition
242 Governance of localised technological knowledge
6
of the appropriate levels of royalties that patent assignees should receive from users. The application of compulsory licensing without a methodology for the definition of the correct price for the use of the new technological knowledge risks being void of any actual use: either inventors would claim royalties that nobody would be ready to pay, or users would try to deny the economic value of the use of the proprietary knowledge. Both outcomes would have negative effects. The methodology based upon the joint assessment of the costs and the demand for technological knowledge and the related identification of fair levels of prices for technological knowledge can generate a clear framework.
In such a legal framework, stemming from our economic framework, patent assignees can try to fix a price for the usage of their knowledge. The owners of technological knowledge are fully informed about the use of their proprietary knowledge since they have received explicit notification by unrestricted users. After a reasonable period of time they can claim from the users their fair royalties. An assessment of the levels of knowledge prices fixed ex ante can be elaborated on the basis of actual economic impact. If and when contractual relations fail to identify a fair royalty, the judiciary system should intervene as the settler of the last resort.
Part III
The introduction of localised technological change
13 Localised technological change The benchmark
1 Introduction In the Schumpeterian tradition of analysis firms are able to change their technologies and to introduce innovations as a part of their competitive strategies. In the effort to generalise the Schumpeterian approach, the introduction of an innovation can be viewed as a part of a broader process of creative reaction to the changing conditions of both the product and the factor markets. Thus both the induced technological change and the post-Keynesian demand-pull traditions of analysis contribute to this generalised approach. Firms occasionally are faced with a mismatch between plans and actual market conditions. The mismatch is determined by the bounded rationality of firms that are not able to foresee all the possible changes that are likely to take place within the life horizon of their irreversible inputs. The introduction of technological innovations by suppliers and competitors is a strong cause of the mismatch. The changes in the factor markets affect the conditions of the firms, as much as the changing macroeconomic context. Firms are unable to foresee all the possible economic changes and the technological changes that will be introduced in the marketplace and yet are forced, at each point in time, to make irreversible decisions about their tangible and intangible production factors. Firms can face unexpected changes in their product and factor markets, changing either their technologies or their techniques. Firms can adjust passively to the new market conditions, moving on the existing map of techniques. Alternatively, they can consider the introduction of new technologies. Irreversibility of tangible and intangible production factors limits the possibility of changing the technique. The changes in techniques require that each firm is able to move on a given map of isoquants. Because of the effects of irreversibilities and limited knowledge, however, technical changes engender some switching costs and some costs in terms of missing opportunities for learning. The introduction of new technologies is a viable alternative when switching costs are high and technological opportunities are good. Localised learning provides the opportunity to introduce new technologies, although in a limited technical space. The introduction of new technologies however is not free: it requires dedicated resources, and specific activities must be carried on. A trade-off between technical change and technological change emerges, whether
246 Introduction of localised technological change changing just the technique, on the existing map of isoquants, or changing the technology and hence the shape of the isoquants. The trade-off will be tilted towards the introduction of technological changes when the access to knowledge is easy and conversely switching costs are huge (Atkinson and Stiglitz, 1969; David, 1975; Stiglitz, 1987; Antonelli, 1995). Because learning is the main source of new knowledge and learning is mainly local, and because of the irreversibility of production factors and layout, technological change is inherently localised. It is induced by changes in factor and product markets that cannot be accommodated by technical changes in a given map of isoquants and the related price and quantity adjustments. Hence it is constrained by irreversibility and based upon the localised opportunities for learning and generating new knowledge (Antonelli 1999a, 2001). Developing the notions of bounded rationality, local search and localised technological knowledge, innovation is viewed as the result of a local search induced by the difference between expectations and reality. Firms are myopic agents affected by bounded rationality and as such they are unable to correctly anticipate all the possible conditions in the world. Myopic firms are not able to rationally calculate all the costs and benefits to be derived from the introduction of innovation; moreover, they resist the introduction of all changes which would increase the burdens and the costly limitations of bounded rationality. Myopic agents however may be induced to innovate and introduce technological change when current conditions seem inappropriate and unexpected events occur.1 Even myopic firms are aware of the costs of not changing their productive and commercial set-up. The costs of not changing are then compared with the costs of introducing new technology. In such a context firms react to all changes in market demand and in the relative price of the factors of production only after some dedicated resources have been used to search for new and more convenient procedures. Consequently, firms make sequential yet myopic choices, reacting to a sequence of ‘unexpected changes’ in their business environment.2 When quasi-irreversibility applies, all changes in current business involve some adjustment costs that have to be accounted for. In such an approach, firms are portrayed as agents whose behaviour is constrained by the irreversible and static character of most of their material and human capital. Moreover the management of firms is affected by bounded rationality, which implies strong limits to their capability to search and elaborate information about markets, techniques and technology. Competence represents the basic irreversible factor of production. In turn competence is embodied both in the organisation of the firm and in its stock of fixed capital (Antonelli, 2003a). Innovation, the introduction of new technology, is the result of creative, reactive and sequential decision making set off by disequilibrium in both product and factor markets. Changes in the relative and absolute prices of the factors of production (as well as changes in demand conditions for their products) force firms to venture away from expected equilibrium conditions. There is a mismatch between the existing production pattern, which is the result of previous irreversible decisions
Localised technological change: the benchmark 247 regarding both fixed capital and labour. Such decisions were based on necessary but myopic expectations and the new situation created by unexpected changes in the product and factor markets. However, firms can adjust by changing their technology, and they can no longer be considered as reacting only by adjusting output or prices. The introduction of technological changes however is not free, and to a large extent it is the result of intentional acts. The introduction of new technology requires the investment of dedicated resources to carry out research and development activities, to acquire external knowledge and take advantage of new technological opportunities, to accumulate and articulate the benefits of experience and to use the tacit knowledge acquired in repeated processes of learning by doing, learning by using, learning by interacting with consumers and learning by purchasing. Each firm moreover cannot be analysed separately when the generation of new technological knowledge and the introduction of new technologies are being considered. The characteristics of the collective networks of innovators and the structure of interactive learning into which each firm is embedded play a major role here. The governance of knowledge, the effective access to external knowledge, and the quality and density of the knowledge communication channels in place between firms and learning institutions, such as universities and public research centres, and among firms – rivals, customers and suppliers – are determinant in assessing the amount of knowledge each firm can rely upon and hence are the crucial factors in favouring the extent to which firms rely on creative reactions instead of passive adjustments. In an economic system characterised by poor knowledge governance, firms can rely only on internal knowledge and hence are induced to face the mismatch more with passive adjustments than with creative reactions. Conversely in an economic system characterised by sophisticated knowledge governance procedures, firms can access high-quality external knowledge and use it in internal recombination processes that favour the creative reaction and the eventual introduction of localised technological knowledge.3 Formal modelling of these basic elements in a standard microeconomic framework provides the economic rationale for understanding the benchmark dynamics of localised technological change. The following sections, 2 and 3, present the elementary versions of the model. The conclusion introduces the rest of Part III and provides a guide to understanding the grafting of additional elements on to the benchmark model as it is implemented (see Chapters 14–17).
2 The basic model with strong irreversibility The notion of irreversibility of production factors is embedded in basic microeconomics. As is well known, since the Marshallian distinction between short-term and long-term, the S-shape of the total cost curve and the consequent U-shape of the average and marginal cost curves are a direct consequence of the irreversibility of fixed capital in the short term. The irreversibility of capital is not a new attribute just discovered, but a founding element of basic microeconomics. It seems appropriate to stress and recall that standard cost theory, the theory of the firm and
248 Introduction of localised technological change consequently the theory of the market are all short-term theories, that is theories that build upon the irreversibility of capital. The very distinction between shortterm and long-term depends upon the stretch of the time period into which it is not possible to change the capital endowment selected at each point in time. The notion of irreversibility, however, has rarely taken a historical understanding. Little analysis has been devoted to the amount of historical time that is necessary for overcoming the constraints of irreversibility. When the necessary historical dimension of irreversibility is taken into account, the analysis of the context in which firms select their conduct, assess their performance and elaborate a strategy needs to include the scope for creative reaction that is the possibility that firms will react to the constraints of irreversibility and change their technology accordingly. The central notion of quasi-irreversibility is based upon the possibility, for firms constrained by the lack of malleability of some production factors, of reacting and adapting to such conditions by means of the introduction of technological and organisational innovations. Irreversibility instead takes place, as in microeconomic textbooks, when, in the short term, firms cannot do any better than accept the negative consequences of the irreversible production factors chosen in previous decision making. When a significant portion of production factors is characterised by strong irreversibility, such that the amount of capital selected in the previous period of time cannot be changed at all, firms, exposed to changes in the relative prices of factor costs and/or in the levels of their demand, incur significant declines in technical efficiency. In these conditions firms are induced to try to find a solution in the form of a new technology. In this case they will incur innovation costs, i.e. the costs of implementing their tacit knowledge and actually changing their technology. More precisely we consider two cases: when changes take place in the product markets and hence the demand for the output of the firm changes and when changes take place in the factor markets and the relative prices of production factors are affected. Let us consider the first case. At time t, the firm has identified the correct levels of output and budget and selects the equilibrium position such that the desired isoquant is tangent to the appropriate isocost, defined by the ratio of wages W to rental costs R. In Figure 13.1 the firm is in equilibrium at point E1. When demand changes, the textbook firm should change the levels of inputs, with a given technique, in order to expand output to the new desired level. In other words, that firm would reach, moving along the isocline, the point E2, on the new desired isoquant placed further to the right on the same map. Such a process however is not possible in the short term because of the fixed production factor K that had been selected at time t–1. In the ‘short’ term or, more generally, when irreversibility applies, the firm cannot move along the isocline so as to reach point E2. In order to produce the new increased level of output the firm can go to point A, identified by the intersection between the new isoquant 2 and the axis of the fixed endowment of the irreversible production factor K. The solution A, however, is inferior and out of equilibrium both because of the ‘wrong’ capital
Localised technological change: the benchmark 249
K
E2 E1 C
A
L Figure 13.1 Change in output with irreversible fixed capital
intensity, which is lower than the levels suggested by the slopes of the isoquant and the isocost and, consequently, because of the higher costs incurred, easily quantified by the distance between C and A, i.e. the intersection between the minimum isocost and the axis of the fixed capital. In such conditions the firm cannot face the competition of other firms which happen to have selected the proper output level and do not want to change it. The firm however can change the situation, by means of a creative reaction, and innovate so as to change its technology. Such a firm can generate and introduce a new technology, such that the equivalent isoquant, producing the same output 2, is tangent to the isocost in point C and compatible with the endowment of fixed capital. The introduction of the new technology is induced by the changes in the levels of output and the constraints imposed by the strong irreversibility of the fixed production factor. Let us now consider the second case, which takes place when the firm is exposed to changes in the factor markets (see Figure 13.2). For the sake of geometric clarity we shall consider a compensating change in relative prices, creating a new level of wages W′ and rental costs R′. When such changes in factor markets take place, the firm of the standard microeconomics textbook would choose the new technique B where the new marginal rate of substitution equals the slope of the new relative prices. Such a solution, however, implies change in the levels of superfixed production factors: typically it is a very long-term solution, which cannot be considered.
250 Introduction of localised technological change It is clear, in fact, that our context of analysis is an extension of the time horizon of the traditional short-term cost and production analysis. The firm with fixed production factors cannot do any better than stay in technique A, far away from point C defined by the intersection between the new isoquant and the endowment axis. In this case, again, the firm is bound to experience a decline in price efficiency, which is measured by the distance between points A and C (Farrell, 1957). Once more, the firm can face the decline in efficiency only by means of the introduction of a new technology which makes it possible to identify an equivalent isoquant that is able to be tangent to the new isocost W′/R′ exactly at the point C and, hence, fully compatible with the endowment of the fixed factor selected with the old isocost in the old factor market conditions. In sum, it should be clear that all changes in the business environment, with respect to expected levels of output and factor costs, generate a reduction in the general efficiency which is directly functional to the distance on the isoquant map between the new desired equilibrium point and the one actually possible, clearly defined, for the new output, by the intersection between the isoquant and the endowment of superfixed production factors. In these circumstances, the implementation of tacit knowledge, the generation of localised technological knowledge and the introduction of new technologies that make possible the re-establishment of the efficiency conditions become a viable alternative to the quasi-losses.4 This inducement is all the stronger, the more competitive the marketplace, the lower the barriers to entry, and the larger the variety of firms in terms of superfixed
K B
C
A
L Figure 13.2 Changes in factor price with irreversible fixed capital
Localised technological change: the benchmark 251 production factors endowment. In a highly competitive marketplace, quasi-losses directly affect the marketplace for incumbents, which face new competitors that are not constrained by the superfixed production factors endowment and hence are more efficient: quasi-losses may soon become losses. The same is true when incumbents differ in terms of the stock of irreversible production factors: the larger it is, the heavier are the quasi-losses, and hence there is the risk of facing not only a decline in profitability but also the emergence of actual losses. Hence the larger is the variety of firms and the more competitive is the marketplace, the more likely is the introduction of localised technological changes that enable firms to restore their output and price efficiency. Finally, and by the same token, the larger is the entropy in factor markets and the larger is their turbulence, the stronger is likely to be the inducement to generate new localised technological knowledge and introduce localised technological changes. On these bases we can now turn to a brief formal exposition of the model. We can write the revenue stemming from the introduction of localised technological changes to cope with the changes in the relative prices of inputs and in the demand for incumbents characterised by relevant superfixed production factors, as follows: (1) RI = f(DAA′) where RI is the revenue in terms of the reduction in the use of the flexible production factor measured by the technical distance between A, on the isocost line, and any new solution A′ towards and possibly beyond C on the endowment axis, determined CTI RI
DAA′ Figure 13.3 Decision making for the introduction of localised technological changes with strong irreversibility of production factors
252 Introduction of localised technological change by factor irreversibility, that localised technological changes make it possible to obtain. We can now turn our attention to the role of technological knowledge and technological change. We assume that, because of the strong role of learning in acquiring the tacit character of the localised knowledge that is necessary to innovate and for the complementarity between tacit knowledge and R&D activities, the search for new technologies is especially productive along the endowment of irreversible production factors expressed by the endowment axis AA′. All new technologies that reshape the isoquant map along the endowment axis AA′ enable the productive factors to be used rationally so as to restore the general price efficiency of the firm and lead to an overall increase in efficiency and eventually in total factor productivity. The introduction of new and better technology however is costly. In order to capitalise on the tacit knowledge acquired by means of the learning that has been going on in the techniques being used, firms have to invest in formal R&D activities. A systematic search for available external knowledge is also necessary, and relevant communication costs are associated with this. This research process can stop when the new technology is such that the firm reaches point C. The firm is again in equilibrium, because the marginal rate of substitution again equals the slope of the new relative prices. Further movements along the AA′ axis, beyond C, however, are welcomed. They are likely to generate an increase in total factor productivity in absolute terms. The costs of innovation activities necessary to move the isoquant along the endowment axis AA′ towards point C, and further to the left, are a function of the leftward distance from the original equilibrium technique: (2) CI = g(dAA′) where CI represents the costs of the resources dedicated to research activities, i.e. to implementing learning procedures, building technological communication channels with other firms and with other research institutions, and operating R&D laboratories and broadly all of the activities directed towards the introduction of the technological changes that are necessary to reshape the isoquant so as to move along the line AA′ and restore an equilibrium condition. A firm that chooses to stay in technique A incurs a decline in the general price and output efficiency but avoids all innovation costs. Conversely, a firm that chooses to introduce a new technology that makes the technique C efficient, i.e. such that there will be a tangent solution along the endowment axis between the new isocost and an equivalent isoquant, is able to avoid the decline in efficiency but incurs substantial innovation costs. We are now in a position to portray the decision process of the firm with the standard tools of profit maximisation. The profit equation for the firm reads as follows: (3) P = RI(dAA′) – CI(dAA′)
Localised technological change: the benchmark 253 where RI stands for the gross revenue from adjusting to the new factor prices and demand levels, measured in terms of the reductions in production costs, with respect to the distance between the technique C and the technique A, made possible by the introduction of technological changes that reduce the price and output inefficiency. In other words the revenue of changing the technology along the line AA′ consists in the reduction of variable costs with respect to A.5 CI(dAA’) are the innovation costs and can be measured in terms of the distance on the endowment line between A and C and beyond. Standard maximisation of the profit equation enables the identification of the ‘correct’ amount of research expenditures a firm can bear. Our geometric approach makes it possible to relate the amount of innovation selected directly to the technical inefficiency arising from superfixed production factors so as to establish a tradeoff between technical inefficiency and technological innovation. Maximisation here identifies the ‘best’ distance on the endowment line a firm can walk by means of the introduction of localised technological changes, induced by the twin constraints of a production process characterised by heavy superfixed inputs and changes in both demand and relative prices in the business environment. With low innovation costs the incumbent may be induced to cover a long distance on the endowment axis and go beyond the intersection between the new isocost and the endowment line so as to introduce significant technological changes which enable the substantial increase of total factor productivity. With high innovation costs, incumbents will be induced to make only incremental innovations so as to reduce the distance on the endowment axis between A and the intersection between the isocost and the endowment lines. In these circumstance incumbents can only reduce quasi-losses and come closer to the best practice. The introduction of localised technological changes along the endowment axis will make it possible for firms that cope with changes in both product and factor markets to adjust the ratio of marginal productivities, increasing the usage intensity of the irreversible production factor. The direction of technological change will be shaped by the endowment of superfixed production factors: the new technology will save on flexible production factors and conversely will be irreversible factor intensive. The rate of technological change in turn will be affected by the levels of entropy of the business environment and the share of superfixed production factors on total costs. The greater both the former and the latter are, the greater is the inducement to rely upon technological change in order to cope with the new price and market conditions. In general we see that the process is likely to reduce the quasi-losses associated with the ‘wrong’ mix of production factors engendered by the presence of superfixed features in the production mix. Specifically, moreover, it is clear that with efficient innovation activities and high levels of technological and learning opportunities firms can go beyond the ‘equilibrium’ point which simply restores the desired ‘Farrell’ efficiency conditions and finds new technological conditions that increase total factor productivity. For given levels of endowment of irreversible production factors and entropy in factor and product markets, an economic system is likely to increase its total factor productivity levels the greater the efficiency of innovation
254 Introduction of localised technological change activities and the more conducive the local innovation system to generating new technological knowledge (Nelson, 1982). At the end of this analysis it is clear that the relationships between irreversibility and innovation appear quite complex: irreversibility stirs the generation of new localised knowledge and the introduction of new localised technological innovations. The introduction of localised technological change becomes a means and a tool to increase the flexibility of firms, copying with fluctuations in both product and factor markets which cannot be accommodated by the existing stocks of installed fixed capital goods and other intangible production factors.
3 The basic model with weak irreversibility Let us now consider the case of the weaker form of irreversibility: fixed production factors can be altered by means of dedicated activities and consequent switching costs, which keep them within a limited technical area and prevent significant changes being made to the input mix. In order to cope with the constraints imposed by factor irreversibility, firms introduce technological changes. Such technological changes are localised both by the specific constraints and by the opportunities provided by the processes of learning about the specific techniques in use and hence improving them and relying upon the surrounding competence of customers, providers and competitors. Myopic firms are induced to deal with the dynamics of product markets6 and hence with the levels of demand for the firm and the dynamics of factor markets and hence with the costs of inputs by introducing technological innovations and making adjustments in response to market fluctuations while retaining the previous input levels as much as possible; hence they change the technology locally, determined by the relative costs of introducing innovation.7 In Figure 13.4 we see that a change in relative factor price affects the viability of the previous equilibrium E1. The firm can either change the technique and move to E2 or change the technology by means of the introduction of technological innovations, so as to find a new equilibrium in the proximity of the isocline O E1, in E3 or (possibly) beyond.8 The outcome will depend upon the levels of switching costs, that is the amount of resources that are necessary to perform all the activities to move from E1 to E2, compared to the amount of resources that are necessary to innovate and move towards and beyond E3.9 The resilience in the old equilibrium point E1 is out of the question: the firm produces at costs that are well above the levels of other firms, typically new firms with lower levels of irreversible factors, which are able to produce in the new equilibrium point E2. The firm is now exposed to a clear decline in the levels of performance and of satisfaction. A reaction is necessary: it can be a passive one consisting in the traditional technical change defined as a movement in the space of existing isoquants or a more creative one so as to include a change in routines and the eventual introduction of innovations. The difference between current profits, after the changes in the marketplace, and the profits that should have been possible without such changes measures the amount of resource the firm is ready to commit
Localised technological change: the benchmark 255
Capital
E2
E1 E3
O
Labour
Figure 13.4 The trade-off between technical change and technological change
in order to bring about the changes that are likely to restore the expected levels of profitability. In other words, because of the mismatch between expectations and the actual conditions in the marketplace, the firm cannot remain in the position that had been planned. A budget for adjustment costs has to be allocated when the mismatch arises. Consequently it is clear that: 1) the larger the mismatch is, the larger the budget available must be; and 2) the greater the flow is of technological innovations being introduced in the markets for inputs and products, the larger the mismatch is for each firm. All adjustments are possible but are costly. Technical change, because of irreversibility of existing production factors and limited knowledge about the existing techniques, requires some switching activity. The introduction of technological innovations is a viable alternative to technical change. Technological change on the other hand, by definition, is not on the shelf, and its introduction in turn requires some innovation activities. The position of the frontier of possible adjustments is defined by the amount of resources that the firm should invest just to move from the previous equilibrium technique to the new one, either on the existing map of isoquants or on a new map. The search for the correct solution is identified as a maximisation process where the firm tries to maximise the amount of changes, including technological innovations, that can be generated with a given amount of resources set by the levels of switching costs.10 In this model all changes in the production pattern and hence all movements on the existing map of isoquants, either on a given isoquant or from one isoquant to another, but still on the same map, require some switching resources (SWR) to be used.11 Formally the following definition is given:
256 Introduction of localised technological change (4) SWR = Z(dK/K, dL/L) where dK/K and dL/L are defined as the changes in the levels of irreversible inputs which are necessary in order to satisfy the new unexpected levels of demand and factor prices, and SW stands for switching costs.12 The firm can either adjust to the new factor market conditions, by changing its position within the existing area of techniques, defined by the existing technology, or react, in a creative way, by introducing an innovation which makes it possible to change the technology and hence the area of techniques.13 The firm is now set to consider the fundamental trade-off between the costs of switching incurred by technical change in the existing technical area and the costs of introducing technological changes which reshape the technical area. The firm can identify the correct solution by means of the standard maximisation of the output, with a given frontier of possible adjustments, when a proper isorevenue is defined. The absolute levels of the revenue generated by all adjustment activities consisting in the revenue made possible by the introduction of new techniques and the revenue made possible by the introduction of the new technologies respectively define the isorevenue. Formally we see the following relations: (5) TC = b(R) (6) SW = c(R) In equation (5) TC measures the amount of technological innovation necessary to change the technical space that the firm can generate, taking into account the internal competence and knowledge accumulated and the external knowledge it can access. In equation (6) SW measures the amount of technical change necessary to match the required switching14 so as to move in the existing technical space and reflects the levels of irreversibility and rigidity of tangible and intangible capital. Much work has been done in the localised technological change approach, to inquire into the conditions, characteristics and determinants of the trade-off between technical change and technological change. The introduction of technological changes is possible only if appropriate amounts of knowledge and competence, both internal and external to firms, have been accumulated and are available to firms. The conditions of the learning processes and the determinants of the eventual production of knowledge have received much attention. The costs of technical change, on the other hand, are influenced by the irreversibility of the commitments made by firms and the costs of adjusting fixed production factors, both tangible and intangible, to new and unexpected product and factor markets. It is clear that the relationship between switching, i.e. changing the technique, and innovation, i.e. changing the technology, is essential to defining the outcome of the search process initiated by the changes in the product and factor markets. It seems clear that the greater the efficiency is in the production of technological changes and the lower the efficiency of switching, the larger is the amount of
Localised technological change: the benchmark 257 innovations introduced. Correspondingly, the smaller the efficiency of research activities is, the smaller will be the amount of innovations each firm will generate. The firm will adjust to the new factor and product market conditions more by means of switching activities than by means of the introduction of new technologies. The extent to which the firm will rely on either switching or innovations will be influenced by the relative efficiency of each activity and by the shape of the relevant isorevenue. This is well represented by the shape of the frontier of possible adjustments presented in Figure 13.5. A symmetric shape would take place when switching activities are as efficient as the research activities that are required for the introduction of new technologies. When the intercept on the horizontal axis is larger than the intercept on the vertical axis, research activities are more efficient than switching ones. In a standard elementary textbook of microeconomics, with no irreversibility, no limited knowledge, perfect access to all the technical knowledge available15 and smooth adjustment processes, it is clear that switching costs are negligible and hence the intercept on the vertical axis would be much larger than the intercept on the horizontal one. At the system level it is clear that the larger is the number of firms that change their technology the greater the discrepancy is for each myopic firm and actual conditions of product and factor markets and hence the greater is the burden of adjustment activities that are required: in this case the position becomes endogenous, as it is affected by cumulative processes at the system level. At the system level the dynamics of knowledge compositeness also matters: the larger the number of firms engaged in complementary research activities is, the larger are likely to be the positive effects of external knowledge on the production of new knowledge: hence the larger will be the horizontal intercept. By the same token
Technical change (SW) E
Technological change (TC) Figure 13.5 The frontier of possible adjustments
258 Introduction of localised technological change it is clear that the more efficient the systems of knowledge governance the larger will be the horizontal intercept. With respect to Figure 13.5, the slope of the isorevenue can be considered equal to unity in the simple case. In this case it is assumed that the unit revenue stemming from switching matches exactly the unit revenue stemming from the introduction of new technologies that make it possible to reach the same new isoquant. By definition the two alternatives yield exactly the same benefit simply because they make it possible to fetch two points that belong to the same new isocost. It should be clear however that, as soon as the introduction of new technologies makes it possible to increase total factor productivity because it makes it possible to fetch a lower isocost so as to go beyond the new isocost towards the origin, then clearly the benefits stemming from the introduction of the new technology are greater than the benefits stemming from sheer switching. Hence the map of isorevenues is clearly bounded: when the amount of technological change is great, the slope of the isorevenue should also change so as to reflect the higher benefits of increased total factor productivity. To make this point more compact, let us now assume that a frontier of possible adjustments can be considered, such that, for a given amount of resources (R) necessary to face the mismatch, firms can generate an amount of either technological change (TC) or technical change (SW). Specifically the shape and the slope of the frontier of creative adoptions reflect the effects of the technological opportunities based upon the localised competence built by means of internal learning by doing and the opportunities offered by the knowledge and the technologies generated by third parties that become available either by means of imitation or by the active push of upstream suppliers. Formally this amounts to saying that: (7) SW = d(TC) In order for standard optimisation procedures to be operationalised, an isorevenue function needs to be set. The revenue of adjustments (RA) compares the revenue that adjustments by switching in the technical space (SW) yield with respect to the revenue of technological change (RTC). Formally we see: (8) RA = s SW + t RTC where s and t measure the unit revenue of switching and the unit revenue of technological change generated with the given amount of resources available to face unexpected changes in product and factor markets and the equilibrium amount of resources that can be identified to fund the introduction of technological change. The system of equations can be solved with the standard tangency solutions so as to define both the mixes of creative adoptions which in each specific context firms are advised to select and the amount of technological change with respect to switching the context suggests selecting. The equilibrium condition is: (9) d′(TC) = t/s
Localised technological change: the benchmark 259 The equilibrium conditions identified by equation (9) capture the essence of localised technological change consisting of creative reactions engendered by the mismatch between plans and actual factor and market conditions for firms that are constrained by the irreversibility of their choices. In Figure 13.4 it can be seen that the firm, originally in equilibrium at E1, should move to E2, because of the increase in wages and the reduction of capital rental costs.16 The switching movement from E1 towards E2 is the standard substitution process and involves the introduction of technical change. The irreversibility of existing production factors and imperfect knowledge regarding remote techniques, albeit on the existing isoquant, make the switching mobility of the firm expensive and resource consuming. Technological change, however, provides a viable alternative. The introduction of technological change makes it possible to restore equilibrium conditions at E3 or at any other point that is part of the new isocost. It is actually better at any point beyond the new isocost. If the firm is able to go beyond the new isocost, technological change leads to an increase in overall factor productivity levels not only with respect to the old equilibrium conditions but also with respect to the production conditions at E2. According to the different hypotheses and empirical contexts of application the slopes of both the frontier of possible adjustments and the isorevenue will change so as to accommodate the specific dynamics at play. The relative ease of introduction of technical changes as opposed to technological innovations, the relative weight of irreversibility, the amount of external knowledge available, the efficiency of the internal generation of new knowledge and hence new technologies, and the relative profitability of introduction and adoption of new technologies according to the character of local factor and product markets are all elements that can be integrated into the framework of the benchmark. The amount of technological change being introduced and its degree of localisation will be determined both by the characteristics of each firm and by the structural features of the system. Thus the model of localised technological change provides a framework that makes it possible to consider creativity as an economic force and yet makes it possible to understand the economic dynamics of its dwelling, the context in which it is likely to take place and the mechanisms that control and command its implementation.
4 Conclusion The frame into which a generalised Schumpeterian approach to understanding the determinants and the characteristics of the endogenous process of introduction of technological change is now set. The notions of quasi-irreversibility, both bounded and procedural rationality, and creative reaction are central. Quasi-irreversibility is the result of irreversibility cum creativity: myopic firms, unable to foresee all the possible future outcomes but able to implement procedural rationality and to react creatively to mismatches between internal fixed factors and new unexpected events in both the product and the factor markets, are able to overcome the consequences of irreversibility by means of the introduction of localised technological
260 Introduction of localised technological change changes. Quasi-irreversibility contrasts with irreversibility, which takes place in microeconomic textbooks, in the short term, when firms cannot do any better than accept the negative consequences of the irreversible character of production factors. Quasi-irreversibility emerges when firms, constrained by the lack of malleability of some production factors, are able to react and adapt by means of the introduction of technological and organisational innovations. Such technological and organisational innovations are clearly localised because of: 1) the localised inducement provided by factor irreversibility, either when fixed factors cannot be changed or when some changes are possible by means of switching activities; 2) the source of internal competence and knowledge based upon learning by doing and learning by using the production factors already in place and the current techniques; and 3) the critical role of external knowledge necessary to implement internal knowledge, provided by the local communication networks. Technological change and creative reaction are now included in the range of possible economic actions. Creativity is admitted and yet it is clearly circumscribed and embedded in a context defined by a number of highly idiosyncratic conditions. An economic analysis of creativity can be performed without falling under the control of either chaotic behaviour with no predictability and poor intelligibility or its opposite, that is into an automatic display of routinised and ubiquitous creativity based upon standardised learning. This model provides the benchmark upon which a number of important extensions, inclusions and applications can be grafted so as to implement a generalised model of localised technological change which articulates the analysis of the dynamics of collective knowledge (see Chapter 14), the inducement mechanisms (see Chapter 15), the mix of product and process innovations (see Chapter 16), the combination of adoption and innovation (see Chapter 17) and the role of idiosyncratic factors in assessing the rate and the direction of technological change. In all these cases it will become clear that the conditions of the system in which the mismatch and the eventual creative reaction take place play a key role. The dynamics of localised technological change cannot be analysed in a vacuum, as the innovative conduct of each firm cannot be analysed without a clear grasp of the systemic conditions in which it takes place. Two important systemic aspects deserve to be stressed at this stage. First, it is now clear that the dynamics of localised technological change at the system level is a self-propelling process. The localised technological change introduced by each agent in the system is at the origin of new mismatches, between virtual equilibrium as anticipated in the necessary plans and actual factor and product market conditions, for other agents. The introduction of new technologies affects the virtual equilibrium of both factor and product markets: the exit of firms does not take place as anticipated, but the derived demand for production factors does not change according to the changes in the factor markets. By the same token the virtual equilibrium of the product markets is altered by each new technology introduced both within each industry and at the system level. Within each industry the introduction of a new product as well as a new process (see Chapter 16) modifies the competitive arena and the
Localised technological change: the benchmark 261 distribution of consumers’ choices. At the aggregate level the introduction of new products and processes change the anticipated and virtual equilibrium because of the effects on the aggregate demand stemming from total factor productivity increases and by the changes in the distribution of revenue stemming from the possible bias in the direction of the new technologies (see Chapter 15). Moreover, the localised generation of new knowledge and the eventual introduction of new technologies, induced by the mismatch between plans and actual market conditions, act as a self-propelling factor because of the powerful effects of knowledge complementarily, fungibility and commutability and related externalities. For given conditions of knowledge governance mechanisms at work within each system, the larger the base of technological knowledge available at any point in time is and the larger the number of agents active in the business of knowledge generation is, the greater the supply of external knowledge for third parties is likely to be and hence the greater the likelihood that new firms will prefer to match the mismatches between plans and actual market conditions by means of the introduction of total factor productivity-increasing innovations, rather than sheer passive adjustments, moving in the existing space of techniques. It is clear that no agent, even if endowed with Olympian rationality, can take into account, when making plans, the effects of innovations being introduced by other agents. Second, mistakes and errors made by sub-rational agents also play an important role in this dynamic picture. In standard microeconomics there is no reason to worry about ‘mistakes’: market forces are expected to be able to sort out all agents responsible for wrong choices.17 Their costs would be larger than those of competitors, and losses would quickly drive them out, with no consequence at the system level. In our approach, instead, mistakes, like innovations, cannot be predicted. Thus mistakes also are a cause of emerging mismatches between plans and actual product and factor market conditions. Once more, other interacting agents can endow no rational agent with the capability to incorporate into his/her plans the mistakes made. Yet, because of irreversibility and limited knowledge, unexpected mistakes and innovations introduced by other agents are the cause of mismatches and hence inducements to creative reactions and ultimately to the introduction of new localised technologies. The characteristics of the system in which agents are embedded are essential for grasping the dynamics of localised technological change to assess both the amount of the inducement forces at play and the access to that external knowledge that makes the introduction of new technologies possible. So far the analysis has considered a single firm. When a population of heterogeneous firms is taken into account, localised technological change can be viewed as a process leading to the construction of a new map of isoquants. In a heterogeneous population each firm has a different size, different age and different composition of the stock of irreversible production factors; hence each firm has different switching costs and different learning capabilities and has access to different knowledge capabilities and different local pools of knowledge. The identity of each firm stems from the historical past, because of different paths of growth, different levels of aspiration, different levels of creativity and different
262 Introduction of localised technological change modes of interaction and communication with other firms and public institutions. When such a population of firms is confronted with a generalised mismatch between expectations, plans and actual product and factor market conditions, a variety of different localised technological changes will be introduced. Such a variety of new localised and superior techniques can be interpreted as dots of a new emerging map of isoquants, i.e. of a new technology. In these conditions the traditional representation of the technology and the basic tool of textbook theory of production, i.e. the map of isoquants, can be considered as a punctual and static step of a dynamic and self-reinforcing process of localised technological changes generated by a population of heterogeneous learning and creative firms that have been generating their technologies in a context characterised by local constraints and local endowments. The rest of Part III is dedicated to applying and implementing the basic guidelines presented in this chapter and to showing their analytical fertility.
14 The system dynamics of collective knowledge From gradualism and saltationism to punctuated change
1 Introduction In the economics of localised technological knowledge each firm has a heterogeneous and distinct knowledge base, rooted in its own ‘locus’ defined by learning procedures that are specific to the techniques in place and the set of fixed tangible and intangible assets. However, when knowledge exhibits high levels of supermodular complementarity and networking costs are low, firms have an incentive to implement the convergence of their own knowledge and competence to increase knowledge complementarity. Commons of collective knowledge emerge when the active participation of firms pushes the direction of internal research and learning activities towards higher levels of complementarity, with the aim of building a systemic integration. Collective knowledge is characterised not only by imperfect appropriability and access to intellectual property rights that are either shared or often not specified or specifiable but also by the role of the intentional networking effort, participation and contribution of each agent. Collective knowledge is the result of the valorisation of the elements of latent complementarity among the bits of knowledge possessed by each localised agent (Antonelli, 2001). Knowledge networking activities help to identify and access the sources of external knowledge. On these bases the firm will select and focus the direction of internal learning and research activities in order to integrate them with the characteristics of the external knowledge available. By means of knowledge networking, firms direct their research and learning efforts towards the emerging commons of collective knowledge. The emergence of technological systems is the ultimate result of such an effort of exploration, creation and exploitation of knowledge complementarities (Antonelli, 2001). The analysis of the dynamics of collective knowledge within the framework of the economics of localised technological change can help to provide a synthesis of the divide between Schumpeterian saltationism and Marshallian gradualism. Smooth, Marshallian dynamics can easily generate major Schumpeterian discontinuities. The gap between the theories of punctuated and gradual growth can be reconciled when the essence of the Schumpeterian and Marshallian approach is properly combined. The economics of localised technological change provides an original framework to model the dynamics of the introduction of new technologies.
264 Introduction of localised technological change In this approach, the introduction of technological innovations is the result of the interaction between the inducement to change the technology, generated by the mismatch between plans and related irreversible commitments and expectations, and the interplay between positive and negative externalities provided respectively by technological spillover and networking costs. The correct appreciation of the interactions between individual action and population dynamics makes room for a system dynamics framework able to explain in a single context both Marshallian gradualism and Schumpeterian discontinuities. Within the context of a punctuated approach, Marshallian gradualism and Schumpeterian saltationism can be considered as two extreme possibilities between which a continuum of solutions can be identified. Small variations in the parameters of the positive and negative externalities and in the feedback affecting the extent of the mismatch and hence the levels of the inducement can generate either gradual or discontinuous changes. The rest of the chapter is organised as follows. Section 2 presents a brief exposition of two contrasting views about continuity in economic and technological change. Section 3 stresses the analysis of positive and negative externalities, and articulates the analysis of the dynamics engendered by the feedback between localised technological change and mismatch. Section 4 provides some anecdotal evidence about the relevance of such dynamics in the understanding of long-term growth in the case of Piedmont, a region of Italy. The conclusion summarises the main results of the work and puts them in a broader perspective.
2 Smooth vs discontinuous change The difference between Marshallian gradualism and Schumpeterian abruptness about the rate and direction of economic evolution at large has strong implications for the economics of innovation and new technologies. According to Alfred Marshall (1920: xiii), ‘Natura non facit saltum’: economic evolution is gradual. Its progress is sometimes arrested or reversed by political catastrophes; but its forward movements are never sudden; for even in the Western world and in Japan it is based on habit, partly conscious, partly unconscious. As though an inventor, or an organizer, or a financier of genius may seem to have modified the economic structure of people almost at a stroke; yet that part of his influence, which has not been merely superficial or transitory, is founded on inquiry that has done little more than bring to a head a broad constructive movement which had long been in preparation. Joseph Schumpeter provides instead the key reference for a very different view: We must recognize that evolution is lopsided, discontinuous, disharmonious by nature – that the disharmony is inherent in the very modus operandi of the factors of progress. Surely this is not out of keeping with observation: the history of capitalism is studded with violent bursts and catastrophes which do not accord well with the alternative hypothesis we henceforth discard, and the reader may well find that we have taken unnecessary trouble to come to
The system dynamics of collective knowledge 265 the conclusion that evolution is in disturbance of existing structures and more like a series of explosion than a gentle, though incessant, transformation.1 (Schumpeter, 1939: 102) The Schumpeterian approach has always stressed the sudden emergence of new technological paradigms and major technological breakthroughs that develop a saltationist interpretation of long-term economic growth and technological change. In the Schumpeterian approach discontinuities in long-term growth are viewed as a consequence of discontinuities in the rates of introduction of radical technological innovations. The views about technological change are at the core of the debate about continuity in economic change. Both schools of thought recognise that technological change plays a major role in their respective interpretations. The gradualist approach emphasises the role of minor innovations. Technological change takes place by means of a variety of small, incremental steps, characterised by substantial complementarity and cumulability. In the long term the sequence of incremental innovations builds up a major change. No discontinuity however can be found in the continual process of introduction of new technologies and their selection by means of imitation, adoption and diffusion. The gradualist approach does not necessarily lead to linear change but it is consistent with the traditional neoclassical views about technological change as manna, descending from exogenous scientific progress, or, in a more dynamic approach, as the result of systematic and ubiquitous learning (Arrow, 1962a, 1962b). The gradualist approach is also consistent with the new growth theory. In this approach technological change is the result of the profit-maximising conduct of firms. They are, in fact, assumed to be able to appropriate substantial portions of the economic benefits stemming from their innovations and augmented by the homogeneous access of firms to externalities spilling from the research and development activities conducted by other firms (Romer, 1986, 1990). Eldredge and Gould (1972) have explored in biology the notion of punctuated change. In biological punctuation the critical event is exaptation, a special form of speciation that consists in the application of existing genetic codes to a different ecological context. Separation of the new context is crucial in determining the evolution of the new agents and the eventual, possible emergence of new species. Such separation may occur by means of geographic remoteness, allopatric speciation or exploitation of the periphery of existing niches. In exaptation, the initial event may consist of a minor change. However, the interaction with the new environment and the sequential accumulation of small, additional changes that reinforce each other in a new genetic drift may eventually lead to the emergence of a new species. Occasionally the new species fits better not only in the new environment but also even in the original one. Punctuation is clearly the result of a gradualist process where incremental changes and feedbacks of the local environment interact and reinforce each other (Gould, 2002). Joel Mokyr has successfully applied the perspective elaborated in biology about punctuated discontinuity to the economic analysis of technological change: as in
266 Introduction of localised technological change the process of biological punctuation, the critical event is speciation, the application of existing technological know-how to a new domain. The new domain differs from the original one, favouring additional changes in a new direction. Eventually the new technology is so successful that it can invade other niches, including the original domain. The analysis of Mokyr synthesises and puts in a broader perspective many years of empirical and theoretical research that confirms the discontinuity of technological change and hence economic growth. According to Mokyr much recent thinking among historians of technology and economists of innovation favours the basic notion of punctuated gradualness; radical inventions, stemming from recombination, raise the marginal product of effort in development and lead to a sequence of further improvements. Consequently an intensification of smaller inventions can be observed in periods following radical inventions. Borrowing from the recent results of new advances in palaeontology and population genetics, Mokyr provides an interesting specification and interpretation of discontinuous technological change. His approach stresses the interactions of radical and smaller innovations with a conducive environment: The clustering phenomenon of radical innovations is widely observed in all cultural processes, and represents a combination of conducive environment and interactions between the agents themselves. In biology, the agents do not respond to mutations in other species directly, yet each species takes changes in others as changes in its environment. These interactions, too, may result in critical mass situations leading to occasionally intensive and sudden outbreaks of specification. (Mokyr, 1990a: 352) Levinthal (1998) has applied the punctuated approach to the analysis of the development of wireless technology from a laboratory device to wireless telegraphy, broadcast radio and finally wireless telephony to show convincingly how the sequential applications of given technological know-how to new domains with specific and heterogeneous characteristics generated a cascade of seemingly discontinuous changes that are in fact the result of the same know-how gradually enriched in different directions by the stimulations of different contexts of application. In a similar vein Loch and Huberman (1999) have elaborated a model of punctuated diffusion. This model is based upon three parameters: the rate of technological improvement of technologies, the level of positive externalities generated by their adoption, and the level of switching costs of adopters. Technologies coexist and are characterised by different rates of technological improvements and different rates of generation of network externalities in a population of adopters that are heterogeneous with respect to the levels of switching costs. In this model the resilience of old, inferior technologies, as well as the sudden switch of all adopters to the new technology, may depend on small changes in the parameters of the three interacting variables.
The system dynamics of collective knowledge 267 Building upon Gould’s punctuationism one can argue that Marshallian gradualism and Schumpeterian saltationism are placed as extremes along a continuum of possibilities. It can be argued that Marshall and Schumpeter provide complementary rather than conflicting ingredients, useful for elaborating a broader, system dynamics approach. Marshall emphasises the role of out-of-equilibrium dynamics based upon technical and pecuniary, positive and negative externalities as key factors in a process of increasing division of labour.2 Schumpeter focuses on the endogenous dynamics of intentional introduction of technological and organisational innovations, as determined by the rivalry among firms in product markets. These two strands of analysis can be brought together in an integrated dynamic framework based upon the methodology of systems dynamics. Gradual and discontinuous growth emerge as extreme cases of a broader punctuated process where innovation is induced by the mismatch between plans and expectations, and it is fed by the changing access conditions to knowledge commons and by the effects of population dynamics on the costs and on the productivity of innovation activities. The advent of new information and communication technologies has provided much evidence for both the discontinuity of technological change and the systemic complexity of the relations among technological innovations. Technological systems emerge in leaps and bounds as a result of a major scientific breakthrough and the introduction of a flow of minor technological innovations that are all related to each other by strong elements of complementarity and cumulability in a few and selected regional and economic sites. The introduction and diffusion of new information and communication technologies has not been and still is not a homogeneous, ubiquitous and steady process across time and economic space. On the contrary, new information and communication technologies provide clear empirical evidence about a strong variance in time and space in the rates of generation of new knowledge, in the rates of introduction and diffusion, and in their effects, across regions, industries, product characteristics and firms. The study of the introduction and gradual build-up of the technological system centred upon new information and communication technologies confirms the key role of the complementarity of an array of minor innovations, mainly based upon learning processes and tacit knowledge acquired in repeated learning processes in doing, in using and in interacting. The empirical evidence on the new technological system also confirms the key role of networking activities among learning firms. Complementarity among innovations in fact is mainly the result of a complex process of knowledge interactions and transactions among firms finalised to enhance the systemic scope of application and integration of each single small technological change. Three elements characterise the emergence of Schumpeterian discontinuities in the new punctuated evidence about information and communication technological change: a flow of small and incremental innovations characterised by high levels of complementarity, cumulability and fungibility; systematic networking activities by firms in the attempt to take advantage of the scope for system integration; and a conducive environment providing easy access to large flows of knowledge communication. The flow of small innovations can generate a quantum jump
268 Introduction of localised technological change without a radical invention, but only within a conducive environment, able to support the necessary flow of incremental innovations. The economics of localised technological change provides an appropriate analytical context in which to understand the mechanisms at work in the interaction between individual action and population dynamics.
3 The complex dynamics of collective knowledge: networking and spillovers The complementarity between internal and external knowledge plays a key role in this context. The variety of firms and learning institutions is mostly important in the generation and circulation of knowledge when the latter 1) is viewed as a collective good, with varying degrees of appropriability, 2) is dispersed and fragmented in the economic system, and 3) is the result of both top-down and bottom-up processes, and when learning by doing, learning by using and learning by interacting with suppliers, customers and all rivals play an essential role alongside intramural research and development activities. Knowledge is currently increasingly viewed as a collective process. The notion of a collective process differs both with respect to the Arrovian tradition of knowledge as a public good and the approach to knowledge as a quasi-private good, implemented by new growth theory. Collective knowledge is a shared activity that can be implemented only by interactive agents that belong to a community of action and understanding. Collective knowledge pays attention to the consequences of knowledge indivisibility and the role of the complementarity among the localised bits of knowledge possessed by each agent that characterises both the generation and the dissemination of knowledge in the system and values the contribution of external knowledge in the production of new knowledge. In this approach the role of technological communication among learning agents is stressed as a major systemic character affecting the capability of each agent to implement its internal knowledge (Allen 1983; Von Hippel 1988). The network structure of knowledge communication networks affects deeply the flows of knowledge communication and hence the availability of external knowledge. There is an array of possible network architectures. In geodesic networks (i.e. networks where each agent has a direct link to each other agent), communication costs are very high; the dissemination of new knowledge is hampered by relevant communication costs and by the decay of knowledge spillovers associated with distance and heterogeneity among agents. Within centred networks based upon many interconnected and hence competitive hubs, knowledge is disseminated far better than in fragmented networks where only a few links connect scattered clusters or in networks based upon monopolistic hubs able to exert control over knowledge flows and to extract rents out of it. The appreciation of external conditions for the generation of localised technological knowledge is an important result of this line of inquiry. Localised technological knowledge is the result of the combination of internal competence and knowledge with the external knowledge embodied in capital goods and
The system dynamics of collective knowledge 269 intermediary inputs provided by upstream suppliers or available in the form of technological information, licences and patents and technological spillovers and made available through technological transactions and technological interactions. The relationship between external and internal knowledge is crucial. The role of communication and transmission of knowledge is a major factor in assessing the rate of generation of new knowledge and introduction of new technologies. Systems differ with respect to the speed and capillarity of the flows of knowledge communication. Percolation analysis (borrowed from physics) and communication theory have provided the basic tools to appreciate the distinctive role of receptivity and connectivity in communication processes. The receptivity of agents and their absorptive capabilities are now appreciated as well as the strength and intensity of the message (Arrow, 1969; Cohen and Levinthal, 1990). The structure of economic systems is analysed from the viewpoint of the knowledge communication flows, the density and duration of the communication channels in place and their organisation within the networks of relations. Knowledge communication is not spontaneous, but the result of intentional and dedicated networking activities. In the economics of localised technological knowledge and localised technological change, knowledge networking is an essential component of broader research activities. Knowledge networking is the result both of the exploration and search for the sources of external technological knowledge, either tacit or codified, and of the intentional direction of internal research and learning activities towards complementary external knowledge. Knowledge networking includes: 1) knowledge transactions (the purchase of knowledge in the markets for technological knowledge where prices are incomplete vectors of information); 2) cooperation among firms based upon an array of contractual forms such in-house outsourcing, technological clubs, patent ticketing, joint ventures, sponsored spin-offs, and open technological platforms; and 3) knowledge interactions that are not mediated by prices but rather based upon proximity in geographic, industrial and knowledge space, constructed trust and reciprocity. Knowledge networking is strictly complementary to internal learning and intramural research and development activities. Much empirical evidence however also confirms that the efficiency of resources invested in internal research activities depends upon the amount of external knowledge available, made accessible by means of networking activities and complementary with dedicated efforts. In turn the levels of external knowledge are influenced by the number of firms engaged in complementary research activities and the extent of their research budgets. The notion of supermodularity introduced by Milgrom and Roberts (1995) applies usefully to knowledge by providing a tool that can grasp the dynamics of knowledge complementarities. Knowledge is supermodular when raising the internal research increases the returns to networking and vice versa. The dynamics of localised technological change is grasped when it is recalled that the amount of resources that become available to fund the adjustment to the mismatch is endogenous and dynamic, as is the efficiency of the resources invested
270 Introduction of localised technological change in the research activities. This amounts to saying that both the shape and the position of the frontier of possible adjustments are endogenous and dynamic.3 Building on the previous analysis (see Chapter 13), two elements must now be stressed. First, a cumulative inducement mechanism is at work because the localised technological change introduced by each firm at time t is the main, if not the single, cause of a new and increasing mismatch between its own plans and its actual product and factor market conditions for other firms at time t+1. Specifically we shall assume that the general amount of mismatch increases more than proportionately: the new localised technologies introduced by each firm will disturb the correspondence between plans and facts for more than one firm. Formally this amounts to saying that the number of firms confronted with the mismatch and hence the choice between switching and changing the technology locally, at each point in time, is greater than the number of firms found in the same conditions in the previous period of time: (1) Nt+1 = j(Nt ) with j′ > 0 and j″ = 0. Therefore, as the mismatch increases, a larger budget for adjustment costs has to be allocated. Consequently it is clear that the levels of the adjustment budget at each point in time can be considered a function of the number of innovations introduced in the previous period. The levels of the adjustment budget at each point in time can be considered a function of the number of innovations introduced in the previous period: (2) Rti = a(It–1) where a′ > 0 and a″ > 0 Number of mismatches (t+1)
Number of innovating firms (t) Figure 14.1 The cumulative inducement mechanism
The system dynamics of collective knowledge 271 where Ri measures the budget for adjustments at time t that are made necessary for each firm to manage the mismatches stemming from the amount of innovations (I) introduced in the previous period of time. The number of firms destabilised at each point in time by the innovation introduced in the previous one is also increasing over time. Moreover the shape of the relationship expresses a dynamics where the introduction of each innovation generates a more-than-proportionate increase in market entropy for other firms. The mismatch between plans and actual market conditions is not exclusively determined by the introduction of technological innovations but it is clearly augmented by innovations, introduced at each point in time. Second, firms do not innovate in isolation, but rely upon knowledge pools. Knowledge networking however is not free. It requires dedicated resources. Much empirical evidence confirms that the unit costs of search activities, including networking activities, are influenced by the number of firms engaged. A clearly negative, pecuniary externality is at work here. Network analysis for social communication systems provides basic elements for understanding the dynamics of communication costs. In a network the maximum number of links (ML) between the agents (n) is given by the following simple equation: ML = (n–1) n/2. Hence the increase of the number of agents in the network engenders a more-than-proportionate growth in the number of links. If coordination costs are associated with the number of communication links, it seems clear that communication and coordination costs increase more than proportionately with the size of the network (Faust and Wasserman, 1994). Much anecdotal evidence about the dynamics of coordination and communication costs within industrial and technological districts confirms that congestion problems do emerge rapidly with the increase in the number and in the variety of agents that participate (Bresnahan and Gambardella, 2005; Patrucco, 2008). This analysis leads to the articulation of the following system of equations:4 (3) TCti = e(Ati (Rti)) The amount of technological change that each firm can generate at time t depends upon the general level of efficiency (At) of its research function and upon the amount of resources invested in research activities by each firm i (Rt). The level of research budgets, for given levels of resources available, depends upon the level of the unit costs for research activities r. The budget available for possible adjustments can fund a level of research and networking activities (R&N) necessary to valorise and mobilise the tacit knowledge acquired by localised learning processes and to identify and internalise the external knowledge available. This depends on the levels of unit costs of research and networking activities (r): (4) R = rR&N The unit costs of research, including networking, that are necessary to identify, internalise and integrate the external knowledge depend upon the number of other
272 Introduction of localised technological change firms (N) engaged in research activities and in the total amount of research activities at work: (5) r = f(N R&D) with f′ > 0 and f ″ > 0 The number of firms engaged in research and networking activities and the general levels of their research activities exert also a positive effect in terms of the amount of external knowledge that becomes available to each firm by means of technological interactions, spillovers and technological externalities of different kinds. Hence: (6) Ati = g(N R&D) with g′ > 0, but g″ > 0 The substitution of equations (4), (5) and (6) into the previous equation (3) leads to:5 (7) TCti = e(g(N R&D)ti (f(N R&D)Rti)) From equation (7) it is apparent that the amount of technological change that each firm can generate at time t is influenced by the positive externalities exerted by the external knowledge made available by the other firms and effects of the negative pecuniary externalities on the unit costs of research and networking activities. The relationship between equations (5) and (6) is crucial for assessing the dynamics of the system. The relationship between positive and negative externalities (i.e. between knowledge spillovers and networking costs) becomes crucial. The amount of technological change that a firm induced to react by the mismatch between plans and facts can generate is determined by three elements: 1) the amount of resources available; 2) the unit costs of the activities that are necessary to valorise the internal learning processes, to make explicit the tacit knowledge accumulated, and to access, by means of networking, the external knowledge available; and 3) the amount of spillovers and external knowledge that can be internalised and their complementarity. Specifically two forces are at play here: the negative effects of pecuniary externalities on the costs of generating new localised technological change and the positive effects of knowledge supermodularity. More specifically, the effects of the positive and negative externalities are well grasped by the analysis of the derivative of TC (the amount of technological change that each firm is able to generate) with respect to N (the number of firms engaged in the same knowledge commons): (8) dTC/dN = [g′(N R&D)(dN R&D + N dR&D) + (f′(N R&D)(dN R&D + N dR&D)R)] 1/dN The value of the ratio of the first to the second term of equation (8) conveys all the relevant information. The first term expresses the positive effects of the number of
The system dynamics of collective knowledge 273 firms engaged in the same knowledge commons in terms of access to the same collective knowledge and hence in terms of positive externalities, while the second measures the negative effects in terms of pecuniary externalities. The value of the ratio can be positive, negative or equal to zero: (9) g′(N R&D)(dN R&D + N dR&D)/(f′(N R&D)(dN R&D + N dR&D)R) = 0 Three cases can be identified: 1
2
3
Fragmented knowledge. Networking costs are very high and the effects of knowledge supermodularity are poor. Technological knowledge exhibits low levels of supermodular complexity, as the single bits of knowledge are not complementary. Each of them is the result of idiosyncratic research and learning activities internal to firms and specific to their conditions. When the negative effects of networking costs are always larger than the positive effects of knowledge supermodularity, firms will rely on external knowledge only to a limited extent. Networking activities will be kept at a minimum. Technological change is occasionally introduced by firms in isolation. This situation can be considered the standard case, consistent with textbook microeconomics, where firms are allowed to change only their techniques, not their technology. Positive feedbacks. The second case takes place when the effects of both positive and negative externalities exhibit a negative derivative with respect to the number of firms engaged in the knowledge commons. Their effects increase less than proportionately with the number of firms engaged. The derivative of pecuniary externalities, however, is smaller than the second derivative of technical externalities. In this case the dynamics is fully endogenous, as the feedback of the introduction of innovation on the adjustments budgets keeps exerting its positive effects, which translate into larger and larger amounts of research activities being conducted. At each point in time the flow of innovations is larger than in the previous one; in addition the efficiency of research activities is larger and the unit costs of research skills keep increasing, albeit at a lower rate than that of the increase of efficiency. The number of firms engaged in research and networking activities keeps growing as the growing number of innovations being introduced destabilises an ever-increasing number of firms. The position and the shape of the frontier of possible adjustment will change to reflect both the greater size of both intercepts and the greater relative efficiency of research and networking activities with respect to switching activities. With a given slope of the isorevenue it is clear that, the smaller the marginal rate of transformation of switching activities into research activities is, the larger is the equilibrium value of technological changes being introduced at each point in time. Much theorising about new growth theory can be accommodated within this special case. Localised supermodularity. The third case is the most interesting. It is the case of endogenous but transient net positive externalities: external increasing
274 Introduction of localised technological change returns stemming from knowledge supermodularity do take place but within a localised and limited context. The fabric of communication channels in place and the endowment of social capital (including the quality of the local scientific infrastructure) and the levels of knowledge supermodularity are such that the potential complementarity among the knowledge base and the competence and the stock of knowledge of diverse firms can be identified, implemented by means of effective networking and well-focused research activities, and fully exploited. Commons of collective knowledge take off. This dynamics however takes place only within a circumscribed region of technological knowledge and for a limited number of firms when the second derivative of positive externalities is positive and the second derivative of negative externalities is non-negative. As long as the positive effects of knowledge externalities, in terms of spillovers and access to external knowledge, are larger than the negative effects of pecuniary externalities in terms of increased networking costs, the number of firms engaged in building complementary knowledge will increase. The number of innovations being introduced at each point in time will be larger with respect to the previous period. Moreover in the following period the amount of the mismatch will be larger, and hence the budget made available to face the new necessary adjustments must be larger. As long as the positive effects of knowledge externalities are larger than the negative effects of pecuniary externalities, the larger budgets for adjustments will translate into larger flows of innovations being introduced into the system. The size of the knowledge commons will increase as long as net positive externalities exist. The number of networking firms will keep increasing, as will the flow of complementary innovations being introduced. Specifically, the negative effects of pecuniary externalities in terms of the increase in the unit costs for research and networking activities, on the one hand, and the positive effects of collective knowledge in terms of the increase in the output of research activities, on the other, can be isolated and directly confronted (see
SW
T TC Figure 14.2 The changing slope and position of the frontier of possible adjustments
The system dynamics of collective knowledge 275
Positive knowledge externalities (TE)
(NC)
(TE) Negative pecuniary externalities (NC)
It4
It3 It2 It1
Number of firms engaged in knowledge networking Figure 14.3 The dynamics of knowledge and pecuniary externalities for research activities
Figure 14.3). The increase of the unit costs of research activities is expressed by the positive and constant slope of research and networking costs (NC) as determined by the number of firms engaged in knowledge networking activities. The number of firms engaged in complementary research activities, within the same knowledge commons, also exerts a positive, albeit decreasing, effect on the productivity of research activities conducted by each firm, as is expressed by the log linear slope of TE. The difference between the slopes of NC and TE is crucial: (10) TE = m(N), with m′(N) > 0, m″(N) < 0 (11) NC = n(N), with n′(N) > 0, n″(N) = 0 (12) p(N) = m(N) – n(N) s.t. m′ > n′ It follows that p′(N) > 0, and p″(N) < 0. Let us recall that, as long as net positive externalities can be found, the number of new firms that engage in networking activities, within the same knowledge commons, increases. Moreover, the need to fund a budget for adjustment activities, and hence to conduct additional research activities, and the size of the adjustment budget, including research and networking activities, are a function of the flows of innovations introduced. Hence:
276 Introduction of localised technological change (13) dN(t)/dt = W(p(N)) and, given the properties of W and p(N), it follows that: (14) N(t) = §t (dN(t)/dt)dt = §t W(p(N))dt Equation (14) establishes a functional relationship between the flow of networking firms engaged in research and induced by the mismatch between plans and actual market conditions and the stock of collective innovators. The p(N) function is S-shaped and has a flexus. Therefore a functional form that is compatible with the specific conditions is: (15) N(t) = α1/1–e–kt where k measures the speed of the process.6 It is now clear that the dynamics of the systems is determined by the feedback of innovations on the levels of mismatch and the interplay between positive and negative externalities. When the effects of positive externalities are slightly greater than the negative ones, the dynamics of the system is smooth. However, when the net positive externalities are large, the region of the inflection point of the S-shaped process has a strong projection of the vertical axis. The speed parameter k in equation (15) plays a key role. Its value can be seen as the effects of two dynamic processes. First, the size of this relationship is augmented by the feedback between innovations and the resources for adjustments. The parameter a of equation (2) here plays a key role. When the feedbacks of innovations on the budget for adjustment activities in the following period are large, the slope of the S-shaped process is large, as is the saturation limit. Conversely, when the flows of innovations introduced in the previous period of time have only a small positive effect on the budget for adjustment activities funded in the following period of time, the S-shaped process remains defined within a small region, with low levels of innovation activities and, hence, small flows of innovations being introduced. Second, net positive externalities have a strong effect in terms of rates of entry of new firms and cross-entry of incumbents from other localisations into the knowledge commons. Intentional profit-seeking mobility of firms in geographic, technological and product markets is an important characteristic of this interpretative framework (Marchionatti, 1999).7 The parameter j of equation (1) here plays a key role. The mobility in knowledge space of firms, albeit limited by the localised roots of their competence, has the twin effect of speeding up the rates of generation of technological change and affecting the shape and the architecture of knowledge networks. The rates of technological change, as long as fertile knowledge commons are generated by the interplay between positive and negative knowledge externalities, will be determined by the catalytic reaction stemming from the interaction between innovation, mismatch and collective creativity. When the positive effects of knowledge externalities match the negative effects of pecuniary networking externalities, the unit costs of innovations will no longer
The system dynamics of collective knowledge 277 decline. Firms will rely less and less on networking and hence on external knowledge to generate new localised knowledge. Internal research and learning activities of single firms will play a larger role. Technological change is now characterised by a flow of smaller, incremental and discrete technologies. At this time, the number of innovations being selected by firms facing adjustment problems (and eventually introduced) will be smaller than the ones introduced in the previous period. The system will not, however, enter a stationary state because of the exponential relationship between innovations and mismatch and hence resources for possible adjustments. When attention is focused upon the rate of innovation it is clear that the dynamics is shaped now by the relationship between the dynamics of the unit costs of innovations and the dynamics of the budgets available to fund innovations. The introduction of radical inventions here is not necessary for the dynamics of the process to be sustained if the amount of the resources generated by the mismatch keeps growing and consequently the shape of the frontier keeps moving towards the right, albeit with a changing and less favourable slope. The analysis of the dynamics of positive and negative externalities, framed in the context of the economics of localised technological change, confirms that Marshallian gradualism and Schumpeterian saltationism can be reconciled when a sequence of small, incremental, but cumulative, innovations takes place in a conducive context where a catalytic and collective process of creative reactions, augmented by a conducive population dynamics, becomes the engine of endogenous changes in the shape and the position of the frontier of possible adjustments and leads to punctuated leaps that shape the path of technological change and economic growth, in terms of both rates and directions.
4 The dynamics of collective knowledge and localised technological change: anecdotal evidence in Piedmont The long-term growth of Piedmont, in north-western Italy, provides much evidence about the key role of collective knowledge and localised technological change in the punctuated technological and industrial development of a region (Castronovo, 1971; Bairati, 1983). Industrial and technological change in Piedmont in the nineteenth and twentieth centuries is characterised by a sequence of discontinuous phases of rapid growth centred upon different yet interrelated technologies (Castronovo, 1995; Barca, 1997; Amatori and Colli, 1999). The displacement of the capital city of the new Italian kingdom in 1865 away from Turin to Florence and eventually to Rome impoverished the region, but left a remarkable endowment in terms of technological and scientific institutions able to provide advanced technological services useful not only for the implementation of the aggressive military sector but also for the business community. The introduction of hydroelectric power was largely due to local scientific capabilities based upon the group guided by Galileo Ferraris at the Politecnico, the school of engineering of the University of Turin, which eventually acquired full institutional and academic autonomy. The Politecnico itself was the eventual outcome of the merging of the
278 Introduction of localised technological change Royal Arsenal and the Royal School of Artillery, where for centuries the command of engineering had been implemented for military purposes. The introduction of hydroelectric energy can be considered the result of a localised search for innovation strongly determined by the attempt to react to the crisis engendered by the displacement of the capital city (a typical failure-induced reaction) by mobilising the local intangible endowments in terms of collective knowledge and to make good use of the local tangible endowment of an alpine region. It succeeded and provided many investment opportunities to a mountain region with several valleys and rivers: a large number of dams were built, and many companies entered into industrial production of energy and upstream into production of electrical machinery. Energy prices fell sharply. The textile industry benefited from the large supply of cheap energy and in turn generated a large demand for textile machinery. The engineering industry grew exponentially at the end of the nineteenth century and the beginning of the twentieth century. The scientific traditions of the Politecnico and the rapid growth of the technological capabilities of the local engineering industry provided major opportunities to contrast with the eventual decline of the textile industry and to take advantage of the technological opportunities in the car industry. The birth of the automobile industry in Italy, centred in Turin, provides one of the clearest historical examples of a highly specialised technological district characterised by a typical S-shaped growth. In the period 1896–1928, 167 car companies were started in Italy, almost 40 per cent of which (i.e. 66 car companies) were in Piedmont. After Piedmont, Lombardy, almost three times as large in terms of size of population and economic activity, was the second region, with 63 new car companies. In the rest of the country, only 38 companies were created. Entry into the car industry in Piedmont followed a typical S-shaped process in these 30 years: after a few years of slow growth with two or three companies per year, the typical flexus took place in the years 1904–06, when 21 companies were created: four in 1904, ten in 1905 and seven in 1906. After 1906 the process slowed down in Piedmont and spread slowly to the rest of the country. Eventually a district champion emerged and internalised systematically the complementary activities, with a clear substitution of internal hierarchy by external coordination. Much empirical evidence provided by economic historians confirms the key role of collective knowledge in such a process: mobility of skilled personnel especially played a key role, both between the Politecnico and the business community and among firms. A web of cooperative forms is documented, especially in user– producer relationships, with the creation of open technological platforms to which firms could contribute their distinctive competence to larger projects. Increasing levels of specialisation and the abundant supply of local advanced suppliers of machinery, specifically machine tools, are recorded as a major conducive factor (Antonelli, 2001). The increasing problems experienced in the dynamic coordination of such a fast-growing industry with large numbers of competing and yet cooperating firms are documented as a cause of the eventual consolidation of the industry and the decline in the rates in technological innovation. The sequence of bursts of phases of technological and industrial change experienced in Piedmont for almost a century can be easily interpreted in terms of a punctuated sequence of
The system dynamics of collective knowledge 279 applications of a given technological know-how to a sequential string of new yet related domains. At each point in time when the technology generated by means of the prior application to a new domain entered a period of decline, with evident downturns in the rates of profitability and performance, new failure-induced attempts of localised applications of the technological competence so far acquired in new, but adjacent, technological domains were made. A flow of localised technological changes were introduced and paralleled by the gradual emergence of new knowledge commons fed by the interactions among a variety of innovating firms in a conducive institutional context based upon the strong scientific traditions of the local environment. As a result, a new wave of industrial and technological growth took place, with sustained S-shaped rates of technological change and industrial growth, fed by the entry of new firms and the introduction of incremental innovations by incumbents, in the window of time in which the positive effects of knowledge supermodularity were not offset by the increase in unit costs of external coordination within the knowledge commons. In the Piedmontese case we see the first step from military engineering to electrical engineering, then followed by applications in textile production and eventually in textile engineering, then from textile engineering to machine tools, and finally from machine tools to the car industry. The last step from mechanical engineering into electronics and telecommunications failed at the end of the twentieth century, with the collapse of a number of key firms. The severe economic difficulties of the region in the passage from the twentieth to the twenty-first century can be identified in the weakness of the link between mechanics, electronics and new information and communication technologies, still missing or not yet fully implemented.
5 Conclusion The economics of localised knowledge and localised technological change provides an analytical framework that can easily accommodate the dynamics of punctuated growth and continuous change. The key element is provided by the endogeneity of technological change. The appreciation of the role of positive and negative externalities and of the relationship between innovation and mismatch makes clear that the position and the slope of the frontier of possible adjustments are both endogenous and dynamic. The dynamics of the system is generated by the interplay between the effects of knowledge complementarity and networking costs. In turn networking costs are very much determined by the endowment of communication channels, both tangible and intangible, of each economic system. Economic systems with a poor knowledge communication infrastructure, low levels of trust and social capital, and high levels of opportunistic behaviour have high levels of knowledge networking costs. For given levels of knowledge complementarity, networking costs are very high. Firms rely mainly on internal knowledge and have no access to pools of external knowledge. For given levels of mismatch and hence resources for either technical or technological change, firms will introduce lower levels of innovations. The dynamics of the process here is fully
280 Introduction of localised technological change determined by the feedback between innovations, mismatch and the amount of resources available for possible adjustments. In economic systems with a rich knowledge communication infrastructure and a large endowment of social capital, firms have low levels of networking costs and hence can access external knowledge more easily. The net positive effects of knowledge spillovers provide the system with the opportunity to introduce a sustained flow of innovations. Sustained rates of technological change take place as long as the increase of networking costs does not match the gross effects of technological externalities. The accelerated generation of a myriad of small technological innovations is the consequence and the cause of the creation of new technological systems. Systemic integration of new technologies is the result of intentional networking. Knowledge networking makes possible the valorisation of complementarities among technological innovations. In turn systemic integration enhances the productivity of resources invested in research activities, and this leads to further increase in the amount of resources available to fund innovation. The policy implications of this analysis are clear. A knowledge communication policy can sustain accelerated rates of introduction of waves of complementary technological changes. Specifically we see that the lower the costs of searching, accessing and internalising reliable sources of external knowledge are, the greater the chances are that a creative reaction as opposed to a passive adjustment will take place to face the mismatch between plans and actual market conditions. A continuum of growth regimes can now be identified, according to the results of the interplay between positive and negative externalities. Punctuated growth is clearly a special case provided by the values of the second derivative of positive and negative externalities. Punctuated and gradual growth are simply two possible outcomes of a broader dynamic process governed by the interaction between irreversibilities, creativity and the characteristics of the systems in terms of connectivity and receptivity. The tools of evolutionary biology can be successfully applied to economics, especially if blended with Lamarckian elements, so that the effects of irreversibility and creativity are taken into account in a context where agents are able to learn and to innovate intentionally, technological change is endogenous and the analysis is dynamic. The fruitful grafting requires that two types of feedback are accounted for: first, the interplay between innovation and entropy in the markets for products and for production factors and hence between innovations and the extent of the mismatch those myopic firms need to face; and, second, the interplay between the context of action and decision making for each agent and the characteristics of the system, in terms of knowledge communication flows. Punctuated growth coincides with the emergence of collective knowledge, that is, the participation of a variety of learning agents in the generation of technological systems. High levels of endogenous knowledge complexity and fungibility characterise technological systems such that the vertical and horizontal complementarity of knowledge are brought together. Latent knowledge complementarities, however, can be brought together and fully exploited only when the negative externalities
The system dynamics of collective knowledge 281 engendered by networking costs are lower than the positive effects of knowledge externalities. On these bases, Marshallian gradualism and Schumpeterian saltationism can prove to be not only compatible but also, even more, complementary tools of a broader systems dynamics approach to modelling economic and technological change. With respect to complex dynamics systems, this chapter has shown how relevant is the role, in understanding the dynamics of the process, of the intentional and strategic action of firms that are induced to change their technology and rely upon external knowledge, in terms of structural change. In so doing, firms search and move in the knowledge space and enter into specific knowledge commons. The topology of the knowledge space is changed by the strategic action of myopic firms, as much as the topology of the technological and industrial space is changed by the introduction of technological innovations. Firms are able to change the structure of the economic space and the architecture of knowledge networks.
15 Factor markets Constraints and inducements to innovation
1 Introduction The induced innovation point of view is a fertile line of approach to understanding what determines the rate and the direction of technological change. Recent advances in the economics of innovation have been grafted on to the induced technological change approach and provide new insights and basic guidance in understanding what determines the rate and direction of technological change. Many important dynamic implications are to be drawn from such an analysis of the interaction between composition effects and technological change and the notions of general and contingent technological change. The divide between the microeconomic model, stemming from the observations of Sir John Hicks (1932), and macroeconomic theory, developed by Kennedy (1964) and Samuelson (1965), can be reconciled within the context of the economics of localised technological change. This chapter provides a systematic analysis of the role played by irreversibility, limited knowledge and localised learning in the inducement mechanism, which together lead firms to introduce new technologies, in a context where the interaction between the rate and the direction of technological change and the specific context of their application are considered. Section 2 reviews the basic results of the induced technological change approaches. Section 3 summarises the notion of composition effects. Section 4 presents the distinction between general and contingent technological change. Section 5, which is the core of the chapter, presents a formal model of localised inducement that combines both inducement caused by changes in factor prices and the inducement stimulated by the levels of factor prices resulting in the direction of technological change. Section 6 explores the implications of the model and applies it to understanding the dynamics of technological change in a comparative and global setting. The conclusion puts the results of the model in a broader perspective and lists the main findings.
2 The induced technological change approach According to the induced technological change approach, new technologies are introduced in response to conditions in the factor markets. First, a distinction has to be made between the models of induced technological change, which focus
Factor markets 283 attention on changes in factor prices, and the models of induced technological change that stress the static conditions of factor markets. In the former approach, following Hicks and Marx, firms are induced to change their technology when the price of a factor of production increases (Hicks, 1932).1 According to the generalisation of the basic hypothesis put forward by Binswanger and Ruttan (1978) and recently updated by Ruttan (1997, 2001), firms introduce new technologies which reduce the use of the factor whose costs have increased. A change in factor prices acts as an inducement, and explains both the rate and the direction of the introduction of new technologies. The introduction of new technologies complements and increases the standard substitution process, i.e. the technical change involving the selection of new techniques, defined in terms of factor intensities, in the existing isoquants. Inducement concerns both direction and intensity. The stronger the increase of wages (or any other factor) is, the greater the effects will be, both in terms of labour-saving intensity and in terms of the amount of innovation being introduced. This approach to induced technological change differs from the macroeconomic version, developed by Kennedy (1964) and Samuelson (1965), in that firms introduce new technologies in order to reduce the use of the factors of production that are relatively more expensive. In this second approach the level of factor price is important, not the rates of change. As Ruttan (2001: 103) points out, however, the macroeconomic approach is not able to explain why and how firms introduce new technologies. For these reasons and the fact that it does not have a microeconomic base, the macroeconomic model has been criticised to the point of being neglected. The macroeconomic model, however, points out one important thing, i.e. that there is a clear incentive to introduce a technology which makes the most intensive use of the resources that are locally most abundant. The recent and fruitful debate on skill-biased technological change is centred on an analysis of the causal relationship between the abundant supply of skilled labour and the eventual generation and introduction of skill-intensive new technologies (Acemoglu, 2002). There is a contradiction here between the macroeconomic model and the microeconomic one. In the microeconomic model, firms are induced to introduce new labour-saving technologies by an increase of wages. In a labour-abundant region, however, there is a strong incentive to introduce labour-intensive technologies, rather than labour-saving ones, even after an increase in wages. The induced approach and its contradictions can be improved by considering the role of composition effects in the global economy and by taking into account what determines and what limits the localised introduction of innovation when there are effects of irreversibility and limited knowledge and learning.
3 Composition effects: the interaction between relative prices and the direction of technological change Analysing technological change in a global economy, characterised by heterogeneity in factor markets, different production functions in use in different countries and
284 Introduction of localised technological change rivalry in interdependent international product markets, facilitates the inclusion of the combined effects of relative factor costs and non-neutral technological changes. When a general-purpose technology which can be introduced and applied widely becomes available, and the global economy is heterogeneous with respect to the characteristics of the production functions in use and of the factor markets, the technology cannot be neutral everywhere (Bresnahan and Traitenberg, 1995; Helpman, 1998). When technological change is biased, the context in which it is introduced is fundamental in assessing its effects in terms of total factor productivity growth. When a new technology is biased, in that it favours the more intensive use of one of the factors of production, its effects measured in terms of productivity growth are stronger the more abundant, and hence less expensive, the most effective factor is. Such dynamics have major effects, in terms of emerging asymmetries in the global economy, on firms operating in heterogeneous factor markets while competing in homogeneous and global product markets. Composition effects define what effect relative factor prices will have on total costs and total factor productivity levels. Differences in factor costs are relevant both when technology is given and when technological change is important. In the former case, it is seen that, when, with a given biased technology, relative factor prices, as distinct from the levels of absolute factor costs, change, average costs also change. Specifically, whenever the relative cost of the most productive factor falls there will be a reduction in production costs. Such changes in production costs, even if they cannot be accounted for by overall factor productivity measures, have a powerful effect on the competitive advantage of rival firms based in heterogeneous factor markets. In the latter case, it is seen that when a new superior technology is introduced in two countries – characterised by different endowments and hence different factor markets with different factor costs – it will have a much bigger effect in terms of raising overall factor productivity in the country where the most productive factor is cheaper. Composition effects are all the stronger when there are changes in both technology and factor costs. Composition effects play an important role in the analysis of technological change in different industries and countries because of the strong effects relative prices have on the actual ‘measured’ overall growth of factor productivity in each country. The static and dynamic interaction between different kinds of change in technology and the levels and changes in relative prices can in fact take different forms. Let us consider some general cases. Let us first consider an industry with a capital-intensive production function in a region where a neutral technological change has been introduced and the general efficiency of the production process has increased. Simultaneously, however, capital rental costs have also increased and wages declined. These two changes tend to have opposite effects. The increase in the overall efficiency should lead to an increase in output for given levels of inputs. The increase in relative capital rental costs, however, leads to a reduction in the use of capital and hence in output. This reduction can perfectly offset the increase in the general efficiency. Similar asymmetric relations take place when relative wages increase and a new neutral – capital-intensive – technology is introduced. The general efficiency of the
Factor markets 285 production function is now raised by the increase in the wage-to-rent ratio and hence by reducing the use of less productive labour and increasing the use of more productive capital. Again, the more capital intensive the production function is, the stronger the effects of the increase in wage levels are. The model becomes even more complicated when non-neutral technological change is introduced. Assume that a smooth incremental technological change with labour-intensive and hence capital-saving features is introduced in an industry based in a region where wages are low and capital rental costs are very high. The composition effects in terms of the increase in overall efficiency are very important. The effect can be much stronger than a radical and biased technological change which is characterised by a major shift in the general efficiency parameter and also by a significant increase in the output elasticity of capital and a reduction in the output elasticity of labour. The latter technology will be less efficient than the former, although overall it should be regarded as better. It is clear therefore that the performance of technology depends very much on its bias and the relative costs of the factors of production. Such an analysis immediately applies in a synchronic context where there is a variety of factor markets across industries and regions. A new and radical capitalsaving technology will have stronger positive effects in a labour-abundant region with low wages than in a high-wage country, and this explains why such technology will be adopted more quickly in such regions. The incremental labour-saving technology will have stronger positive effects and will be adopted more quickly in capital-abundant regions with low relative capital rental costs. Only technological change which is characterised by a bias that is appropriate to the structure of local endowments can strengthen technological variety in international markets where the relative prices of inputs differ because of differences in local factor markets. The global introduction of a new general-purpose technology instead reduces technological variety, with negative effects on the structure of comparative advantage and hence on the distribution of the gains from global trade. The introduction of a global and hence necessarily biased technological change has strong effects in terms of new asymmetries among potential adopters. When a new technology is biased, the increase in efficiency takes place in only a limited area of the map of techniques. In such conditions the new and the old technologies are likely to intersect. Before the intersection, the new technology is superior to the old technology in absolute terms and vice versa after the intersection.
4 General and contingent technological change An analysis of composition effects provides a clear framework within which the distinction between radical and incremental technological change can be analysed. Macro- and micro-inventions, frequently developed in the literature on a qualitative basis, can also be considered (Freeman, 1994; Mokyr, 1990a, 1990b).2 When a new technology is introduced in a heterogeneous economic system with a variety of local factor markets, the effects in terms of the growth of overall factor
286 Introduction of localised technological change productivity are influenced by the local structure of relative factor prices. The ranking of different technology depends on the relative prices of the factors of production. The distinction between general and contingent technological change can now be introduced. A new technology is general when it can be defined as a new production function where the shift parameter increases to such an extent that it is superior to any previous technology, even when the output elasticity of each production factor is affected. Contingent technological change instead affects only the composition and the ranking of production factors in terms of their output elasticities. The effects on total factor productivity are generated by the substitution of more productive inputs for less productive ones, with no (general) shift in the production function.3 The notion of contingent technological change differs from previous specifications of technological change, as it considers a specific and relevant aspect of localised technological change. Technological change is neutral when it consists of a shift effect which leads to a traditional increase in overall factor productivity levels with no effects regarding the composition of the marginal productivity of the factors of production. Contingent technological change, instead, affects only the composition and the ranking of the factors of production in terms of their output elasticities. The effects on overall factor productivity are generated by the substitution of more productive inputs for less productive ones, with no shift in the production function. The accepted tradition of productivity accounting, based upon the path-breaking contributions of Abramovitz (1956) and Solow (1957), makes it possible to calculate a synthetic index of the changes in overall factor productivity levels. Following Salter (1960) and Brown (1966), simple calculations make it possible to split up the overall factor productivity level into two well-defined components: the effects of introducing general technology and hence the shift effect, and the composition effects brought about by the introduction of new biased technologies which change the relative output elasticity of inputs. The procedure is very simple and consists in first calculating the standard residual, based, as is well known, on the calculation of a virtual output at time t1. This is based on the new observed levels of inputs and the old output elasticities. Then it is compared with the actual values, and the difference is attributed to the introduction of new technology. The complementary methodology, aimed at splitting the bias and the shift effects, consists in calculating a new virtual output. The new virtual output is simply the product of the production function at time t1, with the new input levels and the new factor shares. The difference between the second virtual output and the actual one measures the shift effect. In turn the difference between the first virtual output and the second measures the composition effect. Let us consider this with two simple production functions at times t1 and t2 respectively. A new technology is introduced with both shift and bias effects during the given time interval. The output elasticity of capital at time t1 is α = 0.25, and it is a= 0.75 at time t2.
Factor markets 287 (1) Yt1 = K– L– for α = 0.25 (2) Yt2 = Ka Lb for a = 0.75 The standard total factor productivity index (AT) is calculated as follows: (3) AT = Yt2/Kt2α Lt2β The shift component of the total factor productivity index can now be calculated as the ratio between actual output and estimated output, the levels of inputs at time t2 and the new output elasticities. Formally the calculation is as follows: (4) AS = Yt2/(Kt2a Lt2b) The difference between AT and AS can be termed AB: it provides a measure of the joint effects of the changes in the relative prices (if any), and the bias in the output elasticities, and measures in a synthetic way the effects of the changes in the composition and relative efficiency of the factors of production: (5) AB = AT – AS It is important to note that AB may be negative as well as positive. A negative AB occurs when a new general technology with a strong shift effect is introduced in a country although the factor intensities are not really appropriate for the local conditions in the factor markets. When AB is negative, an important opportunity for the eventual introduction of dedicated contingent technologies emerges. The generation of new biased technologies that build around the new shift technology and use the locally abundant inputs more intensively, and hence reduce the use of some locally scarce and costly inputs, may be very advantageous. The introduction of general technologies exerts powerful asymmetric effects in a global economy. Such technologies are characterised by such an important shift effect that they are (almost) always and everywhere more efficient than previous technologies. Nevertheless, they will be more productive in some systems than in others, depending on the relative costs of the most productive factors. The introduction of general technology with high levels of capital intensity in a capitalabundant country yields a larger increase in total factory productivity levels than in a labour-abundant country. It may still be adopted even in a labour-abundant country, but it will register lower levels of overall factor productivity. The bias in technology engenders a strong and long-lasting asymmetric effect. The asymmetry will be stronger, the stronger the bias and the shift, and hence the relative profitability of adoption, are, even in less favourable conditions. The direction of technological change and the context in which it is applied are more important than is generally thought, especially in a global economy, where agents based in heterogeneous factor markets compete in quasi-homogeneous and, in any event, interdependent product markets.
288 Introduction of localised technological change The generation of either contingent or general technological change clearly is not an exogenous event which takes place without any economic inducements or incentives. Instead, the introduction of either contingent or general technological change can be considered as the induced outcome of very specific incentives and constraints exerted and shaped by the structure of the economic system. The necessary tools are provided by traditional analysis being incorporated into the economics of localised technological change.
5 The inducement of general and contingent localised technological change With respect to the benchmark model presented in Chapter 13, we assume here that the identification of two quite distinct classes of technological change with respect to their effects is possible, and this is to develop the analysis concentrating on the logic behind decision making about the generation of new technologies. Two quite distinct rationales can be developed by drawing on traditional analysis of the economics of innovation, to understand the generation of contingent and general technological changes respectively. From the point of view of the incentives it seems clear that the introduction of general technology is likely to yield a much greater benefit than the introduction of contingent technologies. Important differences in terms of costs and constraints to their introduction need to be considered. Four groups of conditions are important here. The first draws on the distinction between top-down scientific opportunities and bottom-up technological opportunities. Technological opportunities are mainly based upon learning processes, while scientific opportunities draw on new scientific advances. The second condition concerns where the sources of new knowledge are and whether they are part of the economic system in which the firm is embedded or mainly external, in other regions or even other countries. In this context, the regime of intellectual property rights and the levels of international protection, as distinct from those of domestic protection, play an important role, in that they determine the conditions of access to external technological knowledge. The third relevant group is the distinction between leaning processes and whether it is a question more of learning by doing or learning by using capital and intermediary goods purchased from other industries often located abroad. The levels of switching costs are the fourth relevant group of variables which affect a firm’s innovative behaviour. They include the costs associated with all changes in the existing stocks of tangible and intangible capital and techniques, including the expertise of the workers as well as the brand and reputation of the firm. For a given set of incentives, technological change will be either general or contingent, depending on the specific values of the parameters for these factors. When top-down scientific opportunities emerge and the frontier of scientific knowledge is brought forward by relevant scientific advances, when internal knowledge is more relevant than external knowledge, when learning by doing is more relevant than learning by using and when both irreversibility and switching
Factor markets 289 costs are low, firms are more likely to introduce general technological changes. Instead, when technological opportunities are more important than scientific ones, when the major sources of technological knowledge are abroad, when learning by using is more fertile than learning by doing, and when the irreversibility of both tangible and intangible factors of production is high, firms are more likely, for given innovation budgets, to introduce contingent technological changes rather than general ones. General technological changes involve a radical shift of the map of isoquants, such that all techniques are now more efficient. They can be thought of as the typical result of scientific breakthroughs and research activities in technological domains where agents are able to improve the productivity of a large array of techniques. A major and radical breakthrough leads to new general-purpose technologies. Significant shift effects and hence high levels of increases in overall factor productivity characterise general technological change. The shift effects are such that the new technology is superior to most, if not all, technologies in use in terms of levels of overall factor productivity, whatever their bias and whatever the local factor costs. General-purpose technologies however are likely to reflect the specific and idiosyncratic factor endowment of the innovators: they are only locally neutral. Hence the factors that are most abundant in the innovating country are likely also to be most productive. The introduction of general-purpose technology can be thought of as the outcome of the localised efforts of innovators aware of new scientific opportunities, and so a general shift on the map of isoquants is induced. When a new technology is locally neutral, its adoption elsewhere engenders a significant diffusion and hence the growth of overall factor productivity across firms based in countries and regions that are characterised by heterogeneous endowments. Such positive effects, however, are asymmetric, in that they are stronger where most productive factors are cheaper. Contingent technological change is the result of the incremental introduction of a myriad of small changes after the main shift effect has been generated. Contingent technology is introduced by firms facing unexpected changes in both product and factor markets when the constraints of quasi-irreversibility of fixed capital stocks are low and hence less important than the switching costs associated with all changes in factor intensities. Markets for inputs are more flexible, capital intensity is lower and therefore the role of inertia engendered by sunk costs is also lower. Firms can change their factor mix with relative ease. Further and most important, contingent technology can be considered to be the result of incremental innovations mainly based on learning by using procedures. Firms learn how to use new general technology, especially when it is embodied in capital goods and intermediary inputs, and eventually they are able to capitalise on the new tacit knowledge. The access to external knowledge through user–producer interaction with advanced but remote sellers, sellers of new capital goods and intermediary inputs, can help adopting firms to invent and improve the factor intensity of the new general technology, so as to make it appropriate to the local structure of endowments. The generation of contingent technology can be considered to be the result of a viable innovation strategy for firms which have limited resources to fund research
290 Introduction of localised technological change budgets. Such firms rely more upon external and tacit knowledge, associated with processes of learning by using new inputs; they operate in flexible factor markets and are able to improve and eventually adopt new technologies, mainly invented elsewhere. Specifically, a sequence between general and contingent technological change will now be described. The sequence begins with the introduction of new generalpurpose, but locally neutral, technology in a leading country with idiosyncratic factor markets, and diffusion occurs very quickly across regions and industries because of the big increase in overall factor productivity levels which result from the adoption of the new technology. However, as the new general-purpose technology is adopted in countries and regions where relative factor prices differ sharply from prices in the country where the technology originated, new adopters and other followers will use the new technology and increase its benefits by introducing contingent technological changes that fit in better with the local endowment of production factors. The benefits stemming from the introduction of contingent technological changes are clearly much lower than the benefits derived from the introduction of general technological changes; their costs are also much lower. The analysis in this chapter makes it possible to consider the range of localised choice. At the firm level, the range of technological innovations can vary between the two extremes of a new general and hence locally neutral technology which only consists in a shift effect and a new contingent technology which only consists in a bias. The choice between introducing general and contingent technology, for a firm constrained by factor irreversibility and bounded rationality but induced to innovate by the new and unexpected conditions in its product and factor markets, can be neatly encapsulated in the analytical framework of a nested frontier of possible adjustments that combines the choice between substitution and innovation and also between introducing general and introducing contingent innovations. The introduction of new technology is the result of research and learning activities. The resources available to cope with unexpected changes in the product and factor markets can be used to generate either general or contingent technology. The investment of the resources available leads in turn to research, learning and communication activities which translate into varying levels of generation of either general or contingent technology, depending on how easy it is to introduce either kind of new technology. With respect to the benchmark model presented in Chapter 13, here we assume that the firm faces two nested frontiers of possible changes when there is a mismatch between expected and actual market conditions. The first frontier of possible changes is the frontier of possible adjustments which make it possible to compare the results of resources invested in either technical or technological change. The second frontier compares the different kinds of technological change, whether it is contingent or general. The absolute level of the revenue generated by all the adjustment activities defines the first isorevenue. This covers both the amount of losses that are avoided by the introduction of new techniques and the increase in output resulting from the introduction of the new technologies respectively. The
Factor markets 291 second isorevenue compares the revenue generated either by general or by contingent technological changes. Standard optimisation procedures make it possible to jointly identify both the correct amount of technological change with respect to the levels of technical change switching and the ratio of biased technological change with respect to a shift in technological change. Specifically it is a case of maximising output with a given isorevenue level which is set by the amount of adjustment costs that are necessary to reduce the mismatch between expected and actual market conditions. Formally the following relations are given: (6) TC = a(research activities) (7) tc = b(switching activities) where TC measures the amount of technological innovation necessary to change the technical area and tc measures the amount of technical change necessary to move within the existing technical area. In economics textbooks the amount of switching activities that are necessary to move within the area of existing techniques is very low because firms are not limited by bounded rationality and limited knowledge. The effects of irreversibility moreover do not limit their mobility within the technical area. The same economics textbooks suggest that the amount of resources necessary to change technology is extremely high. Hence a′ is small and b′ is high. In our analysis, instead, irreversibility and bounded rationality are relevant, as is a firm’s technological creativity. Hence a′ is large and b′ small. The choice between technical and technological change is affected by the specific content of technological change, i.e. whether it is general or contingent: (8) GSC = c(general innovation activities) (9) CBC = d(contingent innovation activities) where GSC measures the amount of shift that can be generated with a given amount of innovation activity necessary to introduce general technological changes and CBC measures the amount of bias that can be generated with a given amount of resources dedicated to innovation in order to introduce contingent technological change.4 Assume that it is possible to consider a frontier of possible adjustments such that, for a given amount of resources necessary to correct a mismatch, firms can generate either technical change (tc) or technological change (TC). Integral to the frontier of possible adjustments there is a frontier of possible innovations that can be obtained through the introduction of either general technologies (GSC) or new contingent technologies (CBC). Formally this means that: (10) tc = e(TC)
292 Introduction of localised technological change (11) GSC = f(CBC) So, to make standard optimisation procedures operational, two isorevenue functions need to be set. The first is defined as the revenue derived from adjustments (RA) and compares the revenue that adjustments involving switching within the technical area yield (SW) with the revenue of innovation (RI). The second isorevenue compares the revenue generated by the introduction of general technological change with the revenues generated by the introduction of contingent technological change. Formally, they are presented as: (12) RA = sSW + tRI (13) RI = rGSC + zCBC where s and t measure the unit revenue derived from switching and the unit revenue derived from innovation; and r and z measure the unit revenue derived from the amount of, respectively, general and contingent technological change generated by the given amount of resources available for innovation and induced by the new, unexpected conditions in the product and factor markets (see Figure 15.1). According to the analysis carried out, the slope of the innovation isorevenue is steep: introducing general technology yields far larger benefits than introducing contingent technologies. The slope of the frontier of possible innovations on the other hand reflects the large differences in the costs of introducing general technology compared with the low costs of introducing contingent technological innovation. The system of equations can be solved with the standard tangency solutions so as to define both the mix of contingent and general technological change which firms are advised to select and the amount of innovation involved in switching, a solution they may prefer to adopt. The system of equilibrium conditions is: (14) e′(TC) = t/s f′(CBC) = z/r subject to5 TC = GSC + CBC RI = rGSC + zCBC The cases of either only technical change or only technological change and alternatively perfectly general technological change, based upon pure shift effects, or purely contingent technological change, based upon a pure bias, seem extreme solutions. Much of the real world can be found between such extremes: technological change includes both a shift and a bias effect. The direction of technological change is influenced by the relative profitability of introducing general technological change with respect to contingent technological innovation, and the relative costs of introduction all play a key role.
Factor markets 293
Technical change (SW) E
Technological change (TC)
E General TC
Contingent TC Figure 15.1 The nested frontiers of possible adjustments and shift/bias technological change
The correct direction of the new technologies being introduced is the result of two different but complementary processes. From an ex ante point of view, myopic but creative firms select technological change which involves both a shift and a bias, which if mixed properly are the most appropriate to the specific conditions found in the marketplace in terms of both the profitability of introducing innovation and the relative cost of introducing them. This includes the levels of switching costs. From an ex post point of view, firms which happen to have introduced a
294 Introduction of localised technological change technological change along the correct direction have a greater chance of surviving. Firms which introduce innovations with the wrong bias are likely to be eliminated by a Darwinian selection mechanism activated in the product marketplace by rivalry among firms.
6 Applications and implications The approach developed so far clearly belongs to the class of models of induced technological change. The inducement hypothesis develops the assumption that firms generate new technologies when factor costs change. Firms can react to disequilibrium in factor markets not only by adjusting quantities to prices and vice versa but also, and mainly, by means of the generation, introduction and adoption of new technologies. Hence the primary inducement to introduce innovation is disequilibrium in the marketplace. This is a Marxian legacy, much developed and enriched by the economics of localised technological change. The levels of relative prices and specifically composition effects however have a strong inducement effect on the direction of the new technologies being introduced. The changes in relative factor prices induce the rate of technological change, because of irreversibility and limited knowledge, while the levels of relative prices induce the direction of technological change because of composition effects.6 In the approach developed in this chapter, any increase in wages (as well as in rental costs) induces the generation of new technology, because of the disequilibrium effects of irreversibility of the factors of production and related switching costs. Standard technical adjustment is inhibited by the costs of switching in the existing area of techniques. The introduction of new technology becomes a viable alternative to loss-making resilience. Here the ‘Hicksian’ inducement to the rate is relevant. The inducement to the rate of introduction of technological innovation, however, is separated from the inducement regarding the direction of the technological change. An increase in wages in a labour-abundant country with a large supply of labour and hence low wages should not induce the introduction of a labour-saving technology, but rather of a labour-intensive one, because of the powerful composition effects. The Kennedy–Samuelson inducement is relevant here. The inducement regarding the direction can be different from what is expected in traditional microeconomic inducement models. The identification of two quite distinct inducement mechanisms, the inducement to the introduction of technological innovation and the inducement regarding the direction of new technology, is relevant on two counts. First, it provides a more articulated explanation of the increased substitution effect engendered by the introduction of new biased technologies. Second, it corrects a basic inconsistency in the basic inducement hypothesis applied to factor markets where the prices of inputs differ sharply and the initial conditions of the production function are asymmetric. Let us analyse them in turn. Distinguishing between the inducement mechanisms seems appropriate to provide a sensible answer to the well-known critique Salter (1960) raised to the inducement hypothesis developed along the lines set out by Hicks (1932). Salter
Factor markets 295 (1960) noted that firms should be equally eager to reduce the use of capital and of labour irrespective of a recent increase in the unit costs of either factor. The basic aim of the firm is to reduce total costs. The approach developed here takes this argument into account. When relative prices change, firms are drawn into disequilibrium. Firms can react by changing either their technology or their technique. Irreversibility and switching costs however induce firms to change their technology, and composition effects induce the direction of the new technologies. In order to increase output levels and reduce average costs, firms will introduce and adopt the new technology which uses the relatively cheaper factor more intensively. This direction-inducement mechanism is activated by the levels of relative prices rather than by their changes. All changes in relative prices induce firms to innovate. The distinction between the inducement to innovate set off by changes in factor markets and the inducement to select a factor intensity for the new technology seems to be capable of bringing together the different strands of the inducement hypothesis and providing a broader and coherent context into which they are complementary rather than alternative (Ruttan, 1997, 2001).7 When composition effects are taken into account, the basic inducement hypothesis according to which an increase in the unit cost of a factor (wage) should induce a specific factor saving (labour saving) is no longer applicable. An increase in wages in a labour-abundant country might induce the successful introduction of a labour-saving technology only if there is also a strong shift effect. In such a country, even if wages have just increased it still seems sensible to introduce labourintensive technologies which take advantage of the low relative cost of labour. The basic hypothesis, as formulated by Hicks, can apply only in a symmetric and single system where both output elasticities and relative input costs are equal. The distinction between the inducement to innovate, due to all changes in relative factor costs, and the inducement to direct the bias of the new technology, as dictated by the composition effects, however, provides the inducement hypothesis with a broader and more articulated context of application. The framework developed so far provides a microeconomic tool with which to understand the role of relative prices as determinants of the direction of technological change at the level of the system. The hypothesis that technology is not exogenous, but is the result of the specific market conditions in which agents operate and reflects the historical process in which the markets interact, has been repeatedly put forward to explain the direction of technological change at the level of the system. Habakkuk (1962) developed the hypothesis that American technology was different from that in Britain because of the differences in the two countries’ factor endowments. A substantial scarcity of unskilled labour and relatively abundant natural resources and skilled labour characterises the American economy. Abundant unskilled labour and the institutional and geographic scarcity of land and natural resources instead characterise the British economy. According to Habakkuk, this difference does not lead only to the obvious variety of factor intensities in the two countries but also, and most importantly, to diverse paths of technological change. American technology is intrinsically biased in a labour-saving direction, while in Britain it is directed more towards capital saving. David (1975) has developed this
296 Introduction of localised technological change line of analysis further, suggesting that economic systems are better able to move along technological paths that lead them to enhance their technology by following and deepening the original bias. This argument, originally put forward by Habakkuk and David, has been questioned and has been subject to a systematic analysis assuming that each system is able to introduce new technologies which are locally progressive and are restricted to a range of techniques, defined by factor intensities, which reflect the relative scarcity of the factors of production (Antonelli, 1995, 1999a, 2001). This approach assumes technology to be endogenous, and its direction is strongly path dependent. According to this line of analysis, technological efficiency is very much dependent on the specific context of the application. Each technology and the related bundle of techniques, defined in terms of factor intensity, are appropriate to a set of idiosyncratic market conditions. The model developed so far provides an interpretative framework which helps us to understand the dynamics of technological change on a comparative and historical basis. According to Broadberry (1997: 5) in his impressive reassessment of the longterm British performance in manufacturing productivity in an international context, ‘the fact that accumulation of physical capital and human capital takes place around specific techniques helps to ensure that initial differences in factor proportions are preserved through time’ and with them differences among countries in terms of comparative productivity ratios. Countries which are able to introduce a general technological change characterised by a wide range of applications, which have such a major shift effect that no alternative technology can survive, can become leaders in the global economy. All the other countries will be forced to adopt the new technology but will have lower rates of overall factor productivity growth and hence higher production costs. Their shares of international markets will decline, as will the opportunities for profitability and growth. In such conditions latecomers can only try to rely upon creative adoption. The new general technology provides the opportunity to introduce an array of contingent technological changes which are aimed at adapting the new technology to local endowments and hence to local factor prices. Firms which are active in factor markets which are radically different from those where the new neutral technology was initially introduced can take advantage of contingent technological strategies and direct funds available for intentional learning and research activities towards the introduction of new technologies which build upon the shift already introduced. Thus, they are mainly directed towards a change in the relative composition of the productive inputs. At the other extreme there are the firms which operate in innovation systems characterised by an effective network of knowledge communication channels implemented by an advanced scientific infrastructure. This means there are headquarters of large corporations which have a well-established academic tradition guaranteeing high levels of research and development expenditure and organisational structures which can select and direct the results of R&D activities. In close cooperation with advanced public scientific centres they can operate as the hubs of knowledge clusters able to foster the local dissemination of new knowledge
Factor markets 297 and to favour their active recombinations by the adjacent smaller nodes. Such corporations already operate close to the technological frontier of production functions which already make the best use of the local endowments and show high levels of output elasticity for locally abundant production factors. These firms can develop technological strategies that are aimed at introducing actual shifts in the map of isoquants. They lead to the introduction of new technologies that are locally neutral. If they are introduced in other countries there will be major asymmetric effects for the adopters, while the innovations will provide additional benefits in terms of barriers to entry and to imitation. Such barriers will be based on big cost differences, and hence big mark-ups in protected demand niches. Further, in these countries, financial markets perform an effective role in both screening and assessing the economic viability of new technologies and are able to give a value to intangible assets and direct financial resources towards technological changes, introduced both by existing companies and by start-ups (Freeman, 1987). Firms based in intermediate countries have a real opportunity to choose between a more contingent and a more general technological change. It is clear that introducing a new general-purpose technology which has the most convenient specific mix of output elasticities for the local endowments is more profitable than introducing contingent technology which improves the local efficiency of a new general-purpose technology introduced elsewhere. The relative cost of introducing a radical shift technology and not a bias technology is a crucial factor which affects the choice firms in intermediate countries make. Access to scientific knowledge, both codified and tacit, plays a major role. When and if the academic and scientific infrastructure is in place and appropriate incentives are at work, technological communication between research centres and the business community is also effective. Further, if and when the general institutional conditions for the acquisition and use of new knowledge, especially in terms of intellectual property rights, and large scientific opportunities are available, firms may be better able to direct their research strategies towards the introduction of more general technology. Similarly, the availability of technological districts and local clusters of firms specialising in complementary research and innovation activities may help such choices. Important technological opportunities offered by the spread of new general technology, which is biased, at least for local adopters, offer them important incentives to direct research strategies towards the introduction of more contingent technologies. The conditions of access to external knowledge possessed by the providers of the new technology are very important here. This is because interaction between user–producers means that tacit knowledge can be shared. Effective protection of intellectual property rights in the global economy can prevent new general technological knowledge from being adopted and so delay the introduction of contingent technologies in other countries. All incentives which make the transfer of technological know-how faster may reduce such risks, as long as intellectual property rights are sufficiently protected. The characteristics of technological knowledge and of its generation process play a key role in this context. When high levels of fungibility characterise
298 Introduction of localised technological change technological knowledge, the introduction of contingent technological changes, focusing primarily on bias effects, is favoured. The introduction of new biased technologies can draw on a wide range of technological applications and the intrinsic versatility of the new knowledge. Instead, when technological knowledge is cumulative, the generation of new knowledge and eventually the introduction of new technology are based primarily upon the accumulation of competence and experience gained in previous vintages of the same knowledge. It is more likely that the introduction of general technological change with no bias effects and which is strictly neutral, at least locally, will be favoured. The third relevant parameter is provided by the specific conditions of the factor markets. In regions and industries where the difference in factor prices is so close that the ratio of relative prices is near to unity, resulting in the slope of the isocost and the former technology being represented as a symmetric production function, the incentive to introduce contingent technology clearly is very low. In such regions a firm’s research strategy is of necessity directed towards introducing technology which does not change factor intensity and is mainly based on a neutral shift. In the opposite case, in regions where the supply of a specific input is abundant and its derived demand is very low, there is a unique set of opportunities to direct research strategies towards introducing contingent technology. Similarly in regions where the market prices of the factors of production are very elastic to all increases in demand, firms are likely to direct their innovation strategies towards introducing neutral technology. This means that a research strategy directed mainly towards introducing and adopting contingent technology can be valid as long as firms are active in regions where the current factor intensity is significantly different from that in the countries where shift technologies have been introduced. The difference in relative prices between countries is a prime factor in determining what kind of innovation strategies is chosen. The levels and duration of transient monopolistic extra profits due to barriers to entry and imitation for potential adopters influence the choice and introduction of general technologies. It is clear that the greater the diffusion lag is and the greater the cost differences among innovators and imitators are, then the higher the incentive to introduce general technology is. The long-term shape of the supply schedule for production factors is also important in this context. The profitability of introducing contingent technological changes can be severely reduced by the rigid supply of the most productive factors, and hence there is a sharp increase in relative costs because of the introduction of new technology. Barriers to entry and exit in upstream sectors may change the relative profitability of both introducing and adopting new contingent technology. In general it seems clear that industrial dynamics and market structures play a major role in determining how profitable it is to introduce either of the technologies.8 The analysis of the effects of local factor markets on the productivity of new technologies and the ranking of the profitability of adoption in terms of the matching between the bias of the technology and the relative abundance of the most productive factor in each local factor market provides multinational corporations
Factor markets 299 with a unique opportunity to take advantage of the localised fitness of innovations (Cantwell and Iammarino, 2003).
7 Conclusion In the traditional analysis of the economics of technical change, the introduction of technological change is presented as the substitution of an old technical map with a new one. A new map of isoquants is associated with a new technology. The introduction of a new map of isoquants has two consequences: first, it may, at given factor prices, set off a substitution process, including the factors of production; second, it enables the levels of output for given levels of inputs to be increased. The substitution of the old map of isoquants with the new one is analysed in a single and static context: a situation in which the heterogeneity of factor markets and production functions at work in a global economy is not taken into account and where there is no change in the relative prices of the factors of production. Moreover, possible overlapping between the old map and the new map is not considered. The analysis is concentrated on the narrow area of techniques defined in the maps of isoquants representing the previous equilibrium, as determined by the tangency between the isocost and the relevant isoquant. The inducement approach has split analysis into two strands: the microeconomic model stresses the role changes in factor prices play as the basic inducement which determines the rate of technological change. The macroeconomic model focuses attention on the levels of factor shares and hence on relative factor prices as the mechanism that induces the direction of technological change at the aggregate level. The economics of localised technological change helps to reconcile and integrate the two approaches. Much of the current analysis of the effects and determinants of each new wave of technological change does not seem to provide a systematic understanding of the static and dynamic role of the structural characteristics of the economic system into which the new technologies are being introduced. More generally, too much attention has been paid to assessing what affects and what determines the rates of technological change. Instead, there has been too little analysis of what determines and affects the direction of technological change. Even less attention has been paid to the interaction between the rate and the direction of technological change in a dynamic and complex context, a situation in which factor costs are allowed to change in time and in area. Composition effects, the actual levels of the measured overall factor productivity of each technology, depend on the specific system of relative prices in each factor market. Composition effects determine the consequences of introducing technological change in each regional system, and are characterised by the specific system of relative factor prices in two ways. The contribution of the economics of localised technological change is very important here. Irreversibility and limited knowledge engender switching costs that limit the mobility of firms within the area of existing techniques. All changes in input costs, in this context, set off a clear inducement to introduce technological
300 Introduction of localised technological change innovation. Firms in each region, induced to innovate, will introduce the technology which best fits in with the specific conditions of the factor markets. Relative factor prices become a selection mechanism which makes it possible to choose technology. Over time firms based in one region will make consistent choices and select technologies shaped by a similar factor bias. Hence, composition effects can be endogenised by potential innovators who direct their technological efforts towards introducing technology which is biased in such a way as to make the best and most productive use of the production factors which are most easily available. Thus they have the lowest prices in each specific region. At a general level, there is technological variety across regions in both cases. This is because the bias in the adoption and the bias in the generation of new technology lead to choosing the mix which is the most appropriate to the specific factor market in each region. Such a bias in the direction of technological change can be thought of as being due to the intentional ex ante decision of innovators who are well aware of the relative scarcity of inputs in their own region. Innovative firms, for a given cost of an innovation, will find it more profitable to introduce new technology which makes a more intensive use of the locally most abundant factor. The bias in the direction of technological change can also be determined ex post by a selection process among innovators. Those firms which happen to have introduced the technologies which use the locally most abundant production factor most intensively will emerge as the winners of the selection process. Replicator dynamics will force the ‘wrong’ innovators out of the market and will favour the ‘correct’ innovators, who will rapidly increase their market share. The direction of technological change in terms of the specific form of its bias sequentially introduced and adopted reflects the specific conditions of local factor markets. In the long term, well-defined technological paths emerge in each region as the result of the selection process in the global product markets. The more rigid and idiosyncratic the endowment and the system of relative prices are, the more specific the technological path of each region is likely to be. The direction of the technological path may change as each economic system is exposed to international competition. After a new radical and general technology has been introduced, the search, in each country, for appropriate technologies may lead to the introduction of new contingent technologies that reshape the form of the production function. In any event, the introduction of new technology is clearly the result of an outof-equilibrium situation, which forces the firm to innovate. Firms will innovate if a number of key systemic conditions exist. Such a situation can provide a unique opportunity to bring together results of the economics of innovation, which is more interested in assessing the rate at which innovation is introduced and analysing the characteristics of new products and new processes and merging them into an analytical framework which develops the role of factor intensities and output elasticities. The basic common thread and unifying element in the above analysis is the out-of-equilibrium approach, which is a distinctive element of the economics of localised technological change.
16 Localised product innovation The role of proximity in the Lancastrian product space
1 Introduction This chapter provides an extension and an application of the notion of localised technological change, traditionally applied to process innovations, to understanding the role of technological and market proximity in the introduction of product innovations. The basic ingredients of the localised technological change approach are applied to understanding the dynamics of not only process innovations but also product innovations. The main contribution of this chapter consists in the use of a variation of Kelvin Lancaster’s (1971) concept of product space to demonstrate how the clustering of product innovations in technical and product space may be beneficial and how it affects the innovative conduct of firms. The implementation of the framework elaborated by Lancaster to analyse the choices of consumers with the analysis of the marketplace based upon monopolistic competition provides a context into which the role of proximity in the product space can be assessed and the dynamics of localised product innovation can be understood. In this way the chapter provides a microeconomic foundation for the analysis of induced innovations and implements it along the lines of the localised approach to analysing technological change. This chapter is structured as follows. Section 2 provides a preliminary discussion about the role of proximity in product innovation from both the technological and the market viewpoint. Section 3 elaborates a simple model which makes it possible to implement the basic intuition that firms are able not only to introduce innovations when facing unexpected events, but also to choose whether to introduce product or process innovations, according to the localised context of opportunities and constraints that are defined in historical time. The conclusion summarises the results and explores some implications for empirical research, innovation policy and innovation strategy.
2 Proximity and product innovation Localisation in multidimensional spaces matters because of four classes of reasons: 1) Agents are characterised by bounded rationality and yet are able to learn. The capability and the competence acquired by means of learning processes are heavily localised in a limited technical space. Technological and organisational innovations
302 Introduction of localised technological change are possible only in the proximity of the specific learning context. 2) Proximity in regional and technological space to other learning agents makes it possible to take advantage of communication flows among complementary innovations and innovative activities and hence of contextual spillovers. 3) Irreversibility of fixed production factors limits the mobility in the technical space and constrains agents to make the best possible usage of existing inputs. 4) Relative factor prices favour the use of technologies that make the most effective usage of locally abundant inputs. The specific contribution of this chapter consists in the integration in the analysis of the role of proximity in the product space. Proximity in the product space is relevant for three classes of reasons: the role of localised learning about consumers’ tastes; the effects of brand loyalty and reputation on consumers’ choices; and the effects of product differentiation and reputation, as sources of barriers to entry and to mobility, on market structure. Let us analyse them in turn: •
•
The role of proximity in the knowledge space and the generation of product innovations. Firms can introduce successful innovations only when sufficient competence and tacit knowledge are available. Such competence can only be acquired in localised learning processes. Because learning is a necessary condition for the efficiency of the innovation activities, the introduction of innovations cannot take place too far away from the context of action and the roots of the competence of the firm with respect to its consumers and competitors. Not only are the competence and experience that are necessary to innovating acquired in the repeated usage of a given set of capital goods and intermediary products and in the production of well-identified products, but the experience accumulated in marketing and interacting with a welldefined set of consumers and competitors in a limited range of products is necessary in order to generate new knowledge and eventually introduce new products. Interactions with customers are a primary source of tacit knowledge about their tastes and needs (Lundvall, 1985). No successful product innovation can be effectively and successfully introduced without some dedicated competence about the marketplace. The distance, in the product space, from the products being traditionally delivered to the marketplace can be considered a strong factor of increasing innovation costs and decreasing efficiency in the generation of innovations. Proximity in the product space matters as the prime source of information about the tastes of customers and their potential interests. Proximity in the product space matters as well as a factor to know more about the capabilities of competitors and their strategic attitude. The introduction of product innovations in market niches that are far away from the source of the experience of each firm is put at risk by the lack of specific competence, and relevant, additional costs should be recognised (Von Hippel, 1988). The role of proximity in the consumers’ choice. The demand for radically new products with a mix of characteristics that is completely different from the experience and competence of consumers is affected by relevant problems. As a huge literature on the diffusion of innovation shows, it takes time for
Localised product innovation 303
•
new products to be fully appreciated. As Bianchi (1998) notes, the choices of consumers are the result of both tastes and capabilities acquired by means of experience, skills and hence localised knowledge. The demand for new products delivered by firms with low levels of reputation and brand recognition has to take into account the burden of relevant switching costs, consisting in information, search and transaction costs necessary to prospective customers in assessing and valuing the new products (Klemperer, 1987a, 1987b). Such costs increase with the distance of the new product in terms of characteristics with respect to the existing ones. Hence it is clear that the demand for new products is negatively affected by their distance in the space of product characteristics from the existing ones.1 It seems clear that the greater the novelty of the product is, with respect to previous innovations, and the lower the reputation of the innovating firm, the greater is the resistance of customers to accepting the new product. The role of proximity in the marketplace. A large literature has explored the role of barriers to entry and limit pricing in homogeneous product markets. Much less attention has been paid to analysing the effects of barriers to entry and to mobility in markets characterised by high levels of product differentiation and hence monopolistic competition (Caves and Porter, 1977). Monopolistic competition cum barriers to entry and to mobility in markets that are structured as overlapping niches with high levels of cross-price elasticity seems a promising area of application and implementation for understanding the localised dynamics of product innovation (Sylos Labini, 1956, 1984; Metcalfe, 1997). The price at which firms can sell in a marketplace characterised by monopolistic competition with barriers to entry and to mobility is affected by the idiosyncratic characteristics of their products including their reputation. Established firms which enjoy a good reputation and produce specialised and differentiated goods with low levels of substitutability can secure large shares of the general demand, and they can sell their products at high price–cost margins. The levels of the price–cost margins are strictly associated with the degree of product differentiation. With high levels of product differentiation, monopolistic competition cum barriers to entry and to mobility applies. Incumbents can control a captive demand and act as local monopolists. The size and the extent of their captive demand are determined not only by the differences in the manufacturing costs of rival producers in adjacent product markets but also by the effects of switching costs for consumers (Sylos Labini 1956, 1984; Swann, 1994; Klemperer, 1995).
A number of counter-arguments about the negative role of proximity in the product space should be considered in order to provide a more balanced view. From the supply viewpoint, proximity may lead to major opportunity costs. Long ago Granovetter (1973) noted that firms which restrict their search for new techniques to a narrow range of sources are likely to miss out on (often very important) changes originating beyond their normal networks. This point was subsequently restated and complemented by other important empirical studies by Pavitt (2000)2 using
304 Introduction of localised technological change US patent data to show that firms are widening their range of technological activities, presumably in part because they need to cover more fields as technologies increasingly cross industry boundaries. The point is not that proximity cannot be valuable, but that it may not always be feasible for firms to confine their learning to familiar terrain.3 On the supply, demand and market sides it is also clear that the characteristics of the economic topology matter in assessing the role of proximity. The distribution of firms and consumers in the product space is a historical factor with important effects. Following Burt (1992), it is clear that ‘weak ties are essential to the flow of information that integrates otherwise disconnected social clusters into a broader society’ (Burt, 1992: 26). It is clear that the introduction of a radical product innovation with a radical new mix of product characteristics can have such strong positive effects in terms of consumers’ welfare as to easily undermine their resistance. In sum, proximity is an important factor from different viewpoints in the conduct of firms when considering the introduction of a product innovation. This is quite a familiar premise in strategic management literature. Rumelt (1974) and many subsequent writers have been arguing that a firm is most likely to be profitable if it can restrict its activities to a narrow and familiar range: ‘the intensive cultivation of a single field has proven, on average, financially more successful than bold moves into uncharted areas’ (Rumelt, 1986 [1974]: 156).4 Specialisation in a single business and unrelated diversification appear as two negative extremes in terms of both profitability and rates of growth. Related business diversification has long been regarded as the proper strategy (Porter, 1985). Building upon the methodology introduced by Lancaster to study the behaviour of consumers and the role of product characteristics, progress can be made with the elaboration of a framework which makes it possible to analyse the role of proximity in the product space as a determinant of the innovative conduct of firms.
3 A Lancastrian model of localised product innovation Much work has been done in the localised technological change approach to inquire into the conditions, characteristics and determinants of the trade-off between technical change and technological change. The introduction of technological changes is possible only if appropriate amounts of knowledge and competence have been accumulated and are available to firms. The conditions of the learning processes and the determinants of the eventual production of knowledge such as the characteristics of the internal organisation and structure of firms, the structure of the local systems of innovations, the channels of communications among firms and between them and scientific institutions, the forms of interactions and cooperation between firms active in the same industry as well as across industries and diverse markets, the working of labour markets as vehicles for the transmission of information and knowledge, the management and the structure of the relations among users and producers, the positive and negative effects of the spillover of proprietary knowledge among rivals and more generally the governance of the appropriability conditions and the structure of intellectual
Localised product innovation 305 property rights have received much attention. Much work has also been devoted to analysing the effects of the irreversibility and duration in historical time of capital goods and intangible assets in shaping the conduct of firms (Antonelli, 2001, 2003a). Because of the key role of learning in the production of new knowledge and the eventual introduction of technological innovation, proximity between the techniques in place and the new technology plays a key role. Learning is eminently localised: firms learn by doing and by using current techniques; they learn by interacting with current customers, rivals and suppliers; they acquire the external knowledge spilling in the close technical and regional surroundings. Technological change is localised also because technological knowledge is localised, that is closely related to localised learning processes and to the context in which it takes place, in terms of local knowledge externalities. The competence acquired in a given localised context declines with distance in regional space, for its key role in interaction among organisations and human beings, in product space, for the emerging complexity associated with the variety of customers and their preferences and the variety of rivals, and in technical space, for the difficulty in adapting to new procedures and processes. Knowledge proximity translates into proximity in technical space, proximity in product space and proximity in regional space (Antonelli, 2001). Along this line of inquiry significant progress can be made when the differences between product and process innovations are considered. The new – localised – technology can concern the process or the product. The trade-off between technical and technological change is affected not only by the opportunity for the introduction of process innovations but also by the prospects for the introduction of product innovations. Firms, in other words, consider two joint choices: the first is whether to introduce technical or technological change and the second is whether to introduce product or process innovations. Process innovations make it possible to increase efficiency and face adverse market conditions with a reduction in production costs. Product innovations make it possible to increase the quantities that can be successfully delivered to the marketplace. In turn the introduction of product innovations requires not only the command of technological knowledge, but also the competence and experience acquired by means of learning by doing and by interacting with the customers. Competition in Schumpeterian markets, where the emphasis is on innovation and rivalry, is based upon product differentiation and product innovation rather than price. In this context, monopolistic competition cum barriers to entry and to mobility proves to be the appropriate analytical framework. Each firm is a local monopolist in a niche product market. The demand for the family of substitutable products is split into a variety of niches. The size of each product niche, that is of the captive demand for each local producer, depends upon the costs and related market prices for potential entrants for their own products, narrowly defined in terms of product characteristics. Incumbents charge monopolistic prices upon their own captive demand. The dynamic distribution of firms in the product space and the costs of mobility and the opportunities for rival entry become key issues (Scherer, 1984, 1992; Swann, 1994).5
306 Introduction of localised technological change The framework elaborated by Lancaster (1971) to study the choices of consumers faced with a variety of products that are differentiated with respect to a variety of characteristics is very helpful in our context. The notion of product space provides a unique context in which the analysis of the role of product innovation can be carried out. Moreover it can be easily used and implemented with a simple analytical and geometric exposition (Gravelle and Rees, 1981). Let us start with the formal introduction of the characteristics of the goods, as distinct from the goods themselves: (1) A = (a1, . . ., an) denotes a bundle of characteristics. The amount of each characteristic depends on the bundle of goods: (2) ai = f(x1, . . ., xn) = f i(x) The goal of the consumer is to optimise his/her utility, as it is generated by the mix of goods that can be acquired and hence by the mix of characteristics that each good contributes, for given prices of the goods and a budget (B). Geometrically, a ray represents each good, and the prices for the good are defined in terms of the distance on the ray from the origin (see the rays F and G in Figure 16.1, where the vertical axis represents the characteristic a1 and the horizontal one the characteristic a2). The budget line is easily identified as the linear combination of all the quantities of the different goods that can be purchased with a given budget B. The standard maximisation easily applies: (3) max u (a1, . . ., an) s.t. (i) Σ pj xj = B (ii) aj′ = f j (x1, . . ., xn)
F
a1
E
G
FE GE
a2 Figure 16.1 Equilibrium in a Lancastrian product space
Localised product innovation 307 It should be clear that the total amount of characteristics produced by the bundle of goods acquired is the result of the sum of the specific characteristics generated by the accessible quantity of each good. The well-known ‘parallelogram rule’ makes sure that a linear combination of goods can always be identified so that the desired and optimum levels of characteristics chosen by the consumers can be ‘generated’ by the appropriate mix of goods. In Figure 16.1 the equilibrium quantities FE and GE are identified by means of the parallelogram rule as the projection of the equilibrium point E. Building upon the analytical context provided by the merging of the analysis of monopolistic competition cum barriers to entry and mobility, we assume that each firm is a local monopolist in that it is the single supplier of each good: each firm produces the good defined by the slope of each ray. Hence each ray represents both a good and a firm. The size of the demand for each product depends upon the utility function and the budget of the consumer and upon the extent to which the product can be substituted by a differentiated and yet rival product. The notion of cross-price elasticity applies here. Let us assume that the equilibrium situation identified with a given set of preferences of consumers, given cost conditions for a given family of rival producers and a distribution of single incumbents in a given set of product niches, is satisfactory for all. Changes to such an equilibrium condition can be brought about by: 1) the reduction of the prices of the goods sold by competitors; 2) the increase in the market price for the firm due to any increase in input costs; and 3) a change in the preferences of consumers, who now prefer other products and other characteristics. These changes lead to a reduction in the quantities delivered in the marketplace by at least one of the firms previously in equilibrium and hence in a reduction of its profits. The reduction can be so sharp that losses emerge. The firm is now exposed to a clear decline in the levels of performance and of satisfaction. A reaction is necessary: it can be a passive one and consist in the traditional technical change, defined as a movement in the space of existing isoquants, or a more creative one, so as to include a change in the routines and the eventual introduction of innovations either in the product or in the process. The difference between current profits, after the changes in the marketplace, and the profits that should have been possible without such changes is indicative of the amount of resource the firm is ready to commit in order to bring about the changes that are likely to restore the expected levels of profitability. In other words, because of the mismatch between expectations and the actual conditions in the marketplace, the firm cannot remain in the position that had been planned. The introduction of technological innovations is a viable alternative to technical change. Both adjustments are possible but are costly. Technical change, because of the irreversibility of existing production factors and limited knowledge about the existing techniques, requires some switching activity. Technological change on the other hand, by definition, is not on the shelf, and its introduction in turn requires some innovation activities. Technological change can consist of either process innovations or product innovations. A combination of product and process
308 Introduction of localised technological change innovations can also be considered. Process innovations can be easily measured in terms of increased efficiency in the production process. Product innovations can be measured in terms of the change in the slope of the ray representing the new product in the Lancastrian space of product characteristics. Specifically here the metrics of the innovation output are measured respectively: 1
2
for process innovations, by the reduction in the price, i.e. cost plus mark-up, the firm can achieve in the marketplace and hence in the distance from the origin along the ray (see in Figure 16.2 the new intersection F2 between the new budget line B2 and the ray F); for product innovations, by the change in the slope of the ray defined in terms of product characteristics, with respect to the original one (see in Figure 16.2 the change in slope between the old ray G and the new ray H).
The choice set is now framed. The firm faces two nested frontiers of possible changes in order to solve the mismatch between plans and real market conditions. The first frontier of possible changes is the frontier of possible adjustments which makes it possible to compare the results of resources invested in either technical changes or technological ones. The second frontier, the frontier of possible innovations, compares the kinds of technological change and whether they consist of product or process innovations. The frontier of possible innovations has a concave and asymmetric shape that reflects the strong effects of distance, in the space of product characteristics, on the introduction of product innovations. This effect is stronger in the introduction of product innovations rather than process innovations (see Figure 16.3). The position of the frontier of possible adjustments is defined by the amount of resources R that the firm should invest just to switch from the previous equilibrium F
a1 F2
H
F1
G
B1 B2 a2 Figure 16.2 Product and process innovation in a Lancastrian product space
Localised product innovation 309 Product innovation (PRD)
Technological change (TC)
Resources (R)
Technical change (SW)
Resources
Process innovation (PRC)
Resources
Resources
Figure 16.3 The production of technological change, technical change, product innovation and process innovation with a given amount of resources
technique to the new one. The search for the correct solution, in other words, is identified as a maximisation process where the firm tries to maximise the amount of changes, including technological innovations that can be generated with a given amount of resources set by the levels of switching costs.6 The firm can identify the correct solution, in the given context of action in the marketplace, with the existing technological opportunities and learning conditions, by means of standard maximisation of the output, for two given nested frontiers, when two nested isorevenues are defined. The absolute levels of the revenue generated by all adjustment activities consisting in the revenue made possible by the introduction of new techniques and the revenue made possible by the introduction of the new technologies respectively define the first isorevenue. The second isorevenue measures the bundle of revenues generated by either product or process innovations. Formally we see the following relations: (4) TC = a(R) (5) SW = b(R) (6) PRD = c(R) (7) PRC = d(R)
310 Introduction of localised technological change where TC measures the amount of technological innovation necessary to change the technical space that the firm can generate taking into account the internal competence and knowledge accumulated and the external knowledge it can access; SW measures the amount of technical change necessary to move in the existing technical space and reflects the levels of irreversibility and rigidity of tangible and intangible capital; PRD measures the amount of product innovation; and PRC measures the amount of process innovation that can be generated with a given amount of dedicated resources (R) defined by the amount of switching activities the firm needs to complete to move from one equilibrium point to the other. It is clear that the relationship between the four production activities is essential to define the outcome of the search process initiated by the changes in the product markets. It seems clear that, the greater is the efficiency in the production of technological changes and the lower the efficiency of switching, the greater is the amount of innovations introduced. Correspondingly, the greater is the efficiency of product innovation and the smaller the efficiency of the generation of process innovations, the greater will be the amount of new products each firm will generate. To make this point more compact, let us now assume that a frontier of possible adjustments can be considered such that, for a given amount of resources (R) necessary to face the mismatch, firms can generate an amount of either technological change (TC) or technical change (SW). Nested to the frontier of possible adjustments we find a frontier of possible innovations that can be obtained with the introduction of either product innovations (PRD) or process innovations (PRC). Specifically the shape and the slope of the innovation frontier reflect the negative effects of the distance in terms of the technological opportunities associated with and based upon the localised competence built by means of localised learning by doing and by interacting in the marketplace with customers. Formally this amounts to saying that: (8) SW = e(TC) (9) PRD = f(PRC) In order for standard optimisation procedures to be operationalised, two isorevenue functions need to be set. The first, defined as the revenue of adjustments (RA), compares the revenue that adjustments by switching in the technical space (SW) yield with respect to the revenue of innovation (RI). The second isorevenue includes the revenue generated by the introduction of product innovations (PRD) and the revenues generated by the introduction of process innovations (PRC). Formally we see: (10) RA = sSW + tRI (11) RI = rPRD + zPRC where s and t measure the unit revenue of switching and the unit revenue of technological change; and r and z measure respectively the unit revenue of the
Localised product innovation 311
Technical change (SW) E
PRDE
PRCE Technological change (TC)
E Product innovation (PRD)
Process innovation (PRC)
Figure 16.4 The nested frontiers of possible adjustments and possible innovations
amount of product and process innovations generated with the given amount of resources available for innovation and induced by the new and unexpected conditions of the product and factor markets. The system of equations can be solved with the standard tangency solutions so as to define both the mixes of product and process innovations which in each specific context firms are advised to select and the amount of technological change which switching the context suggests selecting. The system of equilibrium conditions is:
312 Introduction of localised technological change (12) e′(TC) = t/s f’(PRC) = z/r subject to TC = PRD + PRC and R= RF7 The cases of either only technical change or only technological change and in turn either product innovations or process innovations are extreme solutions. Much of the real world can be found in between such extremes. Firms are induced to innovate by the mismatch between actual and expected conditions of their production set and their market conditions, necessarily built upon irreversible decisions taken on the basis of myopic expectations which are not met by the disequilibrium conditions in product and factor markets. The type of technological change is influenced by the relative profitability of introduction of product innovations with respect to process ones. The slope of the innovation isorevenue reflects the effects of distance in the product space on the levels of price–cost margins and equilibrium demand for product and process innovations respectively. According to the shape of the innovation isorevenue, both the composition of technological change, whether it consists mainly of product or process innovations, and the mix of possible changes, whether they consist mainly of switching activities or technological changes, are affected. There are three possibilities. First, the isorevenue has a single slope: prices do not change with distance. In this case there are no negative effects of the distance in the product space on the demand for the new product and on the prices because of increased competition. Hence the introduction of product innovations will be favoured. Second, the isorevenue is convex: the greater is the distance of the new product from the original one, as measured by the slope of its ray in the product characteristics, the greater are both prices and equilibrium demand. The number of rivals in the product space of rivals is low and the size of customers’ niches interested by the new product is very high. The introduction of product innovations is much favoured. As a consequence also the equilibrium amount of technological change, with respect to switching, is increased. Finally, the isorevenue is concave. This would reflect the case in which, the greater is the distance of the new ray from the original one, the lower are the prices the innovating firm can charge and the smaller are the quantities of the new product the firm can actually sell. Specifically, we see here that, the greater is the distance of the product innovation ray from the origin and 1) the lower are the positive effects on prices of reputation and brand loyalty and 2) the lower are the barriers to entry and to mobility for rivals producing products with characteristics that are closer and closer, the lower is the protection for incumbents provided by reputation and brand loyalty. The reduction in this distance engenders an increase in the substitutability for consumers and an increase in their switching costs. Clearly in this case the firm would prefer to introduce process innovations and possibly rely
Localised product innovation 313 more on switching activities rather than on technological change in order to deal with the mismatch between plans and facts. The equilibrium conditions identified by equation (12) capture the essence of the dynamics.8 The role of proximity in both technical and product space can now be fully appreciated. Firms tend to introduce product innovations in a limited region of the product characteristics space, as they are constrained by the negative effects of distance both on their actual competence and technological capability in the innovation process and on their demand and the conditions of the monopolistic rivalry. New products can be introduced in a broader region of the Lancastrian space of characteristics when the density of firms is lower and the distance from the original specification of the product sold by each innovator makes it possible to deliver products that are much more appreciated by consumers (and when little competition is found). The Lancastrian approach to modelling the choice set of consumers provides a fertile methodology which helps in framing not only the static context of analysis of the role of characteristics of products in the choice of consumers, but also a dynamic context, one where the characteristics of the products can change. In so doing, the Lancastrian methodology becomes an important device to understanding and modelling the dynamics of localised product innovations.
4 Conclusion Much empirical and theoretical work has made it possible to appreciate the strong localised character of technological change and the role of path dependence in the determination of its rate and direction. Such analysis has paid much attention to the factors that localise technological change in the technical space, because of the role of irreversibility of the existing stock of fixed capital and the effects of learning in the accumulation of the competence and technological knowledge that are necessary for the successful introduction of new technologies. Many analyses reviewed in Part II have assessed the role of proximity in geographic and knowledge space as factors in localising technological change because of the key role of complementarity and externalities in the generation of new knowledge and the key role of knowledge governance and communication in climbing and standing on giants’ shoulders. Communication and hence spillovers are easier among agents who are close to each other both in the actual geographic space and in the knowledge space. It is easier for firms to introduce technological innovations that complement each other and take advantage of strong elements of commonality. Complementarity and interdependence among technological innovations and hence technological proximity, in the form of both technological fungibility and technological compositeness, add on so as to build technological systems and favour the localisation of further innovative efforts and additional technological innovations within such systems. Little analysis has been, so far, devoted to understanding the dynamics of localised technological change in the introduction of product innovations. The empirical evidence available9 however suggests that product innovations introduced
314 Introduction of localised technological change by each firm tend to be localised in a limited region of the space of product characteristics. Longitudinal analyses of firms able to survive in the long run confirm that they tend to specialise in a narrow range of products. The introduction of the now traditional set of hypotheses about the role of proximity and distance not only in learning but also in the competitive space characterised by monopolistic competition yields important results that confirm the dynamics of localised technological change also with respect to the introduction of product innovation. The analytical framework introduced by Lancaster (1971) proves to be especially fertile and productive, able to accommodate the analysis of the innovation process in the space of product characteristics. Moreover it seems that the Lancastrian approach used so far as a tool to stretch the localised technological change approach provides a useful analytical context in which much evidence provided by strategic management finds an appropriate economic understanding. It seems clear that product innovation is more localised the more specific and localised is the process of accumulation of competence and the more relevant is the latter in the generation of technological knowledge and the more dispersed the distribution of firms in the product space. Such results can become useful both from the viewpoint of the economics of industrial strategy and from an innovation policy perspective. The argument developed so far leads to articulation of a number of clear and testable hypotheses. First, the levels of R&D activities should be positively associated with the variety of product innovations, as distinct from the quantity of product innovations, introduced by firms.10 Second, constrained diversification is most successful when learning by doing, by using and by interacting with customers are relevant factors that cannot be dispensed with. Third, the lower are the technological opportunities for the introduction of process innovations, for given levels of inducement pressure, the greater is the variety of product innovations. The heterogeneity of product innovations and hence the diversification and differentiation of firms should be greater in traditional industries rather than in high-tech ones. In these industries the introduction of process innovations depends mainly upon technological change introduced upstream and embodied in capital goods and intermediary products. Fourth, joint ventures and mergers and acquisitions are complementary strategic steps for firms willing to extend the ray of their product innovations. The acquisitions of firms already active in other product markets increase the command of technological knowledge and competence in new niches that are far away from the original core of the firm. The introduction of general-purpose technologies, such as in the case of the gale of new information and communication technologies, can have an important impact on the picture elaborated so far. High levels of fungibility characterise the new generic knowledge and make it possible to favour the introduction of an array of new product innovations even in fields that are far away from the traditional localisation of firms. A related hypothesis, hence, can be elaborated and eventually
Localised product innovation 315 tested: the higher is the penetration of new information and communication technologies, both at the firm and the system level, the greater is the variety of product innovations being introduced both by each firm and by each economic system. Finally, it seems clear that the choice between process and product innovations is affected by the role of historical time and creativity, and specifically by the balance between internal and external irreversibility and learning conditions. Demand and competitive effects are external to each firm and reflect the given characteristics of the existing distribution of firms in the product space, as well as of the tastes and density of consumers. The effects of competence and switching costs instead are internal to each firm and reflect respectively the role of localised technological knowledge and its strong association with localised learning processes, and the effects of historical time in the accumulation of tangible and intangible production factors. Here again the role of path dependence can be appreciated. The localisation of firms in the space of product characteristics at time t has major consequences on their innovative behaviour and hence on their location in the space characteristics at time t+1. At the same time the process is not past dependent because the mix of competence and technological knowledge available at each point in time can lead the firm in any possible direction. Specific conditions affecting learning processes, such as localisation in regions where innovation activities cluster and scientific and technological communication among complementary research activities is easier, can favour the introduction of radical product innovations that push the firm far away from the original localisation in the product space but closer to the core competence of the region. The availability of external sources of codified knowledge with a strong scientific content can help firms to venture into new product characteristics far away from traditional ones, so as to complement and compensate for the dwindling support of tacit knowledge built upon learning that is declining with the distance from the traditional set of operation. An effective science and technology policy is a strong enabling factor, which strengthens the inducement mechanisms and helps firms to rely upon innovations when facing adverse market conditions.
17 Diffusion as a process of creative adoption
1 Introduction The study of technological change has made much progress by means of artificial disjunctions between aspects that it is difficult to separate. The traditional divide between innovation and diffusion can be reconciled in the context of the economics of localised technological change, focusing on the analysis of the determinants of the adoption process. The new attention to the active role of consumers and user–producer relations in understanding demand and in shaping technological change brought about by Bianchi (1998), Metcalfe (2001) and Witt (2001) contributes a new approach to the economics of diffusion. The new approach focuses on the role of adoption as an active process. Adoption, like other consumption, cannot be regarded as a passive attitude. It requires, instead, the active participation of users not only in terms of the search and eventual choice among a range of existing products, but also and mainly in terms of a specific and dedicated activity of adaptation of available products, whether brand new, just introduced or existing ones, to the localised and idiosyncratic needs and constraints of users, as shaped by irreversibility, routines and switching costs. Adoption is the result of a complex process of decision making. Firms are induced to change their technology when product and factor market conditions do not meet their expectations and irreversible choices make adjustments expensive. Technological change consists both of the introduction of original ‘never-seen-before’ technologies and the adoption of technologies that have already been put in place elsewhere. Even adoption requires that a number of highly specific and idiosyncratic problems of adaptation and integration are solved. Moreover adoption requires that a number of preliminary activities are carried out, such as the search, the selection, the identification, the adaptation and the integration into the production process and the firm at large. Technological change, for each firm, is the result of both research and imitation activities. Both command resources and engender specific revenues. Localised technological change consists of creative adoption where external knowledge and embodied technologies are implemented, with internal competence and idiosyncratic knowledge acquired by means of learning processes. The identification of the net profitability of adoption as defined by the gross profitability of adoption minus adoption costs contributes the economics of
Diffusion as a process of creative adoption 317 technological change. The analysis of the evolution of the net profitability of adoption in the context of the economics of localised technological change shows that the dynamics of creative adoption is able to generate an S-shaped diffusion path at the aggregate level. The rest of this chapter articulates this approach as follows. Section 2 recalls the basic acquisitions of the economics of diffusion and adoption and elaborates the notions of induced adoption, adoption costs and net profitability of adoption. Section 3 presents the model of localised technological change consisting of both the induced introduction of new technologies and the induced adoption of technologies already available in the marketplace. Section 4 shows that proper modelling of the dynamics of adoption costs and gross profitability of adoption can lead to the standard S-shaped diffusion processes. The conclusion summarises the results of the work.
2 Adoption and diffusion in the business sector The distinction between innovation and imitation was first introduced by Joseph Schumpeter and has eventually become a landmark in the economics of innovation and new technology. A new technology, either a new product or a new process, is first introduced by an innovator and eventually imitated by competitors. Imitators copy the innovation and in so doing enter the market and reduce the excess profits of the innovator. Imitation feeds diffusion and restores perfect competition. The adoption process, that is the mechanism and the duration of the time spell by means of which innovations are introduced and used by all prospective users, has been studied in great detail, and the notion of diffusion was eventually introduced. The economics of diffusion addresses relevant questions about the characteristics, the determinants and the effects of the adoption process. The most controversial issue is why adoption is not instantaneous and all firms do not adopt the innovation at the same time (Stoneman, 1976, 1983, 1987). The analysis here concentrates on adoption and diffusion of new technologies in the business sector, and does not refer at all to the adoption of new products by households, since the decision problem in the case of households differs significantly from that of firms. When households are considered, the innovation under scrutiny can only be a new product. When firms are the potential adopters and imitators, the innovation can concern the full Schumpeterian range of innovations. Adoption consists in the purchase of a new capital good, a new intermediary input or a new organisational procedure that has been supplied by upstream producers. Imitation consists of the replication of new conduct, a product, a process, a market or an organisational procedure first implemented by another firm. The adoption of a new capital good can be the result of the imitation of a process innovation. Imitation, defined as a form of herd behaviour, however is only one of the many possible causes for delayed adoption. Much work has been going on to identify possible factors for delayed adoption on either the demand or the supply side. Other relevant factors include: 1) the reduction of information costs; 2) network externalities; 3) irreversibilities and sunk costs; 4) changes in factor
318 Introduction of localised technological change markets; 5) the decrease of extra profits and hence market prices; 6) the reduction in production costs associated with learning processes or increasing returns and hence the reduction in the market prices; and 7) the introduction of incremental innovations that implement the original innovation so as to better satisfy the needs of additional groups of adopters. Let us analyse in more detail these approaches, with closer attention to the analysis of the dynamics mechanisms at work and the underlying assumptions. When the drivers of the dynamics are found on the demand side, diffusion, here, is defined as the process of delayed adoptions and imitations of a given innovation, with fixed economic characteristics, including the performance and the price, which takes place because of dynamics on the demand side in a population of heterogeneous agents. The well-known epidemic contagion provided the first and most famous frame for understanding the process: in a population of heterogeneous agents, characterised by information asymmetries and bounded rationality, adoption is driven by the dissemination of information about the effective profitability of adoption experienced by all those who have already adopted (Griliches, 1957). As soon as the information about the advantages provided by the innovation becomes available to the potential adopter, the adoption will take place. Diffusion, defined as a sequence of adoption lags, is fully explained by the characteristics of the spreading of the information. By the same token, technological resilience, i.e. the non-adoption, is simply the result of a lack of information (Mansfield, 1968). Technological resilience can be considered also the result of inappropriate levels of the profitability of adoption of a given technological innovation. The change of relevant conditions for the population of potential adopters however engenders an increase in the profitability of adoption and hence leads to the eventual diffusion. A first mechanism to explain diffusion in this approach, where the dynamics takes place on the demand side, but it is not reduced to the epidemic spreading of information, is provided by network externalities. The working of network externalities has a direct bearing on the profitability of adoption of a given innovation, when the number of its users has a direct bearing on its utility or efficiency (Katz and Shapiro, 1986). Network externalities, that is the effects of the stock of users upon the profitability of adoption, can be both positive and negative because of the effects of congestion. The effects of network externalities have been mainly appreciated with respect to final goods. However, network externalities can have a powerful effect both for final goods and for intermediary and capital goods. The profitability of the adoption of computers in the business sector, for instance, is greatly enhanced by the number of other computers in the network and the number of other firms that can receive, send and share information protocols, files and electronic communication at large (Antonelli, 1999a). In turn network externalities can be either direct, when there is a direct effect of the number of adopters of a technology on its own profitability of adoption, or indirect, when the number of adopters of another, yet related and complementary, technology has an effect on the profitability of adoption of the first technology (Smith, 2004). The understanding of the role of network externalities in grasping
Diffusion as a process of creative adoption 319 the dynamics of the adoption process in the business sector seems especially useful at a time when recent advances in the understanding of the exponential growth of Internet networks stress the role of key users or hubs as providers of positive incentives to enter the network. This approach makes it possible to appreciate the relevance of complex system dynamics to understanding the outcome of interactions where agents are heterogeneous also in terms of their size and the related extent of spillover of network externalities (Barabasi, 2002; Pastor-Satorras and Vespignani, 2004). An important engine of adoption can be provided by changes in the factor markets when technological change is biased and there is rivalry between old and new technologies. Here diffusion can be regarded as the outcome of the increase in the profitability of adoption engendered by changes in the factor markets of potential adopters. The profitability of adoption of a superior but biased technology is affected by the relationship between the factor bias and relative factor costs. All changes in factor markets, such as an increase in relative wages, have a direct, positive effect on the profitability of adoption of a more capital-intensive technology. When the new technology is both superior and biased, two equivalent isoquants, extracted from the two maps, overlap. In such circumstances three relevant events can take place: 1) there is a ratio of capital costs to wages for which the two technologies are equivalent; 2) small changes in the slope of the isocost can engender a radical shift with the sudden adoption of the new capital-intensive technology; and hence 3) there is a sharp discontinuity in the levels and rates of increase of the total factor productivity levels of adopters. If and when wages paid by firms are not identical, but distributed with a normal density function, and there is a historical trend of smooth rates of increase in wages, adoptions are likely to be distributed along a dynamic path characterised by three quite distinct regions: the first with a low level but a fast rate of increase, followed by a second where the adoption of averagewage firms takes place with a sharp discontinuity, and finally a region with a high level of adoptions but low rates of increase. An S-shaped process can easily approximate to such a process. The variance in the distribution of wages is determined by the combination of local differences in the bargaining power of trade unions and specialised workers and Marshallian heterogeneity among firms in terms of market power, profitability and age, and an interesting dynamic process can take place. Let us assume that profit-making firms can pay higher wages, above average levels, loss-making firms instead pay wages that are below the average, and the large majority of firms with normal profits pay average wages. In these circumstances profit-making firms are pushed to adopt new capital-intensive and superior technologies earlier than loss-making firms and hence to take advantage faster of higher levels of total factor productivity levels, with a sharp increase in their rates of growth. Such a discontinuity in productivity levels increases profitability and hence the likelihood of additional adoptions of new technological innovations. The initial conditions of heterogeneity are reinforced, and the variance in the population of firms is increased with a self-reinforcing mechanism (Antonelli, 2003a). The irreversibility of capital goods and the historical duration of their economic life provide a third important dynamic factor. The age structure of the stock of
320 Introduction of localised technological change capital goods of each potential adopter plays an important role in assessing the adoption rates. The sunk costs of past vintages of capital goods delay the adoption of new technologies until the variable costs of the production process with the old technology are lower than the total average costs obtained with the new technology. In this context however the rates of investment and more generally the rates of growth of each company have a strong positive effect on the rates of adoption. The expansion of the productive capacity makes it possible to adopt directly the new technology, while substitution is delayed by the effects of sunk costs. In these circumstances an interesting dynamic process can take place: fast-growing companies, in a dynamic macroeconomic context, have more chances to adopt in a timely way the new technologies and because of their timely adoption, and hence more efficient production processes, have more chances to grow faster. The interaction between growth, investment and adoption is likely to engender a strong reinforcing mechanism (Antonelli et al., 1992; Antonelli, 1993). In the supply side approach, heterogeneity of potential adopters consists in their cost conditions (David, 1969; Metcalfe, 1981). Diffusion is now defined by the structure and the sequence of delays in the adoption of a family of closely related technologies with changing economic and technical characteristics, rather than a single and given technology with static features. Potential adopters can be ranked in terms of cost characteristics. Diffusion here is driven by the dynamics on the supply side and specifically by the introduction of an array of events, including: 1) incremental changes in the prototype introduced by the innovator and/or by imitators; and 2) the decline of the market price due to a) the entry of new competitors and the decline of market power and hence mark-up for early innovators and b) the positive effects of increasing returns associated either with sheer economies of scale and density or with the dynamics of learning by doing. The sequence between the introduction of product innovations and the eventual introduction of process innovation to manufacture the new products, articulated by Utterback (1994), also has a direct effect on the decline of the market price for the new products and hence on the increase of their profitability of adoption. Both the decline of the market price and the introduction of incremental innovations can be seen as the effect of the entry of creative imitators in upstream markets (Stoneman, 1995, 2002). In a complementary approach the reduction in the price of the new products and the increase in the scope of adoption is the result of the selection mechanism at work on the supply side. After the introduction of an array of competing product innovations targeting the same product market, a dominant design progressively emerges, with relevant cost advantages in terms of standardisation, specialisation and division of labour, economies of scale, economies of learning and density. Once again diffusion is driven by the dynamics of the supply side (Utterback, 1994). In a similar vein the analysis of the flows of generation technologies shows that often a certain vintage of a technology is superposed by a follow-up technology, for example, Internet-based e-commerce following EDI-based e-commerce, or flexible manufacturing systems (FMS) following flexible manufacturing cells (FMC). If the different vintages are conceptualised as ‘one technology’, the adoption
Diffusion as a process of creative adoption 321 process can be considered as the result of the entry, on the demand side, of new niches of potential customers, attracted by the increasing scope of application of the growing variety of specific applications and customised incremental innovations. Many efforts have been made to combine the supply and the demand side approaches into a single more comprehensive model. Much progress has been made possible by the insight of Metcalfe (1981), where the epidemic, demand-side mechanism is implemented by the shifting conditions on the supply side so as to define the traditional S-shaped process as the envelope of a double shift. More recently Karshenas and Stoneman (1992, 1995) have elaborated a flexible model able to encompass the broad range of possible dynamics that integrates in an equilibrium approach the effects on both the demand and the supply side. In this context the notion of increasing returns to adoption emerges as a key synthetic contribution. Increasing returns to adoption are found both on the demand side, in terms of processes of learning by using the new technology and network externalities, and on the supply side, in terms of processes of learning by doing and economies of scale in the production of the new technology. The negative elasticity of market price to the stock of adoptions, because of the effects of competitive entry and reduction of extra profits in upstream markets, contributes increasing returns to adoption for users. When increasing returns to adoption, on both the demand and the supply side, matter, small events, such as occasional adoptions or changes in the sequences, and the introduction of new standards, especially if they take place at the onset of the process, may have long-lasting, pathdependent effects on the eventual diffusion and especially on the outcome of the selection, in the marketplace, among competing and rival technologies (David, 1985, 1987, 1988, 1990). When diffusion concerns the adoption of an innovation in the business sector, hence by firms rather than by households, the role of adoption costs needs to be considered carefully. The identification of the role of adoption costs paves the way to the distinction between gross profitability of adoption and net profitability of adoption. Adoption costs are defined by the broad range of resource-intensive activities that are necessary to identify an innovation and adapt it to the existing production process. Adoption costs include the costs of search and adaptive research, the costs of scrapping the existing fixed production factors, the restructuring of the production and marketing organisations, the reskilling of personnel, the purchase of the capital good and intermediary input embodying the new knowledge, the purchase of patents and licences, and the costs of technical assistance. Net profitability of adoption is the result of the algebraic sum of the gross profitability engendered by the adoption of an innovation and the costs that it is necessary to carry out in order to identify, select and finally adapt the new technology to the existing production conditions. A closer look at the process by means of which adoption is made seems necessary. First and most important, the notion of induced adoption needs to be considered. The literature on diffusion assumes that firms are always and immediately ready to adopt an innovation as soon as they perceive it as profitable. No room is made for the search for information and more generally for the context in which decision
322 Introduction of localised technological change making takes place. In our approach, adoption is very much induced by a general context where firms consider that a change is necessary in order to meet their expectations and reduce the gap between facts and plans (Antonelli, 1990; Metcalfe, 2005). The adoption of a capital good or an intermediary input is not free, especially for firms. The adoption of a new technology is necessarily the end result of a broader process that includes a preliminary search activity, a comparative assessment and the substitution of existing items, such as capital goods in place, workers, suppliers, customers and other components of the current structure of the firm. Adoption can take place only when some changes and adjustments have been made to the original setting. Such changes affect both the good incorporating the innovation and the layout of the firm as it was before the introduction of the new technology could take place. Adoption can take place only when the profitability of the new layout is confronted with the previous and yields a positive result. This comparative assessment includes the costs of the anticipated scrapping of the existing capital goods and the effects of all the related changes in the investment conduct (Antonelli, 1993). In the context of an induced adoption approach, the dynamics of adoption costs, together with the changing levels of gross profitability of adoption engendered by the introduction of changes in upstream activities, has a direct effect on the net profitability of adoption. Net profitability of adoption is the true driver of the diffusion of innovation. The analysis of adoption costs provides fruitful insight into the understanding of both the actual determinants of adoption and the analysis of diffusion processes (Canepa and Stoneman, 2004). Recent empirical evidence shows that the adoption of an innovation requires the active participation of the firm and thus it is the result of an activity. The characteristics of adoption activity in turn are much closer to the traditional views about original research and development activities than is currently assumed (Antonelli, 1991; Stoneman and Toivanen, 1997; Arvanitis and Hollenstein, 2001). Consistently much empirical evidence confirms that firms engaged in research and development activities are more prone to adopt new technologies, and this seems more relevant when the technologies under scrutiny imply adjustments in firms’ production process, (Faria et al., 2002, 2003). The adoption of a new technology is part of a broader process of technological change. Firms are reluctant to change their technology and are induced to introduce new technologies only when a clear inducement mechanism is put in place. As soon as the routines in place and hence the technology currently in use are questioned, and the inducement mechanism has been initiated by some mismatch between plans and facts, the choice between the introduction of original, invented-here technologies and the adoption of not-invented-here technologies can take place.1 The introduction of all kinds of technological changes by a firm is the result of a range of complementary activities that can be substituted only to a limited extent. At one extreme of the spectrum, technological change is the result almost exclusively of the generation of original knowledge and the novel introduction of a production factor never seen before as such. At the other extreme of the range,
Diffusion as a process of creative adoption 323 there is the traditional passive and imitative adoption where the firm limits itself to purchasing a good incorporating an innovation. The wide gulf of intermediary positions deserves much closer attention. This is the region where creative adoption takes place (Teece, 2005). The economics of localised technological change provides an appropriate analytical context for understanding the mechanisms at work in the case of creative adoption.
3 The role of external knowledge and supply of new technologies With respect to the benchmark model presented in Chapter 13 and along this line of inquiry, important progress can be made when localised technological change is seen as the result of creative adoption, that is the combination of internal competence and knowledge with the external knowledge embodied in capital goods and intermediary inputs provided by upstream suppliers or available in the form of technological information, licences and patents. The introduction of a new technology is induced by the mismatch between expectations and actual market conditions, and the irreversibility of production choices that have been made. The firm initiates a combined process of search and research. All opportunities to change the existing map of isoquants are now considered. The introduction of a brand new technology requires research efforts. The adoption of a new technology into the production process of a firm requires that some efforts to adapt it to the local conditions be made. The combination of the two activities yields the creative adoption of an existing technology, to which a number of changes are made so as to make it more consistent with the specific requirements of the existing production process and hence to reduce the amount of switching costs. The choice set is now framed. The firm faces two nested frontiers of possible changes in order to solve the mismatch between plans and real market conditions. The first frontier of possible changes is the frontier of possible adjustments which makes it possible to compare the results of resources invested in either technical changes or technological ones. The second frontier, the frontier of creative adoptions, compares the kinds of technological change. It defines a range of possible technological changes, all stemming from creative adoptions. The range is contained between the two extremes of a brand new technology: fully original and the ‘passive’ adoption of an external technology. The frontiers, the frontier of possible changes and the frontier of creative adoptions, have the usual concave shape that reflects the effects of diminishing returns in either activity. The shape is defined by the relative efficiency of the activities being considered (see Figure 17.2). The position of the frontier of possible adjustments is defined by the amount of resources (R) that the firm should invest just to switch from the previous equilibrium technique to the new one. The search for the correct solution in other words is identified as a maximisation process where the firm tries to maximise the amount
324 Introduction of localised technological change of changes, including technological innovations, that can be generated with a given amount of resources set by the levels of switching costs.2 With respect to the benchmark model presented in Chapter 13, here we assume that the firm can identify the correct solution by means of the standard maximisation of the output, for two given nested frontiers, when two nested isorevenues are defined. The absolute levels of the revenue generated by all adjustment activities consisting in the revenue made possible by the introduction of new techniques and the revenue made possible by the introduction of the new technologies respectively define the first isorevenue. The second isorevenue measures the bundle of revenues generated by more-or-less creative adoption of existing technologies, that is either the original – innovation – or the passive adoption. Formally we see the following relations: (1) TC = a(R) (2) SW = b(R) (3) OI = c(R) (4) PA = d(R) where TC measures the amount of technological innovation necessary to change the technical space that the firm can generate taking into account the internal
Original innovation (OI)
Technological change (TC)
Resources (R)
Resources
Passive adoption (PA)
Technical change (SW)
Resources
Resources
Figure 17.1 The production of technological change, original innovation, passive adoption and technical change with a given amount of resources
Diffusion as a process of creative adoption 325 competence and knowledge accumulated and the external knowledge it can access; SW measures the amount of technical change necessary to move in the existing technical space and reflects the levels of irreversibility and rigidity of tangible and intangible capital; OI measures the amount of original innovation; and PA measures the amount of passive adoption that can be generated with a given amount of dedicated resources (R) defined by the amount of switching activities the firm needs to complete to move from one equilibrium point to the other.
Technical change (SW) E
OIE
PAE Technological change (TC)
E Original innovation (OI)
Passive adoption (PA) Figure 17.2 The nested frontiers of possible adjustments and creative adoptions
326 Introduction of localised technological change It is clear that the relationship between the four production activities is essential in defining the outcome of the search process initiated by the changes in the product and factor markets. It seems clear that, the greater the efficiency in the production of technological changes and the lower the efficiency of switching are, the greater is the amount of innovations introduced. Correspondingly, the smaller the efficiency of internal research activities and hence the smaller the amount of original innovations and the smaller the efficiency of the adaptation activities are, the smaller will be the amount of innovations each firm will generate. The firm will adjust to the new factor and product market conditions more by means of switching activities than by means of creative adoptions. The extent to which the firm will rely on levels of creative adoption closer to passive adoption or will try to introduce original innovation, still based upon some levels of technological blending and recombination, clearly will be influenced by the relative efficiency of the activities and by the shape of the relevant isorevenue. To make this point more compact, let us now assume that a frontier of possible adjustments can be considered such that, for a given amount of resources (R) necessary to face the mismatch, firms can generate an amount of either technological change (TC) or technical change (SW). Nested to the frontier of possible adjustments we find a frontier of creative adoptions that can be obtained with the introduction of either original innovations (OI) or passive adoption (PA). Specifically, the shape and the slope of the frontier of creative adoptions reflect the effects of the technological opportunities based upon the localised competence built by means of internal learning by doing and the opportunities offered by the knowledge and the technologies generated by third parties that become available either by means of imitation or by the active push of upstream suppliers. Formally this amounts to saying that: (5) SW = e(TC) (6) OI = f(PA) In order for standard optimisation procedures to be operationalised, two isorevenue functions need to be set. The first, defined as the revenue of adjustments (RA), compares the revenue that adjustments by switching in the technical space (SW) yield with respect to the revenue of technological change (RTC). The second isorevenue includes the revenue generated by the introduction of original innovations (OI) and the revenues generated by the passive adoption of innovations and knowledge generated elsewhere (PA). Formally we see: (7) RA = sSW + tTC (8) RTC = rOI + zPA where s and t measure the unit revenue of switching and the unit revenue of technological change; and r and z measure respectively the unit revenue of the
Diffusion as a process of creative adoption 327 amount of original innovations and passive adoption of external technologies and knowledge respectively generated with the given amount of resources available to face unexpected changes in product and factor markets and the equilibrium amount of resources that can be identified to fund the introduction of technological change. It seems clear that the slope of the isorevenue of creative adoptions exhibits the greater unit revenue stemming from the introduction of original innovations. This makes it possible for the firm to command monopolistic market power and hence extra profits. By the same token however it should also be clear that the shape of the frontier of creative adoption should reflect the larger output – for given levels of inputs – in terms of adoptions with respect to the output in terms of introduction of original innovations: passive adoption is easier than the introduction of original innovations. The system of equations can be solved with the standard tangency solutions so as to define both the mixes of creative adoptions which in each specific context firms are advised to select and the amount of technological change with respect to switching that the context suggests selecting. The system of equilibrium conditions is: (9) e′(TC) = t/s f′(PA) = z/r subject to R = RF.3 The cases of either only technical change or only technological change and in turn either fully original innovations or fully passive adoptions are extreme solutions. Much of the real world can be found in between such extremes. Firms are induced to innovate by the mismatch between actual and expected conditions of their production set and their market conditions, necessarily built upon irreversible decisions taken on the basis of myopic expectations which are not met by the disequilibrium conditions in product and factor markets. The type of technological change is influenced by the relative net profitability of the introduction of original innovations with respect to the passive adoption of external technologies. The slope of the innovation isorevenue reflects the relative gross profitability of introduction of invented-here technologies with respect to the gross profitability of adoption of technological innovations introduced elsewhere. According to the shape of the innovation isorevenue, both the composition of technological change, whether it consists mainly of innovations or adoptions, and the mix of possible changes, whether they consist mainly of switching activities or technological changes, are affected. The equilibrium conditions identified by equation (9) capture the essence of the dynamics of localised technological changes consisting of creative adoptions engendered by the mismatch between plans and actual factor and market conditions for firms that are constrained by the irreversibility of their choices.
328 Introduction of localised technological change
4 Creative adoption and the diffusion of innovations The dynamics of creative adoptions is able to accommodate the traditional S-shaped aggregate diffusion process provided that a set of conditions applies. It is sufficient to assume that at each point in time the stock of adoptions exerts two quite distinct externalities. The stock of creative adoptions is likely to exert a negative effect on the gross profitability of adoption. The relationship is shaped by diminishing returns: the gross profitability of adoption, as determined by the market price for the products of the firm, is higher with low levels of adoptions and declines with the increase in the stock of adopters impinging upon the basic technology until it reaches a minimum level. The rationale for this effect is easily found in the typical Schumpeterian competition as a dynamic process. Early adopters can command extra profits associated with the creative implementation of the new technology. Eventually, however, as the number of adopters increases and the rivalry among users of the new basic technology becomes stronger, the market prices for the products manufactured with the new technology are driven to their minimum level and the conditions for perfect competition are finally restored. The understanding of this dynamics is the direct result of the new approach to adoption as the result of a creative and innovative process. There is in fact a continuum of conditions between the first producer of a new good and its first adopter which cannot be cut: early adopters are able to command transient extra profits like early innovators. Like early innovators, early adopters experience the decline in extra profits associated with the increase in the stock of adopters. This is not, however, the single effect of the stock of adoptions. The stock of adoptions is likely to exert also a negative effect in terms of a decrease in unit adoption costs. Here an array of positive effects is at work, including learning processes and increasing returns to scale. Moreover in upstream markets the entry of new competitors is likely to reduce the market prices for the basic technology to be adopted and creatively implemented with the internal and local competence of each firm. The market prices for the basic technology and the capital goods that embody the new knowledge decline as the number of adoptions increases. It is sufficient that the combined outcome of the two external effects respect a number of simple conditions for the net profitability of adoption to follow a welldefined path that is able to generate an S-shaped process. Specifically, as Figure 17.3 shows, the difference between the negative effect of the number of adopters (N) on adoption costs, on the one hand, and their negative effects on the gross profitability of adoption, on the other, can be isolated and directly confronted. The difference in their slopes and specifically the ratio of the values of their first and second derivatives are crucial. It is sufficient that the difference between the two slopes presents a combination of values that engenders a quadratic relationship of the net profitability of adoption with respect to the stock of creative adoptions, such that V1 < V2 > V3, to obtain a typical S-shaped diffusion process. Let us put it formally:
Diffusion as a process of creative adoption 329
Net profitability of adoption (V) V1 Gross profitability of adoption (GPA)
V2 V3 (AC)
Adoption costs (AC)
(GPA)
N
dN
N
N
t
Figure 17.3 The dynamics of gross profitability of adoption, adoption costs and net profitability of adoption and the S-shaped diffusion process
(10) GPA = m(N), with m′(N) < 0, m″(N) > 0 (11) AC = n(N), with n′(N) < 0, n″(N) < 0 (12) V(N) = m(N) – n(N) s.t. m″ > n″ Let us recall that, at as long as net profitability of adoption is found, the number of new adopting firms increases. Hence: (13) dN(t)/dt = W(V(N))
330 Introduction of localised technological change Given the properties of W and V(N) it follows that: (14) N(t) = Et(dN(t)/dt)dt = Et W(V(N))dt Equation (14) establishes a functional relationship between the flow of adopting firms and the stock of adopters. The p(N) function is S-shaped and has got an inflection. Therefore a functional form that is compatible with the specific conditions is: (15) N(t) = α1/1–e–kt, where k measures the speed of the process Equation (15) has its solution in the standard logistic function. The interpretative framework implemented so far is consistent with the empirical evidence. As is well known, much empirical evidence suggests that the time profile of the diffusion of a technological innovation and a family of closely related technological innovations can be easily approximated by a logistic distribution which exhibits a long phase of slow progress, a period of fast adoption by new firms and eventually a stretched period of approximation to the asymptotic levels of saturation (Stoneman, 1983, 2002). The process of diffusion of a new technology can now be considered as the rational result of the dynamics of localised creative adoptions engendered by the continual mismatch between plans and actual market conditions. The rate of the diffusion will be influenced by the dynamics of adoption costs and gross profitability of adoption, but also by the dynamics of localised technological change. When the mismatch between plans and actual market conditions is wide for many firms, when the effects of irreversibility are strong and hence switching costs are relevant, and when technological opportunities are attractive, the inducement to change the technology will be stronger and hence there will be an incentive for creative adoptions. Innovation and adoption are likely to feed each other so as to be complementary aspects of a broader dynamic process. The greater the number of firms is that do change their technology by means of varying degrees of creative adoption, the wider the mismatch between plans and actual market conditions is likely to be and hence the stronger the incentive to adopt and innovate is likely to be. The diffusion process of a given family of technologies will be faster, for given dynamic paths of its gross profitability of adoption and adoption costs, the greater are not only the net profitability of adoption but also and primarily the incentives to change the technology in place. The variance in the speed of the diffusion process of a given family of innovations, across countries and regions, with similar structural characteristics and a similar distribution of asymmetries among firms in terms of cost conditions and access to information, can be explained by the variance in the levels of entropy in factor and product markets and hence in the strength of the levels of inducement for firms to change their technologies. The higher the entropy is and the stronger the incentives to change their technology are, the faster is the diffusion of a given
Diffusion as a process of creative adoption 331 family of technologies. The localised introduction of new technologies will take place also by taking advantage of the adoption of available innovations. This model provides an analytical account which is consistent and compatible with the ‘encompassing model’ proposed by Karshenas and Stoneman (1995). Karshenas and Stoneman elaborated an equilibrium model of diffusion able to take account of both demand- and supply-side factors and their interrelatedness. Like the model presented here, their encompassing model is very flexible and incorporates several types of costs of new technology. At one extreme is purchasing a capital good only; at the other end is a whole set of factors such as costs of learning, switching, upgrading human capital stock and changing the organisation. This model in other words elaborates a frame for understanding the decision making of firms which is likely to generate expected dynamic behaviours that are well represented by the variety of cases integrated by the flexible model of Karshenas and Stoneman (1992, 1995).
5 Conclusion The economics of localised technological change provides a context in which the adoption of new technologies can be considered as the result of an active and intentional undertaking by firms. The adoption of a new technology is the result of a complex process where an inducement mechanism has to be identified, specific activities have to be put in place, and dedicated resources have to be committed. The adoption of a new technology requires a clear effort to adapt it to the preexisting context. There is no adoption without adaptation. In turn, such an adaptation requires considerable levels of competence and creativity. At the same time the introduction of a new technology is always the result of the blending and recombination of elements of technological knowledge both as goods and embodied in capital goods and intermediary inputs, organisational procedures and routines introduced elsewhere. Each innovation builds upon previous innovations. Technological knowledge and technological change as a consequence exhibit strong elements of ‘cumulability’, and both are the result of the incremental introduction of changes added on to previous advances. If there is no adoption without adaptation, it is also true that there is little innovation without some adoption. The economics of localised technological change provides a context in which the inducement to introduce technological changes is the result of the creative reaction of firms exposed to an increasing gap between expectations and actual conditions of profitability. When technology has to be changed, because switching costs impede standard shifts in the existing maps of isoquants and performance is falling below the expected levels, firms can rely on their competence and the knowledge acquired by means of research and development activities carried on intra muros. External sources of knowledge and new technologies embodied in new capital goods and intermediary inputs however do provide essential inputs to the introduction of new technologies by each firm. The introduction of technological change is the outcome of a process of creative adoption where external knowledge
332 Introduction of localised technological change and new technologies made available in the markets are recombined with the knowledge generated internally by means of learning processes and research and development activities. The traditional divide between innovation, adoption and diffusion can be successfully questioned in the context of the economics of localised technological change. Firms are induced to change their technology when product and factor market conditions do not meet their expectations and irreversible choices make adjustments expensive. Technological change is the result of the combination of research and search activities that lead both to the introduction of new technologies and to imitative adoptions. Both command resources and engender specific revenues. Localised technological change consists of creative adoption where external knowledge and embodied technologies are implemented with internal competence and idiosyncratic knowledge acquired by means of learning processes. The identification of the net profitability of adoption as defined by the gross profitability of adoption minus adoption costs contributes the economics of technological change. The analysis of the evolution of the net profitability of adoption in the context of the economics of localised technological change shows that the dynamics of creative adoption is able to generate an S-shaped diffusion path at the aggregate level. The divide between innovation and adoption is less and less realistic at a time when general-purpose technologies (Helpman, 1998), such as new information and communication technologies, characterise the rate and direction of technological change. New information and communication technologies with high levels of fungibility characterise the present trend of innovation at the aggregate level. In this context, firms, induced to change their technology by the dynamics of localised technological change, make use of the fungibility of the new technological system and enter a process of creative adoption. Adoption and innovation are two complementary aspects of a broader process of reaction to the mismatch between expectations and facts and eventual introduction of localised technological changes that build upon the creative adoption and recombination of internal and external technological knowledge. The distinction between gross and net profitability of adoption and the identification of the costs of adoption together with the grasping of their dynamics, including the effects of the stocks of adoption on the evolution of the net profitability of adoption, provide an analytical probe that combines the demand and supply tradition of analysis of diffusion and shows the complementarity between innovation and adoption within the context of the economics of localised technological change.
18 Path dependence and the quest for complexity
1 Introduction The theory of complexity is emerging as a new unifying theory for understanding endogenous change and transformation across a variety of disciplines ranging from mathematics and physics to biology. The theory of complexity favours a systemic and dynamic approach: systemic in that the outcome of the behaviour of each agent and of the system into which each one is embedded can only be understood as the result of the interaction between micro- and macrodynamics; and dynamic because it makes it possible to understand the structural change that stems from the interaction of agents. A theory of economic complexity can be built upon a number of basic assumptions: 1 2
3 4
5
6
Heterogeneous agents. Agents are characterised by distinctive and specific characteristics as well as being intrinsically heterogeneous. Location matters. Location in a multidimensional space, in terms of distance among agents and their density, matters and influences both behaviour and performance. Local knowledge. Each agent has access only to local information and local knowledge, i.e. no agent knows what every other agent knows. Local context of interaction. Agents are localised within networks of relations, including transactions and feedbacks, which are specific subsets of the broader array of interactions that define their behaviour. Creativity. Agents are creative, i.e. agents can follow some rules but they can also change the rules. They do this in response to given feedbacks, according to both their own specific characteristics and the features of local endowments, including the network of transactions and interactions in which they are embedded. Systemic interdependence. The outcome of the behaviour of each agent is strictly dependent on the web of interactions which take place within the system. Hence, at each point in time, the topology of the system, i.e. how the characteristics and the structural interactions of the agents in their relevant multidimensional spaces are distributed, plays a key role.
334 Introduction of localised technological change The dynamics of complex systems is characterised by: 1) non-ergodicity: a little shock at one point in time affects the long-run dynamics of a system; 2) phase transitions, consisting in the occurrence of qualitative changes determined by small changes in the parameters of the system; and 3) emergent properties: the properties of a system that apply at a specific level of aggregation of a system (Arthur et al., 1997; Taylor, 2001; Barabasi, 2002; Wolfram, 2002; Durlauf, 2005). Complexity economics can retain much of traditional economics provided it is able to accommodate the intrinsic dynamic capabilities of complex dynamic system analysis. The integration of the rich and elaborated competence of economics in dealing with systemic analysis, although in a static context, can draw on complex system dynamics, especially when the role of historical time and the intentional behaviour of rent-seeking agents are taken into account and when an understanding of the economics of innovation is integrated. The capability for systemic analysis has always been regarded as one of the main pillars of strength of economics as a science. Ever since its conception, economics has been fully aware of the complexity of the relations that characterise economic interactions. An understanding of the mutual interdependence between workers, consumers, producers, savers and investors lies at the core of economic analysis. This translates into a professional competence based on an understanding of the laws which govern the close interrelation between markets for products and markets for inputs, including monetary and financial markets, between investments and savings and between production and consumption, all actions carried out by a myriad of agents. The introduction of general equilibrium analysis has brought the systemic capabilities of economics to a remarkable level of sophistication. it is at great cost, however, for there is a total lack of dynamics in such a system. In traditional general equilibrium analysis of future prices, there is a far-sighted representative agent who is able to foresee all the possible future prices and to act accordingly with a hyper-rational selection of a trajectory of techniques which can take account of and make the best use of the factors of production according to their present and future price. When, according to Milton Friedman, a variety of agents is assumed, stupid agents can be admitted, provided that perfect competition is at work, a single attractor exists and one best technical combination can be identified. Market forces will take care of selecting from among a cohort of agents. Those who, even by chance and ignorance, have made the correct choice will survive. Competition will sort out those who made the wrong one.1 As soon as the rules change because of exogenous causes, agents can adapt to them, but they cannot generate them internally. By the same token, as soon as changes take place in consumer preferences or in the levels of endowments, or in technology, agents will adapt to them, but they cannot generate or shape them. The economic agent portrayed in general equilibrium analysis is either far-sighted or stupid but always passive.2 However, general equilibrium analysis provides an extraordinary set of tools with exogenous production and utility functions to understand the mechanisms at work in the allocation of given resources among a given group of agents. Neoclassical economics has provided an elaborated and sophisticated framework
Path dependence and the quest for complexity 335 for understanding the conditions for static efficiency. In such a context, growth and development are the consequences of exogenous changes in the shapes of the utility functions, in the characteristics of the technology and in the demographic conditions, as well as the supply of natural resources. The theory of economic growth elaborated in such a context does not address the causes of growth. It is limited to analysing the complementary conditions in terms of rates of growth in the supply of labour and savings that make it possible for exogenous growth to take place. As soon as the assumptions about the exogeneity of production (and utility) functions are relaxed and agents are considered both intelligent and endowed with a specific form of reactive creativity which makes it possible to endogenously change the basic features of the utility and production functions and hence tastes, preferences, technologies and routines, the relevance of general equilibrium analysis declines. It is difficult to conceive a system of future prices which is able to take into account the introduction of all possible new technologies in a given time horizon. There is no longer a single attractor, as firms are now credited with the capability to generate their own technological knowledge and to change their technologies and not only to vary either the quantity they produce or the prices they charge. Economic systems are now considered as complex dynamic mechanisms which are able to grow and have differentiated levels of efficiency. In turn, such levels of efficiency are the outcome of the behaviour of the agents and of the structure of their relations, in that they have a differential capability to change the rules and the network of interactions. Hence, they are able to generate new technological knowledge and to introduce new technologies. An understanding of the working of the economic system as a complex dynamic process is possible as soon as the systemic properties that belong to economics as a science are extended so as to include the possibility of agents and subsystems internally generating new technological knowledge and hence new production technologies and new preferences. What is more, it is not difficult to do this, for most of standard microeconomics can be retained and properly implemented at the agent level: agents do try to optimise within the strong limitations of their subjective conditions. This seems to be the context in which the analysis of the conditions of dynamic efficiency can be considered so that it can become one of the key aims and scopes of contemporary work in economic theory. The merging of complex dynamic theory with a theory of the agent based upon subjective and procedural optimisation implemented by the necessary consideration of creative choices in a context characterised by intrinsic heterogeneity of firms can be productive both for economics and for building a more articulated theory of complex system dynamics.3 Economics can draw on a complex dynamic approach in the analysis of the causes and determinants of the action of agents who are aware of the implications of the key role of economic topology, i.e. the characteristics of the distribution of agents in the multidimensional space and of their structural relations. Creative agents in fact can do many things; they can change not only their technologies but also their location in the multidimensional space. Topological change is relevant both in terms
336 Introduction of localised technological change of the location of individual agents – with all the consequences in terms of creative capabilities that have been explored – and with respect to the overall features of the multidimensional space in which their action is embedded. Topology, that is to say the distribution of agents in space, plays a key role in complex dynamics. Proximity among agents and density in space are in fact a key attribute of complex dynamics. The key features of topology shape interactions. Economics provides the basic tool for a strategic notion of economic topology and topological action to be developed, and hence it is possible to analyse what causes and determines the intentional activity of agents when they move in the relevant space and when they change the given topology. Accordingly it becomes clear that topology does not change only through the stochastic process but also because of the intentional action of agents. The merging of complex dynamic and economic analysis can contribute to the understanding of the interplay between intentional economic action and the topological distribution of agents: the causes and the effects of the sequence of changes that alter the topology of the relevant spaces. Thus this approach may not be able to foresee the final destination of the process but can highlight the paths of change (Foster, 1997, 2005). The notion of path dependence elaborated by Paul David provides one of the most articulated and comprehensive frameworks to move in the direction of an analysis of the conditions that make it possible for an economic system to generate and exploit endogenous growth. Path dependence is an essential conceptual framework which goes beyond the analysis of static efficiency and enters the analysis of the conditions for dynamic efficiency. It applies both to each agent, in terms of the quasi-irreversibility of the agent’s own endowment of tangible and intangible assets, networks of relations in both product and factor markets, stock of knowledge and competence, and to the system level, in terms of general endowments of production factors, industrial and economic structure, and architecture of the networks in place. The identification and articulation of individual and system path dependence make it possible to retain the positive contributions of complex dynamic system methodology and at the same time to overcome its intrinsic limitations stemming from its origins built on natural sciences where human decision making is not considered. Dynamic systemic efficiency matters when the key economic problem is no longer regarded as the identification of the conditions that make it possible to get the best use of given resources, in terms of outputs, but rather the qualification of the conditions that make it possible to get the most effective generation of resources and hence of the output that can be obtained through their efficient allocation and mix. The notion of path dependence proves to be especially attractive for European economists brought up on a tradition that considers growth and change rather than equilibrium as the relevant object of analysis and hence values historical time and philological inquiry as basic tools for studying the dynamics of social events. The rest of the chapter is structured as follows. Section 2 explores in detail the notion of path dependence and identifies its basic ingredients. Section 3 shows how the different combinations of the basic ingredients lead to identifying different types
Path dependence and the quest for complexity 337 of path dependence, especially when the analysis is applied to understanding the introduction and diffusion of technological innovations. Individual and systemic path dependence are identified, their articulations are explored and the implications for the analysis of the conditions of dynamic efficiency are developed. The conclusion summarises this analysis.
2 Path-dependent dynamics According to Paul David, path dependence is an attribute of a special class of dynamic processes. A process is path dependent when it is non-ergodic and subject to multiple attractors: ‘systems possessing this property cannot shake off the effects of past events, and do not have a limiting, invariant probability distribution that is continuous over the entire state space’ (David, 1992: 1). Historical analysis and much empirical evidence in economic growth and specifically in the economics of innovation and new technologies confirm that these characteristics apply and are most relevant to understanding the laws of change and growth of complex systems. Path dependence provides a unique and fertile analytical framework which is able to explain and assess the ever-changing outcomes of the combination and interplay between factors of continuity and discontinuity, growth and development, hysteresis and creativity, routines and ‘free will’, which all characterise economic action in a dynamic perspective which is also able to appreciate the role of historical time. The notion of ergodicity deserves careful examination. A process is ergodic when its initial condition has no persisting and enduring influence on its development and eventual outcome. The general equilibrium framework of analysis is typically ergodic, although the analysis of the competition process and its building blocks such as the theory of costs and of the firm are based upon short-term conditions where some costs are fixed and their temporary irreversibility may have major consequences for the outcome of the interactions among firms in the marketplace. When a process is non-ergodic, initial conditions (and events that occur at early points in the path) typically exert strong effects on its development and on the final outcome. Past dependence or ‘historicity’ is an extreme form of non-ergodicity. Historical, as well as social and technological, determinism fully belongs to past dependence. Here, the characteristics of the processes that are analysed and their results are considered to be fully determined and contained in their initial condition. In the theoretical economics of innovation, this extreme (some would say degenerate) form of path dependence has often been assumed: the epidemic models of diffusion of innovations and the notion of innovations ‘locked in’ to a technological trajectory are typical examples of the deterministic representation of essentially stochastic technological and social phenomena. As such, these nonergodic models are analytically informative but empirically uninteresting. The process takes place within a single corridor, defined at the outset, and external attractors cannot divert its route, nor can the dynamics of the process be altered by transient random disturbances in its internal operations.
338 Introduction of localised technological change
Dynamic processes Non-ergodic processes
Ergodic processes
stochastically convergent
determistically convergent and globally stable
past dependent
path dependent
Figure 18.1 Typology of dynamic processes
Path dependence differs from deterministic past dependence in that irreversibility arises from events along the path, and it is not only the initial conditions that play a role in selecting from among the multiplicity of possible outcomes. The analysis of a path-dependent stochastic system is based upon the concepts of transient or ‘permanent micro-level’ irreversibilities, creativity and positive feedback. The latter self-reinforcing processes may work through the price system, or they may operate through non-pecuniary externalities. The conceptualisation of stochastic path dependence can be considered to occupy the border region between a view of the world in which history is relevant only to establish the initial conditions, after which the dynamics unfold deterministically. It is the conceptualisation of historical dynamics in which one ‘accident’ follows another relentlessly and unpredictably. Path dependence gives economists the scope to include in their conceptual framework the analysis of the effects and determinants of historical forces upon economic affairs without succumbing to naive historical determinism. The sequence of steps becomes a relevant issue in path dependence. At each step the direction of the process can be changed as a result of the influence of new events when selecting among the different attractors of the system. A full understanding of a path-dependent process requires a detailed analysis of the sequence of the steps which have been made and of the interactions between the effects of irreversibility, local externalities, creativity and feedback. Irreversibility pushes towards a trajectory where the initial conditions are replicated and determine its direction. Creativity, local externalities and feedback exert a diverting effect. The dynamic interplay between these elements shapes the characteristics of the process and its direction at each stage and has an influence on the subsequent stages. At each stage the balance between such dynamic forces may differ, and hence so will the direction of the process. In path dependence the sequential interplay between the past-dependent effects of irreversibility and the diverting effects of creativity, local non-market interactions engendering externalities and feedback plays a key role and qualifies the conditions of path dependence itself as an interface between ergodic and nonergodic approaches to economics.
Path dependence and the quest for complexity 339 The careful identification of the stages of each dynamic process and of the changes in their alignment, hence in the interplay between ergodic and non-ergodic forces, and the consequent localisation and explanation of the points of inflection, bifurcation or change in direction and intensity becomes a major task and a fertile area of inquiry. Irreversibility clearly is the prime ingredient and causal factor of the dynamics. Irreversibility consists in the lack of dynamic malleability of production factors as well as tastes and preferences, reputation and routines. When irreversibility applies, it is difficult and costly to change a given set of conditions. Irreversibility engenders specific costs: the costs of switching from one condition to another. Irreversibility provides the past dependence of the process. Quasi-irreversibility matters when initial conditions can be changed by means of creative reactions. Creative reactivity is the dynamic factor in path dependence. Path dependence assumes that agents are able to react to the changing conditions in their environment not only by adjusting prices to quantities and vice versa but also, and mainly, by changing their technology as well as their preferences and tastes. The contribution of prospect theory, according to which the more agents are exposed to frustration the less risk averse they are, is most relevant in this context (Kahneman and Tversky, 1979; Witt, 1998). Path dependence assumes that the conduct of agents themselves within systems is path dependent in that their knowledge, strategies and location in the architecture of the system are the result of a dynamic process which is influenced by the initial conditions and yet is open to a variety of local attractors that shape the characteristics of the process and its eventual outcome. Localised positive feedback and local externalities articulated by the structure of social, non-market, interactions are very important ingredients of path dependence. The outcome of the creativity of agents, affected by quasi-irreversibility and hence induced to innovate, is both shaped and magnified by the effects of local social interactions in terms of externalities and positive feedback. Proximity, as defined by connections and communication channels, qualifies the conditions for interaction and communication among learning agents, engaged in complementary innovation activities, in local contexts where each agent exerts some control over fragments of indivisible knowledge. The interactions with other complementary agents affect the outcome of the innovation process. The chances of introducing innovations that increase total factor productivity depend very much on the local efficiency of the innovation process, the conditions for accessing external knowledge and more generally the conditions for local knowledge communication. In turn, the profitability of adopting innovations is dependent on the idiosyncratic context of action of the agents in each subsystem. The architecture of the system in which such dynamics takes place is itself the product of the dynamic action of firms. Firms are able to change their technologies as well as their context of action: firms diversify, integrate downstream or upstream, can specialise in core business and in doing so change their boundaries by means of strategies of inclusion and exclusion of activities. Firms change their location in regional space by means of spatial mobility. Firms can change their connections
340 Introduction of localised technological change with customers, suppliers, rivals and research institutions. As a consequence, industrial structures change their shape and the division of labour is altered both within and among economic systems. The structure of relative prices of inputs is also affected by new technologies according to their bias and the derived demand. At this point, the major distinction between past dependence and path dependence must be stressed and highlighted so that the key role of knowledge can be understood better. Past dependence defines the elements of irreversibility that become a source of impediment to adapting passively to unexpected conditions in the market, make it difficult to alter the direction of the process and make it possible to learn only within a narrow space. Path dependence defines the results of the action of a variety of interacting agents bounded by irreversibility but able to collectively change their technology and to generate new knowledge.4 While strong irreversibility and internal learning exert a past-dependent, converging effect that keeps the process within well-defined corridors, local externalities and local feedback, and creativity, including the generation of collective knowledge in evolving knowledge pools, and induced changes in factor prices and in the industrial structure exert a path-dependent effect which may become diverging and push the process away from its initial direction. One of the most important results of the analysis carried out in this book is the identification of the conditions that favour the generation and dissemination of technological knowledge. This is done through an analysis of the relations between the laws governing the generation of new knowledge and the rate and direction of localised technological change. Knowledge is itself path dependent. Knowledge is the consequence of localised learning. Localised learning here is defined both in terms of the specific technical conditions upon which learning processes have been taking place within the firm and in terms of the specific location in the web of relations that characterise knowledge networks which define the conditions of access to external knowledge (Rizzello, 2004). Table 18.1 Path and past dependence Path dependence
Past dependence
Quasi-irreversibility
Internal learning
Internal learning
Irreversibility within the firm
Local externalities and feedbacks in changing knowledge pools
Irreversibility of the endowments
Knowledge governance
Given knowledge pools
Creativity
Given architecture of the system
Dynamic coordination Structural change Architectural change of knowledge networks Induced changes in factor prices
Path dependence and the quest for complexity 341 Learning is possible mainly, but not exclusively, when it is based on repeated actions in a well-defined contextual set of techniques and relationships in doing and in using. Hence, learning is mainly local in that it is limited by the scope and the perimeter of the expertise acquired within a given technology, productive conditions and organisational sets, including the wide array of relations and interactions which operate within knowledge networks. Knowledge cumulability shapes and limits the direction of learning and defines the opportunity costs of alternative directions of creativity. The cumulated effects of localised learning by doing and learning by using consist in fact in higher levels of competence and eventually the opportunity to generate new knowledge. All changes to a given set of techniques, production conditions and relationships are likely to engender relevant opportunity costs, in terms of missed opportunities for learning and hence for the acquisition of higher levels of competence and the generation of new technological knowledge (Antonelli, 1999a). Agents generate new knowledge by means of: 1) internal learning and R&D activities; 2) the previous vintages of knowledge accumulated within their organisations; 3) access to knowledge external to each agent, but internal to the local knowledge commons; and 4) the creation of new communication channels and new relations and hence limited mobility in the knowledge space. At each point in time an agent’s level of creativity depends on the interplay between the pastdependent effects of knowledge cumulability and the dynamics of positive externalities and feedback from agents engaged in complementary activities within knowledge networks. The specific characteristics of technological knowledge as a collective economic good have major implications here. Technological knowledge is a highly heterogeneous product characterised by varying levels of appropriability, excludability, compositeness, cumulability, stickiness and fungibility, in its production and in its use, which are more or less articulated in tacit or codified forms. Because of key complementarity between internal and external knowledge, technological knowledge becomes a collective good, available within well-circumscribed ‘commons’ characterised by effective systems of social (non-market) interactions (Guiso and Schivardi, 2007). This has major effects on the behaviour of firms that are able to access the technological commons. Proximity among firms in regional, product and technological spaces and hence the density and structural features – in terms of the architecture of communication channels and connectivity in general – are extremely important if all the technological spillovers and collective knowledge available are to be accessed. However, firms have to make a substantial effort to absorb the external knowledge. The mechanisms of knowledge governance and dynamic coordination play a key role in assessing the amount of technological knowledge each firm is able to generate. The levels of creativity of each agent are substantially shaped by the characteristics of the knowledge commons. Such characteristics in turn exhibit the typical features of quasi-irreversibility: at each point in time they are quasi-fixed, but in the long term they are changed by the intentional action of those agents who are endowed with procedural rationality
342 Introduction of localised technological change and thus are able to modify the architecture of relations which shape the generation of collective knowledge. The new understanding of the characteristics of the system of social interactions in which agents are embedded with respect to the role of knowledge communication and governance, which is described Part II, favours a closer relationship between the analysis of innovation and the analysis of diffusion. There are many common elements in the adoption of a new technology and access to external knowledge. The pervasive role of knowledge externalities leads to the description of the notion of the diffusion of innovative activity. At the system level, the extent to which creative reaction is possible and eventually able to generate sustained rates of introduction of new localised technologies can be regarded as the result of the spread of knowledge externalities. In this context, the identification of the types of external factors at play in dynamic processes is the key to understanding the differences between path-dependent innovation and path-dependent adoption. The identification of path-dependent adoption, as distinct from path-dependent innovation, becomes relevant in this context. Path-dependent innovation can be defined as the set of rules that make it possible to understand why firms are better able to innovate within a limited set of techniques. Path-dependent adoption instead is the area of explanatory analysis that makes it possible to understand why firms adopt some technologies, possibly in the proximity of the existing ones, rather than introduce possible technological innovations. More specifically, pathdependent diffusion assumes that technological innovations have already been introduced and that there is some rivalry and substitutability between them and those already in use, as well as among them. Path-dependent innovation stresses the importance of analysing social conditions for technological creativity. The notion of path-dependent innovation elaborated by Paul David (1975) applies when the analysis is concentrated on the characteristics of the process by means of which firms are induced to both generate new technological knowledge and introduce new technology. The notion of localised technological change contributes to the analysis of path-dependent innovation, as it identifies the conditions that explain the localised character of technological change, namely the mix of irreversibilities (including local endowments), induced innovation, creativity and local externalities. The driving force is clearly the mismatch between the irreversibility of the tangible and intangible stock of sunken inputs and the actual conditions in both factor and product markets, which are affected by the continual introduction of unexpected innovations into the system. Such a mismatch induces a creative reaction and the eventual introduction of new localised technologies. But the structure of feedback and interactions among agents in the markets for inputs and outputs shapes the levels and the magnitude of the chain reaction that takes place after the introduction of one innovation. In a loosely connected system the innovation of one firm can generate only a minor mismatch for a limited number of other firms. In a tightly connected system, the innovation introduced by one firm can induce a major mismatch for many other firms and hence set off a major chain reaction. Myopic but creative agents introduce technological changes that are
Path dependence and the quest for complexity 343 localised by their knowledge base and are built on localised learning processes. Switching costs stem from irreversible production factors and the external conditions of factor markets. While the rate of introduction of new technologies is induced by the mismatch between the irreversibility of production factors and the actual conditions of the markets, the direction of the new technologies is induced by the relative prices of production factors. The conditions of access to technological spillovers and the external knowledge available in the local pools of collective knowledge help to explain why the direction of the introduction of new technologies is localised along a well-designed technological, technical and industrial path. Such conditions of access deeply affect the results of the induced innovation activity of each agent and hence the incentives to innovate. Each agent’s level of creativity is determined by the social context of learning and specifically by the characteristics of the knowledge commons in which it is embedded. Wide knowledge commons, implemented by effective knowledge governance mechanisms, provide agents with the opportunity to build on localised learning and hence to be better able to implement the techniques originally in place. The quality of knowledge commons is the prime factor driving firms towards a creative reaction rather than passive adjustment to unexpected changes in the marketplace. An understanding of the role of technological externalities in the generation of new technological knowledge and in the introduction of new technologies makes it possible to stress the role of systemic path dependence. Firms searching for a possible reaction to unexpected events help each other with localised spillovers and reciprocal knowledge transfer that build on a local knowledge commons based on competence and experience acquired in learning by doing, learning by using and learning by interacting. The direction of the process of each agent introducing new technologies here is seen as the result of the collective knowledge available locally. The notion of path-dependent adoption elaborated by Brian Arthur (1989) and Paul David (1985) contributes to the analysis of the diffusion of new technologies. New technologies are sorted out mainly by the effects of increasing returns to
Table 18.2 Types of path dependence Path-dependent innovation
Path-dependent adoption
Factors internal to firms
Switching costs Localised learning Complementarity of internal and external knowledge
Complementarity of new and old technologies Increasing returns to adoption
Systemic factors
Localised knowledge commons Knowledge communication Knowledge governance mechanisms Architecture of the system
Adoption externalities Composition effects and relative factor prices Reduction in the hedonic prices of the new product
344 Introduction of localised technological change adoption at the system level. In such an analysis the selection and eventual diffusion of new technologies are path dependent in that they are influenced by the timing of their sequential introduction, which in turn affects the relative profitability of their adoption as shaped by the powerful consequences of positive feedback which consists in the interplay between adoption externalities on the demand side and increasing returns in production. Here the choice of the new technology is shaped by conditions in the factor and product markets. The diffusion of new and better technologies can be delayed or barred by lack of compatibility with the internal and irreversible characteristics of potential adopters. Here internal factors play a key role in explaining path-dependent diffusion. The durability and irreversibility of internal factors such as the capital stock, but also the skills of human capital and the location in a given space, as well as relationships, will determine the adoption of new technologies with customers and suppliers. Irreversibility of tangible and intangible production factors and the consequent switching costs exert important pressure on the diffusion of rival innovations. New technologies are sorted out not only by their absolute levels of efficiency but also with respect to their complementarity and compatibility with the installed stock of fixed and irreversible production factors. Firms with an important stock of fixed capital and employees whose competence is irreversible, and which attach great value to their customer base and to the relationships with those who provide intermediary inputs will select the new technology, not only because it is more productive but also because it is more compatible with and easier to integrate into the existing production process and also fits in with the network of relations already in place. Internal factors here play an important role in explaining path-dependent diffusion. Increasing returns to adoption are at work here on both the demand side and the supply side. The greater the stock of adopters is, the lower are: 1) the production costs, because of the effects of increasing returns in the production of the new product; 2) the levels of the mark-up, because of the larger flows of entry of imitators and competitors in the supply of the new product; and 3) the efforts that are necessary to use the new good properly, owing to the effects of social learning to use it. The ability of users to assess the quality of a new product is strongly influenced by the size of users. Moreover, the effects of shared use intrinsically shape the position of many products on the quality ladder: again the size of the stocks of users in place affects the height of the ladder, as perceived both by potential consumers and by incumbent users. Local factor markets are major external factors in path-dependent adoption. The productivity and cost-effectivity of a new technology are influenced by composition. Composition effects are the outcome of the sensitivity of output to the relative scale of each single factor, rather than to the scale of the bundle of production factors. They are positive when the relative scale of most productive factors is augmented and that of the least productive factors is reduced. In general, because of composition effects, the greater the productivity is of the factor which is more widely used, the lower the productivity is of the factor which is less used and the greater are the effects of any changes in the relative levels
Path dependence and the quest for complexity 345 of factor costs. When the most productive factor is cheaper, and hence its use is more intensive, and the least productive factor is most expensive, and hence its use is least intensive, production costs are lowest. The growth of total factor productivity stemming from the introduction of a given technology is higher, the higher is the output elasticity of the productive factor which is locally most abundant. The irreversible features of the economic systems also shape path-dependent adoption where firms are embedded. Local factor markets are major external factors in path-dependent adoption. Composition effects influence the productivity and cost-effectivity of a new technology. Composition effects are the outcome of the sensitivity of output to the relative scale of equilibrium use of each single factor. They are positive when the relative scale of equilibrium use of the most productive factors is augmented and that of the least productive factors is reduced. In general, because of composition effects, the greater the productivity is of the factor which is cheaper and hence more widely used, the lower is the productivity of the factor which is more expensive and hence less used and the greater are the effects on profitability of adoption of new technologies. The profitability of adoption of a new technology is highest when the most productive factor is relatively cheaper and hence its use is more intensive, and the least productive factor is relatively more expensive and hence its use is less intensive. The growth of total factor productivity stemming from the introduction of a given technology is higher, the higher is the output elasticity of the productive factor which is locally most abundant. Composition effects act as sorting devices. For a given supply of new and rival technologies, with similar shift effects, composition effects act as powerful selection devices, and the diffusion of technologies will be influenced by the local conditions of factor markets. Labour-intensive technologies will diffuse faster in labourabundant countries, and capital-intensive technologies will be adopted faster in capital-abundant technologies. The adoption of new technologies that are characterised by high levels of output elasticity of labour but small shift effects might be delayed forever in capital-intensive countries. This analysis is most important when the global economy is considered: in the global economy firms based in highly heterogeneous local factor markets compete in quite homogeneous product markets. Composition effects favour the adoption of technologies that are characterised by high levels of output elasticity for locally abundant production factors (Antonelli, 2003a). Too much emphasis has been put on the effects of path-dependent diffusion in terms of ‘lock-in’. Technological lock-in is a possible outcome of path-dependent diffusion, although new waves of better technologies, possibly introduced by competitors, may eventually break the technological resilience. The real key point is in fact not the ‘lock-in’ effects but rather the ‘lock-out’ ones. In path dependence, firms need to change their current set-up because of unexpected events which they cannot cope with (they change by means of traditional price–quantity adjustments), their action being due to irreversibilities and constraints, on the one hand, and the opportunity to introduce new technologies, on the other. Such dynamics are fuelled by irreversibility and are shaped by the
346 Introduction of localised technological change changing effects of local externalities and feedback, within a defined path. The conditions that determine the generation and spread of collective knowledge and the knowledge governance mechanisms at work are key components of the systemic process that shapes the path-dependent evolution of an economic system.
3 The engine of growth: collective knowledge, localised technological change and path dependence The different forms and factors of path dependence that have been identified so far can be considered as complementary components of a broader dynamic framework which make it possible to understand the engine of growth and the conditions for dynamic efficiency. Path dependence includes the economics of localised technological change and collective technological knowledge and provides an initial, and yet quite elaborated, framework for addressing the key issues in the analysis of the conditions for dynamic efficiency. In so doing path dependence provides a framework which is not necessarily in conflict with a static analysis and more generally with the quest for static efficiency. Path dependence addresses a different set of problems and analyses a different set of conditions (David, 2007). In the standard static framework, growth is exogenous and ergodic. In such a context, growth takes place when and if exogenous changes concerning the technology, preferences and distribution of natural resources and demography occur. The direction of change can be either positive or negative: the system is able to automatically adjust to any new set of exogenous conditions. Such a context does not change even when assumptions about the variety and heterogeneity of agents are taken into account. Low levels of irrationality, at least on the supply side, are also compatible. The irrational behaviour of firms will lead to suboptimal performances, and they will eventually be to forced exit. Rational behaviour will lead to survival in the marketplace. The notion of Brownian movement has been successfully borrowed from physics to characterise such a microdynamic context.5 Perfect and unconstrained mobility of firms in the product and factor markets with no barriers to entry and exit and no switching costs, as well as changes in the levels of output and possibly the exclusion of suboptimal behaviour, is the prime engine of such a Brownian movement. New firms enter the market attracted by the gap between prices and costs, and firms leave the markets when the prices fall below costs. Firms are not supposed to be able to change their technology and hence their production functions. At best, these firms are able to influence the position and slope of supply curves by entering or leaving the market. In the approach elaborated in this book, an approach which builds on Paul David’s path-breaking contributions, growth is endogenous and path dependent. Growth is primarily the result of the endogenous changes in technologies and tastes, hence in production and utility functions, which take place because of the agents’ creativity and reactivity. Agents are characterised both by substantial levels of irreversibility and by high levels of creative capability. Irreversibility exposes agents to substantial losses and rigidities when their plans are not fulfilled and their expectations are
Path dependence and the quest for complexity 347 not realised. Agents, however, are able to elaborate with procedural rationality and creativity, including mobility, albeit in the immediate surroundings limited by switching costs and bounded rationality (Garud and Karnoe, 2001). Creativity makes it possible to consider, alongside entry and exit and changes in output levels, the introduction of innovations, as the other possible reaction to a mismatch between expectations and the actual conditions in the factor and product markets. Creativity can blossom only in a conducive context characterised by highquality knowledge governance and hence a proper mix of the incentives to generate knowledge and to spread it. Individual creativity is possible only when technological knowledge has the characteristics of a collective process, one where the active participation of learning agents in epistemic communities characterised by technical proximity is both the cause and the consequence. Growth is a possible outcome of a system which when it is exposed to a continual mismatch between expectations and actual events is able to organise and structure its creativity so as to change its technology and its psychology. When the creative reaction is not appropriate and consistent, the system is kept within the standard conditions of the static equilibrium. In such a context it is clear that innovation feeds innovation. When and if the creative reaction to any mismatch is conducive to the introduction of new products and new processes, a self-sustaining process of growth and innovation occurs. The introduction of innovation is itself the factor that changes the conditions of equilibrium in both product and factor markets. The stochastic character of the growth process becomes clear. The levels of creativity engendered by the mismatch and the levels of technological change introduced depend on a number of complementary conditions. The mismatch is a necessary but non-sufficient condition to induce the successful introduction of new technologies. Only in special circumstances do all the conditions apply and are they conducive to a strong and positive creative reaction in terms of fast rates of introduction of new and highly productive technologies. Even hyper-rational agents who have access to Olympian rationality cannot be expected to be able to foresee the outcomes of the innovative process. When innovation is taken into account, the mismatch between expectations and actual factor and product market conditions is bound to take place, with varying levels of intensity and different gaps. At each round the mismatch can lead to the generation of new technologies or can decay into an equilibrium adjustment process leading to the conditions of static efficiency. The conditions that are conducive to the introduction of innovations clearly become the central focus of the quest for the conditions of dynamic efficiency. An understanding of the relationship between the amount of entropy within the system, that is to say, the size and the distribution of the mismatch between expectations and the actual conditions in product and factor markets, as well as the amount of creativity, is the first and central area of concern. Low levels of mismatch can easily be absorbed by firms, which can simply adjust prices to quantities and vice versa, with low levels of attrition. High levels of entropy however are also likely to endanger the capability of firms to react appropriately and to be able to introduce successful innovations. Empirical analysis in this area
348 Introduction of localised technological change is still scarce and could provide the basis on which theoretical investigations could subsequently be pursued. The conditions that affect the levels of agents’ creative reactivity, for given levels of entropy, play a major role in assessing the dynamic efficiency of an economic system. The organisation of firms in terms of hierarchical structure and decision making is extremely important, with respect to their effects in terms of accumulation of competence and technological knowledge, to their capability to convert tacit knowledge into timely innovations and, most importantly, to their ability to organise a timely and successful reaction to the changing conditions of their business environment. Such elements can be used to draw up a new layer of path dependence. Individual path dependence and system path dependence need to be identified so that their relations can be articulated. Individual path dependence defines the sequence of the conditions of each agent’s action and reaction. As said above, agents are characterised by substantial irreversibility of their production factors and their location in a given context. Specifically an agent’s action is constrained by his location in the product and factor markets on the one hand and by their location in the architecture of network relations which defines his access to knowledge pools and hence his capabilities in terms of the generation of new knowledge on the other. As soon as a mismatch between plans and expectations emerges, firms are able to try to reduce their opportunity costs by generating new knowledge and changing their technology so as to adapt as efficiently as possible to the context shaped by their irreversible commitments. At each point in time the possibility for firms to generate new relevant technological knowledge is defined by their specific and idiosyncratic access conditions to the existing pools of collective knowledge, by their location within such pools in terms of proximity and density and by the knowledge governance mechanisms that each firm has been able to implement. Specifically, creative firms, aware of the strategic role played by external knowledge in generating new knowledge and consequently aware of the relevance of their location in the space of knowledge networks, rely on procedural and subjective – rather than substantive – rationality in order to improve their location in the relevant multidimensional space. Creative firms, induced to generate new localised knowledge, try to change their location within the web of interactions and the knowledge networks by changing their strategies, their size, their borders and their connections with the rest of the system. Firms are supposed to be able to intentionally change both their own receptivity and the topology of the knowledge space with an array of strategies including entry into and exit from specific local knowledge pools, investment in communication channels, long-term contracts with other firms possessing bits of complementary knowledge, and the creation of technological platforms and technological clubs. Eventually creative firms embedded in a fertile knowledge network will be able to generate new relevant technological knowledge and change their technology locally.
Path dependence and the quest for complexity 349 Creative firms can be successful following such strategies, and this depends on the features of the system in which they are embedded at each point in time. They produce not only localised technological change but new strategies as well. These in turn lead to structural and architectural changes. Strategic conduct, based on the procedural and subjective rationality of creative agents, takes many different forms: 1) changes in the structure of the firm by means of inclusion and exclusion of activities, with strategies of diversification, integration and specialisation; 2) changes in the basic cost asymmetries, with effects on the market structure; 3) changes in the specialisation and division of labour, with consequent changes in the industrial structure; 4) changes in location, with consequent effects in terms of spatial agglomeration; and 5) changes in the organisation of transactions, interactions and connections with rivals, customers, suppliers and other learning organisations. Such a variety of specifications of the notion of mobility in topological space has clear effects on the architecture of the system, with direct effects on the topology of the knowledge pools in terms of their connectivity and their function as vectors of external knowledge for the agents at time t+1. At each point in time individual path dependence is shaped by systemic path dependence and vice versa. There is a continuous interplay between the behaviour of the individual agents and the characteristics of the market structure and the knowledge architecture, and at each point in time it is shaped by both irreversibility and creativity. It now becomes clear that systemic path dependence consists in the sequence of structural conditions. This will be in terms of market structure, industrial structure, regional structure, architecture of knowledge networks, systems of interactions and feedback, as well as dynamic coordination, all of which can generate the mismatches and feedback in terms of local externalities that respectively induce and shape the creative reactions of individual agents. Systemic path dependence is the specific and pertinent form of complex dynamics, where learning and diverse agents are connected in multiple ways. They interact both serially and in parallel to generate sequential as well as simultaneous novelty. Systemic path dependence displays spontaneous self-organisation within structures of interactions and feedbacks that are generated by local interactions and are characterised by weak irreversibility: ‘Emergence [of novelty] occurs in a narrow possibility space lying between conditions that are too ordered and too disordered. This boundary or margin is the “edge of chaos”, which is always far from equilibrium.’6 The analysis of systemic path dependence makes it possible to grasp the historical process at the aggregate level where at each point in time the architecture of the system is the product of the action of agents at time t–1 and yet it exerts a strong, albeit not fully irreversible, effect on the strategic conduct of agents. In turn the creative reaction of firms, in terms of both the introduction of new technologies and mobility in the relevant space dimensions, is able to change the structure as it was and to engender new structures in an ongoing evolving interaction between the system and the agents. The action of no individual agent can command the aggregate outcomes and yet it can influence it.
350 Introduction of localised technological change Table 18.3 The factors of individual and systemic path dependence Individual path dependence
Systemic path dependence
Strong and weak quasiirreversibility
Fixed production factors
Factor and skill endowments Competitive entropy Economic structure Input–output matrix
Local externalities and feedbacks
Access conditions to knowledge pools Proximity Receptivity Knowledge governance
Variety of agents and competence Architecture of knowledge networks Communication and percolation probability IPR to manage the knowledge trade-off
Creativity
Learning Creation of new communication channels Generation of new localised knowledge Introduction of new localised technologies Firms’ strategies
Structural change in industries and product and factor markets Architectural change in knowledge networks Supermodularity within knowledge pools New governance mechansms New technological systems and new technological districts New interactions and feedbacks Dynamic coordination
Systemic path dependence consists of an array of forms of structural changes at the system level where existing architectures shape the direction and the rates of change and yet do allow room for the effects of the strategic action of agents. The tradition of analysis in industrial economics based upon the structure–conduct– performance paradigm provides fundamental elements to feed this line of inquiry. Building upon that legacy, however, it is now possible to stretch the understanding of the notion of economic structure, based primarily on the analysis of the sectoral composition of the economy and the types of market forms, so as to include the endogenous endowment of production factors, the architecture of network interactions, the geographical distribution of activities, the institutional set-up and, most important, the governance of knowledge flows within the system. Thus it deserves to be reconsidered and implemented with the understanding of the feedback effects of the conduct and performance of firms on the architecture of the system.7 In such an extended approach the structural characteristics of the system at each point in time do affect the conduct and the performance of the firm, as in the structure–conduct–performance approach, but now include explicitly the introduction of technological and organisational innovation, and in turn do change as a consequence of the strategies of firms. Firms endowed with creative reactivity
Path dependence and the quest for complexity 351 can change their technologies, their borders, their strategies, their product markets and their industrial sectors, their location in the multidimensional space, the modes of interactions within the knowledge networks and the architecture itself of the networks of interaction and interdependence, including the institutional set-up. In so doing they affect the structure of the system. Path-dependent complex dynamics is fully articulated when the sequence of the structure–conduct–performance paradigm is stretched so as to include, as the last step in a circular flow, the new structure at time t+1.The structural and architectural characteristics of the system at time t+1 have changed under the effects of the creative reaction of firms. Different agents, rooted in different regions, with different endowments and hence different conditions of their local factor markets, may react with similar levels of creativity to similar changes in their current conditions, introducing new technologies with marked differences in terms of factor intensity not only because of the effects of internal localised learning and the access conditions to the local pools of collective knowledge but also because of the selection mechanism stemming from the powerful composition effects. Here composition effects act as an inducement factor that explains both the direction of the introduction of new technologies and their selective adoption and diffusion. Once again, however, feedbacks are at work. The direction of new technologies introduced, in turn, affects the derived demand for inputs and hence their relative price (Antonelli, 2002).
Structure, governance and architecture (t)
Localised technological changes
Firms’ strategies
Structures, governance and architecture (t+1)
Figure 18.2 The evolving interaction between individual path dependence and systemic path dependence
352 Introduction of localised technological change Path-dependent structural change includes more than the changes in the relative prices of production factors as affected by the direction of technological change and its effects on the derived demand for inputs. Other relevant aspects include the changes in the regional and spatial distribution of firms with multiplant and multinational strategies of growth and their consequences in terms of the concentration in urban and metropolitan areas and the creation of new technological districts and the decline of others, the changes in the market forms of product markets in terms of concentration, barriers to entry, cost asymmetries and the consequent changing levels in the price–cost margins, the changes in the industrial and economic structure of an economic system as it emerges from enhanced levels of division of labour with the identification and growth of new industries and the decline of others and the changing vertical flows of interaction among users and producers, the changes in the international division of labour as determined by the flows of exports and imports and the changes in the international specialisation of each system, the changes in the organisation of financial markets, the emergence of new markets, the changes in the organisation and governance of knowledge networks with the creation of new hubs and the decline of others and the changes in the institutional set-up. From this viewpoint it should be clear that the structure of the system, as well as the architecture and the governance of knowledge networks, is neither exogenous nor static, but on the contrary is the result of an endogenous process of evolution (Ormerod and Colbaugh, 2006; D’Ignazio and Giovannetti, 2007). Systemic path dependence may lead a system to higher as well as to lower levels of quality of the architecture of knowledge networks in terms of the conditions of access to external knowledge. Change does not necessarily imply progress. The notion of generative relationship introduced by Lane and Maxfield (1997) is especially relevant in this context. Generative relationships are: constructive positive feedbacks – that have an obvious counterpart: as the structure of agent/artefact space undergoes ripple of changes, new agents and artefacts come into being and old ones acquire new functionalities, so identities change – and hence, old interpretations of identity bear an increasingly strained relationship with observable actions, the facts of the world. Different agents respond differently: some respond to the resulting ambiguity by generating new attributions to make sense of experienced novelty, and so attributional heterogeneity increases – increasing further the possibility that participants in other relationships will achieve sufficient attributional diversity to become generative actually, trust may result from the interactions themselves. (Lane and Maxfield, 1997: 185) Generative relationships lead to the introduction of innovations, and innovations feed structural change in agent/artefact space. The process takes place through a ‘bootstrap’ dynamics where new generative relationships induce attributional shifts that lead to actions that in turn generate possibilities for new generative relationships. The structural characteristics of the system in terms of the distributions of
Path dependence and the quest for complexity 353 agents in multidimensional spaces and of their networks of communication, relationship and interactions qualified by aligned directedness, heterogeneity, mutual directedness, permissions and action opportunities are key elements for the sustainability of the process. The capability of a system to provide dynamic coordination to the creative reaction of the myriad of agents involved is a key element of ‘good’ systemic path dependence. At the system level, changes in the architecture of the network relations that are conducive to fostering the rate of technological change depend upon the mechanisms of dynamic coordination. When dynamic coordination is at work, new technological districts and new technological systems can emerge, with increasing returns both in the use and in the generation of new knowledge. The Schumpeterian gales of innovation are likely to occur when and where dynamic coordination within the new emerging architecture of the system provides support to the creative reaction of firms through the provision of external knowledge and low communication costs. Dynamic efficiency, hence, consists in the emergence of an architecture of the system that is able to provide firms with access to collective knowledge and hence to direct the adjustment process towards the introduction of new superior technologies.
4 Conclusion The notion of path dependence provides a unique framework for analysing the conditions for dynamic efficiency at the system level. Path dependence provides an extraordinary and articulated set of interrelated notions which make it possible to go beyond the analysis of the static conditions for general equilibrium and efficiency. Path dependence builds on a few simple basic elements: irreversibility and historical time, innovation viewed as a creative reaction, brought-together local externalities and feedback. These elements are put in place and integrated into basic economics so as to provide a basis upon which a fully articulated post-Arrovian approach can be elaborated: one where the microeconomic role of subjective rationality at the agent level and of markets as mechanisms for the creation and distribution of incentives is recognised and emphasised and yet contextualised. In such an approach the welfare attributes of equilibrium are no longer valid. As soon as technological and structural changes are endogenous, future prices and future markets are no longer plausible, and equilibrium growth itself is questioned. Sequences of possible and virtual equilibria can be identified and traced in a selfpropelling process of out-of-equilibrium changes that are the endogenous product of the system dynamics. From a different viewpoint it seems clear that path dependence makes possible the application of the theory of complexity derived from natural sciences and biology to economics, as it enriches it with the understanding of the law of systemic interactions among human agents who are able to rely upon procedural rationality and hence can both choose and be creative. The notion of path dependence makes it possible to implement the analysis of the conditions that make it possible for firms to convert the entropy of the systems, as determined by the continual mismatch
354 Introduction of localised technological change between expected and actual market conditions, into new and better technologies and so to fuel a rapid and effective growth. Path dependence provides a framework which makes it possible to go beyond both the deterministic and the static, but fully systemic, world of the Walrasian equilibrium. Moreover, it goes beyond the naive applications of complex system dynamics that are able to understand the dynamic consequences of the stochastic action of creative agents but are less able to take account of the intentional action of human beings. A new landscape appears, one where the dynamic outcomes of the interactions of agents in the marketplace cannot be fully anticipated. A variety of possible outcomes can be predicted, as well as their sequence and their dynamic relationship. Walrasian equilibrium is but one of the many possible outcomes, as well as growth. Growth is the positive, and stochastic, result of a system of interactions in which agents are able to react to the shortcomings of the mismatch between expectations and the actual conditions in the product and factor markets, and this changes the equilibrium conditions of the system. Economic policy now has new horizons. From this viewpoint path dependence proves to be far less hostile and conflicting with traditional microeconomics than often assumed. Path dependence even seems to provide important opportunities to rescue relevant portions of the received tradition, especially when the distinction between short-term and long-term analysis is considered and the sharp differences between subjective and objective rationality are taken into account. The consequences of the interactions of agents in markets now include the generation and use of new and better technologies, rather than the single gravitation around a stable and unique attractor. The definition of the interfaces of compatibility and incompatibility between the two areas of investigation becomes an interesting area for theoretical analysis. The goal clearly is the build-up and maintenance of a social, institutional and economic environment which is conducive to stimulating, implementing, appreciating, valorising and making effective the innovative reactions of myopic but creative and learning agents. From such a viewpoint the economics of localised technological change and hence the economics of path dependence can be considered as a major tool in building an economics of complexity. Creativity is the basic engine of change and development. Irreversibility and local externalities define the key elements in which action takes place. While the economics of non-localised creativity risks engendering the dynamics of chaotic behaviour, the economics of localised knowledge and localised technological change, because of the role of path dependence, provides a combination of ingredients which takes account of an induced and constrained creativity and hence quasi-predictable change. In so doing, it supplies elements which make system dynamics intelligible when applied to analysing economic systems. Past dependence, that is to say the effects of irreversibility and learning, guides the complex dynamics of the system and qualifies the features of the landscape in which the dynamics takes place. Creativity, however, does emerge, in proper conditions, and can change the direction and the rate of the dynamic
Path dependence and the quest for complexity 355 process: here path dependence comes into play, with a clear divide with respect to the deterministic approach of past dependence. Path dependence provides an analytical framework in which it is possible to analyse the results of complex system dynamics when applied to economic analysis as distinct from natural sciences. Economics needs to consider the intentional action of a myriad of heterogeneous and interdependent agents who are rooted in the relevant spaces but are credited with the capability to elaborate strategic choices based upon procedural rationality. Thus path dependence is better able to take account of the analysis of the behaviour of human agents who are no longer credited with assumptions of Olympian rationality but are still able to rely upon procedural rationality and hence to make strategic choices, albeit in a subjective, narrow and limited context of recognition and understanding. The strategic and creative action of such agents is likely to generate new technologies, in an appropriate context, and to affect the structural conditions in the system. Structural changes take place not only because of stochastic dynamic processes but also because of the rentseeking behaviour of agents who are always willing to change their location in the relevant space in which they are embedded. This is an assumption which is true as long as the economic agents are characterised by procedural rationality and creativity as well as by limited mobility. Structural changes at the system level generate new mismatches for individual agents and new opportunities for the generation of new knowledge in a continual process that can be understood only in terms of path dependence.
19 Conclusions Hysteresis and creativity
This book has elaborated the view that innovation is a path-dependent, collective process that takes place in a localised context, if, when and where a sufficient number of creative reactions are made in a coherent, complementary and consistent way. As such, innovation is one of the key emergent properties of an economic system viewed as a dynamic complex system. The combination of hysteresis and creativity is usually regarded as an oxymoron. This book has argued quite the opposite view. The economics of complexity provides a framework into which the oxymoron can be articulated. As Arthur et al. (1997) suggest, complexity consists of six basic ingredients: 1) dispersed interactions among heterogeneous agents acting locally on each other in some space; 2) no global controller that can exploit all the opportunities or interactions in the economy even though there might be some weak global interactions; 3) cross-cutting hierarchical organization with many tangled interactions; 4) continual adaptation by learning and evolving agents; 5) perpetual novelty as new markets, technologies, behaviors and institutions create new niches in the ‘ecology’ of the system; and 6) out of equilibrium dynamics with either zero or many equilibria existing and the system unlikely to be near a global optimum. (Rosser, 1999) This book has elaborated an approach that articulates the role of innovation as an essential feature of a complex dynamic system. Creativity is stirred by hysteresis, yet hysteresis limits the effects of creativity. At the same time, hysteresis defines and qualifies the context in which creativity can take place. Thus creativity changes the conditions of the system and sets the dynamics of the process, but within a narrow scope, shaped by irreversibility. Change and continuity are the two terms between which the path-dependent dynamics of generation of localised technological knowledge and introduction of localised technological change takes place. Building upon the classical, Marshallian and Schumpeterian legacies and elaborating the analysis of the role of constrained creativity, induced by the mismatch between plans and facts and by the effects of irreversibility, and augmented by the effects of localised learning and local externalities, the localised approach
Conclusions: hysteresis and creativity 357 is able to provide a strong endogenous and yet contextual explanation for the complex dynamics of technological knowledge and the introduction and adoption of technological changes. In so doing this approach pretends to go beyond the ubiquitous and automatic determinism of both new growth theories and the evolutionary thinking of Darwinistic ascent about technological change. In the former, as is well known, technological change is the smooth and gradual, incremental result of unconditional and ubiquitous learning processes. Yet the empirical analysis provides consistent evidence about a huge heterogeneity of rates of introduction of new technologies across time and among agents. Moreover, the larger is the desegregation, the wider is the variance found in the rates of introduction of new technologies. Clearly the access conditions to knowledge externalities and the idiosyncratic conditions of each agent, in terms of creativity, play a key role, which is not explained. In the latter, the system is supposed to be able to select among technologies, yet little analysis has been elaborated to explain why and how firms innovate, and consequently a poor foundation has been provided for the claim that firms can change the trajectory upon which they are constrained. The contrast between ex post and ex ante perspectives is sharp: ex post, every sequence of technological and organisational innovations can be claimed to be a trajectory. From an ex ante perspective however it seems clear that at each point in time each firm has in front of it a variety of possible innovations, each of which is constrained by past events but by no means fully determined by them. The many applications of evolutionary thinking to the economics of technological change have been successful in explaining the selection of variety but never provided a clue to grasping the origins of the continual reproduction of variety and the logic of the innovative choice that confronts each firm at each point in time. From this viewpoint it seems that a drift away from Darwinistic evolutionary approaches, where innovation is too often regarded as the result of random processes and attention is mainly focused on selection mechanisms, and the elaboration of an evolutionary approach able to better appreciate the role of intentional decision making and collective learning as part of a process of creative reaction to the changing conditions of the environment could improve the economic relevance of this essential biological reference. Innovation is the rare product of the intentional and constrained creative reaction of firms facing emerging mismatches between expectations and actual product and market conditions, which can take place only when, if and where a sufficient number of complementary and consistent conditions apply. The dynamics of localised technological change is endogenous and takes place only when a number of highly qualified circumstances apply. As such it can be understood only in the context of an economic theory that praises the conditions out of equilibrium and is able to appreciate the role of historical time. Localised technological change is the outcome of the creative reaction, to the mismatch between expectations and actual facts, of myopic firms that are not bounded to quantity–price adjustments, but are able to change also their technology in a limited technical space defined by the pervasive role of irreversibility of fixed production
358 Introduction of localised technological change factors and the effects of bounded rationality and learning processes. The reaction of firms to unexpected mismatches can be creative, as opposed to passive, only when the appropriate knowledge governance mechanisms are in place in fertile system architecture. As a consequence, at each point in time the system is kept away from one possible equilibrium and pushed towards many alternative ones made possible by the innovations introduced in a continual variety of efforts and attempts by heterogeneous and creative agents, surprised by the mismatch between expectations and actual product and factor markets, but able to change both their technologies and their location within the networks of interactions that shape the architecture of the system. The introduction of technological and structural changes is an endless process because each innovation modifies the context anticipated by each other agent and hence induces other reactions and eventually other innovations. The assumptions about the irreversibility of at least some inputs, the key role of learning and the mechanisms of knowledge governance in place at each point in time qualify the process as non-ergodic: historical time matters. The assumptions about failure-induced technological change based upon reactivity, creativity and endogenous innovative capability mark the distinction between a past-dependent process and path-dependent one: each innovation cannot be fully predicted from the past of the innovator. Good knowledge governance mechanisms provide better incentives to the generation of new knowledge and to its dissemination. Good system architectures make it possible to valorise the levels of creativity within the systems and each individual. Good knowledge governance mechanisms push the system to increase the amount of knowledge generated by means of increased levels of division of labour and coordination among the localised learning agents. Better configurations of the network of interactions may emerge out of the changes in the location and in the systems of communication among ‘mobile’ agents. Hence it is clear that the better the architecture of the system is and the better the mechanisms of knowledge governance implemented are, the faster the introduction of localised technological change is. In turn it is also clear that better knowledge governance mechanisms and better system architectures are likely to emerge where and when creative reactions are supported by the system, and rates of generation of new technological knowledge and introduction of effective technological innovations are both fast and consistent. The specific context in which agents are embedded qualifies their creativity and its effects. Learning is not sufficient for knowledge to emerge, and knowledge is not a sufficient condition for technological change to take place. Technological change is the result of creative reaction to emerging surprises. Technological change is also a specific aspect of a broader system dynamics where the interaction conditions of the system in which learning and reactive firms are localised play a key role in assessing the eventual results of active – as opposed to passive – conduct selected from a set of choices which includes innovation and structural change and is constrained both by the past and the burden of irreversibility and by the characteristics of the firms, consumers and rivals with which daily interaction and transactions take place.
Conclusions: hysteresis and creativity 359 The characteristics of the architecture of the system and within it the distribution and organisation of the knowledge commons and communication channels are essential to understanding the scope of the individual creativity of agents. Firms rely upon the multiplicative relationship between internal and external, codified and tacit knowledge, in order to generate technological knowledge. None of them can be disposed of. The generation of new knowledge is the result of four quite distinct and yet complementary processes: learning, research and development activities, socialisation and recombination. The changing mechanisms of knowledge governance and the evolving features of the interaction networks are essential to combining the incentives to the generation and the dissemination of new knowledge and hence to assessing the amount of technological innovation that the creative reaction of firms can generate. The conditions for the effectiveness of the creativity are a key factor in this line of analysis. The provision of knowledge externalities and the general conditions for the generation of new technological knowledge here becomes central. Access to knowledge in general is the key factor that makes it possible for creativity to be valorised and to lead to solutions in terms of higher productivity. Next to the architecture of the system in terms of connections among agents within and among industries and regions, the institutions of intellectual property rights and the quality of the public knowledge infrastructure are major factors in this context. The institutions of labour markets also play a major role. Seniority systems favour the accumulation of competence and make it easier to generate new technological knowledge. Mobility of qualified experts embodying high levels of tacit knowledge and technological competence across firms however is conducive to better dissemination and circulation of technological knowledge. The quality and the density of communication channels and interaction networks among learning agents, the access conditions to the flow of technological spillovers and the quality of technological and pecuniary knowledge externalities as determined by the characteristics of the industrial and economic structure are essential, as much as the organisation of financial markets and access to credit and financial resources at large. The changing architecture of the interaction networks within the system is most important in that it provides the opportunity to new firms to identify new markets and take advantage of the opportunities for generating localised technological change. In this approach the variety of actors and interacting markets matters. Firms are induced to innovate when their myopic expectations do not match the actual markets. Firms, in other words, react to all changes to their myopic expectations, not only with changes in the price–output mix but also with the introduction of new technologies and structural changes. The larger the variety of firms is, the larger both the mismatches and, especially, the variety of alternatives that are elaborated and tested by firms in the marketplace are likely to be and hence the greater are the chances that technological innovations will be introduced that are superior and able to increase significantly the general efficiency of the system. Moreover it is clear that the new technologies reflect the specific context of action. Such a context includes firms active in a broad industrial structure which includes
360 Introduction of localised technological change regions and countries with significant heterogeneity in both technologies and endowments. The larger and the more differentiated the competitive arena is, the greater the incentives are to introduce innovations and to sort the most productive ones: globalisation is likely to be the cause rather than the consequence of accelerated rates of introduction of technological change. In a population of heterogeneous agents, rooted in a heterogeneous economic space, with different local factor markets, different needs and preferences of consumers and different pools of collective knowledge, all discrepancies between the expectations upon which irreversible action is taken and the actual conditions of the factor and product markets are likely to engender the localised introduction of new technologies. The capability of agents to generate successful innovations will vary according to the architecture of the system and its effects on the stock of localised learning they can mobilise and the access conditions to the external pools of collective knowledge. The diffusion of such technologies will vary according to their characteristics in terms of output elasticity and shift effects. Increasing returns to adoption on the demand side and the rates of adaptation of consumers to the new products will play a major role in shaping the diffusion of new product innovations. The introduction and diffusion of innovations however are likely to affect the conditions of the product and factor markets upon which the expectations of other agents were built. New discrepancies arise and new feedback is likely to take place, with an endless process of creative reaction. It is clear that, the wider the heterogeneity within the system, the greater are the chances that expectations will not match the actual conditions and hence the faster the rates of introduction of new technologies’ structural change and possibly the faster the rates of growth, provided dynamic coordination is implemented. Technological change and structural change are both the endogenous result of the dynamics of the system: firms are able to generate new technological knowledge, to introduce technological innovations and to change the architecture of interaction networks within the system in which they are embedded. The dynamic complementarity and interdependence among the creative efforts of each agent are likely to play a major role in assessing the eventual outcome. When the innovative efforts of agents happen to be coordinated so as to add complementary bits of new technological knowledge, new technological systems can emerge. New gales of innovation are introduced, and growth, at the system level, can pick up momentum. When the local conditions change, an issue of coevolution and covariance between the dynamic conditions dictated by static and dynamic irreversibilities, internal to each agent, and the characteristics of the economic environment in which agents operate – and which is hence external to them – emerges. The types and forms of co-evolution of the local environment and the agents affect the dynamic characteristics of the processes in place and can determine major discontinuities and drastic changes in the path without assuming any exogenous discontinuity. The institutions of dynamic coordination play a key role in improving the architecture of the system and hence in assessing the levels of knowledge each agent is able to generate.
Conclusions: hysteresis and creativity 361 The architecture of the system is endogenous, as much as the rate and the direction of technological change. From a dynamic viewpoint it seems clear that the architecture of the interaction networks exhibits higher rates of resilience: past dependence is stronger. Moreover, from a functional viewpoint the quality of the architecture of the system, in terms of connectivity and receptivity, can both decline and improve. Systems’ architecture can become more or less conducive to providing effective access to knowledge pools and to making complementary the efforts of learning agents. The analysis elaborated in this book can be summarised by saying that the complex dynamics of localised technological change is the result of the interplay of the following items: 1 At each point in time, firms are rooted in a well-defined location with respect to techniques, interactions and geography. Learning takes place locally, near the existing technical, product and factor market conditions. Learning provides the basic input into the accumulation of contextual competence and tacit knowledge within firms. 2 At each point in time firms make plans, based upon expected virtual equilibria, that are necessarily myopic and yet they commit irreversible resources, both tangible and intangible, selecting local optima. All mismatches between plans and facts, in product and factor markets, are the cause of emerging frustrations, surprises and losses. Switching costs are necessary to overcome the effects of irreversibility and limit the mobility of agents in the relevant space of techniques, reputation and interactions, as well as in regional space. 3 The intrinsic heterogeneity of agents characterises the economic system and it is the result of their specific and idiosyncratic context of action and learning and of their differentiated timing of commitment of irreversible resources. 4 Firms facing such mismatches may react creatively. To do so they need to implement their internal tacit knowledge with internal codified knowledge and external tacit and codified knowledge by means of research and development, socialisation and recombination. Only the command of appropriate levels of technological knowledge makes the creative reaction possible and conducive to the eventual introduction and adoption of new technologies. A sharp distinction exists between technological knowledge, creative reaction and technological change. 5 Knowledge is inherently dispersed and fragmented, and no agent can claim the full command of all the relevant pieces of knowledge. Radical indivisibility and hence complementarity among forms and modules of technological knowledge are at work. 6 Both the rates and the direction of the generation of new knowledge and the introduction of new technologies are heavily influenced by the access conditions to external knowledge. The structure of the knowledge communication and interaction networks and the governance of technological knowledge within knowledge networks and the economic system at large play a key role. The absorption, acquisition and integration of external knowledge, as an
362 Introduction of localised technological change
7
8
9
10
essential input into the generation of new knowledge, are not free. On the contrary, socialisation and recombination, like the implementation and valorisation of internal learning and formal research and development activities, are resource-consuming activities. The generation of new knowledge is conditional on the structure and the governance of the connections within local knowledge networks. The architecture and the governance of communication and interaction networks within the system are evolving and endogenous. Firms are credited with the capability to change their location in the multidimensional space: firms can establish new communication channels and establish new interaction modes. The knowledge governance of the system changes under the effect of the changing interaction strategies of firms. Localised increasing returns take place as long as the benefits of local knowledge externalities exceed the costs of their acquisition. Such localised increasing returns take place only within technological network systems, provided effective knowledge governance and dynamic coordination systems are in place. The rates and direction of generation of new knowledge and innovation can now be viewed as an emergent property of the complex dynamic system in which agents are embedded. The introduction of new technologies is a major cause of surprise and hence mismatch between plans and actual market conditions for any other ‘rational’ agent in the system. Hence the entropy conditions in the system are endogenous. Creative reaction changes the conditions upon which long-term commitment has been made and causes structural changes in product and factor markets. The intentional activity of creative agents affects the topological features of the system, such as the structure of firms, industries and sectors, endowments and division of labour, the architecture and the governance of social interactions, with important effects in terms of connectivity and receptivity of the knowledge networks. Structural changes parallel the generation of new knowledge and the introduction of new localised technological changes. The introduction of new technologies is itself the result of enhanced rates of generation of new technological knowledge and hence a wider and thicker knowledge base which, in a context characterised by the institutional conditions of knowledge governance mechanisms within local systems, favours the intensity of creative, as opposed to passive, reactions and hence the introduction of further innovations and structural changes, including new institutional set-ups and communication networks. Because of the pervasive role of the intentional action of agents, albeit inherently subjective and localised by quasi-irreversibility in a narrow space, and the ongoing feedbacks between individual action and structural change, path dependence is the form of complex adaptive dynamics best suited to analysing the outcome and the process of dynamic interaction and systemic interdependence among heterogeneous and creative agents that are, at each point in time, embedded in a specific contextual location and yet able to change their technology, to generate new knowledge, to change intentionally their
Conclusions: hysteresis and creativity 363 topological context of action and, in so doing, to change the structural characteristics of the system in which they are embedded. The dynamics of localised technological change is path dependent because of the pervasive role of the interplay between irreversibilities, local externalities and the individual creativity of heterogeneous agents. The dynamics of localised technological change is complex because of the key role of the dynamic interplay between individual creativity and the changing and endogenous architecture and governance of the system and of knowledge networks, where both terms are able to influence each other in the process. Both the technology and the structure of the system, at each point in time, and their dynamics are the product of many, but the project of nobody.
Notes
1 Introduction 1 Herbert Simon (1962) introduces the analysis of the architectural characteristics of a system and stresses its relevance. Since then, the applications to economics of this notion have been increasing (Foster, 2005). 2 See D’Ignazio and Giovannetti (2007) for a recent survey on the economics of endogenous network formation. 3 According to N. M. Shnerb, Y. Louzoun, E. Bettelheim and S. Solomon (2000), ‘Many systems in chemistry, biology, finance and social sciences present emerging features which are not easy to guess from the elementary interactions of their microscopic individual components. In the past, the macroscopic behaviour of such systems was modelled by assuming that the collective dynamics of microscopic components can be effectively described collectively by equations acting on spatially continuous density distributions. It turns out that quite contrary, taking into account the actual individual/ discrete character of the microscopic components of these systems is crucial for explaining their macroscopic behaviour. In fact, we find that in conditions in which the continuum approach would predict the extinction of all the population (respectively the vanishing of the invested capital or of the concentration of a chemical substance, etc), the microscopic granularity insures the emergence of macroscopic localized subpopulations with collective adaptive properties which allow their survival and development. In particular it is found that in 2 dimensions “life” (the localized proliferating phase) always prevails.’ 4 As Stiglitz notes: ‘The observations just made about the ways in which information and knowledge differ from the conventional commodities are general: they apply both to new knowledge, about new products or processes, as well as to information, say about the characteristics of a particular investment opportunity. They have developed in the last 50 years two distinct branches of the subject – the economics of innovation and invention, focusing on what is often called knowledge, and the economics of information. Both have important implications for thinking about economic behaviour’ (Stiglitz, 2000: 1449–50). 5 As Stiglitz notes: ‘In standard neoclassical economics the deep properties of an economy – preferences and technology – determined outcomes. Not just distribution, but history and institutions did not matter. But there are natural irreversibilities associated with the creation of knowledge: history has to matter. Indeed, economies with the same deep properties could have markedly different equilibria’ (Stiglitz, 2000: 1459). 6 This book frames, expands, integrates and develops a set of hypotheses that have been tested in articles and chapters that appeared in: C. Antonelli (2007), The economics of innovation, Routledge, London; C. Antonelli (2005), Models of knowledge and systems of governance, Journal of Institutional Economics 1, 51–73; C. Antonelli (2003), Knowledge complementarity and fungibility: Implications for regional strategy,
Notes 365 Regional Studies 39: 595–606; C. Antonelli (2005), The economics of governance: The role of localized technological change in the interdependence among transaction, coordination and production, in Ken Green, Marcela Miozzo and Paul Dewick (eds), Technology, knowledge and the firm: Implications for strategy and industrial change, Edward Elgar, Cheltenham, pp. 29–50; C. Antonelli (2006), The governance of technological knowledge: To use or to sell, in Jon Sundbo, Andrea Gallina, Göran Serin and Jerome Davis (eds), Contemporary management of innovation, Palgrave Macmillan, Basingstoke, pp. 208–29 (reproduced with permission of Palgrave Macmillan); C. Antonelli (2006), The business governance of localized knowledge: An information economics approach to the economics of knowledge, Industry and Innovation 13, 227–61; C. Antonelli (2007), The system dynamics of collective knowledge: From gradualism and saltationism to punctuated change, Journal of Economic Behavior and Organization 62, 215–36 (copyright Elsevier 2007); C. Antonelli (2006), Localized technological change and factor markets: Constraints and inducements to innovation, Structural Change and Economic Dynamics 17, 224–47 (copyright Elsevier 2006); C. Antonelli (2004), Localized product innovation: The role of proximity in the Lancastrian product space, Information Economics and Policy 16, 255–74 (copyright Elsevier 2004); C. Antonelli (2006), Diffusion as a process of creative adoption, Journal of Technology Transfer 31, 211–26 (with kind permission from Springer Science and Business Media); C. Antonelli (2005), The economics of localized technological change: The role of creative adoption, in Grazia D. Santangelo (ed.), Technological change and economic catch-up, Edward Elgar, Cheltenham, pp. 1–16; C. Antonelli (2006), Path dependence, localised technological change and the quest for dynamic efficiency, in C. Antonelli, D. Foray, B. H. Hall and W. E. Steinmueller (eds), New frontiers in the economics of innovation and new technology: Essays in honour of Paul A. David, Edward Elgar, Cheltenham, pp. 51–69. All the material is reproduced with permission of the respective publishing companies: Edward Elgar, Springer, Palgrave Macmillan, and Elsevier. I acknowledge the financial support of the European Union Directorate for Research within the context of the Integrated Project EURODITE (Regional Trajectories to the Knowledge Economy: A Dynamic Model), Contract no. 006187 (CIT3), in progress at the Fondazione Rosselli; and of the research grants of the Dipartimento di Economia Salvatore Cognetti de Martiis of the University of Torino for the years 2006 and 2007. 3 Localised technological change: a critical assessment 1 A distinction seems useful here. Simon introduced first the notion of bounded rationality in order to stress the limitations of the traditional assumptions about the Olympian rationality of ‘Homo economicus’. Bounded rationality quickly became a building block for the newly emerging economics of information: the acquisition of information and the generation of signals are costly and economics needs to care about them. Subsequently, however, Simon elaborated the distinction between substantive and procedural rationality. The introduction of the notion of procedural rationality had farreaching consequences, as it introduces the notion of sequential decision making. Agents cannot achieve substantive rationality for the burden of the wide range of activities necessary to gathering and processing all the relevant information. Agents, however, can elaborate procedures to evaluate at each point in time and space the possible outcomes of their behaviour, but within the boundaries of a limited knowledge and using satisfying criteria as opposed to maximisation rules. The notion of procedural rationality introduced by Herbert Simon marks a major contribution to the economics of innovation. Olympian rationality is at odds with a context characterised by radical uncertainty where nobody actually knows the outcome of a research project and even less so the next directions of the technological changes being introduced. As a matter of fact the very notions of future prices and future markets cannot even be considered when technological change is taken into account. In such a context only sequential decision making based
366 Notes upon limited information and limited knowledge is possible. The application of the notion of procedural rationality to the economics of innovation leads to a new understanding of the basic inducement of innovation. 2 The well-known Darwinist argument elaborated by the Chicago School about the irrelevance of the actual use of optimisation by single agents, as the market is able to select those firms that, even by accident, have chosen the ‘correct’ equilibrium solutions and to sort out those that did not find themselves at the equilibrium point, applies only when technology is not only exogenous but also given and static. Clearly, when technological change takes place, agents who made the wrong choice can turn out to be better off than agents who selected the correct equilibrium points. When technological change is not only allowed to change but is endogenous, it is clear that the role of the markets as the collective focusing mechanism that pushes the system towards the single static equilibrium level is longer relevant. In a context where firms can change their technology intentionally, the features of the decision-making process at the microeconomic level become crucial to understanding the dynamics of the system. While optimisation cannot provide at the system level an appropriate representation of the dynamics of the system, at the level of the agent and in the short term optimisation is an appropriate representation of the decision making of agents, provided the localised context of action and understanding is appreciated. As Foster (2005) notes, the Marshallian legacy here provides important hints to put optimisation in a proper methodological context: ‘As Alfred Marshall stressed a century ago, optimization theory can be useful in short period and local contexts when it is fairly certain that institutions, culture, politics and general economic conditions are not changing very much, and there is time and other resources available to engage in optimization exercises. Importantly, Marshall was not positing a simplistic model of an economic system, as the general equilibrium theorists that he opposed would do later on, only an analytical device that could be used as an approximate “as if” way to understand movements in prices. He was aware, in an intuitive sense, of the fact that the economy is a complex system which has to be cut in certain ways before optimization theory can be useful in understanding movements in prices’ (Foster, 2005: 881). 3 Let us consider the case of the introduction of a new ‘biased’ technological change. We can now confront the old and the new technologies as expressed by two production functions: (1) Y1 = A1 Ka1 Lb1 (2) Y2 = A2 Ka2 Lb2 where a1 differs from a2 and hence b1 differs from b2. A simple logarithmic transformation makes it possible to identify the intersection: (3) log A1 / A2 = (a1 / a2) log K + (b1 / b2) log L 4 Information economics for the economics of localised technological knowledge 1 See North (1997: 226): ‘Competition forces organizations continually to invest in new skills and knowledge to survive. The kind of skills and knowledge individuals and their organizations acquire will shape evolving perceptions about opportunities and, hence, choices that will incrementally alter institutions . . . While idle curiosity is an innate source of acquiring knowledge among human beings, the rate of accumulating knowledge is clearly tied to the payoffs. Secure monopolies, be they organizations in the polity or in the economy, simply do not have to improve to survive. But firms, political parties, or even institutions of higher learning, faced with rival organizations, must strive to improve their efficiency. When competition is muted (for whatever reasons), organizations will have less incentives to invest in new knowledge and, in consequence, will not induce rapid institutional change.’
Notes 367 2 See Greve (1998), who examines how performance feedback affects the probability of risky organisational changes. His empirical analysis in the radio broadcasting industry shows the consequences of shortfalls of performance on the probability of strategic change and their strong sensitivity to social and historical aspiration levels. Ocasio (1995) provides a theoretical reconciliation of theories of failure-induced change and threat rigidity. The theory explicitly links the cognitive psychology that underlies risk-seeking behaviour and threat rigidity with the social groupings and cultural rules that structure thinking and decision making in organisations. 3 This approach differs from that of Nonaka and Takeuchi (1995), also shared by Cowan, David and Foray (2000), according to which tacit knowledge eventually converts fully into codified knowledge. On the contrary, it is argued here that tacit knowledge remains an essential and non-disposable input which can never be fully codified. 4 As Stiglitz notes: ‘The observations just made about the ways in which information and knowledge differ from the conventional commodities are general: they apply both to new knowledge, about new products or processes, as well as to information, say about the characteristics of a particular investment opportunity. They have developed in the last 50 years two distinct branches of the subject – the economics of innovation and invention, focusing on what is often called knowledge, and the economics of information. Both have important implications for thinking about economic behavior’ (Stiglitz, 2000: 1449–50). 5 Models of knowledge and systems of governance 1 As Stiglitz (1985: 146) puts it, ‘lenders are only concerned with the bottom part of the tail distributions of returns’. 2 The Marshallian analysis of external increasing returns within industrial districts and the Marshallian legacy of the role of technological externalities attract new attention (Marchionatti, 2004). 3 See Barabasi (2002) for a convincing analysis of the effects of the emergence of scalefree networks as the result of the convergence in a hub of qualified relations. See Hayek (1945: 520): ‘It seems to me that many of the current disputes with regard to both economic theory and economic policy have their common origin in a misconception about the nature of the economic problem of society. This misconception in turn is due to an erroneous transfer to social phenomena of the habits of thought we have developed in dealing with the phenomena of nature’. 6 The new dimensions of knowledge indivisibility: fungibility, cumulability, compositeness and stickiness 1 See Antonelli and Calderini (2008) for an empirical analysis. 2 Drawing from telecommunication economics, incremental costs are defined by the costs of adding dedicated and complementary units of knowledge to the existing stock for each specific application (Baumol and Sydak, 1994). 3 Two quite different types of supply-side externalities have been identified in the economic literature: 1) technological externalities; and 2) pecuniary externalities. Technological externalities consist of direct interdependence among producers. Pecuniary externalities consist of indirect interdependence. In the former case the interdependence is not mediated by the market mechanisms. In the latter, instead, interdependence takes place via the effects on the price system. Positive pecuniary externalities reduce the price of production factors below the equilibrium level. The new growth theory elaborates exclusively upon the notion of ‘technological externalities’. Technological externalities take place when technological knowledge is considered as an unpaid production factor that enters the production function. Following Scitovsky (1954), technological external economies apply when ‘The producer’s output may be
368 Notes influenced by the action of persons more directly and in other ways than through their offer of services used and demand for products produced by the firm. This is the counterpart of the previous case, and its main instance is inventions that facilitate production and become available to producers without charge’ (p. 144). Here we explore a complementary analytical path, based upon the notion of pecuniary externalities. Pecuniary externalities consist of ‘interdependence among producers through the market mechanism . . . This latter type of interdependence may be called pecuniary external economies to distinguish it from technological external economies of direct interdependence’ (Scitovsky, 1954: 146). Here we assume that external increasing returns are based on both ‘technological’ and ‘pecuniary’ externalities. 4 The notion of percolation probability can be applied here so as to measure the characteristics of the system in terms of knowledge communication. 7 Knowledge and the theory of the firm: the interdependence among transaction, coordination and production 1 See the useful surveys of Zingales (2000) and Mueller (2003), which, interestingly, do not take into account the role of learning in the theory of the firm and more generally ignore the resource-based theory of the firm. 2 As North and Wallis (1994) put it well: ‘one summary hypothesis of Coase is: given the existence of a specific technology, institutions are chosen by firms to minimize transaction costs. This formulation of Coase, and its shorthand, rule of thumb expression – institutions are chosen to minimize transaction costs – have been extremely productive starting points for understanding the evolution of institutions in the modern economy. But what happens when technologies are not given? What happens when technologies and institutions are simultaneously chosen? Then the rule of thumb proposition that institutions are chosen to minimize transaction costs is demonstrably false. Institutions will be chosen that minimize total costs, the sum of transformation and transaction costs, given the level of output’ (North and Wallis, 1994: 610, italics in original). 3 Incomplete contract theory stresses the role of bounded rationality and the limitations of information impactedness in designing ‘perfect’ contracts and hence the need for internal coordination (Grossman and Hart, 1986; Hart, 1995). Repeated renegotiations however can reduce the costs of the use of the markets as well. Upon this basis, incomplete contract theory seems to be complementary both to transaction costs economics and to agency theory (Brousseau and Glachant, 2002). 4 See Chapters 10 and 11 for a deeper analysis of the principal–agent problem with respect to the generation of knowledge. 5 This model builds upon the basic intuition provided by Riordan and Wiliamson (1985) and North and Wallis (1994), subsequently elaborated in Antonelli (1999b). 6 This makes it possible to integrate the role of organisational knowledge together with technological knowledge (Turvani, 2001). 7 See North (1997: 226): ‘Competition forces organizations continually to invest in new skills and knowledge to survive. The kind of skills and knowledge individuals and their organizations acquire will shape evolving perceptions about opportunities and, hence, choices that will incrementally alter institutions . . . While idle curiosity is an innate source of acquiring knowledge among human beings, the rate of accumulating knowledge is clearly tied to the payoffs. Secure monopolies, be they organizations in the polity or in the economy, simply do not have to improve to survive. But firms, political parties, or even institutions of higher learning, faced with rival organizations, must strive to improve their efficiency. When competition is muted (for whatever reasons), organizations will have less incentives to invest in new knowledge and, in consequence, will not induce rapid institutional change.’
Notes 369 8 The localised generation and appropriation of technological knowledge 1 This analysis provides a clue to understanding the puzzling evidence about the low levels of formal research activities of firms localised in fertile and dynamic technological districts. Here the notion of localisation acquires a strong geographic, cognitive and institutional character. 2 It is important to stress once more that in our approach the firm cannot rely exclusively on its internal competence and internal research activities because of the intrinsic complementarity and the limited substitutability between external and internal sources of knowledge. 3 The quest for idiosyncratic inputs can take place both internally and externally. Building upon Edith Penrose, we see that firms have an incentive to generate internal production factors that are idiosyncratic because they are the result of a specific learning process: as such, they are difficult to replicate for other agents which do not share the very same historical process of growth. Stretching the argument elaborated by Edith Penrose we argue that the identification and exploitation of distinctive external factors that other firms cannot access easily at the same price have similar effects at the regional level in terms of incentives and focusing mechanisms upon the direction of technological knowledge. This is all the more true and interesting in a globalised economy where firms localised in different factor markets compete on quasi-homogeneous global product markets. 4 Absorption costs of external knowledge do matter and affect the research strategies of firms. When technological knowledge has high absorption costs, firms will rely more upon internal inputs (Griffith et al., 2003). 5 See Antonelli (1995) for the full specification of the inducement mechanism within the context of the localised technological change approach. 6 The identification of the best mix of inputs depends upon the specific form of the knowledge production function. When the notion of partial substitution built into the traditional Cobb–Douglas specification is recalled, however, it is clear that standard maximisation procedures make it possible to identify the ‘best’ mix of knowledge inputs, provided that the shape of the envelop curve which defines the region where substitution is possible reflects properly the great size of the regions where complementarity among knowledge inputs applies. 7 The argument becomes stronger when the efficiency of external knowledge is greater than that of research and development activities (a > b), but holds even when the efficiency of internal research and development activities is greater than that of external knowledge (a < b). 8 It seems worth noting that the process can be harmed by the reduction in variety and the consequent decline in creative opportunities. This can become an important guideline for local knowledge policies. 9 To use or to sell technological knowledge 1 In this context all assessment is a comparative one. Upstream transaction costs are assessed in relative terms with respect to the sum of internal coordination costs, production costs and transaction costs in downstream markets. 2 Specifically we need to assume that K is a module of technological knowledge with high levels of fungibility: it can be applied to a variety of different products which are sold in different markets; and Z is a vector of different products with high levels of separability. Geographic distance and hence transportation costs can provide separability, as well as national boundaries and hence customs duties. Barriers to mobility across product markets and high research costs impede the easy application of the same knowledge to different products.
370 Notes 10 The governance of localised knowledge in the business sector 1 Along these lines a knowledge production function and a cost equation can be identified. Internal learning and research and development, external codified knowledge and external tacit knowledge respectively enter a knowledge production function as inputs. Their costs, i.e. the costs of internal research and learning activities, the total costs of external codified knowledge, including knowledge transaction costs, and the costs of external tacit knowledge, including knowledge networking costs, enter a cost equation. Maximisation procedures make it possible for firms to select the ‘best’ mix of inputs. This analysis has many important implications about the role of the local context into which firms are embedded in assessing the quantitative evidence about R&D expenditures and output in terms of rates of generation of new knowledge. It is clear, for instance, that when and where external knowledge is cheap, because of both low purchasing costs in the markets for knowledge and low knowledge transaction and networking costs, firms will rely less on internal learning and research activities. Conversely, when and where the access conditions to external knowledge are less easy, firms will rely more on internal research and learning activities. When external knowledge is available at low cost, clearly firms can generate more knowledge at lower cost. This analysis provides a clue to understanding the puzzling evidence about the low levels of formal research activities of firms localised in fertile and dynamic technological districts (see Patrucco, 2003). 2 Knowledge generated by academic departments within the context of specific contracts with firms risks becoming proprietary, with clear reductions in its dissemination. At the same time, however, according to much economics of information, the working of competition in a market characterised by radical knowledge asymmetries provides an important counterbalancing effect when the role of signalling is appreciated. Academic departments in fact have a strong incentive to signal to prospective customers the quality of the research in progress and to disseminate information about the scientific scores. Academic publication, no longer viewed as the distinctive mission of publicly funded researchers, is now pursued by academic departments as a signal to attract new potential customers for their services. 3 Traditional R&D-intensive countries, such as the United Kingdom, Germany and the United States, exhibit a clear trend towards an increasing use of external knowledge. The ratio of technology payments (TP) to the research and development activities performed by the business sector (BERD) steadily increased throughout the 1980s and 1990s in Germany from less than 15 per cent to 34 per cent at the end of the century. In the United Kingdom, it spanned from less than 15 per cent in the early 1980s to 22 per cent in 1999. In the same time interval in the United States it increased from 1 per cent to almost 5 per cent. France remained stable around a 15 per cent ratio. Countries with lower levels of BERD/GDP intensity, such as Belgium, show an increasing trend, reaching 125 per cent in 1999 from the 72 per cent levels of 1981. Italy remained around the 30 per cent level for all of the period considered. Canada, in the vicinity of 25 per cent in 1981, shrank to 16 per cent in 1999. Countries like Spain instead reduced their dependence on external knowledge: the ratio of TP to BERD shrank from a ratio of 150–160 per cent towards the 50 per cent level. Finally Japan, once a strong importer of foreign technology, reduced the ratio of TP to BERD from 7 per cent in 1981 to 4 per cent in 1999 (Antonelli et al., 2003). 4 As a consequence, R&D statistics are less reliable as an indicator of the effective amount of resources invested in the generation of new knowledge by firms. 5 It is worthwhile and analytically rewarding to spell out the origin of the acronym FLOSS (Free Libero Open Source System) currently used to define the system according to which users have access to the General Public Licence provided they agree to make available the results of their interventions. The insertion of the Latin word ‘Libero’ in Italian or ‘Libre’ in French is made necessary by the lack of specificity in the English
Notes 371 language, which is unable to articulate the distinction between freedom of speech and freedom from charge. The FLOSS is clearly based upon the first but does not imply the second. Users have in fact a barter obligation to make available to others the results of their access to the open source. 11 The new economics of the university: a knowledge governance approach 1 The share of innovations introduced in biotechnology and information and communication technologies by small firms is large in absolute terms and much larger than the share of innovations introduced in mechanics, pharmaceuticals and chemistry by large corporations in the previous waves of technological change. 2 In equilibrium the university should pay more to the scientific worker with a higher reputation for the increased quality of the teaching and the positive effects in terms of higher tuition fees and higher quality of students recruited by the university. 3 This mechanism of indirect reward can also take place in a broader context: often academics become political leaders, or consultants to large banks and financial corporations, if not directly members of their boards; occasionally they are appointed to high-level bureaucratic posts and even to parliaments and governments. 4 Many ‘traces’ in the institutional setting of the academic system reveal that salaries paid by universities are mainly related to didactic activities. In many US universities the university covers a yearly salary based upon nine months of full paid salary. The ‘buy-back’ procedure enables professors to be freed from teaching, paying the university a large, if not substantial, part of their wage. In the United Kingdom many academics have formal appointments within universities; their salaries however are paid from the income generated by the same individuals by means of research contracts with outside parties. In Italy over 90 per cent of top-level academics, especially in the schools of law, business and engineering, switch to part-time positions after achieving the status of full professor, with a reduction of over 40 per cent of their academic salary and no changes in didactic burden, but full freedom to earn professional fees. They often return to full-time positions before the end of their academic career. 5 The measurement of the total marginal product of output where the latter is knowledge cannot be based solely on the price levels. Following the well-known Arrovian analysis, knowledge cannot be fully appropriated and is characterised by non-divisibility, nonrivalry in use and non-exhaustibility: relevant externalities are at work. The output of the open science university can be measured only in terms of the externalities that are effectively received by the system. In the case of the professional university, instead, the private marginal returns, estimated by means of the price mechanism, enter into the picture, although externalities remain relevant. Figure 11.2 makes clear the extent to which public subsidies are necessary to compensate for the value of knowledge that is not accounted for by the price mechanism. 6 Hence in Figure 11.2 on the vertical axis we measure the value of the knowledge rather than just the price. 7 Interesting similarities can be found between the economics of open source and the new economics of the professional university. In both cases the basic mechanism consists in the complementarity between the reputation-seeking behaviour of creative talents who can claim the authorship of an ‘invention’ and the levels of the professional fees that can be earned by ‘inventors’ in the markets for idiosyncratic services, provided that a regime of non-exclusivity of intellectual property rights applies. Creative experts make available their advances in open source software provided that their contribution is acknowledged: the effects in terms of reputation in the markets for specific applications produce sufficient incentives to sustain the system, with clear benefits in terms of the increasing returns at the aggregate level stemming from the cumulability of knowledge and the ensuing economies of density in repeated and distributed usage. More specifically, it can be argued that the working of the open source community shows both
372 Notes how relevant is the non-exclusivity of intellectual property rights and how crucial is the identification of the expertise and ingenuity of the professional competence of intellectual workers (David, 2004b; Antonelli, 2007d). 12 Towards non-exclusive property rights: knowledge as an essential facility 1 See Chapter 9. 2 From a strict legal viewpoint the General Public Licence remains in the domain of the property rule. From an economic viewpoint, however, once the ‘inventor’ has selected the GPL as an appropriate form of intellectual property right for his/her purposes the consequences are similar to those exerted by the liability regime. See Samuelson et al. (1994). 3 The analysis of the reasons for the rebuttal of the ex ante measurement of the value of innovation for licensing purposes is clearly stressed by Richard Nelson (2004: 458): ‘Virtually all empirically oriented scholarly accounts of how technology progresses have highlighted that the process is evolutionary in the following senses . . . First, at any time there generally are a wide variety of efforts going on to improve prevailing technology, or to supersede it with something radically better. These efforts generally are in competition with each other, and with prevailing practice. And the winners and the losers in this competition to a considerable extent are determined through ex-post selection processes. Second, today’s efforts to advance a technology to a considerable extent are informed by and take off from the success and failures of earlier efforts’. This quote makes clear also how unfair would be any attempt to remunerate the use of a successful technology on the basis of its specific research costs. It seems clear that, because only a small fraction of many tentative technologies is sorted out, ex post, in the competition process, the success of one technology should be able to provide the ‘inventor’ with a payback for many aborted efforts. 4 See, for a parallel analysis on the value of artistic talent, Swann (2006). 5 Hence the compensations should not take the form of ‘a set of off-the-rack liability rules allocating contributions to the costs of R&D for unauthorised uses of sub patentable innovations within a specified period of time’ (Reichman, 2000: 1791), but rather they should be based on a royalty scheme. Reichman agrees that, when direct competition between inventors and imitators is considered, a shift from cost-based to revenue-based compensation schemes is appropriate, which ‘can be achieved by imposing a higher percentage royalty than would ever apply in the case of follow-on innovations, by extending the period of compensatory liability beyond that applicable to follow-on innovation, or by some combination of the two’ (Reichman, 2000: 1791). 6 Following our argument it can be claimed that the introduction of the notion of knowledge as an essential facility and the related substitution of the liability rule to the property rule are likely to change the market structure and hence to increase the levels of the total surplus stemming from the introduction of an innovation. With exclusive intellectual property rights, the exclusive owner of the knowledge can take advantage of monopoly power in the markets of the product that use and benefit from the application of the proprietary knowledge. With monopolistic pricing the welfare effects of the introduction of the new knowledge are clearly lower than the effects of the use of the new knowledge with competitive pricing. Competitive pricing is clearly made possible by the application of non-exclusive intellectual property rights. 7 The positive effects of the application of the notion of knowledge as an essential facility and of the liability rule should include the reduction of monopoly power in the downstream markets for the products that use the proprietary knowledge. Now in fact many firms can use it and patents are no longer a cause of downstream monopoly. 8 This conclusion is quite consistent with the new views of Reichman, who originated the proposal for a liability regime. In a recent contribution Reichman has come around to the view that litigation costs could be fatal to it. He now refers instead to ‘compensatory
Notes 373 royalties falling within a specific statutory range for a specified period of time’. So the 3 per cent to 9 per cent royalty rate which in his original article he thought might be appropriate would now have the sanction of law, and he argues that ‘about a couple of percentage points in royalties, ideally before an arbitrator or mediator, is socially preferable to litigating costly actions for infringement’ (Reichman and Maskus, 2005: 360). 9 See Richard Nelson (2004: 462): ‘For our purposes here, the most salient aspect of the economist’s public good concept is that a public good is “non-rivalrous in use”. By that it is meant that, unlike a standard economic good, like a peanut butter sandwich, which either you or I can eat but not both (although we can split it), as a public good can be used by all of us at the same time without eroding the quality for any of us. Knowledge is a canonical case of something that is non-rivalrous in use in this sense, and this is not a proposition conjured by economists. The notion that I can tell you what I know, and then you will know it, and I will too, almost surely has been widely understood by sophisticated persons for a long time. There is no “tragedy of commons” for a public good like knowledge. And to deny access, or to ration it, can result in those denied doing far less well than they could if they had access. In this case in point, if access to certain bodies of scientific knowledge or technique can be withheld from certain researchers, they may be effectively barred from doing productive R&D in a field’. 10 As Stuart McDonald (2004) notes: ‘Patents are even now being taken out so that their owners can make money out of vexatious and aggressive litigation rather than from actual innovation’. 13 Localised technological change: the benchmark 1 The reference to the behavioural theory of the firm, described by March and Simon (1958) and Cyert and March (1963), is clear. 2 Such changes are mainly brought about by other agents introducing innovations in both product and factor markets. 3 Here the analysis of Part II is directly relevant. 4 See Antonelli (2001) for an analysis of the combined effects of a change in both the levels of output and the levels of factor prices. 5 The ‘necessary’ assumption that dRI/dAA′ = 0 seems plausible for the reduction in the costs. The latter can be considered as the revenue of innovation activities which make possible the introduction of localised technological innovations along the fixed capital endowment line, and can be modelled as linear in the distance from the intersection of the isoquant expressing the new desired output level. 6 When the dynamics of product markets is at the origin of the mismatch and the demand for the products of the firm differs from the expected levels, the firm with no irreversibility constraints should simply move towards the new desired isoquant on the map of existing isoquants. Once again, however, such a shift is likely to engender relevant switching costs for firms constrained by weak irreversibility. The alternative is, again, provided by the opportunity to generate new technological knowledge and to introduce a technological innovation which makes it possible to minimise switching costs and possibly to produce the new desired levels of output while retaining the previous position in terms of the size of the inputs. In so doing the firm will be able to produce more output with the old inputs and hence to generate directly an increase in total factor productivity. This model applies directly to all increases in the demand levels but also to possible reductions in the output levels: the firm able to introduce localised technological changes will benefit from a clear increase in its competitive advantage and hence will be able to acquire a larger demand for its products. In this case the general model of choice between switching and innovation applies, but with an important difference: now the isorevenue becomes convex. The introduction of innovations increases the profitability of the firm; hence the larger is the amount of innovations that are under consideration, the larger is
374 Notes
7 8 9 10
11 12 13
14 15 16
17
the expected profitability. The convexity of the isorevenue allows a changing slope that becomes steeper the larger the amount of innovations introduced. Firms need to make irreversible decisions and yet they are not able to anticipate correctly all the possible consequences of their decisions in the long term. In these circumstances firms can rely only upon sequential decision making based upon procedural rationality. Only new solutions beyond E3 can engender an increase in total factor productivity (see Antonelli, 1995, 1999a). See Antonelli (1995, 1999a) for a detailed exposition. The firm can ‘discover’ to its surprise that the equilibrium amount of possible adjustments makes it possible to introduce a total factor productivity increasing technological change which leads the firm beyond equilibrium point E3 (see Figure 13.4). This is clearly a case for procedural rationality as opposed to substantive rationality (Simon, 1982). In other models of this kind only changes in fixed capital were assumed to yield switching costs (see Antonelli, 2001). Appropriate tuning of the parameters of equation (4) can express a range of conditions, including the case in which switching costs depend almost exclusively upon the required changes in fixed capital, or in human capital, or in both. In this model the firm considers the possibility of introducing new technology in all possible technical directions. The direction of the innovation activity is not bound by the techniques in use. Localised learning takes place in the technique, defined in terms of input intensity, in use at each point in time, but it makes it possible to move in all directions so as to reshape the map of isoquants. Equation (4) defines the metrics of the switching activities that are necessary. Technical knowledge here defines the knowledge that is necessary to move from one technique to any other on the existing map of isoquants. For the sake of clarity the rest of the analysis considers only the case where there is a change in equilibrium condition determined by a change in factor costs. The model however can be easily extended so as to be used to analyse the consequences of a change in the levels of the demand for the products of the firm and hence in a parallel shift of the isocosts. The specification of the costs of all possible technical changes in the space of existing techniques can easily accommodate the changes that are caused by any increase or decrease in the demand levels and hence in switching from one isoquant to another. The demand-pull hypothesis can be integrated in the analytics of the economics of localised technological change (Antonelli, 1995). See the well-known debate between Alchian (1950) and Friedman (1953) about the role of mistakes in economic theory.
14 The system dynamics of collective knowledge: from gradualism and saltationism to punctuated change 1 See Rosser (1991: 138–40). 2 See Marshall (1920: 368): ‘nothing of this is true in the world in which we live. Here every economic force is constantly changing its action, under the influence of other forces which are acting around it. Here changes in the volume of production, in its methods, and in its costs are ever mutually modifying one another; they are always affecting and being affected by the character and the extent of demand. Further all these mutual influences take time to work themselves out, and, as a rule, no two influences move at equal pace. In this world therefore every plain and simple doctrine as to the relations between cost of production, demand and value is necessarily false’. 3 We shall assume that the unit costs and the efficiency of switching activities are not influenced by relevant externalities. 4 See equations (5) and (6) in Chapter 13. 5 Equation (7) can be considered the fully specified form of equation (5) in Chapter 13. 6 Equation (15) has its solution in the standard logistic function.
Notes 375 7 Intentional mobility of agents in relevant topological spaces matters for complex dynamic system analysis, where often agents are allowed to change their location only at random (Rosser, 2004). 15 Factor markets: constraints and inducements to innovation 1 Hicks (1976) provides a clear definition of the inducement hypothesis: ‘An induced invention is a change in technique that is made as a consequence of a change in prices (or, in general, scarcities); if the change in prices had not occurred, the change in technique would not have been made. I now like to think of a major technical change (one that we may agree to regard as autonomous, since, for anything that we are concerned with, it comes from outside) as setting up what I call an Impulse. If the autonomous change is an invention which widens the range of technical possibilities, it must begin by raising profitability and inducing expansion; but the expansion encounters scarcities, which act as a brake. Some of the scarcities may be just temporary bottle-necks which in time can be removed; some, however, may be irremovable. Yet it is possible to adjust to either kind of scarcity by further changes in technical methods; it is these that are the true induced inventions. The whole story, when it is looked at in this way, is in time, and can be in history’ (Hicks, 1976/1982: 295–6) (italics in the original text). 2 Thus the early distinctions, in technical progress literature, based on factor intensity can be applied in a new context (Robinson, 1937; Asimakopulos and Weldon, 1963; Blaug, 1963; Amendola, 1976; Besomi, 1999). 3 The notion of general-purpose technology contributes to our understanding of the distinction between general and contingent technological changes. General technological change is in fact a general-purpose technology with a wide range of applications in terms of both products and regions. Contingent technological changes on the contrary apply to a limited range of techniques and have a limited range of application (Bresnahan and Trajtenberg, 1995; Helpman, 1998). 4 The metrics of technological change are defined in terms of rates of overall factor productivity, while the metrics of technical change are provided by equation (6). 5 When the case for output maximisation or cost minimisation respectively applies. 6 Paul David suggested some time ago that the decoupling of the inducement to innovate from the inducement of the direction of technological change was a fertile area of investigation. Little work however has been done along these lines since then. See David: ‘As soon as one is ready to discard the neoclassical conception of technological progress which insists that innovation and factor substitution be viewed as logically distinct phenomena, there is no longer any great difficulty in taking an important step toward this proximate objective. Specifically it becomes possible to indicate how the realized factor-saving bias of “changes in the state of technical arts” may come under the influence of factor-prices directly, as well as indirectly through the medium of choice of technique decisions. In regard to the latter, we may for the present purposes eschew less orthodox “behavioral” approaches to the decision making of firms; the prevailing structure of input prices will therefore continue to be cast in the governing role assigned to them by the traditional theory of rational, cost-minimizing firm’ (David, 1975: 57–8; see Antonelli, 1989, 1990 for an exploratory attempt to develop this point). 7 Strong assumptions about the full rationality and foresight of firms are not necessary. Myopic but reactive and creative firms can innovate in a variety of directions. Only the new technology which makes the best use of locally abundant production factors will be chosen in the product markets. Rivalry in product markets can be considered a reliable selection mechanism – a Schumpeterian Darwinism or even better a Schumpeterian Lamarckism – which is able to choose ex post the correct direction of technological change.
376 Notes 8 The social construction of a technology does in fact take place but it is the lengthy result of the interplay between the rate and the direction of technological change shaped by such a process of selective innovation and diffusion (Bijker et al., 1987). 16 Localised product innovation: the role of proximity in the Lancastrian product space 1 The distinction elaborated by Philip Nelson between search goods and experience goods is relevant in this context: it is clear that proximity matters more for experience goods, especially when durable products are considered (Nelson, 1970). 2 See also Pavitt et al. (1989). 3 I acknowledge the comments and suggestions of an anonymous referee. 4 Even more explicitly Rumelt notes that ‘The implication is that both very little and very great diversity produce equivalent variability in earnings, but carefully controlled diversity is the best form of diversification for reducing fluctuations in earnings. We suggest that the factor responsible for this result is the ability of the diversified, but “constrained” firm to employ the beneficial effects of negatively correlated returns by replacing products that are stagnant or declining, with close functional substitutes that are profitable and are growing for reasons related to the decline of the original products. In other words, the “constrained” firm ties its fortunes to the satisfaction of a particular functional need (such as business-information-processing, permanent-image making, or convenience goods). It then develops a variety of products that relate to the fairly constant need and continuously searches its product line for weak points, innovating to meet changing tastes and needs’ (Rumelt, 1986: 157). 5 Much theoretical work on product differentiation and the effects and causes of product and brand variety has been based upon a model of spatial rivalry derived from the wellknown Hotelling model of spatial duopoly. Little attention has been paid in this context to the analysis of technological change and the choice between product and process innovation (Martin, 1993). 6 The firm can ‘discover’ to its surprise that the equilibrium amount of possible adjustments makes it possible to introduce a total factor productivity increasing technological change which leads the firm beyond equilibrium point E (see Figure 16.1). This is clearly a case for procedural rationality as opposed to substantive rationality (Simon, 1982). 7 RF is set by the amount of resources the myopic firm, unable to anticipate the ‘technological surprise’, should in any case invest in order to switch. 8 Stochastic factors play a major role here. Technological change should be viewed as the lucky outcome of a fragile set of highly qualified conditions. Successful – total factor productivity enhancing – innovations are based upon the localised competence and experience of agents induced to innovate by the mismatch between expectations and real market conditions and irreversibility, but are really possible only when a large number of complementary conditions apply. Among such conducive conditions, colocalisation within technological districts, effective communication flows with other firms active in upstream and downstream industries as well as in the industry, creative interactions with university research centres, complementarity with other innovation processes, and access to financial resources play a key role. When such conditions are lacking, the reaction of firms is limited to sheer technical change, i.e. traditional substitution along existing isoquants. 9 See Lang and Stultz (1994) for a clear result and Besanko et al. (2000) for a general survey on these findings. 10 The statistics available on R&D activities capture quite well the actual amount of codified knowledge firms can command. 17 Diffusion as a process of creative adoption 1 See Nathan Rosenberg: ‘The criticisms which I have levelled thus far against the artificial segregation of invention from innovation apply equally well to the segregation of
Notes 377 invention from diffusion. Innovation is simply the beginning of the diffusion process. However, here again we have inherited from the Schumpeterian framework a sharp disjunction which emphasizes the high levels of leadership and creativity involved in the first introduction of a new technique as compared to the mere imitative activity of subsequent adopters. Here also, as a result, the analysis of the diffusion process fails to focus upon continued technological and engineering alterations and adaptations, the cumulative effects of which decisively influence the volume and the timing of the product’s sale. The diffusion process is typically dependent upon a stream of improvement in the performance characteristics of an innovation, its progressive modification and adaptation to suit the specialized requirements of various submarkets, and the availability and introduction of other complementary inputs which affect the economic usefulness of an original innovation’ (Rosenberg, 1976: 75). 2 The firm can ‘discover’ to its surprise that the equilibrium amount of possible adjustments makes it possible to introduce a total factor productivity increasing technological change which leads the firm beyond equilibrium point E (see Figure 17.1). This is clearly a case for procedural rationality as opposed to substantive rationality (Simon, 1982). 3 RF is set by the amount of resources the myopic firm, unable to anticipate the ‘technological surprise’, should in any case invest in order to switch. 18 Path dependence and the quest for complexity 1 It seems important to note that Milton Friedman did base his remarks upon the results of the first wave of enthusiasm raised in economics by the identification of the replicator dynamics in biology. Friedman provided an insightful interpretation of the replicator dynamics as a mechanism able to lead a system of heterogeneous and sub-rational agents to general equilibrium. Thus the replicator dynamics and Darwinist selection at large were interpreted as the foundations for the claim that an objective rationality exists in the system and it can compensate for the lack of subjective rationality. Provided that agents are allowed to react creatively, this result holds. Surprisingly enough the second wave of applications of the replicator dynamics to economics towards the end of the twentieth century completely missed the results of the first wave. Both yesterday and today it should be noticed however that the mistakes of consumers cannot be cleared by the replicator dynamics of market forces: stupid consumers, unable to choose the right products, will enjoy lower levels of welfare, and no mechanism will take care to force them out of the system (hopefully). 2 As Tony Lawson (2005) notes, this is the basic assumption upon which mainstream economics is built: ‘[mainstream economics] is a version of deductivism that posits functional relations presupposing closures of causal consequence. The latter are closed systems in which the events correlated are such that one set (conceptualized as “independent variables”) are considered to stand in the causal history of the remaining events (“the dependent variables”)’ (Lawson, 2005: 494). 3 The ontological approach elaborated by Tony Lawson seems to share many key elements with the approach elaborated so far: ‘The social realm is also highly interconnected and organic. Fundamental here is the prevalence of internal social relations. Relations are said to be internal when they relate to each other in what they are and/or can do what they do, just in virtue of the relation to each other in which they stand . . . The social real is also found to be structured. By this I mean that it does not consist just in one ontological level. In particular, it does not reduce to human practices and other actualities but includes underlying social structures and process of the sort just noted and their power and tendencies. A further fundamental category of the ontological conception I am laying out is that of emergence. A stratum of reality can be said to be emergent, or as possessing emergent powers, if there is a sense in which it (1) has arisen out of a lower stratum, being formed by principles operative at the lower level, (2) remains independent of the lower strata for its existence, but (3) contains causal
378 Notes
4
5 6
7
powers of its own which are both irreducible to those operating at the lower level and (perhaps) capable of acting back on the lower level. Thus organic material emerged from inorganic material. And, according to the conception I am defending, the social realm is emergent from human (inter)action, though with properties irreducible to, yet capable of causally affecting the latter (Lawson, 2005: 495, 496). The criticisms addressed to the notion of path dependence in a static context of analysis, one where the notions of technological change and creativity are not even considered, have little relevance in our framework of analysis. As Rizzello (2004: 265) correctly considers: ‘Margolis’ & Leibowitz’s criticism addresses at least three different levels. The first one is a formal level: they maintain that ideas similar to the path-dependence in economics are dealt with in physics and mathematics by means of the chaos theory, while in biology analogous phenomena are contingency and the irreversible character of natural selection. Through more sophisticated analytical tools it would be possible to explain phenomena which only appear to be the consequence of suboptimal selection, and the idea of the efficiency of the market’s selective process would be confirmed. The second level of criticism distinguishes three degrees of path-dependence. The first one is quite weak: here “path-dependent” means that what we can observe today critically depends on previously prevailing conditions and on decisions made in the past. The second level refers to those paths with a form of dependence linked to past conditions, which lead to determined and specific results, in order to reach different results it is necessary to bear relevant costs. At the third level of path dependence – the strongest – there are remediable inefficiencies. Whereas the first degree of path dependence simply postulates the existence of an inter-temporal relationship, which does not imply prediction errors or inefficient results, the second degree concludes that inter-temporal effects, along with imperfect predictions, bring about unpleasant – though non inefficient – actions; the third degree of path-dependence implies not only that inter-temporal effects increase errors, but also that these errors could be avoided. The first and the second degrees are very common and they can easily be dealt with by economic theory. The third degree challenges the idea of market efficiency and the “invisible-hand theorems that private optimization leads individuals to wealth maximizing allocation”. As should be clear from the analysis developed so far, our understanding of the notion of path dependence is intrinsically dynamic in that it considers what effects irreversibility and local externality have upon the rate and the direction of the generation of new technological knowledge and the introduction of new technologies’. However, Bachelier introduced it for financial prices in 1900, before Einstein used it in 1905. The reference to complexity theory here is clear. According to Mark Taylor (2001: 142–3), ‘it is possible to identify the following characteristics of complex systems: 1. Complex systems are comprised of many different parts, which are connected in multiple ways. 2. Diverse components can interact both serially and in parallel to generate sequential as well as simultaneous effects and events. 3. Complex systems display spontaneous self-organization, which complicates interiority and exteriority in such a way that the line that is supposed to separate them becomes undecidable. 4. The structures resulting from spontaneous self-organization emerge from but are not necessarily reducible to the interactivity of the components or elements in the system. 5. Though generated by local interactions, emergent properties tend to be global. 6. Inasmuch as self-organizing structures emerge spontaneously, complex systems are neither fixed not static but develop or evolve. Such evolution presupposes that complex systems are both open and adaptive. 7. Emergence occurs in a narrow possibility space lying between conditions that are too ordered and too disordered. This boundary or margin is the “edge of chaos”, which is always far from equilibrium’. See Sylos Labini (1956, 1984); Momigliano (1975); Scherer (1984, 1992); Burt (1992); and Audretsch (1995).
References
Abramovitz, M. (1956), Resources and output trends in the US since 1870, American Economic Review 46, 5–23. Acemoglu, D. (2002), Directed technical change, Review of Economic Studies 69, 781–810. Aghion, P. and Howitt, P. (1992), A model of growth through creative destruction, Econometrica 60, 323–51. Aghion, P. and Tirole, J. (1994), The management of innovation, Quarterly Journal of Economics CIX, 1185–209. Aghion, P. and Howitt, P. (1998), Endogenous growth theory, MIT Press, Cambridge, MA. Akerlof, G. A. (1970), The market for lemons: Quality uncertainty and the market mechanism, Quarterly Journal of Economics 84, 488–500. Alchian, A. A. (1950), Uncertainty evolution and economic theory, Journal of Political Economy 58, 211–22. Alchian, A. A. and Demsetz, H. (1972), Production information costs and economic organization, American Economic Review 62, 777–95. Alchian, A. and Demsetz, H. (1973), The property rights paradigm, Journal of Economic History 33, 16–27. Allen, R. C. (1983), Collective invention, Journal of Economic Behaviour and Organization 4, 1–24. Amatori, F. and Colli, A. (1999), Impresa e industria in Italia dall’Unità a oggi, Marsilio, Venezia. Amendola, M. (1976), Macchine produttività e progresso tecnico, ISEDI, Milano. Amendola, M. and Gaffard, J. L. (1988), The innovative choice: An economic analysis of the dynamics of technology, Basil Blackwell, Oxford. Amendola, M. and Gaffard, J. L. (1998), Out of equilibrium, Oxford University Press, Oxford. Amin, A. and Cohendet, P. (2005), Geographies of knowledge formation in firms, Industry and Innovation 12, 471–92. Anand, N. B. and Tarun, K. (2000), The structure of licensing contracts, Journal of Industrial Economics 48, 103–35. Anderson, P. W., Arrow, K. J. and Pines, D. (eds) (1988), The economy as an evolving complex system, Addison-Wesley, Redwood City, CA. Antonelli, C. (1982), Cambiamento tecnologico e teoria dell’impresa, Loescher, Torino. Antonelli, C. (ed.) (1988), New information technology and industrial change: The Italian case, Kluwer Academic Publishers, Boston, MA and Dordrecht. Antonelli, C. (1989), A failure-inducement model of research and development expenditures: Italian evidence from the early 1980s, Journal of Economic Behavior and Organization 12, 159–80.
380 References Antonelli, C. (1990), Induced adoption and externalities in the regional diffusion of new information technology, Regional Studies 24, 31–40. Antonelli, C. (1991), The diffusion of advanced telecommunications in developing countries, OECD, Paris. Antonelli, C. (ed.) (1992), The economics of information network, Elsevier, Amsterdam. Antonelli, C. (1993), Investment and adoption in advanced telecommunications, Journal of Economic Behavior and Organization 20, 227–46. Antonelli, C. (1995), The economics of localized technological change and industrial dynamics, Kluwer Academic Publishers, Boston, MA. Antonelli, C. (1999a), The microdynamics of technological change, Routledge, London. Antonelli, C. (1999b), The organization of production, Metroeconomica 50, 234–53. Antonelli, C. (2000), Collective knowledge communication and innovation: The evidence of technological districts, Regional Studies 34, 535–47. Antonelli, C. (2001), The microeconomics of technological systems, Oxford University Press, Oxford. Antonelli, C. (2002), Innovation and structural change, Economie Appliquée 55, 85–120. Antonelli, C. (2003a), The economics of innovation, new technologies and structural change, Routledge, London. Antonelli, C. (2003b), Knowledge complementarity and fungibility: Implications for regional strategy, Regional Studies 39, 595–606. Antonelli, C. (2003c), The digital divide: Understanding the economics of new information and communication technology in the global economy, Information Economics and Policy 15, 173–99. Antonelli, C. (2004a), The governance of localized technological knowledge and the evolution of intellectual property rights, in E. Colombatto (ed.), The Elgar companion to property rights, Edward Elgar, Cheltenham, pp. 414–37. Antonelli, C. (2004b), Localized product innovation: The role of proximity in the Lancastrian product space, Information Economics and Policy 16, 255–74. Antonelli, C. (2005a), The economics of governance: The role of localized technological change in the interdependence among transaction, coordination and production, in K. Green, M. Miozzo and P. Dewick (eds), Technology, knowledge and the firm: Implications for strategy and industrial change, Edward Elgar, Cheltenham, pp. 29–50. Antonelli, C. (2005b), Models of knowledge and systems of governance, Journal of Institutional Economics 1, 51–73. Antonelli, C. (2006a), Localized technological change and factor markets: Constraints and inducements to innovation, Structural Change and Economic Dynamics 17, 224–47. Antonelli, C. (2006b), The governance of technological knowledge: To use or to sell, in J. Sundbo, A. Gallina, G. Serin and J. Davis (eds), Contemporary management of innovation, PalgraveMacmillan, Basingstoke, pp. 208–29. Antonelli, C. (2006c), The business governance of localized knowledge: An information economics approach to the economics of knowledge, Industry and Innovation 13, 227–61. Antonelli, C. (2006d), Diffusion as a process of creative adoption, Journal of Technology Transfer 31, 211–26. Antonelli, C. (2006e), Path dependence, localised technological change and the quest for dynamic efficiency, in C. Antonelli, D. Foray, B. H. Hall and W. E. Steinmueller (eds), New frontiers in the economics of innovation and new technology: Essays in honour of Paul A. David, Edward Elgar, Cheltenham, pp. 51–69.
References 381 Antonelli, C. (2007a), The foundations of the economics of innovation, in C. Antonelli (ed.), The economics of innovation: Critical concepts in economics, Routledge, London. Antonelli, C. (2007b), Knowledge as an essential facility, Journal of Evolutionary Economics 17: 451–71. Antonelli, C. (2007c), The system dynamics of collective knowledge: From gradualism and saltationism to punctuated change, Journal of Economic Behavior and Organization 62, 215–36. Antonelli, C. (2007d), The new economics of the university: A knowledge governance approach, Journal of Technology Transfer, forthcoming. Antonelli, C. and Quéré, M. (2002), The governance of interactive learning within innovation systems, Urban Studies 39, 1051–63. Antonelli, C. and Barbiellini Amidei, F. (2007), Innovazione tecnologica e mutamento strutturale nell’industria italiana nel secondo dopoguerra, in C. Antonelli et al., Innovazione tecnologica e sviluppo industriale nel secondo dopoguerra, Collana Storica della Banca d’Italia, Laterza, Roma, pp. 3–358. Antonelli, C. and Teubal, M. (2007), Knowledge intensive property rights and the evolution of venture capitalism, Journal of Institutional Economics, forthcoming. Antonelli, C. and Calderini, M. (2008), The governance of knowledge compositeness and technological performance: The case of the automotive industry in Europe, Economics of Innovation and New Technology 18, forthcoming. Antonelli, C., Petit, P. and Tahar, G. (1992), The economics of industrial modernization, Academic Press, Cambridge. Antonelli, C., Marchionatti, R. and Usai, S. (2003), Productivity and external knowledge: The Italian case, Rivista Internazionale di Scienze Economiche e Commerciali 50, 69–90. Archibugi, D. and Lundvall, B. A. (2001), The globalizing learning economy: Major socioeconomic trends and European innovation policy, Oxford University Press, Oxford. Argyres, N. S. (1995), Technology strategy governance structure and interdivisional coordination, Journal of Economic Behavior and Organization 28, 337–58. Arora, A. and Gambardella, A. (1994), The changing technology of technical change: General and abstract knowledge and the division of innovative labor, Research Policy 23, 523–32. Arora, A., Fosfuri, A. and Gambardella, A. (2001), Markets for technology, MIT Press, Cambridge, MA. Arrow, K. J. (1962a), Economic welfare and the allocation of resources for invention, in R. R. Nelson (ed.), The rate and direction of inventive activity: Economic and social factors, Princeton University Press for NBER, Princeton, NJ, pp. 609–25. Arrow, K. J. (1962b), The economic implications of learning by doing, Review of Economic Studies 29, 155–73. Arrow, K. J. (1969), Classificatory notes on the production and transmission of technical knowledge, American Economic Review 59, 29–35. Arrow, K. J. (1974), The limits of organization, W.W. Norton, New York. Arthur, B. (1989), Competing technologies, increasing returns and lock-in by small historical events, Economic Journal 99, 116–31. Arthur, B. (1999), Complexity and the economy, Science 284, 107–9. Arthur, W. B., Durlauf, S. N. and Lane, D. (eds) (1997), The economy as an evolving complex system II, Addison-Wesley, Redwood City, CA. Arvanitis, S. and Hollenstein, H. (2001), The determinants of the adoption of advanced manufacturing technology, Economics of Innovation and New Technology 10, 377–414.
382 References Asimakopulos, A. and Weldon, J. C. (1963), The classification of technical progress in models of economic growth, Economica 30, 372–86. Atkinson, A. B. and Stiglitz, J. E. (1969), A new view of technological change, Economic Journal 79, 573–8. Attalah, G. (2003), Information sharing and the stability of cooperation in research joint ventures, Economics of Innovation and New Technology 12, 531–55. Audretsch, D. B. (1995), Innovation and industry evolution, MIT Press, Cambridge, MA. Audretsch, D. B. (2006), Entrepreneurship, innovation and economic growth, Edward Elgar, Cheltenham. Audretsch, D. B. and Feldman, M. (1996), Spillovers and the geography of innovation and production, American Economic Review 86, 630–40. Audretsch, D. B. and Stephan, P. E. (1996), Company–scientist locational links: The case of biotechnology, American Economic Review 86, 641–52. Avnimelech, G. and Teubal, M. (2004), Venture capital start-up co-evolution and the emergence and development of Israel’s new high tech cluster, Economics of Innovation and New Technology 13, 33–60. Bairati, P. (1983), Valletta, UTET, Torino. Baldwin, C. Y. and Clark, K. (2000), Design rules: The power of modularity, MIT Press, Cambridge, MA. Barabasi, A. L. (2002), Linked: How everything is connected to everything else and what it means for business, science and everyday life, Plume, New York. Barca, F. (ed.) (1997), Storia del capitalismo italiano dal dopoguerra a oggi, Donzelli, Roma. Barton, J. (2000), Reforming the patent system, Science 287, 1933–4. Baumol, W. J. (2002), The free-market innovation machine: Analyzing the growth miracle of capitalism, Princeton University Press, Princeton, NJ. Baumol, W. J. and Sydak, J. G. (1994), Toward competition in local telephony, MIT Press, Cambridge, MA. Besanko, D., Dranove, D. and Shanley, M. (2000), Economics of strategy, John Wiley & Sons, New York. Besomi, D. (1999), Harrod on the classification of technological progress: The origin of a wild-goose chase, BNL Quarterly Review 208, 95–118. Bianchi, M. (ed.) (1998), The active consumer, Routledge, London. Bijker, W. E., Hughes, T. P. and Pinch, T. J. (eds) (1987), The social construction of technological systems: New directions in the sociology and history of technology, MIT Press, Cambridge, MA. Binswanger, H. P. and Ruttan, V. W. (1978), Induced innovation: Technology institutions and development, Johns Hopkins University Press, Baltimore, MD. Blaug, M. (1963), A survey of the theory of process-innovations, Economica 30, 13–32. Boldrin, M. and Levine, D. K. (2002), The case against intellectual property rights, American Economic Review 92, 209–12. Bonazzi, G. and Antonelli, C. (2003), To make or to sell? The case of in-house outsourcing at FIAT Auto, Organization Studies 24, 575–94. Boschma, R. A. (2005), Proximity and innovation: A critical assessment, Regional Studies 39, 61–74. Bresnahan, T. F. and Trajtenberg, M. (1995), General purpose technologies: ‘Engines of growth’?, Journal of Econometrics 65, 83–108. Bresnahan, T. and Gambardella, A. (eds) (2005), Building high-tech clusters: Silicon Valley and beyond, Cambridge University Press, Cambridge.
References 383 Broadberry, S. (1997), The productivity race: British manufacturing in international perspective, 1850–1990, Cambridge University Press, Cambridge. Brousseau, E. and Fares, M. (2000), Incomplete contracts and governance structures: Are incomplete contract theory and new institutional economics substitutes or complements?, in C. Menard (ed.) (2000), Institutions, contracts and organizations: Perspectives from new institutional economics, Edward Elgar, Cheltenham. Brousseau, E. and Glachant, J. M. (eds) (2002), The economics of contracts, Cambridge University Press, Cambridge. Brown, M. (1966), On the theory and measurement of technological change, Cambridge University Press, Cambridge. Brusoni, S. and Prencipe, A. (2001), Unpacking the black box of modularity: Technologies, products and organizations, Industrial and Corporate Change 10, 179–205. Buchanan, J. M. (1965), An economic theory of clubs, Economica 34, 1–14. Burt, R. S. (1992), Structural holes: The social structure of competition, Harvard University Press, Cambridge, MA. Canepa, A. and Stoneman, P. (2004), Comparative international diffusion: Patterns, determinants and policies, Economics of Innovation and New Technology 13, 279–98. Cantwell, J. and Iammarino, S. (2003), Multinational corporations and European regional systems of innovation, Routledge, London. Cassier, M. and Foray, D. (2002), Public knowledge, private property and the economics of high-tech consortia, Economics of Innovation and New Technology 11, 123–32. Casti, J. (1998), Would-be worlds: How simulation is changing the frontiers of science, Wiley, New York. Castronovo, V. (1971), Giovanni Agnelli, UTET, Torino. Castronovo, V. (1995), Storia economica d’Italia: Dall’Ottocento ai giorni nostri, Einaudi, Torino. Caves, R. E. and Porter, M. E. (1977), From entry barriers to mobility barriers: Conjectural decision and contrived deterrence to new competition, Quarterly Journal of Economics 91, 241–61. Chandler, A. D. (1962), Strategy and structure: Chapters in the history of the industrial enterprise, MIT Press, Cambridge, MA. Chandler, A. D. (1977), The visible hand: The managerial revolution in American business, Belknap Press of Harvard University Press, Cambridge, MA. Chandler, A. D. (1990), Scale and scope: The dynamics of industrial capitalism, Belknap Press of Harvard University Press, Cambridge, MA. Chandler, A. D., Hagstrom, P. and Solvell, O. (eds) (1999), The dynamic firm: The role of technology, strategy, organization and regions, Oxford University Press, Oxford. Chesbrough, H. (2003), Open innovation: The new imperative for creating and profiting from technology, Harvard Business School Press, Boston, MA. Chesbrough, H., Vanhaverbeke, W. and West, J. (2006), Open innovation: Researching a new paradigm, Oxford University Press, Oxford. Clark, G. L. (2000), Pension fund capitalism, Oxford University Press, Oxford. Cohen, W. M. and Levinthal, D. A. (1989), Innovation and learning: The two faces of R&D, Economic Journal 99, 569–96. Cohen, W. M. and Levinthal, D. A. (1990), Absorptive capacity: A new perspective on learning and innovation, Administrative Science Quarterly 35, 128–52. Cohen, W. M., Nelson, R. R. and Walsh, J. P. (2000), Protecting their intellectual assets: Appropriability conditions and why U.S. manufacturing firms patent (or not), NBER, Cambridge, MA.
384 References Consoli, D. (2005a), Technological cooperation and product substitution in UK retail banking: The case of consumer services, Information Economics and Policy 17, 199–216. Consoli, D. (2005b), The dynamics of technological change in UK retail banking services: An evolutionary perspective, Research Policy 34, 461–80. Cooke, P. (2002), Knowledge economies, Routledge, London. Cowan, R. and Foray, D. (1997), The economics of codification and the diffusion of knowledge, Industrial and Corporate Change 6, 595–622. Cowan, R., David, P. A. and Foray, D. (2000), The explicit economics of knowledge codification and tacitness, Industrial and Corporate Change 9, 211–53. Cyert, R. M. and March, J. C. (1963), A behavioral theory of the firm, Prentice-Hall, Englewood Cliffs, NJ. Dalle, J. M. and David, P. A. (2003), The allocation of software development resources in open source production mode, SIEPR Discussion Paper No. 02-27. Dalle, J. M., David, P. A., Ghosh, A. R. and Steinmueller, W. E. (2004), Advancing economic research on the free and open source software mode of production, SIEPR Discussion Paper No. 04-03. Dasgupta, P. and David, P. A. (1987), Information disclosure and the economics of science and technology, in G. Feiwel (ed.), Arrow and the ascent of modern economic theory, Macmillan, London. Dasgupta, P. and David, P. A. (1994), Towards a new economics of science, Research Policy 23, 487–521. Dasgupta, P. and Seragekdin, I. (eds) (2000), Social capital: A multifaceted perspective, World Bank, Washington, DC. David, P. A. (1969), A contribution to the theory of diffusion. Research Center in Economic Growth, Memorandum No. 71, Stanford University. David, P. A. (1975), Technical choice innovation and economic growth, Cambridge University Press, Cambridge. David, P. A. (1985), Clio and the economics of QWERTY, American Economic Review 75, 332–7. David, P. A. (1987), Some new standards for the economics of standardization in the information age, in P. Dasgupta and P. Stoneman (eds), Economic policy and technological performance, Cambridge University Press, Cambridge, pp. 206–40. David, P. A. (1988), Path-dependence: Putting the past into the future of economics. Mimeo, Department of Economics, Stanford University. David, P. A. (1990), The dynamo and the computer: A historical perspective on the productivity paradox, American Economic Review (P&P) 80, 355–61. David, P. A. (1992), Path-dependence in economic processes: Implications for policy analysis in dynamical system contexts, Fondazione Rosselli, Torino. David, P. A. (1993), Knowledge property and the system dynamics of technological change, Proceedings of the World Bank Annual Conference on Development Economics, World Bank, Washington, DC. David, P. A. (1994a), Why are institutions the ‘carriers of history’? Path dependence and the evolution of conventions, organizations and institutions, Structural Change and Economic Dynamics 5, 205–20. David, P. A. (1994b), Positive feedbacks and research productivity in science: Reopening another black box, in O. Granstrand (ed.), Economics and technology, Elsevier, Amsterdam. David, P. A. (2000a), Path dependent learning, and the evolution of beliefs and behaviours,
References 385 in U. Pagano and A. Nicita (eds), The evolution of economic diversity, Routledge, London. David, P. A. (2000b), Path dependence and varieties of learning in the evolution of technological practice, Ch. 10 in J. Ziman (ed.), Technological innovation as an evolutionary process, Cambridge University Press, Cambridge. David, P. A. (2004a), Can ‘open science’ be protected from the evolving regime of IPR protections?, Journal of Institutional and Theoretical Economics 160, 9–34. David, P. A. (2004b), Understanding the emergence of Open Science institutions: Functionalist economics in historical context, Industrial and Corporate Change 13, 571–89. David, P. A. (2004c), The tale of two traverses: Innovation and accumulation in the first two centuries of U.S. economic growth. Gallman Lecture, Duke University. David, P. A. (2007), Path dependence: A foundational concept for historical social science, Cliometrica: Journal of Historical Economics and Econometric History 1: 91–114. David, P. A. and Bunn, J. (1988), The economics of gateway technologies and network evolution: Lessons from electricity supply history, Information Economics and Policy 3, 165–202. Dawkins, R. (1986), The blind watchmaker: Why the evidence of evolution reveals a universe without design, W. W. Norton, New York. Day, R. H. (1983), The emergence of chaos from classical economic growth, Quarterly Journal of Economics 98, 201–13. Demsetz, H. (1967), Towards a theory of property rights, American Economic Review 57, 347–59. D’Ignazio, A. and Giovannetti, E. (2007), From exogenous to endogenous economic networks: Internet applications, Journal of Economic Surveys 20, 756–96. Dosi, G. and Nelson, R. R. (1994), An introduction to evolutionary theories in economics, Journal of Evolutionary Economics 4, 153–72. Dosi, G., Nelson, R. R. and Winter, S. (2000), The nature and dynamics of organizational capabilities, Oxford University Press, Oxford. Durlauf, S. N. (2005), Complexity and empirical economics, Economic Journal 115, 225–43. Eldredge, N. and Gould, S. J. (1972), Punctuated equilibria: An alternative to phyletic gradualism, in T. J. Schopfer (ed.), Models in paleobiology, Freeman Cooper, San Francisco, CA, pp. 82–115. Elster, J. (1983), Explaining technical change, Cambridge University Press, Cambridge. Elster, J. (1986), An introduction to Karl Marx, Cambridge University Press, Cambridge. Etzkowitz, H. (1998), The norms of entrepreneurial science: Cognitive effects of the new university–industry linkages, Research Policy 27, 823–33. Etzkowitz, H. (2002), MIT and the rise of entrepreneurial science, Routledge, London. Etzkowitz, H. and Leydesdorf, L. (2000), The dynamics of innovation: From national systems and ‘Mode 2’ to a triple helix of university–industry–government relations, Research Policy 29, 109–23. Fai, F. and Von Tunzelmann, N. (2001a), Scale and scope in technology: Large firms 1930/1990, Economics of Innovation and New Technology 10, 255–88. Fai, F. and Von Tunzelmann, N. (2001b), Industry-specific competencies and converging technological systems: Evidence from patents, Structural Change and Economic Dynamics 12, 141–70. Faria, A., Fenn, P. and Bruce, A. (2002), Determinants of the adoption of flexible production technologies: Evidence from Portuguese manufacturing industry, Economics of Innovation and New Technology 11, 560–80.
386 References Faria, A., Fenn, P. and Bruce, A. (2003), Count data model of technology adoption, Journal of Technology Transfer 28: 63–79. Farrell, M. J. (1957), The measurement of productive efficiency, Journal of the Royal Statistical Society Series A (General): 253–90. Faust, K. and Wasserman, S. (1994), Social network analysis: Methods and applications, Cambridge University Press, Cambridge. Feldman, M. P. (1993), An examination of the geography of innovation, Industrial and Corporate Change 2, 447–67. Feldman, M. P. (1994), The geography of innovation, Kluwer Academic Publishers, Boston, MA. Feldman, M. P. (1999), The new economics of innovation spillovers and agglomeration: A review of empirical studies, Economics of Innovation and New Technology 8, 5–26. Feldman, M. P. and Audretsch, D. B. (1999), Innovation in cities: Science-based diversity specialization and localized competition, European Economic Review 43, 409–30. Feldman, M. P. and Massard, N. (eds) (2002), Institutions and systems in the geography of innovation, Kluwer Academic Publishers, Boston, MA. Florida, L. and Kenney, M. (1988), Venture capital financed innovation and technological change in the USA, Research Policy 17, 119–37. Foray, D. and Steinmueller, W. E. (2003), On the economics of R&D and technological cooperation: Insights and results from the project COLLINE, Economics of Innovation and New Technology 12, 77–92. Foss, N. (1997), Resources, firms and strategies: A reader in the resource-based perspective, Oxford University Press, Oxford. Foss, N. J. (1998), The resource-based perspective: An assessment and diagnosis of problems, Scandinavian Journal of Management 15, 1–15. Foss, N. and Mahnke, V. (eds) (2000), Competence, governance and entrepreneurship: Advances in economic strategy research, Oxford University Press, Oxford. Foster, J. (1993), Economics and the self-organisation approach: Alfred Marshall revisited, Economic Journal 103, 975–91. Foster, J. (1997), The analytical foundation of evolutionary economics: From biological analogy to economic self-organization, Structural Change and Economic Dynamics 8, 427–51. Foster, J. (2005), From simplistic to complex systems in economics, Cambridge Journal of Economics 29, 873–92. Fransman, M. (2002), Telecoms in the Internet age: From boom to bust to . . .?, Oxford, Oxford University Press. Freeman, C. (1987), Technology policy and economic performance, Pinter, London. Freeman, C. (1994), The economics of technical change, Cambridge Journal of Economics 18, 463–514. Friedman, M. (1953), The methodology of positive economics, in M. Friedman, Essays in positive economics, University of Chicago Press, Chicago, IL. Garicano, L. (2000), Hierarchies and the organization of knowledge in production, Journal of Political Economy 108, 874–904. Garud, R. and Karnoe, P. (eds) (2001), Path dependence and creation, Lawrence Erlbaum Associates, Mahwah, NJ. Geroski, P. (1995a), Markets for technology, in P. Stoneman (ed.), Handbook of the economics of innovation and new technology, Basil Blackwell, Oxford, pp. 90–131. Geroski, P. (1995b), What do we know about entry?, International Journal of Industrial Organization 13, 421–40.
References 387 Geuna, A. (1999), The economics of knowledge production, Edward Elgar, Cheltenham. Geuna, A., Salter, A. and Steinmueller, W. E. (eds) (2003), Science and innovation: Rethinking the rationales for funding and governance, Edward Elgar, Cheltenham. Gibbons, M., Limoges, C., Nowotny, H., Schwarzman, S., Scott, P. and Trow, M. (1994), The new production of knowledge: The dynamics of research in contemporary societies, Sage Publications, London. Gleaser, E. and Scheinkman, J. A. (2000), Non-market interactions. NBER Working Paper Series 8053. Gompers, P. and Lerner, J. (1999), The venture capital cycle, MIT Press, Cambridge, MA. Gompers, P. and Lerner, J. (2003), The really long-run performance of initial public offerings: The pre-Nasdaq evidence, Journal of Finance 58, 1355–92. Gould, S. J. (2002), The structure of evolutionary theory, Harvard University Press, Cambridge, MA. Granovetter, M. S. (1973), The strength of weak ties, American Journal of Sociology 78, 1360–80. Gravelle, H. and Rees, R. (1981), Microeconomics, Longman, London and New York. Greve, H. R. (1998), Performance, aspirations, and risky organizational change, Administrative Science Quarterly 43, 58–86. Griffith, R., Redding, S. and Van Reenen, J. (2003), R&D and absorptive capacity: Theory and empirical evidence, Scandinavian Journal of Economics 105, 99–118. Griliches, Z. (1957), Hybrid corn: An exploration in the economics of technological change, Econometrica 25, 501–22. Griliches, Z. (1992), The search for R&D spillovers, Scandinavian Journal of Economics 94, 29–47. Grossman, S. J. and Hart, O. D. (1986), The costs and benefits of ownership: A theory of vertical integration, Journal of Political Economy 94, 691–719. Guilhon, B. (ed.) (2001), Technology and markets for knowledge: Knowledge creation, diffusion and exchange within a growing economy, Kluwer Academic Publishers, Boston, MA. Guilhon, B. (2004), Markets for knowledge: Problems, scope and economic implications, Economics of Innovation and New Technology 13, 165–82. Guiso, L. (1998), High-tech firms and credit rationing, Journal of Economic Behavior and Organization 35, 39–59. Guiso, L. and Schivardi, F. (2007), Spillovers in industrial districts, Economic Journal 117, 68–93. Habakkuk, H. J. (1962), American and British technology in the nineteenth century, Cambridge University Press, Cambridge. Hall, B. H. (2002), The financing of research and development, Oxford Review of Economic Policy 18, 35–51. Harabi, N. (1995), Appropriability of technical innovations: An empirical analysis, Research Policy 24, 981–92. Hart, O. D. (1995), Firms, contracts and financial structure, Oxford University Press, Oxford. Hayek, F. A. (1937), Economics and knowledge, Economica 13, 33–54. Hayek, F. A. (1945), The use of knowledge in society, American Economic Review 35, 519–30. Hayek, F. A. (1952), The sensory order: An inquiry into the foundations of theoretical psychology, Routledge & Kegan Paul, London.
388 References Helpman, E. (ed.) (1998), General purpose technologies and economic growth, MIT Press, Cambridge, MA. Henderson, R., Jaffe, A. B. and Trajtenberg, M. (1998), Universities as a source of commercial technology: A detailed analysis of university patenting, 1965–88, Review of Economics and Statistics 80, 119–27. Hicks, J. R. (1932), The theory of wages, Macmillan, London. Hicks, J. R. (1976), Time in economics, in A.M. Tang (ed.), Evolution welfare and time in economics. Quotes are drawn from J. R. Hicks (1982), Collected essays on economic theory, Vol. II, Basil Blackwell, Oxford. Holmstrom, B. (1989), Agency costs and innovation, Journal of Economic Behavior and Organization 12, 305–27. Hopenhayn, H. A. and Mitchell, M. F. (2001), Innovation variety and patent breadth, Rand Journal of Economics 32, 152–66. Jacobs, J. (1969), The economy of cities, Jonathan Cape, London. Jaffe, A. B. and Lerner, J. (2001), Reinventing public R&D: Patent policy and the commercialization of national laboratory technologies, Rand Journal of Economics 32, 167–98. Jaffe, A. B. and Lerner, J. (2004), Innovation and its discontents: How our broken patent system is endangering innovation and progress and what can be done about it, Princeton University Press, Princeton, NJ. Johnson, D. K. (2002), Learning-by-licensing: R&D and technology licensing in Brazilian invention, Economics of Innovation and New Technology 11, 163–77. Jones, C. I. (1998), Introduction to economic growth, Norton, New York (2nd edn, 2002). Kahneman, D. and Tversky, A. (1979), Prospect theory: An analysis of decision under risk, Econometrica 47, 263–91. Kahneman, D. and Tversky, A. (1992), Advances in prospect theory: Cumulative representation of uncertainty, Journal of Risk and Uncertainty 5, 297–323. Kaldor, N. (1972), The irrelevance of equilibrium economics, Economic Journal 82, 1237–55. Kaldor, N. (1981), The role of increasing returns, technical progress and cumulative causation, Economie Appliquée 34, 593–617. Karshenas, M. and Stoneman, P. (1992), A flexible model of technological diffusion, Journal of Forecasting 11, 577–602. Karshenas, M. and Stoneman, P. (1995), Technological diffusion, in P. Stoneman (ed.), Handbook of the economics of innovation and technological change, Blackwell, Oxford. Katz, M. L. and Shapiro, C. (1986), Technology adoption in the presence of network externalities, Journal of Political Economy 94, 822–41. Kauffman, S. A. (1993), The origins of order: Self-organization and selection in evolution, Oxford University Press, Oxford. Kealey, T. (1996), The economic laws of scientific research, Macmillan, London. Kennedy, C. (1964), Induced bias and the theory of distribution, Economic Journal 74, 541–7. Kingston, W. (2001), Innovation needs patents reforms, Research Policy 30, 403–23. Kirman, A. (2004), Economics and complexity, in R. Leombruni and M. Richiardi (eds), Industry and labor dynamics: The agent based computational economics approach, World Scientific, New Jersey. Klemperer, P. (1987a), Markets with consumer switching costs, Quarterly Journal of Economics 102, 375–94.
References 389 Klemperer, P. (1987b), The competitiveness of markets with switching costs, Rand Journal of Economics 18, 138–50. Klemperer, P. (1990), How broad should the scope of patent protection be?, Rand Journal of Economics 21, 113–30. Klemperer, P. (1995), Competition when consumers have switching costs: An overview with an application to industrial organization, macroeconomics and international trade, Review of Economic Studies 62, 515–39. Kortum, S. and Lerner, J. (2000), Assessing the contribution of venture capital to innovation, Rand Journal of Economics 31, 674–92. Krafft, J. (2003), Vertical structure of the industry and competition: An analysis of the evolution of the infocommunication industry, Telecommunications Policy 27, 625–49. Lach, S. and Schankerman, M. (2003), Incentives and invention in universities. NBER Working Paper No. 9727. Laestadius, S. (1998), Technology level, knowledge formation and industrial competence in paper manufacturing, in C. Green, G. Eliasson and C. McCann (eds), Microfoundations of economic growth: A Schumpeterian perspective, University of Michigan Press, Ann Arbour, pp. 212–26. Lamberton, D. (ed.) (1971), Economics of information and knowledge, Penguin Books, Harmondsworth. Lancaster, K. (1971), Consumer demand: A new approach, Columbia University Press, New York. Lane, D. A. and Maxfield, R. (1997), Foresight, complexity and strategy, in W. B. Arthur, S. N. Durlauf and D. A. Lane (eds), The economy as an evolving complex system II, Santa Fe Institute Studies, Westview Press, Boulder, CO, pp. 169–98. Lane, D. A., Maxfield, R., Read, D. and Van der Leeuw, S. (2006), The innovation innovation: From population thinking to organization thinking. ISCOM working papers, Università di Modena e Reggio. Lang, L. and Stultz, R. (1994), Tobin’s q, corporate diversification, and firm performance, Journal of Political Economy 102, 1248–80. Langlois, R. N. (1992), Transaction-cost economics in real time, Industrial and Corporate Change 1, 99–127. Langlois, R. N. and Foss, N. J. (1999), Capabilities and governance: The rebirth of production in the theory of economic organization, Kyklos 52, 201–18. Lawson, T. (2005), The nature of heterodox economics, Cambridge Journal of Economics 30, 483–505. Lawton Smith, H. (2006), Universities, innovation and the economy, Routledge, London and New York. Lerner, J. (1995), Venture capitalists and the oversight of private firms, Journal of Finance 50, 301–18. Lerner, J. and Merges, R. P. (1998), The control of technology alliances: An empirical analysis of the biotechnology industry, Journal of Industrial Economics 46, 125–56. Lerner, J. and Tirole, J. (2002), Some simple economics of open source, Journal of Industrial Economics 52, 197–234. Lerner, J. and Tirole, J. (2004), The economics of technology sharing: Open source and beyond, Journal of Economic Perspectives 19, 99–120. Lessig, L. (1999), Code and other laws of cyberspace, Basic Books, New York. Levin, R. C., Klevorick, A. K., Nelson, R. R. and Winter, S. G. (1987), Appropriating the returns from industrial research and development, Brookings Paper on Economic Activity 3, 242–79.
390 References Levinthal, D. A. (1998), The slow pace of rapid technological change: Gradualism and punctuation in technological change, Industrial and Corporate Change 8, 217–47. Loasby, B. J. (1999), Knowledge institutions and evolution in economics, Routledge, London. Loasby, B. J. (2002), The division and coordination of knowledge, in Sheila C. Dow and J. Hillard (eds), Post Keynesian econometrics, microeconomics and the theory of the firm, Beyond Keynes, Vol. I, Edward Elgar, Cheltenham. Loch, C. H. and Huberman, B. A. (1999), A punctuated-equilibrium model of technology diffusion, Management Science 45, 160–77. Lundvall, B. A. (1985), Product innovation and user–producer interaction, Aalborg University Press, Aalborg. Lypsey, R., Bekar, C. and Carlaw, K. (1998), General purpose technologies: What requires explanation, in E. Helpman (ed.), General purpose technologies and economic growth, MIT Press, Cambridge, MA. McDonald, S. (2004), What means become ends: Considering the impact of patent strategy on innovation, Information Economics and Policy 16, 135–58. Machlup, F. and Penrose, E. (1950), The patent controversy in the nineteenth century, Journal of Economic History 10, 1–20. Madden, G. (ed.) (2003), Handbook on telecommunications, Edward Elgar, Cheltenham. Mansfield, E. (1968), Industrial research and technological innovation, Norton, New York. Mansfield, E. (1995), Academic research underlying industrial innovations: Sources, characteristics and financing, Review of Economics and Statistics 77, 55–65. Mansfield, E., Schwartz, M. and Wagner, S. (1981), Imitation costs and patents: An empirical study, Economic Journal 91, 907–18. March, J. C. (1991), Exploration and exploitation in organizing learning, Organization Science 2, 71–87. March, J. C. and Simon, H. A. (1958), Organizations, John Wiley & Sons, New York. Marchionatti, R. (1988), Il cambiamento tecnologico in una prospettiva classica, Economia e Politica Industriale 58, 177–206. Marchionatti, R. (1999), On Keynes’ animal spirits, Kyklos 52: 415–39. Marchionatti, R. (2004), What don’t economists know now that Marshall knew a century ago? A note on Marshall’s ‘sophisticated informality’, Journal of Post-Keynesian Economics 26, 441–60. Marmolo, E. (1999), A constitutional theory of public goods, Journal of Economic Behavior and Organization 38, 27–42. Marquetti, A. (2003), Analyzing historical and regional patterns of technical change from a classical-Marxian perspective, Journal of Economic Behavior and Organization 52, 191–200. Marshall, A. (1920 [1890]), Principles of economics, 8th edn, Macmillan, London. Martin, S. (1993), Advanced industrial economics, Basil Blackwell, Oxford. Marx, K. (1857–58), Grundrisse (Economic manuscripts of 1857–58, First version of Capital), http://www.marxists.org/archive/marx/works/cw/volume29. Marx, K. (1976 [1867]), Capital: A critique of political economy, Penguin Books, Harmondsworth. May, C. (2000), A global political economy of intellectual property rights: The new enclosures?, London, Routledge. Mazzoleni, R. and Nelson, R. R. (1998), The benefits and the costs of strong patent protection: A contribution to the current debate, Research Policy 27, 273–84. Menard, C. (ed.) (2000), Institutions, contracts and organizations: Perspectives from new institutional economics, Edward Elgar, Cheltenham.
References 391 Merges, R. and Nelson, R. R. (1994), On limiting or encouraging rivalry in technical progress: The effect of patent scope decisions, Journal of Economic Behavior and Organization 25, 1–24. Metcalfe, J. S. (1981), Impulse and diffusion in the study of technical change, Futures 13, 347–59. Metcalfe, J. S. (1997), Evolutionary economics and creative destruction, Routledge, London. Metcalfe, J. S. (2001), Consumption preferences and the evolutionary agenda, Journal of Evolutionary Economics 11, 37–58. Metcalfe, J. S. (2005), Ed Mansfield and the diffusion of innovation: An evolutionary connection, Journal of Technology Transfer 30, 139–57. Milgrom, P. and Roberts, J. (1995), Complementarities and fit: Strategy structure and organizational change in manufacturing, Journal of Accounting and Economics 19, 179–208. Mokyr, J. (1990a), Punctuated equilibria and technological progress, American Economic Review (P&P) 80, 350–4. Mokyr, J. (1990b), The lever of riches: Technological creativity and economic progress, Oxford University Press, Oxford. Momigliano, F. (1975), Economia industriale e teoria dell’impresa, Il Mulino, Bologna. Morroni, M. (2006), Knowledge, scale and transactions in the theory of the firm, Cambridge University Press, Cambridge. Mowery, D. C., Nelson, R. R., Sampat, B. N. and Ziedonis, A. A. (2001), The growth of patenting and licensing by U.S. universities: An assessment of the effects of the Bayh–Dole Act of 1980, Research Policy 30, 99–119. Mueller, D. C. (2003), The corporation: Investment, mergers and growth, Routledge, London. Murmann, P. (2003), Knowledge and competitive advantage: The coevolution of firms, technology and national institutions, Cambridge University Press, Cambridge. Nelson, P. (1970), Information and consumer behaviour, Journal of Political Economy 78, 311–29. Nelson, R. R. (1959), The simple economics of basic scientific research, Journal of Political Economy 67, 297–306. Nelson, R. R. (1982), The role of knowledge in R&D efficiency, Quarterly Journal of Economics 97, 453–70. Nelson, R. R. (2004), The market economy and the scientific commons, Research Policy 33, 455–71. Nelson, R. R. (2005), Technology institutions and economic growth, Harvard University Press, Cambridge, MA. Nelson, R. R. and Winter, S. G. (1973), Toward an evolutionary theory of economic capabilities, American Economic Review 63, 440–9. Nelson, R. R. and Winter, S. G. (1982), An evolutionary theory of economic change, Belknap Press of Harvard University Press, Cambridge, MA. Nonaka, I. and Takeuchi, H. (1995), The knowledge-creating company, Oxford University Press, Oxford. Nooteboom, B. (2000), Learning and innovation in organizations and economics, Oxford University Press, Oxford. Nooteboom, B. (2002), Trust, forms, foundations, functions, failures and figures, Edward Elgar, Cheltenham. Nooteboom, B. (2003), Managing exploration and exploitation, in S. Rizzello (ed.), Cognitive developments in economics, Routledge, London, pp. 218–42.
392 References Nooteboom, B. (2004), Governance and competence: How can they be combined?, Cambridge Journal of Economics 28, 505–25. North, D. C. (1997), Some fundamental puzzles in economic history, in W. B. Arthur, S. N. Durlauf and D. Lane (eds) (1997), The economy as an evolving complex system II, Addison-Wesley, Redwood City, CA. North, D. C. and Wallis, J. (1994), Integrating institutional change and technical change in economic history: A transaction cost approach, Journal of Institutional and Theoretical Economics 150, 609–24. Ocasio, W. (1995), The enactment of economic adversity: A reconciliation of theories of failure-induced change and threat-rigidity, Research in Organizational Behavior 17, 287–331. O’Donoghue, T. (2001), A patentability requirement for sequential innovation, Rand Journal of Economics 32, 654–79. Oerlemans, L. A. G. and Meeus, M. T. M. (2001), R&D cooperation in a transaction costs perspective, Review of Industrial Organization 18. Ormerod, P. and Colbaugh, R. (2006), Cascades of failure and extinction in complex evolving networks, Journal of Artificial Societies and Social Simulation 4. Pastor-Satorras, R. and Vespignani, A. (2004), Evolution and structure of the Internet, Cambridge University Press, Cambridge. Patrucco, P. P. (2003), Institutional variety, networking and knowledge exchange: Communication and innovation in the case of the Brianza technological district, Regional Studies 37, 159–72. Patrucco, P. P. (2005), The emergence of technology systems: Knowledge production and distribution in the case of the Emilian plastics district, Cambridge Journal of Economics 29, 37–56. Patrucco, P. P. (2008), Collective knowledge production, costs and the dynamics of technological systems, Economics of Innovation and New Technology 17, forthcoming. Pavitt, K. (2000), Technology, management and systems of innovation, Edward Elgar, Cheltenham. Pavitt, K., Robson, M. and Townsend, J. (1989), Technological accumulation, diversification and organization in UK companies, 1945–83, Management Science 35, 81–99. Penrose, E. T. (1959), The theory of the growth of the firm, 1st and 3rd edns, Basil Blackwell (1959), Oxford, and Oxford University Press (1995), Oxford. Perroux, F. (1964), L’économie du XXe siècle, Presses Universitaires de France, Paris. Pisano, G. P. (1990), The R&D boundaries of the firm: An empirical analysis, Administrative Science Quarterly 35, 153–76. Polanyi, M. (1958), Personal knowledge: Towards a post-critical philosophy, Routledge & Kegan Paul, London. Polanyi, M. (1966), The tacit dimension, Routledge & Kegan Paul, London. Porter, M. E. (1985), Competitive advantage, Free Press, New York. Radner, R. (1996), Bounded rationality, indeterminacy and the theory of the firm, Economic Journal 106, 1360–73. Raymond, E. (1999), The cathedral and the bazaar: Musings on Linux and the open source by an accidental revolutionary, O’Reilly, Sebastopol, CA. Reichman, J. (2000), ‘Of green tulips and legal kudzu’: Repackaging rights in subpatentable invention, Vanderbilt Law Review 53, 17–43. Reprinted in R. Dreyfuss and D. Zimmerman (eds), Expanding the boundaries of intellectual property, Oxford University Press (2001), Oxford, pp. 23–54. Reichman, J. and Maskus, K. (eds) (2005), International public goods and transfer of
References 393 technology under a globalized intellectual property regime, Cambridge University Press, Cambridge. Reitzig, M. (2004), The private value of ‘thickets’ and ‘fences’, Economics of Innovation and New Technology 13, 443–56. Richardson, G. B. (1960), Information and investment, Oxford University Press, Oxford. Richardson, G. B. (1972), The organization of industry, Economic Journal 82, 883–96. Riordan, M. and Williamson, O. E. (1985), Asset specificity and economic organization, International Journal of Industrial Organization 3, 365–78. Rizzello, S. (1999), The economics of the mind, Edward Elgar, Cheltenham. Rizzello, S. (ed.) (2003), Cognitive developments in economics, Routledge, London. Rizzello, S. (2004), Knowledge as a path-dependent process, Journal of Bioeconomics 6, 255–74. Robinson, J. (1937), The classification of inventions, Review of Economic Studies 6, 139–42. Romer, P. M. (1986), Increasing returns and long-run economic growth, Journal of Political Economy 94, 1002–37. Romer, P. M. (1990), Endogenous technological change, Journal of Political Economy 98, S71–102. Romer, P. M. (1994), The origins of endogenous growth, Journal of Economic Perspectives 8, 3–22. Rosenberg, N. (1965), Adam Smith on the division of labour: Two views or one, Economica 34, 127–39. Rosenberg, N. (1974), Karl Marx and the economic role of science, Journal of Political Economy 82, 713–28. Rosenberg, N. (1976), Perspectives on technology, Cambridge University Press, Cambridge. Rosenberg, N. (1992), Economic experiments, Industrial and Corporate Change 1. Rosenberg, N. (1994), Exploring the black box, Chapter 2 in Joseph Schumpeter, Radical Economist, Cambridge University Press, Cambridge, pp. 47–61. Rosser, J. B. (1991), From catastrophe to chaos: A general theory of economic discontinuities, Kluwer Academic Publishers, Boston, MA. Rosser, J. B. (1999), On the complexities of complex economic dynamics, Journal of Economic Perspectives 13, 169–92. Rosser, J. B. (ed.) (2004), Complexity in economics, Edward Elgar, Cheltenham. Rumelt, R. P. (1986 [1974]), Strategy, structure and economic performance, 2nd edn, Harvard Business School Classics, Boston, MA. Ruttan, V. W. (1997), Induced innovation evolutionary theory and path dependence: Sources of technical change, Economic Journal 107, 1520–9. Ruttan, V. W. (2001), Technology growth and development: An induced innovation perspective, Oxford University Press, Oxford. Saint Paul, G. (1992), Technological change, financial markets and economic development, European Economic Review 36, 763–81. Sakahibara, M. and Bransletter, L. (2001), Do stronger patents induce more innovation? Evidence from the 1988 Japanese patent law reform, Rand Journal of Economics 32, 77–100. Salter, W. E. G. (1960), Productivity and technical change, Cambridge University Press, Cambridge. Samuelson, P. (1965), A theory of induced innovation along Kennedy, Weiszacker lines, Review of Economics and Statistics 47, 343–56. Samuelson, P., Davis, R., Kapor, M. D. and Reichman, J. H. (1994), A manifesto concerning the legal protection of computer programs, Columbia Law Review 94, 2308–65.
394 References Scherer, F. M. (1984), Innovation and growth: Schumpeterian perspectives, MIT Press, Cambridge, MA. Scherer, F. M. (1992), International high-technology competition, Harvard University Press, Cambridge, MA. Schmookler, J. (1966), Invention and economic growth, Harvard University Press, Cambridge, MA. Schumpeter, J. A. (1936 [1911]), The theory of economic development, Harvard University Press, Cambridge, MA. Schumpeter, J. A. (1928), The instability of capitalism, Economic Journal 38, 361–86. The pages of the quotes refer to the text reproduced in N. Rosenberg (ed.) (1971), The economics of technological change, Penguin Books, Harmondsworth. Schumpeter, J. A. (1939), Business cycles, McGraw-Hill, New York. Schumpeter, J. A. (1942), Capitalism, socialism and democracy, Harper and Brothers, New York. Schumpeter, J. A. (1947), The creative response in economic history, Journal of Economic History 7, 149–59. Scitovsky, T. (1954), Two concepts of external economies, Journal of Political Economy 62, 143–51. Scotchmer, S. (1991), Standing on the shoulders of giants: Cumulative research and the patent law, Journal of Economic Perspectives 5, 29–41. Scotchmer, S. (1996), Protecting early innovators: Should second-generation products be patentable? Rand Journal of Economics 27, 322–31. Scotchmer, S. (2001), On the optimality of the patent renewal system, Rand Journal of Economics 32, 181–96. Shankerman, M. and Scotchmer, S. (2001), Damages and injunctions in protecting intellectual property, Rand Journal of Economics 32, 199–220. Shavell, S. and Van Ypersele, T. (2001), Rewards versus intellectual property rights, Journal of Law and Economics 44, 525–47. Shnerb, N. M., Louzoun, Y., Bettelheim, E. and Solomon, S. (2000), The importance of being discrete: Life always wins on the surface, Proceedings of the National Academy of Sciences of the United States of America 97, 10322–4. Simon, H. A. (1947), Administrative behavior, Free Press, New York. Simon, H. A. (1962), The architecture of complexity, Proceedings of the American Philosophical Society 106, 467–82. Simon, H. A. (1982), Metaphors of bounded rationality: Behavioral economics and business organization, MIT Press, Cambridge, MA. Smith, A. (1776), An inquiry into the nature and causes of the wealth of nations (London edn, 1976, Penguin Books, Harmondsworth). Smith, M. Y. (2004), A model of linked adoption of complementary technologies, Economics of Innovation and New Technology 13, 91–9. Solomon, S., Weisbuch, G., de Arcangelis, L., Naeem, J. and Stauffer, D. (2000), Social percolation models, Physica A 277, 239–47. Solow, R. M. (1957), Technical change and the aggregate production function, Review of Economics and Statistics 39, 312–20. Spence, M. (1973), Job market signaling, Quarterly Journal of Economics 77, 355–79. Spulber, D. (1999), Market microstructures: Intermediaries and the theory of the firm, Cambridge University Press, Cambridge. Stallman, R .M. (1998), The GNU project, O’Reilly, Sebastopol, CA.
References 395 Stiglitz, J. E. (1985), Credit markets and capital control, Journal of Money Credit and Banking 17, 133–52. Stiglitz, J. E. (1987), Learning to learn, localized learning and technological progress, in P. Dasgupta and P. Stoneman (eds), Economic policy and technological performance, Cambridge University Press, Cambridge, pp. 125–44. Stiglitz, J. E. (1994), Economic growth revisited, Industrial and Corporate Change 3, 65–110. Stiglitz, J. E. (2000), The contributions of the economics of information to twentieth century economics, Quarterly Journal of Economics 115, 1441–78. Stiglitz, J. E. (2002), Information and the change in the paradigm in economics, American Economic Review 92, 460–502. Stiglitz, J. E. and Weiss, A. (1981), Credit rationing in markets with imperfect information, American Economic Review 71, 912–27. Stoneman, P. (1976), Technological diffusion and the computer revolution, Cambridge University Press, Cambridge. Stoneman, P. (1983), The economics analysis of technical change, Oxford University Press, Oxford. Stoneman, P. (1987), The economics analysis of technology policy, Clarendon Press, Oxford. Stoneman, P. (ed.) (1995), Handbook of the economics of innovation and technological change, Basil Blackwell, Oxford. Stoneman, P. (2002), The economics of technological diffusion, Basil Blackwell, Oxford. Stoneman, P. and Toivanen, O. (1997), The diffusion of multiple technologies: An empirical study, Economics of Innovation and New Technology 5, 1–17. Swann, G. M. P. (ed.) (1994), New technologies and the firm: Innovation and competition, Routledge, London. Swann, G. M. P. (2006), Is the world flat or round? Mapping changes in the taste for art, in C. Antonelli, D. Foray, B. H. Hall and W. E. Steinmueller (eds), New frontiers in the economics of innovation and new technology: Essays in honour of Paul A. David, Edward Elgar, Cheltenham. Sylos Labini, P. (1956), Oligopolio e progresso tecnico, Giuffrè, Milano. Translated into English (1962), Oligopoly and technical progress, Harvard University Press, Cambridge, MA. Sylos Labini, P. (1984), The forces of economic growth and decline, MIT Press, Cambridge, MA. Taylor, M. C. (2001), The moment of complexity: Emerging network culture, University of Chicago Press, Chicago, IL. Teece, D. J. (2000), Managing intellectual capital, Oxford University Press, Oxford. Teece, D. J. (2005), Technology and technology transfer: Mansfieldian inspirations and subsequent developments, Journal of Technology Transfer 30, 17–33. Teubal, M., Yinnon, T. and Zuscovitch, E. (1991), Networks and market creation, Research Policy 20, 381–92. Turvani, M. (2001), Microfoundations of knowledge dynamics within the firm, Industry and Innovation 8, 309–23. Utterback, J. M. (1994), Mastering the dynamics of innovation, Harvard Business School Press, Boston, MA. Von Hippel, E. (1988), The sources of innovation, Oxford University Press, Oxford. Von Hippel, E. and Von Krogh, G. (2003), Open source software and the ‘Private-Collective’ innovation model: Issues for organization science, Organization Science 14, 209–23. Weitzman, M. (1996), Hybridizing growth theory, American Economic Review 86: 207–12.
396 References Williamson, O. E. (1975), Markets and hierarchies: Analysis and antitrust implications, Free Press, New York. Williamson, O. E. (1985), The economic institutions of capitalism: Firms, markets, relational contracting, Free Press, New York. Williamson, O. E. (1990), The firm as a nexus of treaties: An introduction, in M. Aoki, B. Gustaffson and O. E. Williamson (eds), The firm as a nexus of treaties, Sage Publications, London, pp. 1–25. Williamson, O. E. (1996), The mechanisms of governance, Oxford University Press, New York. Witt, U. (1998), Imagination and leadership: The neglected dimension of an evolutionary theory of the firm, Journal of Economic Behavior and Organization 35, 161–77. Witt, U. (2001), Learning to consume: A theory of wants and the growth of demand, Journal of Evolutionary Economics 11, 23–36. Wolfram, S. (2002), A new kind of science, Wolfram Media, Champaign, IL. Young, A. A. (1928), Increasing returns and economic progress, Economic Journal 38, 527–42. Zeitlin, J. and Herrigel, G. (eds) (1999), Americanization and its limits: Reworking US technology and management in postwar Europe and Japan, Oxford University Press, Oxford. Zingales, L. (2000), In search of new foundations, Journal of Finance 55, 1623–53.
Index
Abramovitz, M. 286 academic outsourcing 176–7, 208 academic patenting 207 academic salaries 203–4, 371 academic system see universities/ academic system Acemoglu, D. 37, 283 adaptive response 40 adjustment budget 270–1 adjustment possibility frontier 255–9, 273, 274, 290–3, 308–12, 323–6 adoption 11–12, 316–32; in the business sector 317–23; creative absorption and diffusion 328–31; frontier of creative absorption 323–7; path-dependent 342–6 adoption costs 321, 328–30 agency costs 76, 198–201; in academic system 201; in corporations 199–201 agency theory 75–6, 122–3, 198–201 agents 335–6; heterogeneous 333, 361; intentional action 105, 247, 336 Aghion, P. 116 Akerlof, G.A. 77 Amendola, M. 106 American economy 295–6 Antonelli, C. 55, 57, 58, 59, 60, 67, 75, 81, 116, 133, 238, 296 appropriability of knowledge 8–9, 74, 91, 135–49; emergence of direction of technological change 140–6; increasing returns 113–16; role of localised appropriability in knowledge exploitation 138–40; trade-off 233, 234; and transaction costs 157 Arrow, K.J. 7, 42, 53, 75, 76, 85, 88, 108, 123, 136, 138, 165, 195, 215, 222, 228, 229 Arthur, B. 51, 343, 356
articulable knowledge 170; constructed interactions 168–9, 182, 183–91 Atkinson, A.B. 2, 53, 54, 57, 70 Audretsch, D.B. 42 automobile industry 109, 151–2, 174–5, 278 average costs of knowledge 230–2 Barabasi, A.L. 102, 105 barriers to entry 132, 303 Baumol, W.J. 42, 112 benchmark model 11, 245–62; strong irreversibility 247–54; weak irreversibility 254–9 Bianchi, M. 303, 316 bias 53, 291–4, 366 Binswanger, H.P. 283 biology 22–3, 61, 265 biotechnology 110, 181 Boschma, R.A. 101 boundaries, firm 131–3 bounded rationality 49–50, 246, 365 British economy 295–6 Broadberry, S. 296 Brown, M. 286 Brownian movement 346 Buchanan, J.M. 102 bundling 180–1 Burt, R.S. 304 business sector: academic system 195–7; adoption and diffusion in 317–23; knowledge governance 167–92; see also corporations, theories of the firm business services, knowledge-intensive 178 capital, knowledge as key component of 24 capitalism 29–36
398 Index Caves, R.E. 88 centred networks 102, 268 characteristics of goods 306–7 chemical industry 155 children 32 Cisco 181 clubs, technological 184 clusters 24, 55, 58, 164, 187–9 Coase, R. 121, 134 codified knowledge 73, 89, 136–8, 170; coordinated transactions 169, 176–82, 186 cognitive distance 49–50 Cohen, W.M. 80, 101 collective knowledge 11, 24, 30, 69–75, 95–103, 341–2; emergence of localised technological knowledge 98–103; knowledge trade-off 95–8; and path dependence 103–6, 346–53; Piedmont 277–9; system dynamics see system dynamics communication 101, 102–3, 196–7, 214; costs 80–1 communication industry 152, 178–9, 220–1; information and communication technology 109–10, 267–8 communications authorities 221 compensatory liability regime see liability rule competence 130–1, 150, 246, 302; division of labour 18–20 competitive innovation 38–41, 45 complementarity 58, 74, 108–10, 187–8, 229; Marshall 22–9; public science and contract research 204–5 complex system dynamics see system dynamics complexity 3–4, 12, 333–55, 356; basic assumptions 333; basic ingredients 356; characteristics of complex systems 334, 378–9; path-dependent dynamics 337–46 compositeness 7, 58, 74; governance mechanisms 170, 182, 186, 188; knowledge indivisibility 109, 110, 111, 112–13, 119 composition effects 59, 283–5, 295, 345 compulsory licensing 225, 241, 242 constructed interactions 167, 168–9, 182, 183–91 consumers’ choices 302–3 content contracts 184
contingent technological change 285–8, 375; inducement of 288–94 contract research 204–5 contracts: content contracts 184; longterm cooperation contracts 183; procedural contracts 82–3, 184 coordinated transactions 167, 169, 176–82, 186 coordination: costs 156–7; dynamic 105–6, 192, 360; interdependence with production and transaction 124–33 corporate function 127–9 corporate growth 163, 173–4 corporations 85, 86, 87–8; and academic system 195–7; agency costs in 199–201; global 93, 163–4; knowledge cost function 209–11; see also business sector, theories of the firm cost-based compensation schemes 226–7 cost characteristics 320 creative adoption 11–12, 316–32; and diffusion 328–31; frontier of 323–7 creative destruction 41 creative reaction/response 4–5, 40–1, 45, 48, 50–1, 56, 339, 361, 362 creative talents 42 creativity 61–2, 333, 347; academic management of 194–8; hysteresis and 356–63; path dependence 338–9, 348–9, 350, 354–5; progressive decline of sources of 21 credit rationing 82, 94 cross-licensing 185 cumulability 7, 73, 218–19, 229; governance mechanisms 170, 182, 186; knowledge indivisibility 109, 111, 112, 119 cumulative inducement mechanism 270–1 Darwinism 23 Dasgupta, P. 171, 194–6 David, P.A. 2, 3, 51, 53, 60, 70, 171, 194–6, 295–6, 336, 337, 342, 343, 375–6 Dawkins, R. 61 debt finance 179–80 demand 29, 45–6, 118; adoption and diffusion 318–20, 321; demand-supply 76–8; economies of scope on the demand side 78–9; optimum level for knowledge rents 230–3, 234; for research activities and professional university 209–11
Index 399 density, economies of 112, 220–3, 229 derived demand 230–3, 234 diachronic complementarity 109 differentiation 22 diffusion 11–12, 60, 316–32, 360, 376–7; in the business sector 317–23; creative adoption and 328–31; path-dependent adoption 342–6; punctuated 266; role of external knowledge and supply of new technologies 323–7; selective 43 direction of technological change: emergence of 140–6; relative prices and 283–5 discontinuous change 11, 264–8 dispersed knowledge 55, 100, 361 dissemination of knowledge 196–7 division of labour 17–22, 27–8, 45 Dosi, G. 44 Durlauf, S.N. 3 dynamic coordination 105–6, 192, 360 dynamic efficiency 41, 88, 217–18 dynamic knowledge transaction costs 77–8, 156; demand-side 76–8; supplyside 151–4 dynamic legacies 5–6, 17–46; Marshall 6, 17, 22–7, 45, 46; Marx 5–6, 17, 29–38, 44–5, 46; Schumpeter 6, 17, 38–41, 42, 45, 46; Smith 5–6, 17, 17–22, 27–9, 44, 45, 46; towards a full endogeneity of technological change 42–4 economic systems 59 economies of density 112, 220–3, 229 economies of scope 113; on the demand side 78–9, 156; in transactions 78 education 21 efficiency: dynamic 41, 88, 217–18; static 88, 217–18 Eldredge, N. 265 emergent properties 3, 334 empirical analysis 146–8 encompassing model 321, 331 endogeneity of technological change 35, 42–4 entrepreneurship 38, 39–40, 42, 163; scientific 94–5, 172 entropy 347–8 entry, barriers to 132, 303 epidemic contagion 318 epistemic communities 189, 190–1 equilibrium analysis 38–9, 334–5 equity finance 179–80 ergodicity 337–8
essential facility 220–3 Etzkowitz, H. 193 evolutionary approaches 43–4, 91–2, 357 evolutionary biology 23, 61 ex ante standardisation 164 exaptation 265 exchange value 238–9 exclusive intellectual property rights 223–4 expectations/actual conditions mismatch 49, 50, 70, 360, 361 experience (learning by doing) 18–19, 28–9 exploitation of knowledge see knowledge exploitation extent of the market 17–22, 27–8, 45 external knowledge 45, 54–5, 66–7, 71–2, 95–7, 99–100, 118, 136–8, 361–2; increasing returns 113–16; knowledge trade-off 219–20; role in diffusion 323–7 factor markets 11, 48, 282–300; changes in and adoption 319; changes in and irreversibility 249–50, 254–5; general and contingent technological change 285–8; inducement mechanism see inducement mechanism; static conditions of 37, 282–3 factor prices: changes and inducement mechanism 36–7, 45–6, 282–3; levels of 37; relative and composition effects 59, 283–5 Factory Acts 32 Ferraris, G. 277 financial markets 94, 155, 179–81 firm, theory of the see theories of the firm Foster, J. 366 fragmented knowledge 55, 100, 273, 361 fragmented networks 102, 268 Fransman, M. 152 Friedman, M. 334, 377 fungibility 7, 58, 73–4; governance mechanisms 170, 182, 186, 187; knowledge indivisibility 110–11, 113, 119 Gaffard, J.L. 106 general equilibrium analysis 38–9, 334–5 general intellect 36 General Public Licence 83, 189–91, 225, 370, 372 general technological change 285–8, 375; inducement of 288–94
400 Index generative relationships 352–3 generic knowledge 43, 89 generic production factors 143–5, 146 geodesic networks 102, 268 geographical clusters 24, 55, 58, 164, 187–9 Gleaser, E. 5, 55 global corporations 93, 163–4 GNU/Linux operating system 189–91 Gould, S. 265 governance mechanisms 9, 56, 60, 68–9, 74–5, 102, 167–92, 350–1, 358; articulable knowledge 168–9, 182, 183–91; codified knowledge 169, 176–82, 186; hybrid forms 167–71; implications of non-exclusive intellectual property rights 237–9; intellectual property rights see intellectual property rights; liability rule see liability rule; multiple equilibria and 116–18; tacit knowledge 168, 171–6, 182, 186; universities see universities/academic system gradualism 11, 264–8 Granovetter, M.S. 303 Griliches, Z. 97, 136 gross profitability of adoption 328–30 growth: engine of 346–53; firm/corporate 163, 173–4; new growth theory 43, 91, 265, 357 growth economics 52–3 growth poles 187, 188 Guiso, L. 102 Habbakuk, H.J. 295 Hall, B.H. 239 Hayek, F.A. 55, 100, 367 Hicks, J.R. 36–7, 282, 283, 375 historical time 57 horizontal technological clubs 184 Howitt, P. 116 hub-based networks 102, 268 hub companies 174–5 Huberman, B.A. 266 hybrid forms of governance 9, 167–71; constructed interactions 167, 168–9, 182, 183–91; coordinated transactions 167, 169, 176–82, 186; quasi-hierarchies 167, 168, 169, 171–6, 182, 186 hydroelectric power 277–8 hysteresis 356–63
idiosyncratic production factors 139–40, 143–5, 146, 148, 369 imitation 11–12, 317 in-house outsourcing 175 incentive complementarity 204–5 incomplete contract theory 368 increasing returns 28; to adoption 321; knowledge indivisibility and external increasing returns 111–16 incremental costs 112 individual path dependence 348–53 indivisibility, knowledge see knowledge indivisibility induced adoption 321–2 induced technological change 29–38, 45–6, 282–3, 375 inducement mechanism 52, 282–300, 375; applications and implications 294–9 industrial districts 24, 55, 58, 164, 187–9 industrial dynamics 132–3 information and communication technologies 109–10, 267–8 information economics 6–7, 65–83; framework for economics of localised technological knowledge 75–83; localised technological knowledge as interactive and collective activity 69–75 innovation possibility frontier 308, 309, 311, 312–13 instability 116–18 institutional innovations 82–3 integration 22 intellectual property rights 9–10, 83, 87, 88–9, 92, 93–4, 97–8, 181, 215–42; essential facility 220–3; functions 215–17; implications for knowledge governance 237–9; knowledge trade-off 215–20; liability rule as a governance mechanism 223–37; non-exclusive 205–7, 215–42; trade in patents and licences 177, 370–1 intentional action 105, 247, 336 interactions, constructed 167, 168–9, 182, 183–91 interactivity 3, 69–75, 333 intermediary markets 152–4, 158–62 internal knowledge 54–5, 66, 136–8 interstices, technological 187, 188 invention, as a business 35–6 irreversibility 26, 51–2, 57, 60, 61–2, 245–6, 259–60; path dependence 338–9, 348, 349, 350; quasiirreversibility 26, 52, 246, 248,
Index 401
joint ventures 175–6
knowledge rationing 81–2 knowledge rents 222–3; optimum levels 229–37 knowledge space 58 knowledge trade-offs 87–9, 95–8, 195, 215–20, 233, 234; use or sell see use or sell trade-off
Kahneman, D. 51 Kaldor, N. 27–9 Karshenas, M. 321, 331 Kauffman, S.A. 104 Kennedy, C. 37, 282, 283 Kingston, W. 87 Kirman, A. 105 knowledge 7, 54–6, 84–107; appropriability see appropriability of knowledge; articulable see articulable knowledge; codified see codified knowledge; as a collective process see collective knowledge; as essential facility 222–3; external see external knowledge; generic and specific 43, 89; as input and output 24–5; internal 54–5, 66, 136–8; key component of capital 24; as a path-dependent emergent property 106–7; as a quasi-proprietary good 89–95; as a quasi-public good 85–9; tacit see tacit knowledge knowledge cost functions 142–3, 160–1; professional university, corporations and open science 209–11 knowledge exploitation: governance mechanisms 171, 186; role of localised appropriability 138–40; strategies 156, 162–4, 165–6 knowledge generation 8, 135–49, 359, 362; emergence of direction of technological change 140–6; empirical analysis and technology policy 146–8; governance mechanisms 171, 182; information economics 71–3; role of knowledge indivisibility 136–8 knowledge indivisibility 7, 73–4, 99, 108–19, 218–19, 229; increasing returns 111–16; multiple equilibria, instability and governance of knowledge commons 116–18; role in knowledge generation 136–8; unfolding of 108–11 knowledge-intensive business services 178 knowledge production function 142–3
labour substitution by capital 31–6, 44–5, 294–6 Laestadius, S. 71 Lancaster, K. 59, 301, 306, 314 Lancastrian space 59, 140–1; model of localised product innovation 304–13 Lane, D.A. 43, 352 Langlois, R.N. 77 large corporations see corporations Lawson, T. 377–8 learning 42–3, 67, 130–1, 245–6, 358, 361; by doing (experience) 18–19, 28–9; economics of 53–4; information economics 69–73; localised 302, 340–1; by using 18–19 Levinthal, D.A. 80, 101, 266 liability rule 190, 223–37, 238; implementation 226–7; optimum levels of knowledge rents 229–37; proposed regime 224–6; welfare effects 227–9 licences, trade in 177, 370–1 licensing: compulsory 225, 241, 242; cross-licensing 185 limited mobility 5 litigation: costs 239; procedure 227 Loasby, B.J. 90 local externalities 61, 338–9, 350 local factor endowments 146–7, 295–6 localised appropriability 138–40 localised learning 302, 340–1 localised supermodularity 273–4 localised technological change 1, 2, 6, 47–62; basic ingredients 49–56; benchmark model 11, 245–62; collective knowledge, path dependence and 346–53; complex dynamics see system dynamics; multidimensional scope 57–60; path dependence see path dependence; process 48–9 localised technological knowledge 69–83; emergence of 98–103; information economics framework for economics of 75–83; as interactive and collective activity 69–75
259–60, 339; strong 247–54; weak 254–9 isorevenues, map of 159–62, 258–9, 290–3, 308–12, 323–6 Italy 277–9
402 Index location 333 Loch, C.H. 266 lock-in 345–6 long-term cooperation contracts 183 Lundvall, B.A. 96, 302 machinery 30–1; substitution of labour 31–6, 44–5, 294–6; suppliers 19 macroeconomic model 283 mandated interconnection 220–3, 224–5 manufacturing costs 131–2 marginal costs 112 market cascades 178–9 market possibility frontier 161 market prices 131–2 market selection 105 market structure 41, 303 markets: extent of the market 17–22, 27–8, 45; financial 94, 155, 179–81; intermediary 152–4, 158–62; quality of 132–3; for research services 207–8, 213–14; use or sell trade-off 155, 158–62 Marshall, A. 6, 17, 22–7, 45, 46, 264, 267, 366, 374–5 Marx, K. 5–6, 17, 29–38, 44–5, 46 Maxfield, R. 352 Meeus, M.T.M. 77 mergers and acquisitions 181 Metcalfe, J.S. 316, 321 microeconomic model 282–3 Milgrom, P. 112–13, 269 mobility, barriers to 303 modular divisibility 182, 186 modularity 109, 151–4, 165 Moglen, E. 189, 225 Mokyr, J. 265–6 monopolistic market power 87–8, 216, 218, 219, 220 multidimensional scope 57–60 multi-industrial technological districts 188, 188–9 multiple equilibria 116–18 multiple inventions 192 Murmann, P. 97–8 myopic agents 246 national economies 295–7 negative externalities 272–7 Nelson, R.R. 7, 44, 51, 67, 85, 88, 89, 136, 138, 165, 215, 372, 373 nested networking interactions 81 nested transactions 79, 178
nested transformation curves 10; diffusion and adoption 323–7; inducement of general and contingent change 290–3; localised product innovation 308–12 net profitability of adoption 322, 328–30 networking/networks 359, 362; complex dynamics of collective knowledge 268–77; costs 80–1; knowledge communication 102–3; network externalities 318–19; within geographical and technological clusters 187–9 neutrality 52–3 new companies 163, 164 new growth theory 43, 91, 265, 357 non-ergodicity 3, 334, 337–8 non-exclusive intellectual property rights 205–7, 215–42 non-rivalry in use 238–9 Nooteboom, B. 70–1, 101, 130, 131, 133 North, D.C. 368–9, 366–7 numerical control 155, 179 O’Donoghue, T. 97 Oerlemans, L.A.G. 77 open science 171, 194–8; knowledge cost function 209–11 open source software 189–91, 225, 370, 371–2 optimisation 50, 366 optimum levels of knowledge rents 229–37 original innovation 323–7 outsourcing: academic 176–7, 208; in-house 175 ownership, rights of 221 parallelogram rule 307 partial divisibility 170 partial endogeneity 43–4 partial substitutability 126 passive adoption 323–7 past dependence (historicity) 337–8, 340 patent fences 238 patent thicketing 185, 238 patents 87, 215–16, 240–1; academic 207; trade in 177, 370–1; see also intellectual property rights path dependence 3–4, 12–13, 57, 60–2, 119, 148, 333–55, 362–3; collective knowledge and 103–6, 346–53; degrees of 378; dynamics 337–46; knowledge as a path-dependent emergent property 106–7
Index 403 path-dependent adoption 342–6 path-dependent innovation 342–6 Patrucco, P.P. 81 Pavitt, K. 303–4 pecuniary knowledge externalities 115, 272–7, 367–8 Penrose, E. 7, 53–4, 70, 90, 123, 134 Perroux, F. 187 phase transitions 3, 334 ‘philosophers’ 19 Piedmont, Italy 277–9 platforms, technological 174–5 Polanyi, M. 89 policy 192; and academic system 211–14; knowledge generation 146–8; models of knowledge and 85, 86, 90, 92–3 Politecnico, University of Turin 277–8 Porter, M.E. 88 positive externalities 272–7 positive feedback 273; path dependence 338–9 prices 144–5; factor see factor prices principal-agent theory 75–6, 122–3, 198–201 private good, knowledge as 89–95 privatisation 93 procedural contracts 82–3, 184 procedural rationality 5, 49–50, 365–6 process innovations 307–8, 309–13 product innovation 11, 301–15; Lancastrian model of localised product innovation 304–13; proximity and 301–4 product markets 155; changes in 48; changes and strong irreversibility 248–9 product space 59, 301, 304–15 production 33–5; interdependence with transaction and coordination 124–33 production function 127–9, 143–4 production possibility frontier 145, 146 productivity 18, 21, 44; total factor productivity 285–7 professional communities 55, 58, 189, 198 professional university 198, 201–11, 371–2; dynamics of demand 209–11; incentive complementarity between public science and contract research 204–5; markets for research services 207–8; non-exclusive intellectual property rights 205–7 profit maximisation 36, 65, 67; strong irreversibility 252–4
profitability of adoption 319 prospect theory 51 proximity 11, 55, 58, 100–1, 301–15; geographical and technological clusters 24, 55, 58, 164, 187–9; Lancastrian model of localised product innovation 304–13; and product innovation 301–4 public funds 195–7; allocation 197, 212–13 public good, knowledge as 85–9, 165 public science, and contract research 204–5 publication 171, 195–6 punctuated change 265–8, 280 qualifications 201 quasi-hierarchies 167, 168, 169, 171–6, 182, 186 quasi-irreversibility 26, 52, 246, 248, 259–60, 339 quasi-proprietary good, knowledge as 89–95 quasi-public good, knowledge as 85–9, 165 quasi-rents 25 Quéré, M. 81 rationality: bounded 49–50, 246, 365; procedural 5, 49–50, 365–6 rationing, knowledge 81–2 recombination 71–3, 109 reflection effect 51 Reichman, J. 224, 372–3 rents, knowledge 222–3; optimum levels 229–37; quasi-rents 25 replicator dynamics 377 reputation 189, 195, 198, 201, 202–3, 205 research and development 71–3, 172 research services, markets for 207–8, 213–14 resource allocation mechanisms 197, 212–13 resource-based theory of the firm 8, 70, 90–1, 120, 123–4, 150 revenue: of adjustments 258, 310–11; of innovation 310–11; map of isorevenues 159–62, 258–9, 290–3, 308–12, 323–6; strong irreversibility 251–3 revenue-based compensation schemes 226–7 revenue functions 159 reverse engineering 166
404 Index Rizzello, S. 100, 378 Roberts, J. 112–13, 269 Romer, P.M. 43, 89 Rosenberg, N. 376–7 royalties 228–9, 237, 241–2 Rumelt, R.P. 304, 376 Ruttan, V.W. 283 S-shaped diffusion process 328–30 salaries, academic 203–4, 371 saltationism 11, 264–8 Salter, W.E.G. 37, 286, 294–5 Samuelson, P. 37, 282, 283 scale-free networks 102, 105 Scheinkman, J.A. 5, 55 Schivardi, F. 102 Schumpeter, J. 6, 17, 38–41, 42, 45, 46, 48, 56, 91, 139, 264–5, 267, 317 science-push model 86 scientific entrepreneurship 94–5, 172 scope economies see economies of scope secrecy 216–17 sectoral technological districts 187–8 selection 39 selective diffusion 43 sell or use trade-off see use or sell tradeoff shift 291–4 signalling: costs 79; patents and 238–9 Simon, H. 5, 49, 365–6 Smith, A. 5–6, 17, 17–22, 27–9, 44, 45, 46 social capital, endowment of 132–3 social interactions see interactivity social welfare 65, 67, 227–9 socialisation 71–3 soft endogeneity 42–3 software 189–91, 225, 371, 371–2 Solow, R.M. 286 specialisation 22 specific knowledge 43, 89 spillovers 42–3, 268–77 sponsored spin-off 185 Stallman, R. 189, 225 standardisation committees 185 standards, ex ante 164 State 87; academic system 195–7 static efficiency 88, 217–18 static knowledge transaction costs 76–7; demand-side 76–8; supply-side 151–4 stickiness 7, 111; quasi-hierarchies 168, 171–6, 182, 186 Stiglitz, J.E. 2, 6, 53, 54, 57, 70, 75, 87, 94, 179, 364, 367
Stoneman, P. 317, 321, 331 strong endogeneity 44, 47 strong irreversibility 247–54 structure 349–52 structure-conduct-performance approach 41, 350–1 substitution 31–6, 44–5, 294–6 supermodularity 112–13, 269, 272–7 supply 118; adoption, diffusion and 320–1, 323–7; optimum level for knowledge rents 232–3, 233–6; supply-side transaction costs 76–8, 151–4 surplus value, extraction of 29 survival, struggle for 23 switching see technical change Sydak, J.G. 112 Sylos Labini, P. 41 synchronic complementarity 109 system architecture 4, 339–40, 349–52, 358, 359, 360–1 system dynamics 4, 11, 27, 60–2, 103–6, 260–1, 263–81, 361–3; networking and spillovers 268–77; Piedmont 277–9; smooth vs discontinuous change 264–8 systemic path dependence 348–53 tacit knowledge 73, 89, 136–8, 170, 302; quasi-hierarchical command 168, 171–6, 182, 186 taxation 195–6 Taylor, M.C. 61, 378–9 teaching/training, in universities 196, 201–2, 203–4 technical assistance 77 technical change 254–9, 290–3, 307, 309–12, 323–7, 373–4 technical space 56, 57 technical substitution 31–6, 44–5, 294–6 technological clubs 184 technological clusters 24, 55, 58, 164, 187–9 technological knowledge externalities 115, 367–8 technological platforms 174–5 technological resilience 318 technological systems 57–8; see also system architecture, system dynamics technology payments 177, 370–1 technology policy see policy telecommunications 152, 178–9, 220–1 theories of the firm 7–9, 90–1, 120–34; competing 121–4; dynamic context
Index 405 and complexities of interdependence 130–3; interdependence between production, transaction and coordination 124–9; knowledge generation see knowledge generation; resource-based theory 8, 70, 90–1, 120, 123–4, 150; use or sell trade-off see use or sell trade-off time 57 topology 335–6 total cost equation 160 total factor productivity 285–7 total surplus 228–9 transaction costs 8, 76–8, 120, 121–2, 156–7; demand-side 76–8; dynamic knowledge transaction costs 77–8, 156; static knowledge transaction costs 76–7; supply-side 151–4 transactions: coordinated transactions 167, 169, 176–82, 186; economies of scope in 78; interdependence of production, coordination and transaction 124–33; nested 79, 178 Tversky, A. 51 unit organisational costs 76, 200 United States of America 295–6 universities/academic system 9, 68, 87, 92–3, 171–2, 193–214; from open science to academic management of creativity 194–8; outsourcing of research to 176–7, 208; policy implications 211–14; professional university 198, 201–11; solution of principal-agent problems 198–201
usage, rights of 221 use or sell trade-off 8–9, 150–66; demand-side transaction costs 76–8; knowledge exploitation strategies 156, 162–4, 165–6; model 158–62; rationale 154–7; supply-side transaction costs 151–4 user interdependence 219 user value 238–9 value: machinery 30–1; theory of 27, 28, 37; user value and exchange value 238–9 variety 22–9, 39, 101 venture capitalism 82, 94, 180, 181 vertical integration 153, 154, 156–7, 162–3, 165, 217 vertical knowledge interactions 188, 189 vertical technological clubs 184 Von Hippel, E. 91, 302 wages 31–2; academic salaries 203–4, 371 Wallis, J. 368 weak endogeneity 42 weak irreversibility 254–9 Weiss, A. 87 welfare 65, 67, 227–9 Williamson, O. 8, 121, 134 Winter, S.G. 51 Witt, U. 316 workers, number of 33 Young, A. 27