A Summary of "The Co-evolution of Technology, Industrial Structure, and Supporting Institutions" by Richard Nelson

Preface

The following is a discussion of Richard Nelson’s paper “The Co-Evolution of Technology, Industrial Structure, and Supporting Institutions”.   The purpose of these posts is to summarize key innovation themes from academic papers, thereby introducing these concepts to a broader audience.  The key takeaway from this paper is that there are deficiencies in economic models of growth as they overlook the co-evolution of growth and institutions with the dynamics of new technology.  

All of the worthwhile insights in the following paper are attributable to the original authors and all mistakes and omissions are my responsibility.    

Introduction

There is a significant intellectual discrepancy between most economic growth models and the description of growth from economic history.   Contemporary economic models treat economic growth quantitatively considering variables like per capital income, real wages, saving rate, rate of return on capital with very little qualitative analysis.   On the other hand in the economic historical accounts there is lot of qualitative events such as new technology, new organizational forms and new institutions.  

The main reason that Nelson and Winter began to develop an evolutionary theory of economic change was a belief that an appreciative theory of economic growth (including historical accounts) and formal theorizing could be brought closer together.   The language of evolutionary theory seemed natural to describe and explain the detailed historical accounts of growth.  The language abstracts and simplifies the historical picture without distortion and remains amenable to powerful formal modelling.

However, the formal models of evolutionary economic growth remain largely quantitative in their depiction of growth, despite significant appreciative theorizing in many respects.  This essay is an attempt to reduce the gaps in formal theories ability to capture developmental aspects of economic growth at an industry or sector level.  One can now reasonably begins to lay out a story of the growth, and development of a manufacturing sector, from birth, maturity and death that will seem to fit many (but not all) cases and can serve as a target for formalization.

The Co-evolution of Technology and Industry Structure

One of the motivators behind the author's early work with Winter was a desire to incorporate technological advance into an economic growth theory.  They treated technological advance as an evolutionary process, where new alternatives compete with each other and the prevailing practise.  Ex-post selection determines winners and losers, usually with considerable ex-ante uncertainty about which the winner will be.   Nelson and Winter highlighted the elements of uncertainty in technological evolution, however they also stressed the systematic selection mechanism that many markets provide.  Due to understanding of user needs and technological systems, the technological mutations introduced into the market are not strictly random.  These dynamics of technological evolution combined with systematic section provide direction to technological advance.

Nelson and Winter observed that in many fields technological advance was cumulative, with today’s advances being built upon yesterday’s and improving it in various directions.   In many of these cumulative technologies ’natural trajectories’  appear, reflecting both what the technologists believe they can achieve and what entrepreneurs believe users will pay for.  This cognitive structure and the agents within it are called a ‘technological regime’ or ‘technological paradigm’.   In this report the authors use a particular technology or industry (or combination) as the unit of analysis (versus the entire economy) with the effort directed towards developing generalizations that can hold across technologies or industries.

The structure of the analysis begins with discussing the body of research that focuses upon the ‘life cycle’ of technologies.   The basic arguments were presented by Abernathy and Utterback (1978) who suggested when a new technology emerges there will be considerable uncertainty around which of the many technological variants will succeed.   Different technological variants will be trailed by different users, however, over time competition between the variants will lead to a few of the technologies variants dominating the others (I.e. A ‘dominant design’ emerges).  Then attention and resources will be concentrated on those variants at the expense of others.  There are several explanations regarding the emergence of dominant design.  One obvious explanation is that one technological variant is simply superior to others and over time the superiority of this variant will become identified and widely known.   However, there are also more complex reasons for the emergence of dominant designs.

If competing technologies are cumulative, then a technological variant that enjoys an early advantage (which may have resulted only from chance) may lead to the race to a dominant design ending very quickly.   Once a technological variant gains an advantage in the market, there is a rational reason for resources to be allocated towards its advancement, since rival technologies may require major development to make them competitive.   In this context, there is no reason to believe that the dominant design is the optimal choice for the society.   It could well be the case that a different technological configuration would have proven superior if it had been developed adequately.

Another explanation of dominant design stresses systematic aspects such as : the growth in number of users of the variant, the development of skills related to use of the variant, or through investments in complimentary products to the variant.  Again, there is no reason why the ‘standard’ technology that is adopted and ‘locked-in’ by society need be the optimal technology.

Once a dominant design emerges, radical product innovation slows, and product design improvements become incremental.  However, there may be substantial improvements to process technologies and if the process technologies support a particular technological product variant the cumulative process innovation will further lock in this variant.

Abernathy and Utterback developed the dominant design story based upon detailed observation of the automobile industry, and while there are many industries where this approach proves illuminating the authors are skeptical that dominan design dynamics can be applied to all industries (e.g. Pharmaceuticals and chemical industries have structures where dominant design doesn’t seem to apply as well).  Nevertheless, where it does apply dominant design theory raises some interesting and troubling economic issues (e.g.  Can one presume that market forces generate efficient outcomes?  What does one exactly mean by market forces?).   Some researchers have gone so far as to argue that social consensus, rather than economic efficiency, which determines the path of technological adoption.   This raises the question of how far economic selection arguments matter in an evolutionary analysis of economic change.  Also to what extent do political and social forces need to be explicitly taken into account in determining the primary path of technology development (more on this later).

Another body of research focuses upon what happens to firm and industry structure.  The basic ideas tend to be these.  During the early periods of experimentation, before there is a dominant design there is no advantage to incumbency as market demand is fragmented across a range of variants, the producing firms tend to be small, product models change frequently and there is considerable entry and exit from the industry.   However, once a dominant design in established companies that do not produce a variant of the dominant design tend to drop out of the industry or into small niche markets.   As the product design is more stabilized the learning from incumbent firms becomes more cumulative and potential new entrants are increasingly at a disadvantage.     With the market less fragmented and more predictable, companies can try to exploit economies of scale and advances in process technology will reflect and enforce this.  There may be a ‘shake out’ in the industry with the surviving companies being fairly large.

Aside from the dominant design there is another theory about why innovation shifts away from the creation of new products and towards process technology as an industry matures, a theory which has nothing to do with inherent tendencies in technological development.  Rather, product innovation depends upon the number of firms and process innovation depends upon their size.  As the industry evolves firms get larger and there are fewer of them, thus the changes in the character in innovation for an industry is driven by changes in market structure.

The story of changing firm and industry structure that has been presented to this point is tied to the notion of organizational learning and how companies develop a complex of ‘core capabilities’ that allow them to thrive in their particular context.  The basic idea is that companies (at best) have a limited number of things that they can do well, capabilities that emerge from considerable learning by experience and significant investment in these ‘core capabilities’.   The research on ‘dominant designs’ suggests that these capabilities cannot be developed when a technology is still in flux, but only after orientations become clear.  On the other hand, since a companies core capabilities have a limited range of applicability suggests that firms tailor their core capabilities to the requirements of the prevailing dominant designs.     Thus, established firms may have considerable difficulty to adjusting and leveraging new capabilities when important new technology have the potential to replace prevailing technologies emerge.   Of course, companies to not stand alone.  They operate in the context of competition with rivals, being served by suppliers, selling to customers and drawing upon available skills.   For an industry with specialized inputs and skill needs its growth will be strongly conditioned by how rapidly and effectively the support structure evolves in the new technological context.

The Evolution of Supporting Institutions

Generally, the evolutionary processes described above proceed in a market setting and involve competition amongst firms with selection primarily being determined by market forces.  A societies institutions, both general and specific to an industry, will influence the parameters and shape of the processes being modelled.   The features of an institutional environment themselves will evolve with the pushes and pulls inherent by the development of a new industry.   The processes here are not market forces, but involve the creation of collective bodies, decisions of voluntary organizations, government agencies, and political action.  One important development that invariably occurs with the emergence of a new industry is that the people within the new industry become aware that there is a new industry with collective interests and needs.  Industry or trade associations form and may play a role in determining the standards for the industry.  They also can provide the industry with lobbying which can evolve the institutions towards the new industry’s liking (e.g. Protection from outside competition, public support programs etc.).   This feature of an industries evolution can facilitate the locking in of the status quo.

If the technology underlying the industry has novel characteristics, then new technical societies and new technical journals tend to spring up or even a new ‘field of science’ may come into being.   Sciences, in general, do not aim to solve practical problems per se, but rather aim to enhance the basic understanding of nature.  This enhanced understanding does then enable technological advances.  However, quite often a new technology can come into existence with little understanding of the scientific principles that are relevant to its (e.g. Steam engines emerged which then launched the scientific field of thermodynamics; computer science emerge after the development of the computer).  The appearance of these technology orientated sciences tend to tie industries to universities, which provide training and research which can enable further technology development.   The presence of university research dilutes the knowledge advantages that incumbent companies could have over new industrial entrants.   Also, research at universities may just become the source of radically different technological alternatives.

Recognition of the role of technical societies and universities in the development of technology opens the door to seeing a wide range of institutions that can co-evolve with technology.  There can be issues relating to intellectual property or regulations (e.g. Biotechnology).   There may also be new public sector activities or programs required to support the new industry (e.g. Automobiles require roads, radio required spectrum management).  The evolution of institutions related to technology or industry may be a complex process that involves not only companies, but industry associations, technical societies, universities, courts, government agencies etc.   Given the range of interests involved in this process, there is often little hope of institutions responding optimally to accommodate new technologies.

Punctuated Equilibrium

The author wants to discuss what happens to a mature industry when radical new developments occur that require significant change (e.g. Transistors displacing vacuum tubes, biotechnology displacing mature pharmaceutical processes).   A key topic revolves around who adopts and diffuses the new technology, do incumbent companies adopt it or is diffusion dependant upon new industry entrants?  The answer depends largely upon whether the new technology employs the same capabilities and skills as the old technology.   If so, then the incumbent firms tend to adopt the technology, if not new firms enter the industry and the failure rate amongst incumbents tend to be high.   A broader question is whether the larger set of institutions supporting the technology and industry are able to adapt,  if not, then the institutional structure can hamstring both the incumbents and the new industrial entrants.   Thus, to be effective with new technologies a nation requires a set of institutions that are compatible and supportive.

Economic Growth and the Formation of Comparative Advantage at a Sectoral Level

At the start of this paper, the authors noted that all formal growth theory focuses upon the quantitative aspects of growth and recognizes very little of the developmental or historical aspects.   The theory developed above suggests that there is a ‘developmental’ flavour to quantitative aspects (e.g. Capital intensity, total factor productivity) as an industry matures.  There are also interesting connections between these variables and firm size and industry structure.   In particular, one might expect that rising capital intensity and the development of large-scale units will occur after a dominant design or mode of production have been established.   However, if the effective use of the technology requires institutional evolution there may be a significant lag before the new technology results in increased effects on productivity.

The most intriguing question for the author about this approach concern the factors that lie behind the establishment of a comparative advantage in a new industry.  Comparative advantage in certain fields is created rather than being innate in country level variables and public and private actions are needed for comparative advantage to be built.   Furthermore, certain factors may be more important early in an industries history compared to later.   When an industry is new, comparative advantage may be driven by a regions ability to easily crete new firms, provide them with funding and how open the markets are to new sources of supply.  Regions may also differ in the speed by which universities can adopt new sciences, in how quickly legal structures can adapt to the demands of new technologies and how supportive public policies are of new technologies (as contrasted with policies that protect the old technologies).  However, once a dominant design is established different institutional aspects become important such as the ability to finance large scale investment or the ability to train skilled labour.

Source

Nelson, R (1994) Industrial and Corporate Change, Volume 3, Issue 1, Pp. 47-63