Innovation And Industrialization

NORTH- HOLLAND

Innovation and Industrialization: A Long-Term Comparison G. N. V O N T U N Z E L M A N N

ABSTRACT This article aims to link the micro-level changes in firms, as the source of production behavior, with mesolevel changes in industrial structure and macro-level changes in growth and development performance. It focuses on the three great industrial revolutions of the last quarter of the present millennium. These differed among themselves in almost every major way, which inhibits generalization but shows that systems (here the "national systems of production") are very different, and that "convergence" of a later industrializing country upon its predecessor is improbable. Each industrial revolution possesses considerable internal logic but is less flexible in regard to adopting features of its successor. As a result, mismatches arise over time between the specified constituents of the production systems, as demonstrated by economic phenomena such as unemployment and political phenomena such as ideology. The task of resolving such mismatches has fallen back on the micro level of firms and households, which itself has imposed serious strains on the productive system. Such heterogeneity imposes severe limitations on the ability to link technological forecasting and social change in the long term. © 1997 Elsevier Science Inc.

Introduction The last quarter of the millennium just coming toward its close has had one overwhelming characteristic that sets it aside from all of its predecessors: that of sustained industrialization. This does not mean that it has pursued a continuous linear advance throughout this period, and in this article I shall examine the specificities of time and place that mark its discontinuities. What can nevertheless be argued is that industrialization and indeed economic growth in this quarter-millennium has been qualitatively as well as quantitatively very different from anything that preceded it. In this industrial era, change has become the norm, whereas in pre-industrial times change--although it did occur sporadically--was the exception ([1], page 277; [2], page 43). This has meant that any prospective entrepreneur contemplating an industrial development would have to take into account not just the changes he or she was intending, but also the likelihood that the broader industrial environment outside would also be changing. Thus entrepreneurs would have to aim at a moving target. Lest this point seem too obvious, it is worth reflecting on the point that much of our theoretical analysis of industries and economies is still based on the presupposition DR. G. N. VON T U N Z E L M A N N is Reader in the Economics of Science and Technology at the Science Policy Research Unit at the University of Sussex, England. Address correspondence to Dr. G. N. yon Tunzelmann, Science Policy Research Unit, University of Sussex, Mantell Building, Falmer, Brighton BN1 9RF UK. Technological Forecasting and Social Change 56, 1-23 (1997) © 1997 Elsevier Science Inc. 655 A v e n u e of the Americas, New York, NY 10010

0040-1625/97/$17.00 PII S0040-1625(97)00027-9

2

G.N. V O N T U N Z E L M A N N

that change continues to be a one-off affair. The methodological basis of the discipline of economics, to quote the one I happen to know best, remains "comparative static"; that is to say, examining a once-for-all change in order to assess the departure from and especially the subsequent return to equilibrium. The notion that equilibrium is at least moving onward, and perhaps might better be envisaged as disequilibrium, continues to be one from which most orthodox economists shrink. A basic reason for this rejection of everyday observation, that the world does not stay the same and does not generally return to its original point if the change element is reversed, probably lies in the quest for generality and "rigor" [3]. The simpler the underlying set of axioms from which theory is derived, the more powerful the theory is seen as being. But such predictability is obtained at the cost of the inadequacy of the predictions such theory is able to make. The problem of beginning with the contrary supposition that the world is continually changing is that of establishing the rules and patterns according to which it changes. The present article adopts the framework suggested by Dosi [4], itself drawn from work on the philosophy of science, of paradigms, heuristics, and trajectories of change. Although the primary emphasis is on technological change, the basic argument is that change permeates all functions undertaken by firms and by the economies that contain them. The very complexity that emerges defies any straightforward application of covering laws or general principles of economic development. Instead the main emphasis will be on unearthing the "stylized facts" [5] which best describe industrial and technological development. These procedures are applied to what are characterized as the three main industrial spurts ("revolutions") of the last quarter-millennium. The main findings of this article are that: (1) each industrial "revolution" differed in almost every respect--technology, processes, organization, etc.--from the other two; (2) nevertheless there was some degree of internal consistency between this variety of characteristics within each industrial revolution; (3) there was no unilateral development of things getting "bigger and better," and in certain respects the most recent era has shown greater similarity with the first than the second "revolution"; (4) the temporal patterns of change (which functions changed first, etc.) do show some consistency from one industrial revolution to the next, although the causal factors involved are still obscure. It is concluded that interrelating technological forecasting and development with social and other broad types of change is very specific to time and place, although a few generalizations--and not often those that are much espoused--may be possible. A recent book-length study by the present author has tried to give a general overview of the evolution of industrial technological structures [6]. This article draws on those findings (together with other work), but delimits it by restricting attention to the "spurts" of industrialization in the successive industrial leaders. That is, I take the three great "industrial revolutions" charted by various authors, and examine the structural characteristics of industry during each of these "revolutions" [7]. I Although this narrows the diversity of experience, by looking just at the three archetypal leaders-Britain in the first industrial revolution from the late 18th century, the United States in the second from the late 19th century, and Japan in that which seemingly commenced The issue of how revolutionary these "industrial revolutions" were has been considered by many scholars, such as the contributors in [7]. The text is not meant to imply that there is any unanimity of views among scholars about either the number or the extent of these "revolutions"--particularly for the "third," whose shape is still evolving. Some reject any notion of "industrial revolution" at all, though patently this is not the position taken in the present article.

INNOVATION AND INDUSTRIALIZATION

3

I TEc" IQI I I

E I~i s I

sl

"'~

ENTREPRENEURSHIP I--~ .' ,,.._ . 1 E

<--- V

<

Fig. 1. A Micro-level taxonomy for production in the firm. Key: circles = exogenous change-agents, diamonds = endogenous change-agents, rectangles = spheres.

in the latter part of the 20th century2--it is still sufficient to reveal just how great the differences were. The taxonomic structure for this comparative study is based on that used in the earlier work, and is set out in abbreviated form in Figure 1 [6]. 3 The central functions involved in a system of production, shown by rectangles in the diagram, are regarded as being u n d e r t a k e n at the level of the firm. The format displayed in the figure comes partly inductively from historical experience, and partly deductively from this author's view that the basic rationale for the existence of firms is that such organizations are 2This does not mean to imply that Japan will necessarily continue to lead the "industrial revolution" which I believe is currently under way. Leadership might well pass back to the United States or forward to a newer industrializingcountry such as China. 3Compare with Figure 12.1 in [6], which shows greater detail.

4

G.N. VON TUNZELMANN

required to combine and integrate different types of knowledge. Concretely, firms exist to transform knowledge about technologies into knowledge about products. Technological knowledge coming in at the top of the diagram eventually has to square with market knowledge coming up from the bottom of the diagram. The problem is that the ways in which technicological knowledge is structured differ, often dramatically, from the ways in which product/market knowledge is structured. Thus firms are at heart responsible for cross-relating the two different structures of knowledge. Two types of function facilitate this: production processes can be organized to streamline the conversion of inputs into outputs, and administrative procedures (managerial, accountancy, legal, etc.) can be devised with similar objectives in mind. These are shown on the left- and righthand sides of Figure 1. Management, in the sense of an overall coordinative function, is however allotted a central role over and above its administrative duties, and is shown in the ellipse in the middle. This function itself may become "entrepreneurial" in the sense of Schumpeter if management is able to go beyond the reactive role of responding optimally to a given set of competitive conditions, toward a proactive role of changing those conditions [8-10]. A more complete description of these functions is given elsewhere [6, 11]. Such functions include capabilities for changing themselves endogenously, shown by the arrows linking them outward to the "endogenous change-agents," and represented by diamond shapes in the figure. At this level, sectoral differences and meso-levels of analysis are likely to be highlighted. Outside them again lie the "exogenous changeagents," depicted by circles---exogenous, that is, to the individual firms of which the economy is composed. It is assumed, for simplicity, that overall levels of science and technology (S&T), finance, demand, etc., are determined at the national level. Of course it cannot be denied that some firms in some economies may be large and powerful enough to have a significant influence on the national characteristics, but not much is lost by ignoring this feedback here [12, 13].4 Also overlooked here, at least in the specification, is the international level of interaction, which would require considering these circles as not just a "national system of production" but also an international system. Some of the issues arising in this way will be considered below, but for the most part the focus on the successive international "leaders" carries the implication that national and international levels roughly coincide. The specific functions of the national systems are each assumed to change according to a given paradigm or set of paradigms (the differing sets relating to different sectors). There will thus be one or more paradigms relating to technology, to organization, to finance, and so on. The notion of "paradigm" here evidently has an affinity with Kuhn's [14] much-debated notion of the scientific paradigm, although arguably becomes even more controversial when applied to areas such as technology or organization. The paradigm is intended to represent the area most likely to be trawled to solve technological (or other) "puzzles." Within the paradigm(s) change takes place according to certain "heuristics," a term Dosi borrows from Lakatos [15] to indicate the factors which promote searching in particular directions and not in others. The "trajectories" pursued in one sense represent a sub-set of the heuristics, inasmuch as they show the choices actually made from within the domain of possibilities, which has already been narrowed through the selection of particular heuristics. However in another sense they represent a widening of the focus, because they are likely to involve the intersection of each 4Allowingfor it would simplyrequire turning the one-wayarrowsfrom exogenousto endogenouschangeagents into two-wayones. For general discussions of "national systemsof innovation," see [12] and [13].

INNOVATION AND INDUSTRIALIZATION

5

functional domain with other domains. For example, the trajectories followed by technology in a certain country at a certain time will to an extent be influenced by the surrounding circumstances, including resource endowments and labor supplies, provision of capital, consumer tastes, etc., as well as by the momentum of the technological heuristics themselves. Just how far they are thus intersected will be considered below. The schema is built up in micro-to-macro fashion. As a result, there are many issues that arise at the meso or macro level which are not made manifest in Figure 1. Figure 2 takes a top-down instead of a bottom-up approach. The great simplification in this diagram is to assume a closed economy--international aspects will be considered below. That aside, the various components of this schema are not very controversial. What has been the subject of intense debate in economic policy making for the last two centuries have been the arrows depicted in the diagram. In a previous work I have utilized the framework of classical economics, and the "circular flow" of interaction between inputs and outputs, to interrelate the elements it contains ([16], Figure 2.1). Figure 2 aims to be more general. Classical economics as practiced by Adam Smith and his successors down to the middle of the 19th century subsequently split into a number of schools of economic thinking, who have feuded with one another up to the present day. The "neoclassical" tradition began at about the time of the second Industrial Revolution, in the last quarter of the 19th century. The contrasting Keynesian school originated in the interwar works of Lord Keynes, especially his General Theory [16]. A third school, of "evolutionary" economists, originally developed alongside the neoclassical tradition (e.g., in the works of the English economist, Alfred Marshall), although its recent resurgence has led it to take on more heterodox standpoints, influenced partly by the works of Joseph Schumpeter. Without elaborating unnecessarily on these debates, it is perhaps helpful to point out three ways in which these schools of thought differ. The first is that they take different positions regarding the construction of their analytical frameworks. Specifically, the neoclassical school typically aims to start with the micro level, which is enunciated through a set of axioms relating to individual units (lirms, consumers, etc.); the objective being to build toward a macro-level system from these micro foundations. The Keynesian school reverses this and concentrates predominantly on the macro level, although some members of the school have in this context given primary attention to micro aspects (such as the role of "imperfect competition"). The evolutionary school works on each level, although perhaps most often begins from the middle (meso) level, because of its concern with technological and organizational systems. The second difference is in their orientation to equilibrium. The neoclassical school in this sense represents the conventional position outlined above, of assuming oncefor-all changes and subsequent return to equilibrium, dictated mainly by "market mechanisms" of price adjustments, etc. The Keynesian school thinks more of disequilibrium, with market instruments such as interest rates often being powerless to ensure any rapid return to equilibrium. The notion of disequilibrium here is oriented toward the short term, in the light of Keynes' mordant dismissal of the tendency of the neoclassical model toward equilibrium in the long run: "In tile long run, we are all dead." The evolutionary model takes a stronger position on disequilibrium, as it sees the potential for disequilibrium even in the medium to long term. An important reason is that technological trajectories and economic developments can pursue their own, often somewhat independent, historical paths, and fail to synchronize over long periods of time. One branch of the school argues for patterns of "punctuated equilibria" as in

6

G.N. VON TUNZELMANN

Government

I

~ Macroeconomy<

c

\.

"~, I~rices ies

quantities~

/

SECTORS:

RESOURCES:

Labour

Agriculture

Capital

Manufacturing Services

HumanCapital Natural

Production System

Distribution System

Consumption System

Firms

Exchange

Households

Demand ~ forInvestment

FinancialSystem.~._.___.~~/~Supply ofSavings

Fig. 2. Aggregate supply and demand in a dosed economy.

INNOVATION AND INDUSTRIALIZATION

7

some branches of evolutionary biology, and this fits naturally into the present discussion of spurts of industrialization. The third difference is that the neoclassical school sees any change (of the oncefor-all kind) as coming pre-eminently from the supply side. The demand side reacts passively to any changes in supply, thus maintaining equilibrium. The Keynesian school argues for the primacy of changes on the demand side, coming about through fluctuations in entrepreneurial behavior toward investment, government behavior to macroeconomic policy, etc. The evolutionary school focuses mainly on the supply side, in this respect like the neoclassicals, although seeing change as more or less permanent rather than one-off. This latter point is worth pursuing a little further. The subdivisions into resources on the supply side and sectors on the demand side in Figure 2 are merely suggestive, and can be substantially redivided or further subdivided. According to the "crude neoclassical" viewpoint, the supplies of resources in the economy are givens, so they constitute "endowments." In equilibrium, they will be fully employed. This is why an increase of demand, expressed perhaps through rising prices, will not enlarge supply. A contrasting view is that supplies can be augmented ("enhanced"), either exogenously or endogenously. If the enhancement of supply is exogenous, the natural inference is that there is no great likelihood that aggregate supply and demand will correspond. Movements away from equilibrium are likely to predominate as unexpected new sources of supply emerge. Conversely, if demand patterns are shaped to approximate supply characteristics, growth and development are much more likely to be sustained. The extensive investigations of the French "r6gulation school" have been particularly targeted at establishing how aggregate demand has been boosted to satisfy aggregate supply under conditions of 20th-century production processes (Fordism, etc.), and whether this has been breaking down in recent times [17, 18]. Once Figures 1 and 2 are matched, it becomes obvious that a wide range of considerations arise in any particular set of historical circumstances. The technologies evolve within the production system of Figure 2, through the context implied by the figure as a whole (the macro-economy and its characteristics). Complexity increases when we take into account the interaction between economies, (i.e., the international aspects, which can be thought of as a third [spatial] dimension of the figures), and further still when we account for historical evolution, using time as a fourth dimension in each case. The rest of this article assesses the nature of the implicit interaction between micro, macro, and global levels in the light of the constituents of Figure 1. Exegesis of Figures 1 and 2 is here conducted by an attempt to fill the "empty boxes" of those figures through detailed historical evidence. The results to date are set out in Table 1. There is only limited space here to justify the cell entries in this table, in cases where the choices are likely to be especially controversial. Many of the individual inferences are referred to in recent writings [6, 19], although these in turn come from a large range of secondary sources. The table itself and the discussion below have not been presented before. The columns of Table 1 show the features of each successive industrial revolution in regard to the structural characteristic noted in the first column. The order of the rows is such as to move down through Figure 1. Some sections begin with the "exogenous" change-agents (the circles of the diagram) and others with the firms' functions (the rectangles), depending on the relationship to the preceding item. The cell entries are shown in three ways: one in normal (roman) font, referring to aspects that were

8

G.N. VONTUNZELMANN

present more or less from the beginning of the particular industrialization spurt; a second underlined, showing those that emerged roughly midway through the spurt; and a third italicized, showing those that arrived only in the later states of the spurt. This therefore attempts to capture the temporal dimension of the industrialization patterns. The discussion in the text relates to each of the "characteristics" (first column of Table 1) in turn, with the characteristics themselves being shown in italics, with the three industrial revolutions noted by country.

Technology A natural inference from the so-called "linear model" of science and technology (S&T), which envisages a process of science creating discoveries, which get turned into inventions and then innovations before diffusion as technologies, is that scientific breakthroughs constitute the foundation of technological development. Many celebrated scholars have indeed argued this to have been the basic cause of the industrialization spurts (e.g., [20], page 249; [21], page 2; [22]). On this argument, the first industrial revolution--in late 18th-century Britain--can thus be traced back to the "Scientific Revolution" of about a century before. The historical evidence does not appear to support such claims. In the British case, the contribution of explicit scientific findings to technology was minimal: what instead the scientific revolution provided was the experimental method, (i.e., a procedure for logical investigation), together with some of the instruments that allowed such analysis. The major scientific advances that carried direct implications for technology, like the discoveries of the laws of thermodynamics, were more likely to be the result of technology than the cause (the "reverse causation" noted in Table 1). Science, of course, steadily accreted and by the time of the later industrial revolutions had become more practical in its adaptability to technology. But even then, the industrial impact of science was a later rather than earlier feature of industrialization, and technology in this particular sense of "know-why" still owed m u c h - - o r even most--to shopfloor learning ("know-how'). In-house R&D in larger firms became the main means for relating science to know-how and vice versa, for practical use [23]. In recent times new fields that fuse some characteristics of abstract science with others of practical technology have arisen--these are referred to as the "transfer sciences," such as biotechnology [24]. Although it is the major scientific breakthroughs that have been given most attention, the technological contributions have been just as important, such as in the development of purer silicon for the transistor. The technological rather than the scientific paradigms have therefore set the scene for industrial advance. In manufacturing, machinery formed the basis for technological development in both the first and the second industrial revolutions, and was still important at least in the early stages of the third one [25]. Machinery is regarded here as a "paradigm" in the sense that, when a problem ("puzzle") arose in manufacturing, the solution first turned to was likely to be one of developing a machine to overcome the difficulty. Although nowadays machinery and technology are often thought of more or less in the same breath, this was not the case at the beginning of the British industrial revolution, where machines were just one option among a range of possibilities. Indeed, in sectors other than manufacturing, other paradigms continued to predominate. In British agriculture, biological and chemical solutions were more likely to succeed, and machinery was uncommon in the British countryside until a century later ([26], chapter 5); similarly in mining, coal-cutting machinery did not arrive for over a century. Thus the British pattern was one of sectoral diversity of technological paradigms. Under the different resource constraints discussed below, American advance involved applying

INNOVATION

AND INDUSTRIALIZATION

9

TABLE 1 Characteristics of the Industrial Revolution

Characteristic Technology Science and technology

Britain: c. 1760-1815" c. 1790-1840 b

United States: c. 1870-1910 d c. 1890-1930 ~

c. 1820-1860 ~

c. 1910-1940 f

Japan: c. 1950-1990 g c. 1970-h ?~

Scientific method Reverse causation ~

Applied sciences ° In-house R & D ~

Transfer sciences h Continuous improvement ~

Machinery" Steam power b

Machinery ~ Electricity~

lnformatics h Environmentalist h

Sectoral differences ~

Sectoral similarities

Fusion ?

Heuristics

Time-saving a Land-saving"

Time -saving J Labor-saving ~

Time-saving Space-saving s

Trajectories

Labor-using" Energy-using b

Capital-using d Resource-using d

Skilled-labor -~ Resource-saving h

Mechanization (transformation)

A u t o m a t i o n (transfer) ~

A u t o m a t i o n (control) h

Technological paradigms

Internal c o m b u s t i o n f

Processes New processes Capital Labor Learning Engineering design Administration Management

Proletariat ~ Division of labor ~

Taylorism

Decision-making

Division of capital

Multiskills, teams ~

Learning by using a

Vertical learning d

Horizontal learningg

L e a r n i n g b y doing ~

F o r m a l learning I

Interactive learning'

"Shop culture . . . . . Professionalization ~

School culture" d Commercialization

Production engineering Corporate controlg

Entrepreneur ~

Hierarchical~"

Strategic h

Organization

Factory system ~

Big business f

Networks

Scale and scope

Small size" Disintegration"

Large size a Process disintegration d Product integration ~

Decentralized, hubg Network integration ~ Flexibility h

Metals (iron)" Chemicals (heavy) h Textile fibers" Coal ~

Metals (steel) ° Chemicals (organic) ~

Plasticsg Bi0technology h

New products

Products New resources

Synthetics f

S m a r t materials ~

Electricity, oil ~

Silicon h

Import substitutes (textiles)" "Decencies" ~

Import substitutes (metals, chemicals) '~ Consume] durables~

Import substitutes (electronics) g Consumer electronics

Product innovation

"Trickling down" a Quality innovation"

Homogeneity d Mass consumption ~

Differentiation Miniaturization h

D e m a n d structure

Social class a Urbanization ~

Egalitarian J Migration J

Rising wealth g International g

Marketing

Merchants Shops

Mass distribution d Learning by selling d

Merchant housesg Outsourcing h (continued)

10

G.N. VON TUNZELMANN TABLE 1

Britain: c. 1760-1815" c. 1790-1840~ Characteristic Meso/Macro Inter-firmlinkages Infrastructure

1820-1860 ~

(cont'd) United States: c. 1870-1910d c. 1890-1930~

Japan: c. 1950--19908 c. 1970-h ?~

c. 1 9 1 0 - 1 9 4 0 f

Atomistica Industrial districtsb

Oligopolistic Antitrust

Groupsg Consultativeh

Water transport Gas lightingb

Roads, air g

Telegraphy e

Railroadsd Electric lighting° Telephony

Railways c

Roads, air f

Information highways i

Sectors

Differentiationa

Linkages~

Assimilationh

Finance

Merchants Local privateb

Financiersd Stock market°

Network g Global capital~

Government

Laissez-faire?

Business-ledd Select tariffsd

Consultativeg Select technologies~ Export-oriented

a

Free trade (Imperialist) c

Macroeconomic f

Compiledfrom a wide range of secondary sources. See text for clarification. aBritain, c. 1760001815. bBritain, c. 1790-1840. cBritain, c. 1820001860. dUnited States, 1870001910. eUnited States, 1890001930. f United States, 1910001940. gJapan, 1950001990. hJapan, 197000. Japan, future. Notes:

machinery virtually across the board, so that sectoral differences narrowed. Within manufacturing, the developments in S&T alluded to above allowed diversification of industries and of their inputs, including the later adoption of electricity and the internal combustion engine as energy sources. These hugely expanded the portability and flexibility of machinery. Just as the United States was able to borrow many of these technologies from Europe in early industrialization, so Japan was able to draw on the technologies of more advanced countries at the beginning, and later on their S&T in cultivating new industries. According to K o d a m a [27], a leading characteristic of Japanese advance has been the effort to fuse different technological paradigms (e.g., developing mechatronics as the fusion of machinery and electronics). The centrality of machinery is therefore obvious, and the principal h e u r i s t i c underlying the adoption of machinery was to save time. The well-known machines developed for cotton processing during the British industrial revolution often achieved this through utilizing rotary m o t i o n - - a c c e l e r a t i n g circular motion was readily obtainable via machinery, in contrast to the natural horizontal or vertical movements of the h u m a n arm or leg. Secondary paradigms like chemicals also contributed to time saving (e.g., in cotton bleaching and dyeing). These latter changes furthermore entailed savings in land, like the extensive fields previously set aside for bleaching. In agricultural pursuits, the savings of land were more apparent (e.g., through adopting new crops and crop rotations). In subsequent industrializations, the focus changed somewhat. In the late 19th-century United States, with land abundant, the development of machinery brought together the

INNOVATION AND INDUSTRIALIZATION

11

heuristics of saving labor and saving time. 5 In Japan, the savings of time linked with savings of space, which was deficient in town as well as in country, and helped encourage the development of compact radios, television sets, and household appliances. The above are treated as heuristics, in the sense that they represent the logic of path-dependent technological development. The presumption is that in each particular milieu, the technological opportunities were greatest in these respective directions. However the dividing line between technological and wider economic impulses is often very hard to draw. The trajectories, which as stated above reflect the intersection between technological paths and broader socioeconomic conditions, therefore appear fairly closely related to the heuristics. It seems clear enough that the land-saving advances in British agriculture owed not just to fortuitous discoveries (e.g., bringing back the potato from the Americas) but also to the geographical constraints imposed by inhabiting a smallish set of islands off the coast of continental Europe. Japan faced even more serious space limitations, despite having a similar land mass to the UK, because so little of it was cultivable. It would be difficult, though not impossible, to sketch out counterfactual paths which observed the same heuristics but different economic influences [28, 29] .6 The characteristics specified in these cells thus overlap with the heuristics defined above them, but are seen as being to a greater extent the unintended consequences of adopting a particular technological path. The resources absorbed by adopting the respective technologies evidently relate closely to the endowments of each country] Later I shall imply that countries like Japan could, however, make good their deficiencies through enhancements of the factors which they did possess more abundantly.

Processes The nature of the paradigms and heuristics thus isolated carries the strong implication that the outcomes were to be observed primarily in new processes. The bulk of the research into processes, especially that by sociologists using Marxian concepts, has focused on labor processes; but Marx correctly pointed out that industrialization had been associated with a shift in emphasis from labor to capital process ([30], pages 690-695). Each industrial revolution was associated with a stage toward automation, but the stages reached were qualitatively different. The terminology of Bell ([31], see also [32]) is used here to demarcate the stages attained by automation. Bell defines around 18 specific stages, but the three main categories are those set out in Table 1. First, mechanization of transformation of the inputs was achieved by the adoption of machine methods in the first industrial revolution, including the invention of early cybernetic devices like the self-acting mule [33, 34]. Next, automation of transfer came famously with the moving assembly line of Ford in the middle of the second revolution; while finally, automation of control through the use of reprogrammable equipment has characterized recent advances of the third revolution. In terms of labor process, the work force found itself being bound to the machinery on which it worked, with a loss of some of its control over the pace and quality of work, since the capital equipment had become the arbiter [35-37]. However in the case of Britain, the work-gang still maintained a degree of control, exercised typically through the foreman--many critics of the British system indeed see this as the pre-eminent s Attempts to date to argue that the labor-savingwas purely determined by economiccircumstanceshave failed, on both theoretical and empirical grounds. The former issue is rather technical (see [26]); the latter can be indicated by the evident shortage of capital as well as of labor. Cf. the discussionby Ayres [28] of my previouswork on the British steam engines. 7The United States' reliance on land and resource inputs continues unabated (cf. [29]).

12

G.N. VON TUNZELMANN

reason why the UK failed to rise to the American challenge in the late 19th century [38]. The division of labor so carefully analyzed by Adam Smith ([39], page 17) nevertheless permitted workers to improve the processes on which they worked, so long as the incentives offered for doing so were adequate. Under "Taylorism" in later American manufacturing, control was intended to be removed from the individual worker, and usually entrusted to a managerial hierarchy. This is usually associated with moves toward "deskilling" the labor force, although F. W. Taylor himself opposed deskilling ([40], chapter 13, 14). In the Japanese system, control reverts to the work force who become responsible not only for ordering the pace and quality of work, but also for correcting their own mistakes. Workers have to learn multiple skills, and are normally organized to work in teams--sometimes teams of workers, and sometimes "teams" of machines. The original British system involved learning in each of the principal ways identified by recent scholars--learning by doing, learning by using, and formal learning. Indeed, Adam Smith in the course of his celebrated exposition of the advantages of the division of labor noted all three in the course of just one sentence describing its potential for innovation [39].s Recently, MacLeod [41] established that most innovations among those she was able to track down through patent specifications, etc., were led by users rather than producers of the machinery? As the machinery construction began to split off more distinctly from its use in downstream production, learning by doing (i.e., through the production-investment process) increased. By the latter part of the 19th century, the machine tool sector had developed as a recognizably substantial entity. Below I discuss the mutual response between the sector producing machine tools and the various sectors that used them (based on [42])--a pattern described as "vertical learning" in Table 1. Beyond the shopfloor, the United States was developing an extensive educational system, based on local government provision at the primary level and assisted by federal government at the tertiary (university) level; these underlay a broader and more coherent pattern of formal learning than Britain was at that stage able to devise. The Japanese system of production demanded further changes, and Aoki [43, 44] has described the "horizontal decision-making" that takes place inside Japanese factories, with strong communication flows up and down the production line. Learning is highly interactive both at the individual level (through job rotation of individuals around the company) and at the network level. The design of production systems has increasingly fallen to the engineers. Mechanical engineering in Britain during its industrial revolution was typified by what Calvert [40] terms "shop culture," that is, the predominance of the workshop (see also [45]). These workshops, invariably quite small in size, trained the next generations of engineers. Educational institutions like university colleges aimed at supplanting these apprenticeship systems, but had difficulty in reconciling practical with academic concerns [46]. The outcome in Britain was the emergence of engineering as a profession, and its rejection of commercial values. American mechanical engineering had almost identical roots, but by the late 19th century the "school culture" of college-based training was usurping the apprenticeship system. The reason seems to have lain in the positive espousal of commercial values by engineering courses in the colleges, and especially the espousal of the interests of nascent big business, where engineers were to be frequently found in lower and middle management. In the Japanese case, there was little problem with placing engineering on the university curriculum from the beginnings of Of course Smith's terminologywas somewhatdifferent;see ([39],page 21). Exceptions arose mainlyin the more sophisticatedcapital goods,like steam engines.

INNOVATION AND INDUSTRIALIZATION

13

modern universities around the 1880s [47], so that status was much less of a problem. The Japanese manufacturing system emphasized the role of engineers in production as a whole, and they played a central part in designing the high-throughput systems. After defeat in World War I! and the dispersal of the old manufacturing hierarchies, Japanese engineers came to play a central role also in top management.

Administration

Management thus took on quite different characteristics in each of the industrial spurts. The British industrial revolution has typically been seen as the heyday of the heroic entrepreneur--someone who amalgamated the talents of owner, manager, and innovator. Chandler [48] thus regards the British pattern as one of "personal capitalism," and believes that it persisted in the UK long after the industrial revolution, and indeed almost into modern times (others might argue that there were not enough heroes). In his view, there are serious weaknesses in such a personal system as compared with the "competitive managerial capitalism" which arose in the United States during its own industrial spurt. Such hierarchies originated in the railroad networks in the middle of the 19th century, because of the considerable geographical distances the American company was obliged to service, and came to manufacturing only some time later. The early departmental hierarchies, based on separating the functions of the firm and then subdividing each of these (R&D, production, sales, etc.) according to product or region, were eventually supplanted by multidivisional systems (M-form companies), in which the product and/or region was the main basis for grouping the firm. Tiers of middle and lower management came to take the main responsibility for day-to-day operations. This structure in turn lost way to Japanese-style management in recent times, under which managerial hierarchies were drastically pruned--the many tiers were regarded as "unproductive labor" in the sense of Adam Smith. To do so however required passing the duties hitherto performed by middle and lower management down to the employees on the shopfloor, and thus the system of horizontal decision making described above. Top management was then freed to undertake strategic functions. In Chandler's [49] eyes, the U.S. managerial system, and especially its top management, had forsaken the responsibilities required for the M-form structure to work; although another view, and one taken in this article, is that the intensified demands of technology and markets in this new phase of industrial development made the M-form inappropriate. Aligned with management were the systems of industrial organization.The British industrial revolution has traditionally been identified with the "triumph of the factory system" [50]; indeed many scholars regard the industrial revolution as one of a change in organization (to the factory) rather than a change in technology. Here the factory system is taken to have arrived somewhat later than the initial technological breakthroughs; but the view that the factory arose for reasons that had as much to do with control and labor process as with technology is also accepted here [36, 51], ~ for example because many of the early innovations in technology were not intended for use in factories. Even by the mid-19th century, the factories were relatively small, and contrasted sharply with the large plants that came to be associated with the subsequent American system of mass production. Chandler [52] regards "big business" as the fusion of mass production with mass distribution. Japanese organization placed less emphasis ~ Compare [36] and [51].

14

G.N. VON TUNZELMANN

on the centralization involved in mass production, and instead decentralized through networks such as the industrial keiretsu. These successive organizational patterns brought varying advantages in scale a n d scope. Scale is often equated with sheer size; but as Chandler ([52], page 281) has rightly pointed out, scale should be thought of in terms of throughput rather than just size[48, 53]. u The relatively small British factories of its industrial revolution can be considered "Smithian," in the sense that their advantages came from intensive specialization. Spinners of fine cotton yarns did not spin medium or coarser yarns, although those in each range were called upon to alter their precise yarn count quite frequently. In the vertical as opposed to horizontal dimension, spinning and weaving were often carried out in distinct firms and for the most part in distinct regions, so there was little move toward vertical integration--indeed the separation increased in later years of the industrial revolution. The relatively large American firms aimed instead at high throughput via mass production. This entailed a horizontal integration of products, directed partly at process efficiency and partly at controlling markets. Broad scope was impracticable with the combination of process technologies (dedicated assembly lines) and managerial systems involved, and the objective thus generally became one of producing a small product range in very high volumes--thus Ford's well-known dictum of 1909 that customers could have a Model T Ford in any color they liked, so long as it was black ([54], page 72). J2 Despite this pattern of product integration, which extended vertically along the lines of processing the materials, there was nevertheless some disintegration in the process dimension. The "American system of manufactures" was based on interchangeable parts and machine methods, and these machines, developed to produce very high levels of standardization, continued to come from separate machinery and machine tool sectors. In a seminal study, Rosenberg [42] showed how the machine tool sector was characterized by vertical disintegration and horizontal spillover, with the latter involving the application of the principles developed for the tools being supplied to one industry to those for other industries. In this way there arose a dynamic interaction between the suppliers of equipment and the users. The Japanese system utilized this aspect of the American system, and applied it quite generally. The scale-oriented American system was ill-suited to markets characterized by rapid changes in tastes, as Henry Ford was duly to find to his cost. The Japanese system sacrificed a modicum of scale in order to achieve substantial economies of scope. With just-in-time scheduling, total quality control, horizontal decision making, and specific process improvements like "singleminute exchange of dies" and reprogrammability, the objective was to be able to switch the product being produced in a matter of moments [55]. The typical Japanese product line is capable of being produced in hundreds, thousands, and even tens of thousands of variants, each tailored to a very specific set of final customer demands--indeed, recent critics have argued that the Japanese system has gone too far in the direction of diversifying product variants [56]. Products There was thus a close set of relationships between organizational forms and product ranges. The production processes drew on a variety of n e w resources in each industrial 11However Chandler often lapses into equating scale with size (e.g., [48], page 17). See also the theoretical analysis by [53]. ~2In reality, the Model T Ford did appear in a number of versions, though tiny in comparison with, say, the number of versions of modern Japanese automobile lines.

INNOVATION AND INDUSTRIALIZATION

15

revolution: some present at the start, others developed as part of the industrialization effort itself. New materials were required in the construction of the equipment and to some extent the plant that contained it--first iron in place of wood, then steel and other alloys, and eventually plastics and other composites. Chemicals were utilized for feedstock and catalysis during the processes, while also being contemporaneously developed as products. The materials to be worked up into finished products needed to be cheap and elastic in supply [57], as well as amenable to high-throughput production. They themselves were increasingly products of advanced technological development. Similar considerations arose in regard to fuels. The new resources were thus not just a matter of discoveries, but of concomitant technological development. In similar vein, the new products associated with each wave of industrialization were more defined by technology than defining it. Indeed, each industrial spurt began by producing products that were already in production elsewhere, with the industrializing countries adopting their more advanced processes to produce relatively established products. The British in the 18th century used machine methods to produce textile yarns and cloths similar in kind to those produced by hand in India or Continental Europe. The Americans in the late 19th century took over principles of steel and chemicals production from Britain and Germany, and through means such as "hard driving" developed them for mass production. The Japanese a century later imported products like the integrated circuit and rapidly moved toward industrial leadership in products such as DRAMs. The truly novel products that were of consequence came later in the three spurts, and often took the form of new applications of the "imported" product. Product innovation took on different heuristics in each case. The typical British pattern was of first producing for wealthy consumers, who were trying to ape the aristocracy above them in the social hierarchy. The ability to develop goods for factory production that closely imitated the virtues of hand-produced luxury items (e.g., Wedgwood pottery) could lower costs far enough that the lower middle class and even domestic servants became prospective customers [58]. ~3 The American producers by contrast assumed no social differentiation at all--they produced large quantities of highly standardized items, of at least reasonable quality, for the mass market. The Japanese, as already noted, reverted to differentiating products, by producing very large numbers of variants on each main item. Miniaturization permitted lower-income consumers with limited facilities and space to acquire "decencies" such as portable TVs. Such product heuristics reflected different paradigms of demand. The demand structure catered for by the British selection of products was one of fairly deep social divides, in which social imitation could act in dynamic fashion toward continual change. Urbanization at the same time involved some blurring of social boundaries and spread of information concerning novel possibilities. The contrasting American pattern was one of a relatively egalitarian social structure, not least in rural areas, and with information dispersed through migration, both external (the mass migration across the Atlantic) and internal (the "go-west" phenomenon). The standardized products went only so far toward meeting the demands resulting from generally rising incomes and wealth, and the rather restricted nature of American consumerism (competing in very minor product differences) was partially undermined by the emphasis on variety and quality at reasonable cost as purveyed by the Japanese. With greater distances to catch up, the Japanese 13Swann [58] describes this process of producing relatively high qualities at a falling cost premium as "'quality innovation," and applies the argument to the recent case of microprocessors.

16

G.N. VON TUNZELMANN

had to place greater stress at an early stage on producing for the relatively rich export markets. The marketing of these products was undertaken through systems of merchants and shops implying individual attention and service in the British case. Mass distribution, which as mentioned above went hand-in-hand with mass production to give rise to big business in the United States, allowed large time savings in dealing with comparatively undifferentiated markets. This included such characteristics as the rise of department stores, chain stores, and catalog selling, to achieve large volume. While this ran the risk of depersonalizing consumer relationships, the American system of distribution however maintained close relationships with customers by way of well-developed and often aggressive marketing and advertising methods, for example in sewing machines--this has been termed "learning by selling" [59]. The Japanese reverted to the use of middlemen and specialist dealers, like the merchant houses (foreign and then Japanese) used for establishing export markets, and more recently the outsourcing of customer relations. In this area, the Japanese do not ever seem to have matched the Americans, and later East Asian developers like the Taiwanese have experienced especially great difficulties in "learning to market" [60].

Meso/Macro Relationships All three cases of industrialization represented types of competitive capitalism, but there were practicing constraints on aggregate supply. Interfirm linkages of various kinds could curb opportunistic behavior. The British case demonstrated classic examples of shared responses and external economies generated through "industrial districts" of individually small producers [61], like Lancashire cotton or Sheffield steel. The much larger American producers were often constituted as oligopolies, which classically give rise to long periods of stability, interrupted by occasional price "wars." Collusion was formally restricted by federal antitrust legislation from 1890, although hardly prevented. In Japan, the vertically linked networks of keiretsu formed themselves into a small number of sprawling groups, with cross-holdings of shares and other relationships. Vigorous competition remained among the groups as a whole, but regular consultations with government allowed a shared vision of industrial prospects. Firms, markets, and regions were linked externally through the provision of transport and communications infrastructure. These are sufficiently well known not to require any description here. Two points however stand out. The main one is that the forms of infrastructure most often associated with the respective industrial revolutions evolved during their course. Schumpeter [62] considered that in the American case, railroads were built "ahead of demand," but subsequent detailed investigation has discredited this claim [63]. Similar arguments are being made today for building the information superhighway ahead of demand. The second point is that, in this case, the types of infrastructure that finally emerged from the industrial spurt did become the basis for another country in the next spurt. The shift from sectoral consumption patterns in the British industrial revolution, from agriculture toward manufactured items, paralleled the shift on the production side. For a time this became the major focus for political activity, crystallized in the debates over the Corn Laws but more broadly reflected in the shift from aristocratic political and social power in the 18th century to industrial power in the 19th century. By the latter 19th century, British agriculture was being sidelined by free trade and the importation of cheap grain from the New World. In the United States, the agricultural sector was brought into the ambit of manufacturing, not only in the use of primary inputs by the

INNOVATION AND INDUSTRIALIZATION

17

latter but also in the use of manufacturing technologies in the former, as noted above. Japan, as has been seen, lacked the primary resources, and from 1971 sought to reduce or replace these inputs, especially imported ones. The key issue was however becoming the link between manufacturing and services, with a blurring of boundaries between activities that had attributes of both, like computer or communication services [64]. The Japanese were able to stretch the features built into their manufactures to mimic service functions, but, partly as a result of arm's length marketing, have not had so much success in actual servicing. In terms of service and infrastructural functions, finance has sometimes assisted expansion and sometimes appeared to obstruct it. In the Keynesian interpretation of Figure 2, the role of finance is to supply sufficient savings to meet the demands for investment provoked by the "animal spirits" of entrepreneurs. The usual difficulty is that established financial structures are too remote from the day-to-day needs of industry to be proactive about industrial finance. In Britain there was a large geographical (and social) gap between finance centered around London and industry around the north of England. Local pools of capital had to be scraped together, and the general impression is one of enduring capital scarcities. In the United States, the multidivisional firm came to act as an internal capital market ([65], chapter 11], and in Japan and zaibatsu and later the industrial groups did likewise. As with transport and communications, the financial infrastructure that had emerged by the end of each spurt became a basis for another region in the next spurt (the stock market, etc.). Much of the bridging between micro and macro worlds thus still called for the involvement of government. This is one of the many aspects considered that deserves a book in its own right [66]. 14Government policy can be interpreted as working through each of the four functions of Figure 1. S&T policy naturally influences the level of science and technology. Industrial and commercial policies such as those promoting competition, or restricting it, impinge upon the organizational sphere, as also may educational and other polices. Financial and similar polices obviously affect finance. Macroeconomic policies such as Keynesian demand management may affect levels of demand, which will also be influenced by redistributive polices such as tax incidence. Here I limit the discussion to just the coordinative role of government. When the industrialization spurts have been at their strongest, governments seem to have taken the view that "if it ain't broke, don't fix it." Generally when countries have been performing very successfully, government intervention has declined (at least in the most successful areas) and pressure for decontrolling trade, etc., has intensified. The inference drawn in recent years from this experience that governments should withdraw in order to promote success could however be a case of "post hoc ergo proper hoc." Closer analysis suggests that governments intervened to remedy specific shortcomings even at the height of Victorian laissez-faire and its parallels in the other two countries. Above all, as Keynes was arguing, there was no guarantee of aggregate demand being sufficient to absorb aggregate supply. The role of the government remains even more fraught than that of finance to the present day. This can be related to Figure 2, not least in arguing that the enhancement of factors of production is a potentially important role for government. There is nothing necessarily new in this--Adam Smith [39] was one of the first of a long line of classical economists arguing for governments to subsidize mass education. But with the "growth of the knowledge-based society" [67] this would appear to be intensifying. There is considerable t~The issue has been summarized briefly in [661.

18

G.N. VON TUNZELMANN

convergence in the theoretical literature on economic growth in favor of the importance of human capital; although a divergence remains between the orthodox neoclassical school which conjectures that countries will converge on the leader country, and the "new growth" or "endogenous growth" school which sees the enhancements as endogenous and considers that countries may diverge in their growth performance [68, 69]. 15 It is now time to take stock of the detailed findings so far. The first reaction is that the three industrial revolutions varied among themselves in a rather startlingly large number of ways, which evidently makes any summary difficult. Only a small number of themes seem to repeat themselves. In terms of technological forecasting and social change, this may well be thought to be the most important single message that comes out of the comparative discussion. Adding in all the other cases of successful or partial industrialization would simply add to this degree of heterogeneity. The second reaction is that each of the patterns nevertheless has quite a high degree of internal consistency. The technologies, processes, organizational structures, and products associated with each contrasted with the others but possessed considerable coherence in isolation. American processes were (and remain) well suited to the relatively standardized American demands, but less well suited to rapidly changing demands. American hierarchical management befits mass production, integrated firms, and college-educated white-collar employees, but is rather distant from the needs of Japanesestyle "continuous improvement" (kaizen). Each system, when faced by competition from a new one, ignored it at first and then somewhat belatedly tried to copy its more obvious parts, but grafting these parts onto existing older systems inevitably meant some loss of coherence and therefore some of the dynamism they gave in their original locale. Efforts by British and American firms to patch on just-in-time and total quality control systems to their existing administrative systems are cases in point. The third reaction is that the evolution has not been a linear one in any obvious way. In certain respects, the most recent industrialization pattern (attributed here to Japan, although possibly to be taken further elsewhere) has more in common with the first (the British) than with the second (American) one. There has been no unilinear move toward greater scale and scope, or higher product quality, or more intense laborsaving, and so on. Even in cases where cumulativeness would be strongly suspected on a priori grounds, as in the advance of S&T, the pattern has included sharp breaks as well as forward momentum. The fourth type of reaction concerns the time lags involved, which can be assessed by the suggested datings in Table 1. The first point to make is that technological revolutions were not led by a rash of novel products--the new goods tended to come later and sometimes very late in the day. Confusion on this point still seems to reign, as debates continue over whether or not we are at present living through a technological revolution. 16 However the English schoolchild who allegedly claimed that "after 1760 there was a wave of gadgets" ([70]; [71], page 42) was not so wide of the mark in describing the first industrial revolution as often thought. The technological and indeed industrial revolutions were led by major changes in processes, rather than major changes in products--new ways of making existing things. These were likely to show up either in large falls in costs of such things, or in substantial improvements in their quality (or ~5For some surveys of the difference between neoclassical and endogenous growth theories, see, for example. [68, 69]. The endogenous growth school pays greater explicit attention to the work of Schumpeter, and in that sense has some similarities with the evolutionary school. ~6[70] is an example of such basic confusion.

INNOVATION AND INDUSTRIALIZATION

19

of course both, as for semiconductors in recent times). An evident implication is that the conventional product cycle, which posits a lead of product changes over process changes [72, 73], cannot account for this reverse pattern. This subject also warrants a research project in its own right, but my preliminary judgment is that the relationships go beyond the orthodox product cycle in two ways: first in taking a different level of analysis (analyzing the firm, industry, and economy rather than the product), second in hinting at more complex dynamic interrelationships. By drawing these four considerations together, we can attempt to derive some generalizations from a pattern whose overriding characteristic is that it is so overwhelmingly heterogeneous. To do so, I would argue that we need to take the supply (production) orientation of neoclassical analysis, couple it with the demand (consumption) orientation of Keynesian analysis, and try to consolidate them in the historically focused and disequilibrium oriented perspective of evolutionary analysis. In similar vein, we need to build both from micro to macro and conversely from macro to micro, again by drawing on all these modes of analysis. One of the very few constancies in Table I that runs across all three industrial revolutions has to date hardly figured at all in any of these analyses. In terms of their technologies, all exhibited heuristics of time-saving, in my interpretation of that concept. Time-saving change can be thought of as either producing the same output in shorter time or producing a larger output in a given time. The two are mathematically almost equivalent, but the practical implications differ. The former meant increased flexibility, in the sense of being able to adapt processes or products to sudden changes of demand. The latter meant meeting increases of demand, without running into barriers that might be imposed by inelastic supplies. Thus time-saving bridged between both fluctuations and rising levels of demand and the constraints of supply. Time-saving also characterized consumption: consumers sought savings in time spent in consumption, sometimes in obvious ways like fast food, sometimes in indirect ways like increased product reliability. The subsumption of service functions into manufacturing could allow time to be "shifted"; for example, recording by audio or video means permitted the consumption of a service such as an opera where and when the consumer desired [64]. As a technological heuristic, time-saving in principle could be adapted to a range of economic circumstances in determining the precise technological trajectory. The incidence of technological unemployment suggests that either it was itself too weak, or its adaptation was too limited, to square demand and supply. My working hypothesis is that the latter was the main problem. The case of automatic teller machines (ATMs) in British banks is an interesting case in point: these succeeded beyond all initial expectations, because bank customers saw them as hugely saving their time, whereas the banks that installed them saw them at first as saving their labor. In this case the banks' objectives of cutting labor costs were happily outflanked, to the satisfaction of both their employees and their customers [74], but the outcomes were not usually so positive. The technologies may have been capable of being adapted to the prevailing socioeconomic conditions, but authority relations were not necessarily structured to ensure that they would be. This example, one of a huge number that could have been selected, indicates the importance of the pairing between technology and organization. Marx's dialectic between the "forces of production" (i.e., technology and inputs) and the "social relations of production" first brought this issue to the fore. There has been much debate among historians, sometimes although not always dividing on Marxian vs. anti-Marxian lines, about whether organizational or technological changes were the basic driving force of

20

G.N. VON TUNZELMANN

industrial revolutions. The dating in Table 1 suggests that organizational changes did not come until the spurt was already under way. However it has also been argued that their rise was to an extent independent of the technological changes; one might also wish to argue that, when they did come, they were if anything more radical than the changes in techniques, and in addition were probably more difficult to replicate elsewhere (it was easier to borrow or copy technologies than production systems). Given that organization covers processes as well as administration, and the primary importance of processes has already been emphasized, the inference that technology led the broader change could not be pressed very strongly. It is the interaction between the functions which is of more lasting significance. Equally, at times technology and organization appeared to be pursuing divergent paths. Information technology in the last two decades has been following a trajectory toward decentralization (e.g., distributed processing, the internet) and has thus made centralized and hierarchical administrative systems look rather incongruous, despite the increased power it could provide them with. Similar issues arise in relation to the links between the innovation system and the economy at large--the two generally reinforce each other but sometimes appear to be drifting toward serious mismatch. Perez [75] claimed that mismatches in the "technoeconomic paradigm" (which can be regarded as a compound of the technological and economic paradigms) were likely to result in extreme political and social response, such as Nazism in the 1930s. Imbalances between supply and demand over the longer time require consideration in a broader perspective than the Keynesian one, which explicitly focuses on the short term, and in a more dynamic perspective than the neoclassical one. I thus conclude that a long-term evolutionary perspective (perhaps in its "punctuated equilibrium" form) is required for suitable appraisal of the linkages between technological forecasting and development on the one hand and social (and other) change on the other.

Conclusion The article has deliberately covered much territory, and a final summary can only be cursory. The main objective has been to show that each of the three major industrial revolutions generated a distinctive "national system of production"; each with a rather high degree of internal consistency, but very different from the others in almost every respect. The amount of spillover between one and the next was therefore considerably less than modern neoclassical theories of economic growth and "convergence" seem to suppose. The broad branches of industry developed in each case linked back to the preceding revolution and in due course forward to the next one, but the specific forms of technology and organization had much less interrelationship. Closer links were provided through forms of infrastructure, including transportation, communications, finance, and education; but even here the precise forms often had to be substantially altered to suit the new environment. The later part of the article shows that the kind of cohesion that developed at the micro level portrayed in Figure 1 was less sustained at the more macro level suggested by Figure 2. Mismatches were to be found between aggregate demand and aggregate supply, that persisted despite some institutional changes aimed at remedying them. To an extent, the onus of rematching them fell back on the micro level, especially in the kinds of technologies and organizations that were required. In practice, this meant shifting to a wholly new economic environment in which new sets of interlinkages could be forged.

INNOVATION AND INDUSTRIALIZATION

21

The author is grateful to seminars of ETIC (Economics of Technological and Institutional Change) at Maastricht, especially Patrick Cohendet, and at the University of Reading, for helpful comments to all of his students, and to two anonymous referees of this journal. The deficiencies remain my own. References t. von Tunzelmann, G. N.: Technological and Organizational Change in Industry during the Industrial Revolution, in The Industrial Revolution and British Society. P. K. O'Brien and R. Quinault, eds., Cambridge University Press, Cambridge, 1993, pp. 254-282. 2. Mokyr, J.: Technological Change, 1700-1830, in The Economic History of Britain since 1700. R. FIoud and D. McCloskey, eds., Cambridge University Press, Cambridge, 1994, pp. 1243. 3. Latsis, S. J., ed: Method and Appraisal in Economics. Cambridge University Press, Cambridge. 1976. 4. Dosi, G.: Technological Paradigms and Technological Trajectories: The Determinants and Directions of Technical Change and the Transformation of the Economy, Research Policy 11(3), 147-162 (1982). 5. Kaldor, N.: Capital Accumulation and Economic Growth, in The Theory of Capital. F. Lutz, ed., Macmillan, London, 1961. 6. yon Tunzelmann, G. N.: Technology and Industrial Progress: The Foundations of Economic Growth. Edward Elgar, Aldershot, 1995. 7. Mokyr, 3., ed.: The British Industrial Revolution: An Economic Perspective. Westview Press, Boulder, CO, 1993. 8. Schumpeter, J. A.: Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest, and the Business Cycle. (transl. R. Opie), Harvard University Press, Cambridge, MA, 1934; repr. Oxford University Press, London, 1961. 9. Lazonick, W.: Factor Costs and the Diffusion of Ring Spinning in Britain Prior to World War I. Quarterly Journal of Economics 96(1), 89-109 (1981). 10. Schumpeter, J. A.: Capitalism, Socialism and Democracy. McGraw-Hill, New York, 1943. 11. von Tunzelmann, G. N.: The Supply Side: Technology and History, in Industrial Dynamics. B. Carlsson, ed., Kluwer Academic Press, Boston, 55-84, 1989. 12. Lundvall, B.-,~., ed.: National Systems of Innovation: Towards a Theory of Innovation and Interactive Learning. Pinter, London, 1992. 13. Nelson, R. R., ed.: National Systems of Innovation: A Comparative Study. OUP, New York, 1993. 14. Kuhn, T, S.: The Structure of Scientific Revolutions. University of Chicago Press, Chicago & London, 1962. 15. Lakatos, I.: The Methodology of Scientific Research Programmes: Philosophical Papers, Vol. 1. J. Worrall and G. Currie, eds., Cambridge University Press, Cambridge, 1987. 16. Keynes, J. M.: The General Theory of Employment, Interest and Money. 1936; repr. Macmillan/Royal Economic Society, London, 1973. 17. Boyer, R.: Technical Change and the Theory of "Regulation," Chapter 4, in Technical Change and Economic Theory. G. Dosi, C, Freeman, R. Nelson, G. Silverberg, and L. Soete, eds., Pinter, London, 1988. 18. Amable, B., and Boyer, R.: Europe in the World Technological Competition, Structural Change and Economic Dynamics 6(2), 167-183 (1995), 19. von Tunzelmann, N.: Engineering and Innovation in the Industrial Revolutions. STEEP Discussion Paper no. 29, SPRU, University of Sussex, 1996. 20. Kuznets, S.: Modern Economic Growth: Findings and Reflections, American Economic Review 63(2), 247-258 (1973). 21. Rostow, W. W.: How It All Began: Origins of the Modern Economy. Methuen, London, 1975. 22. North, D. C.: Structure and Change in Economic History. Norton, New York and London, 1981. 23. Mowery, D. C., and Rosenberg, N.: Technology and the Pursuit of Economic Growth. Cambridge University Press, Cambridge, etc., 1989. 24. OECD, Technology/Economy Programme: Technology and the Economy: The Key Relationships. OECD/ HMSO, Paris/London, 1992. 25. Patel, P., and Pavitt, K.: The Continuing, Widespread (and Neglected) Importance of Improvements in Mechanical Technologies, Research Policy 23(5), 533-546 (1994). 26. David, P. A.: Technical Choice, Innovation and Economic Growth: Essays on American and British Experience in the Nineteenth Century. Cambridge University Press, London and New York, 1975. 27. Kodama, F.: Analyzing Japanese High Technologies. Pinter, London, 1991. 28. Ayres, R. U.: Technological Transformations and Long Waves, Part I. Technological Forecasting and Social Change 37(1), 1-37 (1990).

22

G.N. VON TUNZELMANN

29. Wright, G.: The Origins of American Industrial Success, 1879-1940, American Economic Review 80(3), 651-668 (1990). 30. Marx, K.: Grundrisse: Foundations of the Critique of Political Economy. 1857/8 (transl. M. Nicolaus), Penguin Books, Harmondsworth Middlesex, 1973. 31. Bell, R. M.: Changing Technology and Manpower Requirements in the Engineering Industry. Engineering Industry Training Board/Sussex University Press, Brighton, 1972. 32. Blackburn, P., Coombs, R., and Green, K.: Technology, Economic Growth and the Labour Process. Macmillan, London, 1985. 33. Catling, H.: The Spinning Mule. David and Charles, Newton Abbot, 1970. 34. Lazonick, W.: Industrial Relations and Technical Change: The Case of the Self-acting Mule, Cambridge Journal of Economics 3(3), 231-262 (1979), 35. Braverman, H.: Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century. Monthly Review Press, New York and London, 1974. 36. Marglin, S,: What Do Bosses Do?: The Origins and Functions of Hierarchy in Capitalist Production, Review of Radical Political Economics 6(1), 33-60 (1974). 37. Beniger, J. R.: The Control Revolution: Technological and Economic Origins of the Information Society. Harvard University Press, Cambridge, MA, 1986. 38. Elbaum, B., and Lazonick, W., eds.: The Decline of the British Economy. Clarendon Press, Oxford, 1986. 39. Smith, A.: An Inquiry into the Nature and Causes of the Wealth of Nations. W. Strahan and T. Cadell, London, 1776; bicentennial edn., R. H. Campbell, A. S. Skinner, and W. B. Todd, eds., Clarendon Press, Oxford, 1976. 40. Calvert, M. A.: The Mechanical Engineer in America, 1830-1910: Professional Cultures in Conflict. Johns Hopkins Press, Baltimore, MD, 1967. 41. MacLeod, C.: Strategies for Innovation: The Diffusion of New Technology in Nineteenth-Century British Industry, Economic History Review 45(2), 285-307 (1992). 42. Rosenberg, N.: Technological Change in the Machine Tool Industry, 1840-1910, Journal of Economic History 23(4), 414 4!.46(1963). 43. Aoki, M.: Horizontal vs. Vertical Information Structure of the Firm, American Economic Review 76(4), 971-983 (1986). 44. Aoki, M.: Information, Incentives, and Bargaining in the Japanese Economy. Cambridge University Press, Cambridge, 1988. 45. Berg, M.: The Age of Manufactures: Industry, Innovation and Work in Britain, 1700--1820. 2nd edn., 1994. 46. Buchanan, R. A.: The Engineers: A History of the Engineering Profession in Britain, 1750-1914. Jessica Kingsley, London, 1989. 47. Gospel, H. F. ed.: Industrial Training and Technological Innovation. Routledge, London and New York, 1991. 48. Chandler, A. D. Jr.: Scale and Scope: The Dynamics of Industrial Capitalism. Belknap Press, Cambridge, MA and London, 1990. 49. Chandler, A. D. Jr.: Corporate Strategy, Structure and Control Methods in the United States during the 20th Century, Industrial & Corporate Change 1(2), 263-284 (1992). 50. Fong, H. D.: Triumph of Factory System in England. Nankai University, Tientsin, 1930. 51. Landes, D. S.: What Do Bosses Really Do? Journal of Economic History 46(3), 585-624 (1986). 52. Chandler, A. D. Jr.: The Visible Hand: The Managerial Revolution in American Business. Belknap Press, Cambridge, MA and London, 1977. 53. Scazzieri, R.: A Theory of Production: Tasks, Processes, and Technical Practices. Clarendon Press, Oxford, 1993. 54. Ford, H.: My Life and Work. With S. Crowther. Heinemann, London, quoted from 1924 edn., 1922. 55. Best, M. H.: The New Competition: Institutions of Industrial Restructuring. Polity Press, Cambridge, 1990. 56. Leonard-Barton, D.: Wellsprings of Knowledge: Building and Sustaining the Sources of Innovation. Harvard Business School Press, Boston, MA, 1995. 57. Freeman, C., and Perez, C.: Structural Crises of Adjustment: Business Cycles and Investment Behavior, Chapter 3, in Technical Change and Economic Theory. G. Dosi, C. Freeman, R. Nelson, G. Silverberg, and L. Soete, eds., Pinter, London, 1988. 58. Swann, G. M. P.: Quality Innovation: An Economic Analysis of Rapid Improvements in Microelectronic Components. Pinter, London, 1986. 59. Thomson, R.: Learning by Selling and Invention: The Case of the Sewing Machine, Journal of Economic History 47(2), 433--445 (1987). 60. Hobday, M.: Export-Led Technology Development in the Four Tigers: The Case of Electronics, Development & Change 25(2), 333-361 (1994).

INNOVATION AND INDUSTRIALIZATION

23

61. Marshall, A.: Industry and Trade: A Study of Industrial Technique and Business Organization. Macmillan, London, 1919. 62. Schumpeter, J. A.: Business Cycles: A Theoretical, Historical and Statistical Analysis of the Capitalist Process. McGraw-Hill, New York and London, 1939. 63. Fishlow, A.: American Railroads and the Transformation of the Ante-Bellum Economy. Harvard University Press, Cambridge, MA, 1965. 64. Soete, L.: The Newly Emerging Information Technology Sector, in Technical Change and Full Employment. C. Freeman and L. Soete, eds., Blackwell, Oxford and New York, 1987, 189-220. 65. Williamson, O. E.: The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. Free Press/Collier Macmillan, New York/London, 1985. 66. von Tunzelmann, N.: Government Policy and the Long-run Dynamics of Competitiveness, Structural Change & Economic Dynamics 6(1), 1-20 (1995b). 67. Foray, D., and Lundvall, B. A.: Employment and Growth in the Knowledge-Based Economy. OECD, Paris, 1996. 68. Romer, P. M.: The Origins of Endogenous Growth, Journal of Economic Perspectives 8(1), 3-22 (1994). 69. Sala-i-Martin, X. X.: The Classical Approach to Convergence Analysis, Econom& Journal 106(437), 10191036 (1996). 70. Hutton, W.: The Familiar Shape of Things to Come, The Guardian, December 28, 1995. 71. Ashton, T. S., The Industrial Revolution, 1760-1830. Oxford University Press, Oxford, 1948; repr. 1971. 72. Abernathy, W. J., and Utterback, J. M.: Patterns of Industrial Innovation, Technology Review 80(1), 40~7 (1978). 73. Utterback, J. M.: Mastering the Dynamics of lnnovation: How Companies Can Seize Opportunities in the Face of Technological Change. Harvard Business School Press, Boston, MA, 1994. 74. Ingbam, H., and Thompson, S.: The Adoption of New Technology in Financial Services: The Case o]'A TMs. School of Management, UMIST, Manchester, 1992. 75. Perez, C.: Structural Change and the Assimilation of New Technologies in the Economic and Social System, Futures 15(5), 357-375 (1983). Received May 29, 1996; accepted February 21, 1997