Brian J. Loasby
Connecting Principles, New Combinations, and Routines Reflections Inspired by Schumpeter and Smith
International Joseph A. Schumpeter Society
Manchester, 27 June 2000
In Joseph Schumpeter’s History of Economic Analysis (1954), Adam Smith is not allotted a leading role. This is entirely appropriate, because the focus of economic analysis has been price theory, gradually refined into a formal and intendedly complete model of the allocation of economic resources; and at the time that Schumpeter was producing his treatise the culmination of this development was to be found in the work of Walras, even though it was subsequently recognised that Walras’s analytical system did not quite realise his vision. In this perspective Smith is indeed not a dominant figure. Despite the claims of some economists who have attempted to rescue his reputation as a theorist, it is seriously misleading to interpret Smith as a ‘precursor’, even unintentionally, of general equilibrium in aWalrasian sense of rationally co-ordinated activities, and even worse to interpret him as a precursor of post-Walrasian theory or of other models which rely on closed-system rationality to deduce equilibria. Nor indeed did Smith consider theory, formal or informal, to be sufficient to explain the working of an economic system. As we shall shortly explain, he had very different ideas about both the content and the method of economics.
However: we should notice that Schumpeter also considered economic analysis as defined above, to be only part of economics; and from his other writing we may infer that he did not believe it to be the most important part. In his Theory of Economic Development he makes a sharp distinction between an analytical focus on the co-ordination of a set of economic activities that is assumed to be well-defined - which is still the central tradition today - and the explanation of economic growth ‘from within’ (Schumpeter 1934, p. 63). Economic analysis provides models of the ‘circular flow of economic life as conditioned by given circumstances’ (Schumpeter 1934, p. 3), and although since Schumpeter’s time these circumstances have been extended to include the specification of a very wide range of locations, future dates and possible states of the world, these circumstances are no less ‘given’ (though it is not at all clear by whom) than in the models with which he was familiar. However, ‘[dl]evelopment in our sense is a distinct phenomenon, entirely foreign to what maybe observed in the circular flow or in the tendency towards equilibrium’ (Schumpeter 1934, p. 64). Co-ordination, Schumpeter asserted, is not the result of rational choice or efficient contract, but is achieved by allowing ‘things [to] have time to hammer logic into men’ (Schumpeter 1934, p. 80); thus although conduct eventually appears to be ‘prompt and rational’, it is simply the manifestation of evolved routines. Walrasian theory may therefore be used for the contingent prediction of behaviour, but it does not explain how that behaviour is generated.
That evolution, not rationality, provides the correct explanation for economic order was the theme of a famous argument by Alchian (1950), which was notably criticised by Penrose (1952) for excluding human purpose. However, Schumpeter did not bother with this issue, preferring to use the dependence of co-ordinated behaviour on time rather than rational choice to explain why economic development is necessarily cyclical. He invokes the limitations of human cognition to justify his claim that ‘stability is indispensable for the economic conduct of individuals’ (Schumpeter 1934, p. 40) - a theme to which we
shall return; but the major innovations which he identified as the prime movers of economic development undermine established routines and thus destroy the established basis of co-ordination. The creation of a new basis is no easy task, and no Walrasian mechanisms are at hand - indeed, as we shall see, no such mechanisms are possible within a Walrasian system; therefore a period of extensive innovation is necessarily followed by recession. In Schumpeter’s own vision of the economic system, the theory of business cycles and the theory of growth are inseparable; it is illegitimate to add cycles to a trend which is independently modeled, for without the cycle there can be no growth to produce a trend. This Schumpeterian proposition was developed at length in his study of business cycles (Schumpeter 1930). That the analysis of business cycles cannot properly be treated as a discrete topic has recently been reasserted, but proposals to assimilate this analysis to models of macroeconomic equilibria or of equilibrium growth are clean contrary to Schumpeter’s position.
Schumpeter’s admiration for Walras who is very firmly located on one side of his dichotomy between ‘economic analysis’ and the theory of economic development ‘from within’ therefore leaves ample room for Schumpeter himself on the other side, which many people especially at this conference, might think more important. However, in a history of the theory of economic growth, Smith is not so easily left in a subordinate place. Perhaps Schumpeter himself realised this in recognising the quality of Smith’s Essay On Philosophical Subjects (1980), and especially his History of Astronomy (Schumpeter 1954, p. 182). Smith’s account of “the principles which lead and direct philosophical enquiries” is essentially a theory of development from within, motivated by the desire for some means of understanding phenomena which cannot presently be explained; Smith exemplifies his theory by analysing as succession of theoretical visions of cosmological systems, not dissimilar from the visions that Schumpeter invoices both in his History of Economic Analysis and in his emphasis on major imovations. In relation to the distinction already made between a theory of co-ordination, which is based on a closed dataset, and a theory of growth, which is based on novelty, it is important to note that for both Smith and Schumpeter this restructuring is not deduced from the evidence; it is a product of the imagination. It therefore belongs in a theoretical system in which radical or structural uncertainty is not suppressed by fictional probability distributions, but is taken seriously.
In the course of his exposition, Smith also observes that growing specialisation within the scientific community leads to greater attention to the detail of theoretical systems, which increases the likelihood that inadequacies will be perceived within those systems; and these perceptions working on the motivation to avoid the psychological discomfort of the inexplicable rather than the search for gain (Smith l980, p.61), provide the incentive for the creation of new patterns which appear to be locally appropriate. The result is an increasing and self-sustaining diversity of increasingly specialised systems of thought. Scientific knowledge grows by division, and this growth is accelerated by the specialised studies that result from a finer division. In the Wealth of Nations Smith transformed this principle of the division of labour from an extension of his psychological theory of the growth of knowledge into the foundation principle of his own theory of economic development, with the emphasis now on productive knowledge (Smith 1976b). Smith
differs from Schumpeter in emphasizing the importance of incremental change, as Marshall was to do (drawing explicitly on Darwin as well as Smith); but we should not forget Smith’s (1976b, p. 22) reference to the innovative role of philosophers or men of speculation who “combine together the powers of the most distant and dissimilar objects” - in other words those who, in Schumpeters (1934, p. 66) phrase, “make new combinations.”
It would be a mistake to treat Smiths interest in economic development, and its relationship with the growth of knowledge, as a side-issue in the history, or the future, of economic thought. In a paper presented to the European History of Thought Conference in Graz this year (Loasby 2000), I have argued for the presence of two substantively different traditions in economics. What might be conveniently labelled the cartesian tradition emphasises the search for proven knowledge, where proof is a logical operation; it is easy to see how this tradition gives emphasis to choice as optimization, where the optimum is deduced by applying a criterion function to a possibility set, and to the representation of any stable situation or stable process as an equilibrium, the existence of which is deduced from the initial specification of the system. (Proofs of stability and uniqueness are also usually thought to be desirable, although multiple equilibria are sometimes welcome because they suggest options for policy). The progress of knowledge in economics then consists in improving the internal coherence of models, in increasing the range of circumstances that are included in the specification, and in finding methods of modelling phenomena or relationships that have hitherto proved recalcitrant. These directions may conflict - as an obvious example game-theoretic models ignore the requirements for proofs of a general equilibrium; but the formal properties of closed-system logic provide a common link across these varieties of formal theorising.
David Hume rejected this emphasis on formal proof as a route to empirical knowledge; indeed he rejected the possibility of proving the truth of any general empirical proposition, either by deduction or induction. This we might call Hume’s Impossibility Theorem. Certainty was not achievable; and axiomatic reasoning was insufficient as a method of scientific enquiry. Instead of a fruitless search for a route to infallible truth, Hume therefore proposed an inquiry into the processes by which people come to develop particular empirical propositions and to believe that these propositions are true. It was precisely such an enquiry that Smith undertook in his History of Astronomy, and the terms of his enquiry led him to see Newton’s theory - astonishingly for his time - not as demonstrable truth but as the product of Newtons imagination, and consequently liable eventually to be superseded. He drew particular attention to its appeal to the imagination of others, and not least to its aesthetic appeal, which endowed it with great rhetorical power, a theme to which we will return in a later section. This exploration of Humes’ agenda subsequently led him to emphasis the psychological influences on human decision making in his Theory of Moral Sentiments (Smith l976a) with due recognition of their fallibility both there and in the Wealth of Nations.
Alfred Marshall studied mathematics at Cambridge when Euclidean geometry was quite widely believed to have demonstrated the power of axiomatic reasoning to demonstrate empirical truth, only to experience the clearest refutation of this belief by the Creation of non-Euclidean geometry (Butler 1991). This refutation was later reinforced by the perception that Cournot’s axiomatic reasoning, which led to the conclusion that falling costs necessarily implied monopoly, was falsified by Marshall’s own statistical and personal observations of the contemporary British economy. Though still impelled towards the development of theory, he became cautious about its application, and extremely cautious about its refinement, because of the likelihood that errors in and exclusions from the specification of any model would be carried through to its conclusions thus creating knowledge which might be dangerously false. (How right he was is abundantly illustrated in the record of economists’ policy recommendations.) He also gave the development of knowledge a central role in economic systems, although this role was underdeveloped in his own work and even proposed treating organisation as a factor of production, because of its importance in the generation and use of knowledge. Allyn Young (1928) presented a forceful summation of the Smith-Marshall theme of the development of knowledge in economic systems, and Penrose (1950) virtually reinvented it. Evolutionary economics belongs in this tradition, and the accounts of economic development based on the growth of knowledge, that are offered by Smith and Schumpeter rely on similar elements similarly structured.
To understand these common conceptual bases; and the contrast with formal modeling, I shall draw on the analysis in Jason Potts’ (2000) forthcoming book. As one might expect of an analytical method that was borrowed from nineteenth-century physics, formal models typically assume an integral space in which every element is connected to every other element. Just as in Newton’s universe every object with mass influences every other such object, so in the full general equilibrium system all preferences are directly connected to all goods and all production possibilities, even when these preferences, goods and production possibilities are distributed across space, time, and conceptual states of the World, and in game-theoretic models all players know the rules of the game, are able to compute the full consequences of any action, and know that all other players have the same knowledge and capabilities. All actions are therefore simultaneously determined, even when they take place in different locations and at different times, and even though some of them never occur at all in many states of the world; the genius of the Arrow-Debreu model is to resolve the problems of distance, time, and uncertainty by abolishing them. It is no surprise that completeness implies efficiency - in relation to the problem specification, and that incompleteness, in the form, for example, of inadequately-specified contracts, is a threat to efficiency, which, however, can only be assessed at a point of time.
Now there are opportunity costs in economic theorizing, as in all choices. The assumption of integral space which makes general equilibrium analysts possible also excludes the prospect of analysing a path to equilibrium. As George Richardson (1960, p.57) warned us forty years ago, “[a] general profit opportunity, which is both known to everyone, and
equally capable of being exploited by everyone, is, in an important sense, a profit opportunity for no one in particular. A carefull reading of Richardson’s argument will show that he was sensitive to the deep foundations of this difficulty; as he was later to comment he “knew enough economics and enough physics to know the difference (Richardson 1998, p. xii). In a fully-connected world, the possible ramifications of any action cannot be calculated; because any or every element is liable to change in response, nothing can be relied on. At the level of the individual, rational choice is only possible in such a world if everyone else is already committed to their choices. Thus the standard practice of explaining rational choice on the assumption that everyone else is already in equilibrium is no innocuous simplification; it is the only assumption on which this analysis is valid. As Alan Coddington (1975, p. 154) pointed out, there can be at most one omniscient person; presumably that person should be appointed as central planner.
For a fully-connected system, the only alternative to omniscient central planning is prereconciled choice, which is isomorphic to it. That, of course, is why simultaneity is essential to proofs of equilibria, and this is what the fixed-point theorem provides. If markets are to be invoked as a possible means of reaching a fixed point, their operations must be shunted off into an unanalysable realm in which neither production nor the physical exchange of commodities is permitted, because any non-equilibrium action changes the initial conditions on which the calculation of that equilibrium depends. Simultaneity is also necessary, though not always sufficient, for standard game-theoretic analysis, typically in the form of Nash equilibria (which as a complete and coherent set of best responses to best actions is by definition fully-connected). Sequences must be precisely, if contingently, co-ordinated in advance, and based on knowledge which is already complete. It follows, as a trivial consequence, that this analytical system excludes any notion of development “from within”. Schumpeter was quite right.
A conceptual switch from integral to non-integral space is a switch from fully-connected to partially-connected systems. In both conceptions connections matter, but in the former connections are ubiquitous and therefore require no explicit attention. In the latter, however, they are both an essential component of analysis and problematic. The human mind is a connected system, but though the connections may be very extensive, they can be only a very small proportion of the connections that exist in the environment - especially when that environment includes many other minds. Formal organisations and social groups may also be highly connected, but their connections will also be meagre in relation to their environment. Many externalities will necessarily be ignored, and also many internalities, or detailed structures which may in some circumstances be extremely important. The gap between the set of mental or organisational connections and the phenomena with which individuals or organisations are trying to cope is filled with uncertainty and the possibility of error - but also the possibility of discovery or the creation of novelty.
All interesting problems lie in this gap. A wide range of heterodox economic inquiries postulate, though not always explicitly, incomplete connections which provide their various foci of interest. For example, Kirzner’s (1973) entrepreneur is alert to profit opportunities that are created by the failure of others to make connections between
markets, Keynsian unemployment is made possible by the absence of any credible connection between present investment and future demand, and Simon explains how problems are decomposed to keep connections manageable and decision premises are developed in order to simplify and co-ordinate decision making. Even property-right explanations of firms rely on a very specific exceptions in the form of information asymmetry, to the completeness of connections which is then remedied by an adjustment which is presumed to leave untouched the standard analysis of production and exchange.
In the broadest sense all these theorists investigate the relationship between structure, conduct, and performance (which indeed is a research strategy followed in different ways in many disciplines), but with a much deeper concern for structure than was common in old-style industrial organisation theory, in which the notion of market structure was rarely extended to include any reference to the institutions which guided the operations of particular market, and did not, as Coase (1988, p.57) commented, provide any explanation of the ways in which economic activities were divided up between firms. What they also have in common is that outcomes all turn on the kind, quality and distribution of knowledge, which is itself a structure of connections; indeed, as Smith saw, knowledge (strictly one should say knowledge claims) are created by the invention of connecting principles, and we may now a days think of these connections as having a physical representation within the brain. The growth of knowledge, and the possibility of innovation, depend on the incompleteness of present connections. To put the same point in the language of George Shackle, it is the pervasiveness of uncertainty that provides scope for imagination and makes possible the joy of discovery.
The basic facts of human cognition are that our brains have the capacity to establish an extremely large number of possible networks of connections, but can actually establish only a small fraction of this potential. As a consequence, the total number of connections and therefore the total amount of knowledge and skills, within a community can be greatly increased if the members of that community act in ways which lead them to make different connections. That is why Smith was right to make the division of labour an accelerator of scientific knowledge and the foundation principle of economic growth, and why Marshall was right to emphasise the role of organization, of various kinds, in the development and use of knowledge. Differentiation between individual entrepreneurs or between firms (sometimes labelled as Schumpeter I and Schumpeter II) then provide the basis for distinctive innovations and thus the foundation of the positive case for competition - and, as Schumpeter (1943) declared, against perfect competition. An important corollary, which is not considered in this paper, is that they also provide the basis for distinctive pathologies, and thus the foundation of the defensive case for competition.
Why, then, in Schumpeter’s account, are routines essential to entrepreneurial processes? The routines that matter in this account are those of the non-entrepreneurs, because these sustain the circular flow of economic activity and thus provide the assurance of predictable prices and quantities which allow the entrepreneur to calculate that the
resources needed for innovation will be available at a price that will ensure a profit, given the prices and characteristics of the products which are to be displaced (Schumpeter 1934, p. 141). If the innovation is in the institutions of marketing or organisational structure and practice, then the established routines similarly permit a reliable evaluation of the advantages of the new system. Why it is important that prices should be sufficiently static is that, because of cognitive limitations (or missing connections) the entrepreneur cannot hope to understand the workings of the economic system that generates these prices and relies on their stability to give him confidence in his calculations. His own cognitive powers will be fully employed in creating the productive, marketing and administrative systems (Chandler 1990) that are needed to realise his vision.
However, the effect of a major innovation is to disrupt these routine; by introducing novel connections to existing businesses, through competition from directions or of a kind that had never been contemplated; understanding what has happened, let alone finding a new pattern of new routines appropriate to the changed circumstances, is a formidable challenge to most people (Schumpeter 1934, pp.81-6). But it is not only those who depend on routines who are affected. The flow of entrepreneurial ideas is continuous (though the volume may vary); but the introduction of these ideas depends on the calculation of their advantage, which requires a stable environment. Not only do entrepreneurs not bear financial risks (Schumpeter 1934, p. 137) which are left to the capitalists (though an entrepreneur may choose to be a capitalist as well, or a capitalist might conceive an entrepreneurial vision, and invest his own funds in it); they do not take risks in launching their projects. The practice of entrepreneurship is therefore suspended until recession has run its course, and a new set of routines has been established (Schumpeter 1934, pp. 235-6). Thus Schumpeterian innovation must be seen as a process which replaces a very large set of interconnected routines with another set. (How we might decide whether the new set is better is a question I will make no attempt to answer in this paper.) We may here see a justification for taking routines as the unit of analysts; but we should remember that in Schumpeter’s theory the essential importance of this stabilisation of routines is that it makes possible the next major innovation.
Now there is a valid underlying logic to this insistence on the prevalence of routines in the decision environment, which derives from the confrontation between complexity and human cognitive limitations. Reasoned assessment - to say nothing of rational choice - is impossible unless one can take a great deal for granted; and the larger the number of changes that are incorporated in the project the smaller the number of changes that can be seriously considered outside it. Businessmen’s appeals for a stable environment maybe overdone, and sometimes exceed what any government or any social system can provide, short of the suppression of all change; but without substantial assumptions of environmental stability, purposeful action is impossible.
Nevertheless, the contrast between entrepreneurial creativity and external routines, though rhetorically effective, and justifiable in those term; is overdone; for there are typically routines within the entrepreneurial project also. This is concealed in Schumpeter’s original presentation because of his desire to emphasise - and justifiably
emphasise - that major innovations cannot possibly be produced by any identifiable procedure from what already exists (this is Knight’s (1921) definition of entrepreneurship, which relies on incompletable connections) and, in particular, that the economic analysis of co-ordination, in which all connections are already in place, has no direct relevance to the explanation of innovation, which depends on the creation of new connections. Schumpeter’s identification of the typical innovator as an outsider epitomises this thesis with striking rhetorical power. However, if we ask what is connected, the answer very often is a set of existing routines. This can be inferred both from Schumpeter’s list of types of innovations, and from his discussion of what does not constitute innovation, notably invention.
As many readers will be aware, the theme of innovative discontinuity was also the core of Thomas Kuhn’s (1962; 1970) theory of scientific revolution, and here too discontinuity was placed in a relationship of complementary antagonism to the routines of normal science. There are important differences, notably in Kuhn’s identification of an accumulation of failures of routine, or ‘normal science’ as the incitement to search; but the process of paradigm creation is no less mysterious in Kuhn’s theory than in Schumpeter’s. Now Kuhn’s explanation of the development of science through a succession of paradigms, linked by incremental changes which eventually produce an unacceptable accumulation of anomalies, at first sight very similar to Smiths; but Smith, unlike Kuhn, sets out to explain the transition, which he does by examining the ways in which the successful creators of new paradigms frame their search and the elements which are carried over from one system to another, typically being modified, and their significance sometimes being transformed, in the process. Instead of ‘paradigm’ Smith refers to ‘connecting principles’ which he insists are principles of cognition rather than ‘the real chains which Nature makes use of to bind together her several operations (Smith 1980, p. 105), and here is the clue, for principles are less constricting than routines; they supply categories for classification, premises for argument rules of procedure and criteria for choice, while leaving scope for imaginative conjectures. Neither the content nor the success of these procedures can be predicted - knowledge cannot be attained before its time - but they may often be readily understood in retrospect.
The difficulty of agreeing on a definition of routines is clearly displayed in the discussion reported by Cohen et al (1996) not least in describing their topic as “[r]outines and other recurring action patterns of organizations”. Among the questions discussed is whether the term ‘routine’ should be applied to a regularly recurring sequence of actions typically in response to a regularly recurring cue, or to single actions whether or not such actions are regularly combined with a particular set of other actions in a standard way. My own view is that both conceptions are needed for different purposes. The second is particularly useful in thinking about the innovation process, because the rearrangement of elementary routines maybe guided by rules. I have noted a tendency for authors to slide between ‘rules’ and ‘routines’ applying both, for example, to institutions; but, although Marshall’s principle of continuity applies here as to most other attempts to define
mutually exclusive categories; it is nevertheless useful to think of routines as fully-specified progammes (which may allow for contingent actions as long as the contingencies and the actions are precisely defined) and rules as devices for limiting the range of actions, either by exclusion (‘We are not in that business’) or by endorsement (‘expectations should always be assumed to be rational’). In any situation requiring a decision, they indicate the sort of things that should not be considered or the sort of things that should. Thus they simplify the ex-ante selection of a problem-definition and the ex-post selection from the options that emerge from this problem definition, or in other words they supply the procedural framework for procedural rationality. In any complex decision process, many of the detailed steps may be committed to routines, tacit or explicitly formal; the selection of routines is guided by rules.
Smith’s (1976a) Theory of Moral Sentiments sets out to explain how a particular category of rules influences human behaviour, not only when interacting with others but in private; and the prevalent rule among the more orthodox new institutional economists that institutions should be explained solely as devices for co-ordinating interactions seems to me an unnecessary obstacle to understanding. As Keynes (1937) pointed out, when we don’t know what to do we often look to others for some sort of rule which might work for us and in general borrowing rules from other people is a highly effective (though sometimes disastrous) means of economising on the scarce resource of cognition by exploiting the division of labour. In a recent article, Ekkehard Schlicht (2000) has reminded us of the importance of aesthetic criteria in choosing both our actions and our theories, and Smith was certainly aware of this throughout his work. To cite only his History of Astronomy, we may note Copernicus’s wish to replace the disorder into which the Ptolemaic system had degenerated with “a new system that these, the noblest works of nature might no longer appear devoid of that harmony and proportion which discover themselves in her meanest productions” (Smith 1980, p. 71), Kepler’s choice of the ellipse because it was the most perfect of shapes after the circle, which clearly would not fit the evidence (Smith 1980, pp. 86-7) and the aesthetic appeal of Newton’s unification of celestial and terrestrial phenomena. In his Lectures On Rhetoric, Smith recommended the ‘Newtonian method’ of deriving all explanations from a central principle as the best way to give an account of a system, because it “is vastly more ingenious and engaging” (Smith 1983, p. 146).
To make sense of innovation processes, it will often be helpful to switch the balance of attention from the contrast between routines and origination to the intermediate range of rule; premise; and criteria, while simultaneously taking a very generous view of the factors which may appear in these rules, premise; and criteria. One consequence of doing so is to reduce the apparent contrast between the inexplicable visions of the outsider in Schumpeter’s earlier work and his later account of the large organisation which apparently delivers innovation to order by following what, at a casual glance, look very like routines (Schumpeter 1943); for though research departments cannot meet innovative targets simply by following routines, nevertheless research sequences and research assessments may make extensive use of routines, guided by rules, premises and criteria which develop as members of each research organisation find their own ways of construing their experiences. Poppler’s logic of scientific discovery is that new ideas,
though conjectural, have implications, and that these implications may be testable; indeed, that ideas should be testable is a fundamental principle of scientific research. Much of the work of a research department dedicated to commercial success may be understood as following a set of procedures, often incorporating mini-routines, that are inspired by this logic. But the task of delivering ‘what is required’ is not as straightforward as Schumpeter (1943, p. 132) seems to suggest. These procedures may generate novel problems which may inspire fresh ideas and the end-product of innovative processes does not often correspond at all precisely to the original intentions. Indeed, although research departments may make great efforts to anticipate the conditions of use, they are unlikely to be completely successful, because of the different ways of thinking which result from the division of labour, and so it is likely to be a mistake, for both practitioners and analysts, to exclude diffusion from the process of innovation.
In a recent paper, Paul Nightingale (2000) investigates the contemporary process of pharmacological research. Although many research targets are very obvious in terms of the desired effects, it is not possible to derive from these effects a specification of the product that will achieve that target; thus the standard rational choice model, which relies on the logic of consequences, cannot be used. It is necessary instead to rely on what James March has called the logic of appropriateness, the basic principle of which was stated by Frank Knight (1921, p. 206). “It is clear that, in order to live intelligently in our world... we must use the principle that things similar in some respects will behave similarly in certain other respects even when they are very different in still other respects”. This principle applies even though we do not know which of the similarities are decisive for behaviour or why the differences don’t matter. There is an obvious corollary: we also assume that things different in some respects will behave differently in certain other respects even though they are similar in still other respects, without knowing which differences are decisive and why the similarities are irrelevant. Of course, these rules sometimes break down, and certain phenomena, or certain actions are transferred from one category to the other as a result of experience. The choice of candidate molecules for testing is based on definitions of similarity that have been imposed on past evidence, and these definitions may vary somewhat between companies, or between their leading scientists. The rules are not very prescriptive – except, crucially, in relation to the vast numbers of molecules which might be investigated, and a long sequence of testing is necessary in order to reject all those that the rule allowed into this procedure. Thus the rule is not a routine; but the testing has become highly routinised, and the sequence of tests is guided by rules which are derived from scientific understanding.
Though Schumpeter suggested that innovation could be managed, he maintained his distinction between innovation and the ongoing management of a business. Smith, though recognising the innovative role of ‘philosophers’ included this among the consequences of the division of labour, together with the improvements produced by workmen and specialist machine makers, and Marshall (1920) identified the principal cause of progress as the ‘tendency to variation’ that resulted from the experiments which were based on the particular experiences and temperaments of ordinary business people. That this progress
included changes in organisation its made explicit in his definition of increasing return (Marshall l920, p.318), and the contrast between increasing return as a process of continuous change and the standard concept of equilibrium was emphasised by Young (1928). Penrose (1959, p. 10) also sharply distinguished her theory of growth from the equilibrium models of price theory, and linked it with Schumpeter’s vision, though observing that “[t]he Schumpeterian “entrepreneur” though more colourful and identifiable, is too dramatic a person for our purposes” (Penrose 959, p. 36n).
Penrosian firms develop capabilities, including both productive and managerial routine; and rules for applying them, in the course of their business; but, unlike the inputs of standard production functions these capabilities are not automatically connected to a prescribed set of productive services. Thus new connecting principles may be created by entrepreneurial conjecture to generate new productive opportunities; like larger-scale Schumpeterian visions they are images in the entrepreneurs mind (Penrose 1959, p. 5). These productive opportunities rest on developed internal routines (and, no doubt, on external routines, though Penrose does not investigate the firm’s connection with its environment); their exploitation necessarily requires much non-routine activity, which crowds out other possibilities (Penrose takes the boundedness of cognition for granted), and so, as in Schumpeter’s theory, but at the level of the firm instead of the economy, the resumption of innovation requires the establishment of a new set of routines. Stability is a pre-condition of innovation.
We began with Schumpeter’s dichotomy between the ones of co-ordination and theories of innovation, and ended with Penrose’s endorsement of this dichotomy. But neither Smith nor Marshall thought of economic theory in this way; for both of them; what had to be co-ordinated was a process of continuous change. Theoretical development since the time of Smith left Marshall much more conscious of the theoretical difficulties in this concept; and Marshal own attempts at compromise were not so much rejected as incomprehensible according to the rules of formal theorising. But whereas in Schumpeter’s scheme periods of innovation and of co-ordination maybe treated as temporally distinct, in Penrose’s theory it is precisely the activity of resource allocation that generates change. If resource allocation, the focus of formal theory, is part of the process of economic development, is there any place for concepts of equilibrium and optimisation? Is there any sense in trying to appraise efficiency at a point of time? We know what Schumpeter (l943, p. 83) thought.
A system - any system, economic or other - that at every given point of time fully utilizes its possibilities to the best advantage may yet in the long run be inferior to a system that does so at no given point of time, because the latter’s failure to do so maybe a condition for the level or speed of long-run performance.
Alchian, Armen A. (1950) Uncertainty, evolution and economic theory, Journal of Political Economy, 58, 211-21.
Butler, Robert W. (1991) The historical context of the early Marshallian work, Quaderni di Storia dell Economia Politica. 9, 2-3, pp. 260-88.
Chandler, Alfred D. (1990) Scale and Scope, Cambridge, MA: Belkap Press.
Coase, Ronald H. ( 1988), Industrial organization: a proposal for research, in V. R. Fuchs (ed.) Policy Issues and Research Opportunities in Industrial Organization, New York: National Bureau of Economic Research, pp. 50-73; reprinted in Coase Ronald H., The Firm, the Market, and the Law, Chicago: University of Chicago Press, pp. 57-74. (Page reference is to this reprint.)
Coddington, Alan (1975) Speaking semaphore and beyond, British Journal for the Philosophy of Science, 26,151-63.
Cohen, Michael D, Burkart Roger, Dosi, Giovanni, Egidi, Massimo, Marengo, Luigi, Warglien, Massimo, and Winter, Sidney, (1996) Routines and other recurring action patterns of organisations: current research issues, Industrial and Corporate Change, 5, 653-08.
Keynes, J. Maynard (1937) The general theory of employment, Quarterly Journal of Economics. 51, pp. 200-23.
Kirsner, Israel M. (1973) Competition and Entrepreneurship, Chicago: University of Chicago Press.
Knight, Frank H. (1921) Risk, Uncertainty and Profit, Boston: Houghton Mifflin. Reprinted Chicago: University of Chicago Press 1971.
Kuhn, Thomas S. (1962, 1970) The Structure of Scientific Revolutions, lst and 2nd editions, Chicago: University of Chicago Press.
Loasby, Brian J. (2000) Content and method: an epistemic perspective on some historical episodes, presented at the Fourth Annual conference of the European Society for the History of Economic Thought, Graz 24-27 February.
Marshall, Alfred (1920) Principles of Economics, 8th edn. London: Macmillan.
Nightingale, Paul (2000) Knowledge, terminology and organisation: understanding the dynamics of pharmaceutical innovation processes, presented at the DRUID Summer Conference On The Learning Economy, Rebild, Denmark,15-17 June.
Penrose, Edith T. (1952) Biological analogies in the theory of the firm, American Economic Review, 42 pp. 804-19.
Penrose, Edith T. (1950) The Theory of the Growth of the Firm, Oxford: Basil Blackwell. 3rd edition Oxford: Oxford University Ppess (1995).
Potts, Jason (2000) The New Evolutionary Microeconomics: Complexity, Competence and Adaptive Behaviour, Cheltenham: Edward Elgar (forthcoming).
Richardson, George B. (1960) Information and Investment. Oxford: Oxford University Press. 2nd edition 1990.
Richardson, George B. (1998) The Economics of Imperfect Knowledge, Cheltenham and Northampton MA: Edward Elgar.
Schlicht, Ekkehard (2000) Aestheticism in the theory of custom, Journal des Economistes et des Etudes Humaines. 10 (1).
Schumpeter, Joseph A. (1934) The Theory of Economic Development, Cambridge MA: Harvard University Press.
Schumpeter, Joseph A. (1930) Business Cycles: a Theoretical. Historical and Statistical Analysis of the Capitalist Process. New York and London: McGraw -Hill.
Schumpeter, Joseph A. (1943) Capitalism, Socialism and Democracy, London: Allen and Unwin.
Schumpeter, Joseph A. (1954) History of Economic Analysis. London: Allen and Unwin.
Smith, Adam (1976a) The Theory of Moral Sentiments. ed. D. D. Raphael and A. L. Macfie, Oxford: Oxford University Press.
Smith, Adam (1976b) An Inquiry into the Nature and Causes of the Wealth of Nations, ed. R. H. Campbell A. S. Skinner and W. B. Todd, Oxford: Oxford University Press.
Smith, Adam (1980) Essays on Philosophical Subjects. ed. W. P. D. Wightman, Oxford: Oxford University Press.
Smith, Adam (1983) Lectures on Rhetoric and Belles Lettres. ed. J. C. Bryce, Oxford: Oxford University Press.
Young, Allen (1928) Increasing returns and economic progress, Economic Journal, 38, pp. 527-42.