Brian J. Loasby *
The organisation of capabilities 
Journal of Economic Behavior & Organization
Vol. 35, 1998, 139-160
The division of labour encourages the development of differentiated knowledge, and therefore of distinctive capabilities, which are ‘knowledge how’ rather than ‘knowledge that’ - a distinction akin to that between technology and science. The consequent problems of co-ordination may be handled by access or control, which suggests a choice between markets or firms; but market relationships must be managed, and no person can know enough to control a complex firm. ‘Managing capabilities’ is itself a capability, and must be limited by differences between knowledge bases, which may be acute even when capabilities are closely complementary.
Before we analyse the organisation of capabilities, it will be helpful to consider how we wish to interpret both terms. We shall follow the traditional practice of most economists by including within the scope of ‘organisation’ both formal structures and the distribution of activities among independent agents; but we shall question the traditional assumption which underlies general equilibrium and transaction cost theories, that the function of economic organisation is to ensure allocative efficiency, suggesting an
* Corresponding author.
1. This paper originated as a presentation to the EMOT Workshop on ‘Technology and the Theory of the Firm: Social and Economic Perspectives, at the University of Reading, 14-16 May 1995. The present much-changed version has benefitted from comments at the Workshop, and also from suggestions by Joan Cantwell and three referees.
alternative emphasis on the development and use of knowledge. This will lead to the interpretation of ‘capabilities’ as a particular kind of knowledge, and to an examination of the consequences for economic analysis of that interpretation. By combining these conceptions of ‘organisation’ and ‘capabilities’ we shall define a (non-exclusive) theme for economic analysis and empirical research as the study of the frameworks which both enable and constrain the processes of endogenous development - a study of the institutions of economic evolution.
The way in which a problem is defined is likely to influence the kind of solution that is proffered. This principle is readily illustrated by reviewing some of the ways in which economists have defined the problem of co-ordination, which for many is the central theme of economics. The most ambitious theory of economic co-ordination focuses on a general equilibrium of activities which are already co-ordinated; but the success of theorists in proving the existence, under somewhat restrictive conditions, of such a general equilibrium, and in extending that proof to encompass multiple locations, time periods, and states of the world, has not been matched by success in demonstrating how that equilibrium might be brought about. As Frank Hahn has frequently reminded us, at least one half of the story is missing. That in the Arrow-Debreu model markets open once only, and close before either production or exchange - except exchange of contracts - begins, implies that the activities by which co-ordination is achieved must take place outside a general equilibrium; and this suggests that the continuing failure to provide a rigorous theory of general equilibration is inherent in the general equilibrium method. This argument, which was suggested by Hayek (1937), has been developed by Richardson (1960),. Since the continual need to adjust patterns of co-ordination in response to change is what justifies the centrality of co-ordination in economic theory (Hayek, 1945, p. 523), the absence of any satisfactory theory of co-ordination processes might reasonably be considered a serious defect
To remedy that defect we need to consider how co-ordination processes are organised. Hayek (1937, p. 49) identified “the Division of Knowledge [as] the really central problem of economics as a social science’’ and looked to markets as the most effective institutions within which the dispersed and incomplete knowledge of a multitude of individuals could be brought into alignment. In the same journal and the same year, Coase (1937) noted that each of these individuals might need to acquire specific knowledge in order to use any particular market effectively, and argued that formal organisations could sometimes do better than markets in coping with a sequence of disequilibria which evoked the need for similar kinds of knowledge. Much later, Williamson recognised that dispersed and incomplete knowledge generated important obstacles to the achievement of incentive-compatibility, and built transaction cost economics on that recognition.
Transaction cost economics currently provides the best developed means of analysing the institutional structures within which co-ordination processes may occur. Its advantage
over agency theory, which rests on the same problem-definition, is its explicit recognition of hierarchy: firms are not simply a special kind of market. Yet in its present form transaction cost economics has three significant limitations: two of them will be considered here, and the third in the following section. First, as with general equilibrium theory the analytical objective is to model patterns of activities in which the strructures which ensure efficient co-ordination are already in place, even though the sequence of precontractual search, contract definition, and postcontractual monitoring suggests a process orientation. Williamson’s (1975, p. 9) early reference to organisational arrangements within which “the future is permitted to unfold” has not led him to consider the effects of such arrangements - or of a series of market transactions - on the knowledge of those involved, including their knowledge of each other, and the consequent emergence of new ideas for economising on transaction costs, which could not have been foreseen by boundedly rational individuals; and the significance in his work of the “fundamental transformation” between large and small numbers as transaction-specific assets are created lies not in the transformation process but in its intelligent anticipation. “An embedding of a transaction in an ongoing process of exchange is required to make TCE coherent’’ (Nooteboom, 1992, p. 285).
The second limitation is that Williamson’s insistence on opportunism as a necessary condition for the emergence of formal organisation entails a theoretical focus on the incentive problems which may result from incomplete and dispersed knowledge, at the expense of the logically prior problems which arise from the imperfections of knowledge, even in a state of universal benevolence. The point is not simply that, as Coase (1991) correctly believed, these problems are sufficient to explain the creation of firms; it is that asymmetric information, which is identified as the source of incentive problems, is itself a solution to the fundamental limitations of each individual’s capacity for knowledge. It is only by encouraging people to specialise in the acquisition of distinctive knowledge that we can increase the total amount of knowledge that is available within society. Incentive problems are a very proper topic for analysis; but the incentive problems on which transaction cost theory currently relies are better regarded as part of the pathology of an effective system for generating knowledge which is the foundation of economic progress. Evolutionary economists should not ignore the fitness of pathogens; but we should not seek to cure the disease by killing the patient This paper will contain only incidental comments on incentives in order to focus on the evolution and co-ordination of knowledge.
The charter for all subsequent economics is provided by Adam Smith’s claim that the principal cause of increasing productivity is the division of labour, both within and between trades. This increasing productivity, which includes new products as well as better methods of production, derives less from the specialist use of natural aptitudes, which had long been recognised, than from the development of new aptitudes as a consequence of specialisation (Smith, 1976b, pp. 28-30); and the course and content of this development depends on the pattern of specialisation and the ways in which individuals interpret their particular experiences. Smith (1976b, p. 17) not only draws
attention to the stimulus that is given by the division of labour to the invention of machinery, but notes (pp. 20-21) that such machines may be invented by three distinct kinds of specialists: workmen who conceive of ways of mechanising their own tasks, those who have developed machine-making as a distinctive trade, and those “whose trade it is, not to do anything, but to observe everything; and who, upon that account, are often capable of combining together the powers of the most distant and dissimilar objects.’’ It is natural to expect from these three different kinds of specialists three different kinds of interpretation, and thus three different kinds of machinery, especially as technology has increased in power and complexity.
Menger (1871) argued that instead of Smith’s sequence in which the division of labour stimulates the development of knowledge we should rather think of knowledge as directing the division of labour; and indeed the effective use of new technology often requires a restructuring of tasks, and sometimes a reallocation of productive activities between firms, as has been demonstrated, for example, by Langlois and Robertson (1995). Smith and Menger are both right, for the causal sequence may be cumulative: ideas for a novel division of labour create the conditions for developing new knowledge, which may inspire a further division of labour. This process of increasing return, which both Marshall (1920) and Young (1928) identified as the basic principle of economic development, naturally encourages the application of a part of the increase in productivity to the creation of new capital, which, if effectively linked to technology and an appropriate division of labour, generates further increases in productivity.
This approach to industrial organisation and change takes us beyond transaction cost theory, and suggests some modification of that theory. Few of those economists who use transaction cost analysis in order to explain patterns of industrial organisation appreciate the implications of recognising that the need for transactions arises from specialisation: what Williamson calls “technological separability” is not a natural given but a human creation. Self-sufficiency entails no transaction costs, since it entails no transactions; apart from an initial exchange of endowments, there is no reason to incur transaction costs of any kind unless they facilitate gains in productivity. It follows that the summation of separately-computed production and transaction costs, which Williamson has always acknowledged to be necessary, but which is normally left as an exercise for the reader, is not enough; the choice between transaction modes should reflect not just the relative costs of transacting, but the relative gains in productivity and in productive knowledge which they may be expected to make possible. Such potential gains must always be conjectural; there can be no rational expectations - in the strict sense of contemporary economics - about discovery. But, because of Smith’s fundamental principle, what we can expect is that different transaction modes are likely to offer not only different prospects of discovery, but also prospects of different kinds of discovery.
In their enquiries about the most efficient governance system for a particular pattern of specialisation, embodying a typical production function, economists too often neglect the historical process by which increased transaction costs have been deliberately incurred to make new patterns of specialisation viable, and subsequently been justified by the increases in productive efficiency which they have made possible. It is precisely such a story of increased productivity, at some increase - often substantial - in transaction or governance costs, that Chandler (1977, 1990) has offered us, exemplifying Marshall’s
(1919, p. 181) observation that production and marketing are part of a single process. All transaction costs are knowledge costs, as is clear in Coase’s (1937) original article, and so production, exchange, and governance are all aspects of the growth of knowledge, the division of knowledge, and the use of knowledge. By taking the transaction as the unit of analysis the relationship between transaction modes and economic development is kept out of focus; thus we should not be surprised that Chandler (1992, p. 85) has identified this difference in the unit of analysis as a serious limitation of transaction cost theory for his own analytical purposes.
To introduce our examination of capabilities, we turn to the third limitation of transaction cost economics. In that theory, as in most other branches of economics, knowledge is treated as information. We shall note, as worthy of separate treatment, the convenient anomaly which exempts knowledge of productive possibilities from the uncertainties which plague transactions, and which thereby helps to legitimise the independent treatment of production and transaction costs. What is crucial to the present analysis is that knowledge of productive possibilities is not enough; effective performance depends not only on sensible decisions about what to do, but on the availability of the skills that are required to do it. This is implicit in Smith’s theory of economic progress. Before we can sensibly consider the ways in which both the development and use of skills can be organised, we need to examine what it is that needs to be co-ordinated.
In order to elucidate the concept of capabilities we shall make use of Ryle’s distinction between ‘knowing that’ and ‘knowing how.’ Ryle’s purpose was to counter the traditional emphasis on the intellectual qualities of the mind, and the accompanying conception of the mind as something quite different from the bodily mechanisms which it controls; he asserted that “In ordinary life… we are much more concerned with people’s competencies than with their cognitive repertoires, with the operations than the truths that they learn” (Ryle, 1949, p. 28). What people can do often matters more than what they can set down on paper. Moreover, the two are often very weakly connected. “We learn how by practice, schooled indeed by criticism and example, but often quite unaided by any lessons in the theory” (Ryle, 1949, p. 41), and even when we are consciously seeking to apply prescriptions which are derived from theory “the intelligence involved in putting the prescriptions into practice is not identical to that involved in intellectually grasping the prescriptions’’ (Ryle, 1949, p. 49). Yet it is this practical kind of intelligence which is crucial to the effective use of any skill which goes beyond mere repetitive action.
The development and application of skills is not well represented by economic models in which appropriate technologies are rationally chosen by logical operations performed on well-defined information sets. Information sets are rarely adequate to characterise ‘knowledge how,’ which emerges from trial and error within a system that is imperfectly
understood rather than by logical deduction from a well-defined probability distribution. The modern concept of neural networks, in which particular connections are reinforced, and others atrophy, as some perceptions are classified together, and become more strongly linked with those actions which are interpreted as achieving satisfactory responses, seems more appropriate. It is worth noting that two famous economists who took a particular interest in problems of knowledge partly anticipated this psychological analysis. Hayek (1952) examined the problematic relationship between the internal order which is developed within each human brain and the external physical order which this internal order imperfectly and sometimes misleadingly represents, and Marshall (1994, pp. 116-132) produced an early conceptual model of a multi-level brain.
In Marshall’s, Hayek’s and modern artificial intelligence models, we have systems by which phenomena are represented, and linkages constructed between perceptions, actions and outcomes; but in none of these models are there any processes available which could ensure that these representations are correct, and that the linkages in the human brain correspond to the linkages which actually operate in the environment. In Marshall’s model, the higher-level brain has the capacity to represent the imagined consequences of contemplated actions: but there is no suggestion that these expectations could be rational, in the modern sense. Marshall’s “business man endowed with genius” (1920, p. 406) is especially skillful in devising experiments, but he cannot know in advance which products of his imagination will be successful. As Smith (1980) knew, patterns are invented and imposed on phenomena; and if cognition is an evolutionary process, we should remember that evolutionary success requires no more than a performance which is satisfactory in relation to rival cognitive systems in the circumstances that have hitherto been experienced. As with scientific hypotheses, what matters is always the next trial, and the result of that can never be guaranteed.
Ryle (1949, p. 46) reminds us that a skill is “a disposition, but not a single-track disposition like a reflex or a habit. Its exercises are observances of rules or canons or the application of criteria’’; and as Nelson and Winter (1982, p. 84) observe, the effectiveness of a particular skill on a particular occasion depends upon the particular combination of variables encountered on that occasion. It is impossible to know in advance what that combination will be, and whether it will lie within the range of application of the skill. Furthermore, as Ryle (1949, p. 59) points out, it is possible, and indeed common, to have partial knowledge of how to do something, or knowledge of how to perform at a particular level; but whether that level of performance will be good enough on the next occasion depends on those - such as one’s customers - who make the judgement, and on the alternatives which are then available to them. The establishment of Japanese-owned car plants in the US and Britain has shown a number of component suppliers that their capabilities were not of a satisfactory standard in the new environment. Capabilities are always conjectural.
Capabilities are the least definable kinds of productive resources. They are in large measure a by-product of past activities, but what matters at any point of time is the range of future activities which they make possible. What gives this question its salience is the possibility of shaping capabilities, and especially of configuring clusters of capabilities, in an attempt to make some preparation for future events, which, though not predictable, may, as Lachmaim and Shackle insisted, be imagined. It is to Menger’s (1871) credit that
he recognised, if unemphatically, the variety of ways in which people can accumulate reserves against an uncertain future; it is to the discredit of the economics profession that so little attention has been given, with some important exceptions, such as Hicks (1976), Leijonhufvud (1973), Littlechild (1986), and above all Keynes (1936), to the accumulation of reserves as an essential feature of economic activity - essential because it is not at all clear how a modern economy could function if its members were unable to draw on reserves of many different kinds. Capabilities, each of which gives the power to act effectively in a particular range of possible future circumstances, are among the most important of such reserves, and the development of such power is a major, though not always consciously intended, function of formal education, and was included in the Ansoff (1965) set of objectives for corporate strategy. It is hardly possible to discuss capabilities sensibly without recognising that each bundle of capabilities is oriented towards a range of possible futures, and that every such orientation is a fallible conjecture. Shell use multiple scenarios of possible futures in order to encourage their managers to consider what orientations may be appropriate.
The immediate reason for this professional negligence among economists is that the skills in which they are trained, and which are reinforced by the system of formal and informal incentives to which they are subject, are themselves oriented towards a range of possible future analyses which excludes such problem-definitions. Within this range models are fully specified: the information currently available to a particular agent may be incomplete, but information sets cannot be. It can therefore never be rational to operate inside the possibility frontier, as that can presently be specified, in order to be able to respond quickly to threats or opportunities which cannot be clearly identified - because what cannot be identified cannot be included in the economist’s problem definition; strategy as a set of decision premises for future choices, which to non-economists is an obviously sensible and useful definition, must be replaced by a game-theoretic strategy of commitment to conditional actions; and capabilities, which are reserves against contingencies that are not predictable and therefore not contractible, must be replaced by production functions.
Capabilities of the kind so far discussed are not sufficient for business success. That depends, in Penrose’s (1959, p. 31) phrase, on a firm’s productive opportunity, which combines capabilities with the perception of a profitable use to which they can be put; and as can frequently be observed, either perception or capability may be missing. Productive opportunities, too, are conjectures. However, although knowledge, whether true or false, of an attractive market should be regarded as ‘knowledge that’ rather than ‘knowledge how,’ nevertheless the ability to seek out relevant information or to draw inferences from newspaper reports, conversations, political or social changes, or even price movements, is certainly ‘know-how’. “To know something is to be able to understand and otherwise make sense of it” (Minlder, 1993, p. 520). ‘Knowledge that’ may be a public good, but the capability of making sense of it, in any one of the ways that are conceivable, is not. However, like other capabilities, this is generally susceptible to analysis in terms of experience and the patterns that have consciously or unconsciously been imposed upon that experience. The alertness which characterises each of Kirzner’s (1973) entrepreneurs is a distinctive capability, and it is because appropriately oriented capabilities are scarce in each local economic environment that alertness is a source of
profit; Casson’s (1982) entrepreneur is especially skilled at interpreting the conjunction of different kinds of knowledge, which, as we have seen, was the special trade of Smith’s ‘philosophers’.
In the most rigorous sciences, the reliability of the knowledge that is produced depends on the skillful performance of particular operations (Ziman, 1978); and the particular operations which are deemed to be appropriate within each science sets limits (which are not always acknowledged by the scientists themselves) to the kinds of knowledge that can be attained. If macroeconomic problems are always to be studied by constructing models of rational expectations equilibrium, the highest level of skill in such modelling will never generate the kind of knowledge that is embedded in Keynes’ General Theory: indeed, highly intelligent economists who practise such skills are unable to make analytical sense of the concept of involuntary unemployment, as Hahn and Solow (1995, p. 2) have observed. In science as in business, we can learn only by restricting our attention to particular skills or particular topics; and we cannot know - though we may guess - what we might have learned had we attended instead to other skills or other topics. (The importance of how we know is examined in Loasby, 1998). Some constraints on learning are always necessary; but their effects on society or an economic system can be mitigated, partly through the effects of specialisation itself, which fosters a diversity of constraint systems within which knowledge can be developed - thus the distinctive advantages of economists’ and sociologists’ particular ways of generating knowledge may offer mutual compensation for their corresponding deficiencies - and partly by the tolerance of variety within each specialised field. To bring these processes within the scope of economic analysis, however, requires a substantial redefinition of the constraints within which orthodox economists work; but, as often happens, the redefinition of constraints reveals a productive opportunity.
Ryle’s distinction may be used to illuminate the relationship between science and technology. The declarative content of science is ‘knowledge that’ (remembering that what is treated as knowledge is a set of hypotheses which have been not yet been refuted), and technology is ‘knowledge how’; so we should not be surprised that science is rarely an adequate guide to the development of the technology which it might seem to suggest, or even that the technological implications of a scientific theory may not be recognised until long after the theory has become textbook material. As Rosenberg (1994, p. 140-141) points out, the magnitude of development costs, even when the development is related to old science, provides an impressive indicator of this mismatch. A less public indicator is the number of scientific problems that are first identified during the development process: in a time of rapid technological change, technology may lead science as often as the other way round (Rosenberg, 1994, p. 141) - not least in suggesting the value of interdisciplinary research (Rosenberg, 1994, pp. 147-149).
We should not forget that each science has its own technology; good research scientists know how to develop testable implications from hypotheses, how to devise experiments which will effectively test them, and how to interpret the results in a way which will foster the continuing process of discovery - all within the evolving conventions of that
science. (The implications of such conventions for the knowledge that a science can produce are explored in Loasby, 1995.) Postgraduate training is probably of more value in helping people to learn how to do a particular kind of science than in transferring to them the substantive knowledge of science already done; and Kuhn (1962, p. x) introduces his concept of paradigm as an exemplar of knowledge how. But this knowledge how is directed towards a different purpose from the knowledge of how to achieve technological goals; indeed, it has effectively become what Smith called a distinct trade, and an example of the division of labour which has led to a remarkable development of specialised skills.
This division of labour and skills is often conveniently organised by a clear separation between university and industrial science, which minimises the operational conflict between very different concepts of what constitutes good practice. But, like any organisational solution, and indeed any choice worth considering, this has its opportunity costs. As a simple example, developed by Rosenberg (1994, pp. 144-146), scientists become extremely skilfull at achieving results in the conditions and on the scale of a laboratory, but rarely give attention to the skills that are required to achieve equivalent results in very different circumstances. The well-known relationship between the volume and surface area of a reaction vessel, for instance, means that exothermic reactions which pose no problem in a laboratory may present formidable challenges to the design and operation of a commercial plant; but there is no reason why most university research chemists should develop any special skill in handling such challenges. Similarly, because the accumulated expertise of economists in developing theories of rational choice has been directed towards modelling various kinds of equilibria, in which no further choices are required, we should not be surprised - though we may be embarrassed - to observe how limited is the help that rational choice theory can give in making decisions, even to rational choice theorists. The gap between the knowledge of how to do research into chemistry and the knowledge of how to develop new chemical technology has been filled by the emergence of a whole new discipline of chemical engineering; the gap between the knowledge of how to develop economic models of choice and the knowledge of how to make good managerial decisions has been filled by that amorphous collection of disciplines known as management studies, although (as we should expect) many of its practitioners are more successfull in developing skills in building theories for classroom exposition than skills in practical decision making.
If technology cannot be developed in an orderly fashion from scientific principles, it must be heavily reliant on trial and error, an incremental process where the crucial questions at each stage are how to interpret the latest results and what to do next. As with Marshall’s (1920) principle of substitution, decisions are made at the margins of knowledge rather than by marginal analysis. This process is likely to be costly in both time and resources, and it is possible to spend a great deal of both without apparently making much progress, except in discovering what not to do - though, as the Duke of Wellington observed of his experiences as a subordinate officer in an unsuccessful campaign, “that is always something.” The effective development of technology seems
to depend less on linear logic than on identifying or imposing patterns and connections; and experience with present products and processes often provides better guidance than scientific principles. It is also much harder to transfer between organisations: even if it can be codified, the code which makes sense to the codifiers may not make relevant sense to the recipients of the message. Where knowledge how is crucial, it is rarely an attractive proposition to try to develop a collection of novel skills rapidly, because it is likely to be extremely difficult to interpret and guide the learning process when many interrelated procedures are being changed simultaneously (Teece et al., 1994, p. 17); if few or none of the component skills are well established within the organisation, there are too many degrees of freedom, and codified science usually does little to reduce them. Schumpeter’s (1934) new combinations are not combinations of new knowledge how; knowing how to manage the combination of existing capabilities is novelty enough.
We would therefore expect to find that the division of labour between various trades has resulted in a number of distinctive technologies, each embedded in a group of firms, that these technologies change rather slowly, along trajectories which are localised to particular ways of organising knowledge, and that the productive opportunities open to any business are quite closely restricted by the technologies which it has acquired and the markets to which that business knows how to direct them. Rapidly-developing fields are thus likely to be dominated by those already expert in the relevant technologies. But because knowledge how is not reducible to logical sequences, but depends on a multitude of evolving connections, we would still expect to find some variety between firms in the way that they do things and rather more in the productive opportunities that they perceive. We would also, of course, expect firms to make mistakes, and the biggest mistakes would be expected when they move into fields to which their technology is not applicable - especially if they do not realise its inapplicability.
All this is what Patel and Pavitt (1995, 1997) did find in their investigation into technological competences in the world’s largest firms. From an analysis of the U.S. patenting record of more than 400 of the world’s most technologically active firms, they conclude that large firms typically develop skills in a cluster of technologies which is closely related to each firm’s product range, that the set of technologies included in each firm’s cluster changes only slowly, but that the specific emphasis within a cluster, and the range of products to which it is applied, may differ between firms: because each productive opportunity is a distinctive combination of technology and vision there is some variety in patterns of diversification. (For a detailed study of such variety, see Fransman, 1995.) The development of each firm’s capabilities is path-dependent, but there is, in general, more than one path available from any particular configuration.
In order to analyse the organisation of capabilities it is necessary to avoid the top-down perspective of the conventionally-defined allocation problem - what is to be produced, how, by whom, and for whom - and to consider co-ordination from the point of view of the individual, as Adam Smith did. Since the productivity-enhancing division of labour, our wants are provided through “the assistance and co-operation of many thousands”
(Smith, 1976b, p.23); but the obverse of this beneficial interdependence is that “man has almost constant occasion for the help of his brethren, and it is in vain for him to expect it from their benevolence only” (Smith, 1976b, p. 26). Due to the common misinterpretation of Smith’s analysis of society, it is worth noting that ‘benevolence’ is not irrelevant, though Smith places much more weight on people’s sense of propriety, or the desire to be worthy of the approval of others, which sometimes inspires help, and more often curbs opportunism (Smith, 1976a). But benevolence and propriety together are not enough; and the problem which Smith poses, of how an individual is to get things done by others, focuses on individual efforts at co-ordination from within. We need not only to know how to do certain things for ourselves, but also how to get other things done for us; and just as productive activities require direct capabilities, so transactions depend on indirect capabilities.
Indirect capabilities are of two kinds: we may be able to get things done for us either by gaining control of other capabilities or by obtaining access to them. We shall soon discover that, like many other apparently sharp distinctions, this highlights relatively distant points on a continuum; nevertheless, as Ménard (1995) has argued, the distinction is analytically invaluable, and this section is intended to complement his work in ‘disentangling concepts.’ The obvious application of the distinction within this paper is to the contrast between markets, which offer access, and firms, which allow hierarchical control; and the immediate conclusion is that control has substantial advantages, but is likely to be more costly than access. We can access more than we can control, and therefore should limit our attempts at control to those capabilities which are both crucial and manageable. What might these be?
Every new example of the division of labour separates what has hitherto been joined; it therefore implies a need to provide some linkage to replace that which was previously embodied in the practitioners of this now-divided skill. Such linkages may be unnecessary, and even sometimes undesirable, if the division of labour allows the separated skills to be adapted to unrelated uses: no one would now advocate a closer integration between surgeons and barbers. But Marshall was surely right to accept the general rule, which he adopted from Herbert Spencer, that evolution tends to favour both greater specialisation and closer integration. Divided capabilities typically need to be used in clusters, or in closely-ordered sequences, if the improvements in each new sub-skill which follow this division are to be guided in compatible directions and effectively used. Formal organisations may be effective instruments for doing this, and, as Marshall saw, could provide an exosomatic equivalent to the multi-level brain which he attempted to represent by his ‘machine.’ Brains and firms are thus self-organising systems in which the development of well-honed skills releases time and energy for more complex problems, and in particular those which do not recur often enough to allow the appropriate connections to be selected and then reinforced by repeated trial and error. Both systems facilitate the development of capabilities and their reconfiguration to match a changing perception of needs; and this, it may be argued, was Coase’s ( 1988, pp. 39-40) basic rationale for the firm.
Marshall’s multi-level model, and Babbage’s (1989, ) Economy of Manufactures, which Raifaelli (1995) argues was a major influence on his thought, naturally led him to recognise the potential advantages of large-scale organisations, in which the employment of able subordinates promised the greatest opportunity for the head of a firm “to keep his mind fresh and clear for thinking out the most difficult and vital problems of his business; for studying the broader movements of the markets, the yet undeveloped results of current events at home and abroad” (Marshall, 1920, p. 284). But Marshall also recognised that this opportunity might not be taken: the energy and enterprise on which the founder of the business drew to build it up is likely to decline as he grows older, and his successors may have less of either. In addition, a large business depends on initiative at many levels, for any capability may be improved, modified, and applied in new ways; and by freeing himself from detail, the head of the business necessarily deprives himself of much of the specific knowledge which is needed to perceive what might be done and to see that it is. Conventional principal-agent models offer limited help in co-ordinating the growth of knowledge, as Minkler (1993) has explained: incentives certainly matter, but the problem to which they should be addressed is not how to induce the agent to take that action which the agent already knows will best serve the objectives of the principal, but how to encourage the agent to recognise unspecified contingencies and to imagine new actions. Agents may prefer the security and ease of the familiar; and formal control systems tend to discourage initiative, and therefore may destroy rather than develop capabilities. It was because he saw the organisation of industry as the organisation of a discovery process (including, we should not forget, that kind of discovery process that we call the formation of character) that Marshall was less than enthusiastic about large firms.
In firms of any size, organisation shapes and constrains the growth of capabilities: in Penrose’s (1959, p. 149) definition, a firm is “a pool of resources the utilisation of which is organised in an administrative framework.” The framework specifies what people are expected to notice and what to ignore, the range of activities which they are expected to perform and the criteria by which their performance will be judged. That is why organisational design is important. But it is not only formal organisation which matters. Informal organisation may reinforce, counteract, or complicate the effects of formal organisation; what is certain is that it contributes substantially to the framework within which capabilities are developed and used. (That is strikingly true of the economics profession, but no less strikingly false of almost all economic analysis, in which personal connections are subject to indictment for restraint of trade.) Indeed, knowing how to appease the formal and informal expectations of other organisation members is an important skill for anyone who wishes to become an effective organisation member - as is sometimes the ability to circumvent, or even possibly subvert, these expectations.
Partly by deliberate prescription, but primarily through the evolution of custom and practice (which, if integrated by a clear, if tacit, framework of connecting principles, merits description as a corporate culture) members of an organisation develop the indirect capabilities - the knowledge of how to get certain things done - which are necessary for them to use their own capabilities effectively. In most organisations they also eventually resign themselves to the absence of some desirable capabilities: there are some things which they find impossible to get done. That is a pathology of any strong culture. (Sometimes they leave the organisation in order to get them done: this is a common origin
of new entrepreneurial businesses.) It is this network of indirect capabilities, much more than hierarchy and fiat, which allows a firm to co-ordinate a range of operations at lower cost than would be entailed by market transactions. The reduction in cost may be particularly great if intrafirm relationships foster the development of sympathetic understanding within and between departments. People come to know who they can rely on.
If one looks within an organisation, one finds that the detailed integration of activities depends on access rather than control, though the underlying pattern (which may be partly illusory) of control may be an important facilitator of access, especially for newcomers to an organisation. No single individual, and no board of directors, can know how to do what any sizable firm can do; nor can they even know how to get it done except in the sense of encouraging others, by deeds as well as words, to use their own capabilities in ways which only the individual knows are appropriate. The knowledge of how to do whatever it is that an organisation can do is distributed among the direct and indirect capabilities of its members, and, as Winter (1991, p. 185) has pointed out, is accessible only through that combined membership. It cannot be planned according to conventional notions of allocative efficiency. It is through multiple access to a distributed network that a firm is able to solve problems of complexity that cannot be fully understood by any single individual.
Hayek’s analysis of cognitive psychology led him to the conclusion
that “any apparatus of classification must possess a structure of a higher
degree of complexity than is possessed by the objects which it classifies”
(Hayek, 1952, p. 185); it is thus not surprising that the division of labour, by permitting the development of a complex
structure of capabilities that would be unattainable by anyone practising the undivided activities, requires processes of integration
which are beyond the conscious control of any person. The administrative structure of a firm may be
thought of as such an ‘apparatus of classification,’ which cannot therefore be
fully represented by a model in the brain of the CEO. (The CEO may nevertheless make a decisive
contribution to the success of a business, both because of the crucial decisions
which no one else can make and by influencing the development and use of the
company network.) That these
capabilities, as Ryle emphasises,
are the product of intelligent processes which differ substantially from those
normally associated with theoretical or factual knowledge, and which are
therefore often difficult to represent in a way which would make them amenable
to formal analysis, reinforces Hayek’s argument .
We should not overlook the contribution of decomposability to the effective management of elaborate systems, which has been rightly emphasised by Simon; but it may reasonably be urged that Simon has underrated the importance of knowing how to achieve results without the conscious use even of boundedly rational procedures. Knowledge-in-use “is embedded in the social practices of a particular community” (Boehm, 1994, p. 162); a firm is one of the most important communities in which such knowledge can be embedded. It is a social institution for enhancing capabilities rather than reducing transaction costs; indeed its effectiveness in enhancing capabilities may justify an increase in transaction costs, as, we noted earlier, Chandler has demonstrated. We should, however, observe that a firm’s operation as a social institution, by encouraging both ‘moral sentiments’ and the acceptance of decision premises which
reflect organisational rather than individual objectives, may also be extremely effective in reducing the costs of intra-company transactions. The role of the entrepreneur in developing such social institutions is explored by Witt (1998).
Since Marshall, as Raffaelli (1995) has shown, had his own psychological theory in mind in explaining how organisation aids the growth of knowledge, it is not surprising that he recognised the importance of developing appropriate relationships within a business, nor that he insisted that this development must evolve over a long time. A nexus of contracts cannot deliver what Marshall expected business firms to deliver; to be fair, economists who propound contractual theories of the firm expect it to deliver very little of what Marshall thought was important.
Both direct and indirect capabilities develop through specialisation. Indirect capabilities reduce the costs of particular transactions - which, let us recall, are costs of imperfect knowledge - by embodying knowledge of particular circumstances and particular relationships; their evolution is guided by individual intelligence and the selection processes of a particular knowledge community. Governance is a particular category of ‘knowledge how,’ and most effectively developed within a limited range; and a firm is a specialised system, with direct and indirect capabilities which have been derived from a particular pattern of experience and which are oriented towards a particular, if ill-defined, set of possibilities. Coase ( 1988, p. 45) noted that the costs of governance “will increase with an increase .in the dissimilarity of transactions,” and devoted two paragraphs of his paper (Coase, 1972) to the proposition that “the costs of organizing an activity within any given firm depend on what other activities the firm is engaged in,” concluding that firms will differ in the range of their activities (Coase,  1988, pp. 63-64). He goes on to observe that analyses of the optimum size of firm and of economies of scale do not address this issue; we may now add that the contribution of transaction cost theory is restricted by its focus on the make-or-buy decision which is a consequence of defining the problem in terms of the compatibility of incentives, and ignoring the compatibility of skills.
It is therefore clearly not sensible to attempt to manage an economy as one enormous firm; it is not even sensible to extend a firm into areas of activity that require capabilities which are significantly different from those already developed, and so it is not surprising that firms so often develop a product portfolio which depends on a range of related skills (Teece et al., 1994). However, this principle (an example of ‘knowledge that’) is not always easy to put into practice; that requires the skill (or ‘knowledge how’) to recognise what differences are significant - and perhaps the vision to recognise what apparently significant differences might be made compatible. But because capabilities are developed through practice their limits are rarely investigated, except by accident; and capabilities are always conjectural. There have been many examples of companies expanding, by takeover or internal growth, into areas to which they believed their capabilities were readily applicable, only to find that they were wrong. Some of the most spectacular instances have been provided by financial institutions, whose specialist skills in appraising investment projects proved to be inapplicable to many of the projects which they initiated themselves.
Though some of these failures are more reprehensible than others, we should recognise that a substantial degree of trial and error is unavoidable, and if the economy is to be
regarded as a selection environment for firms we should also consider each firm as a selection environment in which various combinations of capabilities are tested. Research departments are primarily selection environments in which most of the effort is devoted to falsifying hypotheses about the profitability of conjectured new products or processes, in order to avoid the costs of falsification by the market. The selection environment within each firm is the product of its own evolved capabilities; what new combinations can be effectively managed depends on the particular skills of management that have already been developed. Every extension of indirect capabilities facilitates further extensions of a similar kind (where similarity is effectively defined by the firm’s own evolution); it also impedes extensions of a dissimilar kind (where dissimilarity is likewise defined). The obverse of capability is incapability; indeed it is easy to understand why Patel and Pavitt (1995, p. 27) claim that given a profile of a firm’s technological competencies, “we shall probably be able to predict what it does not make.’’ Firms within a relatively homogenous technological, market, and social environment will resemble each other in many ways, but not in all: their productive opportunities, and thus their range of products, will still be differentiated, though much less so than in firms from diverse technological, market, or social environments.
Dissimilar clusters of capabilities are therefore linked by markets. However, in order to understand these linkages, we need to go well beyond the formal conception of a market as a set of supply and demand correspondences. If we consider markets from the perspective of those who use them, both individuals and firms, we can see that they provide access to capabilities which have been developed by other people and other firms. Without such access, as Smith realised, and the consequent ability to command other people’s labour, specialisation is dangerous. (It is thus easy to see why Smith was attracted by the possibility of a ‘labour-commanded’ theory of value.) Since microeconomists typically think implicitly of a vertically integrated firm selling directly to a consumer, they equally typically fail to recognise that firms require access to capabilities also, and not only in labour markets. All firms depend on the capabilities of their suppliers, and every firm which is not a retailer depends on the capabilities of those who provide its links to the final consumer. If we conceptualise markets as a means of providing access to capabilities, we shall recognise that the provision of access is a capability too; and markets differ in the scope and quality of the capabilities to which they give access. There are transaction costs of using markets, but a developed market is a means of reducing the costs of individual transactions. But how, and by whom, are markets developed?
Hayek (1948, p. 97) pertinently observed that “the function of competition is here precisely to teach us who will serve us well: which grocer or travel agency, which department store or hotel, which doctor or solicitor, we can expect to provide the most satisfactory solution for whatever particular personal problem we may have to face.” Hayek correctly identified the issue as that of discovering how to gain access to relevant capabilities which are controlled by other people, and implicitly showed how the competitive process serves to reduce the transaction costs faced by the buyer; but his
examples deflect our attention from the producers’ interests in discovering who will serve them well, and also in discovering who are likely to be good customers for what they have to sell. Marshall (1919, p. 274) recognised that all those who wished to make use of markets, whether as buyers or sellers, would benefit from the development of market institutions which would reduce the costs of individual transactions, but also that the incentive to bear the costs of developing such institutions would be strongest for those who expected to make the greatest use of markets: merchants and (especially) manufacturers. Casson explored the obstacles to trade which face anyone trying to develop a new business; and his discussion of the methods of reducing or removing these obstacles led him to conclude that “the set-up cost of a market organisation is likely to be quite high compared with its recurrent costs” (Casson, 1982, p. 179). It is often possible to reduce substantially the cost of individual transactions, and thus enhance the indirect capabilities which are provided by a particular market, through appropriate investment.
Marshall was uniquely aware of this, and pointed to the need for every firm to build up what he revealingly called its external organisation (Marshall, 1920, p. 458). This external organisation was more than a means of developing a special market, within which a firm might expect to be able to sell more easily - though rarely at a lower price - to those who could reduce their own transaction costs by becoming its customers; it was also a means of acquiring knowledge on which to base “experiments… to see whether some new method, or combination of methods, will be more efficient than the old” (Marshall, 1920, p. 406), or to create “new wants by showing people something which they had never thought of having before” (Marshall, 1920, p. 280). This knowledge cannot be attained by anonymous contracting, which economists have extolled as the means to efficiency; it requires the development of continuing relationships, in which voice is preferred to exit. Even if the supplier’s knowledge is effectively packaged in the good or service which is supplied, so that the user does not need to know how its effects are produced - as is true of many consumer products - the user still needs to know enough about the supplier to support a judgement that the desired effects will indeed be produced; reputations are important, and can only be established at some cost and over sometime. Thus even in the case which comes closest to the standard model of exchange, suppliers need to make irreversible investments (and thus cannot afford to sell at marginal cost, as Marshall well knew and as Casson (1982) makes clear in his analysis of the market-making entrepreneur). To acquire the knowledge with which to make improvements, greater investment is required; but it is not sensible to economise on transaction costs if that means forgoing benefits which exceed those costs.
Markets allow firms to gain access to other firms’ capabilities, but only if they are prepared to make appropriate investments, and to develop their external as well as their internal organisation. We should therefore extend Winter’s observation to say that the knowledge required to make and sell any firm’s products resides in the structure of direct and indirect capabilities within that firm, supplemented by the structure of indirect capabilities that connect it with other firms. There are notable differences of practice in the extent to which firms seek to develop and use such indirect capabilities.
One of the most significant of Patel and Pavitt’s (1995) findings is the magnitude of the commitment of the world’s largest firms to the continuous development of their capabilities in technologies which lie outside the distinctive core of their business. This
may be explained by a combination of three factors: first, parts of these background technologies are so closely complementary that they need to be actively managed within the business - they are indeed core, though not distinctive; second, the boundary of each firm’s core is not well-defined, and liable to change over time, and so these background technologies provide reserves against uncertainty by enhancing the firm’s capacity to absorb additional skills, and third, firms need to understand technologies which they do not themselves intend to practice if they are to make the best use of the capabilities of those firms that do practice those technologies - in other words, they are developing indirect capabilities. Some of Philips’ suppliers are reported to have complained (personal communication) that Philips does not know how to make the best use of their capabilities. This third reason, which reflects Ryle’s perception (cited earlier) that “the intelligence involved in putting the prescriptions into practice is not identical to that involved in intellectually grasping the prescriptions,” illuminates one aspect of the ‘second face’ of research and development to which Cohen and Levinthal (1989) have drawn attention: the capability of making productive sense of other firms’ technologies.
We shall observe another aspect of this second face if we follow Marshall in extending this network further to include connections with firms who are business rivals, but who, for that very reason, are also potentially valuable sources of knowledge, because, unlike perfectly competitive firms, they do things a little differently - but in ways that are easy to understand. Marshall was particularly impressed by the enhancement of capabilities which he observed in his visits to industrial districts, in which “inventions and improvements in machinery, in processes and the general organisation of the business have their merits promptly discussed: if one man starts up a new idea, it is taken up by others and combined with suggestions of their own; and thus it becomes the source of further new ideas” (Marshall, 1920, p. 271). The importance of industrial districts as networks of capabilities has recently been emphasised by Becattini (1990) and other Italian economists, and Porter (1990) has offered a similar argument at a national level. (As a matter of scientific record, acknowledgement should also be made of Klein’s (1977) anticipation of Porter’s central theme).
From an Austrian perspective we may think of such networks as structures of complementary capitals, recalling that it is the differences between firms in the same trade that cause them to be complementary in developing the capabilities of the industry to which they belong. The two faces of R&D - the development of a firm’s own technology and the attempt to understand and learn from the technology being developed by others - reinforce this complementary structure. Complementarity in the generation of production possibilities may then be transformed into substitution in the production of goods. As in a scientific community, specialisation, competition, and communication within a framework of connecting principles constitute an effective structure for the generation of knowledge - in this case primarily of knowledge how, which cannot be learnt without doing; people need their rivals’ contributions in order to compete with them. No one person, or group of persons can control such a structure without drastically restricting the range of capabilities within it; it is the virtue of such a system that it produces results which no one had specifically intended. We should not, however, forget that even in networks the obverse of capability is incapability; the pathology of networks, especially of industrial districts, has often been exhibited.
If we think of markets as institutions embodying firms’ knowledge of how to get things done and how to improve their own productive opportunities, and as forms of organisation embodying a great deal of investment, then we shall not feel the need for a distinctive explanation of those interfirm relationships which involve close co-operation. Such relationships appear as anomalies in both market and transaction cost theory, but the anomalies dissolve in our present perspective; the density of the network connecting one firm to its collaborators and competitors is a matter of degree. One might perhaps be inclined to distinguish a bartering of ‘knowledge how’ that seems to approach what sociologists call a ‘gift exchange’ from price-mediated transactions; but the careful valuation of the exchanges of technology between ICI and Du Pont under their Patents and Processes Agreement was an ex post reckoning that grossly understated the benefits which both parties derived from their access to each other’s capabilities, not least in helping to improve their own and in sustaining their confidence that critical problems, such as the resistance of important synthetic fibres to dyes, were soluble (Hounshell and Smith, 1988, pp. 203-205).
It may nevertheless be worth noting why one of the standard results of transaction cost analysis, that asset-specificity entails exposure to opportunism in a market relationship and therefore requires the internalisation of that relationship, does not necessarily inhibit the formation of dense networks. The reason was spelt out by Richardson (1972) even before this result became standard; it is that activities that are closely complementary may require capabilities with very different knowledge bases, which are therefore liable to be very difficult, as Coase (1937) suspected, to integrate successfully within a formal organisation - especially if those knowledge bases are difficult to analyse or even to express; and if this is so, protection against opportunism through vertical integration exacts an unacceptable cost in efficiency. Moreover, the protection may be illusory, for how can one control what one does not understand?
The advantages of specialisation result from focus, as Smith well knew: knowledge develops through differentiation. The development of a specialised skill depends on a variety of experience, but a variety which can be encompassed within a network of connections. As Ryle’s discussion of ‘knowing how’ and Marshall’s and Hayek’s psychological models all make clear, these skills form patterns rather than logically-derived structures; and if they are partially codified, the codification rests on assumptions some of which are unstated because they are unrecognised. Thus, if a firm seeks to internalise a particular complementary skill which is dissimilar, not only may it be faced with management problems which it does not understand (and often, as apparently with Barings, and Sony’s purchase of Columbia, which it does not know that it does not understand) but it may find that this complementary skill, once divorced from the variety of experience within its own specialist field, ceases to develop and soon becomes a source of competitive disadvantage. Many firms encourage their suppliers to serve other customers in the belief that they will thereby learn how to become better suppliers than they could ever be by restricting their experiences to a single customer. But if we begin our analysis, not with a perfectly competitive ideal and an emphasis on fully-specified contracts but with necessarily limited direct capabilities and the consequent need to know
how to get things done by other people, then we might recognise the possibility of building relationships to manage closely complementary capabilities, and indeed the substantial advantages in many circumstances of doing so.
These advantages are most obvious if we think, not of reducing transaction costs, but of increasing net benefits, and especially of increasing net benefits through the development of new skills, new methods, and new products (Zajac and Olsen, 1993). Often this is best achieved through interactive shaping by members of different organisations with distinctive but closely complementary knowledge bases; and the advantages may justify increased costs of governance, and even new forms of organisation, in which members of one business may effectively work for another (as e.g. logistics specialists manage the distribution systems of supermarket chains) while preserving the distinctiveness on which their capabilities depend. The increased costs may result less from the need to safeguard against opportunism, which may be attenuated by the development of the moral sentiments which Smith (1976a) thought so important within compact communities, than from the need to recognise, and sustain, the differences, sometimes startling, between the ways of thinking within different specialisms.
The cost of producing a good or service is not defined by a publicly-accessible production funnction, but depends on the capabilities of the particular people who produce it; and the costs of a particular transaction mode likewise depend on the particular people who are using that mode. These capabilities are endogenous, and should be analysed in the context of change: response to change, preparation for change, and the generation of change. They result from specialisation, and need to be co-ordinated, but in ways which do not inhibit their continuing development; thus the capability of managing co-ordination while fostering learning, as in the management of joint ventures, is worth particular attention. But since capability is know-how, manifested in action, and accumulated rather than the product of logical analysis, it does not fit comfortably into conventional economic theory. The training of economists does not foster the skills which are necessary to handle the concept of capabilities. It is presumably no accident that the two economists who began to develop such skills, Marshall and Hayek, did so by enquiring into cognitive psychology, nor that their applications, to the organisation of industry and social institutions respectively, attracted little attention from economists. It is also presumably no accident that both Marshall and Hayek gave their applications an evolutionary turn, nor that the seminal modern study of capabilities is set in the context of An Evolutionary Theory of Economic Change (Nelson and Winter, 1982).
The organisation of capabilities is the organisation of systems for generating and testing new and improved skills. These systems are the institutions of economic evolution, which requires specialisation, but not uniformity within each specialism. There may at any time be ‘one best way’ of achieving a particular kind of result, but to train everyone within a specialism in that ‘best way’ would be a recipe for disaster. (Fortunately, there are always a few who escape or resist such framing.) Diversity is necessarily a system property, and it requires the absence of control; for control frustrates
the development of capabilities to which one might later wish to have access. That is the fundamental argument against central planning, an argument which was, not surprisingly, invisible to those economists who sought to define allocative efficiency on the assumption that all attainable knowledge was already available, without even, as Hayek (1937) observed, asking to whom it might be available. Thus the co-ordination problems generated by the division of labour were to be solved by ignoring the reasons for the division of labour. Marshall’s continued preference for small businesses, despite the advantages which, following Babbage, he recognised might be achieved by operating on a large scale, rested substantially (though not solely) on the importance of maintaining the sources of diversity. “Each man’s actions are influenced by his special opportunities and resources, as well as by his temperament and his associations” (Marshall, 1920, pp. 355-356); and outsiders escaped the tyranny of conventional assumptions (Marshall, 1920, p. 197).
Just as direct capabilities change over time, given the appropriate environment, so do indirect capabilities; people discover different and better ways of getting things done. Thus the organisation of capabilities in any sector of the economy is liable to change, often, as Langlois and Robertson (1995) have shown, as a consequence of capabilities developed within that pattern of organisation; there is no fundamental reason for believing that any industry is permanently best fitted to any particular organisational form. However, to explain, and perhaps occasionally to predict, such changes, it is necessary to go beyond such factors as economies of scale and transaction costs. It may also be necessary to go beyond conventional economic methods of analysis. To improve our understanding of capabilities we may need to develop new capabilities, and also new ways of organising those capabilities.
Ansoff, H. Igor, 1965. Corporate Strategy, McGraw-Hill, New York.
Babbage, Charles, , 1989. Economy of manufactures and machinery. In: The Works of Charles Babbage, vol. VIII, Pickeling, London.
Becattini, Giacomo, 1990. The Marshallian industrial district as a socio-economic notion In: Pyke, Frank, Sengenberger, Werner, Becattini, Giacomo (eds.), Industrial Disticts and Inter-firm Co-operation, ILO, Geneva.
Boehni, Stephan, 1994. Hayek and knowledge: Some question marks. In: Colonna, Marina, Hagemann Harald, Hamouda, Omar F. (eds.), Capitalism, Socialism and Knowledge: The Economics of FA. Hayek, vol. II, Edward Elgar, Aldershot.
Casson, Mark, 1982. The Entrepreneur: An Economic Theory, Martin Robertson, Oxford.
Chandler, Alfred D., 1977. The Visible Hand, Harvard University Press, Cambridge, MA.
Chandler, Alfred D., 1990. Scale and Scope, Harvard University Press, Cambridge, MA.
Chandler, Alfred D., 1992. Orgrnsational capabilities and the theory of the firm. Journal of Economic Perspectives 6(3), 79-100.
Coase, Ronald H., 1937. The nature of the firm. Economica NS(4), 386-405. Reprinted in: Ronald H. Coase, 1988, The Firm The Market, and The Law, University of Chicago Press, Chicago, pp. 33-55.
Coase, Ronald H., 1972. Industrial organization: a proposal for research. In: Fuchs, Victor R. (ed.), Policy Issues and Research Opportunities in Industrial Organization, National Bureau of Economic Research, New York. Reprinted in: Ronald H. Coase, 1988, The Firm, The Market and The Law, University of Chicago Press, Chicago, pp. 57-74.
Coase, Ronald H., 1991. The nature of the firm: Influence. In: Williamson, Oliver E., Winter, Sidney G. (Eds.), The Nature of The Firm: Origins, Evolution, and Development, Oxford University Press, New York and Oxford, pp. 61-74.
Cohen, Wesley M., Levanthal, Daniel A., 1989. Innovation and learning: The two faces of R and D. Economic Journal 99, 569-596.
Fransman, Martin, 1995. Japan’s Computer and Communication Industry, Oxford University Press, Oxford.
Hahn, Frank H., Solow, Robert M., 1955. A Critical Essay On Modern Macroeconomic Theory, MIT Press, Cambridge, MA and London.
Hayek, Friedrich A., 1937. Economics and knowledge. Economica NS(4), 33-54.
Hayek, Friedrich A., 1945. The use of knowledge in society. American Economic Review 35, 519-530.
Hayek, Friedrich A., 1948. The meaning of competition In: Individualism and Economic Order, University of Chicago Press, Chicago, pp. 92-106.
Hayek, Friedrich A., 1952. The Sensory Order: An Inquiry into the Foundations of Theoretical Psychology, Routledge, London.
Hicks, John R., 1976. Some questions of time in economics. In: Tang, Anthony M., Westfield, Fred M., Worley, James S. (eds.) Evolution, Welfare and Time in Economics, Lexington Books, Lexington, MA, pp. 135-151. Reprinted in: Hicks, John R., 1982, Collected Essays on Economic Theory, vol., II, Oxford University Press, Oxford, pp. 282-300.
Hounshell, David A., Smith, John K., 1988. Science and Corporate Strategy: Du Pont R&D, 1902-1980, Cambridge University Press, Cambridge.
Keynes, J. Maynard, 1936. The General Theory of Employment, Interest and Money, Macmillan, London.
Kirzner, Israel M., 1973. Competition and Entrepreneurship, Chicago University Press, Chicago.
Klein, Burton H., 1977. Dynamic Economics, Harvard University Press, Cambridge MA.
Kuhn, Thomas S., 1962. The Structure of Scientific Revolutions, University of Chicago Press, Chicago.
Langlois, Richard N., Robertson, Pain L., 1995. Firms, Markets and Economic Change, Rutledge, London.
Leijonhufvud, Axel, 1973. Effective demand failures, Swedish Economic Journal March. Reprinted in Leijonhufvud, Axel, 1981, Information and Co-ordination, Oxford University Press, New York and Oxford. pp. 103-129.
Littlechild, Stephen C., 1986. Three Types of Market Process. In: Langlois, Richard N. (ed.), 1986, Economics as a Process: Essays in the New Institutional Economics, Cambridge University Press, Camblidge, pp. 27-39.
Loasby, Brian J., 1995. Acceptable explanations. In: Dow, Sheila C., Hillard, John (Eds.), Keynes, Knowledge and Uncertainty, Edward Elgar, Aldershot.
Loasby, Brian J., 1998. How do we know? In: Boehm, Stephan, Frowen, Stephen F., Pheby, John (eds) Economics as the Art of Thought: Essays in Memory of G. L. S. Shackle, Rutledge, London.
Marshau, Alfred, 1919. Industry and Trade, Macmillan, London.
Marshau, Alfred, 1920. Principles of Economics, 8th edn., Macmillan, London.
Marshau, Alfred, 1994. Ye machine. Research in the History of Economic Thought and Methodology, Archival Supplement 4, 116-132.
Ménard, Claude, 1995. Markets as institutions or institutions as markets? Disentangling some fundamental concepts. Journal of Economic Behaviour and Organization 23, 161-182.
Menger, Carl, 1871 . Principles of Economics. Translated by James Dingwall, and Bert F. Hoselitz, Free Press, Glencoe, IL.
Minkler, Alanson P., 1993. The problem with dispersed knowledge: Firms in theory and practice. Kyklos 46, 562-587.
Nelson, Richard R., Winter, Sidney G., 1982. An Evolutionary Theory Of Economic Change, Harvard University Press, Cambridge, MA.
Nooteboom, Bait, 1992. Towards a dynamic theory of transactions. Journal of Evolutionary Economics 2, 281-299.
Patel, Pari, Pavitt, Keith, 1995. Technological competences in the world’s largest films: Characteristics, constraints and scope for managerial choice, IIASA Working Paper, 95-66.
Patel, Pari, Pavitt, Keith, 1997. The technological competences of the world’s largest firms: Complex and path-dependent, but not much variety, Research Policy 26, 141-56.
Penrose, Edith T., 1959. The Theory of The Growth of The Firm, Basil Blackwell, Oxford, 2nd edn. (1995), Oxford University Press, Oxford.
Porter, Michael E., 1990. The Competitive Advantage of Nations, Macmillan, London.
Raffaelli, Tiziano, 1995. The principles of organization: A forgotten chapter of Marshallian economics. Presented to European Conference On the History of Economics, Rotterdam, 10-11 February 1995.
Richardson, George B., 1960, . Information and Investment, Oxford University Press, Oxford.
Richardson, George B., 1972. The organisation of industry. Economic Journal 82, 883-896. Reprinted in: Richardson, George B., 1990. Information and Investment, 2nd edn., Oxford University Press, Oxford 224-242.
Rosenberg, Nathan, 1994. Exploring the Black Box: Technology, Economics, and History, Cambridge University Press, Cambridge.
Ryle, Gilbert, 1949. The Concept of Mind, Hutchinson, London.
Schumpeter, Joseph A., 1934. The Theory of Economic Development, Harvard University Press, Cambridge, MA.
Smith, Adam, !976a. The Theory of Moral Sentiments, Raphael, David D., Macfie, Alec L. (eds.), Oxford University Press, Oxford.
Smith, Adam, !976b. An Inquiry into The Nature and Causes of The Wealth of Nations, Campbell, Roy H., Skinner, Andrew S. (eds.), 2 vols., Oxford University Press, Oxford.
Smith, Adam, 1980. The principles which lead and direct philosophical enquiries: illustrated by the history of astronomy. In: Wightman, William PD. (ed.), Essays On Philosophical Subjects (1795), Oxford University Press, Oxford 33-105.
Teece, D.J., Rumelt, R., Dosi, G., Winter, S.G., 1994. Understanding corporate coherence: Theory and Mdence. Journal of Economic Behaviour and Organization 23, 1-30.
Winter, Sidney G., 199!. On Coase, Competence, and the Corporation.In: Williamson, Oliver E., Winter, Sidney G. (Eds.), The Nature of the Firm: Origins, Evolution, and Development, Oxford University Press, Oxford, pp. 179-195.
Williamson, Oliver E., 1975. Markets and Hierarchies: Analysis and Anti-trust Implications, Free Press, New York.
Witt, Ulrich, 1998. Imagination and leadership - The neglected dimension of an (evolutionary) theory of the firm. Presented to EMOT Workshop On Technology and the Theory of the Firm, University of Reading.
Young, Allyn G., 1928. Increasing returns and economic progress, Economic Journal 38, 523-542.
Zajac, Edward J., Olsen, Cyrus P., 1993. From transaction cost to transaction value analysis: Implications for the study of interorganizational strategies. Journal of Management Studies 30, 131-145.
Ziman, John M., 1978. Reliable Knowledge, Cambridge University Press, Cambridge.