A Critique of Williamson's Transaction Cost

Economics and its relevance to Agricultural Economics

The work of Williamson has given economics a new vigour. By putting forward a synthesis of past forgotten statements ( Williamson 1985 p.6 ) as an agenda for his research program, Williamson has opened a black box that was closed and neglected for many decades. Institutional economics has replaced the neo-classical ( NC ) economics as a framework through which to examine economic phenomena ( Williamson 1985 p.12 ). But the focus of his institutional economics ( transaction cost economics ( TCE ) and new institutional economics ( NIE ) ) has come under attack.

The purpose of this dissertation is to examine the debate between Williamson and his critics which has taken place during the 80s and early 90s. To summarise the outcome, it will be argued that Williamson's TCE has fatal analytical weaknesses that cannot be overcome. Thus the agenda of TCE should be abandoned. Here it is not possible to reconstruct an alternative agenda, though some pointers to one will be given.

In section I, I shall outline the mainstream economic orthodoxy that existed before Williamson's work. The theoretical saliences of TCE will be introduced in section II. Criticisms of Williamson's work will be examined in section III and some footnotes to section II. Finally, the relevance to agricultural economics is explored in section IV; although this work is theoretical, since my interest is methodological, rather than applied or pragmatic.

Section I: Neo-classical Economics

NC economists' ( NCE ) emphasis has been on the study, primacy, and efficiency of markets as allocators of resources. Though NCE do acknowledge that markets can fail, such acknowledgements are rare ¹. Generally, markets are assumed to perform a social function that no other mechanisms can match: intentional and purposeful actions of heterogenous individuals are co-ordinated by " the invisible-hand ", resulting in optimal and stable unintended consequences. NCE approvingly quote Adam Smith : " If every individual maximises his own selfish utility, then, given certain conditions, society will automatically attain optimality " ( my emphasis ). Any deviations from " market economy " will lead, NCE argue tautologically, to inefficient allocation of resources, and so must be resisted.

What are the " given certain conditions " that ensure the efficacy of the market ? Rational, mechanical ², utility-maximising behaviour by all relevant agents in the absence of imperfect information and true uncertainty. The individual is the best judge of his/her own welfare. Individuals' optima are necessary for social optimium. Finally, there are no restrictions of entry or exit into the markets. ³

Anything that did not correspond to the perfectly competitive firm is constituted by the notion of monopoly power. Monopolies were regarded as " social evils ", that required trust-busting in order to make the industry like a perfectly competitive market.

Section II: Transaction Cost Economics

( a ) Transaction Costs

Williamson approached the study of firms and markets from a different angle: " Contrary to earlier conceptions __ where the economic institutions of capitalism are explained by reference to class interests, technology, and/or monopoly power __ the transaction cost approach maintains that these institutions have the main purpose and effect of economising on transaction costs " ( 1985, p.1 ).

" The transaction is the ultimate unit of microeconomic analysis " ( 1975, p.xi ). The definition of a transaction is " a good or service that is transferred across a technologically separate interface " _ ( 1985, p.1 ). In a well-working interface ( or economic system ), transfers occur smoothly and without costs. But such " exchanges " are fraught with difficulties in reality. NC's perfectly operating world does not hold. Williamson challenges the NCE behaviourial assumptions.

( b ) Behavioural Assumptions

Williamson insists that the study of economic institutions requires an informed appreciation of " human nature as we know it; . . . the principal problem in understanding the actions of men is to understand how they think __ how their minds work " ( 1989, p. 138 ). Williamson states that " TCE characterises human nature as we know it by reference to bounded rationality and opportunism " ( 1985, p.44 ). Bounded rationality assumes agents to be " intendedly rational, but only limitedly so " ( quoting H. Simon, 1985, p.46 ). This " semistrong form of rationality " results from limited cognitive competence. Opportunism is the strongest form of self-interest: " self-interest seeking with guile " ( 1985, p.47 ). Its forms of deceit can be active and passive, and ex-ante and ex-post. It includes lying, stealing, and cheating. It also refers to deliberate incomplete or distorted disclosure of information, breaking of contracts and renegotiating the terms. In the NC world, contractual agents accurately disclose information and execute contracts to the letter. Williamson rules out obedience to contracting agents by others, since self-interest is the basis for bargaining.

This has repercussions for Williamson on the types of contracting that are feasible. An " ex-ante contingent claims contract ", which entails comprehensive contracting of all relevant future contingencies ( thus, a contract to handle a complete decision tree / events network ), is infeasible. Bounded rationality precludes this " classical contracting " ( Williamson 1986b, p.103 ). Thus all contracts are incomplete, so that the ex-post side of contracting takes on special economic importance. " The study of structures that facilitate gapfilling, dispute settlement, adaption, and the like thus become part of the problem of economic [ institutions ] " ( Williamson ( 1989, p.139-40 )). Since agents are opportunistic, contracts which only entail promises are naive and uncertain modes of contracting. Thus " ex-ante efforts to screen economic agents in terms of reliability [ i.e. avoiding " adverse selection " ] and . . . ex-post safeguards to deter opportunism [ i.e. avoiding " moral hazards " ] take on different economic significance as soon as the hazards of opportunism are granted " ( Williamson ( 1989, p. 140 )).

Given these behavioural assumptions _, governance structures ( modes of transactional contracting and execution ) attempt to solve an economic problem, different to that of NC: " [ to ] organise transactions so as to economise on bounded rationality while simultaneously safeguarding them against the hazards of opportunism " _ ( Williamson ( 1986a, p.177, my emphasis )). In pre-TCE, economists took the organisation of economic activity as given, and characterised firms as production functions with certain resource constraints and technical relationships, and with objective of profit maximisation. Hence, ways of attaining allocative efficiency were the problem, not the minimisation of transaction costs, as transaction costs did not exist. In this way, the importance of organisations was ignored, and an in-built bias resulted favouring " market operations ".

( c ) Dimensions of Transactions

The striking innovation of Williamson has been to operationalise TCE, by suggesting three principal dimensions of transactions which affect transaction costs, and hence governance structures.

" Asset specificity " has ramifications, in that the continuity of the relationship between the transacting parties is valued because of unrecoverable investments made by the parties in order to execute contract. So " contractual and organisational safeguards arise in support of transactions of this kind " ( Williamson ( 1986a, p. 179 )), giving rise to " complex ex-ante incentive responses . . . [ and ] complex ex-post governance structure responses " ( Williamson ( 1989, p. 143 )), in the face of bounded rationality, opportunism and uncertainty. Thus, NC's conception of exchange where identities of the parties ( " faceless buyers and sellers " ) do not matter is abandoned in the face of asset idiosyncrasy.

Williamson distinguishes between various sources of uncertainty: " [p]rimary uncertainty is uncertainty of a state-contingent kind, while secondary uncertainty arises from lack of communication . . . [ which is ] of a rather innocent and non-strategic kind . . . Uncertainty of a strategic kind is attributable to opportunism [ arising from strategic nondisclosure, disguise, or distortion of information ] ; I will call it behavioural uncertainty " ( Williamson ( 1989, p.181 ). Williamson points out that economic institutions are important in facing and dealing with uncertainty, especially of the behavioural variant. Economic organisation tries to mitigate the costs of transactions that are asset-specific. Such costs arise from the hazards of behavioural uncertainty in an ever changing world.

As discussed above, special governance structures are required for " non-standard transactions " ( i.e. asset-specific with hazards of behavioural uncertainty in a dynamic world ). " But special structures come at a great cost, and the question is whether the costs _ can be justified. This varies with the benefits _ on the one hand and the degree of utilisation on the other " ( Williamson ( 1985, p.60, my emphasis )). Hence the more frequent the transactions, the easier to recover the costs of special governance structures. So more likely that the special governance structure will be brought into existence by the intentional design _ of the parties, after " working out " how to minimise all transaction and production costs.

( d ) Comparative Study of Institutional Alternatives

Following from the analysis of the dimensions of transactions, Williamson goes on to analyse the conditions under which " special governance structures " would supplant standard market exchange. Williamson writes, " markets and firms are alternative instruments for completing a related set of transactions _; . . . [ and ] whether a set of transactions ought to be executed across markets or within a firm depends on the relative efficiency of each mode " ( Williamson ( 1975, p.8 ), my emphasis ). Governance structures are regarded as outcomes of optimisation problem-solving, designed to economise on the sum of production and transaction costs ¹_.

When transactions are highly specific and affected by increasing degrees of uncertainty, market exchange ( " market governance " ) cannot continue, " classical contracting " fails, since a complete contract is precluded by the failure to control opportunism and to overcome bounded rationality. Under such conditions, other institutions have to be designed so as to allow continuing transactions between the parties. Otherwise either standardised goods and services alone will be offered, or specialised trading will stop under prohibitive uncertainty. " Neo-classical contracting " / " trilateral governance " is one possibility, since it allows a third party to arbitrate in disputes, rather than relying on litigation, which would disrupt transactions ¹¹. However, in cases of frequent transactions and recurring adaptions, the costs of a special governance structure can be recovered. " The choice of organising mode then turns on which mode has superior adaptive properties " ( Williamson 1986b, p.116 ), my emphasis ) ¹². That "choice " is between " obligational contracting " / " bilateral governance " and " internal organisation " / " unified governance ", the latter being more benefical in cases of higher degrees of " idiosyncratic transactions ", since " internal adaptions can be effected by fiat ¹³ " ( Williamson ( 1986b, p.114 )), and the costs of a special governance structure can be recovered in recurrent transactions.

( e ) An Application of TCE

Williamson argues that an " authority hierarchy " ( an internal organisation of work ) of the C19th was more efficient than other forms of internal organisation of work: " the question of efficient versus inefficient modes of internal organisation comes down to an examination of their properties in bounded rationality and opportunism respects. Organising modes that economise on scarce information processing and decision-making capability have superior properties in transaction cost terms, ceteris paribus ¹_ " ( Williamson ( 1980, p.12 ), original emphasis ). He dismisses claims from " the New Left " that the concept of hierarchy lacked any efficiency rationale, who contend that the authority hierarchy arose in the service of capitalist power over labour; he argues that transaction costs efficiency provides an alternative explanation of the organisation of work without evoking power struggles and workers' alienation ¹_. He is dissatisfied with a technological deterministic explanation of the hierarchy's existence: " [t]echnology is not determinative of economic organisation if alternative means of contracting can be described that can feasibly employ, in steady state respects ¹_ at least, the same technology. . . . [T]echnology is more usefully regarded as a factor that delimits the set of feasible modes __ the final choice thereafter turning on a transaction cost assessment " ( Williamson ( 1985, p.89 )).

After listing the possible internal organisations available in the C19th, and after erecting a " simple efficiency criteria " by which to measure the transactions costs of each mode in organising " a set of related transactions ", Williamson assigns positive and negative values in terms of transaction costs ¹_ ( see Table I ). Aggregating these values, Williamson proves that the " authority relation " scores the most favourably, in terms of efficiency ( see Table II ). From the table, he concludes, " Modes are listed roughly in the same order as they appear historically. Although it is possible to argue that later modes [ of production ] displaced earlier modes because ' interests ' were determined to stamp out ' autonomy ', an alternative hypothesis is that successor modes have superior efficiency properties to predecessor modes " ( Williamson ( 1980, p.29 ), my emphasis ) ¹_, ¹_.

There are other applications of TCE in areas of the family, the evolution of the modern corporate firm, vertical integration, non-standard contracting and regulation ( Williamson ( 1985, 1989 )).

Section III

In this section I shall discuss why TCE cannot be regarded as a " useful " research tool for " economists ". I shall put forward the proposition that TCE is " a wolf in sheep's clothing ", as it provides a safety line to a discredited NC economics, and omits important aspects of analysis, concerning the social importance and power of institutions and real uncertainty ( e.g. Williamson's treatment of the historical evolution of the organisation of work neglects to endogenise power, social embeddedness and real uncertainty ( 1980, pp.21-30 )). The reasons for this lack of conviction were hinted at in the footnotes for section II, and will be expanded and extended below. This criticism will be organised as follows: the internal logic of TCE; the methodological individualism of TCE and the decision-making processes of other schools of thought; the institutional evolution of TCE and of its critics; the characterisation of institutional power; the social structural embeddedness of institutions; and finally a re-assessment of Williamson's historical evidence for the C19th. ( See Appendix for the structure of evaluating theories, and its weaknesses. )

( a ) Internal Logic

The argument that institutions that exist economise on transaction costs ( or reflect a preference for unmeasurable objectives like " atmosphere " and " dignity " ) runs into difficulty if it is asked under what conditions would the argument be falsified. Williamson writes, " [ A ] presumption of market failure is warranted where it is observed that transactions are shifted out of a market and into a firm; a presumption of internal organisation failure is warranted for transactions that are unshifted ( continue to be market-mediated ) " ( Williamson ( 1975, p.20 )). It becomes tautological and unhelpful if transaction costs ( TC ) economists insist that their theory of minimisation of transaction costs is proved correct by the existence of institutions since the function of institutions is to be transaction costs efficient. Williamson states, " What has hitherto been regarded as a set of diverse and anomalous contracting practices has been shown to be variations on a common theme: economising on transaction costs " ( Williamson ( 1989, p.199 )). This is made even more impossible if I add that Williamson includes several " ad hoc qualifiers " to his main thesis ( see footnotes 5, 12, 18 ), with which to deny that anything can falsify his thesis. He argues that dignity, atmosphere, and participation are important to institutions because they are not only production units with labourers merely as robots. These factors, he argues, can lead to a trade-off between TC efficient institutions and welfare enchancing institutions, but the end-product will reflect contracting parties' preferences. But he fails to endogenise these factors into his analysis, using them instead as ad hoc rationalisations of what exists that fails to be TC efficient. He also fails to explain what happens if the contracting parties do not share the same preferences for a mixture of TC efficient and welfare enchancing arrangements.

Williamson, however, though he provides a description of the principal dimensions of a transaction, offers no means by which to determine their exact value for the different modes of contracting. How do I determine the values of " asset-specificity ", " behavioural uncertainty " and " frequency " in order to carry out comparative institutional studies of relative transaction costs efficiency ? Are the values to be dualistic ( e.g. Table I ), data-based obtained from econometric studies, or ad hoc from questionnaires to the transacting parties concerning their reasons for mode of contracting ? This leaves Williamson open to the charge that his evocation of lower / higher transaction costs for particular modes is only " ex-post rationalisation of observed situations " ( Bharadwaj ( 1989, p.18 )). The danger of " TCE functionalism " is that it blinds economists into " seeing " and " justifying " everything according to Williamson's meta-story, to the exclusion of other interpretations with their own stories to tell ( e.g. economics of power, economics of technology, economics of social psychology and cultural rules ).

As mentioned in footnote 6, there seems be an inconsistency in Williamson's main thesis where, " [t]he criterion for organising commercial transactions is assumed to be the strictly instrumental one of cost economising. Essentially this takes two parts: economising on production expense and economising on transaction costs " ( Williamson ( 1986b, p.110 )). If I am to take " bounded rationality " ( one of Williamson's behavioural assumptions ) seriously, do not the parties suffer from cognitive limitations: restricted computational capacity, long-run and short-run memory storage capacity, and unique, heterogenous, and unshared mental schemata with which to interpret data ? If so, then how can it be possible for parties to devise and agree on a governance structure that " economises " production costs as well as transaction costs ? Williamson writes that when judging the efficiency of contracts, they must be examined " in their entirety ", yet this sits oddly with his rhetoric of bounded rationality. Surely only the socio-cultural salient features of the economic contracts can be accounted for by the biased and limited cognitive minds that devise economic institutions. Hodgson gives an interpretation of Simon's work which conflicts with Williamson's, " Simon's argument . . . is that a complete or global rational calculation is ruled out, thus rationality is ' bounded '; agents do not maximise but attempt to attain acceptable minima instead . . . this ' satisficing ' behaviour does not simply arise because of inadequate information, but also because it would be too difficult to perform the calculations . . . In Williamson's work ' economising on transaction costs ' is part of global, cost-minimising behaviour [ dual of NC standard assumption of maximisation ], and this is inconsistent with Simon's idea of bounded rationality " ( Hodgson ( 1991, p.199 )).

Williamson's response has been that his interpretation of bounded rationality is " correct " as he " does not subscribe to the view that social institutions and arrangements are the adventitious result of legal, historical, social, or political forces ", and he regards, " economising " as fundamental to economics since it gives a deterministic outcome, which forms the basis for the refutation of the hypothesis ( Williamson ( 1987, p.618-9 )). He writes that alternative hypotheses have to give an analysis to reach a deterministic outcome, otherwise it is unrefutable, and by implication not scientific. However, gone are the days of a mechanical, deterministic viewpoint of the physical sciences. Now that " quantum physics " has displaced " Newtonian physics ", economists have to change their notion of determinism, if they wish to continue to use physics as a metaphor. As will be shown below it is impossible to have a pre-deterministic equilibrium in an uncertain world, where decisions are based on " subjective, present imaginings of future sequels to actions ". Economics as a science requires another criterion by which to evaluate theories if " bounded rationality ", " opportunism " and " uncertainty " are to be its basic tenents, because " determinism " clearly fails to be compatible with the tenents. Yet as I have noted above, the implications of Williamson's hypothesis itself is not refutable as it is tautological and has several " ad hoc qualifiers " to protect it against refutations. Williamson should have studied the deterministic outcomes of satisficing, and not economising, a heuristic rule in keeping with bounded rationality, which would still have allowed him to obtain the refutable implications of his amended TCE hypothesis. Even Simon, from whom Williamson hi-jacks the term, " bounded rationality ", is sceptical of Williamson's project when he describes NIE " seriously lacking in credibility " ( Simon ( 1992, p.6)), and that " the steps taken in NIE are rather conservative moves away from the NC framework " ( Simon ( 1992, p.21 )). He objects to Williamson's arguments that organisations are optimal and intentional outcomes of individuals' motivations of self-interest. I shall look into the private intentional design of institutions and functions of institutions in section III ( b ) and ( c ). And I shall also show why Williamson's construction of the tables on work organisation fails to take bounded rationality, and power ( or, as he puts it, opportunism ) seriously in section III ( f ).

As well as there being an inconsistent use of " bounded rationality ", there is an equally selective and inconsistent use of his other behavioural assumption, " opportunism ". Williamson argues that opportunism leads to transaction costs. " The strategic manipulation of information or misrepresentation of intentions . . . are to be regarded as opportunistic and do have comparative institutional significance for assigning transactions to one mode of organisation instead of another " ( Williamson ( 1975, p.26-7 )). He argues that internal organisation is more suited to restrain opportunism, and " is able to settle many such disputes by appeal to fiat, an enormously efficient way to settle instrumental differences " ( Williamson ( 1975, p.30 )). But Dow rightly points out, " TC theorists tend to see authority primarily as a remedy for opportunism, rather than as a device which might be abused in an opportunistic fashion " ( Dow ( 1987, p.20 )). This is an inconsistent omission. Is not information and decision-making concentrated centrally in the hierarchical mode of contracting, so allowing via " use of fiats " a certain " small number " of individuals to use " information impactedness " to their private " opportunistic " advantages, rather than for " the interests of the system " ? Yet, though Williamson acknowledges that information impactedness, small numbers and opportunism can transform trading situations, he is only able to see such transaction costs occurring in market governance, rather than in internal organisation ( Williamson ( 1975, pp.29-30, 35-37 )), without justifying the resistance of the latter to such temptations.

Williamson has responded to the charge of neglecting downward opportunism by managers which is structurally encouraged by acknowledging that " efficient bargaining . . . is upset by such conditions " ( Williamson ( 1987, p.620 )). Yet he feels that " parties to an asset-specific exchange have a mutual interest in perfecting the contracting relation " ( Williamson ( 1987, p.620 )). Yet this statement is inconsistent with the concept of " opportunism " ( as opportunism is not well-behaved ), and flies in the face of his example of the change of governance structure at Carnegie mill, after the union power was destroyed ( see footnote 19 ). Did the union members regard the change to authority hierarchical organisation of work in their " mutual interest in perfecting the contracting relation " ? If so why they did they resist Carnegie and Frick 's measures to change the existing organisation of work ( " inside contracting " ) at the mill, which gave union members greater power in organising production ? I shall expand the point that certain people can wield " power " through " fiat ", informational control, rights, and values in changing governance structures to serve their own " class efficient " interest in section III ( d ).

" Uncertainty " is to economics what " uncertainty " is to physics: easy to theorise but difficult to understand. Although Williamson acknowledges the importance of uncertainty and even categorises it, he fails to understand its full implications. For Williamson, " behaviourial uncertainty " arises from opportunism, and its importance is confined to frequent, asset-specific transactions where bounded rationality precludes a " classical mode of contracting " ( complete contracts ) and so necessitates adaptions to the governance structures due to frequent exogenous disturbances ( Williamson ( 1985, pp.56-60 )). Williamson writes, " [I]ncreasing the degree of uncertainty makes it more imperative that the parties devise a machinery to ' work things out ' " ( Williamson ( 1986b, p.117-8 )), and of course, the designed governance structure will be optimal i.e. at an equilibrium. But this is impossible to achieve if " primary and secondary uncertainty " is included in his analysis. Williamson ignores how uncertain, bounded-rational and opportunistic beings can adjust their behaviour to become consistent with each other, so that an equilibrating governance structure emerges to ensure co-ordination of beings' activities that economises on transaction costs . What process exists that allows for diverse beings to work things out if they ignorant of " reality " and possess selective and biased knowledge ? Furthermore Williamson fails to discuss how bounded rational and opportunistic designers of governance structures can ever be sure of the consequences of their actions and can ever be certain what means to adopt to bring about a given desired end. Decisions are made from " present imaginings of alternative future sequels to action [ and not ] between [ actual ] alternative future sequels themselves " ( Currie and Steedman ( 1990, p.154) in discussing Shackle's ideas ). Will not the " present imaginings " of the consequences diverge from the " actualities " in an uncertain and complex world where individuals cannot compute nor know each others' decisions and actions ? " Errors " arising from imperfect and distorted information, and a " learning process " arising from the response to more knowledge are not discussed in Williamson's analysis of bounded rational and opportunistic actors inhabiting the world of uncertainty ! Williamson neglects that in an uncertain world, " historical time " moves in one direction only, where yesterday's actions cannot be reversed in order to solve today's problems, in the light of more information and knowledge. Like NC, TCE is only " concerned with never-to-be-realised states and with unconvincing explanations of how the economy gets there " ( Clark ( 1987-8, p.275 )). It is the institutional factors ( regulatory influence of institutions, laws, folkways, mores and other habit-patterning arrangements ) and not NC market forces or TCE's ex-ante and ex-post incentive structures of institutions that directs individuals' intentions and actions towards order in an uncertain world, by making their intentions and actions consistent with each other. But by allying himself with methodological individualism and design argument ( those institutions that are designed by contracting parties are realised immediately ), Williamson ignores habit institutions, power, rules, rights, values and norms, and evolved, feedback processes which may co-ordinate sufficiently well to create commonly shared expectations for transactions to take place in real time without too much disappointment being experienced by certain groups ( see section III ( b ) - ( e )). Similarly, Carvalho ( 1983-4 ) argues that there must be constraints to individuals' imaginations of the future otherwise transactions will only ever occur through " accidental coincidences ". Yet Williamson assumes that free individuals can create stable governance structures with no constraints to their imaginations, without detailing how such resource co-ordination can be achieved within market and hierarchical structures.

I will postpone my discussion on the " persistence " of governance structures to section III ( c ), which is crucial to an understanding of how " efficiency " may not persist. There it shall be argued that efficient institutions cannot exist because in an uncertain world stability is more important than optimality in order for institutions to operate.

In this sub-section, I have tried to argue that even by the criterion of Williamson's own use of rhetoric, his thesis cannot stand. He fails to explicate the terms " bounded rationality ", " opportunism ", and " uncertainty ". He also does not provide the refutable implications of his thesis since they are protected by ad hoc qualifiers. And finally his argument is not an argument, but a tautology.

I shall consider other criticisms, by other critics.

( b ) Decision-making Processes and Methodological

" Methodological individualism " is a trade-mark of NC and NIE study. By it is understood that all institutions and other social phenomena are explained by reducing them to theories of individual action where only the psychology of individuals, initial resource endowments, and natural givens are exogenous. References to other social phenomena as explanations are not allowed ( only " psychologism " is valid ). But Rutherford rightly reminds me that " for institutions to act as institutions they have to be genuine constraints on individual action and affect individual action. In addition, institutions embody social values and norms to which an individual is socialised and habituated. Institutions can thus influence the individual's preferences and goals, so that the individual's psychology can no longer be taken as exogenous to institutional development ( Rutherford ( 1989, p.302 )). If there are constraints on individuals' means and ends in devising certain types of governance structures, imposed by other institutions, the optimality ( TC efficiency ) of such governance structures has to be doubted. This is because the power to change institutions to economise on transaction costs is precluded by institutional forces protecting the prevailing governance structures, and hindering and creating other types of governance structures to stop changes in prevailing governance structures, for the benefits of certain vested groups.

Methodological individualism imputes purposes and intentions to individuals, which is sufficient reason for institutions to be realised according to their original conception ( " pragmatic institutions " ). Williamson quotes, " economic actors are assumed to be ' intendedly rational, but only limitedly so ' ( Simon ( 1961, p.xxiv )) . . . Given limited competence, how do the parties organise so as to utilise their limited competence to best advantage ? " ( Williamson ( 1985, p.45-6 )). Langlois, who has his own brand of NIE, correctly argues against Williamson's universal principle of " sufficient reason ", " Are not many institutions in fact pragmatic in nature ? . . . Should not many institutions be explained as the result of the conscious design of individuals ? . . . Surely there are pragmatic institutions . . . We can explain the existence of, say, a small software firm by the intentions of its principals to form precisely such a firm. But even here, the firm incorporates many other social institutions that are organic [ i.e. the unintended consequences of aggregation of intentional beings, such as co-ordination by the price mechanism resulting in an ex-post equilibrium from the different outcomes of the ex-ante actions of market actors ]. [M]ore importantly, the firm may well work out in a manner quite different from what the founders had intended . . . The pragmatic model of institutions is the more narrow and primitive one " ( Langlois ( 1989, p. 287-8 )). Williamson's insistence that all institutions can be explained by a deliberate desire to economise on transaction costs ( Williamson ( 1985, pp.393-5 )), ignores many studies that show many institutions are " organic " ( e.g. Vanberg ( 1989, p.340 ) who gives an account of Menger's " organic origin of the institution of money " where in the process of improving their wealth by trade, people begin to exchange less marketable for more marketable goods in the initially pure-barter economy. An innovative use of a medium of exchange is then made by some pioneers, and this initial discovery is imitated by others as the pioneers become successful. Eventually the practice becomes a socially uniform pattern of behaviour ). And culture-bound designers work within an accepted range of institutionalised behaviour ( e.g. a legal system and an educational system ), with which to create governance structures. In section III ( e ), I shall discuss in greater detail on how governance structures are " embedded " with social, cultural and legal frameworks, without which such institutions would collapse. Moreover, Williamson's focus on static equilibrium institutional forms ( Williamson ( 1985, pp.32-5 )), ignores that individuals are ex-post wrong about many of their ex-ante intentions in an uncertain and complex world. Hence individuals go through " a never- ending institutionalised learning, discovery, creative, process " until they are satisfied with their newly established patterns of behaviour.

Williamson commits " transaction costs reductionism ": seeing everything reduced to transaction costs and regarding transaction costs as supreme over other factors as the essence of phenomena's existence ( Williamson ( 1985, p.17 )). He treats parties to a contracting governance as if they are individuals with the aim of economising on transaction costs. But in large institutions, people involved will have many diverse aims e.g. job security, empire building, high profile enterprise, and political leverage. Further, Williamson fails to draw any distinction between careful, rational, planned actions to, say, economise on transaction costs and other type of actions which involve habits, routines, and behavioural rules ( a hierarchy of cognitive processes, each stage enacted at various time, space, personal needs, and social institutions ). Furthermore, he does not suggest a framework by which to address interpersonal conflict involving aims and / or means. The implication of the Carnegie mill example, for Williamson, was that the party that won was for the best !

This brings me to consider the decision-making processes which conflict with Williamson's image. There exist " habit institutions " which, although they emerge from human actions, are not from human designs ( i.e. fully or partially unconscious, rather than fully deliberative, processes ). For Williamson, efficient institutions are brought about by design ²_. However, Hodgson points out the importance of habits. " One of the functions of habits is to deal with the complexity of everyday life; they provide us with means of retaining a pattern of behaviour without engaging in global rational calculations involving vast amounts of complex information " ( Hodgson ( 1991, p.179 )). This goes against Williamson's conception of the effects of the complexity and the uncomputability of information by individuals: where governance structures are designed to economise transaction costs ( Williamson ( 1986b, pp.110-11 )). Williamson is able to rehabilitate the optimality and intentionality of economic events by arguing that governance structures allow ill-informed and ill-processed beings to carry out TC efficient operations. Yet, Williamson fails to realise that these institutions are designed by these very same imperfect beings, so what guarantees are there that these institutions will be TC efficient ? ²¹.

Hodgson puts forward an alternative explanation for the function of such institutions, " Stabilised and routinised behaviour establishes and reproduces a set of rules and norms ' fixed by habit, convention, tacit of legally supported social acceptance of conformity ' ( Korni ( 1982, p.79 )) . . . Such inflexibilities or constraints suggest to the individual what other agents might do, and the individual can then act accordingly ( Hodgson ( 1991, pp.181-2 )). Thus institutions and routines enable decisions and actions to be realised by providing reliable information and knowledge ²² ( interpreted information reflecting common institutionalised schema ), so that in a complex and uncertain world, regular and predictable behaviour is possible. The informational and cognitive functions of institutions enable satisfactory expectations to be formed, without which no economy would function in an uncertain world. Yet for institutions to overcome the complicated and uncertain world, they require more than knowledge of other peoples' thoughts, restrictions must be imposed on the freedom of peoples' " present imaginings of future sequels to actions " ²³. Dixon writes, " While uncertainty is made possible by the presence of other minds, other choosers, this existence of others is a necessary but not sufficient condition for the presence of expectation to result in the presence of uncertainty. The degree of control which the decision maker has over the consequences of his or her [ and others' ] actions is also relevant . . . [E]conomic agents may ' attack ' uncertainty directly by the adaption of a portfolio stance which will maintain their own flexibility and by the devotion of resources [ e.g. habit institutions, norms, and values ] toward controlling the consequences of their own, and others', decisions " ( Dixon ( 1986, p.587-8 ), my emphasis ). Similarly, Knight and Merriam write, " One individual can choose or plan intelligently in a group of any size only if all others act ' predictably ' and if he predicts correctly. This means . . . that the others do not choose rationally but mechanically follow an established and known pattern, or else that the first party has coercive power, through force or deception . . . Without some procedure for co-ordination, any real activity on the part of an individual, any departure from past routine, must disappoint the expectations and upset the plans of others who count on him to act in a way predicted from his past behaviour " ( Knight and Merriam ( 1947, p.60 )). Note " deception " does not differ from opportunism. Dixon, and Knight and Merriam give a new interpretation of Williamson's example of Carnegie mill and its change of governance structure: Carnegie and Frick ( one side of the two transacting parties ) had to brow beat the union members ( the other side ) in order to have power over their actions, and institutionalise structures and relations for union members to behave predictably. For union members to act with bounded rationality would have resulted in a failure of Carnegie and Frick to predict their actions and so fail to realise their own benefits. Without the exercise of power that governance structures have over individuals, the co-ordination of their actions would fail, because expectations would fail to be realised. Williamson has not written on the enforcement of habits and expectations by governance structures, expecting optimal behaviour under the conditions of bounded rationality and opportunism to be predictable. But Heiner challenges such a misconception.

Heiner argues that habits and behavioural rules arise in an uncertain and complex world to ensure that reliable actions are carried out, " Uncertainty exists because agents cannot decipher all of the complexity of the decision problems they face, which literally prevents them from selecting the most preferred alternatives. Consequently, the flexibility of behaviour to react to information is constrained to smaller behavioural repertoires that can be reliably administered. Numerous deviations from the resulting behaviour patterns are actually superior in certain situations, but they are still ignored because of uncertainty about when to deviate from these regularities " ( Heiner ( 1983, p.585 )). The suggestion that agents will not perform superior acts because they lack the competence to deal with the difficulty of the decision-making ( so resulting in their following behavioural rules ) has implications for Williamson's " Contractual Man ". Contrary to Williamson's suggestions that agents will act to devise TC efficient modes of contracting ( Williamson ( 1985, pp.45-6 )), the implication of Heiner's work is that institutions and other structural behaviour will be not be optimal, because agents, who cannot know when to deviate from rules, will continue to follow existing patterned behaviour which are not optimal for the changing situations ( i.e. optimal and predictable institutions cannot exist, because they cannot function in the uncertain and complex world ). Contrary to Williamson's explanation of the existence of institutions as a way of economising on transaction costs, Heiner writes, " [E]volved institutions are social rule-mechanisms for dealing with recurrent situations faced by agents . . . That is, institutions are regularities in the interaction between agents that arise because of uncertainty in deciphering the complex interdependences created by these interactions " ( Heiner ( 1983, p.573 )).

In a nutshell, either institutions can be TC optimal and flexible to human designs, or be habitual and stable, with power-relations as constraints to human actions. All the economists, save Williamson, in this section have seen the importance of institutions as regular patterned behaviour. But Hodgson ( 1991 ), Dixon ( 1986 ), Knight and Merriam ( 1947 ) and Heiner ( 1983 ) go further in stating that only " habit institutions " can function since they allow expectations to be formed. Optimal institutions cannot possibly exist. Furthermore, their explanations for the functions of institutions ( i.e. habit forming, power-invoking, and uncertainty-overcoming ) differ from Williamson's narrow TC concern. Rutherford ( 1989 ) and Langlois ( 1989 ) further affirm the importance of " institutional individualism " as opposed to TCE's " methodological individualism ". " Rationality " that embodies instrumentalism, enculturation, institutionalisation, and human needs has greater relevance than universal, value-free, and human needs conception of rationality.

Notes

1. Public goods are oft-cited examples of NC's " exceptional cases ", e.g. defence, lighting, education

2. If agents are to be " rational ", then they must believe in the Judeo-Christian God to ensure that they are not sleeping and not being deceived by the Judeo-Christian Devil ( Descartes ( 1986, pp.12-15 )). Or agents can be institutionalised into thinking that their premises of really being able to think are true. But once agents are socialised into having a particular conception of reason, this destroys NC's " methodological individualism " __ see section III ( b ). Furthermore the inference from the works of Kurt Godel and Alan Turing is that we can never prove that we are consistent, because logical system can not prove its own consistency: reason is based on faith. We can infer from Turing that even if we were computers we would be unable to solve every problem objectively. Hence the NC metaphor that agents are mechanical calculators is doomed in that there will be certain problems that they will be unable to solve.

3. In attacking NC's conception of market operations, Hodgson ( 1988b ) notes that G. B. Richardson had argued that " if NC perfect competition did actually exist it could not function for long. The problem being that no individual agent would be aware of the investment intentions of others. The incentive to invest depends in part on the knowledge of a limited competitive supply from other firms, or the establishment of a belief that others do not possess the information regarding the opportunity that is available to the investor. " ( pp. 188-9 ).

4. Williamson hi-jacked the term from Commons. But Commons writes, " Transactions . . . are not the ' exchange of commodities ' in the physical sense of ' delivery ', they are the alienation and acquisition, between individuals, of the rights of future ownership of physical things, as determined by the collective working rules of society. The transfer of these rights must therefore be negotiated between the parties concerned, according to the working rules of society, before labour can produce, or consumers can consume, or commodities be physically delivered to other performed " ( Commons ( 1934, p.58 )). But Williamson fails to discuss how the process of law-making is determined; mistakenly he takes " rights " as given and knowable to all, but this is against the spirit of Commons's own institutional economics. Furthermore Williamson ignores a major part of Commons's work on transaction costs: " [i]f institutional economics is . . . volitional economics, it requires a volitional psychology to accompany it. This is the psychology of transactions, which we may properly name transactional or negotiational psychology . . . The psychology of transactions is the social psychology of negotiations and the transfers of ownership " ( Commons ( 1934, pp.90-1 )). Yet Williamson makes no reference to negotiational psychology, which is most surprising. The neglect of Commons's law-making process and negotiational psychology is a set-back to the understanding of how institutions shape transactions and vice versa. Commons's text is much richer as he deconstructs the term " transaction " into several types: " strategic transaction " ( volitional side of humans ); " routine transaction " ( taken-for-granted-nature-of-reality side of humans ); " bargaining transaction " ( preceded exchanges of wealth and established the terms of those exchanges ); " managerial transaction " ( managers give orders which workers obey ); and " rational transaction " ( legal and quasi-legal rules constructed within which managerial and bargaining transactions take place ). Commons attaches much importance to negotiational psychology as a way to understanding and influencing economic activities because negotiational psychology is the mental processes and associated activities of people which determine transactions ( Biddle ( 1990, pp.3-9 )). Williamson also misuses another hi-jacked term " bounded rationality " from Simon __ see section III ( a ).

5. Williamson in his 1986a paper, puts forwards " dignity " as another of his behavioural assumptions, arguing that " analysing the economising of transaction costs without regard to dignity encourages the view that individuals can be considered strictly as instruments. Such an approach may be excusable in studying the governance process in capital markets and in any intermediate-product markets. But sensitivity to human needs for self- and social-esteem becomes important when the organisation of work ( labour markets ) comes under scrutiny " ( p.177 ). But this rhetoric towards " workers' management " is unexplored and seems only " ad hoc qualifier " to his main thesis ( only bounded rationality and opportunism craft governance structures ), in case the latter does not hold.

6. It is interesting to note that Williamson uses the word " economise ", though there is nothing in his previous analysis to suggest that agents wish or able to " economise ". Hodgson ( 1991 ) picks out this point ( see section III ( a )). I would have thought that the word " satisfy " is in keeping with his previous analysis of bounded rationality.

7. A weakness of TCE is that Williamson does not provide a satisfactory measurement technique to obtain exact values of " costs " and " benefits ". This has to be contrasted with the NC's narrower approach. Thus how am I to " economise transaction costs and NC production costs ", if I don't know how to measure the former ? Williamson in his 1980 paper tries to measure transaction costs of alternative governance structures, only to be rebutted by Jones ( 1982 ) __ see section III ( f ).

8. Williamson believes that governance structures come about not only by human action but also by human design, reflecting optimal and rational decision-making. This comes into conflict with many institutional economists who argue that institutions arise from human actions, but not from human designs, and that optimality of the institution is impossible given real bounded rationality and habitual decision-making process __ see section III ( b ) and ( c ).

9. A comparative study implies that like is compared with like, but different social embeddedness of institutions rules out such a comparative study which assumes social embeddedness to be exogenous and homogeneous to the institutions __ see section III ( e ).

10. It seems that governance structures are trying to " economise " on two levels, which are not compatible: ( i ) " economise on the sum of production and transaction costs " ( Williamson ( 1986b, pp.110-11 )), ( ii ) economise only on transaction costs ( Williamson ( 1985, p.32 )). Yet alternative governance structures cannot economise on both levels if production costs vary for " the related set of transactions ". Williamson does not state which " economising " takes precedence, when a trade-off exists between production and transaction costs, and when production costs significantly vary for different governance structures. Williamson, in his 1980 paper, ignores production costs when he assesses alternative governance structures' relative efficiency __ see section III ( f ).

11. Williamson writes, " Much of the legal literature . . . assumes that the appropriate legal rules are in place and that the courts are the forum to which to present and resolve contract disputes " ( Williamson ( 1989, p.140 ). He dismisses such " legal centralism or court ordering " in favour of " private ordering ", where incomplete contracting / ex-post governance structures use contract as a framework, which is highly adjustable so as to ensure a continuous relationship between the parties. " Neo-classical contracting " and " relational contracting " are two modes of contracting which embody " private ordering ". They differ only in the respect that the latter allows for the whole relationship to be flexible, whereas the former the reference-point for affecting adaptions remains the original agreement. In contrast, " classical contracting, which " embodies " legal centralism ", allows no adaptions of the original contract.

12. Rhetorically, Williamson states, " Care must be exercised, however, lest problems be construed too narrowly. This will occur if net benefits are calculated in transaction-specific terms, when in fact there are interaction effects [ i.e. quasimoral involvement among the parties, the sociopolitical phenomenon of reciprocity, the nonpecuniary satisfaction of mode of governance ] to be taken into account . . . [F]ailure to allow for such interaction effects, when in fact they exist, leads to suboptimisation " ( Williamson ( 1975, p.37 )). He notes that a transaction takes place within a " trading atmosphere " ( i.e. " interaction effects " ). Yet, Williamson does not incorporate this explicitly into his analysis of governance structures, except as an " ad hoc qualifier " in case his main thesis ( that only asset-specific, uncertainty and frequency affect the design of governance structures ) fails.

13. It is noteworthy that " by fiat " internal organisation can overcome many of the problems that face market governance, arising from opportunism, bounded rationality and uncertainty. Yet, Williamson fails to expand how and why this mechanism can command such an absolute respect and obedience, given that he dismisses " obedience " as unrealistic form of self-interest ( Williamson ( 1985, pp.49-50 )) __ see section III ( e ).

14. Though in the world of abstractions, other things can be forced to be stable and unchanging, by exogenising the variables, in real history these variables may not or cannot be seen to be constant. Thus, the onus is on Williamson to prove otherwise if he is to insist that TCE explains the emergence of hierarchy " in history ". Alas, Williamson fails to achieve this as Jones ( 1982 ) points out __ see section III ( f ).

15. Bhaduri points to a different kind of efficiency to that of NC allocative efficiency and TC optimality : " exchange relations may be " efficient " from the point of view of the surplus-extracting class, quite irrespective of whether or not they satisfy the NC postulates of allocative efficiency " ( Bhaduri ( 1983, p.268 )). For Bhaduri, economic institutions and arrangements arise in order to allow the surplus-extracting class to have power over the surplus produced.

Furthermore like the NC's message of the " value-free ", objective efficacy of the market, TCE's message of the " value-free ", objective efficacy of the hierarchy tries to delink socio-political ideology with economic co-ordination in the hope of evading public policy solutions ( Williamson ( 1975, p.253 )). But Dugger ( 1983 ) rightly accuses TCE of being a " new apologia for the corporate status quo ", because Williamson would leave the system alone because he believes it to be TC optimal when seen in its entirety ( e.g. Williamson ( 1985, pp.286-94; 1986b, pp. 236-7; 1989, pp.148-9 )). But value-free study is impossible to achieve because perceptions and interpretations are value-ladened. Dugger makes a more telling point, " [W]hen analysts venture into relevant policy, even ' positivist economists become practicing normativists, that is, they are engaged in deciding what proper policy ought to be . . . And like the other self-proclaimed value-free positivists who venture into policy, Williamson takes on a status quo bias immediately in his analysis " ( Dugger ( 1983, pp.110-11 )). Further Williamson's posturing of value-free, positivist analysis is thrown into doubt when it is realised that transaction costs cannot be measured, and when he deliberately does not analyse state-run hierarchies.

16. Jones ( 1982 ) argues that during the C19th period of Industrial Revolution, the world was far from a " steady-state ", as technology was rapidly changing __ see footnote 14.

17. Williamson writes, " . . . most of the assignment are transparent or are evident from the discussions . . . " ( Williamson ( 1980, p.24 )). Bharadwaj argues, with some justification, that evoking transaction costs are " ex-post rationalisation of observed situations " ( Bharadwaj ( 1989, p.18 )). Jones ( 1982 ) similarly takes up this theme when he accuses Williamson of being historically selective __ see section III ( f ). But Williamson ignores the dangers of " transaction costs functionalism " by seeing every institutions and arrangements as outcomes of transaction costs considerations.

Furthermore, Williamson reaches his conclusion about the efficiency of the authority hierarchy without taking into consideration production costs, contrary to his stated aim. Jones ( 1982 ) points out that if production costs were considered, the authority hierarchy would become an inefficient mode of contracting.

18. Williamson provides another " ad hoc qualifier " to his main thesis about the importance of transaction costs, when he argues that better job satisfaction and less alienation would not be sacrificed to achieve greater efficiency by some workers. Hence history is littered with less efficient modes of contracting coexisting with efficient modes. " Confronted . . . with the need to make trade-offs, valued attributes will be adjusted at the margin " ( Williamson ( 1980, p.34 )). Note that transacting parties determine / " design " the governance structure and it is always for " the best ", since they make appropriate trade-offs reflecting " their preferences " for higher TC efficiency and less satisfaction, or vice versa.

19. Williamson notes that a superior governance structure may not arise peacefully. He gives the example of Andrew Carnegie's and Henry Clay Frick's resolution to challenge the Amalgamated Association of Iron, Steel and Tin Workers in order to remove " costly haggling and inflexibility " of the existing organisation of work, which was " inside contracting ". After locking-out the mill " violence resulted, with members of the union pitted against scabs and Pinkerton agents. The support of state and federal governments helped Carnegie and Frick to prevail " ( Williamson ( 1980, p.31-2 )). Having broken the union's power, new institutional structures were devised, which realised the benefits of the new organisational changes. With this statement, Williamson writes that only through " mutual agreement " do superior structures evolve ( Williamson ( 1986a, p.195-6 )). But clearly in this example ( and many others ), one transacting party had to suffer " adverse oppression ", which only begs the question: " For whom were the resulting benefits of the intentionally devised " superior " structure ? " Here Williamson has unwittedly trapped himself into political economy. Gone is his rhetoric of " value-free " scientific study ( Williamson ( 1975, p.253; 1986b, p.187-8 )). This leads me into accepting Bhaduri's ( 1983 ) conception of " class efficiency ", and Bartlett's ( 1989 ) idea of the importance of power in establishing and sustaining interest groups' rights and values before transacting __ see section III ( d ). Williamson ignores the field of " law-making, right-making, rule-making, and value-making processes ", which is surprising given his focus on " transaction " as a legal concept, and his rejection of the law-less concept of " exchange ".

20. Williamson may think that he can save his conception of TC optimality by evoking a darwinian type of selection theory of institutions. The selection theory states that whether institutions are consciously designed or accidently invented, the selection mechanism will operate to ensure that institutions are TC minimising. But this fails to help Williamson, because he provides no " feedback mechanism " that sustains institutions' quest for efficiency. Further, the selection theory does not provide an explanation of the origin of the institutions. Furthermore, Williamson uses " the selection theory " to justify not paying detailed attention to the real processes involved, as the " minimisation metaphor " is an extremely appealing and powerful positive heuristic device to ignore the issue. Yet Langlois ( 1989 ) argues that many processes leading to the emergence of institutions are not based on the selection for the most efficient institutions, and even if a process does select for efficient institutions, the selection mechanism is presupposed to operate extremely swiftly and efficiently, without ex-post errors. ( Dawkins ( 1982 ) is aware of the naivety of this assumption in zoology ) __ see section III ( c ). Moreover, how do social and cultural beings choose institutions ? e.g. how does " a white, liberal, Christian male " choose between a Christian family unit and a non-Christian family unit ?

21. The chess computer playing an international grand-master is only as good as the computer programmer who feeds it with instructions and procedures.

22. Hodgson ( 1988a ) points out a danger of " informational reductionism ": those economists who emphasise the importance of information ignore the cognitive processes by which information is selected, arranged and perceived by agents. Each cultural and social beings ( via socialisation and experience ) have their own unique mental schema by which to process information and hence decide and act. Yet, institutions exist to ensure sufficient commonly shared schema for people to sufficiently anticipate each others' thoughts and actions, and so enable the economy to function sufficiently.

23. Of course this has implications for the political aspects of the economy. Socio-cultural institutions become important for the schooling they provide for individuals. " Thought police " must not be regarded with horror by white liberals if " sophisticated " organisation of activities are to be realised. Rather " thought police " must be seen as a mechanism towards creating a society based on " politically correct " opinions, where " we can all get along ".