Annotated Bibliography

THEIMPORTANCEOFBEINGTRUSTED-TRANSACTIONCOSTSANDPOLICYNETWORKTHEORY.pdf

Home >Information Systems homework help > Annotated Bibliography

THE IMPORTANCE OF BEING TRUSTED: TRANSACTION COSTS AND POLICY NETWORK THEORY

ANDREW HINDMOOR

Transaction cost economics is used to defend and develop policy network theory. Networks, like markets and hierarchies, are a form of governance structure. As one form of network, policy communities develop and survive because they reduce transaction costs that would otherwise threaten the exchange of resources between government and pressure groups. Policy communities alter outcomes and should be an important part of the explanation of public policy. Whilst of general relevance, the argument is illustrated with particular reference to the frequently hostile but ultimately successful negotiations between the British Medical Association and the Ministry of Health prior to the creation of the National Health Service in 1948.

I INTRODUCTION Transaction cost analysis is one of the more prominent strands of the 'new institutionalism' that has swept through the social sciences (Lowndes 1996). Whether the theory is of the firm (Williamson 1975), bureaucracy (Horn 1995), state (North 1990) or legislature (Weingest and Marshall 1988), the transaction cost story is that institutions evolve to economize on transaction costs and that different levels of transaction costs (predictably) lead to the development of different types of institutions or, in Oliver Williamson's (1985) phrase, governance structures. The usual contrast upon which the theory rests is that between markets and hierarchies. In this article, net- works are seen as a third and alternative form of governance structure and the emergence of policy communities as a particular and important form of network is explained using transaction cost theory. In the market, exchange is mediated through and made possible by the use of legally bind- ing contracts. In hierarchies, an equivalent role is played by authority. In a policy community, exchange is possible because actors trust each other. To understand how and why policy communities develop, it is necessary to understand how and why trust emerges.

The negotiations between the British Medical Association (BMA) and

Andrew Hindmoor is a Lecturer in the Department of Politics at Exeter University. He would like to thank Keith Dowding, Oliver James, Peter John and Stephen Wilks for their helpful comments on an earlier version of this article.

Public Administration Vol. 76 Spring 1998 (25^3) © Blackwell Publishers Ltd. 1998, 108 Cowley Road, Oxford OX4 lJF, UK and 350 Main Street, Maiden, MA 02148, USA.

26 ANDREW HINDMOOR

Ministry of Health (MoH) prior to the creation of the National Health Ser- vice (NHS) in July 1948 are used to illustrate the argument. The example may seem a curious one for, as Martin Smith (1993, p. 169) observes, a closed policy community did not develop until after the NHS had been formed. Indeed during this period, relations far from being consensual were frequently hostile. At various points, the Minister of Health, Aneurin Bevan, accused the BMA of being bigoted and the BMA the Minister of being a dictator (Webster 1988, p. 110). The example is nonetheless an interesting one because an agreement between the BMA and the MoH was eventually reached and reached without the use of either legally binding contracts or the exercise of authority. This article shows how trust sufficient to make agreement possible developed.

The argument is in three parts. First, it is shown why mutually beneficial exchange between governmental and non-governmental actors is possible. Second, it is argued not only that high transaction costs threaten exchange but that neither markets nor hierarchies can adequately economize on these costs. Third, it is suggested that in these circumstances trust facilitates exchange and that the development of trust can be accounted for in terms of the structural setting of relations between actors. As the more general conclusion to the article emphasizes, it is my intention in using transaction cost analysis not to bury but largely to praise policy network theory. Net- works affect outcomes and are not, as Keith Dowding (1995) suggests, sim- ply a metaphor.

II THE POSSIBILITY OF CO-OPERATION

The story of the creation of the NHS is one that has been told many times (Eckstein 1960; Willcocks 1967; Honigsbaum 1989). The decision to build a national health service at the end of the war was first taken by the Minister of Health, Ernest Brown, in October 1941. Following the publication of the Beveridge report in 1942, firm proposals were published in March 1943. After objections were raised by the BMA, a new Minister, Henry Willnick, unveiled a revised plan in March 1945. Finally and following the election of a Labour government, the National Health Service Bill was given its first reading in March 1946. But beyond providing for free and comprehensive medical coverage, the Bill itself did not specify in any detail the administrat- ive structure of the new service (Eckstein 1960, p. 111). Between March 1946 and the inception of the service in July 1948 it was over this terrain that the BMA and fhe MoH fought.

Whilst accepting in principle the need for a national service, the BMA was steadfastly opposed to the creation of a centralized service in which it feared the doctors' role would become equivalent to that of civil servants. Specifically, the BMA sought to affirm and retain (i) the right of general practitioners to buy and sell practices, (ii) the special status of teaching hospitals, (iii) the independent contract system by which doctors were paid according to the number of patients registered with them (capitation) rather

) Blackwell Publishers Ltd. 1998

TRANSACTION COSTS AND POLICY NETWORK THEORY 27

than by fixed salary, (iv) the right of doctors to treat private patients and, (v) the freedom of doctors to practise where they wanted.

The BMA realized that Bevan was unsympathetic to their demands but knew that they were in a strong bargaining position as the MoH needed them to assist in the implementation of policy. The BMA was not the only organization whose interests were threatened by the decision to build a NHS. Insurance companies and the boards of voluntary hospitals also sought to influence policy. Attention is lavished upon the BMA because it was the one voice that government could not ignore. The government may have wanted but it did not need the support of the voluntary hospitals because the hospitals were to be nationalized and it did not need the sup- port of the insurance companies because health care was to be funded from general taxation. If its policy objectives were to be realized, the government did need the support of the doctors (Smith 1993, p. 165). As Charles Webster (1988, p. 107) argues, 'the primary difficulty for the government was per- suading a BMA which was deeply disenchanted with the NHS legislation to collaborate amicably in framing regulations and setting up the adminis- tration of the new service.'

What could the BMA do if government refused to listen? Most obviously, it could simply threaten to boycott the new service and in 1946 the Chair- man of the BMA Council, Dr. Guy Dain, spoke frequently about the possi- bility of a strike. Indeed even before the Bill had been published, the BMA announced the launch of an emergency fund to support striking doctors (BMJ 1946a, p. 358). Alternatively, the BMA could participate in the service but withhold from it what Oliver Williamson (1975, p. 69) calls consummate co-operation: the use of its initiative, judgement and enthusiasm in implementing policy. As an editorial in the British Medical Journal (BMJ 1946b, p. 240) reminded its readers in February 1946 'he [Bevan] himself said that no health service can be run properly except with the whole-hearted co-operation of the doctors of this country . . . legislators can legislate, but only doctors can provide the service' (emphasis added).

Between 1946 and 1948 the BMA and the MoH were forced to bargain with each other to achieve their respective objectives and the nature of this bargaining process is illustrated in figure 1. Here, the horizontal axis (A- C) shows disagreements between the BMA and the MoH over the range of policy issues previously identified. A represents the preferences of the BMA and C the corresponding and opposed preferences of the MoH. For example, A is a service in which doctors are paid by capitation, C one in which doctors are paid by a fixed salary and B by some combination of the two.' The vertical axis (X-Z) shows possible forms of implementation. Z reflects the preferences of the MoH for a service in which doctors co-operate in the implementation of policy. At X, doctors withdraw this co-operation and the burden of implementing policy is carried by the MoH.

Policy outcomes follow from the choice of both policy framework and form of implementation. The MoH most prefers an outcome (M) in which

28 ANDREW HINDMOOR

Policy Framework

Implementation

A B C

Policy Framework

FIGURE 1 The BMA, the MoH and Pareto efficient exchange

doctors co-operate in the implementation of a policy that reflects govern- ment objectives. The BMA most prefers an outcome (O) in which govern- ment is responsible for implementing a policy that reflects doctors' objec- tives. Three indifference curves are shown for government (G) and doctors (D) and ranked such that for government G3 > G2 > Gl and for doctors D3 > D2 > Dl. Each indifference curve joins points of equal total utility, the slope of each representing the required rate of substitution necessary to maintain a particular level of utility. Convexity reflects the standard assumption of diminishing marginal returns and means, for example, that at a point (S) where doctors are already largely responsible for implemen- tation that they will demand significant policy concessions before agreeing to do more.

However abstract the concept of indifference curves, the diagram is use- ful because it reveals the potential - often realized - for conflict between the BMA and the MoH. What outcome to the bargaining process does it predict? To know this it is first necessary to identify the non-agreement point: the outcome that will occur in the absence of compromise. Because

) Blackwell Publishers Ltd. 1998

TRANSACTION COSTS AND POLICY NETWORK THEORY 29

a British government with a sufficient majority can always, as the pro- fession itself recognized (BMJ 1946c, p. 240), pass the legislation it wants and because doctors can always refuse to assist in the implementation of that legislation, N is the non-agreement point. Neither actor will want or need to accept a worse outcome. Compromise and negotiation are likely because this outcome is not pareto-optimal. If the MoH makes policy con- cessions in return for the BMA's assistance in implementing policy, both can be made better off.

Any outcome within the area NQPR is pareto-superior to N: moving doctors away from D2 and toward D3 (D3 > D2) and government away from G2 and towards G3 (G3 > G2). Of these, those along the line QR are pareto-optimal. It is not claimed that such an agreement will necessarily benefit the patients but at these points it is impossible to make the BMA or the MoH better off without making the other worse off. The BMA was not unaware of the possibility of such a compromise. In a public address in 1946, Dain was able to express:

great hopes that we may secure modification of the Bill very materially. After all, it is up to the Minister to provide a service for the country, and he cannot provide a service for 100 per cent of the people unless he has 100 per cent of the doctors. That fact alone may lead to the consider- ation of the sort of modifications which the profession would accept (BMJ 1946d, p. 536).

The focus in this section has been upon the relationship between the BMA and the MoH but the argument is intended to be of general relevance. As Grant Jordan and Jeremy Richardson (1979, p. 172) recognize, 'public policies are the outcome of a process of adjustment between organisations . . . pressure groups and government have come to recognise that they need each other in order to achieve their respective objectives'. When a particular pressure group is opposed to a policy and can credibly threaten to with- draw its co-operation, compromise and negotiation will take place because compromise and negotiation are mutually beneficial. Government may not want to concede ground but as Dave Marsh and Rod Rhodes (1992b, p. 181) show in the case of Thatcherism, 'rejection of consultation and negotiation almost inevitably led to implementation problems, because those groups/agencies affected by the policy, and who were not consulted, failed to cooperate, or comply with, the administration of the policy.'

Ill TRANSACTION COSTS A N D THE THREAT TO CO-OPERATION

As might be expected, the eventual and agreed structure of the NHS was a compromise (BMJ 1948a, p. 937). The BMA did not get all it wanted. Hos- pitals were still nationalized, the sale of practices abolished and a baseline salary for new doctors introduced. But in return for Dain's promise that 'the profession will do its utmost to make the new service a resounding success' (Webster 1988, p. 127), the MoH did make significant policy con-

30 ANDREW HINDMOOR

cessions. Doctors' right to treat private patients was acknow^ledged, the basic structure of the independent contractor system retained, a system of distinction awards for hospital consultants introduced and tbe central direc- tion of medical labour largely forsworn. Of subsequent interest in this arti- cle is not tbe fact that a mutually beneficial exchange took place but tbe method by which it was achieved.

Exchange always takes place within the context of a particular govern- ance structure, defined by Williamson (1986, p. 155) as 'the institutional setting within which the execution of transactions is completed'. Two forms of governance structure are usually identified: the market within which exchange is made possible by the use of contracts and hierarchy within which exchange is made possible by the use of authority. We will consider first the use of contracts. In a contract, both parties specify obligations to each other that can be enforced, if necessary, through the intervention of the courts. Could the BMA and the MoH have used a contract to secure agreement: the MoH agreeing to make X policy concessions in return for Y assistance in the implementation of policy? Given that no such contract was used, the question may appear strange but as Ronald Coase (1937) was the first to observe, in a world of zero transaction costs such an arrangement would be entirely feasible. In the 'real world' of which Williamson (1995, p. 27) is fond of talking, transaction costs - defined as the costs of exchang- ing rather than producing goods and services - are rarely zero and for this reason markets are not always efficient and for this reason are not always employed by actors to facilitate exchange.^ Four different sources of trans- action costs threatening contractual exchange between the BMA and the MoH can be identified: complexity, power asymmetries, information asym- metries and thinness. Whilst attention is again focused upon relations between the BMA and the MoH, the argument is intended to be of general relevance as these transaction costs are a pervasive feature of relations between government and pressure groups.

Complexity A proposed exchange is more complex the larger the number of contin- gencies that have to be considered ex ante by both parties before being able to specify what will ex poste constitute satisfactory performance of an agreement. Because negotiation is costly in terms of botb time taken and legal fees spent and because an incompletely specified contract will leave one or both parties vulnerable to exploitation by the other, the use of con- tracts is most likely when complexity is minimal. For two reasons, any con- tract between the BMA and the MoH would have been extremely complex.

First, the number of contingencies was simply too large. As Rudolf Klein (1995, p. 1) emphasizes, the NHS was a significant policy innovation; one that required the nationalization of over 1,500 hospitals, the employment of over 500,000 people and the extension of general practitioner care to the entire population. Given inevitable limits upon individual foresight, neither

TRANSACTION COSTS AND POLICY NETWORK THEORY 31

the BMA nor the MoH could have known what would and would not work in the new service. No contract could have been written prior to tbe estab- lishment of the NHS because no one could have known with sufficient precision what the contract would need to contain. Second, and as already noted, the MoH sought from the BMA their 'consummate co-operation': the use of judgement, enthusiasm and initiative. As Williamson (1975, p. 79) observes, such qualities are extremely difficult to define in contractual terms. How, for example, could the MoH have shown to the satisfaction of a court that the BMA had not shown 'enthusiasm' for a new policy?

Power asymmetries Contracts are meant to eliminate the possibility of exploitation by commit- ting actors to particular and predetermined courses of action. If one party reneges upon an agreement, tbe other can seek to enforce compliance or achieve compensation through the courts. But government cannot always be constrained in this way because uniquely it can overturn or ignore judge- ments against it (Eggertsson 1990, pp. 317-57). In the British system, the very claim to parliamentary sovereignty and the absence of substantial checks and balances undermines tbe government's contractual credibility.

Consider the BMA's already noted opposition to a full-time salaried ser- vice. Whilst during negotiations Bevan assured the BMA that he would not introduce such a measure he also told the House of Commons that 'there is all the difference in tbe world between plucking fruit when it is ripe and plucking fruit when it is green' (BMJ 1948b, p. 64). As part of tbe eventual agreement reached in 1948, the BMA forced Bevan to introduce amending legislation making it impossible for a minister to introduce a salaried ser- vice without further legislation. But as the BMA must have been aware, this was far from making the introduction of a salaried service itself impossible. However carefully drafted, no contract could have achieved this objective.

Information asymmetries Having signed a contract, government will want to know that doctors are implementing policies in intended ways. Information asymmetries occur and complicate exchange when the underlying circumstances relevant to a trade are known by one or more but not all the parties to that exchange. A familiar example is that of the market for second-hand cars where unless they are a mechanic the potential buyer will fear that they are being sold a 'lemon' by an unscrupulous dealer (Akerlof 1970). In our case, the source of the information asymmetry is government's lack of specialized medical knowledge and physical distance from the 'street level bureaucrats' (Lipsky 1980) implementing policy. This asymmetry can be reduced if government employs additional agents to undertake ex poste monitoring but such a con- tractual safeguard itself represents a significant transaction cost that will reduce the benefits of exchange.

32 ANDREW HINDMOOR

Thinness A transaction is thin the smaller the number of trading partners an actor can deal with to achieve their desired objectives. To this extent, thinness is simply a measure of monopoly; the twist added by transaction cost theorists being the observation that the very consummation of an exchange can leave one or both actors more reliant on the other. One anticipated consequence of creating the NHS was that private medicine would largely disappear as middle-class patients registered themselves with state doctors. As such, the creation of the NHS was expected to increase the doctors' reliance upon the government for employment. On the other side of this equation and as Martin Smith (1993, p. 175) suggests, by eventually creating a health service in which much of the responsibility for implementing policies was del- egated, tbe government became more reliant upon the doctors. This would not have posed any difficulty if the terms of any contractual agreement between the BMA and the MoH could have been completely and indefi- nitely specified but as already noted such a solution was not available.

With respect to the MoH's reliance upon the doctors, how will thinness influence the prospects of reaching an agreement? The answer suggested by Gary Miller's (1992, pp. 110-15) equivalent analysis of labour relations within the firm is that the BMA will avoid delegating implementation in the short term for fear of being exploited in tbe long term. The reasoning behind this conclusion is shown in figure 2. At time (T)l, the MoH has the choice of whether to delegate implementation under the terms of a contract so complex that it can last only until T2. If implementation is not delegated the outcome is C. At T2 and if implementation has been delegated, tbe MoH's dependency upon the BMA will have been increased. The BMA then has the choice of whether to exploit this advantage or remain true to the spirit of the agreement signed at Tl. If it chooses to honour the agreement, the outcome is A and if it chooses not to honour the agreement, the outcome is B. For the MoH B > C > A and for the BMA A > B > C. The dilemma is that whilst the government knows that if it delegates implementation doctors will have every reason to exploit their position (A > B) and that it should therefore not delegate implementation (C > A), this outcome is par- eto-inefficient as for both actors B > C.

When transaction costs are high, transaction cost theorists argue that hier- archy characterized by the use of autbority will be more efficient than mar- kets and be cbosen by rational actors for this reason. Authority is present when 'some form of subordination prevails' (Williamson 1975, p. ix): that is when one actor can order another to bebave in a particular way. Tbe most obvious example of a hierarchical relationship is that found between an employee and an employer. Here, repeated and detailed contract negoti- ations are not necessary as the employee simply agrees to do whatever their employer tells them to. The distinction Williamson draws between markets and hierarchy appears exaggerated as in botb an important role is played by a third party, the courts. The employee agrees to work for an employer

O Blackwell Publishers Ltd. 1998

TRANSACTION COSTS AND POLICY NETWORK THEORY 33

honour agreement A

BMA Delegate implementation

MoH

not honour agreement B

Not delegate implementation

Time 1 Time 2

MoH - A> C> B BMA-B > A > C

FIGURE 2 Co-operation and exploitation

knowing that if the employer fails to pay the agreed wage payment can be legally enforced. The difference between markets and hierarchy appears instead to lie largely in the nature of the decision-making process. In a market, decision-making is decentralized and obligations are precisely defined. In a hierarchy, decision-making powers are given to a limited num- ber of individuals and discretion exercised by them in specifying obli- gations.

Will hierarchy facilitate exchange between government and pressure groups? The simple answer is that it will not. With reference to figure 1, at the start of negotiations, the BMA will be unwilling to concede hierarchical authority to the MoH because they will know that the MoH will use this authority to achieve their most preferred and the doctor's least preferred outcome, M. Equally, the MoH will be unwilling to concede hierarchical authority to the BMA because tbey know that the BMA will use this auth- ority to achieve their most preferred outcome, O. In the case of the employee-employer relationship, hierarchy is attractive to both parties because it is assumed that the employee has no particular preference over

34 ANDREW HINDMOOR

the nature of the tasks they are called upon to perform. The BMA cannot remain indifferent to the direction in which authority is exercised as it is precisely this that they seek to influence.

IV NETWORKS AS A GOVERNANCE STRUCTURE

Complexity, thinness and power and information asymmetries give one actor the opportunity to exploit another. Exchange is threatened because neither actor trusts the other not to exploit this opportunity. It should be noted tbat if the BMA trusted the MoH to act fairly and that if the MoH repaid this trust then the existence of, for example, a power asymmetry would not in itself threaten exchange. Ultimately, it is the lack of trust that threatens exchange. Given Williamson's (1985, pp. 47-9) assumption that actors will always act opportunistically, trust in this context should not be equated witb blind faith. An actor will trust another only if they calculate that the likely benefits of doing so outweigh the likely costs (Williamson 1993). Actors will trust only if they have reason to trust.

Markets and hierarchies generate trust by providing institutional safe- guards. The employee is prepared to work for the employer because they trust that they will be paid and they place their trust not in the personal integrity of the employer but in the efficacy of the legal system that will enforce their agreement. Neither markets nor hierarchies are able to provide safeguards sufficient to generate trust between government and pressure groups. But as critics of Williamson's work frequently observe, networks offer an alternative form of governance structure (Thorelli 1990; Powell 1990; Thompson et al. 1991). In this section it will be argued that the defin- ing characteristic of a network is a trust that does not depend on the pres- ence of formal and exogenous safeguards.

However great the frequently observed disagreement between theorists as to how such basic terms as issue network and policy community should be defined, tbere is, within the British literature, a considerable and long- standing agreement as to bow policy communities as one particular form of network should be conceived. Rhodes (1988, pp. 42-3) distinguishes between different communities in terms of the 'operating codes', 'underly- ing philosophies' and 'rules of the game' that animate relations within them. Stephen Wilks and Maurice Wright's account (1987, p. 305) also emphasizes the importance of rules like informality and the avoidance of disputes that amount to an 'unwritten constitution' governing relationships. In a separate piece, Wright (1988, p. 599) argues that 'in order to try to understand . . . what actually happens in the policy process we have to try to identify those norms of behaviour . . . which influence and determine policy'. A corresponding set of arguments is to be found in Jordan and Richardson's Governing under Pressure in which the authors (1979, pp. 100- 1) seek to discover the 'operation understandings' which influence 'the pro- cess by which and the atmosphere within which . . . policy-making is resolved'. Again, much is made of the importance of rules which allow

TRANSACTION COSTS AND POLICY NETWORK THEORY 35

actors to achieve 'understandings which benefit all participants' (Jordan 1990a, p. 472).

That such rules and understandings exist and are respected clearly facili- tates exchange: they allow actors to eschew formal contracts and careful monitoring in favour of tacit agreements. Such rules are constitutive of a policy community, giving each actor information about how others can be expected to act and so allowing 'a specific item of business [to be] trans- acted in a context where participants already have mutual needs, expec- tations [and] experiences' (Jordan 1990b, p. 326). Because tbey are informal and unwritten, such rules cannot be enforced in the way a legally recog- nized contract can. Neither is compliance ensured by giving one actor hier- archical authority over the actions of another. Clearly the actors in a policy community must trust each other not to exploit their positions and trust each other in the absence of any external safeguards. It is trust that makes the emergence and survival of such rules possible. But why do actors not abuse and so eventually destroy this trust?

It cannot be argued that actors trust each other because tbey do better if they trust each other. Such an argument is correct insofar as trust does indeed make possible tbe pareto-superior move from M to QR in figure 1 but it is also incomplete. Functional explanations of tbis sort are only satis- factory if a feedback mechanism - such as the competitive market - can be identifled that eliminates actors who do not trust each other. Perhaps it could be argued tbat pressure groups that abuse the trust placed in them will not only fail to influence policy but for this reason will eventually lose members and disappear. This would, however, be a fragile foundation upon which to account for the existence of trust and for this reason, the concept must be given micro-foundations.

Actors in a policy community trust each other because their relations are historically, socially and personally embedded in an ongoing system that 'generates standards of behaviour that not only obviate the need for but are superior to pure autbority relations in discouraging malfeasance' (Granovetter 1985, p. 498). The concept of embeddedness which is usually found within the 'sociological' rather than 'economic' new institutionalism can be introduced here through the use of some simple game theory. In figure 3, the two actors, we will once again assume the BMA and the MoH, have the choice of whether to co-operate or defect. If an actor co-operates they will not only trust the other actor but will not abuse any trust placed in tbem. If an actor defects they will not trust the other actor and will abuse any trust placed in them. If both actors defect then no exchange will take place and with reference to figure 1, the outcome will be the non-agreement point of N. If both actors co-operate the eventual outcome will fall some- where along the line QR. If the BMA co-operates with the MoH but the MoH defects by, for example, reneging upon an agreement not to introduce fixed salaries, tbe eventual outcome will be M. If tbe MoH co-operates but the BMA defects by, for example, exploiting tbe government's dependency

36 ANDREW HINDMOOR

upon them once implementation is delegated, the eventual outcome will b e O .

Given the structure of preferences already established in figure 1, it can be seen that the game is equivalent to the prisoner's dilemma. In any single play, it is rational for an actor to defect regardless of whether the other co- operates even though this leads to a pareto-inferior outcome (QR > N). If, for example, the BMA co-operates, the MoH should defect (M > QR) and if the BMA defects the MoH should also defect (N > O). In these circum- stances, there is no reason for one actor to trust another. However, as is now equally well known, when (i) the number of actors is relatively small, (ii) contact with those outside the network is limited, (iii) there is frequent interaction between actors and the expectation of future interaction, and (iv) co-operation in one area can be made contingent upon co-operation in others, co-operation and trust can develop (Axelrod 1984; Taylor 1987; Dasgupta 1988; Coleman 1990; Kreps 1990; Chong 1991). In these cases, relations can be described as being embedded. By giving actors incentives to co-operate, embeddedness can generate a different outcome to the game: one of conditional co-operation. Embeddedness offers an internal solution to the prisoner's dilemma in which no third party is required to enforce co-operation.

In the conclusion to his history of the NHS, Brian Watkin (1978, p. 19) comments that it is difficult to understand why it took two years for the BMA and the MoH to reach an agreement as not only the need for but the eventual terms of compromise had long been recognized by botb parties. Certainly and immediately after the NHS Act had been passed, the Sec- retary of the BMA, Sir Charles Hill, let it be known that if the MoH made concessions on the introduction of a full-time salary that the BMA would accept abolition of the sale of practices. So why did it take until June 1948

]VfoH

Co-operate Defect

Co-operate BMA

Defect

BMA - O > QR > N > M MoH - M > QR > N > O

EIGURE 3 Co-operation and defection

TRANSACTION COSTS AND POLICY NETWORK THEORY 37

to reach essentially just such an agreement? The first part of the answer to this question is that neither contract nor authority could be used to facilitate exchange. The second part is that initially there was insufficient trust to make agreement possible in the absence of such formal safeguards. The BMA did not trust the MoH generally and Bevan in particular to protect their interests once the NHS had been formed. As an editorial in the Lancet (1948, p. 561) noted, 'to the majority of doctors, the service is objectionable far less for what it is than for what it might become'. Specifically, the BMA did not believe Bevan's assurances that a full-time salaried service would not eventually be introduced: assurances whose true value was felt to have been revealed in the Minister's comments about 'picking fruit' (Webster 1988, p. 103).

This Act is the first and substantial step towards a whole-time state medi- cal service. Even those who might be inclined to 'give it a chance' see the grave risks . . . once the profession crosses the Rubicon there will be no retreat... Mr Bevan will have the power by regulation to make medi- cal men whole-time servants of the state. Once in his service we shall be powerless to do anything more than haggle over terms and conditions (BMJ 1948c, p. 153).

Yet by June 1948 and on behalf of the BMA, Dr. Dain was ready to accept Bevan's promise that 'the profession shall always be consulted about regu- lations' and was telling a public meeting that the MoH 'has for thirty years always dealt with us [the BMA] fairly' (BMJ 1948d, p. 1086). Why were relations at one stage so poor and why did they subsequently improve?

Returning to the earlier list, the first two aspects of embeddedness are the number of actors in and the degree of closure of a network. Of interest here is Harry Eckstein's (1960, p. 104) argument that compromise became possible in 1948 not because of the immanence of the deadline for launching the NHS but because Bevan's own room for manoeuvre had increased. When legislation was being piloted through Cabinet and cabinet commit- tees, any agreement between the BMA and the MoH could have been over- turned by either the Treasury who were concerned about the financial implications of the service or Herbert Morrisson who was determined to ensure a greater role for local authorities in the administration of the ser- vice. Later and when subject to parliamentary debate, negotiations were threatened by Bevan's need to appease his own frequently militant back- benchers with a mixture of bellicose threats, denunciations and promises to concede no further ground. In this atmosphere, Eckstein argues, compro- mise became impossible. Relations only started to improve once the Bill had been passed and the MoH given sole responsibility for reaching an agreement: that is when the network was closed and the number of actors reduced. In this way, the very structure of the policy-making process con- tributed first to the erosion and then the rebuilding of trust.

For relations to become embedded and trust to develop, there must also

38 ANDREW HINDMOOR

be frequent interaction between actors. Here at least, Bevan's inheritance was a positive one as intense discussions between representatives of the BMA and the MoH had taken place throughout the war (Klein 1995, p. 10). Furthermore, whilst Bevan himself might have been an unknown quantity having never previously held ministerial office, the Permanent Secretary at the MoH, Sir Douglas William Scott, the Deputy Secretary, Sir John Wrig- ley, and the Chief Medical Officer, William Jameson, all had previous experience of negotiating with the BMA. Initially embedded relations were threatened only when Bevan, citing the requirements of parliamentary sov- ereignty, announced that he would permit comment upon his Bill but not a 'long series of protracted negotiations' (Webster 1988, p. 93). The British Medical Journal (1946c, p. 240) complained that the BMA was 'faced with a peacetime Minister who consults but does not negotiate or, in the ordinary sense of the word, discuss.'

Over the next year and as the Conservative Opposition never tired of observing, Bevan became, in the minds of the BMA's negotiating team, part of the problem and not part of the solution. Indeed it is an indication of how fragile relations became that when negotiations restarted a pivotal and largely covert role had to be played by intermediaries like the President of the Royal College of Physicians, Lord Moran, and the President of the Royal College of Surgeons, Sir Alfred Webb-Johnson. It was, for example, Moran who proposed that as a token of Bevan's good faith amending legislation be introduced to allay the BMA's fears that a full-time salaried service was to be introduced. In a mature policy community, discussion, consulation and negotiation are ongoing. Whilst relations between the BMA and the MoH were far from mature, they were helped by the common knowledge that if the NHS were established relations would not only continue but intensify. By locking the actors into a network, 'thinness' actually made co- operation more likely by turning a 'one-shot' game into an iterated relation- ship. Knowing that there would be a need to co-operate in the future, the incentives to co-operate in the present were increased.

Finally, and as Robert Keohane (1984, p. 129) argues in the case of inter- national regimes, co-operation between actors is easier to achieve if negoti- ations can be broken down into discussion of a series of issues. When negotiations are embedded within other negotiations, trust can develop because actors will realize that defection in any one area will lead to an unravelling of co-operation in others. Throughout the war, discussions between the BMA and the MoH had been conducted in this piecemeal way. Initially, Bevan adopted a similar strategy, linking discussion of the NHS to ongoing and otherwise separate pay negotiations. Relations deteriorated only when Bevan repudiated an earlier understanding, reiterated by Willnick in May 1945, that doctors would be able to buy and sell medical practices. As Klein (1995, p. 16) comments, 'compromises which might just [have been] acceptable [to the BMA] as part of an agreed package were no longer so once the package itself had been unwrapped by the Minister with-

TRANSACTION COSTS AND POLICY NETWORK THEORY 39

out consultation.' The BMA reacted - in Bevan's mind overreacted - by denouncing the entire Bill. When negotiations eventually restarted, compro- mises had to be painfully repackaged and trust rebuilt.

Measured purely in terms of the time taken, the transaction costs of the eventual agreement reached between the BMA and the MoH were high. Furthermore and on several occasions, negotiations came close to collaps- ing. But sufficient trust was eventually established to make agreement pos- sible without having to resort to the use of either markets or hierarchies. The network might have been fragile but it was still the governance struc- ture chosen to facilitate exchange.

V CONCLUSION: EXPLAINING WITH NETWORKS

After a series of case studies examining relations between government and pressure groups and in an effort to clarify discussion. Marsh and Rhodes (1992a, p. 251) construct a typology of pohcy communities and issue net- works the familiar contents of which are reproduced in table 1.

TABLE 1 Types of policy networks: policy communities and issue networks

Dimension

Number of participants Type of Interest

Frequency of interaction

Continuity

Consensus

Distribution of resources (in network)

Internal

Power

Policy community

Membership Very limited number, some groups consciously excluded Economic and/or professional interests dominate

Integration Frequent, high-quality, interaction of all groups on all matters related to policy issues Membership, values and outcomes persistent over time All participants share basic values and accept the legitimacy of the outcomes

Resources All participants have resources, basic relationship is an exchange relationship

Hierarchical; leaders can deliver members

There is a balance of power among members. Although one group may dominate, it must be a positive-sum game if community is to persist

Issue network

Large

Encompasses range of affected interests

Contacts fluctuate in frequency and intensity

Access fluctuates significantly

Some agreement exists, but conflict is ever present

Some participants may have resources, but they are limited, basic relationship consultative Varied, variable distribution and capacity to regulate members Unequal powers, reflecting unequal resources and unequal access - zero-sum game

Source: D. Marsh and R. Rhodes (eds.). Policy Networks in British Government (Oxford, Oxford University Press, 1992), p. 25.

40 ANDREW HINDMOOR

In a generally sceptical but very influential review of the network litera- ture, Dowding (1995) expresses particular reservations about the value of this account. First, it is argued that the labels policy community and issue network cannot explain outcomes as they fail to distinguish between depen- dent and independent variables. What is required is a 'proper model which causally relates the characteristics [in the typology] to each other' (1995, p. 141). Second, Dowding argues that any element of explanation in this account can be traced not to the characteristics of the network but to the characteristics of the actors within it. Only in formal sociological network analysis where 'networks are distinguished from one another by the relations between the actors rather than in terms of the attributes of the actors does the invocation of network explain anything' (1995, p. 152).

Dowding's criticisms provide a benchmark against which the arguments of this article can be judged. First, using the transaction cost perspective it has been possible to relate the elements of the Marsh and Rhodes typology to each other. An initial and necessary condition for the emergence of a policy community is that an actor has a resource that is valued by another. In the example considered, the BMA had the capacity to alter the costs of implementing policy and the MoH the capacity to alter the policy itself. For this reason, relations were potentially positive-sum. Such exchange relationships need not always develop.

If a pressure group is unable to significantly alter the costs of implementing a policy then relations are hkely to be zero-sum and the basic relationship one of consultation. Exchange of resources is threatened by high transaction costs inherent in the relationship between government and pressure groups. Exchange is made possible not by the use of contracts or authority but by the development of trust and trust is most likely to develop when relations are embedded: that is when the number of participants is small, some groups are consciously excluded and there is frequent and high-quality interaction. Once established, outcomes will be pareto-superior to those achieved in the absence of trust and for this reason are likely to be accepted as legitimate by both parties. In this way, the dimensions of (i) the number of participants, (ii) type of interest, (iii) frequency of interac- tion, (iv) consensus, (v) distribution of resources, and (vi) power in table 1 can be placed in a causal ordering.

Turning to Dowding's second argument, what is the explanandum in this account? Certainly and as already acknowledged, the attributes of the actors matter. A pressure group like the BMA can achieve a particular out- come because it has particular and valued resources. This is not only to be expected but to be welcomed for, as critics of formal sociological network theory argue, any explanation that depends entirely upon the nature of relations between actors is inadequate (Friedman and McAdam 1992). But in viewing networks generally and policy communities specifically as a form of governance structure, it is clear that the relations between actors also matter. For David Knoke and James Kuklinski (1982, p. 10), relations

> Blackwell Publishers U d . 1998

TRANSACTION COSTS AND POLICY NETWORK THEORY 41

refer to: 'actions or qualities that exist only if two or more actors are con- sidered together. A relation is not an intrinsic characteristic of either party taken in isolation, but is an emergent property of the connection or linkage between units of observation'.

The argument in this article is that it matters not whether relations between actors exist but whether they are embedded. Whether relations are embedded cannot be discerned with reference to only one actor. To say that relations are between a small number of actors and that they are fre- quent and of a high quality is to say something about the network and not the actors within it. The transaction cost framework therefore encourages explanation at both the 'micro' level of individual actors and the 'meso' level of relations between actors.

NOTES

1. The eventual structure of the NHS is a product of choices made over a series of individual issues, the more salient of which have already been noted. All this presentation demands is that the doctors - represented by the BMA - have preferences over all possible structures and that these preferences are transitive.

2. More formally, transaction costs can be defined as 'the costs of deciding, planning, arrang- ing and negotiating the action to be taken and the terms of exchange when two or more parties do business; the costs of changing plans, renegotiating terms, and resolving dis- putes as changing circumstances require; and the costs of ensuring that parties perform as agreed' (Milgrom and Roberts 1990, p. 60).

REFERENCES

Akerlof, G. 1970. 'The market for lemons: quantitative uncertainty and the market mechanism'. Quarterly

Journal of Economics 84, 388-400.

Axelrod, R. 1984. The evolution of cooperation. New York: Basic Books.

BMJ. 1946a. 'Preparing for the future', British Medical journat, 9 March, 358.

—. 1946b. T h e minister listens', British Medical Journal, 16 Feb., 240.

—. 1946c. 'Principle and unity', British Medical Journat, 2 Feb., 167.

—. 1946d. 'The profession and the bill: address by the chairman of the council', British Medical Journal,

6 April, 533-8.

—. 1948a. 'What we have gained', British Medical Journal, 15 May, 936-7.

—. 1948b. T h e profession and the minister: Dr. Hill's Address', British Medical Journat, 10 Jan., 61-5.

—. 1948c. 'Money and freedom', British Medical Journat, 24 Jan., 153-6.

—. 1948d. 'Safeguards secured', British Medical Journal, 5 June, 1086-7.

Chong, D. 1991. Collective action and the civil rights movement. Chicago: University of Chicago Press.

Coase, R. 1937. 'The nature of the firm', Economica, 4, 386-405, repr. in L. Putterman and R. Krozner (eds.)

1996. The economic nature of the firm. Cambridge: Cambridge University Press.

Coleman, J. 1990. Foundations of social theory. Cambridge, Mass: Harvard University Press.

Dasgupta, P. 1988. Trust as a commodity, in D. Gambetta (ed.). Trust: making and breaking cooperative relations.

Oxford: Blackwell.

Dowding, K. 1995. 'Model or metaphor: a critical review of the policy network approach'. Political Studies

43, 136-58.

Eckstein, H. 1960. Pressure group politics: the case of the British Medical Association. London: George Allen

and Unwin.

Eggertsson, T. 1990. Economic behaviour and institutions. Cambridge: Cambridge University Press.

42 ANDREW HINDMOOR

Foot, M. 1973. Aneurin Bevan, volume two: 1945-1960. London: Davis-Poynter.

Friedman, D. and D. McAdam. 1992. 'Collective identity and activism: networks, choices and the life of a

social movement', in A. Morris and C. Mueller (eds.). Frontiers in social movement theory. New Haven:

Yale University Press.

Granovetter, M. 1985. 'Economic action and social structure: the problem of embeddedness', American Jour- nal of Sociology 91, 481-510.

Honigsbaum, F. 1989. Health, happiness and security: the creation of the National Health Service. London: Rout- ledge.

Horn, M. 1995. The politicat economy of public administration. Cambridge: Cambridge University Press.

Jordan, G. 1990a. 'Policy community realism versus "new" institutionalist ambiguity'. Political Studies 38, 470-84.

—. 1990b. 'Sub-governments, policy communities and policy networks: refilling the old bottles?'. Journal of Theoretical Politics 2, 319-38.

Jordan, G. and J. Richardson. 1979. Governing under pressure: British politics in a post-partiamentary democracy. Oxford: Martin Robertson.

Keohane, R. 1984. After hegemony: co-operation and discord in the world political economy. Princeton: Princeton University Press.

Klein, R. 1995. The new politics of the National Health Service. London: Longman.

Knoke, D. and J. Kuklinski. 1992. Network analysis, London: Sage.

Kreps, D. 1990. 'Corporate culture and economic theory', in J. Alt and K. Shepsle (eds.). Perspectives on

positive political economy. Cambridge: Cambridge University Press.

tjincet. 1948. 'Safeguards', Lancet, 10 April, 561.

Lipsky, M. 1980. Street-level bureaucracy: dilemmas of the individual in pubtic services. New York: Russell Sage.

Lowndes, V. 1996. 'Varieties of new institutionalism: a critical appraisal'. Public Administration 74, 181-97.

Marsh, D. and R. Rhodes. 1992a. 'Policy communities and issue networks: beyond t y p o l o g / , in Marsh and

Rhodes (eds.). Policy networks in British government. Oxford: Clarendon.

—. (eds.) 1992b. Implementing Thatcherite policies: audit of an era. Buckingham: Open University Press.

Milgrom, P. and J. Roberts. 1990. 'Bargaining costs, influence costs and the organization of economic

activity', in J. Alt and K. Shepsle (eds.). Perspectives on positive political economy. Cambridge: Cambridge

University Press.

Miller, D. 1992. Managerial dilemmas: the political economy of hierarchy. Cambridge: Cambridge University Press.

North, D. 1990. Institutions, institutional change and economic performance. Cambridge: Cambridge Univer- sity Press.

Powell, W. 1990. 'Neither market nor hierarchy: network forms of organization', in B. Staw and L. Cumming

(eds.). Research in organization behaviour. Greenwhich: JAI Press.

Rhodes, R. 1988. Beyond Westminster and Whitehall. London: Unwin Hyman.

—. 1996. T h e new governance: governing without government'. Political Studies 44, 652-67.

Smith, M. 1993. Pressure, power and policy: state autonomy and policy networks in Britain and the United States.

Hemel Hempstead: Harvester Wheatsheaf.

Taylor, M. 1987. The possibility of cooperation. Cambridge: Cambridge University Press.

Thompson, G., J. Frances and J. Mitchell (eds.) 1991. Markets', hierarchies and networks. London: Sage.

Thorelli, H. 1990. 'Networks: between markets and hierarchies'. Strategic Management Journal 7, 37-51.

Uzzi, B. 1996. 'The sources and consequences of embeddedness for the economic performance of organiza- tions', American Sociological Review 61, 674-98.

Watkin, B. 1978. The National Health Service: the first phase. London: George Allen and Unwin.

Webster, C. 1988. The health services since the war: problems of health care, the National Health Service before 1957. London: HMSO.

Weingest, B. and W. Marshall. 1988. 'The industrial organization of Congress: or, why, legislatures, like firms, are not organized as markets'. Journal of Political Economy 96, 132-64.

Willcocks, A. 1967. The creation of the National Health Service: a study of pressure groups and a major social policy decision. London: Routledge and Kegan Paul.

Wilks, S. and M. Wright (eds.). 1987. Comparative government-industry relations. Oxford: Clarendon. Williamson, O. 1975. Markets and hierarchies. New York: Free Press.

TRANSACTION COSTS AND POLICY NETWORK THEORY 43

—. 1986. 'Vertical integration and related variations on a transaction-cost economics theme', in J. Stiglitz and

G. Matthewson (eds.). New developments in the analysis of market structure. Cambridge, Mass: MIT Press.

—. 1985. The economic institutions of capitalism. New York: Free Press.

—. 1993. 'Calculativeness, trust and economic organization'. Journal of Law and Economics 36, 453-86.

—. 1995. 'Hierarchies, markets and power in the economy: an economic perspective'. Industrial and Corporate

Change 4, 21-49.

Wright, M. 1988. 'Policy community, policy network and comparative industrial policies'. Political Studies

36, 593-612.

Date received 16 April 1997. Dated accepted 31 October 1997.