Transaction costs, externalities and information technology in health care

HEALTH ECONOMICS, VOL. 5: 25-36 (1996)

ECONOMICS OF HEALTH CARE SYSTEMS

TRANSACTION COSTS, EXTERNALITIES AND INFORMATION TECHNOLOGY IN HEALTH CARE

BRIAN FERGUSON’ AND JUSTIN KEEN’ ’ Centre for Health Economics, University of York; 2Health Economics Research Group, Brunel University

SUMMARY

This paper discusses some of the economic issues which underpin the rationale for investment in information and communications technologies (ICTs). Infomation imperfections lead to significant transaction costs (search, negotiating and monitoring) which in turn confer a negative externality on parties involved in exchange. This divergence in private and social costs leads to a degree of resource misallocation (efficiency loss) which, uncorrected, results in a suboptimal outcome. Traditional solutions to this problem are to rely upon direct government action to reduce the costs of transacting between market agents, or to employ tax/subsidy measures and other legislative action to achieve the desired market outcome.

Three key policy questions are raised in the context of the NHS purchaser/provider relationship. Firstly, what is the optimum level of transaction costs; secondly, can ICTs assist in lowering the level of transaction costs to the optimum level; thirdly, who should bear the investment cost in reducing the level of transaction costs? The issue of property rights in different information systems is discussed and raises interesting policy questions about how much investment should be undertaken centrally rather than devolved to a more local level.

In some ways this economic framework offers a post hoc justification of why different ICT systems have been introduced at various levels of the NHS. Essentially this reduces to the problem of externalities: providing good information confers a positive externality; not providing relevant, timely and accurate information confers a negative externality, by increasing further the level of transaction costs. The crucial role which ICT systems can play lies in attempting to reduce the level of transaction costs and driving the market towards what Dahlman has described as the transaction-cost-constrained equilibrium.

KEY WORDS-information technologies; transaction costs

INTRODUCTION

The publication of Working for Patients’ and Caring for People’ in 1989 heralded major changes in the organisation and delivery of health and social care in the UK. The introduction of the new arrangements has resulted in new organisational structures and information flows; at the time of writing both are in flux, and the NHS could be fairly described as a highly malleable organisation. It is not clear, however, whether these changes will increase the efficiency of service delivery. The issue has been addressed by several

author^,^-^ whose analyses raise questions about the wisdom of some policies, but conclude that it is too early to judge their overall likelihood of success.

All of the authors highlight the crucial role of information, and the difficulties surrounding its acquisition and use, in the new structures in general and in quasi-markets in particular. Information problems manifest themselves in various ways in health care. Information about costs, quality and final outcomes is notoriously dBcu1t to obtain and value.’-* In market transactions, the absence of reliable information on the price, quality or

Address for correspondence: Brian Ferguson, Deputy Director, Centre for Health Economics, University of Yo&, YO1 SDD, UK.

CCC 1057-9230/96/010025- 12 0 1996 by John Wiley & Sons, Ltd.

Received 21 October I994 Accepted 22 November I995

26 B. FERGUSON AND J. KEEN

outcomes of care will distort competition and lead to efficiency losses. From a patient’s perspective, health care can be viewed as an experience good.g Problems relating to information collection and use are, then, crucial to any analysis of current health policies.

Any discussion about information in modem health services will inevitably raise questions about the role of information and communications technologies (IcTs). In common with the general environment within the NHS, ICTs are developing rapidly and in somewhat unpredictable fashion. Most visibly, there has been discussion and specu- lation in the national and specialist media about the potential afforded by the convergence of comput- ing and telecommunications. According to some commentators, a new generation of computers and networks is set to transform every aspect of our lives, whether it be through home shopping, electronic voting or by facilitating working from home. The claim, therefore, is that these systems may have profound economic and social effects, although the scale and nature of these effects remain unclear. This claim is also advanced in the NHS where, it is argued, there is considerable scope for transforming management and clinical practice with the use of ICTs.

Given the importance of information within the reformed NHS, and the pace of technological developments, it is timely to investigate the role and value of ICTs in health care. There appears to have been remarkably little serious evaluation in the past, either at the level of overall policies or of individual systems. This paper discusses the efficiency of health services, and the extent to which information and ICTs may increase or decrease efficiency. The transactions cost approach is used as a framework for exploring the relationships between ICTs and the efficiency of services.

MARKET CONTEXT

This paper focuses upon the new contractual environment within the health service in which purchasers (eg. General Practitioners, Health Authorities) contract with provider hospitals for a defined range of services. The organisational model has been referred to as the ‘internal market’, which is only accurate up to a point: purchasers are allowed to contract with organisations from other sectors (private, independent, voluntary) who are essentially in competition with NHS providers.

The key relationship is nevertheless the purchaserlprovider one and it is useful to focus upon this contractual relationship when consider- ing the transactions cost model.

The authors referenced in the Introduction have provided thoughtful analyses of the economic characteristics of the internal market. For the purposes of this paper another useful contribution has been made by Dawson.” Dawson argues that the NHS has important characteristics in common with markets for industrial goods, where there are high fixed costs and relatively low marginal costs, and generally small numbers of buyers and sellers in any one locality. In industrial markets many contracts are unique, and it is more efficient for the buyer and seller to negotiate directly rather than for the seller to publish a price list. Dawson argues that, by analogy, it may therefore be more efficient for purchasers and providers of health care to negotiate contracts locally, and not be required to publish contract prices.

One possible objection to her argument is that she does not take proper account of one crucial difference between markets for industrial goods and the NHS, namely that the NHS is a public service and must be accountable to external bodies for the resources it consumes. The problem is that it is not clear how Dawson’s model, which assumes local accountability for decisions, can be resolved with the requirement for accountability to the Government and general public. [This problem does not reduce the value of Dawson’s conmbu- tion, since she is in effect challenging politicians and policy makers to clarify the nature of the NHS internal market, and in particular the nature and degree of central regulation. In the context of this paper the key point is that this issue is unresolved, and the NHS has both market-like and hierarchical (vertical accountability) characteristics].

Although this important issue has not been resolved, the NHS Executive has had to devise policies in many areas of NHS activity, including information technology policy. The Information Management and Technology (M&T) Strategy” was published by the NHS Executive at the end of 1992. The main purpose of the Strategy is to provide a framework to guide national and local investments in I n s , and to lay down data and communication standards. Broadly, the NHS Executive is funding a number of initiatives directly, which are designed to facilitate electronic exchange of data across the NHS, while purchasers and providers are charged with implementing

INF0RMA”ION TECHNOLOGY IN HEALTH CARE 27

systems to meet their local needs. Promulgation of national standards should help local sites to implement compatible information systems.

The M&T Strategy requires the development of new systems to support post-reform management processes, rather than simply assuring that the preexisting hierarchical reporting requirements can be met using electronic means. In fact the Strategy does not make clear how information systems can be designed to satisfy the requirements of both the purchaser/provider split and central reporting, and so provides a practical example of the dilemma highlighted by Dawson.

Whether or not the problem is resolved, the clear implication is that there will be substantial expenditure on new ICTs across the NHS over the next few years. It is difficult to quanm this expenditure, as there is a paucity of reliable data about historical expenditure on ICTs. The NHS Executive estimates that hospitals will spend E125M. on purchasing hospital infomation support systems (HISS) in the three years to March 1998. HISS is only one element, albeit an important one, of hospital investments in ICTs, and the total capital expenditure in hospitals may be substantially higher than this figure.

The only clear indication of the likely order of expenditure on the national ICT infrastructure is in a British Medical Journal edit~rial’~ which states that the NHS is spending a nine figure sum on a nationwide computer network; and even here no source for this very large figure is quoted. All of this suggests that any estimate of the capital, implementation and runnjng costs of all NHS systems will be in the order of hundreds of mil- lions of pounds over the next few years.

Indeed, it is not possible to comment in any detail about ICTs in the NHS, principally because there is so little empirical evidence available about commonly used systems. Such research evidence as there is, much of which is from the USA, suggests that ICTs often lead to changes in patterns of service delivery, but there is almost no empirical evidence about patient outcomes. l4 As Glandon and Buck” point out, even where economic evaluations have been attempted, most studies have been poorly designed, and so provide little basis for discussion of the true costs and benefits of investments. There is, therefore, a pressing need for good empirical evidence about the costs and benefits of IT investments. This paper can be viewed as a contribution to efforts to place discussion of ICTs in health care on a proper theoretical footing.

EXTERNALITIES, TRANSACTION COSTS AND IMPERFECTINFORMATION

The economics literature on externalities arose in recognition of the fact that, when all contracts have been entered into by market transactors, there still remain some interactions which ought to be internalised but with which unconstrained market forces cannot cope. According to Dahlman,16 the only reason for this is because the cost of carrying out the actual transaction is greater than the expected benefit: that is, the problem of extemali- ties, which may be said to exist whenever there exists a divergence between private and social cost, arises because of the presence of transaction costs. Reinforcing this view, Dick” stated that ‘it is economic nonsense to talk about externalities in a zero transactions cost world’.

Taking this, and Dahlman’~’~ view that imperfect information is really the sole cause of transaction costs, the argument is as follows. Insufficient or imperfect information increases the level of transaction costs. In the context of the purchaser/provider relationship, the imperfect nature of available information supplied by the provider would impose further costs upon the purchaser (eg. additional direct search costs to improve the information base, or more indirect costs arising from sub-optimal decision-making). This leads to real resource consequences and an inefficient market outcome unless some form of corrective action is taken.

Imperfect information will lead to externalities which would not otherwise exist. The different types of cost (which are discussed in detail later) are not necessarily confined to the single purchaser/provider relationship. In most circums- tances such costs are not confined to a single contract nor are they borne only by the organisations party to that contract (for example, the Regional Office model adopted by the NHS Execu- tive implies some contract monitoringlmarket analysis role which is essentially an additional overhead cost to the NHS, spread across a whole range of regional contracts).

The presence of externalities also draws attention to the question of the extent to which information is a public or a private good. There are, broadly, two types of information available to purchasers and providers. The first is that which is published in books and journals, which has often been generated through research, and approxi- mates to a public good. Such information has the


typical characteristics of pure public goods, for example jointness of supply and nonexcludability in consumption. The second is the information generated within the health service about the volume, cost and quality of services delivered. Historically this has in effect been a private good, owned and controlled by the clinicians or mana- gers who produce it. Requirements for accountability mean, though, that providers must be encouraged to demonstrate how they have used the resources allocated to them.

The policy implication is of course to ensure, through statutory requirements, that (for example) providers collect and make routinely available certain types of information. Over the years statutory minimum datasets have been designed to ensure a minimum level of reporting. The theoretical justification is that a negative level of externality would otherwise exist and, by forcing some players in the contractual process to collect and make available certain information, these external costs become internalised.

Information (and supporting ICTs) may therefore be said to act as a positive externality; systems which operate effectively could have the same effect as a reduction in per-unit taxes on contracts. Two distinct effects may be at work. The first exploits economies of scale in the collection, analysis and distribution of information. Given that the costs of acquiring information are typically higher than the costs of subsequent distribution, it makes sense for a single party to acquire relevant information and distribute it to others, avoiding duplication of expensive effort. In public services, either the Government can be that party or it can delegate the task to others.

The second externality effect derives from standardisation associated with ICT networks. A large network requires a high degree of standardisation of data and technology, and in economic terms can be viewed as equivalent to the introduction of a new standard (or set of standards). David and Greenstein,’* reviewing evidence on the economics of standardisation, posit a law of increasing returns: the more people adopt a standard, the greater the benefit (or reduction in costs) which accrues to each person. In the same way, the more people use a particular network, the greater the benefits which potentially flow to them and others. The expectation of such widespread use of a network is that efficiency gains will be promoted, as the incremental cost of additional users becomes negligible in relation to additional benefits.

David and Greenstein do, however, point to possible efficiency losses sustained because of the need for regulation of standards. For example, some groups systematically acquire more influence than others because they are unequally represented when the issues discussed are of a technical nature. Similarly, FarrellIg argues that standardisation may lead to four potentially important problems : 0 minority interests are not catered for, since the

standards-setting process tends to lead to com- promises in favour of majorities;

0 there is the oft-observed problem of lock-in to old or inappropriate standards;

0 once movement does occur towards a new standard, it may be large and rapid, and hence difficult to manage;

0 success requires a major gamble by one or more agents, since there are no ex ante market (or indeed other) mechanisms to test the long-run efficiency of a new standard. The risk, therefore, is that sub-optimal standards are implemented.

Notwithstanding these problems, there is a strong case for the NHS Executive to investigate whether efficiency gains can be realised through the exploi- tation of positive externalities, particularly so if the regulatory costs of market intervention can be minimised or avoided.

TRANSACTIONS COST APPROACH

Types of transaction costs

Prominent in the (fairly limited) literature on costs and benefits of information systems is an emphasis upon the transactions cost model. In terms of conventional economic analysis the presence of transaction costs means that socially efficient outcomes (incremental costs = incremental benefits) are not achieved. This implies some form of welfare loss to society, the effects of which could be mitigated by reducing the transaction costs themselves. There are, for the purposes of this analysis, at least three major types of transaction costs which are relevant in the reformed MIS: 0 the costs of acquiring information (loosely

0 the costs of bargaining and making decisions referred to as ‘search costs’);

(loosely referred to as ‘negotiating costs’);

INFOFWATION TECHNOLOGY IN HEALTH CARE 29 0 the costs of policing and enforcement (loosely

referred to as ‘monitoring costs’).

This classification is consistent with the distinction made by Williamsonzo-2’ who divides transaction costs into two types: ex ante and ex post exchange. Broadly speaking, the ‘search’ and ‘negotiating’ costs are the ex ante costs associated with reaching an exchange agreement, and the ‘monitoring’ costs relate to the ex post monitoring of outcomes, contract compliance and, possibly, arbitration.

Robinson,6 following Williamson, makes the point that ‘firms and markets can be viewed as alternative methods of economic organisation for handling transactions’. In essence, the choice as to which method of organisation is optimal reduces to a question of relative efficiency. There are three features-bounded rationality, opportunism and asset specificity (described below)-which can be expected to favour internal organisation over market transactions. It is clear that the Government has placed faith in the converse, believing that supply-side competition, leading to system efficiency gains, will more than offset the additional transaction costs. This latter issue is an empirical one and perhaps subsidiary to the basic point that the conditions highlighted by the transactions cost approach are likely to be prevalent in the reformed NHS.

Firstly, bounded rationality simply refers to the fact that decision makers can only act rationally to a limited extent, due to the constraints of future uncertainty and the complexity of problem- solving. This is particularly true in the NHS in which day-today ‘fire fighting’ often takes place at the expense of long-term strategic thinking, and decisions inevitably reflect the best that can be done in the time available. As Gurbaxani and Whang’ point out, decision-making in the absence of relevant information can lead to sub-optimal outcomes, hence imposing additional costs upon the process of transacting. Furthermore, uncertainty is pervasive in the NHS factors such as frequent re- structuring and the job mobility of key staff involved in negotiating contracts mean that effective long-term contractual relationships (possibly the most beneficial of market arrangements-see Hamu) are not always possible.

Opportunism refers to the pursuit of individual self-interest, possibly at the expense of wider corporate or system benefits. While the traditional caring values embodied in the NHS founding

principles still exist, there are incentives for providers to act entrepreneurially, with possible increases in organisational slack and inefficient resource use. This is particularly likely given that providers have no incentive to generate ‘excess’ surpluses, since these can be reclaimed through the Secretary of State’s reserve powers. Surpluses in excess of the target rate of return on capital employed may well be taken by providers in the form of greater managerial perks or higher levels of amenity.

Finally, asset specificity arises when transactions require specific investment in physical and human assets. In essence, this means that long- term contractual relationships are likely to be fruitful since purchasers and providers will have invested human (and other) capital in drawing up service specifications and contracts.

Each of these conditions, to a greater or lesser degree, implies that: 1) it is worth investing human and other resources in forging long-term relationships between purchaser and provider; 2) there is potential for opportunistic behaviour which leads to mismst and the pursuit of goals which may conflict with wider system objectives; and 3) there is a good argument for a monitoring/market analysis role which is external to the contractual relationships per se (ie. it is undertaken by a third party). In tum all of these features will inevitably lead to higher search, negotiating and monitoring costs.

It is within this context that the role of information and communications technology should be viewed. While information systems have multiple roles in an organisation (eg. see Gurbaxani and Whang” who outline five roles: operational, transaction processing, decision support, monitoring and performance evaluation, documentation and communication), Ciborra% points out that ‘the key feature of a mediating technology is ... the possibility of lowering the costs of transacting’. Cibonra describes information technology as a ‘mediating’ technology within his proposed framework for analysis: ie. a technology which links several individuals through standardisation and extension of the linkages.

E@ciency loss The fundamental line of argument therefore

runs as follows. Economic inefficiency (defined broadly as marginal social costs not equalling marginal social benefits) in the NHS is caused at


least in part by imperfect information. Remedying this situation is not costless: indeed, the associated transaction costs are likely to be significant. (Recall Dahlman’~’~ famous article in the Journal of Law and Economics in which he refers to the existence of only really one type of transaction cost: namely, resource losses due to imperfect information.) Reducing these to zero is not feasible, nor is it even likely to be desirable. The problem therefore reduces to a series of key questions: a) what is the optimum level of transaction costs?; b) can ICTs assist in lowering the level of transaction costs to the optimum level (this makes the not unreasonable assumption that existing transaction costs are excessive)?; and c) who should bear the investment cost in reducing the level of transaction costs?

These three questions are discussed in more detail below. Meanwhile consider Figure 1 which highlights the efficiency loss within the context of contract negotiations between a single purchaser and provider for a particular service. The horizon- tal axis represents the volume of services agreed between purchaser and provider, and the vertical axis represents the costs of providing that level (and quality) of service. This simple conceptual framework assumes that it is possible to identify the marginal costs (MC) of providing one more unit of service, and to estimate the marginal valuation (MY) placed upon that service.

The framework also assumes for simplicity that the transaction costs are incurred by the provider,

hence are reflected in a higher cost per unit of service. This simplifying assumption is not unreasonable and not crucial to the analysis: the provider is typically in the position of having to provide detailed breakdowns of the costs of service and evidence that it is meeting qualitative criteria. Hence it is assumed that these information (search) costs are built into the provider’s cost function. In terms of Figure 1 these transaction costs would have the same effect upon the provider as a per-unit tax which resmcted the volume of service which the provider could meet within a given budget constraint. Rather than the optimal level of service (04, where MC= MV), the service is restricted by the amount h4. This reduction in service saves a certain level of resource (hqfe) but there is a loss in benefit, measured by the marginal valuation of hq: area hqfa in Figure 1 (see Ngs). The net loss is the difference between the two areas: shaded area aef. Analogous to measures of consumer surplus in welfare economics, this area represents the efficiency loss arising from transaction costs in this purchaser/ provider contract negotiation process.

Optimum level of transaction costs This returns us to the three key questions posed

earlier, which have assumed a much sharper focus since the NHS reforms. Statutory infomation requirements were more straightforward in a system which did not separate the provision of services from their financing. Hospital (Unit)

Vollmle of service

Fig. 1. Efficiency loss imposed by transaction costs.

INFORMATION TECHNOLOGY IN HEALTH CARE 31 information was effectively the same as District Health Authority (DHA) information.

The reforms have made a clear distinction such that purchaser (eg. DHA) information, while having significant overlaps with provider (eg. NHS Trust hospital) information, has a very different focus. For example, both purchaser and provider will have an interest in ensuring timely and accurate inpatient activity information; it is less clear, however, that providers will be inter- ested in health needs databases, or, conversely, that purchasers will wish to have detailed knowledge of operating theatre information systems.

This arises largely because of the differing objectives which purchasers and providers have in the reformed NHS. The primary role of the purchaser lies in assessing the needs of its resident population and commissioning services accord- ingly; providers, while having many different objectives due to the multiplicity of managerial and clinical decision-makers in their organisations, have to win sufficient contract income to remain financially viable.

In effect, the purchaser/provider split brings into focus the question of who provides what type of information, and, perhaps more important, who should bear the investment cost in the necessary information systems. In short, a provider could reasonably consider that supplying vast quantities of information to a range of purchasers is confer- ring a positive externality. Similarly, a purchaser which did not receive timely and accurate price or waiting list information may legitimately feel that a negative externality exists, in turn requiring intervention by some visible or invisible hand in the market-place.

If it is accepted that, as Dahlman16 states: ‘in the theory of externalities transaction costs are the root of all evil’, this implies that policy intervention should aim to reduce the level of transaction costs. What, then, is the optimum level of transaction costs? The following passage from DahlmanI6 highlights the need to think in terms of a ‘transaction-cost-constrained’ model rather than the perfectly competitive solution to a Walrasian general equilibrium model envisaged by some authors:

‘...if there are costs of transacting present in our world, then it may be too costly to eliminate all externalities, so we should preserve some of them in order to reach an optimum. Hence, as a suitable reference point, we ought to use a transactioncostconstrained

model that describes the relevant and attainable optimum, not the irrelevant and unattainable solution to a model with zero transaction costs.’ Dahlman (p.153)

One of the key contributions made by Coase26 was to recognise that transaction costs are a necessary condition for deviations from an ‘attainable optimum’. In such a situation the presence of transaction costs, combined with informational differences between parties to the exchange, means that it is likely to matter to whom liability obliga- tions and ownership rights are assigned. The ‘attainable optimum’ should now be thought of as the transaction-cost-constrained equilibrium and the policy question becomes a practical one of how to reach that equilibrium. Coase’s analysis implies that the preferred resource allocation can be attained by: 1. direct government action which reduces the

costs of transacting between market agents; or 2. using taxes (or subsidies) and other legislative

action. In the NHS context, an early example of struc-

tural changes which tend to reduce the level of transaction costs has been the merger of old DHAs to form new, bigger Purchasing Authorities, probably driven by perceptions that economies of scale will be achieved in certain functions (eg. Public Health, Contracting), and perhaps also in the degree of risk-pooling which can be achieved. The pace of mergers has undoubtedly been forced by the Government, partly through suggestions that Purchasing Authorities should be responsible for a resident population of around 0.75 million (much larger than the typical old DHA), and partly through pressure upon DHAs and Family Health Services Authorities (FHSAs) to merge such that their boundaries become co-terminous. Examples of mergers can be interpreted as an organisational response to the genuine uncertainty which is currently pervasive in the NHS, and perhaps also as a response to perceived current information deficiencies and the transaction costs which these impose.

Can ICTs assist in lowering the level of transaction costs to the optimum level?

It seems relatively uncontentious to argue that ICTs have the potential to lower the level of transaction costs. The issue of whether ICT


systems fulfil this potential in practice is a key empirical question. Comparing the roles which Gurbaxani and Whang2’ outline that information systems can play in organisations, with the earlier definition of transaction costs, illustrates a high degree of symmetry (see Table 1). In other words, information systems have developed largely to aid the functions implicit in these costs of transactions.

Implicit in the earlier discussion relating to merged Purchasing Authorities/FHSAs and merged provider units is the practical implication that there would be shared information systems and centralisation of data which were previously more dispersed. Integrated information systems would be dealing with the same population base and smaller geographical areas (localities), allowing information to be shared among a wider range of health and social care professionals.

Integrated information systems which function efficiently will undoubtedly have the potential to reduce, for example, search costs and costs associated with contract monitoring; eg. up-to-date and accurate information relating to comparative hospital waiting times for similar procedures will help both with the drawing-up of next year’s contracts and with the monitoring of existing contracts. In short, organisational change which leads to common goals and the effective sharing of information has the potential to reduce the types of transaction costs outlined in Table 1.

The possible effect of ICTs is now discussed in relation to the three features6 which can be expected to favour internal organisation over market transactions. Firstly, ICTs may help to reduce bounded rationality in two ways: (a) by facilitating the collection of data relating to the behaviour of providers, thereby assisting the purchasing function; (b) by exploiting research

Table 1. Roles of ICT systems and types of transaction cost

Roles of Im systems Types of

transaction cost

Operational Search Transaction processing Bargaining Decision support Decision-making Monitoring and performance Policing and

evaluation inforcement Documentation and Information-sharing

communication

and development information from elsewhere on costs, processes and outcomes-for example from on-line databases and national dissemination centres. However, the volume and complexity of contracts suggests that the capacity of purchasers to collect and analyse information will still be limited. Moreover, the costs and effectiveness of most health and social care interventions are simply not known. Hence, problems may be reduced for specific contracts but overall purchasers and providers will continue to operate under seriously bounded rationality. A key question, therefore, is what incentives exist for purchasers/providers to invest time andlor money in obtaining information?

It is possible to envisage using ICTs to identlfy opportunistic behaviour. Some forms of opportunism occur in the management of services, particularly where it involves intentional misrepresentation-for example in claims made by GPs for item-of-service payments. Information technology can be used to analyse such claims and direct attention towards potential abuses more easily than with manual systems (the ICT is used as a pattern-matching device to examine claims behaviour over time). It is worth noting in passing that the costs of detection and potential prosecu- tion may actually be higher than allowing fraudulent claims to continue, even although the latter is not a feasible option. The example serves as a cautionary note that ICT, even where it helps to improve efficiency in some context, may actually increase costs.

Opportunistic behaviour may also occur where information relating to the quality or outcome of services is not readily available, leading to inefficient behaviour by service providers. Patients experience this inefficiency, in the form of sub- optimal treatment for individuals and suboptimal organisation of service delivery for groups of patients, so the problem is far from trivial. If ICTs facilitate the monitoring of service delivery (eg. when individual homes are linked on networks to facilitate the eliciting of outcomes information), then efficiency may be enhanced. This scenario also depends on clinicians releasing data about their own work, or being persuaded to focus and act upon patient-based outcomes data themselves, an unlikely development in the short-term.

The scope for opportunistic behaviour is increased by the fact that those who might benefit from such behaviour are in the position of being able to enter the data themselves. While not

INFORMATION TECHNOLOGY IN HEALTH CARE 33 suggesting that this position is exploited in practice, there is an implicit degree of trust which may be tested in situations where market forces begin to bite and hospital viability becomes an issue. In such cases it would be less easy for purchasers to challenge qualitative data than sharper contract price information, and a status quo-which may not be in the best interests of patients-is maintained.

ICTs will not be useful in situations where discussions are beyond the reach of the information technology itself. Potentially the most important instance is where outcomes cannot easily be measured, nor is it desirable to even try. For example, the palliative care of terminal cancer patients concentrates upon making life as comfort- able for the patient as possible-thereby in some small but important way improving quality of life-but there is no real prospect of improving health outcomes. A system which attempted to do so would be at best irrelevant and at worst offen- sive to patients and their carers. In a Merent context, there is no obvious role for ICT (ironi- cally!) in dealing with the opportunistic behaviour of the type found recently in Wessex Region2’ where vital information was concealed from Members of the Regional Health Authority and the NHS Executive. In such cases the tendency for opportunistic human behaviour is able to prevail however sophisticated the detection systems in place.

It is possible that ICTs will reduce costs associated with asset specificity. Service providers may share information, perhaps over networks; or databases might be developed to help transfer expertise from (say) a hospital physician to a GP. Both examples may assist in reducing the degree of asset specificity in relation to services for individual or groups of patients. Ins may also enable the delivery of services in new settings and thus reduce the costs of certain services, for example through avoiding investments in specialised facilities (although this would need to be tested empirically). For example, telemedicine-involving services delivered over wide-area networks-might facilitate devolution of minor surgery and other services to GP surgeries and health centres. However, the extent to which asset specificity can or should be reduced may be limited, since it is desirable that health professionals invest in acquiring specialised knowledge on our behalf.

The interim conclusion must be that the transaction costs associated with contracting may be

ameliorated through the use of I n s in specific instances, but in general the scope for ICTs to realise efficiency gains must be subjected to rigorous empirical evaluation. This is inextricably linked to the more general question of efficiency gains which can be expected to arise from contracting in the internal market, an issue which has also been characterised by remarkably little empirical analysis.

It is worth noting that one possible consequence of the line of argument in this paper is that ICTs may themselves influence the structure of the internal market. If ICTs substantially reduce the marginal costs of acquiring and using contract information (possibly by building upon information which is routinely collected for clinical purposes), and through linking purchasers and providers together on electronic networks increase the number of providers that can bid for a particular service, then they may serve to increase the competitiveness of the internal market. I n s may thus not only increase the efficiency of existing contracting processes in the ways outlined above, but also change the scope of contracting itself. (It should be stressed that there is no empirical evidence that the marginal costs of infomation acquisition and use will be substantially lower than for existing systems.) The NHS Executive might thus usefully consider the implications of ICTs for the nature of the internal market.

Who should bear the investment cost in reducing the level of transaction costs?

In addressing this question the first and most obvious point is that it depends upon the nature of the information under consideration. For example, property rights in a health needs assessment data- base would most likely be vested in a Health Authority purchaser, and similarly infomation relating to operating theatre utilisation vested in provider Units. It is less obvious, however, who should be responsible for collecting and computeris- ing infomation about comparative GP referral rates, waiting times and prices of particular procedures. Compiling such information carries sigdicant system set-up and running cost implications. In the case of comparative data which will be of value to more than just one purchaser and one provider in contract negotiations, there is clearly an argument for central data collection, whereby the majority of costs are borne centrally and the same informaion is disseminated to all purchasers and providers.


With regard to merent types of information, the NHS Executive has put in place an infrastructure which should help to achieve scale economies. For example, the Cochrane Collaboration analyses and distributes information about clinical trials. Similarly, the Centre for Reviews and Dissemina- tion at the University of York undertakes systematic reviews of the effectiveness of health care interventions. In principle, therefore, the answer to the question ‘who should bear the investment cost?’ is reflected in the NHS Execu- tive’s conscious decision to bear the investment costs of collecting certain types of information. It remains an empirical question whether such economies of scale are achieved in practice. Econometric evidence from both the UK and USA with regard to direct health care services is mixed in this respect,28 and virtually nonexistent with regard to information technology.

The NHS Executive’s IM&T Strategy” involves the creation of an NHS-wide network together with the development and implementation of a range of data and technical standards. The spine of a national network is being funded by the NHS Executive, with local sites (purchasers and providers) charged with linking themselves to it. The standards include Read Codes, the use of Health- care Resource Groups (HRGs) in contracting, Patients’ Charter, Efficiency Index, and so on. The Strategy can be interpreted as seeking to confer positive externalities on the contracting process, in effect by reducing the level of transaction costs within the system.

David and Greenstein’s and Farrell’s points a b ~ v e ’ ~ - ’ ~ can be used to judge how far positive effects might be offset by the regulatory costs of standardisation. On the positive side, the MIS Executive may be viewed as the only body capable of funding and promoting the infrastructure for scale economies of information management and the major standardisation processes required to implement a national IM&T Strategy. Efficiency may be increased as a result of tlus considerable business risk; on the other hand, ‘lock-in’ may occur with new standards lacking the flexibility to permit evolutionary change, or some standards may prove unenforceable in an NHS with increas- ingly autonomous market agents. Whether or not the gamble will pay off, only time-and careful evaluation-will tell.

The alternative is to allow the sums of money devoted to such central data collection to flow down the system so that purchasers and providers

invest in the local systems which they consider necessary to produce relevant infomation to support contracting and other functions. (In the case of a cash-limited system the overall sum of money is of course the same, but, devolved to more local level, could perhaps be more specifically targeted). Comparative information would then cover a restricted geographical area: this is arguably more desirable given the potential confounding factors associated with comparing data from distant parts of the country, in which there may exist marked differences in culture and clinical practice (a point which advocates of ‘evidence-based medicine’ need to take into account). It is important, however, to bear in mind Smith’s point that devolution of responsibility ‘only reduces the need for information flows between tiers, and there remains a crucial need for accurate information within the devolved units if they are to manage their resources successfully. Moreover, if the central authori ty... is to set the devolved authorities.. .realistic and equitable targets, it still requires a large volume of undistorted information.. .’ (p.1 13).8

The interesting question, however, is how the balance is struck between ensuring that national infrastructures are in place to facilitate the operation of the internal market (and indeed to address issues of public accountability), and how much local autonomy remains for purchasers and providers to decide upon-and pay for-information systems which they require to fulfil contracting and other functions. This is a crucial economic question: significant local resources are tied up in completing national returns and ensuring that standards are adhered to in the implementation of ICT systems. The opportunity cost of these resources lies in whether there are more appropriate locally-based systems which organisations would value. Indeed, purchasers and providers locally may decide that the transaction costs associated with gathering certain types of infomation are smply not worth iilcming, and that the resources could be better employed elsewhere- for example in direct patient care.

In practice the balance between local autonomy and central control is difficult to achieve; whichever way the problem is viewed, there is no substitute for careful evaluation of the transaction costs involved. The problem can be summarised as attempting to achieve a balance between positive and negative externalities. The issue is wider than the evaluation of ICTs, and centres upon the

INFORMATION TECHNOLOGY IN HEALTH CARE 35 degree of regulation which the Government wishes to impose on the functioning of health and social care markets.

DISCUSSION

The aim of this Section is to highlight some of the key points and policy implications which arise from the transactions cost approach. The transactions cost framework is intuitively appealing: excessive transaction costs arising from imperfect information lead to inefficiencies in, for example, the purchaser/provider contracting relationship. The question is to what extent ICTs can improve information flows and thereby reduce the degree of efficiency loss. This is largely an empirical question which lies at the heart of discussions around efficiency gains from a more market- orientated MIS. Undoubtedly ICTs have the potential to reduce transaction costs but there are important questions surrounding the optimal level of investment and who should bear the cost.

Given that there is an externality dimension to this-essentially, good information confers a positive externality-there are sound arguments for government intervention to lend a visible hand to the market. The NHS IM&T Strategy fits well with this argument: the government is bearing a proportion of the risk of investment in large systems and the implementation of agreed standards. One interesting question, however, is the extent to which the NHS Executive’s role becomes one of direct control rather than simply regulation of the market. In the area of NHS IT in particular, it is difficult to avoid the conclusion that the NHS reforms have led to greater centralisation, which is somewhat at odds with White Paper calls for devolution to local level. There may be sound reasons for this, but a balance must be struck between central direction of information requirements/systems and local decisions between purchasers and providers regarding the M N ~ of information considered necessary to support contracting and other functions. At present the balance appears to be strongly on the prescriptive side rather than the facilitating role which would perhaps be more appropriate to a managed market system. This may be unfair criticism of ‘the Centre’

which is attempting to put in place the infrastructure to support the market; national networks are a good example of potential positive externalities in

information-sharing and communication. The key question nevertheless remains of what is the optimal level of transaction costs, and hence investment in infonnation systems to reduce these, together with the equally important consideration of the most cost-effective method of achievement.

The fact that little research has been undertaken into the costs and benefits of ICTs means that there are many outstanding empirical questions to be addressed. (Many will of course concern clinical benefits of ICTs and are beyond the scope of this paper.) The arguments in this paper suggest that economic evaluations should investigate whether economies of scope and scale are actually achieved through the introduction of data and system standards. The NHS IM&T Strategy places great emphasis on the potential value of networks linking purchasers and providers, and there will be opportunities for evaluation as new networks are implemented. Major investments, such as in the NHS Clearing Centre for contracting data, would seem to be prime candidates for.evaluation. More generally, areas where there is substantial data traffic, as for example between GP practices and hospitals, are also strong candidates.

In summary, the approach described in this paper poses questions about the real role of ICTs in health and social care, and the extent to which they can achieve the diverse objectives of organisations involved in contracting. It also raises the vital questions of optimal level of investment and the role for government in bearing risks associated with national networks and standards. The organisational flux which appears to characterise the present NHS increases the need for careful evaluation of new and existing ICTs. It seems clear that ICTs will continue to shape organisational development and the implementation of the NHS reforms; it is of crucial policy relevance to know whether ICTs will reduce transaction costs in the process.

ACKNOWLEDGEMENTS

The authors thank Professor Ray Robinson and anonymous referees for comments on an earlier draft.

REFERENCES

1. Department of Health. Working for patients. Cm 555; London: HMSO, 1989.

36 B. FERGUSON

2. Depment of Health. Caring for people: commu- nity care in the next decade and beyond. Cm 489; London: HMSO, 1989.

3. Banlett, W. Quasi-markets and contracts: a markets and hierarchies perspective on NHS reforms. Public Money and Management 199 1.

4. Knapp, M., Wistow, G., Forder, J. and Hardy, B. Markets for social care: opportunities, barriers and implications. PSSRU Discussion Paper 919. Uni- versity of Kent, 1993.

5. Le Grand, J. Evaluating the NHS reforms. In: Robinson, R.. and L.e Grand, J. (eds) Evaluating the NHS reforms. London, King’s Fund Institute., 1994.

6. Robinson, R. Hospitals in the market. In: Keen. J. (ed) Ir3formation management in health services. Open University Press, 1994.

7. Raftery, J. and Stevens, A. Information for purchasing. In: Keen, J. (ed), Information management in health services. Milton Keynes, Open University Press, 1994.

8. Smith, P. Information systems and the White Paper proposals. In: Culyer, A.J., Maynard, A.K. and Posnett, J.W. (eds) Competition in health care: Reforming the NHS. MacMillan Press, 1990.

9. Nelson, P. Information and consumer behaviour. Journal of Political Economy 1970; 78: 311- 29.

10. Dawson, D. Costs and prices in the internal market: markets and the NHS Management Executive Guidelines. Centre for Health Economics Discus- sion Paper 115, University of York, February 1994.

11. NHS Management Executive. Information management and technology strategy: overview. London: Information Management GroupN-ISME, 1992.

12. HISS Central Team. HISS Catalogue 1994. Winchester, NHS Executive Information Manage- ment GroupMSS Central Team, 1995.

13. Anderson R. NHS-wide networking and patient confidentiality: Britain seems headed for a poor solution. British Medical Journal 1995; 311 : 5-6.

14. Anderson, J., Aydin, C., Jay, S. (Eds) Evaluating health care information sytems. Newbury Park CA, Sage, 1994.

AND J. KEEN

15. Glandon, G., Buck, T. Cost-benefit analysis of medical information systems: a critique. In: Ander- son J., Aydin C., Jay S. (eds) Evaluating health care information systems, pp. 164-189. Newbury Park, CA, Sage, 1994.

16. Dahlman, C. J. The problem of externality. Journal of Lao and Economics 1979; 111; 1-44.

17. Dick, D. T. The voluntary approach to externality problems: a survey of the critics. Journal of Environmental Economics and Management 1976

18. David, P.A. and Greenstein, S. The economics of compatibility standards: an introduction to recent research. Economic Innovation and New Tech- nologies 1990; 1: 3-41.

19. Farrell, J. The economics of standardisation: a guide for non-economists. In: Berg, J- L. and Schummy, H. (eds) An analysis of the information technology standardisation process. Elsevier, 1990.

20. Williamson, 0. Markets and hierarchies. New York: Free Press, 1975.

21. Williamson, 0. The economic institutions of capitalism. New York Free Press, 1985.

22. Gurbmani, V. and Whang, S. The impact of information systems on organisations and markets. Communications of the ACM 1991; 34(1): 59-73.

23. Ham, C. Managing the market. Health Service Journal 1994; 1 September: 18-20.

24. Ciborra, C. U. Teams, markets and system: business innovation and information technology. Cambridge University Press, 1993.

25. Ng, Y. K. Welfare economics: introduction and development of basic concepts. MacMillan Press Ltd, 1979.

26. Coase, R. H. The problem of social cost. Journal of LavandEconomics 1960; XXII(1): 141-162.

27. Committee of Public Accounts. Wessex Regional Health Authority regional ir3formation systems plan. PAC 63rd Report, House of Commons. London: HMSO, [Session 19921931.

28. Department of Health. The operation of the NHS internal market: local freedoms. national respon- sibilities. December 1994.

2 : 185-195.

Documents

Transaction costs, externalities and information technology in health care