Transaction Cost

Transaction Cost: Coase, Williamson and the Cost of Contracting for Health Care

In health economics, transaction costs are the resources a health system uses up in organising care rather than in delivering it: commissioners writing service specifications, finance teams checking coded activity, insurers processing claims, patients finding and reaching a provider. They help decide whether a service is bought under contract or produced in-house, and they are the hidden price of reforms that turn budgets into contracts. This page traces the idea from Coase to Williamson's transaction cost economics, sets out the main types, follows them into health care contracting and organisation, works through an illustrative make-or-buy decision, and explains when costing studies and economic evaluations count them.

Coase and the cost of using the price mechanism

Coase asked in 1937 why firms exist if prices coordinate production. His answer was that "there is a cost of using the price mechanism". The most obvious cost is discovering what the relevant prices are, and to this he added the costs of negotiating and concluding a separate contract for each exchange. A firm does not remove contracts, but it greatly reduces them: an employee agrees to follow directions within limits, so "for this series of contracts is substituted one".

Long-term supply adds a further cost. Because the future is hard to forecast, a long contract can state only in general terms what the supplier must do, leaving the details to the buyer, so a firm is likely to emerge where a very short-term contract would be unsatisfactory, especially for labour. Coase, who called these marketing costs, set the boundary at the margin: a firm expands until the cost of organising one more transaction inside it equals the cost of carrying it out on the open market or in another firm.

Williamson: asset specificity, uncertainty and frequency

Oliver Williamson turned Coase's insight into testable predictions. In 1979 he argued that if transaction costs are negligible, the organisation of economic activity is irrelevant, and he named the three critical dimensions on which transactions differ: uncertainty, the frequency with which they recur, and the degree to which durable transaction-specific investments are made.

Asset specificity carries most of the weight. In his Nobel lecture Williamson described specific assets as ones that "cannot be redeployed to alternative uses and users without loss of productive value". Once they are in place, bidding among many suppliers becomes a small-numbers relationship during the contract and at renewal. Complex contracts are incomplete because people are boundedly rational, which Simon, quoted by Williamson, described as "intendedly rational, but only limitedly so". When an unforeseen disturbance arrives and the parties depend on each other, each can bargain for a larger share of the surplus. The Nobel committee calls the resulting threat to specific investments the hold-up problem.

Williamson held that nonspecific transactions, occasional or recurrent, are efficiently organised by markets, while specialised governance is "much easier to justify for recurrent transactions" because its set-up cost can be recovered. As specificity and uncertainty rise, bilateral contracts will often give way to unified ownership. Integration has costs too: the Nobel committee notes that the authority that saves on haggling can be abused.

Search, bargaining, monitoring and enforcement

Health studies often sort transaction costs by stage of exchange. Marini and Street organised their analysis of NHS contracting around ex ante search costs, ex ante negotiation costs, ex post monitoring costs and ex post enforcement costs. Porter and colleagues used Williamson's division into ex ante costs of setting up a contract and ex post costs of managing it.

Patients bear them too: Schneider and colleagues define consumer transaction costs in medical care as the costs of completing a transaction, including the time needed to evaluate, choose and locate a provider and to obtain the service.

Transaction costs in health systems

The purchaser-provider split made transaction costs a policy question. When England replaced locally negotiated block contracts with national tariffs under Payment by Results, the Department of Health, as quoted by Marini and Street, expected a standard tariff to minimise "the transaction costs and conflict involved in local price negotiation". In the same study, commissioned by the Department, they found that price negotiation costs did fall, but six interviewed trusts and primary care trusts reported extra costs of between GBP 90,000 and GBP 190,000 each, mostly for new staff. The money went on managing volumes, collecting patient-level data, checking coding for upcoding and settling more disputes.

Porter and colleagues, studying commissioning for long-term conditions, cited Bartlett and Le Grand's condition that the transaction costs of running a quasi-market should be no higher than any cost savings generated by competition. They found transaction costs uncertain but likely to be high, care hard to divide into neat units for contracting, and commissioning dominated by a relational approach based on trust and collaboration with incumbent providers.

Insurance design changes the administrative bill. The OECD put administration at around 3% of health spending in 2014, double that in France and higher in the United States, and higher where social health insurance funds dominate, although single-fund countries resembled systems with residence-based entitlement. Mathauer and Nicolle, using WHO and OECD health accounts data for 58 countries, reported average administrative costs of 4.2% of total health insurance costs for social security schemes in high-income OECD countries, with private health insurance about three times higher.

Vertical integration is the make-or-buy question for providers. Schneider and colleagues treated orthopaedic practices' ownership of imaging and physiotherapy as a make-or-buy choice, driven by economies of scope and transaction economies. Their study, part-funded by the American Association of Orthopaedic Surgeons, weighed ownership's stronger controls against bureaucratic costs and added use.

Baines and Hale applied new institutional economics to community pharmacy, arguing that its organisation imposes non-price costs on government, pharmacists and patients. They expected the new UK pharmacy contract's clinical governance standards to raise costs, with vertical or horizontal integration a possible response.

Worked example: pathology in-house or under contract

This illustrative example uses invented figures, in GBP per year. A hospital trust can run its own pathology service or buy one from an outside laboratory. Williamson argued that such choices require simultaneous attention to production economies and transaction-cost economies, so each option is fully costed:

$$C_{\text{make}} = K + G \qquad C_{\text{buy}} = P + \frac{S}{n} + M + D$$

where $K$ is the cost of producing the service in-house, $G$ is the internal cost of managing it, $P$ is the annual contract price, $S$ is the one-off cost of tendering and awarding the contract, $n$ is the contract length in years, $M$ is the annual cost of monitoring the contract and $D$ is the expected annual cost of disputes, renegotiation and service failures.

Step 1: in-house. With $K$ = 2,000,000 and $G$ = 100,000:

$$C_{\text{make}} = 2{,}000{,}000 + 100{,}000 = 2{,}100{,}000$$

where the result is the full cost of making.

Step 2: routine tests, many suppliers. The laboratory charges 1,800,000 because of its scale. A five-year tender costs 150,000, monitoring costs 60,000 and expected disputes 20,000:

$$C_{\text{buy}} = 1{,}800{,}000 + \frac{150{,}000}{5} + 60{,}000 + 20{,}000 = 1{,}910{,}000$$

where the annual transaction costs total 110,000. Buying saves 190,000.

Step 3: a dedicated on-site laboratory. Suppose the service needs a rapid-response laboratory built on site and linked to the trust's systems, with uncertain volumes, so few rivals could replace it once built. Monitoring rises to 120,000, and the trust expects a 25% chance each year of conceding 600,000 in renegotiation, plus 50,000 of service failures, so $D$ = 0.25 × 600,000 + 50,000 = 200,000 (the hold-up risk). The price and the in-house cost are assumed unchanged, so only the transaction costs differ:

$$C_{\text{buy}} = 1{,}800{,}000 + \frac{150{,}000}{5} + 120{,}000 + 200{,}000 = 2{,}150{,}000$$

where the annual transaction costs total 350,000. Making is now cheaper by 50,000.

Step 4: contract length. With a one-year contract in step 2, the tender's set-up cost is recovered over fewer years: it would cost 150,000 a year instead of 30,000 and buying would cost 2,030,000, still 70,000 less than making.

A comparison of the contract price with the in-house cost favours buying by 300,000 each time. The full comparison reverses in step 3, as Williamson predicted for specific, uncertain transactions, and frequent re-tendering erodes the margin in step 4.

Counting transaction costs in costing and economic evaluation

Transaction costs are opportunity costs, but many fall outside the prices an evaluation uses. A contract price usually reflects whatever the supplier spends on transacting, while the purchaser's commissioning and monitoring effort usually sits in management overheads. The NICE manual (PMG36) limits reference-case costs to resources under the control of the NHS and personal social services (PSS), including infrastructure changes and, where appropriate, staff training, but names no cost of commissioning or managing contracts.

In an economic evaluation of a multimorbidity care model in Chile, Zamorano and colleagues costed the hours of a university implementation support team by activity (information, finding, bargaining, monitoring, change management), separately from implementation costs, and noted that such expenses are rarely reported in costing studies. Patients' own search and travel time is not an NHS or PSS cost. PMG36 allows unreimbursed costs paid by patients to be presented separately (section 4.4.23), and patients' time is counted under a broader perspective such as a societal perspective.

A cost-effectiveness analysis of a technology costed at its list price usually omits the system's cost of procuring it. An evaluation of a provider payment reform or a new model of care should cost the contracting it creates. Transaction costs should be counted once, either inside a unit cost or as a separate item.

Boundaries with neighbouring concepts

Transaction costs are a category of cost. The Coase theorem uses them to explain when bargaining fails to resolve an externality, one route to market failure. Contract theory asks how agreements should be designed when actions or information are hidden, while transaction cost economics asks which governance structure economises on the cost of making and adapting agreements. The principal-agent problem is one source of monitoring cost. Property rights define what is being exchanged; once assigned, transaction costs determine how cheaply the rights can be traded or enforced.

Limitations and common misreadings

Williamson quoted Stanley Fischer's complaint that "almost anything can be rationalized by invoking suitably specified transaction costs", and answered it by naming the dimensions, such as asset specificity, on which predictions turn; the Nobel committee notes that these can be measured. Measurement remains hard in health care because contracting effort is spread across jobs that do other work too.

• Administration is not the same as waste. The OECD notes that administration also secures patient safety and accountability, and can save money later.
• Lower is not always better. Marini and Street's interviewees were unanimous that the higher costs were justified by clearer payment rules and sharper incentives.
• Administrative and transaction costs overlap but differ. OECD administration figures include system governance but not providers' own administration.

Sources

• Baines D, Hale C. New framework for analysing practice? The Pharmaceutical Journal. 2005;274:56.
• Coase RH. The nature of the firm. Economica. 1937;4(16):386-405.
• Marini G, Street A. The administrative costs of payment by results. CHE Research Paper 17. York: Centre for Health Economics, University of York; 2006.
• Mathauer I, Nicolle E. A global overview of health insurance administrative costs: what are the reasons for variations found? Health Policy. 2011;102(2-3):235-246.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; 2022, last updated 31 March 2026. Sections 4.2 and 4.4.
• OECD. Administrative spending in OECD health care systems: where is the fat and can it be trimmed? In: Tackling Wasteful Spending on Health. Paris: OECD Publishing; 2017. Chapter 6.
• Porter A, Mays N, Shaw SE, Rosen R, Smith J. Commissioning healthcare for people with long term conditions: the persistence of relational contracting in England's NHS quasi-market. BMC Health Services Research. 2013;13(Suppl 1):S2.
• Royal Swedish Academy of Sciences, Economic Sciences Prize Committee. Economic governance. Scientific background on the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2009.
• Schneider JE, Ohsfeldt RL, Scheibling CM, Jeffers SA. Organizational boundaries of medical practice: the case of physician ownership of ancillary services. Health Economics Review. 2012;2:7.
• Williamson OE. Transaction-cost economics: the governance of contractual relations. Journal of Law and Economics. 1979;22(2):233-261.
• Williamson OE. Transaction cost economics: the natural progression. Prize Lecture, 8 December 2009. Stockholm: Nobel Foundation.
• Zamorano P, Espinoza MA, Varela T, Abbott T, Tellez A, Armijo N, Suarez F. Economic evaluation of a multimorbidity patient centered care model implemented in the Chilean public health system. BMC Health Services Research. 2023;23:1041.