Technological Externality

Technological Externality in Health Economics

In welfare economics, a technological externality is the kind of externality in which one person's activity alters what another person can produce or enjoy by a physical or psychological route, without passing through any market price. Infection is the standard health case, because one person's vaccination or treatment changes the chance that other people fall ill. Economists separate such effects from pecuniary externalities, which reach others only by moving prices, because in otherwise competitive markets only the technological kind leaves a real cost or benefit out of the market outcome and so gives a case for correction. For health economists the distinction decides which wider effects belong in a cost-benefit or cost-effectiveness analysis, such as herd protection after vaccination or the spread of antimicrobial resistance.

Real effects and price effects

The distinction began as a correction to Pigou. Liebowitz and Margolis, writing in 1994, recount that Pigou's critics introduced the labels pecuniary and technological to show that the higher payments to inframarginal inputs in a growing industry are rents, not a social cost calling for a tax. Pecuniary externalities work through the price system, so the loss to one side of a market is a gain to the other. With technological externalities, in the authors' words, "actual benefits or costs are imposed outside of market mechanisms", and pollution and congestion are the familiar examples.

Holcombe and Sobel, writing in 2000, state the welfare consequence directly. Technological externalities create resource misallocations while pecuniary externalities do not, in consumption as in production.

Writing the effect into utility and health functions

A technological externality appears when a variable chosen by one person enters another person's utility or production function as an argument. Culyer wrote the caring externality in health care this way in 1971, shown here in compressed form.

$$U_{B} = U_{B}(x_{B}, z_{B}, x_{A})$$

where $U_{B}$ is the utility of person B, $x_{B}$ is the medical care that B consumes, $z_{B}$ stands for B's other goods, and $x_{A}$ is the medical care consumed by person A, a quantity that A chooses and B does not. A utility function with $x_{A}$ inside it means that a change in A's care alters B's wellbeing at unchanged prices and income, which is what makes the effect technological.

The same structure describes infection, with a probability in place of utility.

$$\pi_{B} = \pi_{B}(v_{B}, v_{A}), \qquad \frac{\partial \pi_{B}}{\partial v_{A}} < 0$$

where $\pi_{B}$ is the probability that person B is infected over a season, $v_{B}$ is B's own vaccination or preventive effort, and $v_{A}$ is that of person A. The negative derivative says that A's protection lowers B's risk directly, with no price involved. Gersovitz and Hammer call the effect of a person's own infectiousness the pure infection externality, and the effect of preventive action such as killing mosquitoes, which changes other people's risk whether or not it protects the actor, the pure prevention externality.

Technological externalities in health and health care

Culyer put communicable disease first among the external relationships in health care. He described an individual's failure to be immunised as imposing a higher probability of the disease on the rest of society, a physical externality, and reported general agreement among economists that such activity is best subsidised by government and financed out of general taxation, the case behind vaccination externalities. Coast, Smith and Millar argued that rising resistance to antimicrobials is an important social externality that economic evaluations of strategies for infectious disease had not included as a cost, and found no methodological justification for leaving it out.

The caring externality is a technological effect of a psychological kind. Culyer's last and possibly most important case was that some people, though behaving rationally, may consume too little health care in the opinion of others. Brouwer and colleagues describe this as welfarist economics extended to treat utility interdependence as a kind of externality, used by Culyer to justify subsidised health care. Within extra-welfarism, they explain, a concern about distribution may instead rest on fairness and on non-utility characteristics such as health, so the case for funding another person's care need not run through anyone's utility. Here the technological externality sits at the point where welfare economics and extra-welfarist health economics part company.

Why the technological kind can justify correction

A technological externality leaves a real gain or loss outside the market outcome, a source of market failure that the negative externality and positive externality pages treat in general, so prices alone do not lead decision makers to weigh it. Liebowitz and Margolis note that it may be resolved through property rights, private negotiation or government intervention, whereas for a pecuniary effect the market outcome is already the one that full internalisation would reach. A Pigouvian tax or Pigouvian subsidy is therefore aimed at technological effects only.

A technological externality is still not proof that public action will help. While economists cited bees pollinating apple orchards as a classic externality, Liebowitz and Margolis recall, beekeepers were contracting with orchard owners, the bargaining that the Coase theorem predicts when transaction costs are low. Culyer adds that even Pareto-relevant externalities may be inefficient to remove when the institutions needed to internalise them cost too much. He set out the efficient amount of A's care in the caring case as follows.

$$MV_{A} + MV_{BA} = MC$$

where $MV_{A}$ is A's own marginal valuation of A's care, $MV_{BA}$ is B's marginal valuation of A's care, and $MC$ is the marginal cost of care. Beyond this point marginal cost exceeds the summed valuations, so Culyer concluded that the continuing existence of a marginal externality is not a sufficient warrant for further subsidy.

Counting technological externalities in economic evaluation

In an economic evaluation the perspective decides whose gains and losses are counted, and the NICE reference case (PMG36, section 4.2.7) includes health effects on other people, mainly carers. For infection the deciding choice is model structure, because Brisson and Edmunds showed that a static model misses the herd protection that a dynamic transmission model captures when judging cost-effectiveness.

Cost-benefit analysis values the same effects in money. The HM Treasury Green Book (2026) states that a good with negative externalities is priced too low from a social value perspective and calls for non-market valuation where market prices do not reflect social value. It also states that economic transfers incur a cost to the payer and an equal benefit to the recipient and so do not, in themselves, make society better or worse off. A technological effect is a real social cost or benefit, whereas a pecuniary price change behaves like a transfer.

Worked example with herd protection in an NHS vaccination programme (illustrative)

All numbers in this example are illustrative. An NHS programme vaccinates 100,000 more people against a seasonal infection at GBP 30 a dose including administration. On average over this increase in coverage, each vaccination averts 0.02 infections in the person vaccinated and, by reducing transmission, 0.03 infections among other people. Each infection costs the NHS GBP 400 and loses 0.02 QALYs.

Step 1, programme cost. The programme costs 100,000 × 30 = 3,000,000, or GBP 3.0 million.

Step 2, static model with direct effects only. Infections averted are 100,000 × 0.02 = 2,000, NHS savings are 2,000 × 400 = 800,000, and QALYs gained are 2,000 × 0.02 = 40. The net cost is 3,000,000 − 800,000 = 2,200,000, so the ICER is 2,200,000 / 40 = 55,000, or GBP 55,000 per QALY.

Step 3, the technological externality. Infections averted among other people are 100,000 × 0.03 = 3,000. They save the NHS 3,000 × 400 = 1,200,000 and add 3,000 × 0.02 = 60 QALYs.

Step 4, dynamic model with both effects. Total infections averted are 2,000 + 3,000 = 5,000, savings are 5,000 × 400 = 2,000,000 and QALYs gained are 40 + 60 = 100. The net cost is 3,000,000 − 2,000,000 = 1,000,000, so the ICER is 1,000,000 / 100 = 10,000, or GBP 10,000 per QALY.

Step 5, net health benefit. PMG36 (section 4.2.16) asks for net health benefit at GBP 25,000 and GBP 35,000 per QALY, and the lower value is used here.

$$\text{NHB} = \Delta Q - \frac{\Delta C}{\lambda}$$

where $\Delta Q$ is the QALY gain, $\Delta C$ is the net cost in GBP and $\lambda$ is the value placed on a QALY. At GBP 25,000 the static model gives 40 − 2,200,000 / 25,000, which is 40 minus 88 or a loss of 48 QALYs, and the dynamic model gives 100 − 1,000,000 / 25,000 = 60 QALYs.

Step 6, a pecuniary effect for contrast. Suppose the national order also raises the price that private clinics pay for the same vaccine by GBP 2 a dose on 50,000 doses. The clinics pay 50,000 × 2 = 100,000, or GBP 100,000 more, and the manufacturer receives the same sum, so the social total is unchanged and nothing is added to the programme's cost.

Leaving out the technological externality makes a programme that gains health at GBP 10,000 per QALY look like one costing GBP 55,000 per QALY with a net health loss. A real analysis would take the indirect effect from a transmission model and test it in sensitivity analysis, since infections averted per extra dose fall as coverage rises.

Telling a technological externality from its neighbours

Several effects on third parties look alike but are handled differently. The test is whether the effect changes someone's health, wellbeing or output at given prices, and whether the analysis already counts it.

A price change caused by a programme's purchases is pecuniary, as in Step 6. Pooled insurance is a different channel. Bhattacharya and Sood show that when premiums are not rated for weight, people with obesity use more medical care but pay the same premium, imposing a negative externality on others in the pool that disappears when underwriting on weight is allowed. The link is a shared premium, not a direct effect on anyone's health. Carer health effects counted under PMG36 section 4.2.7 are technological in nature but already inside the analysis, so adding them again as an external benefit would double count.

Common misreadings of technological externalities

The term is easy to apply too widely. Holcombe and Sobel argue that, on a household production view of consumption, many external effects on people's utility functions are pecuniary externalities that public policy should ignore, so a claimed caring effect needs the same scrutiny as a physical one.

The opposite error treats every pecuniary effect as harmless. Mirza and Stephens model buyers who ignore the effect of their health insurance purchases on the future price of care, leading to over-insurance and an excessive price of care, so where moral hazard already distorts a market the rule that only technological effects matter need not hold.

Inside evaluations, a dynamic model has already counted herd protection, so a separate herd-immunity bonus on top would count it twice. Finding an external benefit does not show that a subsidy pays, because Culyer's condition turns on marginal valuations at the current level of care.

Sources

• Bhattacharya J, Sood N. Health insurance and the obesity externality. NBER Working Paper 11529. National Bureau of Economic Research; July 2005. https://www.nber.org/papers/w11529
• Brisson M, Edmunds WJ. Economic evaluation of vaccination programs: the impact of herd-immunity. Medical Decision Making. 2003;23(1):76-82. https://doi.org/10.1177/0272989X02239651
• Brouwer WBF, Culyer AJ, van Exel NJA, Rutten FFH. Welfarism vs. extra-welfarism. Journal of Health Economics. 2008;27(2):325-338. https://doi.org/10.1016/j.jhealeco.2007.07.003
• Coast J, Smith RD, Millar MR. Superbugs: should antimicrobial resistance be included as a cost in economic evaluation? Health Economics. 1996;5(3):217-226. https://pubmed.ncbi.nlm.nih.gov/8817296/
• Culyer AJ. The nature of the commodity 'health care' and its efficient allocation. Oxford Economic Papers. 1971;23(2):189-211. https://www.jstor.org/stable/2662234
• Gersovitz M, Hammer JS. The economical control of infectious diseases. Policy Research Working Paper 2607. World Bank; May 2001. https://openknowledge.worldbank.org/entities/publication/a250f5f6-7f35-58af-b746-2b42fadef054
• HM Treasury. The Green Book (2026). Updated 28 October 2026. https://www.gov.uk/government/publications/the-green-book--2/the-green-book-2026
• Holcombe RG, Sobel RS. Consumption externalities and economic welfare. Eastern Economic Journal. 2000;26(2):157-170. https://ideas.repec.org/a/eej/eeconj/v26y2000i2p157-170.html
• Liebowitz SJ, Margolis SE. Network externality: an uncommon tragedy. Journal of Economic Perspectives. 1994;8(2):133-150. https://doi.org/10.1257/jep.8.2.133
• Mirza A, Stephens E. On the over-provision of medical insurance. Carleton Economics Working Paper CEWP 25-02. Carleton University; March 2025. https://carleton.ca/economics/wp-content/uploads/sites/112/cewp25-02.pdf
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 13 October 2026. Sections 4.2.7 and 4.2.16. https://www.nice.org.uk/process/pmg36