Demand

Demand in the Use of Health Care

In economics, demand is a whole schedule, the amount people would buy at every possible price, with income, other prices and preferences held fixed, as distinct from the number of services used last year. Health economists use it to predict how use of care responds to charges, insurance and income, and to value gains and losses when those change. Health care strains the idea, because people want health rather than care, insurance means the patient seldom pays the full price, professionals often choose on the patient's behalf, and recorded use can differ from both demand and need. Demand in health care therefore has to be read alongside derived demand, insurance, need and utilisation, supplier influence and behavioural hazard, and an illustrative diagnostic test under three insurance designs shows how these fit together, while the demand curve, price elasticity of demand, consumer choice and consumer surplus pages hold the detail.

The demand function and its determinants

Demand joins willingness to pay, which comes from preferences, to ability to pay, which comes from income. In consumer theory the demand function is the solution to the consumer's choice of the best bundle within a budget constraint, which is why prices and income enter it together (see utility maximisation). For a health service it can be written as follows.

$$Q_d = D(p, Y, p_s, p_c, h, z)$$

where $Q_d$ is the quantity demanded per period, $p$ is the price the user pays for the service, $Y$ is income, $p_s$ and $p_c$ are the prices of substitutes and complements, $h$ is health status, and $z$ collects other influences such as preferences, information, insurance terms and the time needed to obtain care.

Varying $p$ with everything else fixed traces the demand curve, so a change in the service's own price is a movement along the curve and a change in any other argument shifts the whole curve. The responses are summarised by the own-price elasticity, the income elasticity of demand and the cross-price elasticity, and market demand is the sum of individual demands at each price. Health status is the argument that sets health care apart. Kenneth Arrow observed in 1963 that demand for medical services is not steady, as demand for food or clothing is, but irregular and unpredictable, because it arises from illness.

Out-of-pocket charges and time as the price that matters

The price in the demand function is the one the patient pays at the point of use. That is usually far below the cost of the care.

Under coinsurance the patient's price is a fixed share of the full price, as below.

$$p = c \times P$$

where $p$ is the out-of-pocket price, $c$ is the coinsurance rate (the share the patient pays, between 0 and 1), and $P$ is the full price or cost of the service. Whether a change in insurance appears as a movement along the curve or a rotation of it depends on which of these prices is on the vertical axis, as the demand curve page explains. Deductibles and out-of-pocket maximums under cost sharing make $p$ change during the year.

Where care is free at the point of use, travel, time and waiting act as the price that rations use. Michael Grossman's model of the demand for health likewise treats the patient's own time as an input into producing health, alongside medical care.

Derived demand and the wish for health rather than care

Grossman's human capital model explains why demand for care behaves differently from demand for most goods. In his Handbook review he states that what consumers demand when they buy medical services is better health, so the demand for medical care and other health inputs is derived from the basic demand for health. Care is an input that people combine with their own time to produce health, as the demand for health and derived demand for healthcare pages set out.

More use of care therefore does not mean more health. Grossman also shows that, under certain conditions, a rise in what he calls the shadow price of health, for instance from faster depreciation of health with age, can reduce the health demanded while increasing the medical care demanded. He ruled out uncertainty in his formal development, so the Grossman model leaves aside the uncertainty that Arrow put at the centre of medical care.

Insurance, cost sharing and the evidence on price response

Insurance lowers $p$, and a lower price raises quantity demanded. Arrow noted that widespread medical insurance increases the demand for medical care and that coinsurance had been introduced into many policies partly to meet this. The economic reading is moral hazard. Insured patients use some care whose value to them is below its cost.

The RAND Health Insurance Experiment is still widely regarded as the gold standard of evidence on how health insurance reforms affect medical spending. People who paid a share of their care used fewer services than those given free care and, in RAND's summary, cost sharing reduced highly effective and less effective services in roughly equal proportions. Its famous estimate that the elasticity of medical spending with respect to out-of-pocket price is $-0.2$ is the figure most often attached to the price elasticity of healthcare, but Aron-Dine, Einav and Finkelstein showed that it was derived from the experimental data combined with further modelling and statistical assumptions, and cautioned against summarising non-linear insurance contracts with a single price elasticity. How such experiments are turned into demand curves is covered on the demand curve page.

Need, demand and utilisation are different quantities

Health service planning separates ideas that economic data often blur. Wright, Williams and Wilkinson, opening a BMJ series on health needs assessment, describe need in health care as commonly defined as the capacity to benefit, which requires an effective intervention, demand as what patients ask for, and supply as the care provided. The three overlap only partly.

People can need care without demanding it. The authors note that many people with angina or multiple sclerosis are not known to their general practitioner or a specialist, and that homeless people and people with chronic mental illness may need care but not demand it. Demand can also be induced by supply, since geographical variation in hospital admission rates is explained more by the supply of beds than by mortality indicators.

Utilisation is the care actually used, which needs both demand and supply, and it is what administrative data record, so a utilisation rate reflects demand, supply and rationing together. Equity principles such as horizontal equity are written in terms of need, not demand or use. The planners' "demand" is also narrower than the economist's. It is the quantity asked for at the prevailing price and access conditions, one point on the schedule.

Supplier influence through agency and induced demand

Arrow pointed out that the cost of medical care is not fixed by the illness alone but depends on the choice of a doctor and the doctor's willingness to use medical services. The physician knows much more than the patient about the consequences and possibilities of treatment, both know it, and the patient delegates much of the choice, the principal-agent problem in healthcare that grows out of information asymmetry. Observed demand for care is in practice a joint decision of patient and professional.

Supplier-induced demand is the case where that agency works against the patient. Johnson's review quotes McGuire's definition, under which physician-induced demand exists when the physician influences a patient's demand for care against the physician's own interpretation of the patient's best interests. Advice that moves demand towards the patient's optimum is part of the physician's job and is not inducement. The two are hard to separate because the right care for each patient is not observed; Johnson describes how Dranove and Wehner, applying a method similar to an earlier study of surgeon numbers and surgery rates, found that more obstetricians went with more births, an outcome physicians do not control. Johnson calls this a powerful critique of the empirical methodology. As a falsification test, it undermined the use of physician density to detect inducement.

Behavioural hazard and demand that does not measure value

Standard theory reads the height of the demand curve as the value of each unit to the user, which is what lets demand curves measure welfare. Baicker, Mullainathan and Schwartzstein argue that people also misuse care through mistakes, which they call behavioural hazard, so that much high-value care is underused even when patient costs are low and some useless care is bought even at full cost; welfare calculations from the demand curve alone can then be off by orders of magnitude or have the wrong sign. The consumer choice page describes the behavioural mechanisms involved.

Worked example with one diagnostic test under three insurance designs

The figures are illustrative. A district compares no insurance, 20% coinsurance and free care for one diagnostic test, and each test costs the health system £30.

Annual demand for the test depends on the out-of-pocket price as follows.

$$Q(p) = 2000 - 40p$$

where $Q(p)$ is the number of tests demanded per year, $p$ is the out-of-pocket price in pounds, 2,000 is the number demanded at a zero price, and 40 is the fall in tests for each £1 rise in price.

1. Quantity under each design. With no insurance the patient pays £30; with 20% coinsurance the patient pays 0.2 × 30 = £6; with free care the patient pays nothing. The quantities demanded are as follows.

$$Q(30) = 2000 - 40 \times 30 = 800, \quad Q(6) = 2000 - 40 \times 6 = 1760, \quad Q(0) = 2000$$

where each term is the number of tests per year at that out-of-pocket price. All three points lie on the same demand curve.

2. Price response. Between no insurance and 20% coinsurance the arc elasticity uses the midpoints of quantity (1,280) and price (£18), as follows.

$$\varepsilon = \frac{(1760 - 800)/1280}{(6 - 30)/18} = \frac{0.75}{-1.333} \approx -0.56$$

where $\varepsilon$ is the arc price elasticity of demand for the test over this range.

3. Spending. Without insurance patients spend 800 × 30 = £24,000. With 20% coinsurance total spending is 1,760 × 30 = £52,800, of which patients pay 1,760 × 6 = £10,560 at the point of use and the insurer £42,240. In expectation the insurer's share is paid back by the insured through premiums, so insurance changes the price at the point of use, not who ultimately pays.

4. The standard welfare reading. On the demand curve, patients value the extra 960 tests at between £6 and £30, or £18 on average, so 960 × 18 = £17,280 in total, against a resource cost of 960 × 30 = £28,800. The shortfall of £11,520, equal to (30 minus 6) × 960 / 2, is the deadweight loss that the moral hazard account attributes to 20% coinsurance. Under free care the extra use is 1,200 tests (2,000 less 800) and the triangle is 1/2 × 30 × 1,200 = £18,000. Each triangle is set against the value of the protection from financial risk that insurance provides, which it does not count, so it measures the cost side of insurance, not its net value.

5. If demand understates value. Suppose clinical evidence showed that the extra tests find disease with few symptoms and that their health benefit is worth £40 per test on average. Their value is then 960 × 40 = £38,400, which exceeds the cost by £9,600. The demand curve shows a loss of £11,520 and the health-based account a gain of £9,600, the change of sign that Baicker and colleagues warn about.

6. Interpretation. The demand curve predicts use and spending under each design whatever the value of the tests, so it suits forecasting and budgeting. Whether the extra use is waste needs evidence on health effects as well.

Demand in economic evaluation, budget impact and HTA

Demand enters evaluation in two roles, as a forecast of use and as a measure of value. The forecasting role is the less contested. The ISPOR good practice task force on budget impact analysis lists among the key elements the current mix of treatments and the expected mix after a new intervention is introduced, which is an uptake question for demand forecasting and budget impact analysis.

The valuation role belongs to cost-benefit analysis, where the area under a demand curve above the price paid gives consumer surplus, alongside willingness to pay from stated preference studies. Cost-utility analysis values outcomes directly in QALYs, so it does not use demand curves to value care, but uptake and adherence in its models are still demand responses.

Demand beside its neighbouring concepts

Several nearby terms are often used as if they meant demand. The table sets out what each describes.

Common misreadings of demand in health care

Several errors recur when demand is used in health economics. Treating utilisation data as demand ignores the waiting lists, gatekeeping and supply limits that cut use below what people would choose at the going price, and treating demand as need ignores people who need care but do not seek it. Reading a fall in use after a charge rises as the removal of low-value care conflicts with the RAND finding that highly effective and less effective care fell in roughly equal proportions.

Sources

• Aron-Dine A, Einav L, Finkelstein A. The RAND Health Insurance Experiment, three decades later. Journal of Economic Perspectives. 2013;27(1):197-222. doi:10.1257/jep.27.1.197. https://pmc.ncbi.nlm.nih.gov/articles/PMC3943162/
• Arrow KJ. Uncertainty and the welfare economics of medical care. American Economic Review. 1963;53(5):941-973. https://www.aeaweb.org/aer/top20/53.5.941-973.pdf
• Baicker K, Mullainathan S, Schwartzstein J. Behavioral hazard in health insurance. Quarterly Journal of Economics. 2015;130(4):1623-1667. Working paper version: NBER Working Paper 18468; 2012. https://www.nber.org/papers/w18468
• Brook RH, Keeler EB, Lohr KN, Newhouse JP, Ware JE, Rogers WH, et al. The Health Insurance Experiment: a classic RAND study speaks to the current health care reform debate. Research Brief RB-9174-HHS. Santa Monica, CA: RAND Corporation; 2006. https://www.rand.org/pubs/research_briefs/RB9174.html
• Grossman M. The human capital model. In: Culyer AJ, Newhouse JP, eds. Handbook of Health Economics. Vol 1A. Amsterdam: Elsevier; 2000:347-408.
• Johnson EM. Physician-induced demand. In: Culyer AJ, ed. Encyclopedia of Health Economics. Vol 3. San Diego: Elsevier; 2014:77-82.
• Sullivan SD, Mauskopf JA, Augustovski F, et al. Budget impact analysis: principles of good practice. Report of the ISPOR 2012 Budget Impact Analysis Good Practice II Task Force. Value in Health. 2014;17(1):5-14. doi:10.1016/j.jval.2013.08.2291
• Wolitzky A. Lectures 3-4: consumer theory. 14.121 Microeconomic Theory I, Fall 2015. Cambridge, MA: MIT OpenCourseWare; 2015. https://ocw.mit.edu/courses/14-121-microeconomic-theory-i-fall-2015/
• Wright J, Williams R, Wilkinson JR. Development and importance of health needs assessment. BMJ. 1998;316(7140):1310-1313. doi:10.1136/bmj.316.7140.1310