Consumer Choice

Consumer Choice Theory and Why Health Care Strains the Standard Model

Consumer choice theory is the starting point for economic thinking about the demand for health care, and a model that health economists routinely qualify. It predicts how use of care responds to the price paid at the point of use and to income, and it gives demand curves their welfare meaning. An illustrative co-payment rise shows the model at work, and the points where it breaks down in health care sit alongside the uses it can still support in economic evaluation.

The standard model of preferences, a budget and the best affordable bundle

The textbook consumer ranks bundles of goods consistently and prefers more to less. Those preferences are represented by a utility function, and the consumer chooses the affordable bundle with the highest utility. In a health application the two goods are units of care, such as primary care visits, and a composite of everything else, so the problem is written as follows.

$$\max_{m,\,x} \; U(m, x) \quad \text{subject to} \quad k\,m + p_x\,x \le Y$$

where $U$ is the utility function, $m$ is the number of visits, $x$ is the quantity of the composite good, $k$ is the out-of-pocket price per visit (the co-payment), $p_x$ is the price of a unit of the composite good and $Y$ is the budget available. The inequality is the budget constraint. With visits on the horizontal axis its slope is $-k/p_x$, which depends on the price the patient pays, not on the full cost of the visit to the insurer or health system.

When both goods are bought and preferences are smooth, the chosen bundle sits where the consumer's own rate of trade-off equals the market rate, as follows.

$$\frac{MU_m}{MU_x} = \frac{k}{p_x}$$

where $MU_m$ is the marginal utility of one more visit and $MU_x$ is the marginal utility of one more unit of the composite good. Solving this condition with the budget constraint gives the demand function below.

$$m^{*} = m(k, p_x, Y)$$

where $m^{*}$ is the utility-maximising number of visits. Tracing $m^{*}$ as $k$ changes gives the demand curve, whose responsiveness is the price elasticity of demand. Under the model's assumptions the height of the demand curve approximates the consumer's marginal valuation of each unit when income effects are small, which is why demand curves are used to measure consumer surplus.

Worked example of a co-payment rise from £20 to £30 per visit

The figures are illustrative. A household has an annual budget of £1,200 for primary care visits and other spending, the composite good costs £10 a unit, and preferences take the Cobb-Douglas form $U(m, x) = m^{0.1} x^{0.9}$, under which the household always spends 10% of its budget on visits. The co-payment rises from £20 to £30 per visit.

1. Choice before the rise. With a co-payment of £20 the number of visits chosen is as follows.

$$m_0 = \frac{0.1 \times 1200}{20} = 6$$

where $m_0$ is the number of visits chosen before the rise, 0.1 is the budget share spent on visits and 1200 is the budget in pounds. Spending on the composite good is the remaining £1,080 and the quantity bought is as follows.

$$x_0 = \frac{0.9 \times 1200}{10} = 108$$

where $x_0$ is the quantity of the composite good and 10 is its price in pounds.

2. Choice after the rise. With a co-payment of £30 the number of visits chosen is as follows.

$$m_1 = \frac{0.1 \times 1200}{30} = 4$$

where $m_1$ is the number of visits chosen after the rise. The composite good stays at 108 units, and the budget line pivots inwards, so the most visits the household could afford falls from 60 to 40.

3. Measure the response. The arc elasticity uses the midpoints of quantity and price, as follows.

$$\varepsilon = \frac{(4 - 6) / 5}{(30 - 20) / 25} = \frac{-0.4}{0.4} = -1$$

where $\varepsilon$ is the arc price elasticity of demand for visits, 5 is the midpoint number of visits and 25 is the midpoint co-payment in pounds. This value is built into Cobb-Douglas preferences and says nothing about how patients actually respond, which is a question for empirical studies of cost sharing.

4. Separate substitution from income. The Slutsky compensating payment, the extra income that would let the household just afford its old bundle at the new price, is calculated as follows.

$$\Delta Y = (30 - 20) \times 6 = 60$$

where $\Delta Y$ is the Slutsky compensating payment in pounds. With a budget of £1,260 at the new price, the household would choose the following bundle.

$$m_c = \frac{0.1 \times 1260}{30} = 4.2$$

where $m_c$ is the number of visits chosen with Slutsky compensation. The fall from 6 to 4.2 visits is the substitution effect of the dearer visit, and the further fall from 4.2 to 4 is the income effect of being poorer. Hicksian compensation, which holds utility rather than the old bundle constant, would be about £49.66 and would lead to about 4.17 visits.

5. Interpret. In the standard reading, each of the two visits given up was worth more than £20 but less than £30 to the household, so the response reveals value. That reading rests on conditions that health care often breaks, namely that the consumer knows the options and their consequences, bears the cost of the choice, chooses without being steered by the seller, faces money price as the main cost, and makes choices as utility maximisation describes. If the household cannot judge what a visit is worth, or the forgone visits were follow-ups with large health benefits, the fall in use carries a health loss that the demand curve does not show.

Arrow's argument on uncertainty, information inequality and trust

Kenneth Arrow's 1963 paper in the American Economic Review is the classic statement of why medical care departs from the competitive model. He argued that the special economic problems of medical care can be explained as adaptations to uncertainty in the incidence of disease and in the efficacy of treatment, and observed that demand for medical services is irregular and unpredictable rather than steady like demand for food or clothing.

Arrow also stressed that the uncertainty differs on the two sides of the transaction. The physician knows much more than the patient about the consequences and possibilities of treatment, both parties know this, and the patient cannot test the product before consuming it, so the relation involves trust. The patient therefore must delegate much of the choice, the root of the principal-agent problem and of the wider concern with information asymmetry. He also noted that widespread medical insurance increases the demand for medical care, which he discussed under the heading of moral hazard. For consumer choice theory the consequence is that observed demand reflects a joint decision of patient and professional.

Grossman and the demand for health instead of medical care

Michael Grossman's 1972 model keeps the consumer-choice framework but changes what is chosen. Individuals inherit a stock of health that depreciates over time and can be increased by investment, which they produce in the household from medical care and their own time, subject to both a money budget and a time budget. In his later Handbook review Grossman puts the central point plainly. What consumers demand when they buy medical services is better health, so the demand for medical care is derived from the basic demand for health.

The reframing widens the price that matters, because the cost of care includes the time it takes, and the Grossman model predicts effects of wages, age and education that a simple price-quantity model would miss. The mechanics are on the demand for health page. Grossman ruled out uncertainty in his formal development, so his model leaves Arrow's concern with uncertainty aside.

Supplier-induced demand, when the agent shifts the demand curve

The standard model treats demand and supply as independent. Supplier-induced demand relaxes that separation. In McGuire's definition, as quoted in Johnson's 2014 review, physician-induced demand exists when the physician influences a patient's demand for care against the physician's interpretation of the patient's best interests. Influence that moves demand towards the patient's own optimum is part of good practice and is not inducement.

Inducement is hard to identify because the benchmark, the care that the physician judges best for the patient, is not observed. Johnson's review describes how early studies relating utilisation to physicians per head could find spurious effects. Dranove and Wehner found that more obstetricians were associated with more births, an outcome physicians do not control. A demand curve estimated from market data may therefore partly reflect provider behaviour.

Behavioural departures in health decisions

Behavioural economics questions whether observed choices reveal the chooser's own interests. Baicker, Mullainathan and Schwartzstein describe a behavioural hazard in health insurance, in which patients may underuse high-value care because of mistakes. They point to the salience of symptoms, since many severe diseases have few salient symptoms, to present bias when costs are paid now and benefits come later, to forgetting to take or refill medicines, and to false beliefs about how well a treatment works.

This is the risk flagged in the worked example. With moral hazard alone, raising co-payments improves the efficiency of demand by reducing overuse. With behavioural hazard as well, raising co-payments can reduce efficiency by deepening underuse, so the demand response alone is no longer enough to set a co-payment and the health response must be considered too. The authors present this as a theoretical basis for value-based insurance design, in which some high-value treatments carry no co-payment.

Patient choice in the NHS without a money price

The NHS Constitution for England states that NHS services are free of charge except in limited circumstances sanctioned by Parliament. For most services, patient choice therefore operates with a money price of zero, and the consumer-choice trade-off runs through waiting time, distance, convenience and perceived quality. The Constitution also gives patients the right to make choices about the services commissioned by NHS bodies and to information to support those choices.

The Department of Health and Social Care's NHS Choice Framework for England, updated in October 2024, describes legal rights to choose a GP practice and, when referred for consultant-led care or to a mental health professional, to choose the provider and clinical team for the first outpatient appointment, with stated exceptions. Patients can see providers' waiting times and Care Quality Commission ratings through the NHS e-Referral Service. In economic terms these arrangements combine consumer choice on non-price attributes with agency through the referral, and the published information addresses information inequality about providers, though not about treatments.

Using consumer choice theory in economic evaluation and HTA

Consumer theory supplies the welfare foundations of cost-benefit analysis, where willingness to pay and compensating variation measure value through the consumer's own trade-offs between health and other consumption. A discrete choice experiment combines random utility theory with consumer theory and assumes, in the characteristics tradition, that an intervention is valued through the levels of its attributes.

Where agency, uncertainty or behavioural hazard are strong, the area under a demand curve can misstate the benefit of a service, one reason many evaluations value health outcomes directly (see welfare economics). In decision models the theory matters most for uptake, adherence and responses to cost sharing, where demand and health responses should be modelled separately.

Boundaries and common misreadings

Consumer choice theory here means the general theory and its application to health care as a whole, while the mechanisms it draws on have their own pages, from utility maximisation and the budget constraint to the Grossman model, supplier-induced demand, moral hazard and cost sharing.

A frequent error is to read any fall in use after a co-payment rise as the removal of low-value care, which follows only if patients can judge the value of the care they give up. A second is to treat one measured price elasticity as a property of health care in general. Arrow's argument is also misread as a rejection of consumer theory. He applied welfare economics to medical care and explained its institutions as responses to uncertainty, so the competitive benchmark remains the yardstick for the departures. With a zero money price, as for most NHS services, the trade-off shifts to time, distance and information, which the model can treat as non-money prices.

Sources

• 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. NBER Working Paper 18468. Cambridge, MA: National Bureau of Economic Research; 2012. https://www.nber.org/papers/w18468 (published in Quarterly Journal of Economics. 2015;130(4):1623-1667)
• de Bekker-Grob EW. Discrete Choice Experiments in Health Care: Theory and Applications. PhD thesis. Rotterdam: Erasmus University Rotterdam; 2009. Chapters 1 and 2. https://repub.eur.nl/pub/21908/
• Department of Health and Social Care. NHS Choice Framework: what choices are available to you in your NHS care. London: DHSC; updated 23 October 2024. https://www.gov.uk/government/publications/the-nhs-choice-framework/the-nhs-choice-framework-what-choices-are-available-to-me-in-the-nhs
• Department of Health and Social Care. The NHS Constitution for England. London: DHSC; updated 17 August 2023. https://www.gov.uk/government/publications/the-nhs-constitution-for-england/the-nhs-constitution-for-england
• Grossman M. The Demand for Health: A Theoretical and Empirical Investigation. NBER Occasional Paper 119. New York: National Bureau of Economic Research; 1972. Chapter 1. https://www.nber.org/books-and-chapters/demand-health-theoretical-and-empirical-investigation
• 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.