Behavioural Economics

Why this matters in health economics

Almost every model in this book — demand, insurance, priority-setting — rests, somewhere underneath, on a picture of a rational chooser who knows their own preferences, weighs costs against benefits, and acts consistently over time. That picture is a useful fiction. It is also, in health, a dangerous one. People do not fill the prescriptions that would keep them alive; they intend to quit smoking next month for a decade; they skip the free screening and buy the extended warranty. If your policy assumes the textbook chooser, it will over-predict prevention uptake, mis-price insurance, and waste money on information campaigns that inform the already-converted.

The stakes are concrete. A large share of the burden of disease worldwide is driven by behaviour — tobacco, diet, alcohol, physical inactivity, and failure to adhere to treatment — and behaviour is exactly where the rational model performs worst. When a diabetic patient stops taking medication that costs pennies and prevents amputations costing thousands, no amount of price theory explains it, but a small tweak to how the choice is presented sometimes fixes it. Behavioural economics is the discipline that studies these systematic departures from rationality and, crucially, how to design around them.

For a director, this matters in two directions. First, as a critique: it tells you where the standard demand and evaluation models (see Chapter 1.2 — Demand for Health and Healthcare) will mislead you, so you discount their predictions appropriately. Second, as a toolkit: it offers low-cost, often highly cost-effective levers — defaults, framing, reminders, well-designed incentives — that can shift behaviour where exhortation and even price have failed. Both come with an ethical edge, because designing choices for people who are not fully rational is a kind of power, and the chapter that owns nudging must also own its limits.

Core concepts

Behavioural economics. Behavioural economics is the study of how psychological, cognitive, and social factors shape economic decisions, and how those decisions systematically depart from the predictions of the rational-choice model. It does not claim people are stupid; it claims they are human — using mental shortcuts that work well most of the time and fail in patterned, predictable ways. In health, those patterns matter enormously because the decisions are high-stakes, infrequent, emotionally loaded, and made under uncertainty.

Bounded rationality. The foundational idea, from Herbert Simon, is bounded rationality: real decision-makers have limited attention, information, and computational power, so instead of optimizing they satisfice — they settle for an option that is good enough. People do not read every insurance clause or compare every provider; they pick the default, the familiar, or the first acceptable option. This is not a flaw to be scolded away but a fixed feature of human cognition that system design must accommodate.

Heuristics and biases. To cope with bounded rationality, people use heuristics — rules of thumb — which produce reliable, direction-predictable errors called biases. Availability makes vivid risks (a plane crash, a rare vaccine side-effect) loom larger than common ones (heart disease); status quo bias makes whatever is currently set feel right. Much of the field is a catalogue of these biases and the situations that trigger them.

Loss aversion and framing. People weigh losses more heavily than equivalent gains — loss aversion, a central plank of prospect theory. As a result, the same fact changes behaviour depending on how it is presented: this is framing. "Nine out of ten patients survive this operation" and "one in ten dies" are logically identical and psychologically worlds apart. Screening messages framed as avoiding a loss often outperform those framed as securing a gain.

Present bias and time-inconsistency. People systematically over-weight the present against the future — present bias. The immediate cost of the gym, the injection, or the salad is felt now; the benefit is distant and abstract. Worse, preferences are time-inconsistent: the plan you make for next week (I will quit) is reversed when next week becomes now. This is the behavioural mechanism behind the chronic under-investment in prevention that the Grossman model's rational investor cannot explain, and it is why willpower-based interventions so often fail.

Defaults and the default effect. Because of inertia and status quo bias, whatever happens when a person does nothing — the default effect — exerts enormous influence. The most cited example is organ donation: countries with an opt-out (presumed-consent) default, such as Austria, register far higher consent rates than otherwise-similar opt-in countries, such as Germany, despite similar underlying attitudes. Defaults are powerful precisely because most people accept them.

Choice architecture and nudges. Any environment in which people make decisions has a design, whether or not anyone designed it deliberately; the person who shapes it is the choice architect. A nudge — the term popularised by Richard Thaler and Cass Sunstein — is any aspect of that architecture that predictably alters behaviour without forbidding options or significantly changing economic incentives. Placing fruit at eye level nudges; banning chocolate does not. The line is that a nudge must be easy and cheap to avoid.

Libertarian paternalism. Thaler and Sunstein's justification for nudging is libertarian paternalism: because some choice architecture is unavoidable, it may as well be arranged to help people achieve what they themselves say they want (paternalism), while preserving their freedom to choose otherwise (libertarian). This is the contested philosophical heart of the field, and it is where the ethics of nudging live.

Best practices

1. Design the default first, because it will do most of the work. In any process where inaction is possible — enrolment, appointment booking, repeat prescriptions — decide deliberately what happens when the person does nothing, because inertia means most will take that path. Set the default to the option that serves the person's own stated goals and public health, while leaving the alternative genuinely easy to reach. Organ-donation registration, pension-style automatic enrolment in screening recall, and opt-out rather than opt-in flu vaccination for staff all exploit this lever ethically.

2. Frame health information around losses and concrete, near-term stakes. Because losses loom larger than gains and the present outweighs the future, translate distant, abstract benefits into immediate, tangible, loss-framed terms wherever the evidence supports it. "Miss this screening and you risk finding cancer too late" often moves more people than "screening keeps you healthy." Test frames rather than assuming; the effect sizes are modest and context-dependent, and framing that manipulates rather than informs crosses an ethical line.

3. Attack present bias with immediacy, not information. When the barrier is time-inconsistency — the intention is there but the follow-through fails — add structure that brings the future forward: reminders at the moment of action, appointments pre-booked rather than requested, small immediate rewards, and commitment devices people opt into voluntarily. A text-message reminder timed to the exact prescription-refill date beats a leaflet about long-term consequences, because the leaflet fights present bias while the reminder sidesteps it.

4. Reduce friction ruthlessly for the behaviour you want, and add it for the one you do not. Every extra form, click, or queue is a filter that removes people, including the ones you most want to keep. Make the healthy or recommended action the path of least resistance — one-step booking, co-located services, pre-filled forms — and, where appropriate, make the harmful action slightly harder, as with placing tobacco out of sight. Friction is a more reliable lever than motivation.

5. Use incentives, but design them for how people actually respond. Financial incentives for health behaviour — smoking cessation, attendance, adherence — can work, but the behavioural literature shows the design matters more than the amount. Small, frequent, near-certain rewards often beat larger, delayed, or probabilistic ones because they counter present bias; deposit-contract schemes that put the person's own money at risk exploit loss aversion. Be honest that effects are frequently modest and can fade or even backfire once the incentive stops, so plan for what sustains the behaviour afterwards.

6. Beware of crowding out intrinsic motivation. Paying people to do what many would have done anyway can convert a moral or health decision into a market transaction and reduce the underlying motivation — a well-documented risk with blood donation and some healthy-behaviour schemes. Before attaching money to a behaviour, ask whether the population is already intrinsically motivated, and prefer recognition, feedback, or friction changes where they are. When you do pay, watch for the signal it sends.

7. Segment by the bias, not just by the demographic. Different people fail to act for different behavioural reasons: some forget (attention), some intend but delay (present bias), some are deterred by complexity (bounded rationality), some are anchored to a default. Diagnosing which mechanism dominates in your population tells you which lever to pull — a reminder, a commitment device, a simplification, or a default change — and avoids the waste of applying the wrong one.

8. Simplify choices; more options is not more welfare. Bounded-rational people faced with too many options defer, default, or choose badly — the paradox of choice seen in over-complex insurance markets and drug formularies. Curate a small, well-ordered set of good options, order them to help rather than to sell, and pre-select sensible defaults. In systems where citizens choose insurers or plans, decision support and smart defaults measurably improve the choices people make.

9. Pilot, measure, and be willing to find no effect. Nudges are cheap to build and easy to over-sell; the honest ones are tested against a control, ideally by randomization, before they scale (see Chapter 2.3 — Health Econometrics). Many published nudge effects are small, context-specific, or fail to replicate, so treat every intervention as a hypothesis. A behavioural unit that never reports a null result is marking its own homework.

10. Publish the intent and preserve the exit. The ethical test Thaler and Sunstein set is transparency and easy opt-out: a nudge should survive being described openly to the people it targets, and it must never trap them. Announce that a default exists and how to change it; make the alternative one step away. Covert manipulation, dark patterns, and nudges that are hard to escape are not in this toolkit — they are the abuse of it.

11. Treat the standard rational-choice model as a critiqued baseline, not a truth. When you inherit a demand forecast, a cost-effectiveness model, or an insurance design built on the rational chooser (see Chapter 1.2 — Demand for Health and Healthcare and Chapter 1.3 — Market Failure), ask explicitly where behavioural realism would change the answer. Prevention uptake will usually be lower, cost-sharing will deter valuable care more than a rational model predicts, and complex choices will be made worse than assumed. Adjust the inputs, and say you have done so.

12. Match the tool to the size of the problem — nudges are not a substitute for structural action. A nudge is a scalpel, not a lever for moving populations off tobacco or reshaping food environments; those need tax, regulation, and reformulation (see Chapter 3.2 — Health Policy). Use behavioural design to complement structural policy, not to let it off the hook, and resist the political temptation to reach for a cheap nudge when the honest answer is a harder, costlier intervention.

Questions to discuss with your team

1. Where in our system does "doing nothing" quietly decide the outcome — and have we chosen that default on purpose? Every enrolment, recall, consent, and renewal process has a default, and inertia means most people accept it, yet in most organizations no one has deliberately chosen what it is. Map the points where a citizen or patient does nothing and see what happens to them: are they enrolled in screening recall or dropped from it, opted into data-sharing or out, defaulted to the cost-effective medicine or the expensive one? The honest discussion admits where the current default was inherited by accident rather than designed, and asks whether it serves the person's own goals. It also names the ethical guardrail — that the alternative must stay easy — so you are setting a helpful default, not trapping people. The pay-off is that a single, well-chosen default often outperforms years of leaflets.

2. When we say a health behaviour is a "personal choice", are we describing a free choice or a predictable failure we have declined to design around? It is comfortable to attribute non-adherence, missed appointments, or unhealthy eating to individual choice, because it absolves the system of responsibility. Behavioural economics reframes many of these as predictable products of present bias, friction, and choice architecture that we built or tolerated. The real tension is between respecting autonomy and admitting that we shape behaviour whether we intend to or not. An honest answer distinguishes genuinely informed, stable preferences — which we should respect — from time-inconsistent, friction-driven failures that people themselves regret, which we can ethically help with. It also forces the equity question: friction and complexity fall hardest on the poorest and least-well, so leaving the architecture unexamined is not neutral.

3. What would make us confident a nudge or incentive actually worked, and are we willing to find out it did not? Nudges are cheap, popular, and easy to claim credit for, which makes them dangerously easy to adopt on faith. Ask what evidence would convince you: a randomized pilot with a real control, a pre-registered outcome, a follow-up after the incentive stops to check the effect does not evaporate or reverse. The tension is that rigorous evaluation is slower and risks embarrassing a favoured scheme, while acting on untested behavioural intuition is fast and flattering. An honest answer commits to measuring against a counterfactual and to reporting null results, and it separates the behaviours where a small nudge is plausibly enough from those that need structural policy instead. It also budgets for the ethical review that asks whether the intervention would survive being described openly to those it targets.

4. How do we frame the health messages we already send — and would we be comfortable if patients saw the reasoning behind the wording? Every recall letter, screening invitation, and consent leaflet already frames a choice, usually by accident and often badly, defaulting to abstract benefit ("screening keeps you healthy") when a concrete, near-term, loss-framed message might move more people. The tension is that the frames that shift behaviour most are precisely the ones that edge closest to manipulation, so the discussion has to hold effectiveness and honesty together rather than trading one for the other. A serious answer asks whether the frame informs a real risk or merely exploits a fear, and it commits to testing frames rather than assuming, because effect sizes are modest and context-dependent. It also confronts the reading-age and language question, since a frame that lands with a confident professional may bewilder or frighten someone else. The honest test is Thaler and Sunstein's: would this wording survive being explained openly to the person receiving it, alongside the evidence for why we chose it?

5. When is a nudge honestly the right tool, and when are we reaching for it because the real fix is harder and costlier? Nudges are cheap, quick, and politically flattering, which makes them tempting exactly when the problem is too big for them — moving a population off tobacco or reshaping a food environment needs tax, regulation, and reformulation, not a better leaflet. The tension is between doing something visible now and doing the structural thing that works but costs political capital and money. An honest answer sizes the problem before choosing the tool, treats behavioural design as a complement to structural policy rather than a substitute for it, and names openly when a nudge is being used as a fig-leaf for inaction (see Chapter 3.2 — Health Policy). It also asks who benefits from the cheaper option being chosen, because "we nudged" can let a system avoid a fight it should have had. The equity angle matters too: structural levers tend to reach everyone, while nudges can widen gaps if they work best for the already-advantaged.

6. Before we attach money to a behaviour, have we checked we are not paying people to do what they would have done anyway? Financial incentives for attendance, adherence, or healthy behaviour can work, but the behavioural literature warns that paying for an act done from intrinsic or moral motivation can crowd that motivation out and leave you worse off once the money stops. The tension is that incentives are easy to authorize and easy to measure, while the motivation they might erode is invisible until it is gone — as documented with blood donation and some healthy-behaviour schemes. A candid discussion asks first whether the target population is already motivated, prefers feedback, recognition, or a friction change where it is, and turns to cash only where the barrier is genuinely material. If you do pay, the design questions follow: small, frequent, near-certain rewards to counter present bias, a plan for what sustains the behaviour after the incentive ends, and honesty that effects are often modest and can fade or reverse. It also weighs the signal the payment sends, because putting a price on a health act quietly reframes it as a transaction.

In practice: a health economics example

Scenario: a fictional national health service in a middle-income Latin American country, "Costa Verde", tackling poor adherence to hypertension and diabetes medication.

The Ministry of Health of Costa Verde has a quiet epidemic. Its public system provides antihypertensive and diabetes medicines free at pharmacy counters, yet audits show that within a year of starting treatment, a large fraction of patients have stopped collecting their repeats. The clinical consequence — strokes, heart attacks, amputations, and dialysis — is both a human tragedy and a fiscal one, since each avoidable complication costs the system many times the price of the pills. The standard economic reading is baffling: the medicine is free and the benefit is large, so a rational patient should adhere. The Ministry's newly formed behavioural insights team is asked to explain the gap and to design something cheaper than more clinics.

The team starts by diagnosing the mechanism rather than assuming one. Interviews and pharmacy data reveal not one problem but several: many patients simply forget the refill date (attention); many feel well and see no immediate reason to keep taking a drug for a symptomless condition, discounting a distant stroke against the small daily hassle (present bias and time-inconsistency); some find the monthly queue at a distant pharmacy a real deterrent (friction); and a subset never intended to continue but felt unable to say so to the clinician (weak agency and social pressure). Crucially, these call for different levers, so a single blunt intervention would waste money on the wrong sub-groups.

The team designs a bundle and, importantly, insists on testing it. For the forgetters, an automated text-message reminder timed to the exact refill date, framed around a concrete near-term stake ("your next month's protection is ready to collect") rather than distant risk. For the present-biased, a shift from monthly to three-monthly dispensing where clinically safe, cutting the number of present-cost moments at which adherence can lapse, plus a default of automatic repeat authorization so that continuing requires no action and stopping is the deliberate step. For friction, community-pharmacy and workplace collection points closer to home. They deliberately avoid a large cash incentive, judging the population partly intrinsically motivated and the crowding-out risk real, and choosing instead a small loss-framed element: patients who enrol in the adherence programme and lapse receive a follow-up call, making the lapse visible.

They roll it out as a randomized pilot across matched districts, with a genuine control arm, and pre-commit to measuring collection rates, clinical outcomes where feasible, and cost per additional adherent patient (see Chapter 2.3 — Health Econometrics). The Ministry's health economist is explicit about the trade-offs and the uncertainty: three-monthly dispensing raises wastage if patients stop, the default change must be paired with clear opt-out so it does not become a trap, and the behavioural literature warns that effects may be modest and may fade. The recommendation is therefore not "nudges instead of clinical care" but a tested, low-cost behavioural layer on top of it, scaled only where the pilot shows a real effect against the control — and paired with the harder structural conversation about salt regulation and food environments that no reminder can substitute for (see Chapter 3.2 — Health Policy).

Four sector lenses

Startup

A digital health start-up — an adherence app, a symptom tracker, a behaviour-change coach — is choice architecture made of code, and every screen is a nudge whether the founders admit it or not. Its agility lets it A/B test framings, defaults, and reminder timings at a speed no ministry can match, which is a genuine strength for finding what works. The danger is that the same tools that improve health can become engagement dark patterns that serve retention metrics over the user's stated goals, and investors reward the former. A responsible start-up sets its defaults to the user's health interest, keeps opt-out one tap away, and resists the temptation to confuse a behaviour it can move with a behaviour worth moving.

Small business

A small but established provider — a general-practice partnership, a single community pharmacy, a care home, a specialist clinic — already shapes patient behaviour through the everyday architecture of its front desk, its recall letters, and how its staff frame advice. Unlike a start-up, it is not chasing growth or A/B testing at scale; its strength is a steady relationship with a known population and the standing to make a small default change and keep it. The practical levers are modest but reliable: switching repeat-prescription authorization and appointment recall to opt-out, timing reminders to the moment of action, simplifying a confusing form, and placing the healthy option in the path of least resistance. It rarely has an evaluation budget, so it should borrow tested designs from national behavioural units rather than invent its own, and watch that a small incentive or a stern letter does not sour the trust that is its main asset.

Enterprise

A large provider or insurer applies behavioural design at portfolio scale, where small effect sizes multiplied across millions of members become material. Insurers use defaults in plan enrolment, framing in benefit communication, and incentives in wellness programmes, and must weigh the reputational and regulatory risk of anything that looks like manipulation or that shifts cost onto the unwary. A hospital group redesigns appointment booking, discharge instructions, and staff-vaccination defaults to reduce no-shows and readmissions. Accountability here is to regulators, actuaries, and boards, so behavioural interventions must be evaluated, auditable, and defensible — and the enterprise must be alert that incentives can crowd out the professional and intrinsic motivation of its own clinical staff.

Government

A ministry or national payer is the largest choice architect of all, and its defaults — organ-donation consent, screening recall, pension-style auto-enrolment, formulary defaults — shape behaviour across an entire population, so a single design choice carries population-scale equity consequences (see Chapter 3.4 — Equity). Government can also do what firms cannot: pair nudges with the structural levers of tax, regulation, and reformulation that move whole populations (see Chapter 3.2 — Health Policy). Its accountability is democratic, which raises the ethical bar: nudges deployed by the state must be transparent, publicly justified, and escapable, because the citizen cannot easily switch away from their government. Many governments now run dedicated behavioural units, following the model of the United Kingdom's Behavioural Insights Team, precisely to bring evidence and evaluation to these choices.

Common failure modes

• Assuming the rational chooser. Building demand forecasts, prevention business cases, or insurance designs on a far-sighted, consistent decision-maker, then being surprised when uptake and adherence disappoint. Fix: treat the rational model as a critiqued baseline; adjust prevention and cost-sharing predictions for present bias and friction, and say you have.

• Nudge as fig-leaf. Reaching for a cheap behavioural intervention to be seen to act, when the problem needs tax, regulation, or structural reform. Fix: match the tool to the scale of the problem; use nudges to complement structural policy, not to excuse its absence.

• Un-evaluated confidence. Scaling a nudge or incentive because it is plausible and popular, with no control group and no willingness to report a null. Fix: pilot against a counterfactual, pre-register outcomes, follow up after incentives stop, and publish what fails.

• Crowding out. Attaching cash to a behaviour people were doing for intrinsic or moral reasons, and eroding the motivation. Fix: check for existing intrinsic motivation first; prefer feedback, recognition, and friction changes; when you pay, watch the signal.

• Covert manipulation. Using hidden defaults, dark patterns, or hard-to-escape architecture that would not survive being described openly. Fix: apply the transparency-and-easy-exit test; announce the default and how to change it; keep the alternative one step away.

• One lever for every problem. Sending everyone a reminder when some forget, some delay, and some are deterred by complexity. Fix: diagnose the dominant behavioural mechanism before choosing the intervention.

Maturity model

Checklist

• Map every point where a patient or citizen doing nothing determines the outcome, and choose each default deliberately.
• For each behavioural problem, diagnose the dominant mechanism — attention, present bias, friction, complexity, or default — before selecting a lever.
• Frame health messages around concrete, near-term, loss-framed stakes where the evidence supports it, and test the frame.
• Reduce friction for the recommended action; add friction to the harmful one where appropriate.
• Design any incentive for how people actually respond (small, frequent, immediate, loss-framed) and plan what sustains behaviour after it ends.
• Check for intrinsic motivation before attaching money to a behaviour, to avoid crowding it out.
• Pilot every nudge or incentive against a genuine control, pre-register the outcome, and commit to reporting null results.
• Apply the transparency-and-easy-exit test: announce the default, keep the alternative one step away, reject covert manipulation.
• Where you inherit a rational-choice model, state where behavioural realism would change the answer and adjust the inputs.
• Match the tool to the scale: use nudges to complement, not replace, structural policy such as tax and regulation.
• Assess the equity impact, since friction and complexity fall hardest on the poorest and least-well.

Key sources

• Richard Thaler and Cass Sunstein, Nudge: Improving Decisions About Health, Wealth, and Happiness — the foundational statement of choice architecture, defaults, and libertarian paternalism.
• Richard G. Frank, "Behavioral Economics and Health Economics" (National Bureau of Economic Research Working Paper 10881, 2004) — a survey of how behavioural findings bear on health economics.
• Thomas Rice, "The Behavioral Economics of Health and Health Care" (Annual Review of Public Health, 2013) — a critical review of applications and limits in the health sector.
• Herbert Simon's work on bounded rationality and Daniel Kahneman and Amos Tversky's work on heuristics, biases, and prospect theory — the psychological foundations.
• The United Kingdom's Behavioural Insights Team and the OECD's work on behavioural insights in public policy — practitioner exemplars of evaluated nudging.

References

1. Behavioral economics — Wikipedia — https://en.wikipedia.org/wiki/Behavioral_economics
2. Bounded rationality — Wikipedia — https://en.wikipedia.org/wiki/Bounded_rationality
3. Loss aversion — Wikipedia — https://en.wikipedia.org/wiki/Loss_aversion
4. Framing (social sciences) — Wikipedia — https://en.wikipedia.org/wiki/Framing_(social_sciences)
5. Present bias — Wikipedia — https://en.wikipedia.org/wiki/Present_bias
6. Default effect — Wikipedia — https://en.wikipedia.org/wiki/Default_effect
7. Choice architecture — Wikipedia — https://en.wikipedia.org/wiki/Choice_architecture
8. Nudge theory — Wikipedia — https://en.wikipedia.org/wiki/Nudge_theory
9. Richard Thaler — Wikipedia — https://en.wikipedia.org/wiki/Richard_Thaler
10. Cass Sunstein — Wikipedia — https://en.wikipedia.org/wiki/Cass_Sunstein
11. Libertarian paternalism — Wikipedia — https://en.wikipedia.org/wiki/Libertarian_paternalism
12. Behavioural Insights Team — Wikipedia — https://en.wikipedia.org/wiki/Behavioural_Insights_Team
13. Richard H. Thaler and Cass R. Sunstein, Nudge: Improving Decisions About Health, Wealth, and Happiness — Yale University Press / Penguin — https://en.wikipedia.org/wiki/Nudge_(book)
14. Richard G. Frank, "Behavioral Economics and Health Economics" — National Bureau of Economic Research (Working Paper 10881) — https://www.nber.org/papers/w10881
15. Thomas Rice, "The Behavioral Economics of Health and Health Care" — Annual Review of Public Health — https://doi.org/10.1146/annurev-publhealth-031912-114353