- Prospect theory
- Prospect theory is a model of choice under risk in which people judge outcomes as gains or losses from a reference point and feel losses more than gains.
- Expected utility theory
- Expected utility theory is the standard model of choice under risk, in which people weight the utility of each possible final level of wealth by its probability.
What behavioural economics adds to the rational model
Can you name the five ideas behavioural economists add to the rational model of choice?
The first four are the parts of prospect theory; the fifth is about time rather than risk.
Reference dependence is judging an outcome as a gain or a loss against a reference point, such as what you expected or already have, rather than by final wealth.
Can you think of an example?
Two colleagues each get a £2,000 bonus. One expected nothing and is delighted; the other expected £5,000 and is disappointed. Their bank balances say they did equally well.
Loss aversion is focusing more on a loss than on an equal gain, so that losing a sum pains people more than gaining the same sum pleases them.
Can you think of an example?
Offered a coin toss that loses £100 on tails, most people want a win of around £200 on heads before they will play, although any win above £100 makes the bet favourable.
Diminishing sensitivity is feeling each extra pound of gain or loss less than the one before, the further the outcome lies from the reference point.
Can you think of an example?
The step from losing £10 to losing £20 stings; the step from £1,010 to £1,020 barely registers. So offered a sure loss of £500, or a coin toss between losing £1,000 and losing nothing, many people gamble.
Probability weighting is treating small chances as if they were larger than they are, and near-certain outcomes as if they were less sure than they are.
Can you think of an example?
A lottery jackpot with odds of millions to one feels worth a ticket, and a rare burst pipe feels worth insuring at well above its expected cost, by the same person in the same week.
Present bias is giving extra weight to costs and benefits that fall now compared with any later date, so plans made for the future are abandoned when it arrives.
Can you think of an example?
On Sunday a student plans to revise on Monday evening. On Monday evening a film wins. Nothing about revision or the exam changed; only which of them was happening now.
Prospect theory keeps the maximising but changes what is maximised
At A-level, behavioural economics usually arrives as a list of biases set against the rational model. At university the sharper question is which assumption fails and what replaces it. Kahneman and Tversky's prospect theory keeps the idea that people pick the option they value most. What changes is the thing being valued: not final wealth, but gains and losses from a reference point. The value function is steeper for losses than for gains, flattens out in both directions, and outcomes are weighted by how likely they feel, which overstates small chances.
Check yourself
A shop that charges more for card payments calls the gap "a discount for paying cash". Why might shoppers mind this less than a "card surcharge" of the same size?
Would you rather have £100 today or £110 in a week? And £100 in a year or £110 in a year and a week?
Steady discounting says a week costs the same wherever it falls, so the choices should match. Many people are impatient about a week that starts today and patient about the same week a year away.
Present bias makes today's plans inconsistent with tomorrow's choices
Present value discounts every year at the same rate, so waiting a week costs the same whether the week starts today or in a year. The £100-or-£110 choice breaks that. The beta-delta model captures it: each future period is discounted by δ as usual, and every future payoff is also multiplied by a factor β below one, so "now" gets a premium no later date shares. When next year becomes this year, the premium moves with it and the plan changes. People who know this about themselves buy commitment devices: a pension they cannot touch until retirement, or a gym membership paid a year ahead.
Check yourself
Which of these works as a commitment device against present bias?
Automatic enrolment shows how far a default can move behaviour
From 2012, UK employers were required, in stages, to enrol eligible staff in a workplace pension automatically, leaving each worker free to opt out. A rational saver would not care which way the form was set, because the pension on offer is identical. Yet among eligible private-sector employees the share saving into a workplace pension rose from 42% in 2012 to 88% in 2025. Not all of that rise is the default: the reform also required every employer to offer a scheme and to contribute to it. But because staff were enrolled automatically and could leave, the fact that most stayed in is the default at work. Inertia and present bias fit: opting in takes effort today, and the reward arrives decades later. The same thinking runs through the Behavioural Insights Team, set up inside the Cabinet Office to test such nudges with trials.
The findings refine the rational model more than they overturn it
Critics make three points. Some effects fade with experience: John List found traders at sports-card fairs showed far less attachment to what they owned than novices. Nudges tested at scale often do much less than in academic papers, as DellaVigna and Linos found when they gathered the trials run by nudge units. And a nudge assumes the planner knows what people really want. Against that, the beta-delta and prospect theory models still have people maximising; they change the function, not the method. The rational model survives best in repeated, high-stakes markets and worst in rare, delayed choices.
Interviewers often pick up a book from your reading, and Kahneman's Thinking, Fast and Slow is a favourite. Expect "so is economics wrong?". Name one finding, say which assumption of the rational model it breaks, give the evidence, then say where the model still holds. Mentioning loss aversion alone is a summary; saying when it fades is an argument.
Exam question
You've read Thinking, Fast and Slow. Does it show that the rational-choice model is wrong? [6]
Partly. The rational model assumes people value final wealth consistently. Prospect theory shows choices depend on a reference point, and loss aversion means a loss of £100 outweighs a gain of £100, so many people refuse a coin toss that loses £100 or wins £150, a small favourable bet a rational chooser would take. Present bias shows plans made for the future are dropped when it arrives, which helps explain why automatic enrolment, with its opt-out default, raised UK pension saving so much. But these models still have people maximising, just a different function, and some effects shrink among experienced traders or when nudges are scaled up. So the model is incomplete rather than wrong: sound for practised, high-stakes markets, weak for rare, delayed decisions.
A strong answer states the model's assumption, applies one or two findings with a mechanism and a real case, and then names where the rational model survives: in experienced, repeated markets. Declaring economics simply wrong, or simply right, misses the judgement the question asks for.
Exam question
A gambler is £500 down near the end of the night and starts placing bigger, riskier bets. Why? [4]
Prospect theory says he judges the night against where he started, his reference point, so he is sitting in losses. Diminishing sensitivity means the next £100 lost hurts less than the first £100 did, while winning back to zero would remove the loss entirely, so in losses people become risk-seeking. The rational model, valuing final wealth, predicts no such switch. The explanation assumes his reference point stays where he started; if he has accepted the loss and reset it, the effect disappears.
The model is reference dependence with diminishing sensitivity; the application is being in the loss region, where people gamble; the assumption that could break it is the reference point. Loss aversion alone predicts caution, not bigger bets, so naming the right idea earns the marks.
Exam question
Why might the same person buy a lottery ticket and insure their phone? [4]
In expected-value terms both are poor deals: the ticket pays out less than it costs on average, and so does the insurance, since the insurer covers its costs and makes a profit. A rational risk-averse person should insure but not gamble. Probability weighting explains both: small probabilities are treated as larger than they are, so the tiny chance of a jackpot and the small chance of a broken phone both loom large. The explanation weakens where people insure because they could not afford the loss, which is rational risk aversion.
The model is probability weighting; the application is that one bias explains two opposite-looking choices; the limit is that insuring against a loss you could not bear is rational anyway. Saying only "people are irrational" names no mechanism and earns little.
Prospect theory says people judge outcomes as gains or losses from a reference point, feel losses more than equal gains, grow less sensitive further from the reference point and overweight small chances. Present bias gives "now" a premium, so plans unravel and commitment devices and defaults such as automatic enrolment work. Critics find some effects fade with experience and nudges shrink at scale. The models refine rational choice rather than replace it.