HPC › Decisions › Glossary
Prospect theory is a behavioural model of risky choice, developed by Kahneman and Tversky in 1979, that describes how people evaluate outcomes relative to a reference point rather than in absolute terms. Losses loom larger than equivalent gains, small probabilities are overweighted, and the value function is concave for gains and convex for losses.
The 1992 extension, cumulative prospect theory, applies rank-dependent probability weighting and resolves stochastic-dominance violations in the original formulation.
How it works
The core departure from classical economics is the value function. Kahneman and Tversky showed that people do not evaluate outcomes by their total wealth; instead, outcomes are coded as gains or losses relative to a salient reference point 1. The function is S-shaped: concave in the gain domain, producing diminishing sensitivity as gains grow, and convex in the loss domain, producing the same diminishing sensitivity as losses deepen. Critically, the function is steeper below the reference point than above it, which is the formal expression of loss aversion. A 1-to-1 trade between a gain and an equivalent loss is never neutral.
The second component is probability weighting. Rather than treating probabilities at face value, people distort them via an inverse-S-shaped weighting function: low probabilities are overweighted, moderate-to-high probabilities are underweighted 12. The inflection point sits near 0.35. This accounts for a puzzle expected utility theory cannot resolve: the same individual who buys insurance against small risks also buys lottery tickets on negligible ones. Both behaviours are consistent under the weighting function.
The 1992 extension, cumulative prospect theory, replaced individual probability weights with rank-dependent decision weights applied separately to the gain and loss distributions 2. This resolved stochastic-dominance violations in the original formulation and extended the theory to uncertain prospects with any number of outcomes. The loss aversion coefficient lambda, estimated across 607 empirical findings from 150 studies, has a mean value of approximately 1.955 4; some risky-choice samples yield lower figures near 1.31, reflecting context-dependence in the effect size.
Example
A company reviewing salary packages presents two options: accept a modest pay increase or avoid losing a previously announced bonus of the same magnitude. Most individuals push harder to avoid the loss than to secure the equivalent gain. The framing is identical in net terms; the effort applied to each is not. The reference point determines whether the outcome registers as something to approach or something to resist.
The disparity is not irrationality; it is loss aversion operating exactly as the value function predicts.
Why it matters
The practical reach of prospect theory extends across economics, finance, and organisational behaviour. In financial markets, investors' tendency to sell winning assets prematurely and hold losing ones too long (the disposition effect) maps directly onto the asymmetric value function: paper losses create disproportionate reluctance to realise 3. The theory's empirical foundations have held up across decades of scrutiny and large-scale replication, making it the best available descriptive account of risky choice in experimental settings 3.
The asymmetry also shapes institutional design. Pension default schemes that automatically enrol workers into retirement savings exploit the finding: non-participation registers as a loss relative to the default, driving participation far beyond what equivalent financial incentives presented as gains achieve 3. Contract and salary negotiations follow the same logic, with loss-framed outcomes producing stronger motivation and greater risk tolerance than gain-framed equivalents of identical value 13.
What is the difference between prospect theory and expected utility theory?+
Expected utility theory holds that rational agents maximise the probability-weighted sum of absolute wealth outcomes. Prospect theory replaces this with a value function defined over gains and losses relative to a reference point, adds loss aversion and an inverse-S-shaped probability weighting function, and describes how people actually choose rather than how they should {{cite:10.2307/1914185}}{{cite:10.1257/jep.27.1.173}}.
What is loss aversion and how large is the effect?+
Loss aversion is the empirical finding that a given loss causes roughly twice the psychological impact of an equivalent gain. Across 607 estimates from 150 studies, the mean loss aversion coefficient lambda is approximately 1.955 {{cite:10.1257/jel.20221698}}. Risky-choice sub-samples yield somewhat lower estimates near 1.31, suggesting context modulates the size of the effect.
How does the probability weighting function work in prospect theory?+
The probability weighting function transforms objective probabilities into decision weights via an inverse-S-shaped curve. Probabilities below roughly 0.35 are overweighted; moderate-to-high probabilities are underweighted {{cite:10.1007/bf00122574}}{{cite:10.2307/1914185}}. The outcome is excessive sensitivity to very small chances and insufficient sensitivity to higher-probability outcomes, explaining simultaneous insurance and lottery purchases.
Where does prospect theory apply in everyday decisions?+
Prospect theory applies wherever reference points, loss framing, or probability distortion arise. Pension default schemes exploit loss aversion to raise savings participation. Investors hold losing assets too long. Negotiations shift predictably when proposals are framed as avoiding losses rather than securing gains {{cite:10.1257/jep.27.1.173}}.
