Prospect Theory (Kahneman & Tversky)

The three departures from expected utility

Prospect theory replaces three pillars of expected utility theory with empirically grounded alternatives:

1. Reference dependence. Outcomes are coded as gains or losses relative to a reference point — usually the status quo, sometimes an expectation, sometimes an aspiration. A $50,000 salary feels like a raise to someone earning $40,000 and a cut to someone earning $60,000, even though final wealth is identical.
2. Loss aversion. The value function is steeper for losses than for gains. Kahneman and Tversky's median estimate from lab choices was a slope ratio of about 2.25 to 1 — meaning losing $100 hurts as much as gaining $225 feels good.
3. Probability weighting. People act on transformed probabilities w(p), not probabilities themselves. The function is inverse-S-shaped: rare events (p < 0.1) are overweighted; near-certain events (p > 0.9) are underweighted.

The value function

Tversky and Kahneman (1992) parameterized the value function as:

v(x) = x^α            for x ≥ 0  (gains, concave)
v(x) = −λ · (−x)^β    for x < 0  (losses, convex and steeper)

Their fitted parameters across many experiments: α = β = 0.88 (mild diminishing sensitivity in both domains), and λ = 2.25 — the loss-aversion coefficient. The function is concave above the reference point (you become risk-averse over gains) and convex below (you become risk-seeking over losses, gambling to recover).

The fourfold pattern of risk attitudes

Combining loss aversion with probability weighting produces a famous prediction: people are not uniformly risk-averse or risk-seeking, but switch in four predictable ways.

	Gains	Losses
High probability	Risk-averse (locking in $90 sure over $100 with 95% chance)	Risk-seeking (gambling on $100 loss with 95% chance to avoid sure $90 loss)
Low probability	Risk-seeking (lottery tickets — $1 for 1-in-a-million $1M)	Risk-averse (insurance — paying $200 to avoid 1-in-1000 $50,000 loss)

Standard expected utility predicts a single risk attitude across all four cells. Prospect theory predicts the diagonal flip — and matches actual behavior in casinos, insurance markets, and stock holdings.

Prospect theory vs expected utility theory

	Expected utility (vN-M, 1944)	Prospect theory (KT, 1979)
Carrier of value	Final wealth	Gains/losses vs reference point
Risk attitude	Constant — concave utility	Risk-averse for gains, risk-seeking for losses
Probability treatment	Used as is	Transformed by inverse-S w(p)
Loss aversion	None — symmetric	~2.25× weight on losses
Endowment effect	Not predicted	Predicted — loss aversion of giving up
Lottery + insurance	Cannot explain (same person)	Explains — fourfold pattern
Equity premium puzzle	Implausible risk aversion needed	Loss aversion + myopic evaluation explains
Empirical fit	Fails Allais paradox, Ellsberg paradox	Fits both

From a Jerusalem hallway to the Nobel Prize

Kahneman and Tversky began collaborating at Hebrew University in the late 1960s. Their first joint paper on the psychology of judgment under uncertainty (1971) was followed by a decade of experimental work documenting violations of expected utility — the Allais paradox, framing effects, the certainty effect. Their 1979 Econometrica paper (rejected by the more prestigious American Economic Review) consolidated these into a single descriptive model. It became the most-cited paper in Econometrica's history.

Tversky died of melanoma in 1996 at age 59. The Nobel committee, which does not award posthumously, gave the 2002 prize to Kahneman alone. Kahneman opened his Nobel lecture by saying "everything we did was joint" and dedicated the prize to Tversky. The Sveriges Riksbank Prize that year cemented behavioral economics as a legitimate subfield — Vernon Smith won the same year for experimental methods.

Variants and extensions

• Cumulative prospect theory (1992). Tversky and Kahneman's update, applying probability weighting to cumulative ranks rather than individual outcomes; preserves stochastic dominance, which the original violated.
• Mental accounting (Thaler 1985). People segregate money into separate accounts (paycheck, vacation, gambling winnings) rather than treating wealth fungibly. Explains why people save in low-yield accounts while carrying high-yield credit-card debt.
• Endowment effect (Kahneman, Knetsch, Thaler 1990). Owning something raises its valuation — people demand 2× to give up a coffee mug they were just given vs to buy one. Loss aversion applied to ownership.
• Status quo bias (Samuelson, Zeckhauser 1988). Inertia from loss aversion of any change. Major driver of pension default-option enrollment design.
• Myopic loss aversion (Benartzi, Thaler 1995). Frequent portfolio evaluation amplifies loss aversion; explains the equity premium without absurd risk aversion.
• Reference-dependent preferences (Kőszegi, Rabin 2006). Endogenizes the reference point as rational expectations, giving prospect theory a formal microfoundation.

Real-world applications

• 401(k) auto-enrollment. The 2006 Pension Protection Act made automatic enrollment the default; participation jumped from ~50% to ~90%. A direct application of status quo bias from prospect theory.
• Insurance markets. Probability weighting explains why people buy expensive low-deductible coverage and warranties — they overweight rare losses.
• Equity premium puzzle. Mehra-Prescott (1985) showed expected utility requires implausible risk aversion to explain the historical 6% stock-bond return gap; Benartzi-Thaler (1995) resolved it with myopic loss aversion at empirically realistic parameters.
• Disposition effect in trading. Odean (1998) studied 10,000 brokerage accounts; investors realized gains 1.7× more often than losses, costing ~4.4%/year in returns.
• Tax compliance. Withholding refunds vs owing money triggers loss aversion — taxpayers receiving refunds report higher subjective happiness even though it represents an interest-free loan to the government.
• Sports decisions. NFL coaches punt on 4th-and-2 even when statistical analysis says go for it — losing a possession feels worse than the equivalent expected gain.
• Behavioral finance industry. Books like Thaler's Misbehaving, Ariely's Predictably Irrational, and Kahneman's Thinking, Fast and Slow built on prospect theory; UK Behavioural Insights Team ("Nudge Unit") and equivalent agencies in 200+ countries apply it to public policy.

Common pitfalls and critiques

• Treating loss aversion as a constant. The 2:1 ratio is a population median; individual variation is large, and the ratio shrinks with stake size and sophistication (List 2003).
• Confusing reference dependence with irrationality. Prospect theory describes behavior, not prescribes it. Failing to update reference points after a shock can be myopic — but is empirically robust.
• Replication concerns. Some classic effects (anchoring magnitudes, ego depletion) have weakened in large pre-registered replications; loss aversion itself remains robust (Brown et al. 2024 meta-analysis).
• Mistaking framing effects for stable preferences. Tversky-Kahneman's Asian disease problem shows preferences flip with frame; this is a phenomenon to manage in survey design, not a stable utility.
• Overweighting in policy. Nudges can exploit biases for paternalistic ends. Sunstein's libertarian-paternalism debate is unresolved.
• Forgetting the reference point. Defining "loss" requires specifying the baseline; ambiguity in the baseline (status quo? expectation? aspiration?) is the model's main empirical wiggle room.