Loss Aversion¶
Core Idea¶
Loss Aversion is the asymmetric valuation pattern in which (1) a decision-maker evaluates outcomes relative to a reference point rather than in absolute wealth terms, (2) outcomes below the reference point are coded as losses and outcomes above it as gains, (3) the subjective weight placed on a loss of magnitude x exceeds the subjective weight placed on a gain of the same magnitude (empirically by a factor in the neighborhood of 1.5–2.5, formally denoted λ ≈ 2–2.5 in canonical prospect theory), and (4) this asymmetry systematically distorts risk-taking, status-quo preference, endowment pricing, and framing-sensitivity in ways that violate expected-utility predictions but are reproducible and quantifiable. The construct rests on a reference-dependent value function v(x) with a kink at the origin: v(x) = x^α for x ≥ 0 (concave, risk-averse in gains) and v(x) = −λ|x|^β for x < 0 (convex, risk-seeking in losses, with λ > 1 determining the asymmetry). This functional form, introduced by Kahneman and Tversky (1979) [1], became the empirical and theoretical anchor for all subsequent loss-aversion research.
How would you explain it like I'm…
Losing Hurts More Than Winning
Losses Sting Worse Than Gains
Losses Weighted Heavier Than Gains
Structural Signature¶
A reference-dependent valuation function with a kink at the reference point: steeper slope on the loss side than on the gain side. Formally, the loss-aversion coefficient λ is defined as the ratio of the marginal value of a loss to the marginal value of an equalsized gain, λ = |v'(−ε)/v'(ε)|, where ε is a small positive quantity. The structural primitive is not "people are risk-averse" (a level claim) but "the same outcome is priced differently depending on whether it is framed as a loss or a gain" (a derivative claim). The value function exhibits (a) reference-point dependence (utility is defined over changes, not absolute states), (b) diminishing sensitivity (concavity in gains, convexity in losses), and © loss aversion proper (the kink, with slope ratio λ). This reference-dependence property extends to riskless choice (Tversky-Kahneman 1991) [2], where the value function governs decisions even without probability weighting. Any domain that inherits this structural kink will exhibit the same downstream signatures: disposition-effect holding, endowment premiums, sunk-cost persistence, default-option stickiness, and framing-effect swings. The asymmetry is quantifiable: in laboratory settings, λ typically ranges from 1.5 to 2.5, with empirical variability across contexts. Cumulative prospect theory (Tversky-Kahneman 1992) [3] extended the framework to handle complex lotteries via rank-dependent probability weighting, allowing λ to interact with probability transformations and enriching the predictive landscape.
What It Is Not¶
Loss Aversion is not general risk aversion — under Prospect Theory, people are actually risk-seeking in the loss domain, because the convex utility curve in losses flips the comparative statics relative to the gain domain. It is not Optimism Bias (#250), though the two can cancel or amplify depending on the framing. It is not Satisficing (#248) — satisficing is about search termination on an aspiration threshold, which can use loss-aversive reference points but is a distinct mechanism. It is not simply "caring about money"; it is a quantifiable asymmetry in how identical-magnitude deviations are weighted. It is not Self-Handicapping (#247), which protects self-image against attributions, whereas loss aversion protects wealth or status against reference-point deviations. It is not identical to ambiguity aversion: loss aversion concerns the asymmetry between gain and loss valuation, while ambiguity aversion concerns unknown probabilities. Loss aversion applies under risk (known probabilities) and is empirically distinct from the cautious behavior induced by ambiguous distributions.
Broad Use¶
The pattern appears in financial decision-making (disposition effect—the tendency to sell winning stocks and hold losers to avoid realizing losses, empirically documented across asset classes), prospect-theoretic asset pricing (myopic loss aversion explains the equity premium puzzle via short evaluation horizons), consumer behavior (endowment effect, default-bias in opt-in vs. opt-out enrollment), negotiation (concession asymmetry), medical decision-making (framing effects in patient choice—e.g., survival rates vs. mortality rates produce different choices despite identical outcomes), public policy (tax framing, nudge architecture for retirement savings), organizational change management (resistance keyed to perceived losses rather than to objective net effect), and machine-learning objective design (asymmetric loss functions for regret-sensitive systems, where under-prediction carries higher cost than over-prediction or vice versa).
Clarity¶
It names the asymmetry precisely and separates it from the level claim of risk aversion, which allows analysts to predict behavior that unadorned risk aversion would get wrong — in particular, the counterintuitive finding that the same person who refuses a favorable coin flip for gains will accept an unfavorable one to avoid locking in a loss. This reflection effect (risk-seeking in losses, risk-aversion in gains) is the hallmark of prospect theory and is directly driven by the kink in the value function. The precision also enables cross-domain prediction: any context where a reference point is salient will show loss-aversive patterns, making the framework a portable diagnostic tool.
Manages Complexity¶
It collapses many seemingly-disparate anomalies — disposition effect, endowment effect, status-quo bias, sunk-cost fallacy, default effects, framing reversals — into a single generating mechanism, so analysts can predict which way a new context will tilt without cataloguing each anomaly separately. The unification is particularly powerful because all these effects fall out of a single parameter (λ) and the reference-point structure; they are not separate biases but manifestations of the same underlying asymmetry. The Kahneman-Knetsch-Thaler (1991) [4] synthesis paper articulated this family of effects and showed how they are theoretically and empirically linked.
Abstract Reasoning¶
Demonstrates that utility is not a function of final states but of changes relative to a reference point; the reference point itself is a modeling choice that determines what counts as a loss. This exposes that framing is constitutive, not cosmetic, and that any measurement that reports "preferences" is implicitly measuring preferences-given-a-reference-frame. The deeper philosophical point is that rationality—if defined as consistency with expected utility—is not the only criterion for evaluating decisions; loss aversion shows that decision-makers apply an additional criterion (the asymmetry) that is consistent across contexts and reproducible, suggesting a stable preference structure even if it violates EU axioms. The reference-dependence also challenges the view that preferences are "true" or context-independent; instead, preferences emerge from the comparison of outcomes to a benchmark, and changing the benchmark changes the preference ordering without changing the underlying decision problem.
Knowledge Transfer¶
| Loss Aversion component | ML asymmetric-loss analogue |
|---|---|
| Reference point | Target value or prior baseline |
| Loss region | Under-prediction or overshoot (domain-specific) |
| Gain region | Correct-side deviation |
| Asymmetric weight (λ ≈ 2) | Cost multiplier on the penalized side |
| Behavioral distortion | Systematic bias in the fitted model |
| Reference-point shifting | Curriculum/re-baselining strategies |
The transfer paragraph: in inventory forecasting, stockouts (losses—lost sales, customer dissatisfaction) are weighted several times as heavily as overstock (gains of slack, flexibility) in the loss function used to fit the forecaster. The fitted model will systematically over-order relative to the symmetric-MSE forecaster, and its behavior will be more sensitive to how "the baseline" is defined than to how the data is distributed. This is loss-aversive valuation instantiated in a parametric loss function: the asymmetry is not a bug, it is the formal declaration of which side of the reference the system must not cross. Analysts who understand loss aversion can read the hyperparameter λ as a direct operational analogue of the Kahneman- Tversky loss-aversion coefficient. Extending the transfer: in cybersecurity, false negatives (missed intrusions, losses) are weighted more heavily than false positives (false alarms, low-cost errors) in the classifier's loss function, shifting the system toward higher sensitivity at the cost of higher false-positive rate. The asymmetry is not a flaw but a transparent encoding of organizational risk tolerance, mapped directly from loss-aversion logic.
Example¶
Formal / abstract¶
Kahneman-Tversky value function and the 50/50 gamble: Kahneman and Tversky (1979) [1] and Tversky and Kahneman (1992) [3] found subjects rejecting a 50/50 gamble with +$200/−$100 while accepting the structurally identical payoff relabeled as "avoiding a loss." The expected value is the same: E[payoff] = 0.5(200) + 0.5(−100) = $50. Under expected utility with concave utility, this should have positive certainty equivalent and be accepted. But the value-function calculation with v(x) = x^0.88 for x ≥ 0 and v(x) = −2.25|x|^0.88 for x < 0 (using canonical parameters) yields: V(gamble) = 0.5 × v(200) + 0.5 × v(−100) ≈ 0.5 × (200^0.88) − 0.5 × 2.25(100^0.88) ≈ 0.5(145) − 0.5(2.25 × 91.4) ≈ 72.5 − 102.8 ≈ −30.3, which is negative, so rejection is rational under prospect theory. The kink (λ ≈ 2.25) makes the loss loom larger than the gain, overturning the EU prediction. This is the foundational empirical phenomenon that launched the entire loss-aversion research program.
Endowment effect: the mug experiment: Kahneman, Knetsch, and Thaler (1990) [5] conducted the canonical mug experiment. They randomly gave half the subjects a mug; the other half received nothing. Then they gave both groups the option to trade the mug for a pen of equal value. The mug-owners demanded roughly twice as much to sell (willingness-to-accept, WTA ≈ $5–6) as non-owners offered to buy (willingness-to-pay, WTP ≈ $2–3). The difference arose not from the mug's objective value (it was the same for both groups) but from the reference point: for owners, the mug is a gain relative to the status quo (they have it), so losing it is a loss, which loss aversion makes painful; for non-owners, the mug is a gain they lack, so acquiring it is a gain relative to their status quo (they don't have it), which is less attractive. The gap is proportional to λ: WTA/WTP ≈ 1 + λ, which with λ ≈ 2 yields the observed ratio of ~2. This is the most direct experimental evidence for loss-aversion asymmetry and demonstrates how the reference point (status quo) structures value.
Mapped back to the structural signature: in both cases, the reference point is the critical variable; the value function's kink (λ) translates the framing (loss vs. gain) into a quantifiable preference difference. The examples show how abstract loss-aversion logic operationalizes into concrete behavioral predictions.
Applied / industry¶
Subscription software company pricing reform with reference- dependent customer response: A subscription software company considering a pricing reform faces two framings with identical net effect. Framing A: "raise base price by $3, add a feature bundle worth $5." Framing B: "remove a feature currently included, sell it separately for $5, keep base price flat." Churn analysis from a prior reform shows Framing B produces 3–4× the cancellation rate of Framing A despite the objectively better customer outcome under B. The reference point (what the customer currently has) makes the removed feature a loss under Framing B, and the asymmetric weighting (λ ≈ 2–2.5) swamps the objective gain. The company's revenue optimizer and its retention model must include explicit reference-point tracking and asymmetric penalties to reason about this correctly. In practice, the company reframes all pricing changes as additions to current offerings rather than removals, exploiting the loss-aversion asymmetry to reduce churn while achieving the same financial outcome.
Equity-premium puzzle and myopic loss aversion: Benartzi and Thaler (1995) [6] explained the equity-premium puzzle (Mehra-Prescott 1985) [7] via myopic loss aversion. The puzzle is that the historical average return on stocks exceeds the risk-free rate by ~6%, far larger than standard risk-aversion models predict (Mehra and Prescott estimated that the implied risk-aversion coefficient should be absurdly high, ~20 or more). Benartzi and Thaler showed that if investors evaluate their portfolios frequently (e.g., annually) and are loss-averse, they demand a higher equity premium to tolerate the volatility of stock returns. The evaluation frequency combined with loss aversion creates a much higher perceived risk, explaining the puzzle. The mechanism: annual volatility of equities is ~20%, so a typical investor sees a 20% loss in one year out of every five or so. The loss looms large due to λ, making the investor willing to accept lower average returns in less volatile bonds, unless equities offer a higher expected return. This represents loss aversion scaling from individual decision-making to aggregate market pricing.
Mapped back to the structural signature: in both cases, the reference point (the prior price, the portfolio value) and the loss-aversion asymmetry drive behavior that simple risk aversion alone cannot explain. The examples show how loss aversion operates in real-world settings where reference points are naturally salient.
Structural Tensions and Failure Modes¶
T1 — Reference-point indeterminacy: The theory requires a reference point but does not, in general, pin it down — is it the status quo, the expected outcome, the aspiration level, the prior price seen, the anchor presented in the problem frame? Different reference-point assumptions produce different predictions, weakening the theory's out-of-sample predictive power without careful reference-point modeling. Köbberling and Wakker (2005) [^kӧbberling-wakker-2005] addressed this by axiomatizing the loss-aversion coefficient λ as a pure preference parameter, independent of the reference point, but implementation still requires choosing or measuring the reference point empirically. Without independent measurement of the reference point — measurement that does not itself depend on the observed choices being explained — prospect theory becomes effectively unfalsifiable: any observed loss-aversion behavior can be retrofitted to a chosen reference point that rationalizes the data. Köbberling-Wakker propose decomposing utility-curvature from probability-weighting from loss-aversion to address this; subsequent literature continues the project, but the core indeterminacy challenge remains live in applied work. Failure mode: the theory is applied post-hoc to explain observed behavior by selecting a reference point that fits; without independent measurement of the reference point, loss aversion becomes unfalsifiable.
T2 — Cross-cultural and individual heterogeneity in λ: The loss-aversion coefficient varies by domain, population, magnitude, and experience. Gächter, Johnson, and Herrmann (2007) [^gächter-johnson-herrmann-2007] measured loss-aversion parameters across multiple countries and found substantial heterogeneity: λ ranges from ~1.2 in some populations to ~2.5 or higher in others, with systematic differences by age, gender, and economic background. Reporting a single λ oversimplifies; the asymmetry is real but its size is not a universal constant. Failure mode: a single empirical estimate of λ is treated as a behavioral universal and applied to new contexts without regard for heterogeneity, producing biased predictions.
T3 — Market experience and the plasticity of loss aversion: List (2003) [8] showed that experienced traders exhibit reduced endowment effects compared to inexperienced subjects, and that the effect can be further attenuated or even reversed with repeated trading. This suggests loss aversion is not a hard-wired psychological trait but plastic, shaped by experience and learning. For markets where traders become experienced (financial markets, commodity exchanges), loss aversion may be muted; for novel or low-frequency decisions (medical, legal), it remains strong. Failure mode: loss aversion is assumed to operate uniformly across all decision-makers and all choices, missing the crucial boundary condition that experience can attenuate or eliminate the effect.
T4 — Methodological critique of the endowment effect: Plott and Zeiler (2005) [9] questioned whether the WTA/WTP gap is truly loss aversion per se or a procedural artifact. They showed that when experimental subjects are given experience in the market before the valuation task, the endowment effect largely disappears. They also argued that implicit demand effects (the subject inferring what the experimenter "wants") and other procedural factors could explain the gap without invoking loss aversion. While their critique has been partially rebutted (e.g., by showing the effect persists in some non-market contexts), it highlights that measurement of loss aversion is method-dependent and can conflate behavioral loss aversion with experimental artifacts. Failure mode: the endowment effect is mechanically attributed to loss aversion without investigating whether procedural confounds (experimenter demand, subject expectations, market norms) are driving the result.
T5 — Loss aversion vs. expected-utility calibration: Rabin (2000) [10] made a calibration argument that small-stakes risk aversion under expected-utility theory implies absurdly strong risk aversion at large stakes, violating intuition. For example, if an agent rejects a 50/50 gamble of +$11/−$10 at low wealth, EU theory implies the agent would reject a 50/50 gamble of +$10,000/−$5,000 at high wealth — a counterintuitive implication. Loss aversion resolves this by allowing utility to be reference- dependent and asymmetric: small-stakes loss aversion (rejecting the $11/$10 gamble because of the kink) does not force large-stakes risk aversion at the same level of concavity. However, the resolution comes at the cost of adding a new parameter (λ) and complicating the utility model, so it trades one theoretical problem (EU implausibility) for another (model complexity). Failure mode: loss aversion is invoked to solve EU calibration problems without recognizing that it introduces new theoretical commitments that may themselves require justification.
T6 — Aggregation to market-level outcomes and dynamics: Myopic loss aversion explains the equity-premium puzzle (Benartzi-Thaler 1995), but it does so by positing a particular evaluation frequency and assuming loss aversion is constant over time. In practice, both evaluation frequency and loss aversion may vary endogenously: when volatility is high, investors may check portfolios less often (endogenous evaluation frequency); when markets are booming, loss aversion may be temporarily suppressed by overconfidence or extrapolation bias. The dynamic interaction between loss aversion and other biases (overconfidence, momentum trading, herding) remains incompletely modeled. Failure mode: loss aversion is invoked to explain aggregate market phenomena without accounting for endogenous feedback effects (changing evaluation frequency, switching between behavioral modes, interaction with other biases) that could dampen or amplify the basic effect.
Structural–Framed Character¶
Loss Aversion is a hybrid on the structural–framed spectrum. Part of it is a bare pattern that means the same thing in any field — an asymmetric valuation around a reference point, with losses weighted more heavily than equal-sized gains; part of it is a frame, a vocabulary of decision-makers, value, and choice, inherited from behavioral economics.
The structural core is crisp and portable: a value function with a kink at the reference point and a steeper slope on the loss side, captured by a coefficient near two, is an abstract asymmetry that could in principle describe any reference-dependent response, including how control systems or organizations react to setbacks versus equivalent gains. But the prime is rooted in its behavioral home. Its native vocabulary — reference points, framing, prospect theory, the psychology of gains and losses — comes from the study of human judgment, and it carries an implicit reading of a cognitive bias, a departure from "rational" valuation. Its origin lies in behavioral economics, not in a purely formal relation, and applying it imports that view of human decision-making. With a sharp formal asymmetry but a substantial psychological frame, it sits on the framed side of the middle of the spectrum.
Substrate Independence¶
Loss Aversion is a narrowly substrate-independent prime — composite 2 / 5 on the substrate-independence scale. The structural idea — reference-dependent valuation with asymmetric weighting of gains and losses, the familiar λ ≈ 2.25 — is sound, but it is framed overwhelmingly in the language of behavioral economics. Its breadth is confined to decision-theoretic domains and does not reach physical, biological, or computational substrates in any technical sense. The transfer evidence is thin because the phenomenon is intrinsically behavioral and cognitive; it stays tethered to the human valuing mind it was discovered in.
- Composite substrate independence — 2 / 5
- Domain breadth — 2 / 5
- Structural abstraction — 3 / 5
- Transfer evidence — 1 / 5
Relationships to Other Primes¶
Parents (2) — more general patterns this builds on
-
Loss Aversion is a kind of Preference
Loss aversion is a kind of preference specialized by a reference-dependent, asymmetric value function: outcomes below the reference point are coded as losses and weighted more heavily than equivalent gains above it. It inherits preference's general commitment to an ordering over a choice set on an evaluative dimension, and supplies the specific case where the evaluative dimension is gain/loss relative to a reference point and the ordering relation systematically privileges loss avoidance — producing reproducible deviations from expected-utility predictions in framing, endowment pricing, and status-quo bias.
-
Loss Aversion is a decomposition of Asymmetry
Asymmetry names the structural property of a relation whose two sides are not interchangeable — swap the positions and something changes. Loss aversion is the specific shape this pattern takes around a reference point in the value function: subjective weight on a loss of size x exceeds the weight on a gain of the same size by a factor of roughly 1.5 to 2.5, producing a kinked value function. It is a structurally-particularized instance of directed imbalance whose specific diagnostic is the failed swap-test between gain frame and loss frame at the reference point.
Path to root: Loss Aversion → Preference
Neighborhood in Abstraction Space¶
Loss Aversion sits in a sparse region of abstraction space (81st percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.
Family — Preferences, Utility & Marginal Behavior (8 primes)
Nearest neighbors
- Risk Aversion — 0.79
- Deadweight Loss — 0.76
- Time Preference (Discounting Future) — 0.75
- Expected Utility — 0.75
- Marginal Utility — 0.75
Computed from structural-signature embeddings · 2026-05-29
Not to Be Confused With¶
Loss Aversion differs fundamentally from Risk Aversion, though they are often confused because loss aversion is frequently presented as an explanation for certain forms of risk-averse behavior. Risk aversion is a level claim: it states that a decision-maker prefers certain outcomes over gambles with equivalent or higher expected value. A risk-averse person turns down a 50/50 gamble of +$200 or −$100 (expected value +$50) in favor of receiving $50 with certainty. Risk aversion captures the general preference for safety, and it is typically modeled as concavity in the utility function—the marginal utility of wealth decreases with wealth. Loss aversion is a derivative claim about the asymmetry in how losses and gains are weighted. Under loss aversion, the same person might accept the 50/50 gamble when it is reframed as "avoid a loss," because the loss of $100 looms larger psychologically than the equivalent gain of $200. Prospect theory integrates both: the concave utility function in the gain domain produces risk aversion (people prefer certainty); the convex utility function in the loss domain produces risk-seeking behavior (people will take risks to avoid locks-in losses). A loss-averse person is not necessarily uniformly risk-averse; they are risk-averse in gains but risk-seeking in losses. The distinction is crucial because it explains the "reflection effect"—the reversal of risk preferences depending on whether the problem is framed as gains or losses. Risk aversion alone cannot predict this reversal; loss aversion can. Understanding the difference allows analysts to predict that a person who refuses a favorable gamble in gains will accept a structurally identical gamble framed as a loss, violating simple risk-aversion models but consistent with loss-aversion logic.
Loss Aversion is wholly distinct from Deadweight Loss, though both are concepts in economics frequently invoked in policy contexts. Deadweight loss is an allocative inefficiency—the total loss of economic surplus (consumer plus producer surplus) that results from a market distortion. When a tax, monopoly, or regulatory constraint prevents mutually beneficial trades from occurring, the unrealized gains become deadweight loss. Deadweight loss is a structural property of markets or policies, measured objectively by computing the area of lost surplus in a supply-demand diagram. Loss aversion, by contrast, is a behavioral preference pattern—a description of how individual decision-makers subjectively weight outcomes relative to a reference point. A market can have zero deadweight loss (all mutually beneficial trades occur) yet exhibit high loss aversion (traders value losses twice as heavily as gains of equal magnitude). Conversely, a market can have substantial deadweight loss yet involve decision-makers with no loss-aversion bias (if their preferences were merely concave utility functions without reference-dependence). The concepts operate at different levels: deadweight loss is about the efficiency of market allocation; loss aversion is about the subjective experience and decision-making of individuals within that market. Policymakers sometimes confuse these by attributing loss-aversion-driven behavior (status-quo bias, endowment effects) to deadweight loss, when in fact the behavior is driven by reference-dependent preferences, not by allocative inefficiency. Loss aversion can cause observable market behavior that looks like deadweight loss (people refusing trades that would increase total surplus), but the underlying mechanisms are distinct.
Loss Aversion is also distinct from Indifference Curves, which describe the locus of combinations of goods that yield equal utility to a decision-maker. An indifference curve maps out all the bundles (e.g., "2 apples and 3 oranges," "1 apple and 5 oranges," etc.) that leave the decision-maker equally satisfied. The shape of an indifference curve encodes preferences: a steep curve indicates willingness to trade a lot of one good for a small amount of another; a shallow curve indicates the opposite. Indifference curves are static representations of preference orderings at a given moment. Loss aversion, by contrast, is a temporal or comparison-based phenomenon—it describes how the same outcome is evaluated differently depending on whether it is framed as a loss or gain relative to a reference point. Indifference curves assume the decision-maker has a well-defined preference ordering over bundles of goods; loss aversion reveals that this ordering is not fixed but depends on the framing and the reference point. An indifference curve shows that 5 apples and 2 oranges yields the same utility as 3 apples and 4 oranges; loss aversion shows that the willingness-to-accept for selling 5 apples and 2 oranges is higher than the willingness-to-pay to acquire the same bundle, because reference points matter. In principle, indifference curves could be constructed that incorporate loss aversion (by allowing them to be reference-dependent), but standard indifference-curve analysis treats preferences as context-independent. Loss aversion challenges this assumption: the preference ordering itself shifts depending on what the decision-maker currently has and how the choice is framed. The distinction matters for modeling: indifference-curve analysis assumes stable, well-defined preferences; loss-aversion analysis reveals that preferences are constructed dynamically based on comparison to a reference point.
Solution Archetypes¶
Solution archetypes in the catalog that build on this prime — directly (this prime is a source ingredient) or as a related prime.
Built directly on this prime (1)
Also a related prime in 9 archetypes
- Approach–Avoidance Decomposition
- Bias-Specific Decision Audit
- Change Resistance Diagnosis and Support
- Goal Valence Decomposition and Separation
- Inertia Harnessing
- Negative-Mere-Exposure Reversal for Disliked Targets
- Opportunity Cost Surfacing
- Present-Bias Countermeasure
- Probabilistic Risk Weighting
Notes¶
Loss aversion has been used foundationally across behavioral economics and policy design for four decades since Kahneman and Tversky (1979) [1], but recent replication work (e.g., Gal and Rucker 2018 challenge; Ruggeri et al. 2020 multi-site replication partially supportive) has refined the claim from "universal 2× asymmetry" to "robust but context-dependent and magnitude-dependent asymmetry." Bernoulli (1738) [11] provided the precursor framework (expected-utility theory with concave utility), which loss aversion modifies via reference-dependence and asymmetry. The construct remains productive without needing the contested flag, but Pass B solution-archetype work should engage the magnitude literature and cross-cultural heterogeneity (Gächter et al. 2007) rather than cite the classic 2× figure uncritically. The tight pair with risk_aversion should be noted: risk aversion is about concavity (U''(w) < 0); loss aversion is about reference-dependence and the kink (asymmetry between gains and losses). Both are present in Prospect Theory but are analytically distinct properties. Camerer (2005) [12] provides a broad evidence review across multiple domains (finance, consumer behavior, insurance, negotiation), confirming loss aversion's empirical robustness despite methodological caveats.
References¶
[1] Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291. Foundational behavioral-economics result: outcomes are evaluated as gains and losses relative to a reference point rather than in absolute terms, with diminishing sensitivity and loss aversion — making the choice of baseline (and the contrast it creates with the treatment) constitutive of perceived value and decision behavior. ↩
[2] Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: A reference-dependent model. Quarterly Journal of Economics, 106(4), 1039–1061. Formalizes reference-dependent preferences and the loss-aversion mechanism that underlies the asymmetric weight given to abandoning prior investments versus accepting forward-looking opportunity costs. ↩
[3] Tversky, Amos, and Daniel Kahneman. "Advances in Prospect Theory: Cumulative Representation of Uncertainty." Journal of Risk and Uncertainty, vol. 5, no. 4 (1992): 297–323. ↩
[4] Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. Journal of Economic Perspectives, 5(1), 193–206. Documents the behavioral mechanism making defaults sticky in practice: parties value rights they hold under the default substantially more than equivalent rights they would have to bargain for, so defaults are not neutral starting points but quasi-anchors shaping final allocations. ↩
[5] Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler. "Experimental Tests of the Endowment Effect and the Coase Theorem." Journal of Political Economy, vol. 98, no. 6 (1990): 1325–1348. ↩
[6] Benartzi, Shlomo, and Richard H. Thaler. "Myopic Loss Aversion and the Equity Premium Puzzle." Quarterly Journal of Economics, vol. 110, no. 1 (1995): 73–92. ↩
[7] Mehra, Rajnish, and Edward C. Prescott. "The Equity Premium: A Puzzle." Journal of Monetary Economics, vol. 15, no. 2 (1985): 145–161. Documents that observed equity risk premia are too high for standard expected-utility equilibrium; foundational for equity-premium-puzzle literature and subsequent model extensions. ↩
[8] List, John A. "Does Market Experience Eliminate Market Anomalies?" Quarterly Journal of Economics, vol. 118, no. 1 (2003): 41–71. ↩
[9] Plott, Charles R., and Kathryn Zeiler. "The Willingness to Pay-Willingness to Accept Gap, the 'Endowment Effect,' Subject Misconceptions, and Experimental Procedures for Eliciting Valuations." American Economic Review, vol. 95, no. 3 (2005): 530–545. ↩
[10] Rabin, Matthew. "Risk Aversion and Expected-Utility Theory: A Calibration Theorem." Econometrica, vol. 68, no. 5 (2000): 1281–1292. ↩
[11] Bernoulli, Daniel. Hydrodynamica, sive de viribus et motibus fluidorum commentarii. Strasbourg: Joannis Reinholdi Dulseckeri, 1738. Establishes the principle of energy conservation in fluid flow: pressure and kinetic energy are inversely related. Bernoulli's equation (P + ½ρv² + ρgh = const along streamline) remains the foundation for steady, incompressible flow analysis across engineering and physics. ↩
[12] Camerer, Colin F. "Three Cheers—Psychological, Theoretical, Empirical—for Loss Aversion." Journal of Marketing Research, vol. 42, no. 2 (2005): 129–133. ↩