Behavioral Economics: Heuristics, Prospect Theory, Biases & Incentives

Bounded Rationality

Idea that people make decisions that are less than perfectly rational due to cognitive limitations (load, attention), time constraints, and inadequate information
to cope with these cognitive and informational limits, we deploy heuristics

Satisficing

Search for options only until you find one that clears your aspiration level, then stop
Rather than using heuristics to find the absolute best choice (maximizing), people use them to satisfice

Simon's Scissors

Metaphor: rationality is like a pair of scissors — one blade is the mind's limits, the other is the structure of the environment
Heuristics can look irrational alone, but make sense once we add the environment back in

Kahneman & Tversky's heuristics-and-biases program

Views mental shortcuts as error-prone deviations from formal logic and probability. Emphasized information processing, focused on how individuals make judgements and when those judgements are inaccurate.
Main concepts:
- Availability heuristic: Involves relying on information that comes to mind quickly or is most available to us
- Representativeness heuristic: Tendency to judge the frequency or likelihood of an event by the extent to which it resembles the typical case
- Adjustment and anchoring bias: People become tied to the anchor which influences their estimate even when they know its irrelevant

Gigerenzer's fast-and-frugal heuristics

Heuristics cannot be considered irrational or inferior to a solution by optimization. Views them as adaptive, "ecologically rational" tools that perform well in real-world environments with limited time and information.
Fast = can solve a problem within a few seconds; Frugal = requires little information.
Main concepts:
- Recognition heuristic: Decides by relying on their recognition of something rather than other information
- Less-is-more effect: When recognition validity (α) is greater than knowledge validity (β). Having less information, time, or computation can lead to more accurate judgments and inferences than having more

Take-the-best heuristic

Where you decide between alternatives based on the first useful cue you find. Cue validity = how often that cue alone predicts the correct answer. Used in compensatory environments - one strong attribute overweighs any combination of other cues.

Tallying

Unweighted alternative to Take-The-Best: search cues in any order, then count which option has more positive cues. Exploits compensatory environments, where roughly equal cue weights let several weak cues outweigh one strong cue

Prospect Theory

Theory that people choose to take on risk when evaluating potential losses and avoid risk when evaluating potential gains.
vs. expected utility theory: people are basically rational, so if they have all of the relevant information, they will make a decision that results in the most beneficial outcome.
4 core features:
- Reference dependence: Decisions about loss/gain made from reference points, not in absolute terms
- Loss aversion: Greater sensitivity to loss than gains of the same value
- Risk and uncertainty aversion: Certain outcomes preferred over uncertain ones, even when EU is lower
- Framing effects: The way information is presented influences behavioral outcomes

Certainty effect

Outcomes obtained by certainty are overweighted relative to uncertain ones
Violates substitution axiom of EU: asserts that if B is preferred to A, then any (probability) mixture (B, p) must be preferred to the mixture (A, p)

Reflection effect

People are risk averse in the positive domain (i.e., gains), risk seeking in the negative domain (i.e., losses)
Violates EU invariance axiom: people's risk preferences flip from risk-averse to risk-seeking

Isolation effect

Tendency to disregard any elements that are common to both options
Violates EU assumption that choices between prospects are determined solely by the probabilities of the final states

Core model of prospect theory

overall value or expected utility of outcomes = subjective value of the outcome (value function) x weight of decision (probability weighting function)

Value function

Defined over gains and losses relative to a reference point. Concave for gains (risk averse) and convex for losses (risk seeking)
Transforms outcomes (dollar amounts) into subjective value

Probability weighting function

Inverse S shape - small probabilities are overweighted (π(p) > p); moderate/ large probabilities are underweighted (π(p) < p)
Transforms objective probabilities into decision weights

Fourfold pattern of risk attitudes

1. Relative gains, high probability = risk averse (prefer the certain lower win over big win)
2. Relative gains, low probability = risk seeking (overweigh small chance of winning big) e.g. buying lottery tickets
3. Relative losses, high probability = risk seeing (gamble to avoid a near-certain loss)
4. Relative losses, low probability = risk averse (overweight the small chance of a big loss) e.g. buying insurance

Probabilistic insurance

Insurance policy involving a small probability that the consumer will not be reimbursed
Concave utility function: people generally find it unattractive even though expected utility theory implies that probabilistic insurance is superior to regular insurance
Probability insurance appears intuitively riskier, prefer contingent insurance because of pseudo-certainty

Editing phase of choice

People decide which information to use in the evaluation phase. Influenced by order, wording, or method in which the choices are presented (e.g., framing effects)

evaluation phase of choice

people weigh the probability of each outcome and take actions based on the perceived likelihood and desirability of each outcome. We evaluate edited prospects and choose one with highest (psychological) value

Anchoring Effect

Cognitive bias, tendency to rely too heavily on the first piece of information they receive (reference dependence)

Accounts of why anchoring happens

1. anchor-and-adjustment: people use the anchor as a conscious starting point and then mentally move away from it until they reach a plausible answer
2. Selective accessibility: asking "Is the answer the anchor?" brings anchor-consistent information to mind
3. Anchors as information: we assume a number someone gives us (list price, first offer, sentencing demand) is informative, and judge relative to it

Reasons why we adjust too little from the anchor

Original account (T&K): we start from an anchor and adjust toward a plausible answer, but the adjustment is insufficient
Self-generated anchors: when people supply their own starting point, they adjust step by step and stop at the first plausible value
Resource-rational adjustment: adjusting takes mental effort, so stopping early can be a sensible trade-off between time and accuracy

Factors that influence effect of anchoring

Mood - being sad makes more susceptible to anchoring
Knowledge - Knowledgeable people less influenced by anchors
Incentives and forewarning - Anchoring is reduced (but not eliminated) with incentives and forewarning
Personality - People high in agreeableness, conscientiousness, & openness to experience and low in extraversion more influenced by anchors
Cognitive ability - Anchoring decreases with higher cognitive ability

Endowment effect

Phenomenon in which people tend to keep (and value more highly) the things they already have
May not be human universal: Hadza living in isolate regions do not demonstrate it

WTP and WTA

Willingness-to-pay (WTP): Amount of money an individual is willing to pay for a good
Willingness-to-accept (WTA): Amount of money an individual is willing to sell a good for
WTP-WTA gap: If an individual's subjective value of a good does not change by having it, all else being equal, WTP should equal WTA, but evidence suggests WTA > WTP

Kahneman, Knetsch, & Thaler (1990): Experimental tests of the endowment effect

Randomly give mugs to half a class and give them all cash tokens
Actual (V) / predicted trade volume (V*) = below 1 = undertrading because sellers' WTA > buyers' WTP

What causes endowment effect

1. Prospect theory account: loss aversion applied to riskless choices about what we own. Reference dependence where what we own is the reference point, status quo bias.
2. Psychological/mere ownership account: Feeling something is "mine" makes it more valuable; touching, customizing, or imagining owning it can trigger this
3. Memory and attention account: Owners first think of reasons to keep the item, so positive features come to mind more easily
4. Artifact accounts: Strategic overstating, misunderstanding the task, or lack of market experience

Factors that amplify the endowment effect in consumers:

Physical Touch: Physically interacting with products strengthens endowment effect
Customization: Personalizing products increases sense of ownership and perceived value
Imagination: Prompting consumers to imagine owning a product triggers ownership feelings
Time: The longer someone possesses an item, the stronger the endowment effect becomes

Mental Accounting

Cognitive operations used by individuals and households to organize, evaluate, and keep track of financial activities - sorting money into separate "accounts" and treating them differently despite fungibility

Fungibility

money is interchangeable; every dollar is worth the same no matter where it came from or what it's for

Transaction vs acquisition utility

Acquisition utility: the value of the good minus the price paid
Transaction utility: how good the deal feels, comparing the price paid to a reference price. Leads us to buy things we don't need because the deal is too good to pass up

Sunk costs

money or effort already spent that can't be recovered, so it should be irrelevant to future choices
Sunk cost effect: continuing something because of what you've already invested

Payment decoupling

Psychologically separating payment from actually getting or using a product, such as paying later via a credit card or loan.

Mental budgeting

People set (often implicit) budgets for categories like food, entertainment, and clothes, and track spending against them. Violate fungibility

windfalls

sudden, unexpected pieces of good fortune
People spend windfalls (tax refunds, bonuses, gifts, winnings) more freely than regular income

Overestimation

An inflated view of one's own skill or standing

Overplacement

Believing you are better or more deserving than your others

Overpercision

Too much certainty in one's own knowledge or assumptions

Calibration vs. discrimination

Calibration: does your confidence match your hit rate?
Discrimination: are you more confident when you're right than when you're wrong?

Why people are overconfident

Information deficits: People simply don't possess the information necessary to assess themselves accurately. We can't see the errors, risks, or better solutions we never thought of, are rarely get prompt, unambiguous feedback

Above-average effect

Cognitive bias wherein people overestimate their own qualities and abilities compared to others

Dunning-Kurger effect

When a person's lack of knowledge and skill in a certain area causes them to overestimate their own competence

Entrepreneurship and market entry

- Many new businesses fail within a few years, but most entrepreneurs rate their own odds far above those of similar businesses
- Leads to excess entry

Negotiation impasse

- When both sides in a dispute are overconfident about winning at trial or in negotiation, settlements that would benefit both parties are missed
- Self-serving bias: reading the same case, each side predicts the judge will favor them

Planning fallacy

- Systematic underestimation of the time, cost, or risk required to complete future tasks
- Because we take the "inside view" (this plan's details) and ignore the "outside view" (how long similar tasks took)

Consequences of overconfidence

Health:
- Unrealistic optimism about personal risk
- Pluralistic ignorance - College students overestimate how comfortable their peers privately are with heavy drinking
- False-consensus effect - smokers and other risk takers overestimate how many peers share their behavior
Education
- Self-assessed academic performance correlates only modestly with actual performance; peer assessment much better
- What feels like learning often isn't: cramming feels effective, but its an illusion
Workplace
- Employees systematically overrate their own performance
- Organizational feedback systems are structurally poor vehicles for correcting this - only given once a year, experienced as threatening, too late
- CEO overconfidence costs shareholders through bad mergers and -risky accounting, but can pay off in innovation

Self-deception

Favoring welcome over unwelcome information in ways that serve our goals
Evolved to facilitate deception of others
Allows one to behave self-interestedly while, at the same time, falsely believing that one's moral principles were upheld

Reducing overconfidence

Better feedback
Forced prediction before you know the outcome
Reference from others
Record your reasoning at the time to review later
Peer review

Monetary and nonmonetary (social and moral) incentives

People often engage in costly behaviors even when monetary reward is absent
Monetary incentives more effective in private than public
Monetary incentives for socially valued behaviors can decrease or eliminate behaviors driven by social incentives

Intrinsic and Extrinsic Incentives

Extrinsic: motivation to perform a behavior in order to receive a reward or avoid punishment from others
- Can crowd our intrinsic motivation e.g. temporarily paying people for an interesting task
- Can promote or reduce prosocial behavior
Intrinsic: motivation to perform a behavior unrelated to an external reward or punishment
- Different from social incentives, which can be extrinsic (eg getting praise)

Standard economic theory predictions of incentives

Incentives work like prices: raising reward for a behavior → more of it; raise cost → less of it
People work until marginal benefit = marginal cost
Predicts that performance rises monotonically with pay, and financial incentives only add to other motives, never subtracts

Social incentives

Often more powerful than monetary = people engage in prosocial behavior that contradicts their self-interest
Share brain structures with monetary incentives: same brain areas are activated = we experience them as psychologically similar

Gneezy and Rustichini (2000) study on monetary incentives

A3¢ bonus per correct answer led to fewer correct answers than no bonus. 30 ¢ bonus increased correct answers, but no difference between that and 85 ¢
Students asked to collect donations collected 36% less if they were told they would get 1% of collections as a bonus as opposed to nothing. 10% beat 1% but not no pay

Day care study

Day care centers study where parents were fined $3 each time they arrived late to pick up their children
More parents picked up their children late after the fine was imposed
Imposing a cost made parents feel like they were paying for a service, removed the social incentives for picking up your kid on time
Short-term monetary incentives can lead to long-term crowding out of social incentives and norms

Moral crowding out

Frey & Oberholzer-Gee (1997): 51% of Swiss residents accepted a nuclear waste site in their town, reduced to 25% when compensation offered = money signaled risk and turned civic duty into a sale
Mellström & Johannesson (2008): paying ~$7 cut women's blood donation from 52% to 30%. Letting them give the payment to charity restored it; no effect for men

Designing effective incentives

Be careful of when incentives can signal distrust, reframe the situation, or dilute image motivation
Using loss framing e.g. teachers paid a bonus up front, returned if students didn't improve → math scores rose
Using commitment devices for present bias e.g. smokers make their own deposits, forfeited if they relapse
Multitasking problem: rewarding what's measurable pulls effort from what isn't

Incentives can often backfire when:

The task is inherently interesting, financial incentives can reduce subsequent willingness to engage in it
If a task is "noble"
Paying a very high wage on the successful completion of a task
Paying a very low wage or charging a low fee

Cross-cultural differences in incentives

Social incentives are general social goals that develop from early infancy, throughout childhood and adulthood
Monetary incentive more effective than psychological incentive in US & UK
Nonmonetary incentive (e.g., a social norm message) less effective in US than India

Framing effect

When our decisions are influenced by the way information is presented
Framing sets a reference point, loss aversion drives the need to erase the loss, reflection effect is the resulting behavior - avoiding or seeking risk

Tversky & Kahneman (1986): four axioms of expected utility theory

Cancellation—ignore states that give the same outcome whatever you choose
Transitivity—if A ≻ B and B ≻ C, then A ≻ C (or you become a "money pump")
Dominance—if A is better in one state and at least as good in all others, choose A
--> Violated: failure of rational dominance when psychological weighting is distorted
Invariance—different descriptions of the same problem should yield the same choice
--> Violated: two formulations of the same problem elicit different preferences

Economists' defense of rationality and T&K's replies

Incentives will eliminate errors → violations persist with real money; incentives fix inattention, not misperception
Learning will eliminate errors → only with fast, accurate feedback, which big one-off decisions rarely provide
Markets will correct errors → not always: racetrack "place" and "show" bets stay mispriced despite expert bettors

Risky-choice framing

Gain vs. loss descriptions of risky options, effectiveness of gain or loss frames depend on certainty of outcomes
Certainty effect - people give more weight to outcomes that are certain than to outcomes that are uncertain
- Pseudocertainty - An outcome that is actually uncertain is treated as if it were certain

Attribute frames

Describing a single attribute by its positive or negative features
E.g. 75% lean beef rates tastier than 25% fat

Goal frames

A goal becomes focal by describing the advantages of acting or the disadvantages of not acting
E.g. Women told what they'd lose by skipping breast self-exams were more likely to do them

Social vs nonsocial frames

Framing a decision using socially laden language (e.g., a social context) shifts perceptions of normative expectations
Social framing shifts decision to a cooperative one = effective for promoting prosocial behavior
e.g. community vs wall street game
Ellingsen et al study of PD cooperation: results support coordination hypothesis that social framing effects are rational - only occur when both players chose to cooperate

Temporal framing

presenting information in terms of time to influence decision-making
Ex: A message that frames health risks as occurring every day can make the outcome seem more immediate than long-term risks

Reference point framing

providing specific reference points or baselines
Ex: Expecting nothing and getting a $5000 raise from your current salary vs expecting a $10,000 raise and getting a $5000 raise

Factors influencing effectiveness of frames

Individual differences:
- Need for Cognition (NC) - extent to which people engage in and enjoy effortful thinking
- Thinking style - intuitive-experiential more likely to show effects than analytical-rational processing
Development: people become more sensitive to negative framing as they age
- Socioemotional selectivity theory—as people age, their motivations, goals, and preferences shift based on their perceived time horizon
Context: different across domains (health, financial, nature)

Application of framing to Moral decisions

People more inclined to utilitarianism in the positive or neutral frame; more inclined to intuitionism in the negative frame

Intertemporal choice

Process of deciding between outcomes that will be experienced at different points in time
Time discounting: phenomenon that a desired result in the future is perceived as less valuable than the same result now
Time preference: extent to which you discount the future

Exponential discounting

When an agent discounts future costs and benefits at a consistent rate through time
Discount factor δ: decimal multiplier
- Patience = value close to 1, impatience = close to zero
Utility U0(u) = ut (utility you receive at time t) x δ

Calculating discounting

To calculate discount factor:
Find indifference point: say you are indifferent between payout a now and payout b in a year
ua = sqr root of monetary amount a
ua = ub x δ; δ = ua / ub

Hyperbolic discounting

Time inconsistent discounting - In reality, people do not have time consistent preferences
People have greatest impatience near the present
Waiting a week now feels costly; the same week a year from now barely registers
Pre-frontal cortex plays an essential role in ability to make decisions far into the future

Quasi-hyperbolic discounting

Combines aspects of exponential and hyperbolic discounting; accounts for present bias and preference reversals
β: present bias
When β = 1, the model reduces down to the delta model
However, when β < 1, all outcomes beyond the present time get discounted more than under exponential discounting

Loewenstein & Prelec, 1993: Violations of discounting

People exhibit negative time preferences = preference for later rewards
- People typically favor outcome sequences that improve over time
- Savoring and dread: for gains, improving sequences allow savoring the best outcome, for losses, getting undesirable outcomes over with reduces dread
- Adaptation and loss aversion: evaluate new stimuli based on baseline, so improving sequences experienced as a continual series of gains
- Recency effect: we overweigh the final outcome in a sequence

Mechanisms underlying intertemporal choice

- Representation: how a choice is mentally represented, impacts intertemporal tradeoffs & heuristics. Can evoke specific choice heuristics that increase or decrease the salience of delayed rewards and make waiting more or less aversive
- Anticipation: period between decision and outcome matters, produces immediate hedonic consequences, even when the anticipated consumption event is delayed in time
- Self-control: limitations of self-control can cause preference reversals, while good control is used to resist temptations to reverse patient plans

Delayed gratification and self control

Marshmallow test: distraction, "cool" thoughts about the treat (marshmallows as puffy clouds) rather than "hot" ones (sweet, chewy), covering them rather than in plain sight
Delay ≠ just willpower: it also reflects beliefs about whether the future will pay off. Taking the marshmallow early can be rational when the future reward is uncertain
Moffit et al: Childhood self-control predicted physical health, substance dependence, personal finances, and criminal offending outcomes

Individual differences in temporal discounting

Age: older people are more patient
Self control: Higher self-control associated with weaker discounting
Impultivity: Higher impulsivity associated with steeper discounting
Intelligence: Steeper temporal discounting with lower intelligence
Life-history theory: Harsh, unpredictable environments favor a present focus - if resources are sporadic, a future payoff may never arrive e.g. children from low SES backgrounds

Commitment device

Incentives that force us to overcome our present bias
Can work through 3 channels:
- Forcing the agent to maintain the optimal course of action
- Depressing the value of the undesired action
- Increasing the value of the optimal action