Behavioral Economics: Biases, Prospect Theory, Utility
Session 1: orthodox neo clsscl: based of rational choice, may not consider people deviate from this model ≠ Behavioral eco: behaviors deviates in systematic and predictable ways (irrational) ealry neo classlc: hedonic psych maximize pleasure minimize pain post-war neo:
refernces to unobservable mental states were unscientifc, they focused on choices of individudals that “mirrored preferences” / Methodo of Behavioral eco: laboratory experiments/field expirements/ process measures (brain scans&data -> model of decision making process )
Rational choice under certainty
Is anxiomatic theory/ axiom: a basic proposition that is taken for granted -> true or false/ Economics based on preference relation
:
Weakpref/ strongpref/indifference/ rational preference:
complete (x<=y or y<=x) and transitive/ indiferrence:
reflexivity, symmetry,transitivty/strong preference:
transitivity, antisymmetry, irreflexivity/choices under certainty-> rational choices ordering: chooses the most preferred item -> rational person self-interested (choices own preference)/ Utility function: strength of preference
6 principles:1)
people try to choose the optimal -optimizers and don’t succed bc experienced decional makers better,
2)comparer circumstances to reference point+ loss aversion(kahneman 79), we suffer from losse twice as they benefit from a gain of equal magnitude + discourages trade+ people hold endowment
3)People have self control problems -> present bias:
tendency to give stronger weight to immediate rewards than to futures ones, even waiting brings greater benefits.
4) we care about others -> social preferences = negativite reciprocity:
the tendency to punish unfair or hostile behavior, even if punishing costs you something/
6)Limiting choices (paternalism)
Can protect against biases, but heavy control is unpopular and often fails+ nudges > option.
Session 2: Violations of rational choice theory:
Sunk cost fallacy (concorde)
( cost cannot be recovered at the time when the decision is made), failure to consider opportunity costs –> Rational choice = pick the option where benefit ≥ opportunity cost (best alternative forgone)-> why people ignore:
difficult to consider all options, framed choices removes or diminishes reference to OC, considering all possible alternatives of OC is demanding/
Decoy effect:
(adding a worse option (never chosen) makes one option look better(target)→ irrational (violates expansion condition)/ similarity effect:
A similar option mostly splits choices with its closest competitor/
Diversification bias = we plan to spread out choices, but in practice we repeat favorites/compromise effect:
prefer the middle option to avoid extremes (reason-based choice strong when a compromise product is available bc it gives a defensible reason to choose
)
loss&endownement:
we demand more to give up something (WT Accept > WTPay ) because losses feel worse than gains (𝑣(−1) > 𝑣(+1)/ anchoring and adjusting/Aspirational treadmill:
More wealth → higher expectations → little extra satisfaction.
Status quo bias = preference for current state; can block efficient deals (contradicts Coase Theorem, which says with zero transaction costs parties should always bargain to efficiency)/Anchoring = starting from an initial number and adjusting too little → judgments stick close to the anchor/Social comparisons
:
Happiness depends on how outcomes compare to others (bronze medal > silver medal satisfaction).
Session3:
Outcome space:
all possible outcomes ≠ outcome-> subset outcome space, Probability function:
assigns each outcome a number in [0,1], Equiprobable rule:
if they occur with same probability→ each = 1/n, OR rule:
Pr(A) + Pr(B) – Pr(A and B)-> either one or the other AND rule:
all at the same time-> if independent (Pr(A) × Pr(B)/ Everything rule:
total probability of all outcomes = 1, NOT rule:
Pr(not A) = 1 – Pr(A), Mutually exclusive:
can’t happen together, Independent:
proba of one doesn’t have to affect the other, conditional probabilities:
Proba that A happens given that B already happened =Pr(A│B) = Pr(A & B)/Pr(B)
Bayes’ Rule:
Pr(A|B) = [Pr(B|A)×Pr(A)] / Pr(B) (total probability)
/ Bayesian Updating: updating beliefs based on Baye’s rules-
> Posterior = (Likelihood × Prior) / (Evidence)
/ Base-rate neglect = ignoring population frequency → people overestimate rare events after seeing evidence-> probabilité contraire = (1-p(A)) /Confirmation Bias : racism, gamblers, prediction)≠ to prevent: expose ideas on the other side, pay as much intention, apply same standarts of evidence / Availability heuristic: mental shortcut -> ease to recall rather than data.
Vivid recent or significant leads to overestimation of their actual probability.
Calibration= subjective probability = objective frequency (in long run) -> if it happens only 50% – overconfident , happens 90% -> underconfident/ Independence = events don’t influence each other; error = assuming independence when false, or dependence when false/ Gambler’s fallacy:
Believing past outcomes affect independent future events-> thinking a Tail is more likely after a streak of Heads, even though each flip stays 50/50.
/ Conjunction Fallacy:
Believing a detailed scenario is more likely/Disjunction Fallacy:
a single outcome is more likely than “A or B” rises or fall by 10% (we want one answer)
Session4:
Risk vs Uncertainty:
Risk = outcomes with known probabilities, Uncertainty = outcomes with unknown or meaningless probabilities/ Minimax Regret:
Choose the option that minimizes the maximum regret (avoid the biggest “what if”), Utility function:
tell us about consumer’s ranking ordering of bundles of goods,
expected utility= average satisfaction (with risk/ the gamble), expected value:
average outcome(money)-> Risk-averse:
concave utility ∩
-> flattens out= EV > EU = racine carré, LOG/ Risk-neutral:
linear utility, maximized EU = maximized EV/Risk-seeking:
convex utility -> get steeper = EU > EV = X^2 , exponentielle = ∪
Certainty equivalent:
amount of money $ that u are indifferent between the gamble and taking the money (the amount that gives same utility as gamble)= u(CE)=EU(G) -> remplace X by CE =u(x)
of proba function amount (not sure gain price)
/ Prospect Theory
: People evaluate gains and losses relative to a reference point
Session 5: Framing occur-> preferences and behaviour are responsive to how options are described (gain/losses) -> changes relative to a reference point (+1000$)/ integrate
–
> evaluates net value of the gains/losses (- 30$ (loss) & +2$ =28$ loss)
segregate-> evaluates each gains/losses separately.
gains
–
> concave (∩)
→ diminishing sensitivity to extra gains/ losses
:
Convex (∪)
→ diminishing sensitivity to extra losses
./
Diminishing sensitivity:
the further you are from the reference point, the less you feel the marginal change.
V(+10)−v(0) is larger than v(+1010)−v(+1000)
because small gains matter more when you have little/ ∣v(−10)∣−∣v(-0)∣ is larger than ∣v(−1010)∣−∣v(−1000)∣ because small losses matter more when you’re near 0/ Standard utility (total wealth): bundling doesn’t matter / Mental accounting:
People bundle/segregate money into categories (clothing, food, etc.), which shapes choices and can prevent overspending but violates fungibility.
Bundling effect
Same-category items are integrated (one big outcome), while cross-category items are segregated (separate outcomes)/ Allais Problem
: People overweight sure outcomes (certainty effect), leading to choices that violate expected utility & the sure-thing principle/ Ellsberg Problem (test-> ambiguity aversion)
:
People avoid unknown probabilities (ambiguity aversion), violating the sure-thing principle/The Sure-Thing Principle:
if you prefer A in every case, you should prefer A overall.People often violate this because of certainty effect or ambiguity aversion/ Puzzle (gambling + insurance)
: explained by probability weighting— people overweight small probabilities (lotteries, rare disasters) and underweight large ones.
Session 6:
Exponential discounting: people prefers rewards now than later but each future payoff weight less than same payoff today ->f δ is high (close to 1), the curve is flatter → small discounting (patient) ≠ If δ is low, the curve drops sharply → you don’t value the future much (impatient)/ Discount Factor (δ):
Multiplier (0<δ<1) that reduces future payoffs’ value (e.G. Tomorrow’s $1 = δ today)/ Discount Rate (r):
% showing how fast we discount/ Utility Stream:
Sequence of utilities over time u={u0,u1,u2,…}u={u0,u1,u2,…} Part II:
Partitioning & self-control:
Coupons/wages split into envelopes act as mental accounts/ More envelopes =fewer bets/spending (stop points)/ Opened envelope = used up fully; unopened = friction to continue/ Generalizable:
Same for food/chocolate → partitions slow consumption/ Field (Soman & Cheema 2011):
Partitioning > earmarking → more saving, less temptation spending/ Takeaway:
Simple partitions = cheap commitment device, boost self-control & savings.