Behavioral Economics

Challenges the rational-agent assumption in standard economics. People aren’t Econs (rational, patient, consistent) — they’re Humans (emotional, impatient, easily influenced).

graph TD
  Rational[Standard Economics<br/>Rational agent: always optimizes] -->|but real people...| Humans

  subgraph Humans[How humans actually behave]
    Prospect[Prospect Theory<br/>losses hurt 2x more than gains]
    Anchoring[Anchoring<br/>first number sets reference]
    Present[Present Bias<br/>want it now, procrastinate later]
    Mental[Mental Accounting<br/>money is not fungible]
    Herding[Herding<br/>copy the crowd]
    StatusQuo[Status Quo Bias<br/>stick with default]
  end

  Prospect -->|explains| RealWorld1[holding losing stocks<br/>housing market freezes<br/>insurance overpayment]
  Anchoring -->|explains| RealWorld6[salary negotiation<br/>real estate prices<br/>retail discounts]
  Present -->|explains| RealWorld2[credit card debt<br/>undersaving for retirement<br/>gym memberships]
  Mental -->|explains| RealWorld3[payday loans + savings account<br/>theater ticket vs cash]
  Herding -->|explains| RealWorld4[asset bubbles<br/>bank runs<br/>restaurant queues]
  StatusQuo -->|explains| RealWorld5[organ donation rates<br/>401k enrollment<br/>sticky contracts]

  Nudge[Nudge Theory<br/>change defaults, not choices] -->|uses| StatusQuo
  Nudge -->|uses| Framing[Framing<br/>how options are presented]
  Nudge -->|uses| Salience[Salience<br/>make info visible]

Prospect Theory (Kahneman & Tversky 1979, Nobel 2002)

The experiment

Pick A or B:

• A: Guaranteed $500
• B: 50% chance of $1,000, 50% chance of $0

Most pick A (safe, sure thing for gains).

Pick C or D:

• C: Guaranteed -$500
• D: 50% chance of -$1,000, 50$0

Most pick D (gamble to avoid a sure loss).

Standard economics: $500 is $500 regardless of sign. People should be consistent. They aren’t.

The S-curve

Value (subjective)
     ^
     |      /  (gains — concave, risk-averse)
     |     /
     |    /
     |   /
 ----+-------> Outcome
     |   \
     |    \
     |     \   (losses — convex, risk-seeking)
     |      \

• Loss aversion: losing $100 hurts ~2x more than gaining $100 feels good
• Diminishing sensitivity: difference between $0 and $100 feels bigger than between $1,000 and $1,100
• Reference point: you judge outcomes relative to where you started, not the final total

Real-world applications

Anchoring

The experiment

Spin a wheel of fortune (rigged to stop at 10 or 65). Then ask: “What percentage of UN countries are African?”

• People who saw 10 → guessed ~25%
• People who saw 65 → guessed ~45%

The random number anchored their estimate. Even though they knew the wheel was random, their brain started from that number and adjusted — but never adjusted enough.

Why it works

The brain doesn’t estimate from scratch. It picks the first available number (anchor) and adjusts up or down. The adjustment is almost always insufficient — you don’t move far enough from the starting point.

Crucially: anchors work even when they’re completely irrelevant. The UN question had nothing to do with a 10 or 65, but those numbers still biased the answer.

Real-world applications

Connection to Prospect Theory

Both involve reference points:

• Prospect Theory: reference point is where you start (gains vs losses)
• Anchoring: reference point is the first number you see

Together they explain the disposition effect fully: you anchor on the purchase price AND you’re loss-averse relative to that anchor. Result: you won’t sell until the stock returns to your anchor, even if fundamentals say you should.

Present Bias / Hyperbolic Discounting

The experiment

• “Would you rather $100 today or $110 in a week?” → most pick today
• “Would you rather $100 in 52 weeks or $110 in 53 weeks?” → most wait for the $10

Same trade-off (wait 1 week for $10 extra), different time horizon. The discount rate between now and next week is huge. Between two future dates it’s small.

Standard economics: discount rate should be constant. Real people have a present bias — today is special.

Why this matters

Commitment devices

The only fix: lock yourself in before temptation arrives.

• Automatic 401k deduction (can’t spend what you never see)
• “Save More Tomorrow” (Thaler): commit future raises to savings before you feel richer
• Ketchup (or other apps): bet money you’ll lose if you don’t achieve a goal

Mental Accounting (Thaler, Nobel 2017)

The experiment

• Lost a $50 theater ticket. Buy another? Most say no.
• Lost $50 cash. Still buy the ticket? Most say yes.

Same $50 loss. Same net result (you’re down $50 plus pay $50 for the ticket either way). But one feels like “entertainment budget already spent” and the other feels like “extra cash.”

Standard economics: money is fungible. $1 = $1. Brains don’t work that way

Real-world examples

Herding & Information Cascades

The restaurant street

You’re in a new city. Three restaurants:

• A: empty
• B: 5 people
• C: 20 people

You pick C. Not because the food is good — you’re using other people as a signal. But they picked C for the same reason. The crowd is just copying the crowd.

This is an information cascade: after the first few people choose, everyone after them ignores their own information and copies.

Cascade dynamics

1. First person has a signal (good food?)
2. Second person has an independent signal
3. Third person: hears two choices, ignores own signal → copies the crowd
4. Everyone after copies too

Now the whole crowd is wrong. The first few people might have been delivery drivers waiting for orders, not actual customers.

Anti-herding / contrarians

You see C packed, choose empty A. Now others see you in A → they follow → A fills up.

Contrarians exist but are rare because it’s psychologically hard to go against the crowd. Your gut says “20 people can’t all be wrong.”

Bubble mechanics

Prices up → "everyone is buying" → more people buy → prices up more
→ eventually last buyer runs out of money
→ prices down → "everyone is selling" → more people sell → prices down more
→ overshoot below fair value → cycle repeats

Housing 2006, crypto 2021, tulips 1637 — same cascade in both directions.

Nudge / Choice Architecture

Nudge ≠ mandate. No options banned, no prices changed. Just change how choices are presented.

The organ donation split

Two neighboring countries, similar culture:

Same people, same values, same religion. Just the default direction flipped.

Why defaults work

1. Status quo bias: doing nothing is easier than doing something
2. Procrastination: “I’ll sign up later” → never
3. Implicit endorsement: “The default must be the recommended option”

Key nudges

Criticism

• Paternalistic: who decides what the “right” default is?
• Transparency: should people know they’re being nudged?
• Fragile: works in one context, flops in another

Summary: Econ vs Human

Related

• Supply & Demand — the rational consumer this challenges
• Welfare & Efficiency — consumer surplus assumes rational choice
• Market Intervention — nudge as third way between tax and ban