Behavioural & Experimental Economics – Mental Accounting & Myopic Loss Aversion

Overview

• Course/lecture context: ECO331 – Behavioural & Experimental Economics; focus on Mental Accounting & Myopic Loss Aversion (Robert Gazzale, University of Toronto).

• Four-part road map:
  • Recap big ideas in Thaler’s “Mental Accounting Matters”.
  • Show how mental accounting explains sunk-cost fallacy & hedonic editing.
  • Explore mental accounting’s impact on risk attitudes ⇒ myopic loss aversion (mental accounting + prospect-theory loss aversion).
  • Apply insights to labour-economics and public-policy questions.

Mental Accounting: What It Is

• Definition: “A set of cognitive operations people use to organize, evaluate & keep track of financial activities.”

• Three core questions:

  1. What belongs in the same account? (category, size, frequency).

  2. When are accounts opened/closed? (“Registering” utility occurs at closing.)

  3. How does the chosen frame affect risk attitudes?

One vs. Many Accounts – Breaking Homo Economicus

• Homo economicus: uses the broadest frame; utility only over final net wealth.

• Mental accounting: people frame decisions in narrower, often multiple, accounts ⇒ behaviour departs from standard theory.

• Illustrative classroom questions:
  • “One more hour of ECO331 studying?” – Utility depends on incremental change vs. overall lifetime satisfaction.
  • Coin-flip wagers: +10/−10 framed alone vs. embedded in large wealth W alter acceptance rates.

Classroom ‘Anomaly’ Examples

• Concert scenario (leave home with 50):
  • Lose 20 cash en route; ticket costs 20 ⇒ 50 % go.
  • Lose pre-purchased 20 ticket; replacement costs 20 ⇒ much fewer go.
  • Homo economicus prediction: identical.

• Five-block walk for 5 saving:
  • Computer cable (price drop 10→5) vs. sweater (price drop 100→95).
  • Higher willingness to walk in small-price domain although absolute gain equal.

Mental Accounting & Relativity of Costs

• Marketing framing (“Pennies-a-Day”, “Dollar-a-Day”): people compare to same-size/frequency expenses.
  • Small ongoing requests compared to coffee/lunch ⇒ higher compliance.
  • Large one-time sums compared to vacations or appliances.

• Gourville (1998) study: temporal frame × request size influences donation likelihood (see Figure 2, Study 1).

Narrow Categories: Empirical Evidence

• Kooreman (2000) – Dutch child benefit vs. other income:
  • Child benefit account: \text{mpc}{\text{kids’ clothes}}=0.12, \text{mpc}{\text{adult clothes}}=-0.02.
  • Other income: 0.01 and 0.04 respectively.

• Landsberger (1966) – German restitution payments to Israelis:
  • Small payments ⇒ \text{mpc}=2 (!)
  • Medium \text{mpc}=0.5–0.6.
  • Large \text{mpc}=0.2.

Income Accounts & Permanent-Income Violations

• Standard PIH: single marginal propensity to consume.

• Mental accounting yields three:

  1. Current-income account – high \text{mpc}.

  2. Asset account – medium \text{mpc}.

  3. Future-income account – near-zero \text{mpc}.

• Windfall framing (bonus vs. tax refund): more splurging/repayment when framed as bonus; more saving when framed as refund.

Real-World Policy Anomalies

• U.S. fiscal stimulus design:
  • Tax rebates (lump-sum cheques) – mostly saved.
  – Shapiro & Slemrod 2009 survey: \text{mpc} \approx 0.33; low-income used to pay debt.
  – Parker et al. 2011: \text{mpc} \approx 0.7 but concentrated on durables.
  • Reduced withholding (smaller, repeated pay-cheque boosts) – \text{mpc} \approx 1.
  • 2009 ARRA relied on withholding cut, consistent with mental-account predictions.

• SNAP benefits (Hastings & Shapiro 2018): because funds restricted to groceries, they reside in “food” account ⇒ spending patterns differ from cash although 84 % of recipients already spent more than benefit on food.

Hedonic Editing – Opening & Closing Accounts Over Time

• Based on Prospect Theory value function (concave for gains, convex for losses, steeper for losses):

  1. Break up gains into smaller pieces.

  2. Offset small losses with larger gains.

  3. Segregate big losses from small gains.

  4. Aggregate multiple losses.

• Creative classification maximizes perceived utility.

Loss-Realization & Disposition Effect

• Reluctance to close a losing account ⇒ ‘loss realization aversion’.

• Odean (1998): investors sell winners far more than losers; held-onto losers subsequently under-perform.

Sunk-Cost Bias as Mental Accounting

• Hypothesis: individuals want next benefit to share account containing past unrecoverable cost ⇒ continue investing.

• Concert-snowstorm thought experiment: paid-for ticket motivates attendance; free ticket does not.

• Gourville & Soman (1998): health-club attendance spikes right after semi-annual dues charged, then decays.

Motivated & Costly Bracketing

• Narrow bracketing can be adaptive (self-control budgets, transaction utility) but can also:
  • Lead to sub-optimal choices (ignoring substitution possibilities).
  • Enable motivated reasoning (e.g., rationalising indulgent purchases).

Risk Aversion, Narrow Framing & Rabin’s Calibration

• Expected-Utility Theory (EUT): risk aversion arises from diminishing MU of wealth.
  • Arrow (1971): for small stakes, rational EUT agent ≈ risk-neutral unless wealth stakes life-changing.

• Observed small-stakes risk aversion ⇒ extremely steep MU curvature if EUT true.

• Rabin (2000) proof outline:

  1. Assume person rejects 50\% chance of +21 vs. -10 (gain > loss).

  2. Iterate wealth upward by \$21 chunks; by concavity MU must drop geometrically.

  3. Result: if MU(W)=1000, then MU(W+1000)\approx10, MU(W+2000)\approx0.1 ⇒ absurd.

• Resolution proposals:
  • Narrow framing (evaluate each gamble separately).
  • Prospect theory loss aversion.
  • Myopic loss aversion = loss aversion + short evaluation horizons via mental accounting.

Myopic Loss Aversion & Samuelson’s Bet

• Single gamble: 50\% lose 100 / win 200 often rejected.

• 100 independent identical gambles: EUT predicts same rejection, yet colleagues accept bundle.

• Explanation: assess each gamble in isolation (narrow frame) within short-run mental account; longer horizon allows aggregation, showing favourable long-run distribution (only 1/2300 chance of any loss, 1/62000 of losing more than 1000).

Formal Expressions & Inequalities

• Marginal Propensity to Consume examples: \text{mpc} \approx 0.33,\ 0.7,\ 1.

• Rabin inequality example (first step):
  u(W+21) - u(W+10) \le u(W+10) - u(W)
  After adding 21 again:
  u(W+42) - u(W+31) \le u(W+31) - u(W+21)
  Iteration leads to near-zero MU rapidly.

• Samuelson bundle probability of any loss: 1 - (0.5^{100} + \dots) \approx \frac{1}{2300}.

Connections to Other Lectures/Theories

• Prospect theory groundwork: reference depend-ence & loss aversion central to mental accounting.

• Relation to Keynesian consumption models (mpc heterogeneity) & PIH deviations.

• Links with -economics topics (hourly wage framing, bonus vs. salary; upcoming material).

Ethical, Philosophical & Practical Implications

• Policy design: framing tax cuts, benefits, & subsidies to match consumer mental accounts can change spending multipliers.

• Consumer protection: marketers exploit pennies-a-day framing; regulators must weigh manipulation concerns.

• Personal finance guidance: encourage broader framing to combat sunk-cost fallacy & bad portfolio decisions, or exploit narrow framing for self-control (e.g., envelope budgeting).

Key Empirical Papers & Data Points Mentioned

• Thaler (1999) “Mental Accounting Matters”.

• Kooreman (2000) – Dutch child-benefit MPCs.

• Gourville (1998) – donation framing experiments.

• Landsberger (1966) – restitution payments & MPC.

• Shapiro & Slemrod (1995, 2009); Parker et al. (2011) – tax stimulus MPC.

• Hastings & Shapiro (2018); Hoynes et al. (2015) – SNAP framing.

• Odean (1998) – disposition effect.

• Gourville & Soman (1998) – gym attendance & sunk cost.

• Arrow (1971); Rabin (2000) – risk neutrality critiques.

Study Tips & Take-aways

• Always ask: “Which mental account is being used?” & “When does the account close?”

• Translate real-life anomalies into the triad: category, timing, risk attitude.

• In exam scenarios, justify anomalies via: reference dependence, loss aversion, narrow bracketing, hedonic editing.

• Use formal inequalities or MPC data to support qualitative claims.