Behavioural Flashcards
How analyse welfare if agents not rational and selfish?
- positive models of behaviour (e.g. hyperbolic discounting)
- develop bounds on welfare (based on observed choices)
Key deviations from homo economicus?
- social preferences
- bounded rationality
- bounded willpower/impatience/procrastination
- reference-dependent preferences
Type of discounting if individuals ARE time consistent?
Exponential
Type of discounting if individuals are NOT time consistent?
Hyperbolic
Discount factor between current period and future period w/hyperbolic discounting?
Beta * Delta (to the power of whatever time period it is)
Discount factor between 2 future periods w/hyperbolic discounting?
Delta
Theories about which discount factor to use in welfare calculations?
- Long-run preferences (Beta = 1 and Delta < 1, i.e. exponential, not hyperbolic, discounting)
- Short-run actual preferences (even if that means hyperbolic discounting)
- Multiple selves (implication - focus on long-run self)
Explain the idea of multiple selves (in terms of discount factor to use for welfare calculations)
- long-run self (the ‘planner’) w/discount factor delta
- series of short-run selves (‘doers’) that live for 1 period and are totally myopic (discount factor gamma = 0)
- long-run self sets incentives for short-run selves
- short-run selves choose action, given incentives
- short-run selves are the only selves that derive utility
- utility of long-run self = discounted sum of utilities of all short-run selves
WELFARE IMPLICATION - FOCUS ON LONG-RUN SELF
- What is an ancillary condition in Bernheim and Rangel’s (2009) bounds model?
- Example?
- Affects choice behaviour, but not experienced utility
2. e.g framing, salience
Broad types of possible intervention if people suffer from behavioural biases?
- paternalism (direct regulation)
2. behaviouralism (nudges)
….. and ….. (….) created a model to derive bounds on welfare using revealed preferences
Bernheim and Rangel (2009) created a model to derive bounds on welfare using revealed preferences
- Problem of ‘looseness of bounds’ on welfare
- Solution?
- Evaluate solution
- Problem - if behavioural issues important (e.g. framing, salience), then bounds so ‘loose’/wide that they become uninformative
2a. Solution - discard certain d’s as being ‘contaminated’ for welfare analysis
2b. e.g. intoxication reduces rationality so discard choices made when drunk
2c. Fewer d’s means tighter bounds on welfare
- Problem – refinements require positive theory of behaviour…(original problem!)
Outline Bernheim and Rangel’s (2009) model for deriving bounds on welfare
- Standard models: agents choose x (from choice set X)
- Behavioural models: agents choose (from generalised choice set G, where G = (x,d))
3a. ‘d’ = ancillary condition (affects choice behaviour, but not experienced utility)
3b. d could be framing, salience, default option
- Revealed preference: prefer x over y, if x always chosen over y FOR ANY D
Provide simple example of deriving bounds on welfare using revealed preferences
- Example – car insurance (3 different plans: 1, 2 and 3)
- Run 2 experiments (experiment A and B could represent different framings of survey, or different salience of options)
3a. Preferences in experiment A: 1 > 2 > 3
3b. Preferences in experiment B: 2 > 1 > 3
- Plan 3 never optimal for all d (whatever the ancillary condition is, e.g. framing), regardless of positive theory, so optimum bounded between 1 and 2
Chetty et al (2009)
AFFECT OF TAX SALIENCE
- Experiment:
(i) Vary salience of sales tax
(ii) Including sales tax of 7.35% in displayed price decreases demand by ~8%
- Empirical analysis:
(i) Exploit state-level changes in excise/sales taxes
(ii) US consumers more responsive to increase in excise duty on alcohol (included in displayed price) than sales tax (not included in displayed price)
Chetty et al (2009)
- Experiment:
(i) Vary salience of sales tax
(ii) Including sales tax of …..% in displayed price decreases demand by …..%
- Empirical analysis:
(i) Exploit state-level changes in excise/sales taxes
(ii) US consumers more responsive to increase in ….. (included in displayed price) than ….. (not included in displayed price)
Chetty et al (2009)
- Experiment:
(i) Vary salience of sales tax
(ii) Including sales tax of 7.35% in displayed price decreases demand by ~8%
- Empirical analysis:
(i) Exploit state-level changes in excise/sales taxes
(ii) US consumers more responsive to increase in excise duty on alcohol (included in displayed price) than sales tax (not included in displayed price)
Experimental evidence that consumers are more responsive to changes in more salient taxes
Chetty et al (2009)
Experiment:
(i) Vary salience of sales tax
(ii) Including sales tax of 7.35% in displayed price decreases demand by ~8%
Welfare analysis with salience effects - preference recovery assumptions
(1) taxes affect utility only through effects on chosen consumption bundle
(2) when tax inclusive prices fully salient, agent chooses same allocation as fully-optimising agent
- If tax salience affects behaviour, how might a less salient tax be more efficient?
- Under what assumption?
- Less salient tax affects behaviour less, leading to smaller fall in consumption and lower DWL
- Assumption = no income effects (dx/dz = 0)
- Under what assumption might a less salient tax decrease efficiency (if salience affects behaviour)?
- Example?
- Assumption = income effects (dx/dz > 0)
2. Agent who ignores complex tax on cars might under-consume food due to income effect (lowering welfare)