Behavioural Economics essay type questions

What is projection bias and what does it violate in the context of intertemporal choice?

the tendency to project current preferences into the future as if future preferences will be the same as current preferences In DUM assume instantaneous cardinal u is stationary

a) Explain social preference theory b) Are lab experiments an effective tool when considering social preferences?

a) Real life evidence suggests people act as if they care about others' welfare as well as their own, i.e. have "social preferences." Violations in analytical game theory due to A. Inequity aversion (Fehr and Schmidt, 1999) B. Pure Altruism C. Competitive preferences D. Rawlsian preferences E. Utilitarian preferences: (not an exhaustive list) b) Lab experiments pros and cons -separate B and standard explanations -control + T = causal -variation e.g. Smith and market focus. -assumption: insights gained in the lab can be extrapolated to the world beyond -Hawthorne effect *Game explained* Gift exchange experiments Explained 1. Similar to the trust game 2. But the money passed by the first mover (offer), is a pure lump-sum transfer. 3. First mover requests desired effort/quality level in return for offer 4. Second mover chooses effort/quality 5. Effort is costly but increases the first movers payoff *Context* Used experienced sports-card traders as subjects. *Behavioural findings* LAB EXPERIMENT FINDINGS -offer is typically > min allowed - If social preferences are present for the card sellers then buyer who offer more money should be awarded with better quality cards -find + relationship between offer and quality REFINED LAB EXPERIMENT FINDINGS -\robust\ to even when trading real basketball cards requires certain level of knowledge to know value -where prices are sensitive to condition of card NATURAL EXPERIMENT FINDINGS -sellers unaware being watched -Little statistical relationship: when belief cards could not easily be graded little chance further interaction Implication (7) -much of the *observed behaviour consistent* with *social preferences* is due to reputational concerns -other confounders also e.g. Hawthorne effect -results still not generalisable even with refined lab experiment !!! -social pref effects are lessened -field data reveals underlying conditions / causes -more imitates self-interest theory -social pref theory better suited less market pressure e.g. charity giving

Are people overinfluenced by the weather at the time they make decisions?

(Conlin, O'Donoghue, and Vogelsang, 2007) EXOGENOUS VARIATION Definition + people understand qualitatively how tastes change over time, but underestimate the magnitudes Do: Field evidence. Compare catalogue order purchases and returns of weather-related apparel, regression analysis Why weather (3) -clear and salient change in context -weather is exogenous -shows revealed PB in the context of consumer behaviour Classical prediction (3) fully rational ind, identical choice irregardless of weather, over-arching pref matter Behavioural prediction (2) 1. greater likelihood of returning cold-weather items if ordered during a cold time as project pref for cold-weather items into future. 2. Likelihood returning items: declines in order-date temp, inc in return-date temp Findings (4) - people's decisions are overinfluenced by the current weather - if the order-date temp falls by 30°F, the return probability increases by 3.95 percent (negative relationship) -robust days of the week -extrapolated to other contexts Alternative explanations (4) 1. Learning about adequacy of current clothing - lower order-date temperatures should then decrease the likelihood of return over time 2. Current weather affects search propensity but no effect for non-cold-weather items 3. Current temp predict future temp hence lower order-date temp should decrease chance of return 4. Current temp affect beliefs about future temp: Control for temperature 7 days prior to order (but still fundamentally unable to distinguish PB from bias of the order-date temp in mispredicting future temperatures

What is the empirical relationship between income and subjective well-being in the United States? Include reference to the results of Deaton and Kahneman (2010) and Luttmer (2005). Deaton method Deaton findings explanation of easterlin paradox

*Easterlin paradox* Definitions (2) Decades of recent growth does not transpire into *higher reported well being measures* despite *consistent relationship* between *happiness and income* Positive relationship in the cross section but not over time N.B: Objective measures have improved *Graph Clark et al* - gradient between Y and H is steeper within country than across time - Due to status benefit of high Y country which has no aggregate impact on country level H -t0 relatively poor hence steep, t2 country is richer hence flatter - Individual income and individual happiness yields steeper relationships - Formulas - MU tends to one as relative income rises -tell relative income story Reasons 1. Easterlin explains that it is relative Y or C that matters for w-b 2. Could be due to shifting reporting function leading to differences in interpretations Layard (2016) highlights three main reasons within the literature 3. Social comparisons A- Positional externalities Generally, income is a positional good but leisure is not These ppl prefer more income than others in their reference group. People spend too much time working to achieve what is at best a temporary gain in relative income B- Relative income effect upwards comparison for self-improving downwards comparison for self-enhancing Trade off and compare ourselves to people broadly similar C- Luttmer paper (PUMA and reported H) Negative income in area and your H Mimic work longer hours than socialise Disagreements about finance Bias overvalue extrinsic rather than intrinsic values Leisure declines in relative Y Costs housing not a mechanism (as positive rel earnings in area and satisfaction of town) Measures robust different w-b measures D- Frank 1997 town with 3000 square feet garden c.f. town with 5000 square feet garden d not interact no sig inc SWB from increasing garden size for all 4. Adaption - All living organisms respond to external changes in ways that restore their internal balance. - harder to secure *permanent* increases in happiness through increases in income - (Layard, 2005) Uses survey data to show that income increases happiness in S/R but not the L/R - Income is partially *addictive*. Having once experienced a higher sofL, hard revert to back and feel the same - value based on social norms, as growth increases so do aspirations which constrains w-b benefit -Clark et al spike graph 5. Tastes social factors can affect our *ordinal* preferences and may also affect the *cardinal happiness* we get from a given consumption bundle, even if they have no effect on our indifference curves. Limitations - certain threshold income -Stevenson and Wolfer: no empirical evidence of satiation point (critical value in which income no longer impacts w-b) Only use eval measures - Oswald does find positive relationship with another dataset -Vary dependent on SWB Empirical findings 2. Deaton and Kahneman -Emotional w-b vs life evaluation definitions -Data (450k residents) -Method -Findings (6) -adaption graph 3. Relative income effect - upward and downward comparison -> comparison similar people -Endogenous -Literature uses mean ideal measure is weighted mean over ref group by exposure 4. Luttmer findings ( 7)

Explain why models in Behavioural Economics and Traditional economic theory are "as if" models and how Neuroeconomics can inform the economic theory? As if (5) Inform eco theory (5)

1. *'As if models'* (5) -Economics based on axioms of rationality -The brain is the ultimate "black box" working on as if assumptions -The foundations of economic theory assumes that details about the functioning of the brain's black box would not be known. -Models in BE and Traditional economic theory consider the economic behaviour on the basis of "as if" assumptions (i.e. "as if" individuals are rational, "as if" they have hyperbolic preferences, etc.). -BE developed alternative models to further represent behaviour -but methods are similar i.e. observe actual behaviour 2. *Neuroeconomics definition* According to David Laibson it is the study of biological micro-foundations of economic cognition -bio micro-foundations = neurocehmical mechanisms e.g. brain systems -Biological micro-foundations are neurochemical mechanisms, like brain systems, neurons, genes, heart rate, skin resistance, and neurotransmitters -Eco cognition includes mental representations, emotions, expectations, learning, memory, preferences, decision-making, and behaviour -focused on the process of choice and how agents come to decide certain aspects, providing the basis for such models 3. *Inform economic theory* (5) -inform eco theory in two key dimensions If inexpensive: 1. explaining how some key assumptions in models may be unfounded 2. via ability to observe not just the outcome but the process -> improved measurements of u catalyse and direct new models speed up the progress of model development provide new empirical methods that will sometimes provide new tests provide different tests for robustness of theories 3. eventually be able to prevent mispredictions by agents and help people to better understand and manage themselves 4. Neuroeconomics questions In terms of undermining particular assumptions, neuroeconomics questions the measures of [[discount rates, assumed cognitive abilities and capacity]] •Declining discount rate - McClure •Money provides u only indirectly from the g/s it can buy - same dopaminergic reward activity is found in drugs, money, funny cartoons and attractive faces so it may be that due to these systems, people don't perfectly compute the value of money in terms of what it can buy o why people are quite careless with "play money" and other times companies disguise payments or people buy health club memberships to avoid the emotional elements of these •Not unlimited cognitive capacity - numbers and fruit cake, self-control

What is the Allais paradox and how should standard expected utility theory be modified to explain the paradox? (18 Marks)

1. Allais paradox matrix -anomoly EUT in experimental data 2. Common consequence effect -Subjects first presented A/B -Independence axiom implies same payoff in 3rd state world -Ignore 3rd state world -A/B are identical to C/D EUT prediction -But violated 3. Common ratio effect -Choices between A/B and C/D same payoff and relative probabilities -EUT prediction -But violated despite the ratio of probabilities being the same 4. Most common explanation for the paradox is: - D-Mers weight the prob of outcomes via a function (π) - that overweights small values of p - underweights large values of p -i.e. Prospect theory. Even though the likelihood of a costly event may be miniscule, we would rather agree to a smaller, sure loss — in the form of an insurance payment — than risk a large expense. 4. Define prospect 5. Value function explained (6) 6. General formula 7. Graph and interpretation (8) 8. Weighting function intuition i.e. p is non-linear (7)

How does behavioural economics differ from neoclassical economics? Is one better than the other for policy?

1. Definitions Neoclassical economics: normative approach, axioms of rationality, u only observable by choice, ordinal utility Behavioural economics: psychological insights, human behaviour, economic D-M. 2. Neoclassical (6) Separated Psychology and Economics Rejected assumptions of Benthamite Utility Revealed pref: infers pref from observing choice Welfare + choice identical Reasons for choice are irrelevant Anti-paternalistic culture 3. Obj Behavioural • The objective is to modify, supplement and enrich economic theory by adding insights from psychology. • Aims to do is provide some psychological foundation to such economic choices 4. Examples of differences Individuals care about fairness (social pref) Risky outcomes are not weighted linearly Present biases Mistakes / preference reversals 5. Process AAEICI A. Identifies normative assumptions of models e.g. Bayesian updating or EUT B. Identifies anomalies where there are clear violations of the model. C. Rule out other explanations e.g. t.c D. Use the anomalies as inspiration to create alternative theories that *generalise* existing models E. construct eco models of behaviour using *behavioural assumptions*, derive fresh implications and test them F. Fully integrate them into eco analysis by embedding old and new assumptions as special cases of general models (enhance not replace) 6. Similarities (3) new BE assumptions studied using standard mathematical methods tested using standard statistics /metrics judged by standard scientific criteria 7. For Classical theory (Gul and Pesendorfer, 2008) -objective tool to present models (could argue a lot of normativity embedded though...) -sufficiently flexible that irregularities explained in model - e.g. mistakes explained by changing feasible set of strategies (American tourist) -Own opinion: despite limits to classical validity / applicability often appropriate 8. For BE (rebuttal to G&P by Camerer, 2008) • argument lies on (subjectively v. narrow) defined boundaries of economics • BE illustrates progress • auxiliary assumptions (tourists) are 'ad hoc' • advocates neuroeconomic o brain studies can identify non-price signals that will affect economic choice o brain studies offer a method for choosing between competing theories that explain equally well some economic phenomenon o brain studies can assist in explaining why things like "mistakes" occur (where a mistake is a choice not aligned with preference). o Addiction o Enhancements o Questions assumptions • Tourist example is stroop task, learning effect, hence not due to change in feasible set • economics is wrong (assumptions) but strong (evaluation) • Explores biases and complementary welfare suggestions 5. (Atkinson, 2011) BE is neglected due to (3) -assume away difference between ppl -assume agreement in welfare criterion (max CS) -assume obj to identify set poss PE outcomes -value not CHOICE matters (although overlap) -argues that welfare should be restored -role in informing policymakers

Explain the disposition effect using the paper by Odean (1998). Why is the disposition effect is usually not observed during the month of December? robust alternative solutions

1. Definitions (2) Tendency to hold losers too long and sell winners too soon coined by Shefrin and Statman (1985) In this paper defined as: the tendency of investors to hold losing investments too long and sell winning investments too soon 2. Implication of Prospect Theory • Value function brief explained • Status quo is coded relative to an expectation / aspiration • Example: Suppose there is a stock that an investor believes will yield high enough return for the risk. _If the stock appreciates_ and the investor continues to use the purchase price as a ref point -> more concave/risk-averse point in V. If the investor lowers expectation of the return she is likely to sell _If the stock depreciates_ -> price in risk-seeking, convex point in V. The investor will continue to hold the stock to *justify its original price.* So investors belief must fall further before she will sell 3. Data Analysing trading records for 10,000 accounts at a large brokerage house 4. Methodology Investor one Five stocks: ABCDE A,B worth more than paid C,D,E worth less than paid Sells A and C Investor two Three stocks: FGH F,G worth more than paid H worth less than paid Sells F Paper gains: B,G Paper losses: D,E,H Realised gains: A,F Realised losses: C PGR =RealisedGains / (RealisedGains + PaperGains) PLR = RealisedLosses/ (RealisedLosses + PaperLosses) Large difference in PGR and PLRM -> indicates investors are more willing to realise either gains or losses. Test of the disposition effect is a joint test of the hypothesis that people sell gains more readily than losses Assumptions (2) -Ref point is purchase price -Independence assumption: investors realise gains too soon and losers to long *means each account is weighted by the number of realised and paper gains/losses in that account* 5. Alternative explanations (2) • MEAN REVERSION Investors either rationally (if future expected return loser > winner) or irrationally believe that current losers will *outperform* current winners in the future and hence sell winners to rebalance portfolios • TRANSACTION COSTS Refrain from selling losers due to higher *t.c* of trading at lower prices 5. Results (3) • strong preference for realising winners rather than losers (exhibit disposition effects) • *robust* when controlling for rebalancing and share price behaviour not motivated by a desire to re-balance portfolios or to avoid higher trading costs from lower priced stocks ie. disposition effect still holds • losses realised at higher rate relative to gains in Dec 6. Why results not observed in December Investors engage in tax-motivated selling Finance stuff: investors should postpone taxable gains by holding their profitable investments and capture tax losses by selling them Shefrin and Statman: investors choose to sell losers due to self-control measure- reluctant to sell for a loss but recognise tax benefits so wait till deadline Tax-swap: reconcile tax-loss selling with aversion to realise losses by selling a losing stock and purchasing stock with similar risk characteristics Thaler (1985) people tend to segregate different gambles into separate mental accounts, which are evaluated separately for gains/losses (hedonic framing) One clear intuition is that a realised loss is more painful than a paper loss. When a stock is sold, the gain or loss has to be 'declared' both to the tax authorities and to the investor. Because closing an account at a loss is painful, a prediction of mental accounting is that people will be reluctant to sell securities that have declined in value. 7. Limitations Do not distinguish between rational mean reversion and disposition effect argue ind themselves do not differentiate the two Independence assumption: If an investor repeatedly decides not to sell a stock these decisions are endogenous and not independent

Consider the general utility function for social preferences specified in Charness and Rabin (2002). What are the main empirical results from this paper? Include reference to the parameters of the model. Motivation, Method, Model

1. Motivation (2) Questions inequity aversion -retaliation -BUT Pareto damaging behaviour that is plausible lowers equality Questions social welfare -driven by social welfare of inequity aversion? -difficult as we only see the eff sacrifice 2. Method (3) -evidence from 29 games -test the role of diff social pref explanations -eliminate confounds e.g. Ultimatum to Dictator and research if poss of negative reciprocity without threat of retaliation 3. Model Utility function -u is the weighted outcome of our own payoff (whether ahead or behind) and a weight of the other player's strategy either to misbehave or cooperate -Embeds different theories σ <= ρ <= 0 Competitive σ < 0 < ρ < 1 Inequity Aversion 1 >= ρ >= σ>0 Social welfare θ >0 Reciprocity 4. Results (7) DICTATOR GAME (400,400) vs (750,400) - <1/3 exhibit inequity aversion even when costless ULTIMATUM GAME (800,200) vs (0,0) - strip reciprocity and find nobody goes for equal split but more strictly positive split -few sacrifice money to increase equality -only minority favour equality even when costless -50% make inequality increasing sacrifices when eff and inexpensive -Inequity aversion better predicts when ahead -social welfare better prefernce explanation otherwise

Describe the trust game analysed by Vanberg (2008). What interpretation does the author offer to explain why behaviour in experimental settings often differs from the prediction offered in classical game theory? Indirect and direct explanation Issue Do (3) confusing sentence.. Prediction

1. Trust is important (3) 2. Hypothesis (2) -pre play communication enhances cooperative behaviour (guilt aversion / expectations) promises affect behaviour only indirectly-> changes in the payoff expectations -preference for honesty (lying aversion / commitment) promises affect behavior directly, because people have a preference for keeping their own word. Issue is they correlate... 3. Do -Game tree (two stage mini dictator game with random dictatorship by nature) -Dictator choice identical to trustee choice in C&D -Opp prior communication -How distinguish between lying and guilt aversion -Switch group measures expectation -No switch group measures lying -to achieve independent variation in promises and promise-induced changes in second-order beliefs. 4. Predictions Expectations-based reasoning: switch or no switch condition is identical Commitments-based reasoning: Cooperation more likely in no switch as YOU would be lying if deviate 5. Findings -a dictator's own promise significantly affects behaviour -promises made by others lead to significant changes in second-order beliefs, but do not affect behaviour - so effects of promises are irrelevant in switch -/> changes in payoff expectations -hence people pref for honesty

How are the habit formation and the reference point model joined?

1. where zt in HF is the ref point 2. Direct C or ref point effect on utility 3. Main difference is that in HF there is no assumption of loss aversion or DMS in these combined models

Explain how Neuroeconomics can help to explain the hyperbolic discount rates.

1. Uses the multiple system hypothesis (4) a. brain makes decisions (constructs value) by integrating signals from multiple signals. b. these systems process info in qualitatively different ways c. in some cases differently weight attributes of rewards d. sometimes system are in cohesion other times conflict 2. Past literature talks about the duality of individuals wrt interests vs passions 3. Neuroscience formalises this: Affective system fast, unconscious, impulsive involves emotion typically results in desire or drive to do sthing e.g. hungry so eats Analytical system slow, conscious, forward-looking thought processes that are almost free of emotion e.g. driving a car 4. Intertemporal choice (IT) (5) - IT driven by two systems: one more cognitive and deliberative - rational part of the brain is associated with the standard exponential portion or time-consistent element of hyperbolic discounting - whereas the automatic and affective system is associated with the present bias. - During decisions involving both immediate and future payoffs, different areas of the brain are activated. Hypothesis of McClure key hypothesis is that the pattern of behaviour summarised by the two parameters • β (which reflects the special weight placed on outcomes that are immediate) and • δ (which reflects a more consistent weighting of time periods) Method McClure examined the brains of individuals whilst they made decision between monetary rewards of different values and delivery times Immediate payoffs/rewards -(beta) -the limbic system or NAcc is activated where there is short-run impatience and a greater focus on instant satisfaction -This is the area that is typically activated during drug addiction. Future payoffs -long-run patience is mediated by the DLFPC -able to trade-off more abstract costs and benefits -including those in the future (delta) During IT decisions, there is often conflict between these two areas and the relative strength of activation can be indicative of which will overrule the other. McClure findings - β areas are activated disproportionately when choices involve an opportunity associated with limbic structures and are known to be rich in dopaminergic innervation. -Hyperbolic discounting makes individuals impatient in the short run which would be seen by higher activation of the limbic system -found the limbic system showed greater activation for the immediate rewards. -N.B: This experiment also showed that it is possible to show impatience towards goods with immediate satisfaction, like money, not just drugs and therefore this impatience is not due to the addictive nature of the good. 8. Findings (3) -Hyperbolic discounting implies we are always impulsive when faced with right incentives (usually immediate cost and benefit) -BUT not the case, competition between analytical and affective system -Neuroeconomics suggests the effects can be moderated by factors which strengthen or weaken each of the two systems. For example, cognitive overload can take away potential from the cognitive system and make it easy for the automatic system to override (E.g. stress from work can inhibit one's ability to control eating habits, taking drugs accentuates activation of affective states). -can explain the uniquely human propensity to experience fear / anxiety when thinking about the future, giving rise to impulsivity vs workaholics.

(a) What contribution has Neuroeconomics made to our understanding of present bias? (14 Marks) (b) What other biases affecting decision making can be explained by the dual brain system? (14 Marks)

a) Intertemporal choice question b) Ultimatum game: trust / risk aversion

What are the implications of time inconsistent preferences in the real world?

A person with time inconsistent preferences may or may not be aware that they have such preferences PROCRASTINATION TCs, Naifs, Sophisticates Crucial findings: -Naifs exhibit solely present biased effects, when there is a trade off between two future outcomes there is a stronger relative weight on the earlier moment as it gets closer / overestimate the benefit of waiting -Sophisticates will do earlier than Naifs irrespective of if it is an immediate cost or reward / have tendency to mitigate procrastination as pessimistic about future behaviour -TCs stick to their optimal lifetime plan WORKERS It has been found that workers with self-control problems do not work as hard as they would like. This contradicts agency theory as it is found that workers incorrectly belief that future behaviour will reflect current beliefs, and they therefore end up slacking. But instead, using a one year (!) field experiment three key findings were made: 1. large no. of workers are sophisticated and have a preference for a dominated contract to motivate higher effort. (penalises low output but does not reward high output) (dominate effect) 2. Effort increases as payday gets closer (payday effect) 3. These effects are heterogeneous- those with greater payday effects are more likely to choose the dominated contract Present bias among workers encourage firms to implicitly adopt commitment devices either via *high-powered incentives* or *impose work rules to monitor effort*. This has a direct negative effect on profits and therefore there is interest on both sides to mitigate procrastination. STOCK MARKET Literature analysing the impact of the weather on the stock market have found a strong significant correlation between sunny weather and stock returns. Thus stock markets are systematically influenced by investor psychology, whilst a rational market would not respond to this biased sample of news Mechanism There is a psychological bias: sunlight affects mood, meaning people tend to *evaluate future prospects* more optimistically when they are in a good mood c.f. bad mood hence NOT fully rational price setting People in good moods tend to generate more unusual associations, perform better in creative problem-solving tasks, and show greater mental flexibility These results are found to be causal and robust as shown by (Saunders, 1993) ADDICTION S are more likely not to consume addictive goods as a self-control measure. This is because they are pessimistic about future behaviour and know they will overconsume. Those with present biased preferences are more likely to be addicted.

Discuss the Bernheim and Rangel (2004) model of addiction, what new contributions do the authors give to the study of addiction? HFM process (5) MDS (4) 1 phrase I like Story (6) 4 forget Findings (3) 2 forget

Addiction is defined as repeated, and unwanted use despite clearly harmful consequences. Addiction has been shown to interfere with normal brain processes in the rewards part of the brain -> strong misguided cue-triggered behaviour. The process is the Hedonic forecasting mechanism which links cues with short-term hedonic responses. - HFM learns through *experience* - it associates an action with *anticipatory biochemical responses* - leads to systemastically skewed D-M - An addictive substance *acts* directly on learning process of HFM - Causing the brain to *grossly* exaggerate pleasure responses MDS process where this process occurs MDS usually fires in response to reward not a cue But over time if reward abstained fires to a cue instead Intensifes over time -> cue conditioned dopamine response Model 3 premises based on psychology / neuroscience i) use of addictive good is a mistake ii) sensitivity to environmental cues is triggered by past experience iii) addicts understand they can be triggered by cues and try to manage addiction with sophistication Once exposed to enviro cues... 1. the ind may enter a *hot* D-M mode where always consumes the substance irrespective of underlying preference 2. the ind may enter a *cold* D-M mode where he considers all alt and consequences, inc the effects of current choices on the likelihood of entering hot in the future. (cognitive control) 3. Story A. An ind first selects a "lifestyle" activity (a); B. he allocates resources between an addictive substance (x {0, 1}) and a nonaddictive substance (e >= 0) C. He enters each period in the cold mode and chooses his lifestyle activity rationally. D. This choice, along with his history of use and other environmental factors, determines the *probability* with which he *encounters* cues that trigger the hot mode. E. If triggered, always uses substance even if this is not his best choice. F. If he is not triggered, he rationally decides whether to indulge or abstain. Findings 1. C depends systematically on the *characteristics* of the individual (including aptitude for cognitive control), the *substance*, and the *environment*. 2. *Experienced uses take more hard core drugs* when one substance is more addictive than another, then (ceteris paribus) the more addictive substance is associated with less consumption among relatively new users, but with more consumption (both intentional and accidental) among highly experienced users. 3. For *highly addictive products* If w-b declines rapidly, never choose substance e.g. crack cocaine If w-b declines temporarily and accelerates -> start to consume -> take part in half-hearted abstention -> precommitment to abstention through rehabilitation can lead to re-offence If w-b flattens now completely addicted to the good without hope of abstention

Give application examples of the alternative models of Instantaneous Utility

Applications of Habit-Formation models (6) 1. effect p on addictive products 2. asset pricing anomalies (EPP) 3. Movement in asset prices in business cycle 4. Growth rates -> more saving 5. aggregate spending hump shaped IRF 6. After shock C is sluggish in S/R only Applications of Reference Point models 1. Shifts C undesirable due to loss aversion 2. Delay speed-up asymmetry 3. Loss aversion emphasises aversion to delay 4. PIH (shock to Y leaves C growth unchanged, but RP model says unwilling to lower C now -> fall in growth of C in future)

What are the some of the standard assumptions in analytical game theory, and how do they relate to the empirical concerns of behavioural game theory? Forget (1)

Assumptions -Mental presentation of the relevant game -Unbounded rationality -Self-interest -Equilibrium is reached instantly Concerns BGT: -Questions how we interpret a game -Limits strategical reasoning / flaws in backward induction -Fairness / Social preferences -Introduces a learning effect / comments on likelihood of MSE / whether NE is reached Examples 1. Iterated games (rationality) 2. Minimum efficiency coordination (likelihood of MSE) 3. Traveller's dilemma (sensitivity cost parameter)

What are each of the models explaining and not?

Becker and Murphy Cannot explain: commitment (rational), mistakes (rational), present bias (TCs), random re addiction because (cyclical) Laibson Cannot explain: mistakes (too costly to delay) cravings (difference between anticipate and avoid)

Difference between guilt aversion and lying aversion

Both concepts agree that people feel guilty about breaking promises Guilt aversion assumes that guilt is triggered by behaviour that leads to an outcome which is inconsistent with others' expectations Lying aversion here guilt is triggered by behaviour that is inconsistent with contractual or moral obligations concerning what "should" be done

Describe a model in which consumption of an addictive good with known detrimental longer-term consequences is compatible with rationality.

Becker and Murphy (1988) *Rational model of addiction* Addiction is defined as repeated, and unwanted use despite clearly harmful consequences Utility function and graph Ut(c𝜏, a𝜏, s𝜏) a- parameters s𝜏 stock of past consumption= d(s𝜏-1 + a𝜏-1) state of addiction, its accumulation is positively dependent on current C, negatively depends on addictive decay of the past state of C (δ), influences u directly and indirectly via wages c𝜏 present consumption a𝜏 addictive good b- The key set up of the model looks at a *time path for consumption* where at any moment in time a consumer is in a state described by a function related to the level of past consumption of the addictive good. Consumers are considered to be time consistent (exponential discount rate) meaning that their decisions are not some form of imperfect rationality. c- plans addiction by max intertemp expected u, choosing c𝜏, a𝜏, s𝜏 in each period d- c𝜏 is based on how addictive the good is and what will likely happen to its future price First looking at addictiveness of the good: Addiction is characterised by *adjacent complementarity (A.C)* i.e. current consumption influences future C. Changes in prices also impact consumption. - The main idea is that of *forward looking* behaviour. - If goods are *addictive and future price is set to inc, should consume more now.* - If goods are less addictive, not expect price inc, can consume now - inverse relationship current C of a𝜏 and all prices - *Long term effect* of price on demand is greater for addictive goods and exceeds short term effect shown in graph. - Current C is also impacted by future prices (negative cross price effects) hence it takes time for policies to work Graph o suppose agent is at Co=δS along A1 o A fall in price of the addictive good -> increase in D curve from A1 to A2 o C initially rises from Co to C1 o C then grows further over time as C1 is above the steady state line. o Grows to new steady state C2=δS' Discussion (4) 1. key point is that decisions to consume the addictive good can be a rational D-M process. Consistent with observed data and individuals are perfectly aware of the harmful effects of consuming the addictive good. 2. Due to its rationality there is no need for gov intervention except to correct for negative externalities e.g. second hand smoking 3. These individuals are maximising their own welfare and anything we do will harm them! 4. Different people with different time preferences make different choices. o Present-biased individuals more likely to be addicted o Sophisticates do not take the addictive good as a self-control measure as pessimistic about future behaviour Limitations (4) -can't explain desire for commitment as time consistent -present bias means not stick to original plan -model says agents can smoke more if they think cigarettes are less addictive, but underestimate addictiveness due to conformity bias / law of large nos. -overestimate chance of quitting

What are the three reasons that Layard (2006) gives to explain the claim that people in the West are no happier than 50 years ago. What policy implications does he draw?

Believes that economic policy needs radical reform to incorporate psychological insights Believes the prime purpose of social science should be to discover what helps and hinders happiness SOCIAL COMPARISONS *Issue* 1. We compare our incomes with those of others. 2. If others become richer, this reduces our *satisfaction* with whatever we have. 3. The conventional wisdom is that people compare themselves mainly with people who are close to themselves in the income distribution, but if income distribution is reasonably stable the reference group will be proportional to average income 4. It is sometimes suggested that people are more concerned with their *rank* order in the income distribution than relative income. (Although experiments with the US General Social Survey suggest otherwise) *Theory to include this* 1. Hence, U=u(y-αy ̅,h) 2. Average income tends to attract a large and significant negative coefficient. (whether cross-section or time series) 3. In some studies the negative effect of average income is almost as large as the positive effect of own income. *Policy implication* This is a case of *negative externality* and so the external cost can be internalised via: a corrective tax which will reduce work effort to a level where the incentive to raise relative income has been fully offset Use of average income needs further debate as if people only compared their incomes with *incomes above* their own, the optimum tax would be more *progressive* than otherwise. Need to rethink measures of 'excess burden' where any tax wedge is distorting as people will overwork without a tax. ADAPTION *Issue* 1. All living organisms respond to external changes in ways that restore their internal balance. 2. Adaptation does make it harder to secure *permanent* increases in happiness through increases in income 3. (Layard, 2005) Uses survey data to show that income increases happiness in S/R but not the L/R 4. Income is partially *addictive*. Having once experienced a higher sofL, hard revert to back and feel the same 5. *Theory to explain this* 1. U=u(y-β_y-1, h) 2. Add lagged income to the happiness function, with a negative effect 3. In a Swiss study looking at happiness and income, lagged income is a major influence on income-aspirations, and this has been confirmed by numerous studies by Van Praag et al. *Policy implication* 1. Habituation to income is only a problem for public policy if this effect is *unforeseen* 2. Frank calls this a negative internality 3. Evidence shows that people *overestimate* the extra H gained from extrinsic values -> people will work too hard and consume too much 4. Suggest a possible corrective tax: the required correction is towards lower work effort and thus lower C 5. Loss aversion makes this sort of stabilisation policy difficult TASTES *Issue* Economics normally assumes that tastes are given 2. But social factors can affect our *ordinal* preferences and may also affect the *cardinal happiness* we get from a given consumption bundle, even if they have no effect on our indifference curves. *Theory to explain this* 1. U=u(y,h,T) 2. There are many other taste variables to include *Policy Implication* 1. Limit advertising: provides information, but also makes us feel we need more money than we should otherwise have felt we needed. - US General Social Survey provides data on how a person perceives their position in the income distribution. If we regress this estimate on a person's actual income and the hours he watches TV, we find that watching TV makes a person feel poorer. 2. Reduce performance related pay: which sought to align interests of principal and agent - But Kube et al example / studies find that extra financial rewards reduce internal motivation as tastes change over time - Competing with colleagues often causes a lot more strain with uncertain gains 3. Hollistic approach: more broadly explicit consideration that values matter (including the values which we expose our children to as they develop, in schooling and elsewhere)

How does the B+M rational model differ from the Cue model

Cue Model Definition Factors which people experience *(cues)* in the environment can sometimes elicit *changes in their preferences* e.g Smell of food can trigger people's desires to consume it Process (6) A. A current cue is complementary with current C if that cue has been associated with C in the past e.g. smell smoke fumes and nicotine cravings B. Each time period choose between primary activity (done before) and alternative activity (no conditioning effects) C. Consumer learns of a binary level of cues which occurs with random probability D. These are cue signals which are green or red E. Each time period, a cue is realised and the consumer chooses between a primary or alt activity F. consumers will seek to manage cues through expending resources Predictions (2) o Cue model predicts *falling discount rates for primary activity* and *constant non-zero discount rates for alternative good* / Predicts that agents will exhibit high levels of short run impatience when exposed to reward-availability cues. When a subject is faced with a cue she has a pref to consume now t0. If she cannot consume now this *does not affect her pref in future states*. i.e. the cue only impacts decisions at time t0 Experimental evidence on children show that patience is shorter when exposed to the reward. Had to decide between having a small reward now (marshmallow) or a larger reward later (cookies). o Cues initiate psychological changes where it inc MU of C of that good -> delay being costly e.g. ice in a glass and whiskey Findings o Explains why preferences change rapidly from moment to moment o It generates multiple steady states of the compensatory variables: (0,0), (1,1), (1,0), (0,1) - first two indicate no addiction and addiction. - The last two are the interesting ones in which actions are cue sensitive. Whereby a consumer engages in primary activity in red OR green state only. o Desire for cue management: consumers have a desire to control response to red or green signals 3. Differences Model provides a micro-foundation for the complementarity affects in the BM model. 1. A new model to explain variations in cravings. Becker predicts that binges are cyclical (by assigning weight K as sub good and eating K as complementary). But they argue re-addiction is random, slight variation in a cue -> large effects on MU In Becker staying on steady state line is optimal, this is unlikely to actually happen. Therefore cues model shows how consumers manage their environmental cues. e.g. hide your cigarettes 4. Policy implication Since cues can affect MU, it can change C patterns. Using this cue mechanism to nudge people into having "better" behaviour is thus a form of paternalistic libertarianism. People want to manage cues but are unable to. For example increased cues of healthy foods in schools so as to encourage their consumption 5. Limitations - Cannot explain people wanting to quit but failing regarding past actions as mistakes - or people anticipating cravings (e.g. going for in-house treatment to avoid but they think they will have cravings) Bernheim and Rangel allows for this via a process where a hot state forces the person to consume the addictive good.

Chetty et al 2014

Data: 41 million observations on savings for the population of Denmark Useful as contains information on saving in all accounts, including changes in defaults Do: Chetty et al (2014) contrast impact on retirement savings of: *Neoclassical tools (t), tax subsidies for retirement saving* In 1999, the Danish government reduced the tax deduction for contributions to capital accounts for the top tax bracket from 59 to 45 cents per Danish Kroner rely upon individuals to take an action to raise savings targets active savers who are financially sophisticated *New policy tools (n), defaults and automatic enrolment plans* policies that raise retirement contributions with no action i.e automatic enrolment targets passive savers Results: They show that the effectiveness of retirement saving initiatives by government depends on if the policy changes saving rates for passive or active savers Neoclassical results: - in align with what as expected reducing the subsidy lowered contributions -Individual-level responses suggest the entire effect is driven by the approx. 20% who are 'active savers' i.e. switch contributions from capital pension accounts to annuity pension accounts or other savings -80% unresponsive 'passive-savers' -Hence, each $1 of tax expenditure on retirement savings subsidies increases retirement saving by approximately $0.01 (in this context) Behavioural results: - Default options are effective e.g. Chetty (2015) - Madrian and Shea (2001) where automatic enrolment increases participation from 20% to 80% -$1 increase leads to increase in net savings by $0.85 Hence behavioural is better as: (1) subsidies induce relatively few individuals to respond (2) they generate substantial crowd-out conditional on response (3) they do not increase the savings of passive individuals, who are least prepared for retirement

What are framing effects? Explain with some examples.

Definition Choices can be presented in a way that highlights the positive or negative aspects of the same decision -> changes in their relative attractiveness. Types: (Levin et al., 1998) 1. *risky choice* framing (e.g. the risk of losing 10 out of 100 lives vs the opportunity to save 90 out of 100 lives) 2. *attribute framing* (e.g. beef that is 95% fat vs 5% lean) 3. *goal framing* (e.g. motivating people by offering a $5 reward vs imposing a $5 penalty) Kahneman & Tversky, 1979 framed gambles in terms of losses or gains -> risk-seeking vs risk-averse Certainty effect: when probably outcomes are underweighted relative to certain outcomes which leads to risk aversion to in sure gains and risk seeking in choices with sure losses

Explain the notion of overconfidence and give two applications of it

Definition: often referred to skill distribution biased beliefs lead to suboptimal decisions Three reasons behind overconfidence 1. Advertising -Selling to overconfident consumers -evidence that this systematic bias is the norm -creates incentives offer zero MC tariffs -with subsequent steep marginal charges -e.g. mobile phones 2. Excess entry -overconfidence in firm behaviour -experimental setting with basic features of business entry situations -success depends on relative skill -Most subjects who enter believe total profit of all entrants will be negative, but their own profit will be positive -compared to random device there is more entry -more firms enter than what is profitable

Compare between the discounted utility model and hyperbolic discounting, and how can the latter explain preference reversal? Use algebra and graphs were applicable. Assumptions (3) Preference reversal intuition (3)

DUM 1. Instantaneous utility function D(k) = (1/1+p)^k p is discount rate= inds pure rate of time preference Assumption: *Time is discounted at constant rate* delaying or accelerating two dated outcomes by a common amount should not change preferences between the outcomes Violations - Assumption is violated when we look at the discount rates in short time horizons and long term horizons (over time). This model cannot explain why £110 in 31 days is preferred to £100 in 30, but £100 today is preferred to £110 tomorrow ie. there is *preference reversal* - Present bias extensions are O'Donahue and Rabin, 1999 (S+N) - Discount rates also vary across IT choices Empirical findings •small outcomes discounted more than large •gains outcomes discounted more than losses •sequences of multiple outcomes discounted differently to single outcomes Solutions •Hyperbolic discounting can explain, whilst DUM cannot •Hyperbolic discount rates imply a decreasing intertemporal discount rate (pn) •pn empirically found not be constant •So the implicit discount factor over longer time horizons is higher than the implicit discount factor over shorter time horizons • D(K) increases graph • D(k) formula •Implicitly DF between now and next period is βδ •Implicit DF between two future periods is δ •hence discount rate declines and = DF increases • Discount immediate futures heavily due to impatience Assumption: *IT consumption is independent* indicates that a person's w-b in period t+k is independent of her consumption in any other period. Past consumption violations Violated when looking at addictive behaviour: addictiveness depends on the stock of past consumption. Future consumption violations Violated when looking at preferences for sequences ??? Solutions: 1. Explained using idea of a ref point (z) stock of past C u(c;z)=u(c-z) 2. Another example is the habit formation model 3. Anticipation utility Assumption: *Utility function is stationary* there is no change in preferences over time cardinal u, u(ck) is constant across time w-b generated by any activity is the same in different periods Violation 1. Visceral influences such as hunger, sexual desire, physical pain, cravings etc. indicates that utility function is not stationary and make different decisions -masturbation and risk-taking - Projection bias (Focalism (walk dog) , failure to anticipate adaption (pay increase)) -reed and van leeuwan Edmund et al- stock prices increase from winning a football match 2. Temptation utility (preference for having something now Laibson (2006) Cue management: predicts that individuals have a preference for commitment and are willing to spend a substantial amount of money to manage cues and avoid temptation Solution • instantaneous u takes the form u(𝑑𝜏,𝑐𝜏) where 𝑑𝜏 represents the vector of visceral states in period 𝜏 • Projection bias utility function • Temptation utility

Why are unlikely events overweighted

Even though the likelihood of a costly event may be miniscule, we would rather agree to a smaller, sure loss — in the form of an insurance payment — than risk a large expense.

Illustrate the main difference between Expected Utility Theory and Prospect Theory and describe in brief what the latter is capturing that is omitted in standard Expected Utility Theory (18 Marks)

EUT 1. Four axioms 2. Definition 𝑉 = ∑ p(𝑥)E(u) 3. Example of calculating EUT with graph Properties 1. Independent of how it is described 2. Linear in probabilities 3. Defined over wealth Anomalies which are solved - Independent of how described (framing - Linear in prob (overweight, insurance) - Defined over wealth (loss aversion, endowment, risk-seeking in losses averse in gains) T & K produced ground-breaking work in BE in 1979 called prospect theory which aims to consider these violations. PT 1. Definition Prospect Theory provides a model for simple prospects with monetary outcomes and stated probabilities, it has two key aspects to the model a value and weighting function to consider violations listed above. 2. Two stage process in which prospects are edited including segregation or coding followed by evaluation where the prospect with highest value is chosen 3. Value (4) • Value function is the overall value of edited prospect, expressed in two scales. V= [f v(x) , pi(p) ] • V measures deviation from ref point. • Weighting function π associates with each probability p a decision weight π(p) = reflects the impact of p on the overall value of prospect • value function v assigns to each outcome (x) a number v(x) = reflects subjective value 4. General formula 𝑉 = ∑ 𝑣(𝑥)𝜋(𝑝) 5. Graph + Intuition (8) -risk aversion due to π(p) and diminishing marginal sensitivity (DMS) -DMS determines shape -DMS exists in both gains and losses -concave in gains, convex in losses -steeper for losses than for gains due to loss aversion -risk averse in gains, risk lovers in losses -Reflection effect -at critical values e.g. ability to pay rent, V is convex in gains Differences (5) 1. individuals assign a value to a prospect not expected utility 2. p is linear vs non-linear (7) -shows deviation from ref point -measures desirability not just likelihood of outcome - not behave at end points - non-liner to remove likelihood of bad outcome e.g. Russian roulette -overweight small probabilities to guard against losses -π(p)= p is expectation axiom -tener cuenta certainty effect 3. EUT defined over wealth PT over gains/losses - outcomes are defined relative to a ref point -ref point is zero point of scale - v measures the value of deviations from ref point 4. EUT preferences are independent of manner which prospects described. PT framing effects

How does mental accounting solve the anomalies of EUT?

EUT definition 𝑉 = ∑ p(𝑥)E(u) Graph Axioms 1. Completeness (define preferences) 2. Transitivity (consistent) 3. Independence (additional third) 4. Continuity (mix between good and bad) Properties of EUT 1. linear in probabilities 2. defined over wealth 3. Preferences are independent of how their described Definition: Mental accounting cognitive operations which organise, evaluate and keep track of financial activities Prospect theory considers the gains and losses in an experiment as separate from the other wealth and income the person has. The idea of mental accounting generalises on this idea and argues that to *make decisions manageable* we keep them apart and hold separate "mental accounts" for different choices. The decision to buy a coffee after class is not integrated with the decision of what apartment to buy even though the money comes from the same budget set. 1. Framing effects (violates transitivity, independent of how described) Types: (Levin et al., 1998) A. *risky choice* framing (e.g. the risk of losing 10 out of 100 lives vs the opportunity to save 90 out of 100 lives) B. *attribute framing* (e.g. beef that is 95% fat vs 5% lean) C. *goal framing* (e.g. motivating people by offering a $5 reward vs imposing a $5 penalty) Solution: Transaction utility 2. Loss aversion (violates utility over final wealth) - "losses loom larger than gains" the pain of losing is psychologically more powerful than the pleasure of gaining - people are willing to act take risks or act dishonestly Camerer paper b. Endowment effect (violates utility over wealth) Mug experiment paper c. Risk-seeking in losses, risk averse in gains (violates utility over wealth) Odean paper Solution: The authors highlight myopic loss aversion: when investors take a view of their investments that is strongly focused on the short term, leading them to react too negatively to recent losses, which may be at the expense of long-term benefits. Examples Derived from Samuelson's $200 vs -$100 with 50:50, would reject when done once but would accept if done multiple times. Suppose an investor must choose between a risky asset that pays an expected 8 %/ year with a sd of 20% (like stocks) vs safe asset which pays a sure 1 percent. The *attractiveness* of the risky asset will depend on the *time horizon* of the investor. The longer the investor intends to hold the asset, the more attractive the risky asset will appear, as long the investment is not evaluated frequently. 4. Overweight probable outcomes Insurance (violates linear in probabilities) Solution: assigning to different accounts- e.g chancel repair liability 5. Allais (violates independence) 6. Ellsberg (violates linear probabilities)

What is the law of small numbers?

Exaggerate how likely it is that a small sample resembles the parent population *Model* A. Person observes a *sequence of binary signals* of some underlying quality e.g. a sequence of good or bad investments by a mutual-fund manager B. Assume that each value of the signal is generated randomly from a stationary probability "rate." C. The person is a Bayesian and has *correct probabilistic priors* about this rate. D. But, whereas in reality these signals are generated by an i.i.d. process, the person believes that they are generated by random draws without replacement from an "urn" of N < infinity E. This captures belief in the LSN, F. person believes that the proportion of signals must balance out to the population rate before G. N signals are observed. H. *As N becomes infinitely large, the person becomes fully Bayesian; the smaller is N, the more he believes in the LSN* Implication (4) 1. neglecting the likelihood of small sample extremes 2. Belief in Fictitious Variation observing a small number of signals from a large population of different sources, overinference leads to exaggerate variation in rates among sources e.g. observe two performances from large number of analysts. LSN will always believe there are some good and bad ones, whilst Bayesian will realise over time they are all average 3. -> Gambler's Fallacy (opposite of overinference): an assumed relationship between obs, so that they should 'even out' over a small sample (i.e. second draw of a signal is negatively corr with first draw) 4. Main crux of the model is that there is a tendency to Overinfer from short sequences when there is uncertainty "the Hot Hand Effect" - after 2 signals an ind with LSN has higher on average beliefs - Exaggerating the likelihood: a short sequence resembles the underlying rate --> exaggerating likelihood the underlying rate resembles the short sequence Examples of Gamblers fallacy: - Subjects ask simulate random toss of a coin or draw (with replacement) from deck of cards Switching rate was around 58% in both experiments and declined Pr(AIB) 58.5% vs Pr(AIAAA) 29.8%. -Terrell [1994] studied a pick-three lottery run in NJ. Each day, bettors are allowed to buy tickets guessing the exact three-digit number the state would draw that day. The state divides 52% of money evenly among those bettors who choose the winning number. Hence amount distributed to winners is both indicator of both the number of bettors on that number and the cos of betting wrong. Tickets cost 50c each, so that if equal numbers of bettors choose all numbers, winnings on a given day should be $260. *If the winnings are sig higher than that, it indicates that too few bettors are betting that number.* 25% fewer lottery players in NJ bet on a number that won a week ago than if it won nine or more weeks ago - investors may under predict repetition in short strings of performances: i.e. assume mean reversion -> underreaction in S/R to announcements by firms -> induces underpricing and so continued stock price rises and there are short term positive correlation in returns (momentum) Examples of overinference: - Tversky and Kahneman [1971] Mathematicians were told that pattern of behaviour matching theory was statistically sig for sample of 20. They then also believed this would be the case for sample of 10. - If I that an average fund manager is successful once every two years, then he believes that a fund manager who is successful two years in a row must be unusually good. -Camerer [1987] Subjects observed three draws being drawn with replacement from one of two urns. Urn X which contained x2 black to red, and urn Y x2 red to black. After observing the three draws, the subjects participated in an asset market experiment, whose value depended on which urn was generating the draw of the balls. Hence betting on beliefs of likelihood of Urn X or Y. When three draws reflected proportions from either Urn X or Y there was over exaggeration that they were playing according to that urn. -investors overpredict repetition of longer strings / stock prices overreact to consistent patterns of good or bad news -> overreaction in M/R or L/R to announcements -> induces overpricing and so a reduction in the stock price and there is long-term negative correlation of returns (L/R reversal) Limitation Model would not calibrate well as the only degree of freedom is size of the urn in the model

What is the Ellsberg paradox and how should standard expected utility theory be modified to explain the paradox? (18 Marks)

Imagine an urn contains 90 balls from which 30 are known to be red and the remaining 60 are either black or yellow. One ball is to be drawn at random from the urn: (Gamble table) Common consequence effect -Subjects first presented A/B -Independence axiom implies same payoff in 3rd state world -Ignore 3rd state world -A/B are identical to C/D EUT prediction -But violated Ambiguity aversion -sharp probabilities: prob of winning / losing is v.clear -Ambiguous probabilities: chances of winning/losing are uncertain Pend tend to avoid ambiguous probabilities in the domain of gains, and seek them in losses How relates to prospect theory: U is measured via the value function Concave in gains and convex in losses

Explain mental accounting using the paper by Thaler (1985).

Goal: to develop a richer theory of consumer behaviour than standard eco theory Definition: set of *cognitive operations* used by individuals and HH to organise, evaluate, and keep track of financial activities Components: 1. (perception) captures how outcomes are perceived and experienced / decisions are made and evaluated Acquisition utility: measure of the value of the good obtained relative to its price- similar to consumer surplus Transaction utility: perceived value of the 'deal' defined as the difference between the amount paid and the 'reference price' for the good e.g. buy beer from a friend people are willing to pay more for the beer from the resort because the reference price in that context is higher at the resort For example: -Ask subjects 'Who is happier, someone who wins two lotteries that pay $50 and $25 respectively, or someone who wins a single lottery paying $75?' 64% say the two-time winner is happier. Framing, a sale as a 'rebate' rather than a temporary price reduction might facilitate the segregation of the gain in line with principle -beer $2.65 (resort) and $1.50 (hotel) 2. (categorisation) assignment of activities to specific accounts *Minimal account*- entails examining only the differences between two options and disregarding all common features *Topical account*- relates the consequences of outcomes to ref level (tend to do) *Comprehensive account*- incorporates all other factors inc current wealth, future endowments. The one we use most e.g. drive to save $5 on a calculator 3. (evaluation) concerns the frequency with which accounts are evaluated and 'choice bracketing' Accounts can be balanced daily, weekly, yearly, and so on, and can be defined narrowly or broadly, choice of how to bracket gambles influences the attractiveness of the individual bets / possibility of buying e.g. more likely to buy something from account in 'the green' Hedonic framing is a way of evaluating joint outcomes to max u (1) Segregate gains (because the gain function is concave). (2) Integrate losses (because the loss function is convex). (3) Integrate smaller losses with larger gains (to offset loss aversion). (4) Segregate small gains (silver linings) from larger losses (because the gain function is steepest at the origin, the utility of a small gain can exceed the utility of slightly reducing a large loss). *All components violate fungibility* Example: Fishing trip example Violates fungibility- i.e. the ability of a good or asset to be interchanged with other ind goods or assets of the same type Yet couple behaved the way they did because the £300 was put into both "windfall gain" and "food" accounts Applications 1. Incentive savings accounts Different MPC • current income (where MPC is high) • current assets (where MPC is intermediate) • future income (where MPC is low) C overly dependent on current income Savings help to smooth C Mental accounting studies suggest that unlabelled and easily available money will be spent more freely than money that is "accounted for,'' leading to very low saving rates among the un-banked. Can lead to negative interest saving schemes e.g. rent-to-own

What are heuristics

HEURISTICS *Definition* methods of arriving at satisfactory solutions with modest amounts of computation: can be thought of as mental 'rules of thumb' that people employ for all kinds of judgements. For example cut a cake amongst even split into 4 instead of according to ind preferences -> Simplify problems and overweight the immediate evidence in predicting the future TYPES *Representatives heuristics*: degree to which sample is similar to the characteristics of overall population e.g. Steve is shy so more people believe he is a librarian 1) //Prior probabilities are neglected// - e.g. With Steve there are more farmers than librarians in the country so should be factored into D-M - Experiment: told distribution of lawyers to engineers yet people disregarded this and made decisions according to the degree to which this description was representative of the two stereotypes 2) Neglect sample size - predict average height of a man (6ft) same in population as the sample 3) Misconceptions of chance People expect that a sequence of events generated by a random process will represent the essential characteristics of that process even when the sequence is short. - e.g. considering tosses of a coin for heads or tails, people judge the sequence H-T-H-T-T-H to be more likely than the sequence H-H-H-T-T-T 4) Neglect predictability 5) Illusion of validity people express great confidence Steve is a librarian, even if the description is unreliable or outdated. 6) Misconceptions of regression *Availability heuristics* exaggerated importance of info that can be easily recalled ● Biases due to the retrievability of instances -e.g. things like fame or salience will make things more available, and thus be considered more numerous -e.g. when listing celebs, in a group were men were relatively more famous subjects guess more males in a group ) ● Biases due to the effectiveness of a search set -e.g. judging whether more words start with 'r' or have 'r' as the third letter ● Biases of imaginability ● Illusory correlations - e.g. believing an individual with with shifty eyes his booky, because "suspiciousness" and "eyes" have a strong association, even if they might not predict anything true about the diagnosis *Anchoring heuristics* states that people estimate values by starting at one value (the anchor) and adjusting either up or down. Usually, people fail to fully adjust, and thus their estimates are biased towards the anchor ● Insufficient adjustments - e.g. in multiplying 8×...×1 vs. 1×...×8, people will report higher estimates for the first product and lower estimates for the second product ● Biases in the evaluation of conjunctive and disjunctive events - people tend to overestimate the prob conjunctive events - e.g. drawing a red marble seven times in succession (with replacement) from a bag containing 90 percent red (w/ 45% chance) preferred to draw red once (w/50% chance) ● Anchoring in the assessment of subjective probability distributions - e.g. starting at current (mean) estimate and adjusting upward to get to the 90th percentile

Give empirical examples of projection bias in the literature

Limitations (4) 1. Difficult to estimate as need current pref, predicted future pref and realised future pref 2. Individuals may not mispredict future preferences, but deem current preferences more valid 3. Future preferences may not be known 4. Observed diversification bias: - People seek more variety when they choose multiple items for future consumption simultaneously than when they make choices sequentially, i.e. on an 'in the moment' basis. - Sequential choices lead to greater experienced utility. -Halloween candy experiment. 1. Read and van Leeuwen (1998) asked office workers to choose a healthy snack or an unhealthy snack to be delivered a week later Group 1: asked when hungry Group 2: asked when satiated In the first group, 78 percent chose an unhealthy snack, compared to 42 percent in the second group. 2. Another example shopping on an empty stomach -> overbuying 3. Endowment effect -don't yet own an object, underestimate change in valuation once you own it -In the lab subjects overestimate loss aversion

How can self-control be used to explain addiction?

O'Donoghue and Rabin (1999) "Doing it now later" 1. Definition: A person with time-inconsistent preferences may or may not be aware that her preferences will change over time 2. Types of agents Time Consistent (TCs) (3) -time consistent preferences -have exponential p -any point in time choose today's behaviour by determining optimal lifetime plan given today's pref -pref are constant so stick to plan Naifs (6) -time inconsistent preferences -any point in time choose today's behaviour according to today's pref -incorrect belief that they will behave in the future according to current beliefs -mimic TCs -procrastinate immediate cost activities -preoperate (do too soon) immediate reward activities -not aware decreasing discount rates hence biased toward's present Sophisticates (S) -same time inconsistent pref as naifs -any point in time choose today's behaviour according to today's pref -they correctly predict how they will behave in the future -have a preference for commitments -mitigate procrastination -exacerbate preoperation -aware of decreasing discount rates so have pref for commitments 2. Objective of paper Explain self-control problems modelled as time consistent present-biased preferences 3. Definition of present biased preferences when considering trade offs between two future moments, present biased pref give stronger relative weight to the earlier moment as it gets closer 5. Findings -naifs effected solely by present bias effect -sophisticate does the activity sooner than naif with same preferences irrespective of immediate cost / reward as pessimistic about future 6. Immediate cost vs immediate reward IC Sophisticate mitigates tendency to procrastinate. Naif will incorrectly believe and procrastinate. IN FACT: If soph effect > present bias effect, do onerous activity sooner than if no self-control problem IR Sophisticate exacerbate the tendency to do it too soon Naif will overestimate benefits of waiting O'Donaghue and Rabin, 2001 Using a similar model the authors make the following findings wrt to self-control issues in addiction: Findings *When utility function is stationary* utility function ut(a; k) depends on his current addiction level k but not on the specific period t e.g. first hit of a cig yields same u for 20 year old vs 60 year old (unlikely) TCs hit iff they prefer hitting always to never hitting NAIFS person with present bias is more likely to be addicted and overconsume the addictive good c.f. TCs. More likely to procrastinate quitting. SOPHISTICATES Whether they consume more than N depends on pessimism effects vs incentive effects More likely to develop an addiction due to feelings of inevitability (incentive effects dominate) *When utility function is dynamic* it is more likely that the temptation to consume may vary over time in systematic or random ways i.e. temptation to hit is larger earlier in life NAIFS Naifs, in contrast, think in their youth that they can indulge in the large youthful temptation and later quit, but this unfortunately leads to a lifetime of hitting SOPHISTICATES Given an addictive product eventually loses its intrinsic appeal, S hit < N. S aware that indulging in 'the youthful temptation' -> lifetime of hitting, and so refrain to induce good behaviour in their maturity Empirical study Gruber and Köszegi (2001) Addiction is defined as repeated, and unwanted use despite clearly harmful consequences RQ: Whether D for cigarettes changes when taxes are announced Method: Using monthly data, exploit that state excise tax increases have been legislatively enacted but are not yet effective. Use rational model that incorporates time inconsistent preferences. Time inconsistent preferences: -Self control formalised as time inconsistent pref -TCs vs Naifs vs Sophisticates -> overconsumption in naifs -> for sophisticates depends on pessimism / incentive effects: of resist today more likely to resist future temptation Findings (2) 1. preferences are time inconsistent -ind do not realise difficulty of quitting -so look for self-control devices to help them quit. 2. smokers are forward looking announced but not yet effective tax -> increased sales and decreased consumption which is consistent with forward-looking behaviour. Policy implication optimal gov policy should depend not only on the externalities that smokers impose on others but also on the "internalities" imposed by smokers on themselves. Therefore, the optimal tax per pack of cigarettes should be at least one dollar higher than in the rational addiction case.

What policy implications do Bernheim and Rangel (2004) give to combat addiction?

POLICY IMPLICATION N.B: only looks at social welfare and ignores the negative externalities created N.B.B: works iff efforts to abstain have failed, and therefore depends on usage of ind Policy objectives 1. The authors emphasise the role of policy in the ability to avert mistakes caused by enviro cues. 2. An individual is uncertain about their future state of addiction based on exposure to envrio cues, this translates into monetary risk. As expenditure relies on addictive state. Government should magnify monetary risks which cannot be insured against. Policy tools 1. Laissez-faire is therefore the best policy for substance users who make *no serious attempt* to abstain Benchmark case 1. It is optimal to subsidise (rehabilitate) a when the likelihood of use *rises* with the level of past experience. 2. Provided the substance is sufficiently cheap, it is optimal to tax when the likelihood of use declines with the level of past experience. 3. Under weak conditions, a small subsidy for rehabilitation is beneficial, and a small tax is harmful. 4. When substance taxation is optimal, under some conditions criminalisation can perform even better 5. Programs that make addictive substances available on a prescription basis have potentially large benefits 6. Restrictions on advertising, marketing and public consumption also beneficial

Intertemporal choice example

Schedule of cinema: A mediocre movie this week (3 units of utility) good movie next week (5 units) a great movie in two weeks (8 units) a Johnny Depp movie in three weeks (13 units) Must do homework in 4 weeks, hence must skip a movie. Which week do you skip the movie and do the report? Your utility as of time t of a stream of instantaneous rewards 𝒖𝒕 𝒖𝒕+𝟏, ... . 𝒖𝑻 is given by 𝑼(𝑪𝒕 ... . 𝑪𝑻) = 𝒖𝒕 + 𝛃δ∑(s= t+1, T) 𝒖𝒔 Let δ=1. \Exponential discounting : β=1\ You work on your report on the first Saturday \Naifs: β=0.5\ Skip the Johnny Depp movie and do the report in week 4 _Week 1_ u(skip mediocre)= 0.5(5+8+13)=13 u(attend mediocre)= 3+ 0.5(8+13)=13.5 Same Week 2 _Week 3_ u(skip great)= 0.5(13)= 6.5 u(attend great/skip Johnny)= 8 Hence have to skip Johnny Depp to get report in on time Sophisticates: β=0.5 Self aware and know in week 3 if not done work skip Johnny. So in week 2 you consider whether to go to the good movie or not. In this case you skip the good movie and work in week 2. _Week 3_ u(skip great)= 0.5(13)=6.5 u(attend great/skip Johnny)= 8 Hence skip Johnny _Week 2_ u(skip good)=0.5(8+13)=10.5 u(attend good / skip Johnny)=5+0.5(8)=9 But now skip good movie _Week 1_ u(skip mediocre)=0.5(5+8+13)= 13 u(attend mediocre)= 3+ 0.5(8+13)=13.5 But do not skip mediocre

Explain how Neuroeconomics was important in understanding the concepts of risk aversion, ambiguity aversion, loss aversion and social preferences

RISK AVERSION •A lot of risk aversion is driven by fear. •Compare test between patients with prefrontal damage and those without -damaged patients often do not store the pain of remember losses -Often go back to the "high risk" deck, c.f. to normal patients. -Those with more emotional reaction also chose "low risk" deck •Also shown that the 2D:4D ratio, which is a marker for prenatal androgen effects, has predictive effects for the risk level of individuals -49 financial traders from London and using their P&L statements to measure success -find higher ratios have: higher net profits and predict amount of time they typically spend in the industry and the long-term profitability -this is because higher prenatal androgen exposure inc risk-performance, induces fast viscumotor motor skills and physical reflexes (Kosfield et al, 2005) Method 1. Trust game slightly adapted so trustee offer is a random mechanism, but investors face the same risk as in trust game 2. Placebo study as control 3. Inject investors with Oxycontin and compare Explain risk In the trust game, the risk of investor due to the uncertainty of the trustee's behaviour—that is, a social interaction with a specific trustee constitutes the risk. Hypothesis originally risk averse individuals, should be less risk averse with Oxytocin Results No difference between groups only 10% of the subjects with oxytocin choose the maximal transfer level in the risk experiment AMBIGUITY AVERSION • When individuals experience discomfort or fear, which often comes from gambles with unknown probabilities (such as Ellsbergs), this is transmitted to the OFC. When comparing patients with OFC damage to those without, OFC damaged individuals typically show more "rational" decisions • Activation of insula is different when people choose certain money amounts compared to when they choose ambiguous gambles (insula processes information like physical pain, hunger, pain of social exclusion). LOSS AVERSION losses loom greater than gains / the tendency for losses to have greater hedonic impact than comparable gains evidence that different areas of the brain are activated when losses or gains are experienced. SOCIAL PREFERENCES Breakthroughs in neuroscience reveal that *feelings and thoughts can be measured directly*. A natural question is whether we can find neural traces of the special reward value stemming from the mutual cooperation outcome. •Cooperation o Neuro-imaging studies report activation in the ventral striatum when subjects experience mutual cooperation with a human partner c.f. computer. o These studies suggest that mutual cooperation with a human partner is especially rewarding •Punishment Social preference theories also predict that subjects prefer punishing unfair behaviour such as defection in public good. In this view, it is natural to hypothesise that the act of punishing defection involves higher activation of reward circuitry. •Negative reciprocity: Sanfey et al., 2013 o By scanning brains of subjects in ultimatum games, unfair offers cause activation in the DLPFC (associated with planning, cognitive), insult cortex (associated with pain, disgust, negative emotions) and ACC (to resolve conflict. o Found that often the result of whether unfair or fair offers is accepted or rejected can be predicted by the relative strength of activation in the DLPFC and insula cortex. o This two-system of decision making can help inform economics in predicting and understanding decision making, particularly when mood-altering events take place. oUnfair offers by humans were also found to lead to more activation than by computers, which can explain the higher rejection rates •Trust (Kosfield et al, 2005) o In a placebo controlled hormone study to provide casual information about impact of hormones o they played a game between investor and trustee where investor has initial endowment of 12, which can send any amount 0,4,8,12 to trustee, total triples and trustee can send whatever amount back. o They found that of investors injected with oxytocin, 45% showed the maximum level of trust and only 21% gave an amount less than 8. Whereas those in placebo only showed maximum 21% of time and 45% showed trust levels below 8. Limitations Very small sample

What is the equity premium puzzle

The equity premium is the difference in the rate of return on equities (stocks) and a safe investment such as T bills. The puzzle is that this difference has historically been very large. In the USA the equity premium has been roughly 6% per year over the past 70 years. This means that a dollar invested in stocks on 1 January 1926 was worth more than $1800 on 1 January 1998, whereas a dollar invested in treasury bills was worth only about $15 (half of which was eaten up by inflation). Part of this difference can be attributed to risk aversion but not all. LOSS AVERSION (prospect theory) Losses loom greater than gains / greater hedonic impact to a loss than its comparable gain. T& K (1991) find the ratio of slopes of a gain or loss in money to be 2:1 -> MYOPIC LOSS AVERSION (mental accounting) Due to loss aversion more willing to take risks when they evaluate their performance less frequently (longer evaluation period) e.g. Samuelson dilemma: Implication Suppose an investor must choose between a risky asset with expected pay 7% per year with a sd of 20 % (like stocks) and a safe asset that pays a sure 1% The attractiveness of the risky asset will depend on the time horizon of the investor How long does the evaluation period have to be for investors to be indifferent between a risky and riskless asset? Found that for nominal returns the equilibrium evaluation period is about thirteen months, while for real returns it is between ten and eleven months.

What is the traditional method of measuring probabilities and what violations exist? Give applications where appropiate

Traditional approach: *Bayesian probability estimation* Probabilities are estimated correctly, given the relevant information and update correctly given a sequence of prior outcomes (e.g. always predict 0.5 when tossing a coin) Violations Many people tend to incorrectly assume that the prior outcomes do affect the probability of the next outcome hence estimate p<0.5 for heads again HEURISTICS LAW OF SMALL NUMBERS

Does experimental evidence regarding the one-shot ultimatum game and the dictator game contradict the predictions of money-maximising utility? In the ultimatum game, what is the role of subgame-perfect equilibrium refinement?

ULTIMATUM GAME 1. Game tree 2. Analytical prediction -NE is (10,0) using backward induction -If we assume agents are money max 3. Behavioural findings -rarely pick zero -results are very regular -mean offers 30-40% -offers below 20% are rejected about half the time 4. Reasoning for inconsistency -argue that inconsistency due to narrow nature of game -But + outcomes are also consistent with NE -e.g. If P2 plans to reject all offers < £4 and accepts all others, then P1 should offer£4, which is a NE -Hence there are many NE 5. Reasoning for inconsistency by BE (5) -assuming money maximising only -P2 threat to reject a positive offer is not credible -hence one SPE: P1 offers the smallest amount possible and P2 accepts -evident receiver has social pref if reject 10-z>0 -proposer less certain could be beliefs or social pref DICTATOR GAME 1. Game - similar ult game, but 2nd stage eliminated -Allows us to decipher pref of proposer 2. Analytical prediction -assume strict money max -NE: P1 offers lowest possible 3. Behavioural findings -proposer offers less than ult -but does share 10-30%

Illustrate the main methods used in Neuroeconomics and briefly discuss their main problems. (28 Marks)

There is no perfect way, yet, of measuring brain activity The general procedure for neuroeconomics is 1. Measure brain activation making decisions 2. Identify voxels which are particularly active 3. Cross check these with when these voxels are active under other circumstance (E.g. when happy, sad, angry) 4. Create a link between how that voxel is activated and the behaviour that was seen 1. *Brain imaging methods* few ways to measure activation Brain scans: involves taking pictures of an active brain EEG measures electrical potentials at the scull caused by neural activity Pros -EEGs can record very accurately the time of brain activity Limitations (5) -interior brain activity is not recorded (do not know where the activity was) -inaccurate in picking up activity in some parts of the brain -requires large number of repetitions -sensitive, simply moving your eye leads to electrical activity do hard to distinguish between causes from the experiment and normal body functions -costly fMRI / PET FMRI: most common method which uses the fact that haemoglobin have different magnetic properties depending on oxygen in blood PET: measures flow of blood to the brain fMRI Explained further... an inc in neural activity -> inc in use of O2 -> fall in level of O2 in activated area -> overcompensated by oxygen-rich blood which moves to the activated area (PET measures the blood flow in the brain) \Pros\ -can record much more accurately where the brain activity takes place \Limitations\ -less accurate on the time of activity -not very user friendly quite obtrusive -costly 2. *Pharmacological methods* - studies involve administering a drug that blocks (anatagonist) or stimulates (agonist) a particular kind of neurotransmitter receptor, to then compare behavioural performance on and off the drug - used to test the sufficiency and necessity of neurochemical systems for mental processes 3. *Study of animals* -used to analyse addiction -observe decisions taken -used in single neuron measuring: where tiny electrodes are sent to the brain each measuring a single neuron's firing \Pros\ (2) -brain structures are similar -has shed far more light on the basic emotional and motivational processes that humans share with other mammals than on higher-level processes such as language \Limitations\ (2) -insertion of the wires damages neurons, hence restricted to animals -Non-human may respond differently than human 4. *Markers* (4) -associations between ind D-M and features of neuroanatomy - Diff tendencies in D-M, e.g. tolerance for risk, linked to neurobiological measures, e.g. morphometry -neural features may be suitable as relatively stable biomarkers for different decision traits -and improve the prediction of eco behaviour across diff contexts \Limitations\ (2) -Markers are not always precise -limited field of application

Using social preference theory, what is the reasoning for violations of analytical NE in the Ultimatum Game and Dictator Game respectively?

ULTIMATUM GAME & DICTATOR GAME DESCRIBED Explanations 1. Inequity aversion (Fehr and Schmidt, 1999) Utility function + graph -u is assumed decreasing in the difference between you and another -α>β (care more if I have less) -assumes linearity in the difference -very flexible and bad predicting, as either optimising at equality or take everything 2. Pure Altruism 3. Competitive preferences 4. Rawlsian preferences 5. Utilitarian preferences: Predictions PROPOSER - Those with strong diff aversion/ Utilitarians/ Rawlsians prefer money to be evenly split -Those with competitive / positional pref / egoism, prefer receive all the money -Altruists (concerned with social welfare, but who give more weight to their own outcome) prefer positive split e.g. £7, £3 RECEIVER Cannot be altruistic or would accept anything hence inequity aversion explains both.

Violations of EUT

Violations (6) 1. Allais paradox: violates independence, common ratio and consequence 2. Ellsberg: favour certain outcomes due to ambiguity aversion 3. Initial perception / intuition affects D-M 4. Expectation axiom: Underweight outcomes that are probable compared to certain outcomes -> certainty effect / reflection effect 5. Probabilistic Insurance: reducing the prob from p to p/2 is less valuable than reducing from p/2 to 0 6. Isolation effect: discard components share by all prospects

How might we measure well-being? What problems are there with reported measures of subjective well-being, and what gives us some reassurance that the answers are not merely random? Assumptions (3) PE bias / duration neglect definitions validity

Utility can be measured as cardinal, having an *absolute value*, or ordinal, where utility is only relative and we can only tell *preferences between outcomes*. 1. When assuming SWB is a good measure of utility: - we assume utility itself is *measurable* and therefore cardinal - we assume that SWB *represents* our preferences and also choices - we assume people *report* honestly and accurately. - better predictor of utility when *discrepancies* between decision utility and experienced utility are small 2. Benjamin et al summary Although differences do occur, Benjamin et al use hypothetical questions about choices over two scenarios, asking respondents which they would choose and which they believe would give a higher SWB. They find SWB is still the *biggest predictor of choice*, conceding *83%* of the time, however there are systematic causes of discrepancy which include desire for status or trade off for family happiness. 2. Different types of well being measures: hedonic, evaluative, eudainomic vs global, instant u Different Qs -> diff conceptual answers 2. Measuring tools survey (GSS common) reports face-to-face momentary evaluations day reconstruction 3. Econometric methods When looking at SWB as the dependent variable, we can do so in numerous ways: -OLS assume continuous variable -logit (below or above 5) -ordered logit 4. Measurement issues (8) - Interpret Qs (particularly with translation issues) - Interpret scale -Scale bounded -Observed sensitivity (sensitive to context / mood e.g. find a $1 coin) -Ordering matters (talk about politics) -social desirability -*systematic errors: errors in recollection and aggregation* (Kahneman 1997) */Peak-end bias/: psychological heuristic where people judge an experience large based on how they felt at its peak (most intense) and its end* Duration neglect: duration of experiences has little or no independent effect on remembered u -reporting function 5. Results are valid - *common patterns* (u shaped in age oswald wu 2004, female happiness premium, religious participation, black people, children, unemp, widowed) - *corr between SWB and physical indicators* (responses often correlate with other measures such as heart rate, sweat or physical actions such as a laughing) - *corr with objective measures* -> Oswald and Wu 2010 compare life satisfaction data of 1.3m randomly sampled Americans (GSS), with state level imputed quality of life rank and the authors find strong corr -> Gabriel et al 2013 following compensating differential literature calculate state level quality of life using only objective data 6. Results are reliable correlation of life satisfaction results between 2 weeks is around 0.59 so moderate correlation 7. Limitations ( 1. Endogeneity (reverse causality issue with prod) 2. Very careful when talking about happiness function as not observed therefore not deterministic 3. SWB is measure of an assumed unobservable experience quality construct

What Do Laboratory Experiments Measuring Social Preferences Reveal about the Real World?

Workplace example *Premise* Firms achieve higher profits by treating workers 'kindly' (fair wage). Workers reciprocate positively to gifts by exerting more effort. Past \lab experiment\ results broadly concluded positive reciprocity but recent papers have contradicted this... (Kube et al., 2012) Do? 1. Controlled field experiment 2. The libraries of two economic chairs at a German uni had to catalogued 3. Job advertisement emphasised one-off job opportunity (remove reputational concerns) 4. repeated data entry tasks which are simple and made it easier to measure productivity 5. Examine the effect of different gifts (e.g. bottle, 20% inc wage, choice, origami) Results 1. 20% wage inc, inc prod by 5% stat insig 2. Bottle treatment group: stat sig 25% rise in prod, even though compensation was only 20% (robust if price of bottle is known) 3. When offered choice between 7 euros and bottle > 80% chose cash. Found a 25% increase in line with bottle treatment. 4. Origami also had a statistically sig productivity gain Implication 1. Previous literature explains gift exchange experiment but does not decipher the difference in monetary vs non-monetary gifts 2. Simply offering the bottle was enough to trigger reciprocity 3. Signalling goodwill from an altruistic employer 4. Also could be an Inequity Aversion story: inc outcome inequalities between workers and the employer, by reducing the employer's rent -> employee internalising the cost/effort to get the bottle 5. cash as a gift puts an exact monetary value on a relationship, re-framing a social exchange to a market relation 6. Doesn't matter if low (bottle) or high value (origami) Evaluation 1. worker-employee long term and gift effect -> habituation 2. (Bandiera et al, 2005) Do: compare prod under relative incentives, where ind effort imposes a neg ext on others, with piece rates where it does not. Externalities can be internalised due to altruism (care about colleague's payoffs) or collusion (fear of punishment) If externalities are all internalised it is due to altruism. Find: -prod is at least 50% higher under peace rates -as workers internalise the neg ext their effort imposes on others under relative incentives -effect more pronounced when with friends -internalisation is too small for complete social welfare story and too large to assume ext are ignored. -prod is sig lower in relative incentive schemes when able to monitor each other and internalise negative ext -> implicit collusive agreements which rules out pure altruism as only cause

Selfish behaviour is compatible with dishonesty and can be understood within the confines of standard economic analysis. However, field-study evidence suggests that economic agents often act honestly and neuroscience has also shed light on factors that might promote honesty. (a) Discuss field-study evidence that has looked into these issues. (14 marks) (b) What hypotheses have been forwarded in the field studies in (a) to explain honesty? (7 marks) (c) What does neuroscience suggest about the role of Oxytocin in respect of dishonesty? (7 marks)

a) Charness and Dufwenberg (2006) Vanberg (2008) b) Due to inequity aversion Fehr and Schmidt (1999) Due to social welfare Charness and Rabin (2002) c) (Kosfield et al, 2005) Definition: Oxytocin is a hormone which induces labour in mammals during lactation of the young and during mating Method (3) -Trust game 1. two anonymously interacting trustee and investor game 2. Both subjects receive an initial endowment of 12 monetary units 3. Investor can send 0, 4, 8 or 12 MU 4. experimenter triples each MU the investor transfers 5. the trustee has the option of sending any amount back to the investor 6. hence s.t. risk that the trustee will abuse trust -Placebo study as control to find causal effect -Inject investors with Oxycontin and compare behaviour with control Hypothesis: Oxytocin increases the trusting behaviour of investors (higher money payoffs) Results: 45% in the Oxycontin group showed the maximal trust level c.f control 21% The investors' average transfer is 17% higher in the Oxycontin group

What reasonings are there for an individual to exhibit overconfidence and which theory is most likely based on empirical evidence?

(Burk et al, 2013) Intro few people place themselves in bottom 40%, biased beliefs -> suboptimal decisions REASON 1: Uncertainty and Bayesian updating from common prior-> overconfidence REASON 2: Self image concerns Consumption value vs motivation value. Threshold model. REASON 3: Signalling easier positive external representation Findings -Not reason 1: if judgements are from Bayesian info, then testable restrictions placed on the joint distribution of beliefs and actual ability if ind place themselves in ability quantile k , the largest share of them must actually be from quantile k: REJECT THIS H0 ( inds from an ability quantile j < k are more likely to think in k) -Not reason 2: offer feedback. Reject imp prediction of self-image models: ind with high beliefs are uniformly more likely to demand info about their ability. Threshold model: ind low self-assessment seek info as long as chance of improvement. So overconfident should not ask. Reverse is found. -Above supports signalling theory: ind enjoy acquiring evidence confirming a positive belief, and enjoy sending public signals based on such evidence Robustness checks Issue: socially rooted bias if strict pref positive signal -relax assumption that reveal true beliefs -examine how personality differences affect judgements -social potency scale Definition: which captures the extent to which an ind likes to dominate and influence others, and derives pleasure from being in the limelight -social closeness scale -stress reaction scale -find that more socially dominant individuals have more confident judgements holding ability constant - This suggests that one of the mechanisms behind overconfidence is socially motivated biased interpretation

In addition to the possibility of failure to adequately recognise the impact of exogenous variation in context of preferences explain a paper that considers endogeneity of context and what that leads to.

(Dalton + Ghosal, 2012) ENDOGENOUS VARIATION any endogeneity of context suggests a further potential mechanism for systematic mistakes 1. Do: In the model frames (e.g. ref point) are pref parameters and endogenously related to action 2. Rational / sophisticated agents prediction: take this feedback of an action into account in their pref parameters 3. BE agent prediction: insufficient regard of this feedback (an overweighting of present pref parameters in projected future pref parameters) -> systematic biases 4. This idea can be extrapolated to social reference- when automatic then the ref group for an ind are dependent on the actions of the individual via exposure. If feedback considered then rational. Graph. Limitations -not directly observable or even inferred from choice behaviour

Discuss applications of Prospect Theory in real life phenomena, are there are any limitations / alternative explanations to the theory?

*Investors* (Odean, 1998) *Labour Supply* (Camerer et al., 1997) Life cycle models of labor supply predict a positive relationship between hours supplied and transitory changes in wages Why taxi drivers Wage: transitory in nature, correlated within days but uncorr between days, fluctuate (weather, subway breakdown, holidays etc.) Hours: flexible, trip sheets yield better measure Findings: 1. Little evidence of intertemporal substitution (predict workers to work more or supply more labour hours when demand is high) 2. 2/3 elasticity samples are negative 3. cab drivers work long hours during sunny days (low demand) and short hours during rainy days (high demand) 4. Due to daily income target based approach and risk aversion Why prospect - consistent with loss aversion - cab drivers set a loose daily income target (ref point) and quit working once they reach that target -learning effect / reference adjustment: inexperienced exhibit this behaviour more

What defines the approach to public policy that Thaler and Sunstein (2003) describe as Libertarian Paternalism? What false assumption and two misconceptions do Thaler and Sunstein claim anti-paternalistic stances are generally based on?

*Libertarian Definition* - In 'Nudge' (2008) authors define Libertarianism as inc agency. - T&S consider a policy as "paternalistic" if it is selected with the goal of influencing the choices affected by parties in a way that will make them better off. *Paternalism Definition* Paternalism as influencing choices to "make their lives longer, healthier, and better" Libertarian Paternalism seeks to adapt choice architecture where decisions are made or increase freedom of choice *Summary* The key idea then is that Nudges: 1. shape the choice environment in an attempt to improve outcomes 2. without altering the choice set or significantly changing economic incentives. 3. Examples of Nudge policies include things like moving healthy food next to cafeteria check-outs, instead of junk food or altering savings defaults discussed later *False Assumptions* 1. People always make choices in their best interest But people often do not exhibit: -RE, don't forecast according to Bayes rule, use heuristics, exhibit preference reversals and are sensitive to framing/wording effects; as well as dynamic inconsistency, present bias and self-control problems *Misconceptions* (2) 1. There are viable alternatives to LP In many situations, some organisation or agent must make a choice that will affect the choices of some other people. Example *food placement in a cafeteria* impacts decisions made. Hence the cafeteria manager has three options: place where it most benefits customers, place at random, or place where it least benefits customers. The first option is LP and the authors suggest the other options would not be advocated by society 2. Paternalism always involves coercion (e.g. Moving fruit before deserts in school cafeteria does not force people to not eat dessert) If there is no coercion, some types of paternalism should be acceptable *Process* When choosing between systems, the steps taken are: 1. Compare cost-benefit analysis Benefits: employees join who would otherwise not, some made better off, avoid not saving enough Costs: some made worse off (as highly liquidity-constrained), save too much for retirement 2. Choose approach that majority would choose if explicit choices were revealed or preferences 3. Choose approach that would force people to make their choices explicit. BUT Forcing choices in context of retirement plans: risk that choices that are actually elicited will be inadequately informed or not promote welfare 4. Choose approach that minimises the number of opt-outs *Implication* (3) -allows already rational/optimal agents to continue unhindered (consistent with Libertarian principles) -whilst Nudging those s.t. errors / biases to influence choices in a Paternalistic way that will make choosers better off, as judged by themselves. *Example* "Default option" A simple example is automatic enrolment of 401(k) employee saving plans, where individuals must actively opt out. This is paternalistic if employer believes that most employees would /prefer/ to join if they took the time to think about it and didn't loose the enrolment form. Then "forcing" is acting paternalistically as attempting to /steer choices/ in directions that promote welfare. But since no one is forced to do anything, this is libertarian paternalism. The employer /MUST/ choose some set of rules, and either plan affects employees' choices. In a /rational/ world there is no need as agents pick best decision for themselves. Having a default has a large effect as the chosen plan caries /legitimacy/ for many employees, as it seems to have resulted from some conscious thought about what makes most sense to people. To opt out requires time and effort of changing from the /status quo/ when default rules are "sticky" and therefore affect choices, this inertia may cause the large effects of higher enrolment. *Inevitable* Sometimes paternalism is unavoidable and when it is absent, it is usually because the starting point appears so natural and obvious that its preference-shaping effects are invisible to most observers. The inevitability is most clear when the planner has to choose default rules.

Briefly explain the differences in results in trustee and investor behaviour Kosfeld et al (2005)

*Mechanisms for oxytocin and trust* What mechanisms might be involved in generating the effect of oxytocin on trusting behaviour? Evidence found that oxytocin in animals increases prosocial inclinations in non-human mammales thus this could be a mechanism for trust 1) Could be due to a general increase in prosocial inclinations 2) implies should affect behaviour of investor and trustee 3) Trustees with oxytocin should give more back but not the case 4) Thus, oxytocin does not increase the general inclination to behave prosocially just trusting behaviour of investor In contrast, the trustees can condition their behaviour on the basis of the investors' actions Oxytocin make subjects more optimistic about likelihood of good outcome.

Illustrate the concept of Loss aversion, within the framework of the Prospect Theory.

/Definition/ "losses loom larger than gains" the pain of losing is psychologically more powerful than the pleasure of gaining /Realised/ Value and graph explained Utility is steeper for losses than for gains The ratio of the slopes of the value function in two domains, for gains and losses of money, is about 2: 1 (Tversky and Kahneman, 1991) /Implication/ People are more willing to take risks / behave dishonestly (Schindler & Pfattheicher, 2016) to avoid a loss than to make a gain. Status quo bias Papers: Endowment effect (mug, lottery ticket) Camerer et al., 1997 Odean Equity premium puzzle

What are limitations of social preference theory?

1. Can include everything and explain nothing 2.Analytical game theory is a benchmark 3. Ecological validity -degree of anonymity varies results -stakes matter (when larger in pre-industrial societies NE is more likely) -framing effects 4. Social preference or fairness social norms? 5. Environment matters -social pref vary by context and the process / intentions matter -the game people are playing depends on their utility functions 6. Process matters - Respondents less likely to reject computer generated offers but we should be indifferent 7. Valid elsewhere? -human sociality (selfishness axiom) is rejected in 15 small scale societies globally -market integration degree impacts benefits from cooperation

Explain the idea of a projection bias in the context of intertemporal choice and how it is violated in real life In reality Why important Example

1. DUM assumption 2. In reality agents are conceptually aware that preferences change but we underestimate size. Underappreciate degree to which a present C surge changes ref C level 3. Definition:overestimate the degree to which current pref reflect future preferences 4. Important as it allows us to make IT decisions that correctly predict how our tastes will change 5. Healthy snacks paper Read and van Leeuwen (1998) asked office workers to choose a healthy snack or an unhealthy snack to be delivered a week later (in the late afternoon). Workers were asked either when they were plausibly hungry (in the late afternoon) or when satiated (after lunch). In the first group, 78 percent chose an unhealthy snack, compared to 42 percent in the second group.

What are some of the most common methods used in Behavioural Economics?

1. Definition of BE 2. Objective 3. Process 4. Approaches (4) Experiments: -separate B explanations and standard ones -control + T = causal -variation e.g. Smith and market focus. Field data: -these studies involve observing real decisions that people make in their lives. -For example, auctions, betting on horses, investment decisions. -No concern of ecological validity (Dutch auction) -Confounds e.g. buy cheap stuff CC not discounting future cost savings at a high rate Neuroeconomics -definition -brain scanning and imaging techniques (PET/EEG) -brain activity, light up, stimuli -Ad: high level of resolution -DisAd: not relevant, emergent phenomenon, reverse inference

What is projection bias, and what are the possible implications of projection bias for consumption over the life-cycle, and for the consumption of durable goods?

1. Definition of PB 2. PB model Utility function - parameters -preferences are time inconsistent LIFE CYCLE 4. Rational predictions If C is habit-forming then optimum C path is upward sloping in general, but somewhat balanced. 5. Behavioural predictions A. PB -> an underappreciation of the impact of current consumption on future utility i.e. a negative internality via adaptation (ref graph) B. PB-> underappreciated habit formation -> repeated readjusting of immediate C/ ie. continually increasing -> delay C C. Both aspects leads you to consume too much earlier on in life and too little later D. If an ind inc C > previously thought -> fall in savings DURABLE GOODS Satisfaction derived from durable goods may vary day to day Classical prediction: consider expected daily valuations and ignore temporary valuations on the specific day they happen to buy the good Behavioural prediction: PB-> overvalue on high value days -> if multiple opp to buy then overbuy -> explains impulse purchases PB better explains overconsumption of durable good than hyperbolic discounting. Because overconsumption on cumulative small-scale decisions e.g. crisp , where the net effects of repeated purchases can be vast overconsumption of crisp less good at explaining long-term C (durable goods). Application -sales tactics ice cream front of the store when hot

Explain the concept of 'self awareness of changing time preference'. Based on the example in the paper by O'Donoghue and Rabin (1999), explain when an individual is Naïf and when is she sophisticated with respect to her intertemporal preferences? Definition present biased preferences Limitations (2)

1. Definition: A person with time-inconsistent preferences may or may not be aware that her preferences will change over time 2. Types of agents Time Consistent (TCs) (3) -time consistent preferences -have exponential p -any point in time choose today's behaviour by determining optimal lifetime plan given today's pref -pref are constant so stick to plan Naifs (6) -time inconsistent preferences -any point in time choose today's behaviour according to today's pref -incorrect belief that they will behave in the future according to current beliefs -mimic TCs -procrastinate immediate cost activities -preoperate (do too soon) immediate reward activities -not aware decreasing discount rates hence biased toward's present Sophisticates (S) -same time inconsistent pref as naifs -any point in time choose today's behaviour according to today's pref -they correctly predict how they will behave in the future -have a preference for commitments -mitigate procrastination -exacerbate preoperation -aware of decreasing discount rates so have pref for commitments 2. Objective of paper Explain self-control problems modelled as time consistent present-biased preferences 3. Definition of present biased preferences when considering trade offs between two future moments, present biased pref give stronger relative weight to the earlier moment as it gets closer 4. Go through example in seminar 5. Findings -naifs effected solely by present bias effect -sophisticate does the activity sooner than naif with same preferences irrespective of immediate cost / reward as pessimistic about future 6. Immediate cost vs immediate reward IC Sophisticate mitigates tendency to procrastinate. Naif will incorrectly believe and procrastinate. IN FACT: If soph effect > present bias effect, do onerous activity sooner than if no self-control problem IR Sophisticate exacerbate the tendency to do it too soon Naif will overestimate benefits of waiting 7. Which one is better? Depends on IC or IR IC: Welfare loss from present bias in Naifs IR: Welfare loss from present bias in S 8. Limitations (2) -In many scenarios being aware of self control problems worsens the issue -Not all people are sophisticates as assumed, and results can lead to odd outcomes if we assume Sophistication always 9. Welfare implications ADDICTION S may not consume an addictive product at all as a means of self control, as they know they will lose control if they try to consume in moderation

What is the Easterlin paradox, and (to the extent that it holds), what might explain the phenomenon?

1. Easterlin paradox Definition Graph Explanation why it exists (5) -BEYOND CERTAIN THRESHOLD, relative y/c not absolute -adaption over time -social comparison means change in aspirations/expectations which constrains W-B BENEFIT -value based on social norms -shift in reporting function 2. Clark et al paper Graph Explanation (gradient steeper for a country point in time than over time, status benefit, ind Y and H yield steeper relationships) Formula + intutiton (5) 3. Limitations - certain threshold income -Stevenson and Wolfer: no empirical evidence of satiation point (critical value in which income no longer impacts w-b) Only use eval measures - Oswald does find positive relationship with another dataset -Vary dependent on SWB 4. Evaluation (6) -Results are asymmetric - Satiation varies with SWB - Evidence suggests constrains to SWB of affluence -Modern growth does no lead to increased reported SWB but improved SWB inequality -Types of growth impact SWB -Objective measures have improves

Give two examples of a field experiment which tests the effect of a commitment device. Intro (3) Summary of lit (2) Why commitment device? Finding SEED (1) Objective second paper

1. Intro A person with *time-inconsistent preferences* may or may not be aware that her preferences will change over time. Hence, policy might be aimed in providing: *Commitments* (impose a retirement scheme) *Educating* (making ind more sophisticated) Efforts to indirectly prove time inconsistency have focused on the use of external "commitment devices" that *limit future choice sets*, because the use of such devices provides *smoking guns that prove time consistency wrong* 2. TCs vs Naifs vs Sophisticates S like commitment devices Can be useful for inds with self control problems Ashraf, Karlan and Yin (2006) 3. Objective test effect commitment savings device 4. Data Existing / former clients of a bank in the Philippines (N=1777) 5. Method -randomised control methodology -baseline survey -allocated T and C groups (SEED which restricts access to deposits but did not compensate for this restriction, marketing where inds were encouraged to save, control) -follow up survey 6. Findings (4) -women appear time inconsistent in baseline survey - 28.4% take up SEED -more hyperbolic inds take up SEED especially females - average savings balances increased by 81% for SEED (without crowding out effects) -the longer-term impact (2.5 yrs) fell to 33% increase and no longer sig -Marketing T had no effect Royer, Stehr and Syndor (2015) 7. Objective To examine if *self-funded commitment contracts* can improve the L/R effects on an incentive program 8. Data 1000 employees at a Fortune 500 company 9. Method -randomised control methodology -commitment contract: pledge to continue using gym over 2 months after initial incentive period -Allocate T and C groups (control, incentive $10 per visit up to 3 visits/week, incentive + commitment also put money in bet continue after 2 months) 10. Findings (5) -attendance rates doubled in incentive period -all groups reduced frequency after -attendance prob of incentive+commit remains elevated -prolonged effect even years after the program -incorporating temporary incentives with unincentivised periods that leverage habit formation through techniques, such as commitment contracts, will likely be more cost effective than consistent incentives.

Describe the concept of a projection bias and formalise a new model to explain it

1. Projection bias definition overestimating the degree to which current preferences reflect future preferences 2. Traditional theory Utility function (5) -formula -maximise u with given info -preferences are time consistent -affected only by own payoffs -independent of framing Assumptions -correct about p(st) -Bayesian updaters 3. New Model Utility function (2) -parameters -preferences are inconsistent -conceptual idea but overestimate 4. *Explained further* People experience an IMPACT BIAS overestimate the intensity / duration of their emotional reactions to an event otherwise known as affective forecasting One cause is focalism the other is failure to anticipate adaption Focalism. Focalism - underestimate the extent other events will influence our thoughts and feelings Examples 1) Football fans only think about the joy of winning a match and neglect other factors that will influence happiness e.g. deadline at work 2) Belief that buying a dog will bring nothing but joy but neglect influence of happiness e.g. going for walks in the rain 3) Having a child (evidence has shown that children have insignificant effect on H only grandchildren). Adaption: failure to anticipate how quickly they will make sense of things that happen to them in a way that speeds emotional recovery 1) Spinal chord injury patients adapting to a new way of life 2) Winning the lottery Implication We often end up making decisions that satisfy preferences that exist in the present but not in the future. This is at the heart of our failure to meet long-term goals and our engagement in harmful behaviours such as smoking, over-eating, and over-consumption. Limitations (4) 1. Difficult to estimate as need current pref, predicted future pref and realised future pref 2. Individuals may not mispredict future preferences, but deem current preferences more valid 3. Future preferences may not be known 4. Observed diversification bias: - People seek more variety when they choose multiple items for future consumption simultaneously than when they make choices sequentially, i.e. on an 'in the moment' basis. - Sequential choices lead to greater experienced utility. -Halloween candy experiment.

What policy recommendations has the literature looked at to combat addiction?

1. Rational model (nothing) B + R (2004) agree with this conclusion if the ind makes no attempt to abstain 2. Educating / getting individuals to be more sophisticated 2. Due to smoking paper: optimal gov policy should depend not only on the externalities but "internalities". Tax rate $1 higher than normal rational model. B+R (2004) also agree with this justification iff the likelihood of use declines with the level of past experience. B+M (1990) agree also, wrt cigarettes. Doubling the federal excise tax -> in L/R to a 4-6% fall in C 4. Due to Karlan and Zinman paper (commitment device) - Test effectiveness of commitment device for smokers -RCT: T is to sign up to CARE program where bet to pass a urine test, group received cue cards reminding them of bad smoking effects -3-5% more likely to pass urine test -Persistent -Limited incentive and leveraging habit formation leads to lasting effects 5. Due to Bernheim and Rangel, 2004 Subsidise & tax A. It is optimal to subsidise (rehabilitate) a when the likelihood of use *rises* with the level of past experience. B. Provided the substance is sufficiently cheap, it is optimal to tax when the likelihood of use declines with the level of past experience. C. Under weak conditions, a small subsidy for rehabilitation is beneficial, and a small tax is harmful. D. When substance taxation is optimal, under some conditions criminalisation can perform even better E. Programs that make addictive substances available on a prescription basis have potentially large benefits F. Restrictions on advertising, marketing and public consumption also beneficial

What role can behavioural economics play in improving public policy?

1. Thaler's - Definitions -Assumptions -Misconceptions -Implication is that we can nudge those in the right direction without hindering already rational agents -Default option example 2. Chetty's Model - Formula - How to measure experienced utility non-paternalistically - allows us to use BE as concerns that perceptions of exp u override decision u - good for model uncertainty - treats 'biased' individuals only 3. Layard (2006) We are less happy!! - Social comparisons: tax to internalise external cost of working too much - Adaption: tax to internalise negative internality - Tastes: limit advertising, performance-related pay and hollistic appraoch

Describe the trust game analysed by Charness and Dufwenberg (2006). What interpretation do the authors offer to explain why behaviour in experimental settings often differs from the prediction offered in classical game theory? Trust is important Do Behavioural explanation sentence Definition guilt aversion Communication boots (5) Evaluation (3)

1. Trust is important to (3) -EXCHANGE RELATIONSHIPS (or -> misselling e.g. PPI + asymmetric info) -INVESTMENT IN RELATIONSHIPS (incentives to invest) -TRUST (any transaction without complete contract, Hugh-Jones contract substitute) 2. Do Examine the impact of communication in a one-shot principal-agent game with hidden actions 3. Game 4. Analytical GT findings -Trustee no obligation to return, if money max choose 'Don't' -SPE, backward induction, (Out,Don't) -Main treatment is, (B makes promises to A) communication, with a credible threat 5. Behavioural GT findings (4) -show experimental evidence, act honestly, expectations, avoid guilt aversion -new equilibrium (In, Roll) reached -Definition guilt aversion: influences second order beliefs -> the more the agent believes his principal expects to be helped, the more likely the agent is to actually help. -Communication prior boots outcome: A. words affect agent's beliefs B. this changes guilt experienced C. A promise may feed a self-fulfilling series of beliefs D. so makes the promise credible E. so this is a commitment device. 6. Evaluation (3) A. Non-promising messages are not effective B. ecological validity: extends to lawyer and client / husband and wife B. guilt-aversion is a microfoundation of social norms -Social norm is a moral expectations that people are inclined to live up to -Guilt-aversion provides the motivation to behave according to the social norm

Limitations of Kahneman and Tversky (2)

1. Validity issues as considered riskless scenarios only assumed same behaviour for risky scenarios 2. Lab experiment drawbacks Good -separate B and standard explanations -control + T = causal Bad -assumption: insights gained in the lab can be extrapolated to the world beyond -Hawthorne effect

Describe how games, in reality, deter from predictions made in analytical game theory and what explanations does behavioural economics give?

Aim behavioural game theory is to show what actually happens aiming to inform and expand analytical theory. Iterated Games 1. Analytical predictions -Optimal behaviour, backward induction, dominant strategy equilibrium, eliminating dominated strategies 2. Violations p-contest -infinite number of possible iterations -iterated dominant NE is 0 -Research found average is 35, robust -Players preferred only one or two steps of iteration 3.Reasoning (4) -People are unable to iterate beyond a couple steps 'limited computability' -Belief that other people are incapable of doing so -Failure of backwards induction in K model theory: Level 0: hierarchy begins with some very naive type. Non-strategic player will choose actions without regard to the actions of other players. Level 1: sophisticated player believes that the other players will act non-strategically; his or her action will be the best response consistent with those first-order beliefs. This goes on throughout the levels -Cognitive hierarchical theory Does not fit into any type, but assumes some % of the population fits each type, and BR accordingly. 2. Minimum efficiency Coordination 3. Travellers' Dilemma

In what way does the observed behaviour in the Travellers' Dilemma and Minimum effort coordination games seen in the lecture differ from the predictions of analytical game theory?

Aim behavioural game theory is to show what actually happens aiming to inform and expand analytical theory. Travellers Dilemma 1. Explained -Fails to acknowledge sensitivity due to cost parameter -2 players simul choose number 180-300 -Lowest number is rewarded for not overestimating item -Winner receives min + R /Loser receives min -R -BR is to undercut 2. Analytical prediction Analytical GT says that iterated play will lead to the dominant strategy being picked as dominated strategies are eliminated and as long as R> 0. NE is to play the LB. 3. Behavioural findings -fails to consider "irrational" agents - Actions vary with R (cost being too optimistic) -R is low, less incentive to act rationally -R is high close to NE prediction Minimum efficient coordination game 1. Explanation -Fails to acknowledge likelihood of multiple equilibria -Players siuml choose effort level 110-170 with associated cost -Lowest effort level is chosen -Min - cost x ind effort level 2. Analytical predictions -Not one pure strategy but multiple equilibria, where cost levels of each ind are the same. -Ind are indifferent about cost size Overcut: lowest is still chosen 120-120c vs 120-121c hence worse payoff Undercut: less payoff as c<1 120-120c - 120-119c = 1-c >0 3. Behavioural findings -fails to consider "irrational" agents -actions vary with c -Low cost -> high effort as productivity of task is high -High cost -> low effort as greater risk of picking high effort

Consider an additive comparisons model (such as the basic case in Clark and Oswald, 1998) where utility is increasing in the difference between consumption of good x and a reference level of consumption of good x (i.e. a comparison term). In what way does the optimal response to a change in the reference consumption depend on the curvature of the comparison term?

Clark and Oswald, 1998 1. Do: Create a model of rational individuals who care about the actions of others/ their relative position in society ie STATUS MATTERS 2. Obj: is to find reasons that would cause an ind to imitate beyond the mere desire to fit in 3. Findings (3) - relative position is not sufficient reason for imitation - action to deviate or conform depends on curvature of the comparison term in utility function - comparison-concave u -> follow 4. Formulas / Algebra FOC and implicit differentiation -> U= sv(a-a*) + (1-s)u(a) - c(a) FOCs: s'v(a-a*) + (1-s)u'(a) - c'(a) da/da*= s''v(a-a*) ----------------- sv(a-a*) + (1-s)u(a) - c(a) -denom is negative due to concavity requirement of max problem -if numerator is <0 the comparison-concave so follow 5. Graph if comparison-concave individual -increase in reference a0 to a1 -relative position worsens -MU of status improves -BR is to imitate and increase a also 6. Leaders vs followers graph -Leaders are comparison-convex -Followers are comparison-concave -A good becomes more attractive, so + shock to MU -Model absent of comparisons -> increase a until MB=MC -But with status effects higher a: reduces MU status for leaders and inc MU for followers. -Both derive utility from status however varying responses depends on curvature... 7. Evaluation The magnitude of changing a depends on s

What are violations of EUT and what models can fix them?

EUT definition 𝑉 = ∑ p(𝑥)E(u) Graph Axioms 1. Completeness (define preferences) 2. Transitivity (consistent) 3. Independence (additional third) 4. Continuity (mix between good and bad) Properties of EUT 1. linear in probabilities 2. defined over wealth 3. Preferences are independent of how their described 1. Framing effects (violates transitivity, independent of how described) Types: (Levin et al., 1998) A. *risky choice* framing (e.g. the risk of losing 10 out of 100 lives vs the opportunity to save 90 out of 100 lives) B. *attribute framing* (e.g. beef that is 95% fat vs 5% lean) C. *goal framing* (e.g. motivating people by offering a $5 reward vs imposing a $5 penalty) Solution: reference point in prospect theory, transaction utility in mental accounting 2. Loss aversion (violates utility over final wealth) - "losses loom larger than gains" the pain of losing is psychologically more powerful than the pleasure of gaining - people are willing to take risks or act dishonestly (Schindler et al., 2016) Camerer paper 3. Endowment effect (violates utility over wealth) Mug experiment paper 4. Risk-seeking in losses, risk averse in gains (violates utility over wealth) Odean paper Solution: steeper in losses, DMS, mental accounting 5. Overweight probable outcomes Insurance (violates linear in probabilities) Solution (non-linear probability / weighting function) 6. Allais (violates independence) 7. Ellsberg (violates linear probabilities)

Discuss whether there is evidence of forward looking behaviour when modelling addiction.

Gruber and Köszegi (2001) Addiction is defined as repeated, and unwanted use despite clearly harmful consequences RQ: Whether D for cigarettes changes when taxes are announced Method: Using monthly data, exploit that state excise tax increases have been legislatively enacted but are not yet effective. Use rational model that incorporates time inconsistent preferences. Time inconsistent preferences: -Self control formalised as time inconsistent pref -TCs vs Naifs vs Sophisticates -> overconsumption in naifs -> for sophisticates depends on pessimism / incentive effects: of resist today more likely to resist future temptation Findings (2) 1. preferences are time inconsistent -ind do not realise difficulty of quitting -so look for self-control devices to help them quit. 2. smokers are forward looking announced but not yet effective tax -> increased sales and decreased consumption which is consistent with forward-looking behaviour. Policy implication optimal gov policy should depend not only on the externalities that smokers impose on others but also on the "internalities" imposed by smokers on themselves. Therefore, the optimal tax per pack of cigarettes should be at least one dollar higher than in the rational addiction case.

What alternative models of Instantaneous Utility exist (3) zt formula + intuition (2) ref point formula

Habit formation 1. Definition: utility from current C can be affected by the level of past C 2. Formula -all effects of past consumption for current utility enter through a state variable (𝑧𝜏) -𝑧𝜏 is the *exponentially weighted sum of past consumption* Reference point model 1. Definition: According to prospect theory, outcomes are evaluated using V defined over gains /losses 2. Formula 𝑢(𝑐𝜏, 𝑟𝜏)= v(𝑐𝜏 −𝑟𝜏 ) rt the ref point could be status quo, past C etc.. Utility from anticipation 1. Definition: instantaneous u= utility from C now plus function of the discounted utility of future C 2. Formula 3. Implication - discount diff goods at diff rates -creates downward bias on discount rate -bias is larger the larger the anticipatory u -put off pleasurable anticipation -reflects emotions e.g. anxiety uncertain future -Empirical study: Sunday ranked below Friday Visceral states 𝑢(𝑐𝜏,𝑑𝜏) vector of visceral states in period t

Briefly describe the simple representative-agent model that Chetty (2015) provides to characterise the planner's public policy objective. What additional constraints are typically imposed by neoclassical economics?

Pragmatic, policy-orientated approach: start from policy question (e.g. how to increase savings rates) and incorporate behavioural factors only to improve empirical predictions and policy decisions (opposes Atkinson) Formalised via a representative agent model. Planner's objective: choose set of tax rates (t) and nudges (n) which max experienced u of the agent s.t. revenue requirement and standard incentive-compatibility condition Formula • parameters • ancillary condition is nudge • c max decision u • want to max experienced u overall • Revenue constraint: revenue = taxes * consumption • C must maximise decision utility • max experienced utility of society subject to constraint that decision utility means c is actual chosen *Back to neoclassical* To nest neoclassical framework within this three additional restrictions are imposed: 1. planner has no nudges thereby relaxing this gives more nudging tools 2. decision u is a smooth, increasing, and concave function of C choices -i.e. no discontinuity, reference or diminishing MUc - relaxing this gives better predictions by incorporating nonstandard preferences and ancillary conditions - also should be application specific i.e. wouldn't separate u when studying C patterns around retirement 3. experienced u = decision u Relaxing all three gives more welfare implications (e.g. Neglect adaptation, positional externalities, if agents have reference-dependent preferences To model differences between experienced and decision utilities, write planner's objective as v(c) + e(c) where e(c) = u(c) - v(c) aka the internality, measured by identifying impact of agent's choice on his own experienced utility (as opposed effect on other's for externality

Explain why the value function in the Prospect theory is S-shaped.

Utility in Prospect Theory is expressed in terms of a Value function 1. Value (4) • Value function is the overall value of edited prospect, expressed in two scales. V= [f v(x) , pi(p) ] • V measures deviation from ref point. • Weighting function π associates with each probability p a decision weight π(p) = reflects the impact of p on the overall value of prospect • value function v assigns to each outcome (x) a number v(x) = reflects subjective value 2. General formula 𝑉 = ∑ 𝑣(𝑥)𝜋(𝑝) 3. Graph + Intuition (8) -risk aversion due to π(p) and diminishing marginal sensitivity (DMS) - DMS implies that the effect of the change diminishes with distance to the reference point. -DMS determines shape -DMS exists in both gains and losses -concave in gains, convex in losses -steeper for losses than for gains due to loss aversion -risk averse in gains, risk lovers in losses -Reflection effect -at critical values e.g. ability to pay rent, V is convex in gains For example: A difference between a gain of 100 to 200 appears larger than a difference between 1100 and 1200. The difference between a loss of -100 and -200 appears larger than difference between -1100 and -1200.

What are the three main ways in which behavioural economics might be relevant for public policy, as suggested by Chetty (2015)? Provide examples for each.

Welfare Consequences From Polices Differ From Predictions Of Neoclassical model!!!! *Measuring experienced utility* Suggests 3 non-paternalistic ways to measure experienced u 1. Take SWB as a measure of experienced utility 2. Choice-theoretic approach: when we believe choices credibly reveal preferences (e.g. opportunities with learning, choices not being sensitive to ancillary conditions) 3. Building a structural model of difference between decision and experienced utility So what does this mean for policy? A. This allows us to use behavioural models for policy decisions B. previously there were concerns of paternalism giving policymakers' *perceptions* of an ind's experienced u precedence over own choices C. Sometimes we don't know if there are behavioural biases there or not! - in presence of model uncertainty optimal policy can be to use nudges (e.g. changes in defaults/framing) - as it changes behaviour and inc welfare if you do suffer from biases, without distorting behaviour if agents optimise. -This is a form of libertarian paternalism D. Supports Thaler's claims that nudges are justifiable policy tools as they expand but do not restrict current choice sets.

Discuss the arguments in the literature for and against choice/ decision utility and what alternatives are there? Use at least one field experiment to explain your answer

Welfare defined by choice distinct from welfare defined by quality of outcome e.g. faulty toaster 1. Choice based utility explained as: -classical, normative, axioms rationality, Bentham -reveal pref: pref shown through choices -reason for choice irrelevant -suitable approximation to true welfare -ordinal utility 2. (Bernheim and Rangel, 2007; 2009) possible to extend decision utility to welfare of analysis of inconsistent choices by allowing for revealed pref only when people are insensitive to context i.e pref purification (ancillary condition) 2. BE objective 3. Anomalies and Process 4. SWB assumptions and optimal (measurable, reported, represents, little discrepancy) 7. General SWB validity (Benjamin et al., 2012) RQ: Do you always choose the option that makes you happier (choice vs outcome) Methodology: -13 hypothetical questions e.g. $80k with 7.5hrs sleep vs $150k with 6 hours -two alternatives *-explore tradeoffs between factors considered imp in SWB -analyse if rankings coincide* Results (5) -do not always max SWB (83% choice=SWB) -SWB most sig factor but others -clear distinction between two -SWB represents idealised revealed preference utility: what they would choose if SWB were unbiased Limitation 1) Issues of self control problems and reversals if asked a new Q (28%) 8. Carter and McBride although slight variation SWB (experienced) and decision u have fundamental similarities e.g. when pooled regression both have S shaped utility function over gains and losses 9. There are various factors that can be considered but moment based is best (Kahneman, 1999) • For example, within experienced u have predicted utility (based on beliefs) and remembered utility (based on past experiences) e.g. family move state • considerable evidence people wrongly forecast utility • depends on what measure you use can lead to different conclusions • *Self-selection* and *dissonance reduction* would predict high SWB among voluntarily move / *Treadmill effects* predict that people who moved will eventually return to their characteristic level of SWB • memory-based measures do not tell us if movers really are happier • need momentary reporting KEY LIMITATIONS 1. Endogeneity (productivity) 2. (Kahneman, 1997) Hand in warm water experiment 69% chose longer trial of 60 seconds 14 Celsius water (less hot than short trial) Peak end Bias Duration neglect 3. Context Money is salient more likely to pick that outcome than higher SWB 4. Other factors rational agents trade-off other things (family)