Economic Psychology Review Class Notes PDF

Two fields and their intersection Economics: social science dealing studying production, distribution and consumption of services and goods (Krugman & Robin, 2012). Theory driven – top down – deductive. Psychology: individual behavior and their underlying processes. Lacks overarchin...

Two fields and their intersection Economics: social science dealing studying production, distribution and consumption of services and goods (Krugman & Robin, 2012). Theory driven – top down – deductive. Psychology: individual behavior and their underlying processes. Lacks overarching theory – bottom up - inductive 3 Summary: Economics and Psychology Economics Psychology Based on few fundamental assumptions: Predominantly an inductive approach, empirical Revealed choice – stable preferences theories at low levels Goal is maximizing utility Explanations of individual behavior All economic laws are derived from these assumptions Intensive effort to describe details Mathematical language and models Experimental and statistical methods, surveys Observational data, self-reports and experiments Objective data, controlled, neutral experiments often with context Assumptions on individual behavior to assist predicting Assumptions about individual behavior must be phenomena (“as if” assumptions) realistic (description and prediction) Some simple and robust psychological/behavioral Contextual, structural, and systemic variables are less concepts are narrowly defined and incorporated into relevant models 4 BE and EP complement conventional economics rather than substituting E.g., world is paralyzed taking major steps in combating climate change  Conventional economics: free rider/externalities problem and financial incentivizes to deal with them. E.g. carbon tax  But! Much is left unexplained: one example here in the exam 1) why there is too little protest, outrage (vs. terrorism?) o BE/EP adaptation People have adapted to gradual changes in their environment over time and thus find it harder to react strongly to the slow and incremental nature of climate change, unlike the immediate and shocking impact of terrorism. o Wishful thinking Many individuals believe that technology will magically solve climate change without requiring significant personal or societal changes, thus reducing their sense of urgency or need to protest. Some people downplay or deny the scientific consensus on climate change because accepting it would o Self-servingly questioning scientific evidence mean acknowledging the need for inconvenient lifestyle changes or economic sacrifices. - it is always a little bit worse o Intangibility - The effects of climate change, such as rising sea levels or increased average temperatures, are less tangible and immediate compared to the vivid and direct impact of terrorist attacks, making it harder to elicit strong emotional reactions or protests. 2) why greenhouse gas negotiations are so difficult o Self-serving beliefs about fairness: Fair burdens Countries may perceive fairness differently - losing something hurts more than gaining the same amount o Loss aversion - Countries are more sensitive to the potential economic losses from reducing emissions (e.g., higher costs, reduced industrial output) than to the potential gains from preventing climate change, making them reluctant to commit to significant emissions cuts 3) Remedies through cheap and easily implementable behavioral interventions (nudge) --> Default Options: Automatically enrolling people in green energy programs while allowing them the option to opt-out. This simple change can significantly increase the adoption of renewable energy sources without requiring drastic changes in behavior or significant financial incentives. 5 Economic assumptions about preferences Good 2 complete and transitive explaining Good 1 Preferences are stable Preferences are coherent (e.g., complete, transitive):  Complete – you can always rank alternatives or bundles of goods – you know whether you prefer apple over pears or vice versa or you are indifferent Coherent preferences (complete and transitive) allow individuals to make consistent and rational decisions. If preferences were not  Transitive – A>B and B>C then A>C complete, you might be unable to choose between options. If preferences were not transitive, your choices could be inconsistent and lead to paradoxical or irrational behavior. We know our own preferences e.x. consuming now than later how much risk we tolerate we care about the other person ; I am helping because it makes me feel better or I really want to be a good person and help Preferences take specific forms – e.g., time preferences, risk preferences, social preferences, (self-control problems) (reference dependance) --> such as altruism and fairness influence social preferences  We are (mostly) risk averse When sure thing worth less than the EV of a risky choice, we prefer less but sure thing. Bank account with minimal gain p while π(p) + π(1-p) ≤ 1 Low ps are overweighted, moderate and high ps are underweighted. what are the regularities? - graph zeichnen und erklären können - what is reference dependence - what do we mean to meximize value - value function and its properties and pi ist wichtig!! Tversky & Kahneman, 1981, Science --> hier unterschied zwischen pi and p erklären können 30 Value function and its properties 1. Impossible events are discarded π(0) = 0 2. Scale normalized π(1) = 1 3. At endpoints not well behaved 4. Very low p  π(p) > p while π(p) + π(1-p) ≤ 1 Low ps are overweighted, moderate and high ps are underweighted. 5. For any fixed probability ratio, the ratio of decision weights is closer to 1 when the probabilities are low than when they are high. E.g., π(0.1)/ π(0.2) > π(0.4)/ π(0.8) Tversky & Kahneman, 1981, Science 31 Endowment effect (Kahneman, Knetsch & Thaler, 1991, JEP) We overvalue something in our property regardless of its market value Or, WTA > WTP after we own the good=buying price of a good is lower than selling price. o Shifted reference points once we own the good Effect disappears when we make a decision for others 33 Mental accounting (MA) Accounting: "the system of recording and summarizing business and financial transactions in books, and analyzing, verifying, and reporting the results." (Merriam-Webster Dict) Mental accounting: a description of how individuals do these things (mentally): Sources / purposes of the money, etc. Purposes of MA: – keep track of where there money is (and is going) – self-control device – diminish 'pain of paying' (not discussed by Thaler) 34 Mental accounting violates … – consumption decisions are independent of financing decisions (in both directions) – consumption enjoyment is independent of financing decisions – purchases should always be financed to minimize Net Present Value (NPV) – money is fungible Any unit of money is substitutable for another. So, consumption decisions are based on the consumer’s total wealth. 35 Pros and cons Pro: 1. A big bonus  saving for something big vs higher monthly salary  upwards adjusting consumption 2. One can use it as a self-control device 3. Feel rich and poor simultaneously Cons: 1. Rigidity of tapping across accounts: We do not want to repay a car loan from the money put aside for our kids college education even though we could save paying a bunch of interest. 36 Consequences of different accounts 1. When sharing multiple good news (gains):  tell them separately, wrap Xmas gifts in different boxes 2. When sharing multiple bad news (losses):  Tell them all at once 37 Recall two things 1. Traditional assumptions about preferences 2. Different approach between economics and decision theory 39 2. Different approach between economics and decision theory Micro-focus of standard econ and decision research often different ECON DECISION RESEARCH good 1 good 2 good 3 attribute attribute attribute - problem: asymmetric dominance bundle 1 - different attributes better for different goods, e.x. one house has better lights, the 1 2 3 other house has a better distance ect bundle 2 good 1 etc. good 2 etc. good attribute 2 2 good attribute 1 1 40 1. Economic assumptions about preferences Good 2 Good 1 Preferences are stable Preferences are coherent (e.g., complete, transitive) We know our own preferences Preferences take specific forms – e.g., time preferences, risk preferences, social preferences, Preference uncertainty Traditionally: (recall!) no uncertainty. We know and consistently reveal preferences by choice to maximize value and preferences are stable. Preferences are independent from context in which they are presented (i.e., other alternatives). So, if I prefer Apple over banana when only these two are available, I will still prefer A over B when orange is also available. But! This is often violated. Sometimes, a market share of a brand increases (vs decreases) after another brand is introduced. Marketers want to foresee how introducing a brand/product impacts existing ones and what is going to be the market share. 42 Preference uncertainty Softer approach: Preferences are uncertain nevertheless they do exist and just need to be discovered = Discovered Preference Hypothesis (DPH) (Plott, 1996) – we reveal them and they become stable over time. Harder approach: Preferences are uncertain, and there is nothing there until one is faced with a choice. Then, we construct them, Constructed Preferences (Slovic, 1995) 43 Constructed Preferences Given that people are uncertain about their own preferences, when called upon to make a decision, they use any available information to help resolve the uncertainty. This opens the door for many deviations from standard predictions about choices and the influence of the context in which options (mostly multiattribute) are presented. 44 Constructed Preferences Given that people are uncertain about their own preferences, when called upon to make a decision, they use any available information to help resolve the uncertainty. This opens the door for many deviations from standard predictions about choices and the influence of the context in which options (mostly multiattribute) are presented. This violates/contradicts regularity ( adding an alternative should not influence preference for other alternatives) 45 Asymmetric dominance – important concept to know A is better than B on one attribute (light) and B is better than A on one attribute (location) Example: Tough choice: Does the advantage in the location compensates for the darkness? light A B location 46 In marketing, one can exploit context dependence Suppose, you want to increase share of a product. Context dependence suggests that expanding the set can lead to preference shifts toward original set members. Two ways: 1. The added option (the decoy) which is inferior to the target (of which you want to increase sales) while competitor asymmetrically dominates the target – decoy effect 2. The added option makes the target seem as a compromise – compromise effect 47 Trick1: Attraction /Asymmetric dominance/Decoy effect (Huber et al, 1982 and 1983) You want to increase sales of A (target). B is your competitor. Product A B Price 400 $ 300 $ Storage 30 GIB 20 GIB 48 Trick1: Attraction /Asymmetric dominance/Decoy effect (Huber et al, 1982 and 1983) You want to increase sales of A (target). Product A B C B is your competitor. Price 400 $ 300 $ 450$ You introduce C – decoy: Storage 30 GIB 20 GIB 25 GIB - Inferior to A (more expensive and less storage) - Asym Dom B (more exp but more storage) You observe sales increase A when C is also offered. Violates independence of irrelevant alternatives! 49 Trick2: Compromise effect The closer the inferior alternative to the relative superior one, the more stronger is the compromise argument (Simonson, 1989, JCR). extras Super fancy car medium Low budget Inferior option price 50 Why these tricks work? People seem to care about justifying their choices to themselves and to others – They seek good reasons for explaining their actions o Idea from social psychology – motives are various:  Enhancing self-esteem  Avoiding regret  Decrease cognitive dissonance  To perceive themselves are rational beings with good reasons So, giving people support/ways to justify their choice makes them feel happier 51 Evaluation mode reversals: Joint (JE) vs separate evaluation (SE) (Hsee, Loewenstein, Blount and Bazerman, 1999, Psych Bull) JE: Options are presented concurrently SE: Options are presented sequentially Experimental example from Hsee et al. Two candidates for a programming job. All subjects are asked how much salary they would pay each candidate, but JE subjects evaluated both candidates, whereas in SE each subject evaluated only one candidate at a time. Experience GPA joint separate Candidate J 70 KY programs in last 2 years 2.5 33.2K 26.8K Candidate S 10 KY programs in last 2 years 3.9 31.2K 32.7K Note: GPA-scale is anchored at 1-5 (5 is the best). 52 Two candidates for a programming job. All subjects are asked how much salary they would pay each candidate, but JE subjects evaluated both candidates, whereas in SE each subject evaluated only one candidate at a time. Experience GPA joint separate Candidate J 70 KY programs in last 2 years 2.5 33.2K 26.8K > < - case for preference reversal Candidate S 10 KY programs in last 2 years 3.9 31.2K 32.7K Note: GPA-scale is anchored at 1-5 (5 is the best). 53 What accounts for the preference reversal btw JE and SE presentation? The evaluability hypothesis candidate Att1: Number of Att2: Number of Att3: GPA JE SE KY years J 70 2 2.5 33.2 K 26.8 K S 10 2 3.9 31.2 K 32.7 K In SE mode: How many KY programs he has written is hard to judge – no background info about how many KY programs is reasonable to expect from someone. So, GPA which is meaningful for everyone guides behavior. In JE mode: KY attribute gains meaning and this information guides choice. 54 In sum… Preferences are reversed between JE to SE evaluation In JE – options are evaluated comparatively, in respect to each other. Attributes gain meaning in respect to each other. In SE – options are evaluated separately. Hard-to-understand attributes are ignored/gain less attention 55 Some applications: Coherent arbitrariness (Ariely, Loewenstein and Prelec, 2003, QJE) Implications of JE/SE research: Absolute judgments are often difficult to make Relative judgments simpler and more natural Problematic to examine absolute attitudes with relative methods Normatively irrelevant factors influence valuation Sometimes, normatively relevant factors do not affect valuations Findings challenge the idea of fundamental value in economic theory 56 Coherent arbitrariness Absolute valuations are arbitrary and sensitive to normative and nonnormative influences. But, once valued this serves as an anchor  imprinting. Then, relative valuations are coherent to the initial arbitrary valuation  locally coherent. Make sure you can explain this through an example. 57 Recall that there is a difference between risk and uncertainty are involved Risk = we know the probability distribution of an outcome – e.g., lottery numbers, roulette wheels – Observable  Probabilistic risk Uncertainty = we do not know the probability distribution of an outcome – horse races, or most things in life - Not Observable RISK: You have 10Euros and place a bet on rolling 2 on a dice. If UNCERTAINTY: You can bet $10 Euros on a specific horse without knowing the probability you get 2 you make 10 time of your bet (1000 Euros). You have 1/6 to get 1000Euros – EV = 1/6*1000Euros distribution of the outcome (final ranks among horses in the race). Or, you are standing at u-bahn station at night and have no idea/schedule/possible obtained information on when the next u-bahn is coming. Or, you kick out your tenant from the appt you own and rent out and have no idea when he finds a new place. 59  Ambiguity Aversion (AA) Tendency to prefer risky over uncertain events. Preferring or betting on events where the probability distribution is known, over when it is unknown. Ambiguity also effects probability judgments: 1. Subadditivity: probability judgments add to less than 1. 2. Most people prefer non-ambiguous gambles over ambiguous ones expect at very low probabilities 3. Ambiguity decreases willingness to act 60 Two urn problem – Ellsberg paradox Two urns: Urn1 – 100 balls in total: red and black, but red/black = ? [0,1] Urn2 – 100 balls in total: 50 red and 50 black, red/black = 1 You are asked to bet. If you say, e.g., you bet on Red1, you mean you would choose from Urn1. If red is drawn, you get 100 Euros, if black you get 0 Euros. Somebody wants to measure your subjective probabilities to win the bet. S/he follows Savage (SEU) axioms. 61 Four questions asked 1. Which one do you prefer to bet on, Red1 or Black1, or are you indifferent? ( Drawing a ball from Urn1 on which even would you place a bet? Red, black, indifferent?) 2. Which one do you prefer to bet on, Red2 or Black2, or are you indifferent? ( Drawing a ball from Urn2 on which even would you place a bet? Red, black, indifferent?) 3. Which one do you prefer to bet on, Red1 or Red2, or indifferent? 4. Which one do you prefer to bet on, Black 1 or Black2, or indifferent? 62 Responses and what they mean 1. Which one do you prefer to bet on, Red1 or Black1, or are you indifferent? ( Drawing a ball from Urn1 on which even would you place a bet? Red, black, indifferent?) 2. Which one do you prefer to bet on, Red2 or Black2, or are you indifferent? ( Drawing a ball from Urn2 on which even would you place a bet? Red, black, indifferent?) 3. Which one do you prefer to bet on, Red1 or Red2, or indifferent? 4. Which one do you prefer to bet on, Black 1 or Black2, or indifferent? (P)Red1 = P(Black1) = 50%, P(Red2) = P(Black2) = 50%, P(Red2) > P(Red1), P(Black2) > P(Black1) – HOW IS THIS POSSIBLE? - People are inconsistent. We are unable to infer the probabilities they assign to event from their choices ( Savage Sure Thing Principle) 63 What is intertemporal choice = choice over time? Any choice made between consuming at two or more different points in time, i.e., across temporal prospects Or, when the benefits and the costs of a choice are separated by time – smoking, borrowing money, etc. Many (if not all) of our daily life choices are intertemporal choices: o Grocery shopping today for the next week  predict food preferences o Lifestyle choices: smoking, food consumption, exercise o Borrowing money with interest o Studying hard now to get a good job later o Committing to a relationship o Etc…. Standard approach: Discounted Utility Model (DU) (Samuelson, 1938) Forget about the uncertainty of the future and focus on evaluating future consumption! Time-invariant preferences captured in Discounted Utility (DU) model ∞ Xt – utility of consuming X in time t U(x) = ∑δ tU(x t ) t =0 δ – discount factor, where δ=1/(1+r) and r is the discount rate larger r yields to smaller δ  Decision maker whose “r” is.8 will care less (i.e., values future consumption less) about the future than whose “r” is.2. DU is very simple (thus appealing), since all human factors (patience, impatience, feeling states, emotions, memories, etc.) are captured in constant “r” 66 Assumptions of DU 1. Final consumption what matters – we integrate new options into existing plans. 2. Utilities across periods are independent 3. Temporal and atemporal preferences are independent 4. Independence of atemporal preferences from outcomes in other periods 5. Constant discounting 6. Stationary atemporal preferences 7. Positive time preference 8. Diminishing marginal utility of consumption over time Key terms Time-variant preferences: Preferences depend on the time of the consumption. When we discount consumption, we steeply discount those that need to be postponed from now to future BUT we do not discount the same consumption so much when it needs to be delayed from a future time point to a more remote time point. Present biased behavior: Consequence of time-variant preferences. We prefer consuming now than in the future. Dynamic inconsistency in the behavior: When we plan a consumption ahead of time (e.g., starting diet) we discount a future consumption much less than how much we discount it when we have to consume that thing on immediately. E.g., : Diet: You think that starting diet next Monday is no problem and eating carrots all Monday will not be as bad as when we actually arrive to next Monday. When, however, you find yourself in next Monday you value eating French fries more than the carrots. You discount diet on the spot so much that the value of French fries is higher. In other words … (1) We prefer “less sooner” over “more later” (e.g., one apple today over two apples tomorrow (Thaler, 1981) (2) We prefer immediate gratification over delayed gratification (e.g., cake today VS cake tomorrow) (3) We procrastinate on tasks that impose an immediate cost and delayed reward (e.g., starting going on a diet, saving money, etc.) Illustrating how time-variant preference leads dynamic inconsistent behavior Smaller reward: S Larger reward: B Preference reversal t0 (Ainslie, 1975, Psychol Bull) When both rewards are far off B (in t2) is preferred over S. When you are at t0 and think about t2. However, as t1 becomes closer in time its relative value increases and at t* it dominates B. Illustrating how time-variant preference leads dynamic inconsistent behavior Smaller reward: S Larger reward: B IN OTHER WORDS, WE HAVE SELF-CONTROL PROBLEMS CAPTURED IN DYNAMIC INCONSISTENCY Preference reversal t0 (Ainslie, 1975, Psychol Bull) When both rewards are far off B (in t2) is preferred over S. When you are at t0 and think about t2. However, as t1 becomes closer in time its relative value increases and at t* it dominates B. What this means? Time preferences are not invariant but they vary by how remote the consumption is: o Imminent consumption is heavily discounted o Remote consumption is not so heavily discounted People do not stick to their plans because when you have to give up something good on the spot hurts more than as you anticipate it will hurt to give it up in the future 72 Predicting future taste/preferences We often need to make choices now which are realized (i.e., outcome utility) at a later time point o Whom we get married o Where to live o How many kids to have o What to study o Where to spend a vacation o How much to save for retirement But! How could we know our future preferences?  We try to predict them. But, we can only make predictions based on what we know and feel. 73 The answer is: Not always = Projection bias When predicting future tastes/choices (i.e., preferences), we anchor our predictions to our current states.  STATE DEPENDENT PREFERENCES Order food hungry – we may order a lot and do not take into account satiation. Ordering a very warm sweater when very cold Signing up for another kid when it is very easy with the current one(s) WHY IS THIS? We underappreciate how much our tastes will change and project our current tastes/preferences in the future. This leads to a systematic under/over-appreciation of current states and a hot-cold empathy gap In a hot state we cannot imagine how we will behave in a cold state and vice versa. Implications on many behaviors involving hot states: Sexual behavior – trying to get people to use condoms in the heat of the moment on a class education… Food/Drug craving –a satiated mindset/physiology vastly differs from a non-satiated one… Mental health – interventions – when? When depressed? When not depressed? And many more 74 Intuitively Generalization to all forms of state-dependent preferences Future tastes often differ from current tastes. Recall why! Due to (1) psychological and physiological reasons, (2) habit formations, (3) addictions, (4) maturation. Taste change can be described by changing “states” Ideally, when relevant tastes for decisions that affect the future are the tastes that prevail when consequences are experienced Projection bias = People under-appreciate effects of changes in future states. They project their current preferences onto the future. More formally 𝑈𝑈𝑡𝑡 (𝑐𝑐𝑡𝑡 , 𝑠𝑠𝑡𝑡 ) Utility at time t, 𝑈𝑈𝑡𝑡 , depends on 𝑐𝑐𝑡𝑡 - what you consume and 𝑠𝑠𝑡𝑡 what state you are in. State can be anything that influences your consumption utility: hunger, cold, sex drive, drug craving, past consumption, etc. You need to predict at time t (now) your future utility from consumption 𝑐𝑐𝜏𝜏 when you are in 𝑠𝑠𝜏𝜏 , where 𝜏𝜏 > 𝑡𝑡. 𝑢𝑢 (𝑐𝑐𝜏𝜏 , 𝑠𝑠𝜏𝜏 ) = αu(𝑐𝑐𝜏𝜏 , 𝑠𝑠𝑡𝑡 ) + (1- α)u(𝑐𝑐𝜏𝜏 , 𝑠𝑠𝜏𝜏 ) α is the degree of your projection bias, 0 ≤ α ≤ 1. Measures how much how much you mispredict your future taste. The overestimation is coming from: 1. Resemblances = you overestimate how much your future taste will resemble to your current taste. 2. Impact = you under/overestimate how much that state will influence your utility. This is called impact bias. Choice architecture to nudge people towards optimal behaviors “…any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives. To count as a mere nudge, the intervention must be easy and cheap” (Sunstein & Thaler, 2008) Easy to implement Cheap to implement Overt and not covert No restriction on choice – preserve the freedom of choice No forcing into any options - preserve the freedom of choice No changing of the incentive structure Remember: Nudge is only one of multiple tools to incorporate behavioral insights into public policy! 79 3. Organ donation (Johnson & Goldstein, 2003, Science) US - appr. 85% approves organ donation and yet 28% granted access to his/her organs. Sweden – 85% approves organ donation and 85% grants access to his/her organs 80 Traditional approach and behavioral economics have divergent predictions on how defaults would influence choice 1. Traditional approach = people’ choice depicts people preference for organ donation – they find little value in donating their organs This implies, whatever default people are enrolled into, should not matter. They would opt-in/opt-out according to their preferences. 2. Behaviorally informed economics = as per constructed preferences, choice will be influenced by how it is presented. Because: (1) this is the status-quo (norm), (2) effortless to accept (inertia), (3) this is the suggestion 81 10 SC 2 open end question 82 (Johnson & Goldstein, 2003, Science)