Behavioral Finance Final

Main Heuristic Driven Biases (Availability, Representativeness, Overconfidence)

1. Availability bias: overweighting information that is more readily available or more conceivable, which alters perceptions of event likelihood 2. Representativeness: judgement based on stereotypes and how close an event or sample is to anther event in mind, or the processes which it was generated by (law of small numbers). Ex: baby boys born in a hospital, should know that sampling variable decreases in proportion to sample size. Could lead to OVERREACTING to new information 3. Overconfidence: setting narrow confidence bands. Found out that men tend to be more overconfident, which led to them to trade more and lose more (esp. single men vs. single women)

(Kahneman Tversky 1973) Retrieval of occurrences and constructing scenarios

All empirical studies have an objective procedure for enumerating instances (e.g., words that begin with K or paths in a diagram), and hence each of the problems had an objectively correct answer, may not be so in real life. However, availability heuristic may be used to evaluate probability to events through constructing possible scenarios, or thinking of past occurrences (memory searching). Constructing scenarios when you didn't experience them before. More scenarios you can imagine makes you think that it is more likely. It is exceedingly difficult for the human mind to apprehend sequences of variations of several interacting factors. We suggest that in evaluating the probability of complex events people tend to simplify them or only consider simple cases. Chess: you see the other person's actions as constant, yours as changing. These considerations suggest that a player is susceptible to the fallacy of initiative-a tendency to attribute less initiative and less imagination to the opponent than to himself. Becoming more preoccupied with certain outcomes because you have experienced them in the past. Consequently, availability provides a mechanism by which occurrences of extreme utility (or disutility) may appear more likely than they actually are.

(Mental Accounting and Consumer Choice, Thaler) Conclusion:

Developing concepts in coding gains and losses through "who is happier" questionnaires and looking at uncertain choices, evaluating purchases through transaction utility which refers on reference prices , and budgetary rules, need more data on more HHs. Mix of econ and psych, most applicable in marketing

(Mental Accounting and Consumer Choice, Thaler) Budgeting Implications from gift giving

People tend to give gifts that someone would not normally buy for themselves, goes against standard theory which predicts that they would buy something the other person already buys Shadow prices are not equal, categories seen as luxuries with a high budget Some people would also prefer to receive a tangible gift rather than cash, counter to normal theory. Some people paid with products as well as money Can also influence people to increase their mental budget by introductions reasons to increase them such as coupling other pros.

(Money Illusion, Shafir et.al) Fairness and morale

People tend to look at profits and judge if they are fair on nominal terms. They also judge fairness in wage cuts nominally, extended into views regarding if workers will quit or not

(Money Illusion, Shafir et.al) Transactions real and nominal, buying a house example

People tend to want to sell a house in times of rising prices over buying a house in times of rising prices even though there may be an implied real loss —> people are influenced by nominal changes People said they are more likely to sell for a higher nominal price Theory used in marketing, telling consumers prices are likely to rise

Frame Dependence: Money Illusion, Currency effects

People think of nominal values of money rather than real value of money —> people are not the same as rational agents as in econ theory. Idea under quantity theory of money: prices are stick and don't adjust right away Think about different currencies real value changing dramatically under high inflation times. Adapting to different metrics may make price differences seem larger in some currencies than others even though they are same % change. Inflation plays a key role in value of $.

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Imputations and prediction mechanism of hedonics (how to get net utility)

Purchase of a durable good: (1) Start from utility as if the car is a gift (2) subtract debt financing during each payment. (3) imputed costs of consumption: consumption seems to cost less as time goes on (4) imputed benefits of payments: increase as loan payments become easier as debt balance reduces —> consumption and payments as experienced, utility increases at the end, payments become easier at the end too

(Mental Accounting and Consumer Choice, Thaler) purchase decisions through process

Purchasing decision thought process: (1) response to local temporal budget constraints such as monthly budget (2) expenditures grouped into categories and purchases considered in categories — violates fungibility (3) consumers evaluate each purchase as the situation arises w(z,p,p*)/p > k (budget) (4) usually high budget for addictive or seductive goods, low budget for things good for you in the long run (exercise, education)

(Kahneman Tversky 1973) on "true" probabilities

The "true" probabilities of such events are elusive, since they cannot be assessed objectively. The subjective probabilities that are assigned to unique events by knowledgeable and consistent people have been accepted as all that can be said about the likelihood of such events. Although the "true" probability of a unique event is unknowable, the reliance on heuristics such as availability or representativeness, biases subjective probabilities in knowable ways. A psychological analysis of the heuristics that a person uses in judging the probability of an event may tell us whether his judgment is likely to be too high or too low. We believe that such analyses could be used to reduce the prevalence of errors in human judgment under uncertainty.

Risks that limit arbitrage

(1) Fundamental risk: market may do very well even if a stock is overvalued (2) Unpredictability of future resale prices

Extensions of BPT:

(1) design of securities by corporations in which capital structure and dividend poly approached from the demand side (2) analyze risk and relationship to time-diversification (3) equilibrium asset pricing model can be derived

(Prospect Theory and Asset Prices, Barberis et.al) consumption based approach and its problems

(1) it does not come close to capturing the stock market's high historical average return and volatility (2) doesn't predict striking variation in expected stock returns over time (3) low correlation of returns and stock growth

(Can MLA explain equity premium puzzle? Larson et.al 2016) Conclusion

(i) we find that professional traders exhibit MLA in their natural domains (ii) Can rule out that the lack of investment flexibility is a reason for the MLA. Implications: (1) expected utility theory may not be descriptive of professional traders' strategies, behavioral finance as alternative explaination (2) can explain asset price volatility, payout puzzles, irrational discounting (3) revealing information on a less frequent basis likely means that investors will be better able to meet their savings goals for retirement (4) individuals, investors, and traders all want timely information and flexibility in making decisions, but in doing so harm themselves and impact the workings of the financial sector

(Can MLA explain equity premium puzzle? Larson et.al 2016) Difference in experimental design from GP97, Haigh&LIst, Thaler et.al

**Traders can place investments in the risky asset at any time, infrequent can decide investment as often as the frequent. ** We do, however, use the same size of the time period to compare the groups. We choose the minute for reporting, but there is no reason we could not choose another interval (our results are virtually the same when we do). Risk allocation determined by how much of portfolio is risky asset

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) double entry mental accounting theory

-b/c payments and thoughts of benefits can interfered w/ each other, create two accounts: net utility from consumption and net disutility of payments NET UTILITY FROM CONSUMPTION Imputed cost of consumption: how much is the pleasure costing a person, can detract from consumption pleasure. Differs in how much influence this has . Experience utility: utility of consumption when the good is free - imputed cost * payment/utility conversion parameter (MU of money, depends on each person's $ situation) Lambda*p is a decision criterion lambda*p_hat is the psychological burden of paying NET DISUTILITY FROM PAYMENT Disutility of purchasing minus imputed benefit derived from each payment (lambda*Pc) Assuming loss aversion

Behavioral Finance 3 Themes

1. Heuristic Driven Biases: processes of which people make estimates and find thing out cause some systematic errors. 2. Frame dependence: Decisions are affected by how information is presented, including language 3. Inefficient markets: as a result of 1 and 2 markets fail to be efficient

Arbitrage and short bracketing

Assumption that they have short time periods to consider because they need to borrow cash/securities to trade and have to pay per period fees Culminating borrowing fees and can add up to large amounts —> bias towards short term Prices may not ever converge to fundamental values due to inelastic demand. (If arbitrageur is risk adverse, his demand for underpriced stocks are limited —> demand curve is no longer perfectly elastic )

How to prove loss aversion (ex-ante vs ex post)

Authors appealing to MLA assume that payoffs are calculated ex-post, net out of endowment b/c if w included in monetary payoffs, ex-ante editing means loss aversion plays no role Bankruptcy games and experiments where subjects are supposed to lose money— experimenter should show that if losses are covered by endowments, reference point is moved

Larson et.al 2016 graph:

Average investment by treatment shows significant difference by the last 2days of experiment— LF traders invest more

(Behavioral Portfolio Theory, Shefrin & Statman) two versions of BPT

BPT-SA single mental account : integrate portfolios into a single mental account and consider covariance like mean-variance investors, but not within the portfolio viewed as a integrated single account BPT-MA diff mental accounts: segregate portfolio into mental accounts and over look covariance among those mental accounts

(Shefrin & Statman) Behavioral Portfolio Theory compared to Markowitz's mean variance portfolio theory and CAPM investors

Behavioral Portfolio Theory is a positive portfolio theory on foundations of SP/A theory and prospect theory (choice under uncertainty) addressing Friedman/Savage puzzle Markowitz's (1952a) mean-variance portfolio theory is the only one inconsistent with the Friedman- Savage puzzle. The two other portfolio theories, Markowitz's (1952b) customary wealth theory and Roy's (1952) safety-first theory, are consistent with the puzzle. Indeed, Markowitz (1952b) introduced customary wealth theory to deal with some unrealistic implications of the Friedman-Savage framework. Using the idea of an efficient frontier, compare w/ mean-variance frontier and shoe that they are different (generate diff portfolios) BPT investors choose portfolios on expected wealth, security/potential, aspiration levels, probabilities of achieving aspiration levels while BPT frontier portfolios consider mean and variance only CAPM investors: combine market portfolio and risk-free securities (bonds and lotteries)

(Behavioral Portfolio Theory, Shefrin & Statman)Comparison of Bernoulli utility function, Friedman-Savage utility function, markowitz's customary-wealth utility function, Kahneman and Tversky's prospect theory utility function

Bernoulli: negative upward sloping FS: squiggle, upward sloping Markowitz: squiggle, inflection point at the middle customary wealth Prospect theory: should remember

Structural estimation: 3 models

CRRA MLA linear MLA Model 1 (CRRA) imposes ex ante reference point while other impose ex-post reference point

(Behavioral Portfolio Theory, Shefrin & Statman) Portfolio Selection in BPT

Combining SP/A theory and mental accounting. Both BPT-SA and mean-variance investors consider the portfolio as a whole, namely as a single mental account. They do so by considering covariances. Mean-variance efficient space achieved by maximizing mu for fixed sigma, and BPT-SA is obtained by maximizing Eh(W) for fixed Prob(W<A) BPT-SA not typically mean-variance efficient

Noisy trader approach

Consider a market consisting of two types of investors: (1) "arbitrageurs"- also called "smart money" are investors who form fully rational expectations about security returns (2) "rational speculators" (3) "noise traders" subject to systematic biases Although riskless arbitrage ensures that relative prices are in line, it does not help to pin down price levels of, say, stocks or bonds as a whole. These classes of securities do not have close substitute portfolios, and therefore if for some reason they are mispriced, there is no riskless hedge for the arbitrageur. For example, an arbitrageur who thinks that stocks are underpriced cannot buy stocks and sell the substitute portfolio, since such a portfolio does not exist. The arbitrageur can instead simply buy stocks in hopes of an above-normal return, but this arbitrage is no longer riskless. If arbitrageur is risk adverse, his demand for underpriced stocks are limited —> demand curve is no longer perfectly elastic

Emotion and Cognition's effect on noisy traders

Demand shifts due to noisy traders will only matter if they are correlated across all noisy traders because if it's random their trades will cancel out However, don't always cancel out b/c many trading strategies based on pseudo-signals, noise, and popular models are correlated, leading to aggregate demand shifts. Think: trading algorithms

Overconfidence and noisy traders

Experimental subjects tend to: (1) be overconfident and take on more risk (2) extrapolate to the past time series (regression to the mean or chasing trends) (3) Subject may underweight old info and overweight new info (overreaction)

Noisy traders over time

First, noise traders might be on average more aggressive than the arbitrageurs - either because they are overoptimistic or because they are overconfident—and so bear more risk. If risk-taking is rewarded in the market, noise traders can earn higher expected returns even despite buying high and selling low on average. Thus, they may not disappear out of the markets

EMH and rational markets

It is often asserted that the EMH holds in rational markets. Such a market is one for which information regarding its assets is freely available to all investors. This, however, does not mean that investors will act rationally based on this information nor does it mean that markets price assets correctly. Even rational markets are subject to the greed and fear of its investors —> bubbles and crashes. Rational markets can behave irrationally because information is not knowledge!

MLA graph versus CRRA

Linearized MLA similar to CRRA. MLA graph is curvy and S-shaped

(Money Illusion, Shafir et.al) Multiple representation

Many ways to represent a situation leads to systematically diff responses: (1) gains and losses (prefer status quo) vs. reporting final assets (prefer risky prospects) think K&T framing effects (2) People tend to adopt the particular frame that is presented and that is guided by availability bias (what info is easy to get and more natural) (3) People may entertain multiple representation at the same time and response is a mix of those induced assessments, weighted by salience - money illusion (people mix real and nominal)

Inefficient Markets

Markets are efficient when prices = intrinsic value, but heuristic driven biases and frame dependence render markets inefficient. Main sources of fallacies: (1) Representativeness and the Gambler's Fallacy (2) Conservatism and regression to the mean (3) Frame dependence (4) Departure from fundamentals (5) Overconfidence

(Behavioral Portfolio Theory, Shefrin & Statman) Conclusion

Proposing two BPTs: (1) BPT-SA where portfolio is integrated (2) BPT-MA where portfolio is segregated such that covariance is overlooked among mental accounts BPT-MA portfolios resemble layered pyramids each with a diff aspiration model ranging from avoiding poverty to getting rich. Since BPT-MA investors overlook covariance between layers, they might combine a short position in a security in one layer with a long position in the same security in another layer. Optimal securities for BPT investors resemble combinations of bonds and lottery tickets. The bonds for the low aspiration mental account resemble risk-free or investment grade bonds, while the bonds for the high aspiration mental account resemble speculative (junk) bonds.

Conclusions:

Studying asset prices in an economy where traders derive utility from consumption and fluctuations from wealth, which they are loss averse over and how loss averse dependents on prior investment performance NEED BOTH MLA AND PRIOR Model replicates asset prices, high mean stock returns, excess volatility, weakly correlated w/ consumption and growth Questions for further analysis: (1) this model only has one risky asset. With many risky assets, what are investors loss averse about? "do they feel loss averse over changes in the value of individual securities that they own, or only over portfolio fluctuations?" A question for their own mental accounting (2) to what extent do our preferences explain not only financial data but evidence on attitudes to risky gambles? (Currently model high curvature utility to explain equity premium puzzle, but only known to be consistent o small scale gambles but predict that people will reject attractive large scale gambles)

Bayes' Theorem

The probability of an event occurring based upon other event probabilities. P(A|B)=P(B|A)P(A)/P(B)

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Preferences for Prepayment

Usual econ theory: think about if the discounted PV of the utility stream is greater than discounted value of loan payments to decide to buy or not. Implications: (1) prefer to pay later (2) financing doesn't get affected by type of product purchased, only by minimizing cost of buying Data shows some goods prefer to be prepaid, some goods people prefer to pay for later. People prefer to have sequences of events that get better with time. People can distinguish between hedonic impact of payments on consumption and consumption on payments. Debt aversion not limited to luxury goods and not always want to expedite financial transaction

Myopic loss aversion graph and equation

Value function u(z)= z for z>0 (gains) and lambda*z for z<0 (losses) where lambda>1 Explanation between subjects, different lambdas will mean that people should invest nothing (high lambda) or everything (low lambda)

Efficient Market Hypothesis (EMH)

WEAK EMH: Given the information available when an investment is made, no investor will consistently beat market returns on a risk-adjusted basis over long periods except by chance or "1. Markets for large classes of assets will lie on the efficient frontier. 2. You can therefore test the weak EMH with MPT. If we understand "market" to mean a capitalization weighted collection of assets (i.e. an index fund) then the EMH can be tested by checking whether index funds lie on or near the efficient frontier. 3. You will see that this is often the case, but not always." If the weak EMH is true, then the only way to beat the market in active managing is by chance or taking more risk Semi strong and strong have bolder claims such as markets represent all information, even those that are not public. The only way an investor can possibly obtain higher returns is by chance or by purchasing riskier investments. Assertion about markets, not investors, but related to investing behavior. It is a common misconception that MPT (modern portfolio theory) assumes the weak EMH. It does not, which is why the EMH can be checked with MPT!

(Behavioral Portfolio Theory, Shefrin & Statman) Roy's safety first portfolio theory

Wants to minimize probability of ruin, when terminal wealth lower than subsidence level Investor w/ normally distributed returns chooses portfolio which minimizes objective function (s-mu_p)/sigma_p He argues that all optimal safety first portfolios lie on mean-variance frontier, but prob bot the case Markowitz's expected utility function: E(W) - c prob(W<s)

MLA prediction

When consequences of three lotteries evaluated in combination, the lotteries are more attractive. Note: EV of a single lottery is greater than zero when lambda < 1.25 for HF and lambda <1.56 for LF. Thus LF prefers riskier asset

(Money Illusion, Shafir et.al) Survey asking to evaluate on economic terms and who is more likely to look for a new job

When prompted to think in real terms, people usually evaluated correctly in real terms. However, when thinking about more vague metrics, they seemed to think nominally. Evaluations in real terms dominate when being prompted to think in economic terms, while less transparent judgements usually biased towards minimal representations

Modern Portfolio Theory and GP97 (mean variance analysis derivation)

diversifying a portfolio can reduce its risk without changing its expected rate of return, using covariant between different assets. Let x= x1, x2,x3, ... xn denote the portfolio composition in terms of the share of the capital invested in each asset. The expected utility hypothesis states that the individual will choose the portfolio weights such that the expected value of utility is maximized Max utility, find that U(w)=E(w)-b(W^2) = E(W)-b(mean+variance) Mean-variance analysis

(Behavioral Portfolio Theory, Shefrin & Statman) SP/A theory

psychological theory of choice under uncertainty. SP/A theory is a general choice framework rather than a theory of portfolio choice. However, SP/A theory can be regarded as an extension of Arzac's version of the safety-first portfolio model. In SP/A theory, the S stands for security, P for potential, and A for aspiration. Lopes' notion of security is analogous to safety in safety-first, a general concern about avoiding low levels of wealth. Her notion of aspiration relates to a goal, and generalizes the safety-first concept of reaching a specific target value. Safety is a decumulative distribution function (prob of wealth falling below S) Fear —> overweighting bad outcomes over good outcomes, causing individuals to act as if computed expectation with values of p that are too high for bad outcomes. Also can be in the reverse of too much hope. Emotions affect weight in function

narrow bracketing

the process of grouping individual choices together into sets, and individuals who narrow bracket tend to make shorter term choices. People who frame outcomes narrowly will evaluate gains and losses more frequently

(The Red and The Black: Mental Accounting of Savings and Debt, Prelec and Loewenstein) Traditional Econ Analysis of Consumer Choice

(1) Assumed to finance expenses to minimize PV of all payments vs their story: debt is unpleasant (2) Costs and benefits of paying off a loan only a financial matter vs their story: you lose utility of continuing to pay off the loan over time (3) Traditional econ gives no explanation for disposing entire load w/ single payment and feel relief after Key idea: pleasure from consumption vs. timing and magnitude of payments which give you disutility. Thinking about payments when you buy diminishes utility of the purchase and thinking about what you will gain when buying will lessen the pain of consuming — two way hedonics

(Mental Accounting and Consumer Choice, Thaler) Methods of raising prices without losing from transaction utility

(1) Increasing reference price, suggest a higher price by increasing perceived cost p* (2) Increase minimum purchase and/or tie purchase to some other purchase (3) obscure p* to make transaction utility less obvious by selling it in an unusual format or way Ex: suggested retain price, usually selling lower than SRP for items whose quality is hard to judge and less sold

(Mental Accounting and Consumer Choice, Thaler) Applications of mental arithmetic to Marketing

(1) Segregate gains: multiple uses, bonus items (2) Silver lining principle: rebate program even if you pay tax on rebates, segregate savings (3) Integrate losses: automobile and house buying, add on buying to another bigger purchase, framing insurance as extra on top of healthcare bill

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Effects on Coupling

(1) Single payment: tight coupling (2)card: reduced coupling, alpha reduces (3) diversity of benefits/payments: membership fees are decoupled more than fee per use (4) earmarking certain accounts to particular expenditure or category of expenditures (college savings plan, retirement, etc.) —> people earmark desirable things when saving and undesirable things when borrowing (5) gifts: decoupling consumption from payments (lowering alpha) willing to spend more on gifts even from same account (6) buyers vs. sellers: coupling desirable for sellers of services or labor but not for buyers Credit cards: prospective accounting (payments applied to past consumption) and decoupling (payments have no unique benefits). Alpha and beta will both be low, creating tendency to want to pay for consumption but not for paying them off

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Three mental accounting assumptions from imputed costs and utility functions

(1) prospective accounting: imputed costs and benefits depends on timing of consumption and payments (2) prorating assumption: simple amortization rule of single payment over time or single utility over time (3) coupling assumption: allows for imperfect imputation of costs and benefits

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Conclusion

-Lack of consistent discount rate -Mental accounting concepts give clues of different in types of peoples -Measuring consumption habits: how much they think about payments when consuming (alpha) and tendency to think about consumption when paying (beta) -People have different mental conventions for defining moment of paying (ex: withdrawing cash and spending it all) causing increased or decreasing coupling -Standard model: anticipated overall utility vs. anticipated overall cost, costs: (1) incurred in future (2) vague form of forgone consumption -This model: people experience immediate pain of paying and once again when future utility is forgone — "two payments" Systematic mismatching: (1) Benefits of a purchase typically more immediate than costs, customer who discounts future steeply finds prospect of future sacrifice not enough to deter impulsive spending (2) Underweighting opportunity costs to out of pocket costs, especially if the decision is for a small amount (slippery slope) (3) Hedonic standpoint consumers want to minimize cost and need to think about the cost of consumption too, but this requires more thought (4) Poor people have increased lambda, higher pain of paying, prefer decoupled, easier to lose from not gaining from mental accounting (5) society level: salience of costs vs. efficient distribution (6) everyone wants to feel like enjoyment/consumption is free

(Money Illusion, Shafir et.al) Effects on the Solow model's claim of efficiency of wages

Assuming people consider nominal and real wages, real wages will respond to the inflation rate b/c the size of the nominal wage increase changes the model.

Main Heuristic Driven Biases (Cognitive dissonance, confirmation bias, conservatism bias, anchor and adjustment, Ambiguity aversion)

4. Cognitive Dissonance:psychological stress by a person who has contradictory believes or values. Causes humans to want to reduce cognitive dissonance by avoiding contradictory information and ideas 5. Confirmation bias: tendency to search, interpret, favor, and better recall information upholding ones' existing beliefs, failure of inductive reasoning. Stronger for emotionally charged, salient issues 6. Conservatism Bias: investors tend to cling to their previous investment decisions despite new information (underadjusting), perhaps to avoid the stress of cognitive dissonance. In conflict with representative bias, where investors may overreact to new information. 7. Achor-and Adjustment: fixation on a target number or value (prior) such as expected prices or econ forecasts, investors often under adjust the original anchor, causing final value (posterior) to be biased towards priors. 8. Ambiguity aversion/Ellesburg Paradox: violation of Bayesian theory, ambiguity aversion prevalent for gains, and ambiguity-loving behavior dominates for losses.

(Can MLA explain equity premium puzzle? Larson et.al 2016) Experiment results: profits

At the end of the experiment, the average trader in the Infrequent group accrued 615,715 units of profit, and the average trader in the Frequent group accrued 401,634 units of profit—a difference of more than 50%. Using a non-parametric Mann-Whitney statistical test, we find that these profits are significantly different at the p < 0.05 level. LF/I made more than HF/F traders. Even though HF traders receive more info, they are worse off profit/earnings wise Examining professional trading in Tobit regression models: regress allocation with trading experience i) those traders with more years of market experience tend to invest less in the risky asset, though the results are too noisy to reject the null hypothesis ii) years of professional experience have little influence on the treatment effect regressing profits on trading experience: (i) not statistically significant (Ii) profits not as penalized as frequent feedback by experience

(Can MLA explain equity premium puzzle? Larson et.al 2016) Design of natural field experiment

Beta testing of new trading platform and they actually get paid for earnings ~1400 per person at the end, ~35% of weekly salary. Enrolled 342 people. Consistent with other natural field experiments, we did not reveal any University affiliations and subjects did not know the request for the beta test was for an academic study. When each group finished we sent out an email informing subjects of their performance and their award. The platform also displayed this final information.

(Barber BB, Odean T, 2001) Calculating returns

Calculating returns: calculated both gross and net returns monthly by estimating bid-ask spread and using a benchmark called "turnover rate" generated from the individual people. Turnover rate = 1/2 sales turnover + 1/2 purchases turnover, and it compares the actual returns a person made at the end of the year versus if they never traded at all (keeping position of the beginning of the year). This helps avoid matching wrong risk model to specific individuals, they choose their own riskiness already by determining beginning of trade purchases

(DeBondt and Thaler 1985) Does the Stock Market Overreact? Theoretical background

Bayes Theorem: Overreaction: In revising their beliefs, individuals tend to overweight recent information and underweight prior (or base rate) data. K&T: representativeness heuristic, People seem to make predictions according to a simple matching to their impressions Debont&Thaler: actual expectations of professional security analysts and economic forecasters display the same overreaction bias Miller&Modigliani: excess volatility is caused by overreaction to short term economic developments over dividend trends P/E earnings ratio anomaly: Stocks with very low P/E earn higher risk adjusted returns than high P/E stocks, perhaps explained by companies with very low P/E are thought to be temporarily "undervalued" Markets may not be totally efficient: existence of quasi-rational agents (Russell&Thaler)

Boys Will be Boys: Gender, Overconfidence, Common Stock Investment (Barber BB, Odean T. 2001) theoretical background

Behavioral economics theory that men tend to be more overconfident than women Traditional finance research: overtrading leads to underperformance, want to extend this and look at more and less oveconfident traders in relation to trading volume. Furthermore, people tend to be more overconfident when they are faced with hard ,ambiguous tasks s.t. Their beliefs cannot be quickly and clearly tested —> picking common stocks a perfect example of this

(Do Professional Traders Exhibit MLA? Haigh&List) Results:

Both traders and students tended to bet higher amounts of wealth in LF/I than HF/F. Treatment effect towards MLA is much higher for traders. Statistically relevant and shows up in regression too. "For example, in specification (2) we find that traders in Treatment F bet approximately 38.5 fewer units than traders in Treatment I, and this difference is significant at the p < 0.01 level. The evidence is slightly weaker for students, where Treatment F subjects bet approximately 25 (13.4-38.5) fewer units than students in Treatment I, a noteworthy differ- ence, but one that is significantly less than the 38.5 unit difference observed between traders."

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Impact of time discounting, when is prepayment/postpayment preferred?

Consumers don't always prepay b/c of time discounting effect, which prefers delaying payments. Trade off between consumption and payment experiences under post/pre payment. Prepayment preferred when discount factor + coupling coefficient alpha >1 (both delta and alpha are big, patient and think about prices) Postpayment preferred when discount factor is small and/or coupling alpha is low (impatient and don't think about prices)

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) paper objective

Create a theoretical model for payment consumption interactions. Postulate that mental accounts of consumption and payments are linked by specific acts of consumption/financial transactions that generate pain or pleasure. Red=owe something, black=balanced/no debt Prediction: consumers do not prefer to consume before spending, they actually have debt aversion and want to pay first. Also have diff. Spending patters with cash/credit and they associate diff money accounts for specific purposes

(An Experiment on Risk Taking and Evaluation Periods, Gneezy&Potters 1997) Conclusion

Direct experimental test on MLA and its prediction that a longer eval period makes a risky option look more attractive, and test shows it is true. Makes people evaluate risk in an greater way and less likely to back off from losses, resulting in a higher payoff. Also supports MLA explanation for equity premium puzzle Giving investors w/ less frequent info will be more positive marketing for fund managers, decreases chance for loss to be experienced Departure from investor case: (1) this only deals with risk with known probabilities, not uncertainty with unknown probabilities (2) Decisions made in a very short amount of time (3) Much lower financial stakes

(Money Illusion, Shafir et.al) Mental accounting

Effect of past nominal values on current decision is a form of money illusion even without inflation. (Wine example) people show that they have conflicting intuition about current values of good bought in the past even with diff accounting methods such as FIFO and LIFO which rely on historically prices

(Do Professional Traders Exhibit MLA? Haigh&List) reason for their research

Equity Premium Puzzle persists but past experiments were done on college students, random people, not experts while decision making usually are made y people with intense market experience. Found that professional traders exhibited stronger MLA behavior than undergrads — suggest that market prices of risky assets may be much higher if feedback frequency and flexibility of decision making is reduced. Institutions may be able to affect human decision making through policies to reduce flexibility and info.

(An Experiment on Risk Taking and Evaluation Periods, Gneezy&Potters 1997) Introduction

Equity premium puzzle: risk-return relationship for stocks so much higher than bonds yet people still hold bonds Bernartzi and Thaler: MLA explains this and Samuelson;s problem, combo of loss aversion and mental accounting. More sensitive to losses than gains and evaluating outcomes of sequences of gambles together. Investors seem to weight losses 2x gains This paper seeks to show prescience of MLA by having subjects make a string of risky choices. Separate them into high frequency choices vs. low frequency choices. MLA predicts that low frequency subjects make riskier choices b/c trade off between losses and gains favors risk in the long run Regular subjective expected utility theory doesn't predict any difference in HF and LF

(Money Illusion, Shafir et.al) Traditional theory and the three classes of anomalies observed

No money illusion models: neutrality proposition in quantity theory but also the assumption of money illusion/no money illusion is invoked and revoked many times Three classes of anomalies: (1) sticky prices, changes in money supply affects prices with a time lag (2) indexing not widely used (3) people talk and write to indicate confusion between real and nominal money —> proposing psychological account of money illusion to help understand and model it

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Prepayment and other strategies for pushing costs out of mind

Foreign currency, play chips allow for prepayment for purchases, makes it easier to spend Mental prepayment via mental budgets/accounting even with time Fixed fee pricing— pricing schemes to eliminate uncertainty in payout amount such as in a grand prixe menu Hypothetical example test: fixed rate contracts superior —> flat rate bias, another form of mental prepayment and shifting cost of payment away from consumption

(Can MLA explain equity premium puzzle? Larson et.al 2016) Design of two groups and what they saw and decisions they can make

Frequent group: can look at prices every second Infrequent ground: can look at prices every 4 hours Both saw a line hat with all past prices up to most recently available, can see their portfolio values changing. Thus the Frequent group saw more negative draws than the Infrequent group. some authors have criticized previous MLA experiments for varying both subjects' ability to make decisions and their information feedback simultaneously **Our approach allows us to identify feedback alone as the driver of MLA.** Further, unlike in some previous experiments, and in line with how trading actually occurs in the field, our traders do not need to make an active decision to hold units of the risky asset. **If they own units, they can continue to hold units by doing nothing, similar to normal stock investment.**

(Do Professional Traders Exhibit MLA? Haigh&List) Background and implications of MLA

MLA= loss aversion + mental accounting Loss aversion: individuals have a value function defined with respect to the status quo, + concave over gains and - convex steeper over losses Mental accounting: refers to how individuals aggregate choices (explicitly or implicitly). In particular it refers to how often transactions/portfolios are evaluated over time and cross-sectionally (whether they are evaluated as portfolios or individually). It helps determines both the outcomes of decisions as well as the framing of those decisions. An agent who frames his decisions narrowly will tend to make shorter-term choices and an agent who frames outcomes narrowly will evaluate her losses and gains more frequently An individual is, therefore, myopically loss averse if he evaluates gains and losses separately as soon as the information is consumed, rather than pool-ing the returns into a lifetime portfolio. This makes individual lottery sequences more attractive to a loss averse individual than aggregate lottery sequences

(Kahneman and Tversky 1973) Availability: A Heuristic for Judging Frequency and Probability Theoretical Background

Little is known, how- ever, about the psychological mechanisms by which people evaluate the frequency of classes or the likelihood of events. We propose that when faced with the difficult task of judging probability or frequency, people employ a limited number of heuristics which reduce these judgments to simpler ones. (Ex: representativeness, an event is judged probable to the extent that it represents the essential features of its parent population or generating process) Alternatively, one may estimate probability by assessing availability, or associative distance. Life-long experience has taught us that instances of large classes are recalled better and faster than instances of less frequent classes, that likely occurrences are easier to imagine than unlikely ones, and that associative connections are strengthened when two events frequently co-occur. Thus, a person could estimate the likelihood of an event, or the frequency of co-occurrences by assessing the ease with which the relevant mental operation of retrieval, construction, or association can be carried out. Availability is an ecologically valid clue for the judgment of frequency because, in general, frequent events are easier to recall or imagine than infrequent ones

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) MIT vacation experiment results

Group A: prefers some prepayment and some interspersing payment and vacation, least desirable is vacation time before paying Group B: Prefers same as A, least desirable is as asymmetrical where you either pay first or vacation first Group C: traditional PV max decision maker, prefers paying vacation off later Discounting model: assumes overall preferences are additive across c and payment. Ratings Regression model: c and preferences are able to be added in the end Mental accounting model: not additive, best according to regression and most exhibition negative loss aversion coefficient (they like pay as you go) Regression: shows that when controlling for debt aversion, subjects prefer later payments and earlier consumption (+ time preference) and good correlation

(Barber BB, Odean T. 2001) Hypotheses and alternative hypothesis

HYP1: Men are more overconfident in women, thus they trade more than women do HYP2: Because they trade more, men are hurt by it in terms of net gain from trading Alternative hypotheses: (1) The difference can be explained by different risk profiles/ different risk aversion (2) Difference in trading performance is caused by skill in selecting better/worse performing common stocks

LCTM Loss Aversion example

Hedge fund:

Frame Dependence #3: Hedonic Editing and consequences

People max psychological well being when making decisions affecting gains and losses —> value function acting on a reference point which separates gains and losses. Hedonic editing —> 1. Loss aversion: value function has v(x) < -v(-x) 2. Gain and loss satiation: v(x) concave in gains, convex in losses, diminishing marginal sensitivity Gains and losses are segregated. Makes sense because value function is concave in gains versus convex in losses. Smaller losses are groups with larger gains because of loss aversion. Small gains are separated from larger losses because of loss aversion. Think: concave gains means that people want gains in small pieces, convex losses means they want losses all at once

(Money Illusion, Shafir et.al) Times money illusion most likely to come up

Holding real change constant, people's reactions usually determined by nominal change Money illusion usually happens when relative prices change and nominal prices may/may not change -anchoring effects onto nominal cost -loss aversion relative to historical reference point -reference point can be nominal and consume people

Transactional Utility Theory, how people evaluate decisions (two types of utility)

How individuals evaluate and asses their choices through a two stage process: (1) Judgement (2) Decision Two kinds of utility: 1. Acquisition utility v(p_bar- p)—> some goods are purchased because they are really good deals 2. Transactional utility v(-p|-p*)—> some goods that would make the person better off are avoided because you have a overinflated reference point and find it too costly Ex: Beer at hotel vs. at run down store The idea of fairness: how much are you willing to pay for a good

(Mental Accounting and Consumer Choice, Thaler) Hedonic Editing, coding gains and losses

How we code gains and losses to make yourself the most happy, implications of prospect theory Should segregate multiple gains and increase in gains b/c v(x) concave Should combine losses or increase in losses (integrate) because v(-(x+y))>-v(x)-v(y) A decrease in gains should be integrated (cencellation) Combine mixed gains Either segregate or integrate mixed losses, make sure to show the small gains from the large losses or small reduction in loss (silver lining principle)

(DeBondt and Thaler 1985) Results

Loser portfolios of 35 stocks outperform the market by, on average, 19.6%, thirty-six months after portfolio formation. Winner portfolios, on the other hand, earn about 5.0% less than the market The overreaction effect is asymmetric; it is much larger for losers than for winners. Secondly, consistent with previous work on the turn-of-the-year effect and seasonality, most of the excess returns are realized in January —> January Effect One method that allows us to further accentuate the strength of the January effect is to increase the number of replications. random noise. The outstanding feature of Figure 3 is, once again, the January returns on the loser portfolio. The effect is observed as late as five Januaries after portfolio formation! Agrees w. Tax-loss hypothesis. In surprising agreement with Benjamin Graham's claim, the overreaction phenomenon mostly occurs during the second and third year of the test period. The overreaction hypothesis predicts that, as we focus on stocks that go through more (or less) extreme return experiences during the formation period, the subsequent price reversals will be more (or less) pronounced. Table I confirms the prediction of the overreaction hypothesis. As the cumulative average residuals (during the formation period) for various sets of winner and loser portfolios grow larger, so do the subsequent price reversals, measured by [ACARL,t - ACARw,,] and the accompanying t-statistics. For a formation period as short as one year, no reversal is observed at all. For all the experiments listed in Table I, the average betas of the securities in the winner portfolios are significantly larger than the betas of the loser portfolios. Thus, the loser portfolios not only outperform the winner portfolios; if the CAPM is correct, they are also significantly less risky. Results are likely to underestimate statistical significance and magnitude of overreaction effect.

Frame Dependence #1: Loss Aversion

If an investment problem is presented in two different (but really equivalent) ways, investors often make inconsistent choices, showing that they are influenced by other factors including language used People would rather accept sure losses but want to take gambles on gains. Think about prospect theory, which has concave less steep gains, convex steep losses. Although two situations may have same expected outcome, you would prefer sure loss over gambling —> direct consequence of framing. "Get-Even-itis" investors want to not sell at a loss, shows consideration fo sunk costs, which traditional econ theory says shouldn't be evaluated. The investments are evaluated against some arbitrary reference point and gains are treated differently from losses against this point Ex 1: Apple Newton product, kept producing even though it was making a big loss. Cognitive dissonance in play, emotional loss Ex 2: LCTM hedge fund by Nobel prize winners and other famous people, collapsed after Asian and Russia financial crises of 1997 and 1998

(Money Illusion, Shafir et.al) Investment example, imaging you are portfolio manager choosing between bond/stock

Immediate feedback, investing under inflation and no inflation. People tend to invest in risky fund way more when there is inflation b/c of nominally positive return feedback. Also exhibited some loss aversion

(Mental Accounting and Consumer Choice, Thaler) theoretical background

Implicit mental account systems used by individuals hand households have certain consequences, which can lead to a richer theory of consumer behavior that can explain violations of fungibility (money spent is all the same regardless of labels) and others. Departure from traditional microeconomic theory which doesn't think about mental accounting, framing, sunk cost analysis, etc.

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Imputed costs and benefits

Imputed cost is what fraction of all payment for consumption is applied to the consumption at any time Similar sum indicates what fraction of consumption utility is applied to payment at time c (Wbc, Wcb) mental accounting system described by coordinates, specifies how costs match up to benefits.

(Can MLA explain equity premium puzzle? Larson et.al 2016) Experimental results: risky asset investment

MLA effect takes some time to arise, traders need time to learn expected value of risky asset. "the end of the experimental time period, traders who receive infrequent information invest significantly more in the risky asset than traders in the frequent information treatment—at a rate roughly 40% higher. This difference is significant at the p < .05 level using a Mann- Whitney test." Robustness check: matching pairs at randomization for a second empirical test showing slightly larger treatment effect. Aggregating investment patterns of last 20% of experiment: while traders invest on average 36.4% of their portfolio in the risky asset in the LFI treatment (n=73), they invest only 27.4% of their portfolio into the risky asset in the HF/F treatment (n=78). Consistent w/ MLA

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Planned vs. unplanned debt, explaination for procrastinating on paying off debt

People may fail to borrow enough or borrow too much depending over/underestimation. Also may put off paying off old debts. Possible explanation: hyperbolic discounting creates incentives to postpone any painful event. Procrastination may be incentivized under hyperbolic discounting even with debt aversion

(Kahneman and Tversky 1973) Availability for Retrieval

In this section we discuss several studies in which the subject is first exposed to a message (e.g., a list of names) and is later asked to judge the frequency of items of a given type that were included in the message. Subjects attempt to recall some instances and judges overall frequency by availability, i.e., by the ease with which instances come to mind. Study 8: Fame, Frequency, and Recall, given recorded list consisting of names of known personalities of both sexes. After listening to the list, some subjects judged whether it contained more names of men or of women, others attempted to recall the names in the list. Fame and frequency were inversely related in all lists. Subjects tended to recall more famous than nonfamous names. Among the 99 subjects who compared the frequency of men and women in the lists, 80 erroneously judged the class consist- ing of the more famous names to be more frequent. In concluding the discussion of the apparent frequency of repetition, it is important to emphasize that the availability heuristic is not the only method by which frequency of repetition can be estimated. The strategies employed to estimate the frequency of a single item can also be employed to estimate the frequency of an item-pair. In addition, the repetition of a pair strengthens the association between its members. The subject may, therefore, use the strength of the association between the members of a pair as a clue to its frequency. (Illusory correlation effect) that an assessment of the associative bond between two items is one of the processes that mediate the judged frequency of their co-occurrence. Thus, when a person finds that the association between items is strong, he is likely to conclude that they have been frequently paired in his recent experience. According to this account, illusory correlation is due to the differential strength of associative bonds. The strength of these bonds may reflect prior association between the items or other factors. Thus, the various sources of illusory correlation can all be explained by the opera- tion of a single mechanism-the assessment of availability or associative strength. The proposed account of the judgment of the frequency of CO- occurrences is tested in the last two studies. Study 9: Illusory Correlation in Word Pairs, establishes the relation between judgments of the frequency of pairs and cued recall. Highly related were recalled better than unrelated words and people thought they occurred more frequently even when low related occurred more often Study 10: illusory Correlation in Personality Traits, Assessed recall. Same results as study 9

(Money Illusion, Shafir et.al) Degree of satisfaction on salary compared to coworkers experiment

Individuals likely to say that person is more happy when they earn more relative to their coworkers even if they are earning less than others in other companies in real terms. (Discrepancy between absolute and comparative job evaluation) Similar to money illusion because when prices change relatively a person's buying power changes too. If judging on the number of dollars made, then preferences correlate to nominal, not real changes

(Do Professional Traders Exhibit MLA? Haigh&List) Concluding Remarks:

Inference for experiments for MLA subject to critique that treatment effects seen in studied done on undergrads may not reflect reality of economic decision makers. Found that traders exhibit MLA more than students. Implications: (1) First, our findings suggest that expected utility theory may not model professional traders' behavior well, and this finding lends credence to behavioral economics and finance models, which are beginning to relax inherent assumptions used in standard financial economics (2) findings have direct implications on the communication strategies for fund managers, whereby revealing infor- mation on a less frequent basis means that the likelihood of incurring a loss is reduced (3) less freedom to adjust (i.e., inducing agents to think in a more aggregated way) might reduce the likelihood that a sell-off en- sues after a minor setback. (4) From GP 97: market info becoming more readily available at a lower cost—> one might expect it to be used more often—> affecting behavior over riskier assets and therefore relative prices.

Examples of applications

Infomercials: gains are segregated Auto rebates: silver lining principle applied, you pay a lot of money but you are happier about decision because of a small amount of money Gym memberships: integrating losses (yearly payment) segregating gains (all the perks)

Frame Dependence #2: Concurrent Decisions

People tend to be risk averse in gains and risk loving in losses. Framing matters a lot, and the claim in that two identical situations in terms of outcomes can be perceived as different problems causing difference choices by one decision maker. Concurrent decisions: two decisions that correspond to the same problem being conceived as two different problems and may cause choice reversals. Relevant theories: prospect theory, loss aversion, hedonic editing, probability weighting

(Mental Accounting and Consumer Choice, Thaler) New Behavioral Econ based utility

New model: (1) utility function replaced by value function from prospect theory, concave gains, convex losses (2) reference price is introduced, developing transaction utility, and gains and losses are separated, framing effects (3) Fungibility is also relaxed and marketing implications of theory derived, not all money is the same

(Money Illusion, Shafir et.al) Wage cuts and money illusion

Not a continuous relationship between raising wages and effort workers put in due to nominal wage cuts. Money illusion may affect allocation of workers across jobs and aggregate employment. Workplaces may want to reduce money illusion created lower worker morale

(An Experiment on Risk Taking and Evaluation Periods, Gneezy&Potters 1997) Procedure

Notable parts: subjects were actually paid final endowment. For LF subjects, they were shown all three outcomes after 3 time periods at once.

(Can MLA explain equity premium puzzle? Larson et.al 2016) Introduction

Once again introducing equity premium puzzle and the MLA explanation in which investors pay too much attention to short term volatility of asset portfolios. Lab experiments, finding individuals receiving LF feedback tend to invest more in risky assets than HF feedback people. However, no evidence on if professional traders behave in line with theory in their work. ESP marginal traders are the price setters in markets. Notable problem: hard to observe traders in their natural environment. This paper gets data from partnering with beta testing trade platforms where it's a natural setting and traders don't know they are being beta tested. Using randomization of info feedback as identification and giving stakes that are similar to their normal salary (can ear up to 35% of their weekly salary) Found trading patterns matching with MLA, traders w/ infrequent price info invested ~33% more in risky assets —> higher profits. Implications: (1) Leads to puzzle: traders seek more info in field resulting in lower returns and profits (2) Evidence for MLA to solve equity premium problem (3) lend insight to asset volatility, dividend payouts, discounting (4) Natural experiments seem to estimate policy/theory parameters

Frame Dependence #6: Regret Theory

People anticipate regret if they make a wrong choice, and this relates to prospect theory. Fear of regret may cause people from taking certain actions because they pay attention to sunk costs (sunk cost fallacy). Avoiding risk in this way may cause investors to keep poor investments too much — related to disposition effect

GP 97 Investment game and later experiments

People bet more in LF treatment, effect is stronger for professional traders and ambiguity

(Kahneman and Tversky, 1974) Adjustment and Anchoring and the biases they may cause

People make estimates by starting from an initial value and adjust that to the final value. This anchoring and adjustment effect could cause biases including: (1) insufficient adjustment, final value greatly affected by initial estimation, which could be incorrect. Also can be a result of incomplete computation, in which assumptions about a final product depends on mentally computing the first few products (2) Biases in the evaluation of conjunctive and distinctive events, People tend to overestimate probability of conjunctive (drawing red marble 10 times in a row) events and underestimate probability of disjunctive (drawing a red marble at least once in 7 tries) events. The simple event is the probability anchor point, people fail to adjust enough for disjunctive and conjunctive events. This causes people to underestimate probability of failure in complex systems which rely on conjunctive systems (3) Anchoring in the assessment of subjective probability distributions, people fail to adjust their initial probabilities upwards or downwards enough to get proper probability distributions. Asking someone to adjust from a value stated in the question versus what people think is right gives less extreme odds. Procedure of elicitation greatly affects calibration

Why should practitioners know about behavioral finance?

Practitioners are often unaware that they make mistakes b/c (1) They rely too on rules of thumb too much (2) Perception of risk and returns are biased by cognitive processes, especially how situations are framed (3) As a result, market prices often deviate from its fundamental values —> markets are inefficient. Knowing about behavioral finance will allow practitioners to predict biases in a systematic way and possibly correct

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Debt Aversion: durable vs. nondurable

Prepayments (1) decreases sum of residual payments, increasing net enjoyment (2) increases sum of residual utilities, decreases pain of payment Durable goods: little cost of delaying payments b/c still get additional utility from the good. People want to keep accounts out of debt —> accelerate prepayment for non durable items such as falling property values or less utility over time

(Kahneman and Tversky, 1974) Representativeness causes errors such as

Probabilities are gauged by how representative A is of B or how similar A and B are. This causes errors because (1) Insensitivity to prior probability of outcomes/base-rate. For example, it may be the case that all people of a sample are just more likely to have a certain characteristic. (2) Insensitivity to sample size, on average results are more likely to happen under a larger sample size (larger sample size gives a normal distribution) (3) Misconceptions of chance: expecting chance to correct for variability in anticipation for certain patters, gambler's fallacy, belief in the 'law of small numbers,' small numbers are not very representative of the larger sample/population they are drawn from (4) Insensitivity to predicability: different evidence have different levels of predictability, extreme ness and range of predictions are controlled by considerations of predict abilities. If there is no evidence for a certain thing happening, predictions should be the same. The higher the predictability, the wider the range of predicted values, and intuitive predictions ignore this. (5) illusion of validity: Selecting the outcome that is most representative of the input independent of its predictive accuracy. Info may be scanty, unreliable, outdated, so this confidence is unwarranted. This is more likely to occur under highly consistent patterns, even if each event may be independent. (6) Misconceptions of regression, regression towards the mean doesn't happen when people expect, and when it happens people may invent other ways to explain it.

(Money Illusion, Shafir et.al) Discussion

Propose that economics agents consider nominal and real costs, but are usually biased towards nominal costs. Money illusion also shows up in relation to framing, anchoring, mental accounting, and loss aversion. People use nominal values so much because it is salient, easy to gauge, and presents a close enough estimate to what is going on in real terms usually. Also found that money illusions also happen when there is no inflation. Also interacts to caring about sunk costs, under weighing opportunity costs relative to out of pocket costs Money illusion could result in larger inflation, poverty among elderly, multinational trade, tourism, etc.

(Barber BB, Odean T) Results

Proved both hypotheses that (1) Men were overconfident, which leads to them trading more than women (2) The overtrading led to a net loss in investing profit not due to skill in investing, but rather just trading volume (more trade —> more losses) Additionally: (1) This effect was more dramatic in single men vs. single women, presumably because married couples may influence each others' investing decisions (2) Regression of performance and other factors including age, income, gender: gender most significant, age somewhat significant (3) Risk Profiles: men also tend to invest in riskier stocks (stock volatility, portfolio volatility, beta, size of company), especially high income men (4) Risk aversion may explain why people may trade more regarding unsure hypotheses, but needs overconfidence to explain why they trade despite trading lowering their expected returns (5) Gambling related behavior such as risk seeking or entertainment accounts also cannot explain the effects seen in the data Thus, people behave irrationally in a predictable manner contrary to traditional investing theory

(An Experiment on Risk Taking and Evaluation Periods, Gneezy&Potters 1997) Results

ROUND 1 (1) Average bets are larger for treatment L than for H in each round, significant since the first round and consistently significant — statistically significant (2) Average level of bets are fairly stable over each round, a little lower in the middle three rounds (3) Subjects are somewhat forward looking when evaluating risky decisions (prospective theory) (4) Hypothesis: experiencing gains and losses affects subjects' risk behavior, but disproved because no significant differences between the diff rounds. ROUND 2 Endowments the same across rounds but differed across individuals, look at amounts and % of endowment (1) Treatment L still bet more on risky lottery, in relative and absolute terms (2) Increased willingness to take risk pays off— treatment L significantly higher payoff

(The Red and the Black: Mental Accounting of Savings and Debt, Lowenstein & Prelec) Prospective Accounting, imputed costs, alpha and beta

Relationship btwn imputed cost and timing of the payment. Imputed cost is highest if payment due right after the vacation, then declines some time. Alpha- attenuation (link between payments and consumption) Low alpha means that person doesn't really think about prices when buying Beta- buffering (degree to which payments change pleasure of consumption and how much pleasure of consumption changes pain of payments. Low beta means that they don't get much solace from thinking about benefits of consumption Likely to differ across methods and situations of payment.

(DeBondt and Thaler 1985) Conclusions

Research in experimental psychology has suggested that, in violation of Bayes' rule, most people "overreact" to unexpected and dramatic news events. Consistent with the predictions of the overreaction hypothesis, portfolios of prior "losers", are found to outperform prior "winners" even though the latter are significantly more risky. Several aspects of the results remain without adequate explanation; most importantly, the large positive excess returns earned by the loser portfolio every January. Much to our surprise, the effect is observed as late as five years after portfolio formation.

Larson et.al vs. GP97 setup and constraints

Same as GP97 in setup except: (1) natural field experiment w/ 300+ pro FX traders (2) Unique asset in the market based on euro/$ exchange rate where price is subsidized to mimic positive return (3) Random assignment to HF and LF groups (4) Can trade at any time (5) short selling allowed In class slides: (1) risk (GP97) vs, Ambiguity (LLM16) (2) LLM16: diff cohorts experience different price paths (3) GP 97 lottery outcomes are kid (4) LF is a constraint for decision timing in GP97 but not in LLM16 (5) Positive endowment in GP97, no endowment in LLM16 (6) Can't lose money in both experiments technically (7) lab data shows a more pronounced frequency effect (8) 10% of subjects are practically inactive at all times

(Barber BB, Odean T, 2001) Security selection and implication

Security selection: Split them up into male sales and purchases, female sales and purchases. Remember the hypothesis is that men trade more and underperform as a result. Use data to show that both males and females lose from trade somewhat equally so it is not a case of one being better at picking stocks. Also related to disposition effect and get-even-itis which both men and women experience

History of Behavioral Finance

Selden (1912): prices may be dependent on mental attitudes Festinger (1956) Cognitive dissonance, when two simultaneously held belief are inconsistent, it is unpleasant and the person will change their beliefs to match one or the other Kahneman and Tversky Heuristic driven biases (1974) such as a result of heretics such as adjustment and anchoring, availability, representativeness Prospect theory (1979) Framing (1981) Thaler (1980): new theory of consumer behavior w/ underweighting of opportunity costs, failure to ignore sunk costs, search behavior, choosing not to choose, regret, precommitment, self control DeBondt and Thaler (1985) markets overreact and are inefficient Thaler (1985) mental accounting Samuelson and Zeckhauser (1988) Status quo as a reference point and decreasing from baseline is a loss Herd Behavior and information cascades: people tend to disregard private information in favor of a more general view —> Bayes Rule application Benartzi and Thaler (1995): Myopic Loss Aversion (MLA) Behavioral CAPM Disposition effect

(An Experiment on Risk Taking and Evaluation Periods, Gneezy&Potters 1997) Design, what MLA predicts, what SEU theory predicts

Sequence of three independent but identical lotteries, individual puts different weights on the loss probability. Expected utility of a single lottery is positive for a higher number than for a sequence of lotteries b/c probability of loss in succession is less, making lotteries more attractive, predicted by MLA. Thus, want to replicate this in subjects by manipulating evaluation period Subjects fully informed about probabilities of winning and losing, and of gains/loss amount. Also each round is independent, can't bet money from previous rounds Two different treatments: H(high freq) and L (low freq). Treatment H plays rounds one by one while treatment L plays every 3 lotteries. Each were notified of win/loss after they play. Treatment L has less freedom, info, choice, feedback —> mimics betting in an more aggregated way, should make them more likely to bet NULL HYP: SEU theory: behavior of L is different from H because H has more information on current wealth level, can adjust bets along the way, etc. but still should be no systematic diff between HF and LF ALTERNATE HYP: MLA theory

(Kahneman & Tversky 1974) Aim of Paper

Showing people rely on a limited number of heuristic principles that reduce complex tasks of assessing probabilities and predicting values to simple judgement operations, which is useful but may produce systematic errors. Talks about three heuristics: Representativeness, Availability, Adjustment/Anchoring.

(Money Illusion, Shafir et.al) Money Illusion

Tendency to think in terms of nominal rather than real monetary values, implies a lack of rationality. Presence of inflation, nominal accounting methods, relative prices, etc. affect decisions Propose that people think about econ transaction in nominal and real terms, bias towards nominal evaluation anyway. Analyze interactions w/ loss aversion, risk attitudes, fairness concerns Bias in the assessment of the real value of econ translations due to ease, universality, and salience of nominal representation (real changes tempered by nominal changes)

(Kahneman and Tversky, 1974) Availability and the errors it can cause:

Situations where people judge the frequency of a class or the probability of an event by ease of which instances or occurrences can be brought to mind. Useful clue because large classes usually are recalled better and faster than less frequent classes. However, prone to biases such as: (1) Biases due to retrievability of instances, frequency is sometimes unrelated to how easily remembered an instance is. Salience (how impressionable), familiarity pose problems. Ex: lists of well known personalities of both sexes tend to remember more famous people better and skew perception of gender proportions (2) Biases due to effectiveness of a search set, it may be easier to search for data mentally one way, biasing impression of probability of frequency. This can be affected by context (abstract>objective examples) (3) Biases of imaginability, when one cannot come up with examples from personal experience, they form hypothetical examples and construct instances. Some are more easily constructed than others, which do not actually reflect actual frequency. (4) Illusory correlation, two sets of data provided together may cause some implied correlation even when they are said to be unrelated.

Frame Dependence #5: Self-Control

Some investors use mental accounting as a form of self control. But then it results in situations where they think of dividend gains as an income instead of a capital gain. Income comes from selling stocks, which actually reduce capital while capital gains don't reduce capital. Mental accounting principals can help brokers make financial products more attractive when they separate the different type of gains into call premium, dividend, and capital gain on stock.

(Can MLA explain equity premium puzzle? Larson et.al 2016) Design of risky asset

Started with some amount, and they received a fraction of that amount at the end. Make trades with a large bid/ask spread (2% of price). 10 trading days. Fund A: positive expected returns, upward shifts in price increased fund more than decrease. Based off a live price of a real asset for naturalness. Fund A returned ~236% for each period.

(Kahneman and Tversky 1973) Assessments of Availability

Study 1: word construction problems where some were asked to estimate how many words they could construct beforehand. The mean number of words produced varied from 1.3 to 22.4 with a grand mean of 11.9. The mean number estimated varied from 4.9 to 16.0 , with a grand mean of 10.3. The product-moment correlation between estimation and production, over the sixteen problems, was 0.96. Study 2: Brainstorming things that fit into a category, measuring retrieval. The product- moment correlation between production and estimation over the 16 categories was 0.93. Discussion: In the above studies, the availability of instances could be measured by the total number of instances retrieved or constructed in any given problem.5 The studies show that people can assess availability quickly and accurately. How are assessments made? (1) cumulative retrieval of instances is a negatively accelerated exponential function of time (2) subject may assess availability without explicitly retrieving or constructing instances

(DeBondt and Thaler 1985) Data Construction

Study assess the extent to which systematic nonzero residual return behavior in the period after portfolio formation (t > 0) is associated with systematic residual returns in the preformation months (t < 0). We will focus on stocks w/ extreme losses or gains over periods Winner (W) and Loser (L) portfolios formed conditional upon past excess returns. Efficiency market hypothesis estimates that the expected returns of both W and L are equal to 0. Overreaction hypothesis suggests that E[returns of W|market info]<0, E[returns of L|market info]>0. Data uses 3 return residuals: market-adjusted excess returns; market model residuals; and excess returns relative to Sharpe-Lintner version of CAPM but all three returned similar results we use market-adjusted returns, which are more likely to bias against overreaction hypothesis Monthly data from NYSE manipulation: (1) Residual returns estimated for stock w/ >85 months of return data, returned to populate list w/ more securities (2) compute the cumulative excess returns for the prior 36 months, ranked from low to high (top 35) and winning/losing portfolios formed (low 35) (3) compute the cumulative average residual returns of all securities in the portfolio, stock is dropped if return is missing in a month after portfolio is created —> rebalancing (4) Average CARs calculated for both portfolios for each month and then make a pooled estimated of population variance in CAR by forming t statistics (5) calculate standard deviation for winning and losing portfolios to see if the average residual return makes a contribution to either average CARs

(Money Illusion, Shafir et.al) Contracts example

Subjects reversed choices for risky vs riskless contracts when framed in real vs. nominal choices because perceptions change with the frame although a=c and b=d choices. When presented with neutral framing, they chose reckless again. People naturally evaluate contract in predominately nominal terms and try to avoid nominal risk. For buying rather than selling, again frame dependent risk aversion and trying to opt for sure nominal value

(DeBondt and Thaler 1985) Implications

The January phenomenon is usually explained by tax-loss selling (see, e.g., Roll [23]). Our own findings raise new questions with respect to this. First, if in early January selling pressure disappears and prices "rebound" to equilibrium levels, why does the loser portfolio-even while it outperforms the market-"rebound" once again in the second January of the test period? Secondly, if prices "rebound" in January, why is that effect so much larger in magnitude than the selling pressure that "caused" it during the final months of the previous year? Possible answers to these questions include the argument that investors may wait for years before realizing losses, and the observed seasonality of the market as a whole. With respect to the PIE effect, our results support the price-ratio hypothesis discussed in the introduction, i.e., high PIE stocks are "overvalued" whereas low PIE stocks are "undervalued." However, this argument implies that the PIE effect is also, for the most part, a January phenomenon.

What is Behavioral Finance?

The application of psychology to financial behavior, aka the choices of the practitioners.

(Kahneman and Tversky, 1974) Discussion

The paper talks about cognitive biases stemming from reliance on judgement heuristics: (1) representativeness, (2) availability, (3) anchoring and adjustment Even when subjects were awarded for more correct answer, the errors in judgement still continued to occur, even for those who are skilled in statistics. For example, people skilled in statistics failed to recognize regression towards the mean or the effect of sample size on variability. Thus, more studies need to be done on how people make decisions knowing about the biasing tendencies of the heuristics they choose to use.

(DeBondt and Thaler 1985) Discussion

The requirement that 85 subsequent returns are available before any firm is allowed in the sample biases selection towards large, established firms. But, if the effect under study can be shown to apply to them, the results are, if anything, more interesting. In particular, it counters the predictable critique that the overreaction effect may be mostly a small-firm phenomenon. The decision to study the CAR's for a period of 36 months after the portfolio formation date reflects a compromise between statistical and economic considerations, namely, an adequate number of independent replications versus a time period long enough to study issues relevant to asset pricing theory. Portfolio creation month is essentially arbitrary.

(Kahneman and Tversky 1973) Availability for Construction

The subject is given a rule for the construction of instances and is asked to estimate their total (or relative) frequency. The subject is given a rule for the construction of instances and is asked to estimate their total (or relative) frequency. Study 3: Judging word frequency that has a certain rule. Results showed a bias towards the first position to be more likely for a majority of the letters because it is easier to think of words starting with a letter although later parts may be more common. Study 4: Permuations, people tend to think that there are more paths through A than B because the paths to imagine are shorter although same # paths actually Study 5: Combinations, Consider a group of ten people who have to form committees of r members, where r is some number between 2 and 8. Max the different committees of r members formed is when r=5. Easier to imagine smaller groups, so bias towards them. Numerosity of committees decreases with their size. Analogous bus stopping patterns tests the same theory, the apparent number of combinations generally decreases with T, in accordance with the prediction from the availability hypothesis, and in marked contrast to the correct values Study 6: extrapolation, We asked subjects to estimate, within 5 sec, multiplication ascending and descending on board. The median estimate for the descending sequence was 2,250. The median estimate for the ascending sequence was 512. Both the underestimation of the correct value and the difference be- tween the two estimates support the hypothesis that people estimate 81 by extrapolating from a partial computation. Study 7: Availability vs Representativeness, availability in the evaluation of binomial distributions, asking about columns of Xs and Os. We propose they glance at the diagram and estimate the relative frequency of each path-type by the ease with which individual paths of this type could be constructed. When presented same ratio of Xs and Os as a proportion, people estimated more correctly. Thus, different representations of the same problem elicit different heuristics. Frequency of a class is likely to be judged by availability if the individual instances are emphasized and by representativeness if generic features are made salient.

Frame Dependence #4: Emotion and Cognition

Think CRT test, which is an cognitive emotion test. Every financial decision is affected by cognitive and emotional component. The cognitive on processes information and people's feelings are affected by info coming in. Behavior depends on how problems are framed and naturally some prefer one frame over another. Outcomes don't directly go to utility numbers, utility theory looks at in comparison to a reference point. Dual systems: automatic system 1 VS. Conscious processing system 2 (slower, calculating) —> parallel thinking that collaborate and take over different processes

(Do Professional Traders Exhibit MLA? Haigh&List) Experimental Design

Traders and students in HF and LF treatments with F=HF and I=LF. LF/I treatment group: Following Gneezy and Potters (1997), we restricted the bets to be homogeneous across the three rounds. Most importantly, after subjects placed their bets, they were informed about the combined realization of the three rounds. This contrasts with our assignment of gains and losses after each round in Treatment F, and provides heterogeneity in the evaluation period. Gave traders higher stakes than students. Remember: Gneezy Potters showed that LF resulted in higher percent of wealth invested.

(Mental Accounting and Consumer Choice) Transactions and Utility Implications:

Transaction utility: loss of utility from raising prices Sellouts and scalping: many markets don't clear (labor, concert tickets, game tickets, etc.) b/c of transaction utility under market clearing price higher than normal price and wanting to maintain relationship between buyer and seller Many alternatives for each item, consumer would be offended by a higher price and buy some alternatives. Also want to keep selling over time (short term gain vs long term loss) Allow reselling, want the reputation of giving a fair price and reaping relationship benefits

(Barber BB, Odean T, 2001) Data

Two data sources: large discount brokerage end of the month position statements + demographics report of these households from another provider (age, gender, etc). + self reported data including experience in trading, net worth, etc.

Beauty Contest Guessing Game

Uniques equilibriums reached through target based iterative process, best reply process, and iterated deletion of dominated strategies Winning choice of 13 implies massive off-equilibrium play. Market environments require a consistent belief on the rationality of others

Behavioral Bayesian Updating, changing Bayes rules and explanation of why

Want to incorporate new evidence into previous system of prior beliefs. Stats+game theory+psychology+finance: want to create a more relevant Bayes Rule essentially for stock market, which is shown to overreact. Note that Kahneman and Tversky's representativeness heuristic causes overreaction, while anchoring and adjustment can also cause under reaction to new info. Bayes Rule: P(A|B)= [P(B|A)P(A)]/P(B) from conditional probability New formulation of Bayes Theorem: P(A|B)= [P(B|A)P(A)]/[P(B|A)P(A)+P(B|~A)P(~A)] where P(~A) is belief against A and P(B|~A) belief in B given A is false. Explanation of new theory: (1) people overreact to new info by neglecting base-rate info (notice that P(A)P(A|B) / P(A)P(B|A)+P(~A)P(B|~A) cancels out the P(A) from top and bottom) (2) inverse fallacy, confusing P(A|B) with P(B|A) which causes constant base rates for alternate hypotheses (3) Overly optimistic? (4) Asymmetry in updating info based on news/established reference points (5) Psychological game approach

(DeBondt and Thaler 1985) Does the Stock Market Overreact? Hypothesis

Want to test if the overreaction hypothesis is predictive and if it can explain the P/E effect or asset price dispersion (1) Extreme movements in stock prices will be followed by subsequent price movements in the opposite direction. (2) The more extreme the initial price movement, the greater will be the subsequent adjustment. Both hypotheses imply a violation of weak-form market efficiency