Behavioral Economics: An introduction

1. Behavioral Economics:An introduction Edward Cartwright

2. Outline of the session • A brief introduction to behavioral economics. • Fairness and reciprocity. • Learning from new information. • Coordination problems. • A very brief look at other areas in behavioral economics. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

3. Why behavioral economics? The workhouse of economic modeling is homo-economicus, an agent who: • Optimally maximizes his expected utility. • Optimally updates his beliefs according to Bayes rule. • Is selfish and without emotion, or, more formally, does not care about the consumption and utility of others. This approach has yielded fantastic insight, but… Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

4. The motivation behind behavioral economics • Do people behave like homo-economicus? • If not, how do they behave? • What are the implications for economic theory and policy? Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

5. What behavioral economics does? It adds to the standard model of economics some reality about how humans behave. In particular, it adds, • bounded rationality, • biases in interpreting information, • interdependent preferences, • emotions, • Learning, • …. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

6. What behavioral economics is not? Part I • It is not about throwing away the economics textbook to start from scratch. • Behavioral economists fully recognize the crucial role played by models based on homo-economicus. [Many of them have helped to develop them.] • They want to work with and adapt these models to take account of human behavior in those instances where it seems important to do so. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

7. What behavioral economics is not? Part II • It is not about reinventing psychology. • Behavioral economics does, and should draw on psychology but • is focused on different questions to psychology, • retains the methodology and mathematical rigor familiar in economics and game theory. • The ‘mindless economic’ debates continue on how much neuroscience and evolutionary psychology, and the like, really add to economics. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

8. The basic nature of behavioral economics • We can find that people do behave as if homo-economicus. • We can find that people have inter-dependent preferences, and emotions, but are behaving ‘rationally’ relative to these. • We can find that people are biased in choices and how they interpret information. • We can say something about settings where outcomes are ambiguous with homo-economicus. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

9. The methods of behavioral economics • Experiments • lab based, • in the field, • neuroscience • Theoretical • game theory • decision theory • evolutionary theory • Simulation • Agent based models Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

10. What about policy? • By its nature behavioral economics should be relevant in all areas of economic policy. • If policy is about influencing individuals (even if they are within a corporate or other structure) then behavioral economics is crucial to get things right. • Policy makers should be worried about a science built on Friedman’s positive methodology. Behavioral economics is diametrically opposite to a positive methodology. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

11. A not too serious example • A problem for primary schools and nurseries is parents picking their children up late. The school must play the role of baby sitter. • Suppose that we fine parents for picking their children up late? • The result can be more parents leaving their children late because the fine makes it ‘ok’ to put a burden on the school. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

12. The results of an experiment • Gneezy and Rustichini (2000) report an experiment in day care centres in Haifa, Israel in 1998. In week 4 a fine was introduced and in week 17 it was removed. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

13. A sketch history • Behavioral economics naturally emerged with game theory in the 50’s and 60’s. The likes of Vernon Smith, Kahneman and Selten showed it’s power. • From the 80’s onwards behavioral economics has been the fastest growing area of economics. Partly due to dissatisfaction with the ‘standard model’. Partly due to the breadth of talent that has worked in the area. • But note that behavioral economics is not new. Historically, economists, including Adam Smith, Keynes and Marshall talked a lot about behavioral tendencies. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

14. Topic I: Fairness and reciprocity • Have at look at the hypothetical scenarios that you have been given. • How do you think people behave in these scenarios (go with your instinct)? • How would homo-economicus behave? • Why are these scenarios different? • Why are they similar? Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

15. Dictator and ultimatum games • The Nash equilibrium in all of these games is simple: • The receiver should accept any positive amount, because something is better than nothing. • The proposer should propose that he will keep all of the money, bar some minimal amount, because the receiver will accept any offer. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

16. Ultimatum game results, part I • What we typically observe is that • Median and modal offers are 40-50%. • The mean offer is 30-40%. • Offers below 20% are rejected about half the time. • High stakes, reputation and anonymity do not change the results. • Demographic variables have weak effects Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

17. Representative results from Eckel and Grossman (2001) Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

18. Tipping, a real ultimatum game • How much would you tip? • In a restaurant they visit frequently the mean amount suggested was $1.28. • In a restaurant in another city the mean amount suggested was $1.27. (Kahneman, Knetsch and Thaler 1986) Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

19. The dictator game, part I • What we typically observe is that • Around 60% of people give money to the other player. • The mean amount given is around 20% of the endowment. • The amount given is less than in the ultimatum game but still positive. This is despite no threat of rejection. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

20. Different cultures • People all over the world have now been subjected to the ultimatum game! • Cultural differences are significant and range from competitive gift giving, to no sharing in sharing societies. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

21. Story so far Low offers create a negative response. ‘I would rather have nothing than accept such an unfair offer.’ A fear of provoking a negative response can increase offers. ‘I need to give him enough that he will not reject’ Generous offers are made even if no chance of rejection. ‘It seems a bit unfair that I should get everything’ Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

22. Context matters in the ultimatum game • With uncertainty offers are lower and less rejected (but, in scenario 9, offers of $9 were rejected more often than those of $8!) • With discrete choice low offers are rejected less (Falk, Fehr and Fischbacher 2003): Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

23. How about chimps? • Chimpanzees do behave according to Nash equilibrium. They propose an unequal split and it is not rejected (Jensen, Call, Tomasello 2007). Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

24. How about children? • Murninghan and Saxon (1998), Harbaugh, Krause and Vesterlund (2007) looked at behavior in the ultimatum game with uncertainty with children and young adults.

25. Context matters in the ultimatum game 2 • When $1 given equals $3 received we see different behavior (Andreoni and Miller 2002), 30-50% try to max the min and 20-30% try to max the total. • Offers from a third party are rejected less (Blount 1995): Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

26. Context matters in the dictator game • The possibility to take money reduces positive offers. • Earning the money reduces offers (List 2007): Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

27. The data of List (2007). Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

28. Groups propose less • When a group of individuals play the role of proposer and responder in ultimatum and dictator game they typically offer less, and reject less. • Further, interacting in a team lowers individual offers (Luhan, Kocher, Sutter 2009). 1. Obvious discussion, don’t you think? 3. Sure, all for ourselves. 1. I am no good Samaritan. 3. Transfer 0. … 3. No.2, do you share our opinion? 2. I think we should be fair. …. 3. So, for heavens sake, make that 1! Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

29. What we observe • Reciprocity: many people seem to desire reciprocity: ‘If someone does good (or bad) to me then I want to do good (or bad) to them. • Fairness: people care about outcomes, relative to others: ‘Why should I get less than him’, ‘Why should I get more than him?’ • Why is this different to the ‘standard model’. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

30. Fairness and reciprocity matter • In the standard model, utility is a function of consumption, $10 is always as good as $10. • Behavioral economics emphasizes that it matters: • where the $10 comes from; $10 stolen induces guilt and shame while $10 earnt induces pride. • How much others are getting; $10 when others are getting $20 might be annoying but $10 when others are earning $5 might induce guilt. • Note the important interaction between these two effects. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

31. How to model fairness • There are now lots of models to model fairness and reciprocity. We highlight and compare two: • The Fehr-Schmidt model of inequality aversion. This is simple and transparent but ignores the importance of context. • The Rabin model of fairness emphasizes the role of motives but is cumbersome in applications. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

32. Fehr-Schmidt model of inequality aversion • Given an allocation (x1, x2, …, xn) a person’s utility is where 0 ≤ αi < 1 and βi ≤ α i. • So, players feel envy, as given by α i and guilt as given by βi. • This is a very simple model that fits some of the experimental data well. • But, it ignores motives and context. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

33. Rabin model of fairness • The kindness of player A towards player B is given by • The utility of player A is then. • Utility thus depends on kindness given and perceptions of kindness received. • This model does a very good fitting data that is hard to fit (e.g. people are more likely to cooperate if they believe others will cooperate). • But, it is a complicated model to apply. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

34. A neuroscience perspective • Unfair offers A activate different areas of the brain to fair offers B (Sanfey et. al. 2003). • After applying low frequency rTMS to disrupt dorsolateral prefrontal cortex (Knoch et. al. 2006). Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

35. An evolutionary perspective • Much effort has been put in to explaining the evolutionary origins of ‘altruism’. • Much ‘altruism’ can be explained as avoidance of punishment. For example, offers in the ultimatum game reflect the chance of being rejected. • This, however, raises a second order effect where ‘to punish is an altruistic act’. • Evolutionary models show that first order altruism is unlikely to emerge, but second order altruism can, primarily because punishment should be rare. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

36. Applications and policy • Fairness and reciprocity have wide ranging applied and policy consequences. • One area with important consequences is pricing and wage setting. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

37. Fairness and firm pricing • In Scenario 13: 68% would walk to save on the calculator and 29% to save on the jacket (Kahneman and Tversky 1984). • 825 of respondents thought that the hardware store in Scenario 15 was unfair (Kahneman, Knetsch and Thaler 1986).. • Ski resorts should charge high prices at christmas. Tickets for the cup final should be more expensive. …. Thaler (1985) compares some sports events: Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

38. Wage effort games • Similar to the ultimatum game (but more like a game of trust) the Nash equilibrium is that: • The worker never puts in effort. • The employer, therefore, pays a low wage. • A bonus should never be paid so that makes no difference. • A fine should provide a direct incentive for worker effort. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

39. High wages are reciprocated • We observe that high wages induce more effort. • Fines do not work. Bonuses do. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

40. High wages are accepted by employers • If given the choice between hiring a low wage or high wage worker employers may prefer the high wage worker (see also Bewley 2000). Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

41. Fairness in wages matters • In scenario 19, the change in real income is similar but 62% thought a nominal reduction unfair and only 22% a nominal increase unfair (Tversky and Kahneman 1986). • In scenario 20, 62% thought it unfair to cut the wage but 78% acceptable to hire a cheaper worker. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

42. Some conclusions • Fairness and reciprocity are an important factor in price and wage setting. Supply and demand are not context independent. • Some policy thoughts,: • Care is needed measuring collusion, • Wage setting in the public sector is even more important than we might have thought, • People really need to be explained ‘why’ things are as they are. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

43. Topic II: Interpreting new information • Here is a new set of hypothetical scenarios. • How do you think people behave in these scenarios? • Would you expect any biases in judgements? Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

44. The Law of Small Numbers • People exaggerate how closely a small sample will represent the population. • In scenario A, 22% said the large hospital, 56% said no difference, and 22% the small hospital. • In scenarios B and C people are usually reluctant to do a run of the same coin (Rabin 2002, Walker and Wooders 2001) Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

45. A model inference with the law of small numbers - Rabin 2002. • A person observes a sequence of binary signals from some i.i.d. process. • The person believes that they are generated without replacement from an urn with N signals. • The urn is replaced every other period. • The smaller is N the larger the bias. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

46. Consequences of the law of small numbers • We overestimate the importance of small samples. • We underestimate the importance of large samples. (In Scenario D the real likelihood is 1% but the average guess was 10%). • There is regression to the mean. • We read too much into long streaks of success or failure. (There are no hot hands in basketball). • The sequence matters beyond averages. • We tend to think there is more variation in different things then there really is. • If signals are endogenous we may underestimate the rate of success. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

47. Confirmatory Bias • People tend to be too inattentive to new information contradicting their hypothesis: • They can ignore contradictory evidence, and • Misread it as supporting their hypothesis. • In Scenario E, proponents of capital punishment became more in favor and those against less in favor having read the reports. • In Scenario F, those who though she was from a poor background and saw the second video estimated a lower amount of 3.71. Those who thought she was from a well-to-do background estimated higher at 4.67! Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

48. A model of confirmatory bias – Rabin and Schrag 1999 • A person receives a series of signals a or b. • The person perceives each signal as α or β. • After each signal the person updates their belief about the hypothesis. He currently believes in the hypothesis he perceives to have received more signals. • If he currently believes in hypothesis X then he: • Correctly perceives a signal supporting the hypothesis. • With probability q > 0 wrongly perceives a signal against the hypothesis. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

49. Consequences of confirmatory bias • Information contradicting a hypothesis can be ignored or miss-interpreted. This is particularly the case if the information is ambiguous. • Hypothesis based filtering. People can used filtered evidence inappropriately. • A person who has recently changed his mind can be under-confident in a hypothesis. • The confirmatory bias need not be eliminated by increasing information. Edward Cartwright, Behavioral economics, GES Summer School, University of Kent

50. Health Care • Both patient and practitioner must form hypotheses based on constantly changing information. • The law of small numbers and confirmatory bias could cause biased decision making. • The evidence is that we do observe such biases (Frank 2004). Edward Cartwright, Behavioral economics, GES Summer School, University of Kent