Experimental Economics Introduction and Examples

Transcription

1 Experimental Economics Introduction and Examples Dr. Christoph Bühren University of Kassel Nora-Platiel Str Kassel Germany Mail: c.buehren@unikassel.de

2 Traditional assumptions of human behavior in economics Pure rationality common knowledge Self-control Egoism Stable preferences 2

3 Mankiw s Principles, Translated 1. People face tradeoffs 2. The cost of something is what you give up to get it 3. Rational people think at the margin 4. People respond to incentives 5. Trade can make everyone better off 6. Markets are usually a good way to organize economic activity 7. Governments can sometimes improve market outcomes 8. A country s standard of living depends on its ability to produce goods and services 9. Prices rise when the government prints too much money 10. Society faces a short-run tradeoff between inflation and unemployment Translation ( ture=related) 3

4 Experimental economics: principles and implementations Principles: Controlled economic environment Replicability Anonymity No defection Induced Value Theory (Smith 1976) Implementaions: Validation of a theoretical model Hypotheses (derived from theory) Careful selection of participants Randomized allocation of subjects to treatments Ceteris paribus variations in treatments ( KISS ) Pretest Control questions Incentives Detection of stable anomalies to develop new 4 theories

5 Experimental Designs Structure, nature, and amount of incentives Money? Show-up fee? Probabilistic incentives? z-tree vs. paper and pencil Treatments m x n design (factorial design) Within-subject vs. between-subject One-shot vs. repetitions How many repetitions? Last round effect? Strategy method vs. real choice Laboratory- vs. field experiment Stranger vs. partner-treatment Individual vs. group decisions (cf. corruption experiment China/Germany) Loaded vs. neutral framing Single blind vs. double blind Priming? Experimenter effect? 5

6 Data Analysis I Internal validity How good is the fit between my (statistical) model and my data? How good can I explain the variation of my dependent variable by the variations of the independent variables? Convergent validity Do I get the same results with different methods? External validity Can I generalize the findings of my sample? 6

7 Hypotheses testing Data Analysis II Example for methods: Descriptive Statistics: mean, median, mode, quantile, expected value, variance (standard deviation), covariance, skewness, kurtosis, range contingency tables t-test Anova Correlation (Pearson, Spearman, Kendalls Tau) Hit rates Regression analysis Non-parametric tests: Mann-Whitney U test Kolmogorov-Smirnov test Wilcoxon test binomial test Chi-Quadrat Test Kruskal-Wallis Test Software: e.g. STATA, EViews, SPSS, Excel 7

8 The first (?) experiment Daniel Bernoulli (1738): St. Petersburg Paradox 8

9 Preferences/Utility/Willigness to pay Stable vs. constructed preferences Stated vs. revealed preferences Traditional utility function with positive but decreasing marginal utility vs. prospect theory Becker/DeGroot/Marschak (1964): Incentive-aligned mechanism to elicit willingness to pay 9

10 Decision Theory vs. Behavioral Decision Theory Examples: Preference Reversals Allais Paradox Ellsberg Paradox Certainty Effect Decision Heuristics Framing Effect Status Quo Bias Endowment Effect Loss aversion Losses Utility Gains Prospect theory 10

11 Game Theory Prisoners Dilemma Prisoner j confess deny Prisoner i confess -7; -7 0; -10 deny -10; 0-1; -1 Prisoner j confess deny Prisoner i confess -7; -7 ; -10 deny -10; -1; -1 11

12 get out of the way Player B Game Theory Chicken Game go on Player A get out of the way 0; 0-1; 1 go on 1; -1-2; -2 The Chicken Game as a sequential game: Backward Induction! -2; -2 B A 1; -1 B -1; 1 0; 0 12

13 Selten (1978) Chain Store Paradox: Behavioral Game Theory Backward Induction? Entrant Out In Incumbent 20 repetitions Fight Cooperate (0, 10) (-1, -1) (5, 5) 13

14 Behavioral Game Theory Backward Induction?? Rosenthal (1981) Centipede Game: McKelvey/Palfrey (1992); Palacios-Huertay/Volij (2009): Palacios-Huerta/Volij (2009) vs. Bühren/Frank (2010) 14

15 Behavioral Game Theory Bargaining Ultimatum Game Introducing a competitive framing in a mini-ultimatum game: Bühren/Frank/Krabel/Werner (2011) 15

16 Behavioral Game Theory Social Preferences U Fehr/Schmidt (1999): i αi βi = w i max{w j w i, 0} max{w n 1 n 1 i j i j i w j, 0} α = envy" ß = guilt" 0 ß i < 1 ß i < α i [Daruvala (2006): α = 0,95; ß = 0,34] Bolton/Ockenfels (2000) ERC (Equity/Reciprocity/Competition)-preferences: U i = a i w i - b i (w i / w k - 1/n)² Andreoni/Miller (1998/2002); Charness/Rabin(2000/2002) Quasi-Maximin: U i = (1 γ )w + γ[ δ min{ w 0 γ i 1 0 δ i 1 i 1, w 2,..., w n } + (1 δ) n j= 1 w j ] 16

17 Danke für Ihre Aufmerksamkeit 17

18 Utility function with loss aversion (prospect theory: Kahneman/Tversky 1979) Utility Losses Gains

19 Group decision making in a corruption experiment: China and Germany compared Björn Frank, Sha Li, Haiying Qin, Bo Li, Christoph Bühren

20 One of the results: Groups seem to act more corrupt than individuals 4,50 Individual 4,00 3,50 Average bribe 3,00 2,50 2,00 1,50 Deutschland Germany China 1,00 0,50 0,00 Individual Group decision

21 Natural field experiments Psychological Pressure in Competitive Environments Soccer (penalty kicks): Apesteguia/Palacios-Huerta (forth. AER); Kocher/Lenz/Sutter (2010) Basketball (free throws): Feri/Innocenti/Pin (2010) Ice-hockey (penalty shots): Kolev/Pina/Todeschini (2010) Tennis (Grand Slam tournaments): Bühren (2011)

22 Choose a number between 0 and 100! You have won if your number is the closest to 2/3 of the average of the numbers given by all participants: Beauty Contest Among Chess Players Christoph Bühren Björn Frank

23 Choose a number between 0 and 100! You have won if your number is the closest to 2/3 of the average number given by all participants: Beauty Contest Among Chess Players Christoph Bühren Björn Frank 非常感谢, 您能参加我们的调查研究! 这里我们为您提供一个有奖小游戏 : 您的任务非常简单 : 请您选择任意一个大于等于 0 同时小于等于 100 的数字, 这个数字不一定要是整数 如果您选择的数字最接近于由所有参加者所选数字的平均数的三分之二, 那么你就获胜! ( 例如 : 只有 4 位参加者, 由他们所选的数字分别是 11.9;23.45 ;40 和 68.65, 那么这 4 个数的平均数是 36, 而平均数的三分之二则是 24 第二位参加者则获胜 ) 为第一位获胜者 ( 所选数字最接近平均数的三分之二 ) 准备的奖品是一张价值 200 欧元的国际象棋基础 - 赠卷, 第二名和第三名可以分别获得价值 100 欧元和 50 欧元的赠卷

24 1st round results of the experimental beauty contest n Mean Std. Dev. Median Min Max Rating < ,37 23, Rating < ,01 21,68 30, Rating < ,12 21, Rating < ,98 21,73 27, Rating < ,15 20, Rating * 95 30,56 22,47 22, Grandmasters 28 32,96 25,03 26, All ,15 22,22 29, *including Grandmasters Other Experiments: Nagel (1995/Labor): 36,73 Schou (1995/Politiken): 32,4 Thaler (1997/Financial Times): 18,92 Selten/Nagel (1998/Spektrum): 22,08 Rubinstein (2007/Internet): 36,2 Kassel (Seminar.): 33,09; only one opponent: 35,83 Kassel (Colloquium): 22,06

25 Decision-Making in Competitive Framings - Strategic Behavior of Chess Players in Mini-Ultimatum Game Chess Puzzles Christoph Bühren, Björn Frank, Stefan Krabel, and Alexander Werner

26 8 5 0

27 1. Dxd7+ 1. Df7+

28 1.... Kxd Kf Kd Kxf7 2. Lg6# 2. Lg6# 2. Dxf8# 2. Lg6# 5 : 5 0 : 0 8 : 2 0 : 0

29 Results Significant framing effect for the responders: Neutral framing Chess framing Accept Reject Fisher s exact test (two-tailed): p=.002 insignificant framing effect for the proposers: Neutral framing Chess framing Offer 2 13 [3] 29 [13] Offer 5 41 [9] 52 [8] Fisher s exact test (two-tailed): p=.185 [p=.071] [in brackets: subsample of those asked for beliefs about responder rejection rates]