Corruption, Social Norms and Decay in Social Capital

Size: px

Start display at page:

Download "Corruption, Social Norms and Decay in Social Capital"

Kristopher Golden
6 years ago
Views:

1 Corruption, Social Norms and Decay in Social Capital Ritwik Banerjee Aarhus University, Denmark December 2014 Banerjee Corruption, Norms, Trust Nov, / 28

2 Motivation 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

3 Motivation Motivation Behavioral Economics as an antithesis to the Rational Actor Model. altruism, guilt aversion, positive self image, social norm Social Appropriateness Norm shared understandings about actions that are obligatory, permitted, or forbidden Ostrom (2000) socially held beliefs about what is morally the right thing to do Krupka and Weber (2013) introduces a a coordination game to measure social appropriateness norm of an action. Understand the link between corruption and trust. Banerjee Corruption, Norms, Trust Nov, / 28

4 NOR SWE Trust Corruption WVS Wave 5 Within country gradient Between country gradient Trust FIN NZL CHE AUS NLD CAN HKG USA DEU CHL GBR ESP CYP FRA JPN VNM CHN IDN THA JOR ITA KOR URY UKR RUS IND ETH BGR ROM SVN POL ARG ZAF MLI EGY MDA COL MEX BFA GTM MAR ZMB IRN MYS BRA GHA PER TUR Corruption IRQ

5 Motivation Motivation Transition from high trust - low corruption to low trust - high corruption. The cross sectional correlation is negative for wave 5. Corruption and trust are negatively related, over time and space. Political Economy literature Lack of trust leads to corruption La Porta et al. (1997); Bjornskov (2011), Xin et al. (2004) Corruption leads to lack of trust Bowler, S. et al. (2004), Anderson et al. (2003); Chang et al. (2006); Della Porta (2000) Mutually reinforcing causality Uslaner (2002), Morris et al. (2010) Exogeneity of the instruments used in dynamic macro panel not aboveboard Banerjee Corruption, Norms, Trust Nov, / 28

6 Motivation Motivation Experimentally tease out causal connection though behavioral spillovers if observed behavior in a game di ers when played together with another game as opposed to when played in isolation The interpretation of the World Value Survey trust question Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Does measure trust Fehr et al. (2003); Bellemare and Kroger (2007) Does not measure trust Glaeser et al. (2000); Lazzarini et al. (2004) Expectations about trustworthiness Sapienza et al. (2013) We re-examine this issue in a cultural context which is di erent from that is standard in literature analyze under what condition the relation may or may not be valid. Banerjee Corruption, Norms, Trust Nov, / 28

7 Motivation Overview Causal link from corruption to trust is established through behavioral spillovers Bribery Game and a strategically identical Ultimatum Game followed by a Trust Game and vice versa. Theoretical framework: Posterior beliefs of bayesian agents depend on whether they meet a good or a bad guy. Mechanism: Social appropriateness violation æ negative expectation shock æ trust deficit. WVS-trust does measure expectations about trustworthiness in the baseline and in the ultimatum game but not in the bribery game. Does not capture the expectation shock. Banerjee Corruption, Norms, Trust Nov, / 28

8 Experimental Design Design Outline 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

9 Experimental Design Design Intuition You are traveling from New Delhi railway station in Rajdhani Express. Someone picks your wallet as you are about to board the train. Choose one of the statements. 1 You will be much more careful with your belongings for the rest of the journey in Rajdhani Express. 2 There will be no change in your alertness level. Reason? Behavioral Spillovers!! Banerjee Corruption, Norms, Trust Nov, / 28

Experimental Design Design Design Random assignment to Ultimatum Game or the Bribery Game followed by baseline trust Randomized to roles of Citizens or Public O

10 Experimental Design Design Design Random assignment to Ultimatum Game or the Bribery Game followed by baseline trust Randomized to roles of Citizens or Public O Participant A or Participant B in UG cials in BG and Citizens and Participant As performed a real e ort task to win a trophy code Banerjee Corruption, Norms, Trust Nov, / 28

11 Experimental Design Design Part I : Bribery Game Citizen performs a task, needs to complete it in 10 minutes Unable to complete the task C: 200 PO: 600 Successfuly completes the task, C wins a prize of 400 PO decides whether to ask for a bribe and if yes how much. b={0,100,200,300,400} C decides whether to accept or reject the demand for bribe Accept Reject C: b PO: 600+b C: 200 PO: 600 Social Norm Banerjee Corruption, Norms, Trust Nov, / 28

12 Experimental Design Design Part I : Ultimatum Game Participant A performs a task, needs to complete it in 10 minutes Unable to complete the task P-A: 200 P-B: 600 Successfuly completes the task P-A qualifies for the next part of the game where P-B decides how much to share with P-A x={0,100,200,300,400} Accept Reject P-A decides whether to accept or reject the offer P-A: x P-B: 600+x P-A: 200 P-B: 600 Social Norm Banerjee Corruption, Norms, Trust Nov, / 28

13 Experimental Design Design Part II : Trust Game In our game the sender is endowed with M400. She then decides how much to send, t œ{0, 50, 100,...400}, toa receiver. Any amount sent is multiplied by 3. The receiver decides how much to return, w œ{0, 50, 100,...3t}, tothe sender. As a result, the payo of a sender equals M(400 t + w) whereas that of the receiver equals M(3t w). The predominant interpretation in the literature is that t is a measure of trust, while w is a measure of trustworthiness. Subjects play only one role in BG/UG but make decisions both as a sender and a receiver in Trust Game. Banerjee Corruption, Norms, Trust Nov, / 28

14 Experimental Design Design Additional Data Social Norm incentivized social norm elicitation tool (Krupka and Weber, 2013) Subjects asked to rate actions of a Public O cial in BG and that of Participant A in UG on a social appropriateness scale from +3 to -3. Modal Response was rewarded. World Value Survey - trust question Risk Attitude How willing are you to take risks, in general? studied by Dohmen et al. (2011) Altruism: Suppose you win a lottery of Rs How much of it will you be willing to share with a charity of your choice? Demographics Banerjee Corruption, Norms, Trust Nov, / 28

15 Result Behavior in BG and UG Outline 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

16 Result Behavior in BG and UG BG and UG Treatment UG BG Bribe/Ultim Amount Amount Acceptance Percentage UG BG Bribe/Ultim Amount UG BG 80% demanded bribe and 100% demanded ultim ( 2 -test, p < 0.001) Mean bribe lower than mean ultim for full sample (t-test, p < 0.001) and corrupt sample Di erence in Acceptance rates statistically significant at bribe M200 and M300 Banerjee Corruption, Norms, Trust Nov, / 28

17 Result Behavior in BG and UG Social appropriateness of BG and UG Mean Norm Rating/Difference Amount Frequency/Difference Amount Difference BG Difference BG UG UG Social Norm governing UG and BG very di erent. Correlation between the di erence in social appropriateness measure and di erence in actual behavior is Banerjee Corruption, Norms, Trust Nov, / 28

18 Result Trust Behavior following BG and UG Outline 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

19 Result Trust Behavior following BG and UG Trust Game Frequency Amount shared UG BG Amount UG BG Base Amount UG BG Base No statistical di erence in zeroes of trust decision Mean trust is greater in UG than in BG in full sample (t-test, p-value=0.04) and corrupt sample (t-test, p-value=0.08). Banerjee Corruption, Norms, Trust Nov, / 28

20 Result Trust Behavior following BG and UG Expected Return on Trust and Return Ratio Expected Return on Trust Hypothetical Amount Sent Return Ratio Amount Sent by the Sender BG Baseline UG BG UG Baseline Definitions Expected Return on Trust = Return Ratio = Expected Return for a Trust level - Trust level Trust Level Amount Sent Amount Received Back Banerjee Corruption, Norms, Trust Nov, / 28

21 Result Mechanism Outline 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

Assume that an honest person is more likely to be trustworthy a corrupt person is less likely to

22 Result Mechanism Theoretical Framework Citizens meet good/honest guys or bad/corrupt guys in the first stage. Assume that an honest person is more likely to be trustworthy a corrupt person is less likely to be trustworthy a citizen updates belief by observing examples of public o behavior citizen subsequently takes the trust decisions cial s Banerjee Corruption, Norms, Trust Nov, / 28

23 Result Mechanism Structural Equations Predictions If a citizen meets a corrupt (honest) public o. cial, then her belief about a matched partner s trustworthiness decreases(increases) her trust level decreases (increases) Linear approximation of trust decision is t ú = µ µ Á µ = 1(H)+ The structural equations to be estimated are given by TRUST = 0 + 1E( ERoT BG, NORM)+ 2X Á ERoT = 0 + 1BG + 2NORM + Banerjee Corruption, Norms, Trust Nov, / 28

24 Result Mechanism Joint Estimation Table: Joint Estimation - Dependent Variable: Trust (1) (2) (3) ERoT * * (223.93) (238.93) (286.23) age (6.99) (7.00) female ** * (29.82) (31.30) catscore (3.51) (3.50) fam_income (19.29) (19.41) altruism 0.03 (0.18) risk 17.64** (7.03) Constant *** ** * (17.56) (363.05) (358.56) First Stage bg -0.13* -0.12* -0.11* (0.07) (0.07) (0.07) normmean 0.03* 0.03* 0.03* (0.02) (0.02) (0.02) Constant Yes Yes Yes Observations Banerjee Corruption, Norms, Trust Nov, / 28

25 Result Mechanism Joint Estimation (contd.) First Stage The priming e ect of the bribery game treatment negatively a ects the expectations. Expectations increase with Social Appropriateness of the demand ERoT then a ects the trust decision Coe cient for ERoT is positive and significant in col (1) and has p-value=0.10 in col (2) Is there an additional priming e ect in the second stage? No Trust decisions have no e ect on the subsequent bribe demand decision. Banerjee Corruption, Norms, Trust Nov, / 28

26 Result Interpretation of World Value Survey - Trust Outline 1 Motivation 2 Experimental Design Design 3 Result Behavior in BG and UG Trust Behavior following BG and UG Mechanism Interpretation of World Value Survey - Trust 4 Conclusion Banerjee Corruption, Norms, Trust Nov, / 28

27 Result Interpretation of World Value Survey - Trust WVS-trust and ERoT in C, P-A and Baseline Citizens, P As, Baseline Participant As, Baseline Coefficient of ERoT Hypothetical Trust Level Coefficient of ERoT Hypothetical Trust Level Coefficient of ERoT Participant As Hypothetical Trust Level Coefficient of ERoT Citizens Hypothetical Trust Level Banerjee Corruption, Norms, Trust Nov, / 28

28 Conclusion Concluding Remark Causal spillover from corruption to trust but not from trust to corruption. Mechanism - Norm Violation leads to a negative expectation shock (surrounded by norm violators) Reduced ERoT drives down trust levels WVS-trust measures expectations about trustworthiness in a di erent cultural context WVS-trust is not related to trust in trust games Implication: WVS-trust cannot be used to measure trust the way risk question is used to measure risk. WVS-trust measures expectations about others trustworthiness. However, response to WVS-trust comes from a global reference point but expectations about others trustworthiness is locally Banerjee Corruption, Norms, Trust Nov, / 28

Housing markets, wealth and the business cycle

Housing markets, wealth and the business cycle Nathalie Girouard copyright with the author OECD Economics Department DG ECFIN workshop: Housing and mortgage markets and the EU economy Brussels, 21 November