The Role of Framing, Inequity and History in a Corruption Game: Some Experimental Evidence
2. Experimental Methodology
2.1. Experimental Design and Procedures
2.2. Treatments and Research Questions
Firms offered a bribe in 107 out of 170 cases (63%) and the average bribe offered was 6.73 experimental dollars.The bribe was accepted by the official in 82 cases out of 107 (77%).Out of the 82 cases where the firm offered a bribe and the official accepted, the citizen decided to punish in 59 cases (72%) at an average punishment amount of 8.76 experimental dollars.
3.2. Analysis of the Role of Framing, History and Inequity Aversion
3.3. Bribe and Acceptance Behavior
3.4. Punishment Behavior
4. Concluding Remarks
Conflicts of Interest
Appendix A: Instructions for the Loaded Inequitable Treatment
Detailed Instructions for Firms
Detailed Instructions for Officials
Detailed Instructions for Citizens
Appendix B: Instructions for the Neutral Inequitable Treatment
Detailed Instructions for Player A
Detailed Instructions for Player B
Detailed Instructions for Player C
- Cameron, L.; Chaudhuri, A.; Erkal, N.; Gangadharan, L. Propensities to engage in and punish corrupt behavior: Experimental evidence from Australia, India, Indonesia and Singapore. J. Public Econ. 2009, 93, 843–851. [Google Scholar] [CrossRef]
- Alatas, V.; Cameron, L.; Chaudhuri, A.; Erkal, N.; Gangadharan, L. Gender and Corruption: Insights from an Experimental Analysis. South. Econ. J. 2009, 75, 663–680. [Google Scholar]
- Alatas, V.; Cameron, L.; Chaudhuri, A.; Erkal, N.; Gangadharan, L. Subject pool effects in a corruption experiment: A comparison of Indonesian public servants and Indonesian students. Exp. Econ. 2009, 12, 113–132. [Google Scholar] [CrossRef]
- Zizzo, J. Experimenter demand effects in economic experiments. Exp. Econ. 2010, 13, 75–98. [Google Scholar] [CrossRef]
- Abbink, K.; Irlenbusch, B.; Renner, E. An experimental bribery game. J. Law Econ. Organ. 2002, 18, 428–454. [Google Scholar] [CrossRef]
- Azfar, O.; Nelson, W. Transparency, Wages, and the Separation of Powers: An Experimental Analysis of Corruption. Public Choice 2007, 130, 471–493. [Google Scholar] [CrossRef]
- Barr, A.; Lindelow, M.; Serneels, P. Corruption in public service delivery: An experimental analysis. J. Econ. Behav. Organ. 2009, 72, 225–239. [Google Scholar] [CrossRef]
- Jacquemet, N. Corruption as Betrayal: Experimental Evidence; Mimeo, Université Paris 1 Panthéon-Sorbonne: Paris, France, 2012. [Google Scholar]
- Andreoni, J.; Croson, R. Partners versus strangers: Random rematching in public goods experiments. In Handbook of Experimental Economics Results; Plott, C., Smith, V., Eds.; Elsevier: Amsterdam, North Holland, The Netherlands, 2008; Chapter 82; pp. 772–776. [Google Scholar]
- Kreps, D.; Milgrom, P.; Roberts, J.; Wilson, R. Rational cooperation in the finitely repeated prisoners’ dilemma. J. Econ. Theory 1982, 27, 245–252. [Google Scholar] [CrossRef]
- Benoit, J-P.; Krishna, V. Finitely Repeated Games. Econometrica 1985, 53, 905–922. [Google Scholar]
- Abbink, K.; Hennig-Schmidt, H. Neutral versus Loaded Instructions in a Bribery Experiment. Exp. Econ. 2006, 9, 103–121. [Google Scholar] [CrossRef]
- Barr, A.; Serra, D. The effects of externalities and framing on bribery in a petty corruption experiment. Exp. Econ. 2009, 12, 488–503. [Google Scholar] [CrossRef]
- Cooper, D.; Kagel, J. The Impact of Meaningful Context on Strategic Play in Signalling Games. J. Econ. Behav. Organ. 2003, 50, 311–337. [Google Scholar] [CrossRef]
- Fehr, E.; Gächter, S. Cooperation and Punishment in Public Goods Experiments. Am. Econ. Rev. 2000, 90, 980–994. [Google Scholar] [CrossRef]
- Fehr, E.; Gächter, S. Altruistic Punishment in Humans. Nature 2002, 415, 137–140. [Google Scholar] [CrossRef] [PubMed]
- Di Tella, R.; Schargrodsky, E. The role of wages and auditing during a crackdown on corruption in the city of Buenos Aires. J. Law Econ. 2003, 46, 269–292. [Google Scholar] [CrossRef]
- Rauch, J.; Evans, P. Bureaucratic structure and bureaucratic performance in less developed countries. J. Public Econ. 2000, 75, 49–71. [Google Scholar] [CrossRef]
- Treisman, D. The causes of corruption: A cross-national study. J. Public Econ. 2000, 76, 399–457. [Google Scholar] [CrossRef]
- Van Rijckeghem, C.; Weder, B. Bureaucratic corruption and the rate of temptation: Do wages in the civil service affect corruption, and by how much? J. Dev. Econ. 2001, 65, 307–331. [Google Scholar] [CrossRef]
- Svensson, J. Eight questions about corruption. J. Econ. Perspect. 2005, 19, 19–42. [Google Scholar] [CrossRef]
- Van Veldhuizen, R. The influence of wages on public officials’ corruptibility: A laboratory investigation. J. Econ. Psychol. 2013, 39, 341–356. [Google Scholar] [CrossRef]
- Armantier, O.; Boly, A. Comparing corruption in the laboratory and in the field in Burkina Faso and Canada. Econ. J. 2013, 123, 1168–1187. [Google Scholar] [CrossRef]
- Jiang, T.; Lindemans, J.W.; Bicchieri, C. Can trust facilitate bribery? Experimental evidence from China, Italy, Japan and the Netherlands. Soc. Cognit. 2015, 33, 483–504. [Google Scholar] [CrossRef]
- Schotter, A.; Sopher, B. Social Learning and Coordination Conventions in Intergenerational Games: An Experimental Study. J. Political Econ. 2003, 111, 498–529. [Google Scholar] [CrossRef]
- Chaudhuri, A.; Schotter, A.; Sopher, B. Learning in Tournaments with Inter-generational Advice. Econ. Bull. 2006, 3, 1–16. [Google Scholar]
- Chaudhuri, A.; Schotter, A.; Sopher, B. Coordination in inter-generational minimum games with private, almost common and common knowledge of advice. Econ. J. 2009, 75, 91–122. [Google Scholar] [CrossRef]
- Chaudhuri, A.; Graziano, S.; Maitra, P. Social Learning and Norms in an Experimental Public Goods Game with Inter-Generational Advice. Rev. Econ. Stud. 2006, 73, 357–380. [Google Scholar] [CrossRef]
- Schotter, A.; Sopher, B. Advice, Trust and Trustworthiness in an Experimental Intergenerational Game. Exp. Econ. 2006, 9, 123–145. [Google Scholar] [CrossRef]
- Schotter, A.; Sopher, B. Advice and Behavior in Intergenerational Ultimatum Games: An Experimental Approach. Games Econ. Behav. 2007, 58, 365–393. [Google Scholar] [CrossRef]
- Schotter, A. Decision making with naïve advice. Am. Econ. Rev. Pap. Proc. 2003, 93, 196–201. [Google Scholar] [CrossRef]
- Banerjee, R. Corruption, Norm Violation and Decay in Social Capital. J. Public Econ. 2016, 137, 14–27. [Google Scholar] [CrossRef]
- Trivedi, V.; Shehata, M.; Lynn, B. Impact of Personal and Situational Factors on Taxpayer Compliance: An Experimental Analysis. J. Bus. Eth. 2003, 47, 175–197. [Google Scholar] [CrossRef]
- Cadsby, C.B.; Maynes, E.; Trivedi, V.U. Tax compliance and obedience to authority at home and in the lab: A new experimental approach. Exp. Econ. 2006, 9, 343–359. [Google Scholar] [CrossRef]
- Abbink, K. Laboratory experiments on corruption. In International Handbook on the Economics of Corruption; Rose-Ackerman, S., Ed.; Edward Elgar Publishers: Cheltenham, UK, 2006; Chapter 14; pp. 418–437. [Google Scholar]
- Serra, D.; Wantchekon, L. (Eds.) Research in Experimental Economics Volume 15: New Advances in Experimental Research on Corruption; Emerald Group Publishing: Bingley, UK, 2012.
- Banuri, S.; Eckel, C. Experiments in culture and corruption: A review. In Research in Experimental Economics Volume 15: New Advances in Experimental Research on Corruption; Serra, D., Wantchekon, L., Eds.; Emerald Group Publishing: Bingley, UK, 2012; Chapter 3; pp. 51–76. [Google Scholar]
- Chaudhuri, A. Gender and corruption: A survey of the experimental evidence. In Research in Experimental Economics Volume 15: New Advances in Experimental Research on Corruption; Serra, D., Wantchekon, L., Eds.; Emerald Group Publishing: Bingley, UK, 2012; Chapter 2; pp. 13–49. [Google Scholar]
- Cameron, C.; Miller, D. A Practitioner’s Guide to Cluster-Robust Inference. J. Hum. Resour. 2015, 50, 317–373. [Google Scholar] [CrossRef]
- Cragg, J.G. Some Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods. Econometrica 1971, 39, 829–844. [Google Scholar] [CrossRef]
- Cameron, A.C.; Trivedi, P.K. Microeconometrics Methods and Applications; Cambridge University Press: Cambridge, UK, 2005. [Google Scholar]
- Wooldridge, J. Econometric Analysis of Cross Section and Panel Data; MIT Press: Cambridge, MA, USA, 2002. [Google Scholar]
- Chaudhuri, A.; Paichayontvijit, T.; Li, Y. Context, Common Knowledge, Trust and Reciprocity; Working Paper; Department of Economics, University of Auckland: Auckland, New Zealand, 2016. [Google Scholar]
- Samuelson, L. Economic Theory and Experimental Economics. J. Econ. Lit. 2005, 43, 65–107. [Google Scholar] [CrossRef]
- Levati, M.; Miettinen, T.; Rai, B. Context and interpretation in laboratory experiments: The case of reciprocity. J. Econ. Psychol. 2011, 32, 846–856. [Google Scholar] [CrossRef]
- 1Alatas et al. (2009a)  use the same game, undertaken across the four different locations, to look for gender differences in behavior. Alatas et al. (2009b)  look at differences in behavior in this game between Indonesian students and Indonesian civil servants. These two papers are not immediately relevant to our study and hence we refrain from elaborating on them.
- 2We chose the random re-matching protocol in our repeated game set-up in order to preserve the essence of one-shot interactions while allowing for learning and gathering experience. Andreoni and Croson (2008) , make this point, albeit in the context of public goods games, by stating: “A common way to deal with this has been to rematch subjects randomly into groups for each iteration of the game, hence forming a repeated single-shot design and avoiding the repeated-game effects.” Further, as Kreps, Milgrom, Roberts and Wilson (1982)  and Benoit and Krishna (1985)  note, the subgame perfect equilibrium in the one-shot stage game may not necessarily be an equilibrium under fixed matching even with finitely repeated play.
- 3This is the same conversation rate as in Cameron et al. (2009). At the time the experiments were run in 2007 NZ $1 was roughly equivalent to US $0.77.
- 4In the most recent CPI, published in 2015, New Zealand is 4th while Australia is 13th. New Zealand was ranked second in each of the previous three years. Australia has typically ranked somewhere around 10th in each of those years.
- 6It is important to note the following. For firms, taking this average is easy since all we need to do is to look at whether a particular firm offered a bribe or not in each of the ten rounds. However, for each official, we average over acceptance decisions only in those cases when a bribe was offered and the official had a decision to make: whether to accept or reject; similarly, for citizens we look at only those cases, when the bribe was offered and accepted and the citizen actually had a decision to make. This implies that, while we have the same numbers of firms, officials and citizens in each treatment, for the officials and the citizens the actual number of observations over which we are averaging will differ from one subject to another.
- 7Compared to the Loaded inequitable treatment, there is greater inequity between the firm and the official in the Loaded equitable treatment. It is possible, along the lines of the results reported by Jiang et al. (2015) , that the changing inequity in the payoffs of the firms and the officials also played a role in the firm’s decision to offer a bribe or not. We thank an anonymous referee for pointing this out.
- 8We note that these observations are not independent and this is likely contributing to the high significance levels; but the general point is still valid. Given that the earnings from not bribing is constant, we use a t-test here.
- 9As a robustness check, we re-ran our regressions with standard errors clustered at the level of individual subjects. The results are similar. We report non-clustered standard errors because Cameron and Miller (2015)  argue that in order to obtain precise estimates with clustering one requires a large number of clusters and fixed size in each cluster. This is not really true in our case. We do not have many clusters and our cluster sizes vary. Cameron and Miller (2015) argue that performance can deteriorate significantly with a relatively small number of clusters and/or where the cluster size is not fixed.
- 10The results for , , and estimated coefficients are not reported in the tables.
- 11In terms of the interaction between treatments and round, we note that the variables are not demeaned. Given that the regression includes both main treatment effects and treatment*round interactions, the main treatment effects reflect the treatment difference in a hypothetical round 5.5.
- 12Taking logs generates more precise results. The results are similar if we take levels; the only difference without logs is that we have less significant estimated effects. The joint log likelihood for the two-part model with a random effects maximum likelihood regression on a log-normal model presented in Table 4 is −125.72, the joint log likelihood for the two-part model with a random effects maximum likelihood regression on the actual bribe amount is −1029.44, and the joint log likelihood for two-part model with a random effects Tobit model is −897.46. This shows that the model we presented in Table 4 provides the best fit for our data.
- 13If a firm does not offer a bribe in round t-1, the value for lag ln(bribe amount) is replaced with the log-normal of the bribe amount in the most recent round before that. An anonymous referee correctly pointed out that this is a strong assumption and suggested that this may affect our estimates; e.g., by increasing the distance between the lagged bribe amount and the bribe amount for firms that are less likely to bribe at all. We thank the referee for raising this point. But, we could not find or think of another way to deal with missing observations. We had to choose a particular way of addressing this and to us, it made sense, to insert the last value experienced by an individual as the lagged value.
- 14Likelihood ratio test shows this specification is better than one without additional interaction terms between treatment dummies and bribe amount ( (3) = 10.28, p = 0.0163).
- 15Likelihood ratio test shows this specification is better than one with additional interaction terms between treatment dummies and bribe amount ( (3) = 4.34, p = 0.2275).
- 16The joint log likelihood for the two-part model with a random effects maximum likelihood regression on a log-normal model presented in Table 4 is −338.20, the joint log likelihood for two-part model with a random effects maximum likelihood regression on the actual bribe amount is −730.72, and the joint log likelihood for two-part model with a random effects Tobit model is −637.12. This shows that the model we present in Table 4 provides the best fit for our data.
|Treatments||Abbreviation||Number of Subjects (Number of F-O-C Triplets)||Exchange Rate||Number of Plays of the Stage Game|
(Control treatment) Loaded language with different conversion rates
|Loaded inequitable||51 (17)||Firm: Exp$60 = NZ$1 |
Official: Exp$40 = NZ$1
Citizen: Exp$30 = NZ$1
|Treatment 2 |
Neutral language with different conversion rates
|Neutral inequitable||60 (20)||Firm: Exp$60 = NZ$1 |
Official: Exp$40 = NZ$1
Citizen: Exp$30 = NZ$1
|Treatment 3 |
Loaded language with uniform conversion rate
|Loaded equitable||42 (14)||Firm, Official, Citizen: Exp$40 = NZ$1||140|
|Treatment 4 |
Loaded language with different conversion rates and history
|Loaded inequitable plus history||57 (19)||Firm: Exp$60 = NZ$1 |
Official: Exp$40 = NZ$1
Citizen: Exp$30 = NZ$1
|One-Shot Game in Cameron et al. 5||This Study Round 1 only||This Study Average over 10 Rounds|
|% of firms bribing||87.8||83.6||93.3||78.3||58.8||62.9|
|% of officials accepting||88.9||96.1||92.9||78.7||60.0||76.6|
|% of citizens punishing||42.2||57.1||21.2||59.5||66.7||72.0|
|Panel 1: Impact of Framing|
|Loaded inequitable (n = 17)||Neutral inequitable (n = 20)||p-Value|
|% of firms bribing||62.9||98||0.0001|
|Bribe amount if chose to bribe||6.67||6.85||0.71|
|% of officials accepting||78||90||0.10|
|% of citizen punishing||73||55||0.14|
|Punishment amount if chose to punish||8.5||6.7||0.08|
|Panel 2: Impact of Inequity in Initial Endowments|
|Loaded inequitable (n = 17)||Loaded equitable (n = 14)||p-Value|
|% of firms bribing||62.9||79||0.22|
|Bribe amount if chose to bribe||6.67||6.20||0.16|
|% of officials accepting||78||72||0.18|
|% of citizen punishing||73||71.2||0.76|
|Punishment amount if chose to punish||8.5||7.8||0.35|
|Panel 3: Impact of History of Past Plays|
|Loaded Inequitable (n = 17)||Loaded Inequitable Plus History (n = 19)||p-Value|
|% of firms bribing||62.9||77.9||0.23|
|Bribe amount if chose to bribe||6.67||6.23||0.37|
|% of officials accepting||78||76.6||0.95|
|% of citizen punishing||73||73.3||0.98|
|Punishment amount if chose to punish||8.5||7.3||0.31|
|(1) Bribe (0/1)||(2) Log-Normal Bribe Amount||(3) Accept (0/1)||(4) Punish (0/1)||(5) Log-Normal Punishment Amount|
|Neutral inequitable dummy||4.19 ** (1.86)||−0.02 (0.07)||−1.67 (1.26)||−1.91 * (1.11)||−0.32 (0.27)|
|Loaded inequitable plus history dummy||0.19 (0.62)||0.02 (0.08)||−0.40 (1.17)||−0.27 (1.19.)||−0.14 (0.30)|
|Loaded equitable dummy||1.45 * (0.76)||−0.07 (0.08)||−2.66 ** (1.28)||−1.06 (1.17)||−0.15 (0.32)|
|Round||−0.07 (0.05)||0.01 (0.01)||0.04 (0.07)||−0.26 ** (0.13)||−0.03 (0.03)|
|Neutral inequitable * Round||−0.25 (0.21)||0.001 (0.01)||−0.13 (0.09)||0.23 * (0.14)||0.05 (0.04)|
|Loaded inequitable plus history * Round||0.03 (0.07)||−0.01 (0.01)||−0.16 * (0.09)||0.02 (0.15)||0.04 (0.05)|
|Loaded equitable * Round||−0.14 (0.09)||0.01 (0.01)||−0.04 (0.11)||0.17 (0.15)||0.04 (0.05)|
|Lag bribe (0/1)||−0.72 *** (0.27)||−0.003 (0.04)||---||---||---|
|Lag bribe with acceptance (0/1)||1.00 *** (0.33)||0.04 (0.03)||---||---||---|
|Lag bribe with punishment (0/1)||−0.71 ** (0.30)||−0.03 (0.02)||---||---||---|
|Lag ln(bribe amount)||---||0.20 *** (0.05)||---||---||---|
|Bribe amount||---||---||0.02 (0.12)||0.27 *** (0.07)||0.05 * (0.03)|
|Neutral inequitable * Bribe amount||---||---||0.42 ** (0.17)||---||---|
|Loaded inequitable plus history * Bribe amount||---||---||0.18 (0.15)||---||---|
|Loaded equitable * Bribe amount||---||---||0.50 *** (0.19)||---||---|
|Lag acceptance (0/1)||---||---||−0.005 (0.26)||---||---|
|Lag acceptance with punishment (0/1)||---||---||−0.36 * (0.21)||---||---|
|Lag Punishment (0/1)||---||---||---||0.25 (0.24)||---|
|Lag ln(punishment amount)||---||---||---||---||0.37 *** (0.06)|
|Constant||0.69 (0.53)||1.17 *** (0.14)||1.93 (1.37)||3.50 * (2.01)||0.37 (0.43)|
© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
Chaudhuri, A.; Paichayontvijit, T.; Sbai, E. The Role of Framing, Inequity and History in a Corruption Game: Some Experimental Evidence. Games 2016, 7, 13. https://doi.org/10.3390/g7020013
Chaudhuri A, Paichayontvijit T, Sbai E. The Role of Framing, Inequity and History in a Corruption Game: Some Experimental Evidence. Games. 2016; 7(2):13. https://doi.org/10.3390/g7020013Chicago/Turabian Style
Chaudhuri, Ananish, Tirnud Paichayontvijit, and Erwann Sbai. 2016. "The Role of Framing, Inequity and History in a Corruption Game: Some Experimental Evidence" Games 7, no. 2: 13. https://doi.org/10.3390/g7020013