Crawford School Seminar Experimental evaluation: Conditions for citizens punishment of government corruption Associate Professor Fiona Yap Crawford School of Public Policy, ANU fiona.yap@anu.edu.au Tuesday 7 August 2012 Research partly funded by the ARC Visiting Scholar award and Keeler IntraUniversity Professorship, University of Kansas, 2011-2012
Study of Corruption in Government Definition: Misuse of public office for private gain Interest: Relation to governance capacity (criminology, justice, law and society) State-building or rebuilding (political development, stability in emerging democracies)
Roadmap Background: from definition to institutions Citizens & institution-building Why experiment? The experimental set-up Results and findings Implications and extensions
Background: The Literature Defining corruption: public versus private (government v business) Public (Susan Rose-Ackerman 1999) (a) bribery, (b) extortion, (c) fraud, (d)embezzlement, (e) nepotism, (f) cronyism, (g) appropriation of public assets and property for private use, and (h) influence peddling
Corruption: The Focus on Institutions Study of public corruption: Individual ethics Culture Economic loss (Kaufmann; Rose-Ackerman; Klitgaard; Schleifer & Vishny; Goudie & Stasavage) Corruption formula: (Klitgaard 1998) Corruption = Monopoly + Discretion Accountability
Institution-Building and Corruption Kaufmann s Worldwide Governance Indicators (Worldbank Indicators): Voice and Accountability Political Stability and Absence of Violence Government Effectiveness Regulatory Quality Rule of Law Control of Corruption
Citizens & Institution-building Citizens participation: important corollary to institutional/political development Governments even in less-democratic countries accede to citizens demands in order to stay in office Strategic interaction between citizens and government (contrasts to decision-making models) (Haggard 2004; Mason & Clements 2002; Gang 2007; Robinson 2006; Howard and Roessler 2006; Gandhi 2008; Yap 2005a, 2005b, 2010, 2011, 2012, 2013)
Citizens Role in Corruption Control The necessary political will may still be generated albeit more slowly by efforts to enhance public awareness and mobilize constituencies of civil society (Grey & Kaufmann 1998) 1) Punishment critical in curbing corruption (Azfar and Nelson 2007; Cameron et al 2009; Fisman and Miguel 2007; Abbink et al 2002) 2) However, citizens demand for punishment inconsistent
Theory & Prediction of Citizens Participation Two competing outcomes: 1) Prisoner s dilemma: 2 individuals, both gain if they do not confess to crime. If one confesses before the other: confessor gains, the other loses substantially If both confess, however, both lose. Expected outcome: both confess because of benefits of confession Conflict between individual rationality and
Theory & Prediction Two competing political economy theories: 2) Stag hunt: 2 individuals, both gain if they hunt stag instead of hare Both gain if hunt stag If both hunt stag, succeed; if hunt hare, substantially smaller gains Expected outcome: both hunt stag for strong benefits Consideration is between mutual benefit and
Assessing Conditions for Citizens Demand of Punishment Why experiments: Makes up for lack of variation, controls or substantive overlap. Consider: The necessary political will may still be generated albeit more slowly by efforts to enhance public awareness and mobilize constituencies of civil society (Grey & Kaufmann 1998) Necessary variants: Awareness; Mobilization of constituencies
Hypothesis: The Experiment Public awareness leads to mobilization Questions: Do citizens make the initial thrust to challenge if there is public awareness? Is this based on expectation that the others will challenge? (mobilization) Video shown to public comprising groups Survey questions to evaluate responses based on awareness and expectations of what groups will do
Cases United States, Australia as baselines Extensions to Singapore, Taiwan, and South Korea Why these cases? Consider their Transparency Int l Corruption Index: Singapore (9.2) & Australia (8.8) Taiwan, South Korea (6.1, 5.4) US (7.1)
Results and Findings, US Table 1. Response based on expectations of one other group Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 16 (8.5) 7 (14.5) 23 Expect 50% or more from own group to challenge 5 (12.5) 29 (21.5) 34 Total 21 36 57 Pearson Chi-square= 17.74 (pr=0.00) Fisher s exact, probability =0.000
Results and Findings, US Table 2. Response based on expectations of one other group, controlling for personal loss Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 5 (2.2) 1 (3.8) 6 Expect 50% or more of own group to challenge 4 (6.8) 14 (11.2) 18 Total 9 15 24 Pearson Chi-square= 7.17 (pr=0.007) Fisher s exact, probability =0.015
Results and Findings, US Table 3. Response based on expectations of one other group, controlling for no loss or some gain Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 11 (6.2) 6 (10.8) 17 Expect 50% or more of own group to challenge 1 (5.8) 15 (10.2) 16 Total 12 21 33 Pearson Chi-square= 12.17 (pr=0.000) Fisher s exact, probability =0.001
Results and Findings, Australia Table 4. Response based on expectations of one other group Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 30 (22.0) 17 (25.0) 47 Expect 50% or more from own group to challenge 27 (35.0) 48 (40.0) 75 Total 21 36 122 Pearson Chi-square= 8.99 (pr=0.003) Fisher s exact, probability =0.003
Results and Findings, Australia Table 5. Response based on expectation of other group, controlling for personal loss Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 6 (3.3) 1 (3.7) 7 Expect 50% or more of own group to challenge 23 (25.7) 32 (29.3) 55 Total 29 33 62 Pearson Chi-square = 4.81 (pr=0.03) Fisher s exact, probability = 0.04
Results and Findings, Australia Table 6. Response based on expectations of other group, controlling for no loss or some gain Expect less than 50% of other group to challenge Expect 50% or more from other group to challenge Total Expect less than 50% of own group to challenge 24 (18.7) 16 (21.3) 40 Expect 50% or more from own group to challenge 4 (9.3) 16 (10.7) 20 Total 28 32 60 Pearson Chi-square= 8.57 (pr=0.003) Fisher s exact, probability =0.006
Table 6. Logit regression of likelihood to challenge Dependent Variable: Likelihood to challenge (0=no challenge; 1=challenge) US (standard errors) Australia (standard errors) Perception of personal loss or gain (-1=loss; 0=no loss or some gain) Perception of other group s voucher loss or gain Expectation that other group will challenge -5.49 ** (2.07) 1.0 (0.86) 4.29*** (1.53) -3.50 *** (0.62) 0.75 ** (0.34) -0.02 (0.01) Cite 6 3.98** (1.50) 2.62** (0.71) Gender (Male=0, Female=1) 0.38 (0.91) 0.81 (0.58) Age (Categories) -2.55 (1.60) 0.48 (0.69) Marital status (categories) -0.77 (3.80) -0.02 (0.45) Income (categories) -0.01 (0.03) 0.02 (0.03) LR Chi 2 (probability) 45.34*** (0.001) 74.84*** (0.001) Pseudo R 2 N 0.59 56 0.44 126 * p < 0.05 p < 0.01 *** p < 0.005
Conclusions & Implications Experimental study to evaluate conditions that motivate citizens punishment of corruption in government. Are citizens willing to make the initial thrust to challenge if there is public awareness?. Is this based on expectation that the others will challenge? (mobilization) Hypothesis: Public awareness leads to mobilization
Conclusions & Implications Theories & Predictions: prisoner s dilemma v stag-hunt Results: a) citizens willing to incur costs to punish or demand punishment of misconduct/corruption in government officials. b) Highly correlated with expectations that other groups will mobilize c) not dependent on personal losses/gains
Extensions Beyond the US & Australia to East & Southeast Asia (evaluate the conventional wisdom that (government) corruption is tolerated) Additional controls in experimental design: a) information b) expectation of penalty c) economy (poor versus strong) Does citizens influence enhance confidence in institutions & government?