Journal of Public Economics 111 (2014) 17–28

Contents lists available at ScienceDirect

Journal of Public Economics

journal homepage: www.elsevier.com/locate/jpube

Letting the briber go free: An experiment on mitigating

harassment bribes☆
Klaus Abbink a, Utteeyo Dasgupta b, Lata Gangadharan a, Tarun Jain c,⁎
a
Department of Economics, Monash University, Australia
b
Department of Economics, Franklin and Marshall College, USA
c
Finance, Economics and Public Policy Area, Indian School of Business, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper examines the effectiveness of using asymmetric liability to combat harassment bribes. Asymmetric
Received 18 January 2013 liability is a mechanism where bribe-takers are culpable but bribe-givers have legal immunity. Results from
Received in revised form 16 September 2013 our experiment indicate that while this policy has the potential to significantly reduce corrupt practices, weak
Accepted 13 December 2013
economic incentives for the bribe-giver, or retaliation by bribe-takers can mitigate the disciplining effect of
Available online 4 January 2014
such an implementation. Asymmetric liability on its own may hence face challenges in the field.
JEL classification:
© 2013 Elsevier B.V. All rights reserved.
C91
K42

Keywords:
Harassment bribes
Experiment
Asymmetric penalty
Retaliation

1. Introduction reduce trust in good governance, which arguably reduces social welfare.
Insofar that the poor are more likely to use public services, such bribes
Bribes (also called speed money) are not limited to situations where act as a regressive tax and prevent access to basic services. As a result,
a citizen pays them to receive undue favors in contracts or services from developing effective anti-bribery mechanisms is an important policy ob-
public officials. Bribes are also frequently exchanged for delivering jective. While refusing to pay bribes remains an option for citizens,
entitled services such as admission to a public hospital or approval changing the legal framework has a potentially direct role for mitigating
of a passport.1 Basu (2011) characterizes the latter exchanges as harassment bribes.2 This paper examines the impact of different liability
harassment bribes. In this case, although the officials cannot deny ser- regimes on the incidence of harassment bribery.
vices legally, they can aggravate delivery or threaten to delay the Different countries vary in the legal approaches to curbing bribery.
service beyond a point where it becomes useless to the citizen. Harass- While in the United States, United Kingdom, France, Germany and
ment bribes raise the effective price of the public goods and services and India, the bribe-giver and recipient are both equally culpable and face
penalties (we call this symmetric liability), prescribed legal punishment
for the bribe-giver is comparatively mild in China, Japan and Russia (see
☆ We gratefully acknowledge financial support from Monash University and the Indian
School of Business. We thank Antonio Callari, David Owen, David Huffman, John Smith, Engel et al., 2013 for a discussion). Harassment bribery is reported to be
Martin Dufwenberg, Subha Mani and participants of the ESA meetings at NYU, Germany, particularly rampant in public services. Data collected over 21 months
and workshop participants at CSSS Kolkata, Monash University, University of from an Indian anti-graft website (www.ipaidabribe.com) showed that
Queensland, University of Tasmania, University of Virginia, JNU, World Bank, ISI Delhi about half a billion rupees was paid in bribes to lodge police complaints,
and ISB for their comments and suggestions. We also thank V. S. Rao, N. K. Sharma and
Veer Singh for their help in facilitating the experiments at their respective universities.
receive land purchase documents, marriage certificates, electricity con-
Urvashi Jain, K. Jayashree, Megha Juneja and Preethi Rao provided excellent research nections or registration documents for home purchases.3 Transparency
assistance. International (2012) ranks India 94th among the 176 countries and
⁎ Corresponding author at: AC8 Room 8119, Indian School of Business, Hyderabad AP
500032 India. Tel.: +91 4023187267.
2
E-mail address: [email protected] (T. Jain). “A novel way to combat corruption: Who to punish”, The Economist, May 5th 2011.
1 3
Since the seminal work by Rose-Ackerman (1978) on corruption, economists have “Rs. 11.42 crore and counting… Is what Bangalore paid in bribes”, The Times of India,
been interested in this area (Shleifer and Vishny, 1993; Mauro, 1995; Bardhan, 1997; June 6th, 2012. These kinds of bribes are observed in many other parts of the world as well.
van Rijkeghem and Weder, 2001; Rose-Ackerman, 2006; Rose-Ackerman and Soreide, For example, about 44% of parents in several African countries were asked to pay bribes for
2011). admission to public schools for their children, (Transparency International, 2010).

0047-2727/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jpubeco.2013.12.012
18 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

finds that citizens pay bribes especially in dealing with health and tax sec- Subsequent work uses this basic structure to answer corruption related
tors, real estate, legal courts and the police. Transparency International's questions and test policy instruments such as staff rotation (Abbink,
(2013) report on bribery in India points out that 54% of citizens report 2004), top-down vs. bottom-up monitoring (Serra, 2012), four-eye
paying bribes for common government services, and that 80% of these principle (Schickora, 2011), and the use of bribes to motivate inspec-
bribes were paid to avoid harm rather than to gain any advantage. The tions (Lowen and Samuel, 2012). There has also been research on the
current legal environment in many countries, which features symmetric impact of framing, subject background and culture on behavior in the
liability for both the bribe-giver and receiver, seems to further exacerbate context of corruption (Abbink and Hennig-Schmidt, 2006; Alatas et al.,
this situation. The typical bribe-giver, who is an ordinary citizen, is in a 2009a,2009b; Barr and Serra, 2009; Cameron et al., 2009; Banuri et al.,
dilemma when faced with a bribe demand from a public official. Refusal 2010). For surveys relating to the experimental literature on corruption,
to pay implies considerable inconvenience or loss due to a certain delay see Abbink (2006) or more recently Banuri and Eckel (2012) and Serra
in receiving the service, while succumbing to bribe-giving makes her and Wantchekon (2012).
legally culpable should the transaction be discovered. The paper most akin to ours is the study by Engel et al. (2013), who
To remedy this situation, Basu (2011) suggests a punishment system also compare symmetric versus asymmetric punishment regimes, albeit
with asymmetric liability, prosecuting and punishing only the public in the context of collusive bribery. In their bribery game, a citizen can
official (bribe-taker) and imposing no legal liability for the ordinary offer a bribe to an official who can reciprocate by manipulating his deci-
citizen seeking the service (bribe-giver).4 He hypothesizes that offering sion in the citizen's favor. In this framework, the results show that
the citizen legal impunity for whistle-blowing even if she has paid a asymmetric punishment increases the frequency of corrupt exchanges.
bribe can encourage more frequent reporting. In turn, this should dis- However, collusive and harassment bribery fundamentally differ from
courage officials from demanding bribes in anticipation of the whistle- one another. Under symmetric punishment, citizen and official both
blowing. Consequently, Basu predicts reduced incidence of harassment have an interest that their bribe payment stays undetected. Letting the
bribery in equilibrium.5 briber go free breaks this common interest. In collusive bribery, the
Critics of Basu's proposal argue that impunity for givers would make common interest stems from the exchange of favors, and this common
bribery morally acceptable, discourage refusals of bribe demands, and as interest still exists even if liability is not symmetric anymore. Hence, the
a result, increase the incidence of bribe-giving instead of reducing results from Engel et al. (2013) cannot necessarily be transferred to
it (Dreze, 2011). From a practical perspective, Drèze also points out harassment bribes.
that refunding bribes can be complicated and difficult, and that feeble Our paper is novel in several ways. First, to the best of our knowl-
prosecution rates and slow delivery of justice in developing countries edge, this is the first experiment addressing harassment bribery which
such as India are significant barriers that will discourage citizens is conspicuous in the public sector in many countries and has a large
from reporting bribe demands. Further, due to lax enforcement, bribe- detrimental impact on developmental outcomes. Second, our experi-
givers after reporting will be apprehensive of future harassment if the mental design helps us gain better insight into decision-making in the
official is still in office, and therefore, refrain from whistle-blowing. context of corruption. For instance, demands for large bribe amounts in-
This paper uses experiments to examine the effectiveness of the crease the costs of public services more than small amounts, so indig-
asymmetric liability mechanism in combating harassment bribes. Our nant citizens might be more likely to report these instead of staying
experimental approach offers an alternative to traditional survey or quiet. Finally, the treatments examined in the paper allow us to evaluate
field data analysis. We present a typical harassment bribe scenario in a the relative impact of monetary versus non-monetary factors that can
stylized game played by participants in a laboratory. By varying the in- influence the propensity for whistle-blowing.
stitutional environment across treatments, we can identify conditions Our findings provide qualified support for asymmetric liability as an
under which policy measures may or may not work. The laboratory al- anti-corruption mechanism. When bribe-giving is legalized, reporting
lows us to observe corrupt decisions empirically, an endeavor that is no- increases and demands for bribes decrease. However, reporting de-
toriously difficult in the field since everybody involved in such decisions creases when the official has the option to retaliate. Our results indicate
has good reasons to remain silent. further that refunding bribes after prosecution might not be necessary
While the effect of leniency programs has previously been studied in for the success of a leniency program since intrinsic motivation is the
the context of antitrust policies (Apesteguia et al., 2007; Bigoni et al., main driver of citizens' reporting behavior. Overall, our experimental
2012), only recently has this idea captured the attention of the anti- findings suggest that although asymmetric impunity schemes are
corruption literature. Dufwenberg and Spagnolo (2011) introduce a promising, they should be complemented with other measures to suc-
theoretical model to evaluate Basu's proposition of asymmetric penal- ceed as anti-corruption mechanisms.
ties. They conclude that the effect of Basu's policy depends on the qual-
ity of the indigenous institutions, would work better if bureaucrats are
under pressure to perform, and if citizens incur very low costs for 2. The harassment game
reporting bribery.6
The growing experimental literature on corruption, starting with Consider a simple sequential-decision game between an official and
Abbink et al. (2002), mainly studies collusive bribery, i.e., when citizens a citizen for the delivery of a service that the latter is entitled to. The
and officials exchange favors at the cost of the public. Collusive bribes official is obliged to grant the service, but de facto has the discretion to
are paid to obtain a favorable service that the briber is not entitled to. deny it, or delay it indefinitely. This gives him the opportunity to
demand a bribe for speedy delivery. The citizen can refuse to pay, but
4
A citizen feedback model being trailed in Pakistan (Callen and Hasanain, 2011) and this is very costly. The game below is purposefully parsimonious to
Ghana's whistle-blower act (Amegashie, 2013) have some similar features, with protec- allow us to shed light explicitly on the efficacy of asymmetric liability
tion being granted to citizen whistle-blowers and measures taken to empower them by as described by Basu (2011), and evaluate some of the associated
registering their complaints.
5 concerns.
Policymakers in other countries have attempted such asymmetric liability rules in
contexts other than corruption. For example, the United States outlawed distribution Fig. 1 describes the harassment game. In stage 1, the official can either
and sale of alcohol during the Prohibition era (1920–33), while consumption remained le- provide the service to the citizen without taking a bribe (B = 0), or
gal. As a result, customers could testify against their suppliers, which would have been dif- choose to specify a bribe amount (B) to ask from the citizen. In stage
ficult had consumption also been illegal (Miron, 1999). Sweden, followed by Iceland and 2, if the official asks for a bribe, then the citizen can choose any of the
Norway, introduced anti-prostitution laws with a similar spirit of asymmetric impunity.
6
Amegashie (2013) focuses on consumer complaints and its impact on social welfare
three actions: (1) refuse to pay the bribe, (2) pay quietly, or (3) pay
when officials may be corrupt or honest. His model also separates out corruption with and report the bribe. Actions (2) and (3) lead to a probabilistic discovery
theft and without theft. The latter situation is akin to harassment bribes. of bribery and the final payoffs in these two cases depend on whether
 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28 19

Fig. 1. The game tree—treatments without retaliation.

the act was discovered or not. The act of bribery is more likely to be rupees (Rs.). This outcome is the most efficient (in terms of joint payoff
discovered if the citizen has reported the demand. for the citizen and the official) and also equitable. Hence, it is salient that
We conduct two treatments with the above set-up. The treatments this is the socially preferred outcome.
differ in the payoffs the players receive for their actions. First, in a sym- If the official asks for a bribe, he also needs to specify how much to
metric liability treatment, both citizen and official are fined if caught. ask for in multiples of Rs. 10 up to a maximum of Rs. 200. The citizen
This represents the legal status quo in most countries, including India has three options: “Refuse to pay”, “Pay quietly”, or “Pay and report”.
where our study was conducted. In a contrasting asymmetric liability Refusing to pay the bribe is extremely costly for the citizen. In particular,
treatment, only the official gets prosecuted and pays a fine while the the citizen's payoff drops to Rs. 50, while the official retains 450 rupees,
briber enjoys impunity and gets back his bribe. Clearly, in the latter and the game ends. The decrease in the official's payoff is motivated by
case, the bribe-giver is no longer discouraged from reporting a bribe detrimental effects that bribery has on social efficiency, which we
demand, and in fact has a strictly positive incentive to report. This, in wanted to capture in our design. These detrimental effects have been
turn, should deter the official from asking for bribes. previously captured in bribery experiments by imposing payoff reduc-
To examine potential obstacles in the way to a successful implemen- tions on other players in the lab (Abbink et al., 2002; Barr and Serra,
tation of the impunity scheme above we conduct two additional treat- 2009; Cameron et al., 2009) or charities (Frank and Lambsdorff, 2010).
ments. To address the concern about spiteful retaliation from the For our game, this approach did not seem appropriate since in contrast
official's side, the first of our additional treatments introduces retaliation to collusive bribes, harassment bribes, it can be argued, have less of an
in the asymmetric liability scenario; now officials who escape conviction immediate effect on a third party. Yet the long-run detrimental effects
even after being reported can retaliate and reduce the citizen's payoff. In of bribery, such as the loss of trust in good governance are undeniable.
our game although retaliation is costly to the official, its mere availability Each act of bribery contributes to those damages, and everybody includ-
might dissuade citizens from reporting bribe demands if they anticipate ing those participating in bribery, suffers from them.7 Hence, to capture
spiteful actions from the official. The second of our additional treatments these features in our game, we reduce the total payoffs available to the
addresses the practicality of bribe-returns to the citizen. We implement two players when the official asks for a bribe, and thus impose an overall
the asymmetric liability framework with bribes no longer returned, decrease in social efficiency. We further wanted to keep the game as
while officials continue to have the option of retaliating if conviction simple as possible, hence we chose this way of implementing the harm-
fails after whistle-blowing. Since monetary incentives to report bribe ful effects of corruption.8
demands are removed, this last treatment poses the toughest behavior- If the citizen chooses to either pay quietly or pay and report the
al challenge to asymmetric liability proposal among the four situations. bribe, the game enters the next stage. Then a lottery determines wheth-
We next describe the four games (Symmetric, Asymmetric, Retaliation er the act of bribery is detected and fines are imposed. The probability of
and No-Refund) in detail along with the payoffs and experiment detection and prosecution depends on whether the citizen has reported
parameters. the bribe exchange. If the citizen has paid quietly, then there is only a
small chance that the act is discovered which we set to 5%. Paying and

2.1. The Symmetric treatment

7
In fact if the official holds a monopoly on access to a particular service (such as public
health care or passport delivery), the possibility of corruption could potentially raise the
Fig. 1 presents the extensive form of the game. First, the official price of necessary services, and further inhibit mutually beneficial exchanges, leading to
decides whether or not to demand a bribe. If he does not demand a some distortions.
bribe, the game ends and both players receive a payoff of 500 Indian 8
Since our experiment is one-shot, simplicity is of particular value in our design.
20 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

reporting increases this probability to 40%.9 Note that even if a bribe maximizing equilibrium prediction remains as before, no-bribery in
payment is reported, detection and consequent punishment are far equilibrium. A plethora of previous experiments however suggest that
from certain.10 retaliation even if costly, is carried out often; sometimes to encourage
If detected however, officials are fined in the symmetric as well as socially desirable outcomes (Gächter et al., 2008), and sometimes to en-
the asymmetric treatments. For the experiment to be meaningful, we force outcomes that are socially inefficient (Abbink et al., 2010). Hence,
needed to parameterize such that behavioral effects can show up in ei- we conjecture that even though not a part of the sub-game perfect equi-
ther direction in each treatment. We conjectured that a monetary fine of librium, a behavioral threat of retaliation can be credible and possibly
Rs. 250, leading to a final payoff of Rs. 200 for each player in case of pros- diffuse the deterrent effects of the leniency program.
ecution, would achieve this. Our chosen fine, along with the detection
probabilities reflect the current situation in India, where they do not 2.4. The No-Refund treatment
serve as an optimal deterrence mechanism (Debroy and Bhandari,
2012), i.e., they are high enough to serve as a deterrent, and yet not so On paper, our fourth treatment provides the toughest challenge for
high that expected payoffs become so trivial that no one would consider the asymmetric liability mechanism. The theoretical analysis of the
engaging in bribery. The only difference between the symmetric and game, as described by Basu (2011), relies heavily on monetary incen-
subsequent asymmetric treatments is that citizens do not pay a fine in tives offered to the citizen to come forward and report a bribe. In partic-
the asymmetric treatments. ular, the prospect of getting her bribe money back when the official is
A simple backward induction analysis of the game suggests that the convicted creates the positive incentive to self-report. However, this
monetary incentives are such that the citizen faced with a bribe demand feature might be naïve as, in reality, bribers rarely get a receipt for
will always pay quietly. The official anticipates this and chooses his best their payment so verifying the exact amount that changed hands is dif-
response—demand the maximum possible bribe. ficult. Therefore, in practice a leniency program is more likely to rely on
the citizen's intrinsic motivation to report a bribe. Such motivations can
2.2. The Asymmetric treatment exist, and in fact a citizen who is sufficiently upset about the unfairness
and immorality of the situation might be willing to report even without
This treatment introduces the briber leniency approach (as pro- the incentive of material benefits.12 The victim can also consider taking
posed by Basu, 2011), and as mentioned before, differs from the sym- action if she believes to be serving a greater societal benefit.13
metric treatment only in the treatment of the citizen in case of The no-refund treatment allows us to examine whether a leniency
detection. Here, only the official is fined (the same amount as in the policy can work in the absence of monetary incentives. The treatment
symmetric treatment).11 The citizen is not held responsible and the differs from the retaliation treatment only in the citizen's payoff in
bribe she has paid is returned to her. Consequently, monetary incentives case of the successful prosecution of the official. We retained the oppor-
change now in a way that the citizen would always like to report if she is tunity for officials to retaliate in this treatment to design an extreme
asked for a bribe. The backward induction logic in turn predicts that of- case scenario in which citizens would face the threat of retaliation and
ficials will shy away from demanding bribes, since not asking for a bribe also not have any monetary incentive to report. We introduce payoffs
is his best response to the citizen's “pay and report” choice. Hence, we for the citizen such that it is not strictly better for the citizen to report
should observe no bribes demanded in equilibrium. Fig. 1 summarizes a bribe as she does not get back the bribe even if whistle-blowing
the extensive form of the above two treatments. turns out to be successful. The equilibrium prediction is now indetermi-
nate. The citizen is indifferent between paying quietly and reporting the
bribe payment.14 As a result, multiple sub-game perfect Nash equilibria
2.3. The Retaliation treatment are possible, and behavior in the experiment can offer some insight on
preferred equilibrium behavior. Fig. 2 describes the extensive form of
The analysis above provides sufficient optimism on the deterrent ef- the above two treatments. Table 1 summarizes the design features
fect of the leniency program, at least in theory. In reality, however, a rel- and predictions of the four treatments.
atively low conviction rate might discourage whistle-blowing While several alternative treatments relating to corruption would be
considerably, especially if the citizen is apprehensive of further harass- interesting to examine using the above framework, we restricted our at-
ment from the official in the future. In this case, the prospect of getting tention to scenarios that are specific to harassment bribes and are cur-
the bribe back after successful prosecution must be weighed against po- rently being considered in the field in different parts of the world.
tential retaliation if the prosecution is unsuccessful, leading to the offi-
cial to remain in office. We consider such a situation in the retaliation
treatment where the official has the option to be spiteful and reduce 2.5. Behavioral predictions
the citizen's payoff if whistle-blowing is unsuccessful. We assume
though that the official needs to incur costs to do so. In particular, the of- The theoretical predictions laid out in the previous subsections are
ficial has to spend Rs. 50 to reduce the citizen's final payoff by Rs. 150. derived assuming pure own-income maximization, which is standard
This ratio (1:3) of cost to damages is consistent with previous experi- game theoretic reasoning. In a morally loaded context like corruption,
ments in which punishment options were studied (Gächter et al., 2008). payoffs perceived by the agents may not be only monetary. For instance,
Since it is costly, an official would never retaliate in the sub-game on citizens and officials may weigh potential material gains against their
the basis of monetary payoff gains alone. Hence, the theoretical money- discomfort from engaging in immoral acts. So, citizens' best responses
may be to refuse to pay even if it is monetarily costly and officials may
9 optimally forego profitable bribe opportunities. If we take these
The probability of detection chosen is conservative but realistic as empirical estimates
suggest that the fraction of reported bribe payments that leads to penalties is about 40% or
12
slightly less in India (National Crime Records Bureau, 2010). See for example, citizen efforts in reporting bribes in India through the web portal
10
Tummala (2009) in a careful documentation of prevalent corruption in India suggests www.ipaidabribe.com.
13
that although there are a plethora of government bodies that have been set-up over the In that case, one might even envision a situation where monetary incentives for
years, the law remains vague and open to different interpretations about the nature of whistle-blowers would prove to be counterproductive if such extrinsic motivations have
punishment. In particular, Section 165 of the Indian Penal Code states the punishment a tendency to crowd out intrinsic motivations (Frey and Oberholzer-Gee, 1997).
14
for bribery as follows, “…shall be punished with imprisonment … for a term which may It may even be costly for the citizen to report, in which case the citizen would have a
extent to three years, or with fine, or with both.” strict incentive not to report. We abstracted from such costs, assuming a situation where
11
Basu (2011) originally proposed that the fines for the official be doubled to keep the reporting facilities are put in place that make reporting costs trivial. Our goal was to study
overall fine the same. We did not implement this because it would have been difficult to the effectiveness of leniency out of pure intrinsic motivation, where there are no monetary
separate effects of the leniency program from effects of the higher fine. incentives for or against reporting.
 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28 21

Fig. 2. The game tree—treatments with retaliation.

considerations into account, then the equilibrium predictions become to pay and report in the symmetric case, we obtain that paying quietly is
less clear-cut. However, as we shall see, we still find that asymmetric better than reporting if
liability should encourage reporting.
Consider the citizen's decision problem. Following the classic ap- CR –CP N0:35B–87:5
proach by Rose-Ackerman (1975), we model attitudes to bribery as
moral costs that are incurred when engaging in corrupt acts. We assume Reporting is only preferable if the costs of paying quietly are much
that refusing does not incur moral costs (or equivalently, normalize higher than the costs of reporting, or if the bribe demand is very high.
them to zero). Let CP be the moral cost of paying quietly and CR be the In the asymmetric treatment, paying quietly is better than reporting if
moral cost of paying and reporting. Note that we cannot a priori say
CR –CP N0:35B
which one is greater. For instance, a citizen might feel CR b CP, because
she has done the right thing and reported a corrupt official, or CR N CP
Now the citizen would choose not to report only if she has a strong
if the citizen is averse to turning in another person even if it is a corrupt
aversion to turning in somebody. Note that regardless of what CR and
official. In the treatments with retaliation, the citizen must also take into
CP are, a citizen who reports under symmetric liability will always
account the subjective probability that the official retaliates, which we
report in the asymmetric case. Thus, even when citizens feel averse to
denote by q. Table 2 lists the payoff lotteries a citizen receives when
reporting, we should still expect more (or at least as much) reporting
choosing one of the three options, considering the moral costs.
under asymmetric liability than under the symmetric status quo.
Comparing the expected values of the lotteries (assuming risk
In the Retaliation treatment, the condition becomes
neutrality), we can derive conditions for the values of the moral costs
required to make one of the options optimal. If we compare pay quietly CR –CP N0:35B–90q

If the citizen does not expect any retaliation, then this is identical to
the asymmetric treatment (q = 0), otherwise reporting becomes,
Table 1 as expected, less attractive as the subjective probability of retaliation
Treatments. increases. Note that for the parameters of our experiment, we still
expect at least as much reporting with retaliation as in the symmetric
Treatment Description Predicted bribe
demands
baseline, except for extremely high perceived retaliation probabilities
(q N 87.5/90 = 97.2%).
1. Symmetric Official and Citizen both liable and pay penalty; 100%
In the no-refund treatment, the citizen prefers paying quietly over
bribe refunded.
2. Asymmetric Only Official liable and pays penalty; 0% reporting if
bribe refunded.
3. Retaliation Only Official liable and pays penalty; 0% CR –CP N−90q
bribe refunded; Official can retaliate.
4. No refund Only Official liable and pays penalty; Indeterminate Thus, if no retaliation is expected, the optimal decision solely de-
bribe not refunded; Official can retaliate.
pends on which action incurs the greater moral costs. Not surprisingly,
22 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

Table 2 (see the experimental instructions in the Appendix A).16 The use
Payoffs with moral costs. of context-specific instructions also improves the external validity of
Pay quietly Pay and report Refuse our results. Furthermore, we used real currency in the experiments
ensuring that participants could comprehend and relate to the decision-
Symmetric 200 − CP 200 − CR with 40% 50
with 5% making more easily.
450 − B − CP 450 − B − CR with 60% Subjects were asked to fill in their decision sheet where they made
with 95% decisions for every situation they could be in during the game. We
Asymmetric 450 − CP 450 − CR with 40% 50
chose to elicit complete strategies for two reasons. First, this method
with 5%
450 − B − CP 450 − B − CR with 60% allows us to gather decisions for all possible decision nodes, including
with 95% those that are not reached in realized play. This increases data-
Retaliation 450 − CP 450 − CR with 40% 50 effectiveness dramatically. Second, strategy elicitation compresses the
with 5% multi-stage game into one with a single simultaneous move for each
450 − B − CP 450 − B − CR with (1 − q) · 60%
player. Our time-constrained setting did not allow us to shift decisions
with 95%
300 − B − CR with q · 60% back and forth between the players.17
No-refund 450 − B − CP 450 − B − CR with 40% + (1 − q) · 60% 50 After the experiment was over, the subjects completed a survey
300–B–CR with q · 60% questionnaire first before collecting their experiment earnings. To com-
pute earnings, the decision sheet for each official was randomly
matched with a citizen who participated in the same treatment at the
same location. Based on the decisions of each player, the payoffs were
reporting becomes less attractive as the citizen assesses higher threats calculated and distributed to the participants in a sealed envelope. The
of retaliation.15 subjects earned an average of Rs. 493. At the time of the experiment,
the exchange rate to other major currencies was approximately 1.95
US dollars, 1.54 euro and 12.50 Chinese yuan for 100 Indian rupees.
3. Procedure

All experimental sessions were conducted in Hyderabad, India, with 4. Results

undergraduate and graduate (Masters) students. Three hundred and
sixty subjects participated in the experiment. Subjects were recruited 4.1. Survey results
by sending email invitations to the student association listservs in
each of three prominent institutes in the city—BITS Pilani's Hyderabad We find that even though our subjects were relatively young (aver-
Campus, NALSAR Law University and the University of Hyderabad. We age age of 22 years), 55% admitted to paying a bribe to obtain household
selected these institutions since they enroll students from all over services such as electricity, water or a telephone connection, financial
India allowing us to recruit a subject group with a range of backgrounds. services in a bank, post office, insurance company or transport office
All four treatments were conducted in each institution. Each subject and educational services at a school or college. Additionally, the partici-
participated in only one of the treatments. pants seemed to be well aware of anti-corruption laws in India, with 63%
No economics experiments have been conducted at any of these reporting “If caught, both the bribe-giver and taker are committing an
institutions before, so contamination from one session to another was illegal act”. They also report feeling uneasy about bribery, with only
a potential concern. Since anonymous recruiting systems were not in 22% supporting the statement, “Do you think that it is useful to have a
place, participants in later sessions could potentially be influenced by system where there is a way to get what you want even if you have to
what they heard from participants in earlier sessions. To rule out such bribe”. The data from the survey indicates that the subject pool was
effects, the experiment was conducted in a single afternoon at each reasonably informed, had some exposure to corruption and had views
location. We held two large sessions with 60 subjects back-to-back, and concerns about it, enabling an accurate examination of the effec-
with the second session starting while the participants of the first tiveness of different corruption policies. Note that the subjects answered
session were still in the classroom. All subjects first met in a large lecture these questions at the end of the experiment after all experiment
theater and were given general instructions. They were then divided decisions had been made.
into two groups of 30 subjects and led to one of two classrooms
where we had prepared visual separation between the desks. The two 4.2. Experimental results
groups participated in different treatments, which we conducted simul-
taneously. Detailed instructions were given by an instructor and to We organize our results below in terms of the two players and their
avoid possible experimenter effects, which might stem from the inevita- behavior. Citizens' actions are reported first, since leniency proposals
ble use of different instructors for the different groups, we made sure are directly aimed at changing their behavior. Officials' anticipatory
that the assignment of instructors to treatments was as balanced as behavior is reported next.
possible.
The subjects within each group were randomly assigned to the 4.3 . Do Citizens report under impunity?
role of an official or a citizen. The experimenter read the instructions
aloud. Our objective was to simulate the context of a corrupt transaction Figs. 3 and 4 show the percentage of subject decisions that were “Pay
and evoke associated emotional and moral responses. Therefore, we and report”, “Pay quietly” and “Refuse to pay” in each of the four treat-
used in-context language consistently throughout the instructions ments. We find that consistent with the predictions earlier (as well as

15 16
We do not derive a full-fledged equilibrium analysis including the officials' best re- There is mixed evidence about the impact of context-specific instructions in corrup-
sponses. Extending the analysis in this way is straightforward, but can be very lengthy, tion experiments. While Barr and Serra (2009) find that bribes are less likely to be offered
as many cases need to be considered. The official's optimal bribe demand depends on and accepted in the laboratory when the experimental instructions explicitly describe a
his expectation of the reporting probability, which depends on the distribution of moral bribery scenario instead of a more abstract description, Abbink and Hennig-Schmidt
costs he expects in the population of citizens, and his own moral costs of demanding a (2006) and Krajcova and Ortmann (2008) find no significant differences between neutral
bribe. However, all other things equal, the payoff structure of the game implies that bribe and bribery frames.
17
demands become less attractive if the probability of reporting crowds out the probability Brandts and Charness (2011) survey experiments that have been conducted with both
of paying quietly. Hence, we would expect that a policy that encourages reporting should strategy elicitation and spontaneous play. While magnitude effects can sometimes be
not increase the frequency of bribe demands, but likely decrease it. found, no study reports that treatment comparisons would be affected.
 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28 23

100% high bribe demands may be seen as unkind and trigger reporting. This is
observed in the symmetric treatment where reporting a bribe demand is
90%
actually costly for the citizen, and yet high bribe demands are typically
80% reported. In contrast, low demands are possibly perceived as a relatively
kind act from the official; so in the asymmetric treatment, when reporting
70% the bribe demand is not costly and citizens would in fact improve their
expected monetary payoffs, they tolerate low bribe demands and
60%
choose not to report.
50% The pairwise tests of treatment differences in Table 2 show that
for the “pay quietly” option, behavior is significantly different across
40%
treatments for the first two comparisons, but not for the comparison be-
30% tween the retaliation and no-refund treatment, suggesting that taking
monetary incentives away may not affect reporting behavior signifi-
20% cantly. These tests also indicate that consistent with Fig. 4, the percent-
age of subjects who “refuse to pay” remain unaltered across treatments.
10%
This suggests that at least in our experiment, citizens on an average do
0% not change their behavior due to the introduction of leniency programs
Symmetric Asymmetric Retaliation No refund that might signal moral acceptability.
Pay and report Pay quietly Refuse to pay
4.4 . Do officials demand fewer bribes?
Fig. 3. Citizens' decisions—all bribe offers.

We next examine whether giving immunity to citizens induces any

anticipatory change in bribe demands by officials. Recall that according
Basu's proposal), the percentage of citizens who pay and then report the to Basu's hypothesis asymmetric liability ought to not only change
bribe demand jumps from 25% to 59% under the asymmetric liability citizens' behavior, but also discipline officials' bribe demands. Accord-
policy (a Mann–Whitney test rejects the null of equal means ingly, the backward induction analysis in our asymmetric and retaliation
with p-value = 0.001). Note that the increase in this percentage is games predicts that in the unique sub-game perfect Nash equilibrium,
not due to any perceptible change in the percentage of subjects who re- officials do not demand any bribes. Overall, in our experiment the
fuse to pay the bribe (see Fig. 3). Rather it is caused by the decrease in officials seem to show a weaker response to variations across treat-
the percentage of citizens who pay quietly (58% to 19%). ments compared to citizens. One possibility is that they might not be
The increase in reporting behavior however, goes down significantly thinking in a backwardly inductive manner. Officials might also face
from the asymmetric to the no-refund treatment where officials are an additional level of strategic uncertainty compared to citizens, as
allowed to retaliate and strict monetary incentives from reporting are they have to foresee the citizens' response while citizens can condition
removed (a Mann–Whitney test rejects the null of equal means with their choice on the official's demand. This additional level of uncertainty
p-value = 0.01). In fact, we find that the average percentage of citizens can possibly dilute some of the treatment effects.
“paying and reporting” goes down towards the mean of the symmetric Fig. 5 shows a distinct tendency towards a reduction in bribe demands
treatment (a Mann–Whitney test fails to reject the null of equal means by officials. We find that the percentage of officials who demand a bribe
with p-value = 0.65). These results suggest that allowing the possibili- drops from 38% in the symmetric liability situation to 24% in the asym-
ty of retaliation by officials deters citizens' reporting behavior. metric liability situation (a Mann–Whitney test rejects the null of equal
Further, an examination of the distribution of bribe demands sug- means with p-value = 0.08). The average bribe amount demanded by
gests that irrespective of the treatment, citizens make their choices con- the officials also goes down from Rs. 152 to Rs. 135 rupees, although
tingent on the officials' bribe demand amounts—higher bribe demands this is not statistically significant.18 The drop stems from fewer bribe
are typically met with a “pay and report” while smaller bribe demands demands, not the amount that is asked for. Fig. 5 describes the average
are often paid and not reported. Fig. 4 describes the subject behavior bribes asked in each treatment, conditional on a positive bribe demand.
in the four treatments. These averages are very similar across all treatments and no difference
To investigate citizens' behavior in more detail, we estimate a multi- is statistically significant. One might hypothesize that officials may try
nomial logit model where citizens' choices are a function of the amount to compensate for the higher risk of being reported under asymmetric
of bribe demanded, controlling for the education institutions as well as liability by demanding higher bribes, but our data do not provide any
some of the demographic characteristics of the participants obtained indication for this.
from the survey. Results of this estimation are reported in Table 3. We When officials are allowed to retaliate, 44% chose to pay money
find that irrespective of the treatment, a one-rupee increase in the (Rs. 50) to reduce the citizen's payoff (by Rs. 150) in the retaliation
bribe demand significantly decreases the relative risk ratio of being in treatment and 33% in the no-refund treatment (this difference is not
the “pay quietly” group compared to being in the base comparison statistically significant). The proportion of officials demanding bribes
group of “pay and report”. For example, in the symmetric treatment, a increases as anticipated. In fact, behavior in the retaliation and no-
rupee increase in the bribe-ask leads to a decrease in the relative risk refund treatments suggests that the disciplinary effects of briber im-
of the subject being in the “pay quietly” group compared to the “pay punity dissipate away considerably for officials. Fig. 5 shows that 38%
and report” group by a factor of 0.995 (see row 1 and column 3 of and 27% of officials demand bribes in the two treatments. The average
Table 3), and similarly for all the other three treatments. amount of bribes demanded are Rs. 149 and Rs. 148 respectively,
The results confirm our observation that across treatments, as the lower than in the symmetric treatment but higher than the asymmetric
size of the bribe demand increases it is more likely to be reported. Put treatment.
alternatively, a small amount of bribe demand is less likely to be report- We next report the results from a probit regression with robust stan-
ed. The latter observation reveals that subject behavior is suggestive of dard errors (see Table 4) that examine the determinants of official
an established norm among the subject population where a large
bribe demand is considered to be unfair, while a small bribe demand 18
Despite the high bribe amount, the overall frequency of bribes is low, with at least 62%
is acceptable and does not violate subjects' fairness perceptions. Conse- of officials not demanding bribes. This might indicate either aversion to corruption or risk
quently, we observe an element of reciprocity in citizens' choices where aversion.
24 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

Fig. 4. Citizens' decisions conditional on bribe demanded.

behavior, controlling for variables from our post-experiment survey.19 The results are depicted in Fig. 6.20 It can be seen that expected pay-
The results confirm our earlier observation that asymmetric liability de- offs generally increase with the bribe demand. Thus, the citizens' higher
creases the likelihood of bribe demand compared to the baseline situa- propensity to report high bribes is not high enough to offset the officials'
tion of symmetric liability. Interestingly, we find that the subjects who extra income from receiving the bribes. In three of the four treatments
report that they are agreeable to giving bribes in the post-experiment the most profitable strategy for officials is to demand the maximum
survey are significantly more likely to ask for bribes. Also, the male sub- bribe. Only in the asymmetric treatment without retaliation, expected
jects seem to ask for more bribes than the female subjects, a finding con- payoffs for all bribe levels remain below Rs. 500 that an official would
sistent with previous experimental findings (Alatas et al., 2009a,2009b; have obtained if he chose not to demand a bribe. This would suggest
Frank et al., 2011) as well as other empirical findings (Lambsdorff and that for a risk-averse official, not demanding a bribe can be an attractive
Fink, 2006). Chi-square tests for mean differences in treatments indicate proposition from payoff considerations alone. Note though, that Fig. 6
further that official's behavior is only marginally different between the suggests that for the average bribe of Rs. 150 asked in the experiment,
symmetric and the retaliation treatment; official's behavior is also the expected payoffs are lower than what an official could have
not significantly different between the retaliation and the no-refund achieved if he decided to go for the no-bribe outcome (Rs. 500). A pos-
treatments. sible reason for this finding might be that while officials correctly esti-
Between the symmetric and the no-refund treatment, there are three mate that citizens are less likely to report bribe demands that are not
features that are different; namely, addition of the asymmetric liability, very high (and therefore ask for less than Rs. 200), they underestimate
official's ability to retaliate, and removal of strict monetary incentives the threshold and ask for smaller than optimal bribes.
for the citizens. Therefore, we cannot discern causality in the difference
in officials' behavior between the two treatments. However, the chi-
square tests of mean differences establish that officials demand signifi- 5. Conclusion
cantly lower bribes compared to the symmetric treatment. We infer
from this result that even in the strongest test of the asymmetric liability While there has been extensive research on using different ways to
mechanism, the disciplinary effect remains present. fight corruption, the use of citizen reporting as a means of tackling
The question arises whether asymmetric liability affects the profit- corrupt government officials has received relatively little attention
ability of bribe demands. Since we elicited complete strategies from until recently. Whistle-blowing by citizens can be particularly useful
the citizens, we can calculate the expected empirical payoff from each in situations where the citizen is held hostage by the government
bribe level based on the citizens' propensity to report at every bribe level. official who refuses to provide services that the citizen is entitled to.
This corruption situation is characterized by harassment bribes.
Using laboratory experiments, we evaluate the effectiveness of an
19
In our specification we included controls for religion (Hindu or non-Hindu), as well as asymmetric liability policy that provides legal immunity to bribe-
interaction terms of the religion dummy with the treatment. Although our results seem to givers to encourage whistle-blowing. We incorporate two extensions
suggest that Hindus ask for bribes significantly more often, we realize that our total sample to the basic policy proposal to provide behavioral evidence on the effec-
is relatively small (49 out of 147 Hindus ask for a bribe in all the four treatments together, tiveness of the policy when bribe-takers have the option to retaliate on
and 8 out of 33 non-Hindus ask for a bribe in all the four treatments together) to convinc-
ingly conclude whether religion affects the propensity to engage in bribe-taking behavior.
20
Accordingly, we have not reported the dummies related to religion and its interaction with We do not include money officials may spend on retaliation in the expected payoff
the treatments. calculations.
 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28 25

Table 3 Table 4
Multinomial logit estimates of citizen decisions. Probit estimates of Official behavior.

Refuse to pay Pay quietly Bribe demand

Coefficient (S.E.)
RRR Std. err. RRR Std. err.
Symmetric treatment 2.27*** (0.94)
Bribe amount in symmetric treatment 1.000 0.002 0.995*** 0.002
Retaliation treatment 0.90 (0.75)
Bribe amount in asymmetric treatment 0.996 0.002 0.977*** 0.003
No-refund treatment −0.02 (0.85)
Bribe amount in retaliation treatment 0.999 0.002 0.989*** 0.002
Nalsar −0.42 (0.26)
Bribe amount in no-refund treatment 1.000 0.002 0.992*** 0.002
University of Hyderabad −0.50 (0.32)
NALSAR 1.371 0.507 0.876 0.214
Male 0.46** (0.23)
University of Hyderabad 4.244*** 1.420 0.974 0.261
Scheduled caste 0.25 (0.33)
Hindu 1.306 0.540 1.144 0.306
Income (in thousands) 0.001 (0.003)
Scheduled Caste 0.817 0.345 1.072 0.345
Agreeable to giving bribes 0.64*** (0.23)
Male 1.113 0.332 1.373 0.300
Age 0.06 (0.06)
Chi-square test for mean treatment differences when choice = Refuse to pay
Chi-square test for mean treatment differences
(Bribe amount in) symmetric treatment = p-value = 0.18
Symmetric = retaliation p-value = 0.09
asymmetric treatment
Retaliation = no-refund p-value = 0.21
(Bribe amount in) asymmetric treatment = p-value = 0.35
Symmetric = no-refund p-value = 0.01
retaliation treatment
Pseudo R square 0.11
(Bribe amount in) retaliation treatment = p-value = 0.64
Number of observations 180
no-refund treatment
Notes: “Asymmetric” treatment, “Female” and “BITS” are the omitted categories from the
Chi-square test for mean treatment differences when choice = Pay quietly independent variables. Robust standard errors reported in parentheses. Specification
(Bribe amount in) symmetric treatment = p-value = 0.00 includes a religion dummy along with a set of interaction dummies between treatment
asymmetric treatment dummies and the religion dummy. These are not reported here for ease of exposition.
(Bribe amount in) asymmetric treatment = p-value = 0.00 **p b 0.05 and ***p b 0.01.
retaliation treatment
(Bribe amount in) retaliation treatment = p-value = 0.26
no-refund treatment in such situations, both bribe demands and reporting return closer to
Pseudo R square 0.102
Number of observations 3571
the original levels of the symmetric liability case.
Our finding suggests that as an anti-corruption measure, asymmetric
Notes: RRR is Relative risk ratio. “Pay and report” is the base outcome from the dependent
liability should be implemented along with complementary measures
variables. “BITS” is the omitted category from the independent variables. Standard errors
clustered by participant. ***p b 0.01. such as policies rotating officials in different posts to mitigate the effec-
tiveness of retaliation against citizens who report bribe demands (see,
for example, Abbink, 2004). In addition, to protect citizens' vulnerabili-
whistle-blowers if the prosecution is not successful, and the impact of ty, whistle-blowers may need to be given protection such as anonymity
non-monetary incentives in reporting bribe demands. for reporters and timely prosecution of the accused. Hence, bolstering
We find that compared to symmetric liability, allowing legal immu- the institutional set-up is important to realize the full benefits of this le-
nity to the bribe-givers increases reporting of bribe demands and re- niency policy.
duces the demand for bribes. We also find that a substantial minority A related concern with implementing a policy that reduces liability
of citizens refuse to pay bribes, across treatments despite significant for bribe-givers is that greater reporting may impose large costs on al-
monetary costs of doing so. This is not surprising, since refusals to pay ready over-burdened judicial systems. Indeed, the incidence of reporting
may be driven more by principles rather than incentives. An implication bribery increases in our experiment between the symmetric and asym-
of this is that asymmetric liability does not change the moral authority metric liability treatments. The costs of prosecuting corruption cases in
of the law on citizens' behavior and consequently the proposed change developing countries are difficult to estimate reliably and if the increase
in the liability does not have to be interpreted as a “license to bribe”. in reporting leads to higher overall costs, then the effectiveness of the
Comparing behavior in the retaliation and no-refund treatments mechanism can potentially be reduced. However, while reporting
shows that strict financial incentives do not necessarily drive reporting might indeed increase in the short term, we also expect that bribe de-
behavior. Non-monetary factors can motivate reporting behavior as mands decrease in the longer run. Insofar as citizens' reporting is driven
well. This could be beneficial in the field as monetary incentives (such by officials' demands, we should then observe fewer cases of reporting
as return of the bribe money) are often difficult to operationalize. and therefore lower prosecution costs. To ensure that citizens' threat to
An analysis of officials' behavior suggests that significant challenges report remains credible, practical implementation of the policy should
to implementing an asymmetric liability rule emerge when officials are
able to retaliate against citizens who report bribe demands. We find that 540

520
Expected payoff (rupees)

Percentage of Average amount demanded

officials demanding bribes conditional on bribe demand
100% 200 500

80% 480 Symmetric

150
152 149 148 Asymmetric
60% 135 460 Retaliation
100
40% No refund
440
37.8% 37.8% [no bribe]
20% 50
24.4% 26.7%
420
0% 0 400
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200

Bribe demand

Fig. 5. Officials' behavior. Fig. 6. Expected payoff to official conditional on bribe demand.
26 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

perhaps coincide with efforts to reduce costs of reporting, such as great- decide not to ask for a bribe, then you get Rs. 500, and the Citizen gets
er use of information technology to crowd-source complaints. Rs. 500.
Moreover, any leniency policy would need to be implemented in the If instead, you decide to ask for a bribe, you have to decide how much
field with caution. While citizen reporting is useful to identify corrupt to ask for. You can ask any amount (B) between 10 and 200 in multiples
officials, citizens themselves could misuse this leniency measure and re- of Rs. 10.
port honest officials. The focus of our study is on deterring harassment The Citizen decides whether or not to refuse to pay the bribe, or to
bribes; we did not examine spiteful behavior of citizens when they pay the bribe without reporting, or to pay the bribe and report your
have the opportunity to report officials. In future research, it would be bribe demand. Reporting determines the probability with which you
interesting to explore if citizens report honest officials and attempt to are fined. If the Citizen does not report the bribe, then the probability
use the asymmetric liability rule for their benefit by harassing the of you being fined is 5%. If the Citizen reports the bribe, then this prob-
official. ability increases to 40%.
After the Citizen has decided, a random draw determines whether
Appendix A . The experimental instructions (For the Retaliation there is sufficient evidence that you are fined. If there is not sufficient
treatment, other treatments are analogous) evidence, then you receive Rs. 450 plus the bribe you have asked
for. The Citizen receives Rs. 450 less the bribe s/he has paid to you.
Player No. _________ If there is sufficient evidence, then you receive a fine and you must
You have been randomly assigned the role of Citizen/Official in return the bribe and in addition pay a fine. Your payoff after this stage
today's experiment. is then Rs. 200. The Citizen gets the bribe back; hence his/her payoff is
Rs. 450.
If the Citizen has reported the bribe, but there has not been sufficient
General
evidence, then the game enters a third decision stage. You can spend Rs.
50 from your final earnings to reduce the Citizen's payoff by Rs. 150; in
Welcome to today's economics experiment. This is an experiment in
that case, you will receive as your final payments Rs. 400 plus the
decision-making which will provide you an opportunity to earn money.
amount of bribe you had asked for earlier; the Citizen will end up receiv-
The amount of money you earn depends on your decisions and a ran-
ing Rs. 300 less the amount of bribe s/he paid to you. If you choose not to
domly matched participant's decisions in the experiment. Your earnings
reduce the Citizen's payoff, then the final payoffs are the payoffs after
in the experiment will be paid to you in cash privately, at the end of the
stage 2: You receive Rs. 450 plus the bribe, the Citizen receives Rs. 450
experiment. Please do not talk to each other during the experiment.
less the bribe.
In the experiment you will be matched with another player in the
As an Official you make decisions at up to two stages. In the begin-
room for the rest of the experiment. You will not know who you are
ning you decide on whether you ask for a bribe and if so, how much
matched with, either during or after the experiment. You and the
you demand. If you decide to ask for a bribe, then it is possible that
matched player will be presented with an economic decision-making
you need to make another decision at the third stage. In case that the
situation that resembles a real-life situation. One of you will be random-
Citizen reports the bribe but you do not get fined you can choose wheth-
ly assigned a role as a Government Official, and the other as a Citizen.
er or not to reduce the Citizen's payoff. We ask you to make this decision
You will be provided with a Personal Record Sheet that will state the
already in the beginning. It is possible that your decision for the third
role you have been assigned in the experiment.
stage is not carried out, depending on the decisions of the Citizen and
the outcome of the random draw. We nevertheless ask you to make a
Overview of the Experiment decision for this case beforehand, such that we do not need to return
the decision sheet to you until the game is completed.
In this experiment, the Official can decide to ask the Citizen for a
bribe or can decide not to ask the citizen for a bribe. Detailed Instructions for Citizens
If the Official decides to ask for a bribe, then the Official has to choose
the amount he wants as bribe from the Citizen. The Citizen then has If you are randomly assigned the role of the Citizen in today's exper-
three options. S/he can refuse to pay the bribe, pay the bribe, or pay iment, it will be the Official who makes a decision first and you respond
the bribe and report the bribe demand. Reporting the bribe makes it to it. First the Official decides whether or not to ask for a bribe. If s/he
much more likely that the Official is caught and fined. does not ask for a bribe, then you get Rs. 500, and the Official gets Rs.
If the Citizen has reported the bribe, but the authorities have not 500.
found sufficient evidence to fine the Official, then the game moves to If instead, the Official decides to ask for a bribe, you are told how
another stage, in which the Official can reduce the Citizen's income by much the Official asks for.
incurring a cost. If the Official asks for a bribe, you have the following options. First,
Though the game has up to three stages in which one player needs to you can refuse to pay the bribe. In this case the game ends and your pay-
make a decision, everybody needs to fill in the decision sheet only once. off is Rs. 50; the Official's payoff is Rs. 450. If you decide to pay the bribe,
You will make decisions for every situation in which you can be during you can decide whether or not to report the Official's bribe demand.
the game. We will then collect your decision sheets and pay you accord- Your decision to report determines the probability with which the Offi-
ing to your decision and the decisions of the other participant you are cial is fined. If you do not report the bribe, then the probability of the Of-
matched with. ficial being fined is 5%. If you report the bribe, then this probability
Attached is a Figure which summarizes the structure of the experi- increases to 40%.
ment. The sheet labeled “Questions” provides some examples that After you have decided to pay the bribe, a random draw determines
might help you in understanding the payoffs associated with different whether there is sufficient evidence that the Official is fined. If there is
decisions. However, before looking at the examples, let us first look at not sufficient evidence, then you receive Rs. 450 minus the bribe you
the detailed instructions for each participant. paid out. The Official receives Rs. 450 plus the bribe you have paid to
him/her. If there is sufficient evidence, then you get back the bribe
Detailed Instructions for Officials you have paid and your payoff after this stage is Rs. 450. The Official
receives a fine, his/her payoff after this stage is then Rs. 200.
If you are assigned the role of the Official in today's experiment, you If you have reported the bribe, but there has not been sufficient evi-
have to first decide whether to ask the Citizen for a bribe or not. If you dence, then the game enters a third decision stage. The Official can
 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28 27

spend Rs. 50 from his/her final earnings to reduce your payoff by Rs. ▪ Hindu
150; in that case, you will receive as your final payments Rs. 300 less ▪ Muslim
the amount of bribe you paid to the Official. The Official receives Rs. ▪ Christian
400 plus the amount of bribe you have paid to him/her. If the Official ▪ Sikh
chooses not to reduce your payoff, then the final payoffs are the payoffs ▪ Other (Please specify__________________)
after stage 2: You receive Rs. 450 less the bribe, the Official receives Rs. ▪ Don't know
450 plus the bribe. 4. What is your caste?
As a Citizen you make decisions at the second stage, after the Official ▪ Scheduled caste
has decided on the bribe demand. The Official can either not demand a ▪ Scheduled tribe
bribe, in which case you do not make a decision. If the Official demands ▪ Other backward classes
a bribe, s/he can ask for twenty different amounts of bribe from 10 to ▪ Upper caste
200 (in steps of 10). We ask you to make a decision for each bribe ▪ Other (Please specify _________________)
amount asked from you beforehand. Your decision sheet comprises a ▪ Don't know
table with all twenty possible amounts. For each amount you tick a 5. What is your program and year of study at the University or Institute?
box whether you want to refuse to pay the bribe, pay without reporting, (Mark only one)
or pay and report the bribe demand if the Official demands this amount. ▪ Bachelor's student (BA, BSc, BE, etc.)
We will then collect the Official's decision sheets together with yours, o Circle Year 1/2/3/4/5
and carry out the decision you specified for the amount the Official ▪ Master's student (MPhil, MA, MSc, MTech, MBA, LLB etc.)
has chosen (if any). o Circle Year 1/2/3/4/5
▪ Other (Please specify _________________)
Control Questions 6. What is your field of study (specialization) in the program?
______________________________
(These questions aim to help you understand the experiment better and 7. Last year, what were your average marks/grades in the program?
should not be used as a guide for decision-making in the experiment.) _____ out of _________
8. How much work experience do you have? (Mark all that apply)
1. Assume that the Official asks for a bribe of Rs. 150. Suppose that the
• None
Citizen he is matched with decides to report the bribe and the Official
• Internship __________ months.
is caught and fined. What will be the earnings of the Official and the
Employer(s):_________________________
Citizen in this group?
• Full-time work __________ years.
a. Official: Rs
Employer(s):_________________________
b. Citizen: Rs
9. In the last year, how much did you earn? Include all sources. Rs.
2. Suppose that the Official asks for a bribe of Rs. 80. The Citizen reports
______________
the bribe demand. The Official is not caught however and he decides
10. What were the source(s) of this income? (Mark all that apply)
to spend Rs. 50 to reduce the payoffs of the Citizen. What will be the
• Employment (part-time/full-time job)
earnings of the Official and the Citizen in this group?
• Allowance from family
a. Official: Rs
• Scholarships
b. Citizen: Rs
• Other (Please specify _________________)
3. What will be the earnings of the Official and the Citizen in the group,
11. How do you most hear about corrupt behaviour? (Mark only one)
if the Official does not ask for a bribe?
• Through personal experience.
a. Official: Rs
• Through the experiences of family or friends
b. Citizen: Rs
• By reading magazines or the newspaper
4. Suppose that the Official asks for a bribe of Rs. 200. The Citizen does
• By listening to the news on TV or radio
not report the bribe demand. The Official is not caught. What will be
• Through an academic course
the earnings of the Official and the Citizen in this group?
• Other (Please specify _________ ________)
a. Official: Rs
12. In what context do you most hear about corrupt behaviour? (Mark
b. Citizen: Rs
only one)
5. Suppose that the Official asks for a bribe of Rs. 100. The Citizen re-
• Corruption scandals involving politicians and bureaucrats
fuses to give the bribe. What will be the earnings of the Official and
• Corruption scandals involving companies and rich individuals
the Citizen in this group?
• Harassment of ordinary people for basic services
a. Official: Rs
• Other (Please specify _________________)
b. Citizen: Rs
13. In what context do you most experience corrupt behaviour? (Mark
only one)
Post-Experiment Survey
• I receive poorer quality public infrastructure because of corrup-
tion
Player Number: ________
• Other people get ahead in education and career because of cor-
Instructions: Please answer ALL of the questions on this survey as
ruption
accurately as you can. All responses will be kept confidential by the
• I have to give bribes frequently for basic government services
researchers and will not be revealed to any authorities within the
• I have to give bribes frequently for services by private service
university or outside. Leave blank if you do not wish to answer.
providers
1. What is your date of birth? • I have to give bribes occasionally for basic government services
• I have to give bribes occasionally for services by private service
Month: ___________ Year: ___________
providers
2. What is your gender? • Other (Please specify ________________)
▪ Male 14. In which contexts have you ever given a bribe? (Mark all that apply)
▪ Female • To get household services such as electricity, water or telephone
3. What is your religion? connection
28 K. Abbink et al. / Journal of Public Economics 111 (2014) 17–28

• To get services in a bank, post office, insurance company or trans- Bardhan, P., 1997. Corruption and development: a review of issues. J. Econ. Lit. 35 (3),
1320–1346.
port office Barr, A., Serra, D., 2009. The effects of externalities and petty corruption experiment. Exp.
• To get educational services at a school, college or for a scholarship Econ. 12 (4), 488–503.
• Other (Please specify _________________) Basu, K., 2011. Why, for a class of bribes, the act of giving a bribe should be treated as
legal. Ministry of Finance Working Paper.
• I have never given a bribe Bigoni, M., Fridolfsson, S., Le Coq, C., Spagnolo, G., 2012. Fines, leniency and rewards in
15. Which of the following best describes the anti-corruption law in antitrust: an experiment. RAND J. Econ. 43 (2), 368–390.
India? (Mark only one) Brandts, J., Charness, G., 2011. The strategy versus the direct-response method: a first
survey of experimental comparisons. Exp. Econ. 14, 375–398.
• If caught, both the bribe-giver and taker are committing an illegal Callen, M., Hasanain, A., 2011. The Punjab model of proactive governance: empowering
act citizens through information communication technology. (http://odta.net/post/the-
• If caught, the bribe-taker is committing an illegal act, but the punjab-model-of-proactive-governance-empowering-citizens-through-information-
communication-).
bribe-giver is not responsible Cameron, L., Chaudhuri, A., Erkal, N., Gangadharan, L., 2009. Propensities to engage in and
• If caught, the bribe-giver is committing an illegal act, but the punish corrupt behaviour: experimental evidence from Australia, India, Indonesia
bribe-taker is not responsible and Singapore. J. Public Econ. 93 (7–8), 843–851.
Debroy, B., Bhandari, L., 2012. Corruption in India—the DNA and the RNA. Konark Pub-
• If caught, neither the bribe-giver nor taker are committing an illegal
lishers, New Delhi.
act Dreze, J., 2011. The bribing game. Indian Express. (April 23, 2011).
• I don't know anything about the anti-corruption law in India Dufwenberg, M., Spagnolo, G., 2011. Legalizing bribes. Eller College of Management
Working Paper No. 11-09.
16. Do you think that it is useful to have a system where there is a way
Engel, C., Goerg, S., Gaoneng, Y., 2013. Symmetric vs. Asymmetric Punishment Regimes for
to get what you want even if you have to bribe? (Mark only one) Bribery. MPI Collective Goods Preprint, No. 2012/1.
• Yes Frank, B., Lambsdorff, J., 2010. Bribing versus gift-giving—an experiment. J. Econ. Psychol.
31 (3), 347–357.
• No Frank, B., Lambsdorff, J., Boehm, F., 2011. Gender and corruption: lessons from laboratory
• Don't know corruption experiments. Eur. J. Dev. Res. 23 (1), 59–71.
17. In which of the following situations have you jumped or cut a Frey, B., Oberholzer-Gee, F., 1997. The cost of price incentives: an empirical analysis of
motivation crowding-out. Am. Econ. Rev. 87 (4), 746–755.
queue? (Mark all that apply)
Gächter, S., Renner, E., Sefton, M., 2008. The long-run benefits of punishment. Science 322,
• While waiting to buy a ticket 1510 (5 December).
• Boarding a bus Krajcova, J., Ortmann, A., 2008. Testing leniency programs experimentally: the impact of
“natural” framing. CERGE-EI Working Paper Series No. 372.
• Boarding a train Lambsdorff, J., Fink, H., 2006. Combating corruption in Colombia: perceptions and
• In government offices achievements. Passauer Diskussionspapiere No. V-44-06.
• Waiting at the bank teller Lowen, A. And, Samuel, A., 2012. Bribery and endogenous monitoring effort: an experi-
mental study. East. Econ. J. 38 (3), 356–380.
Mauro, P., 1995. Corruption and growth. Q. J. Econ. 110 (3), 681–712.
References Miron, J., 1999. The effect of alcohol prohibition on alcohol consumption. NBER Working
Paper No. 7130.
Abbink, K., 2004. Staff rotation as an anti-corruption policy: an experimental study. Eur. National Crime Records Bureau, 2010. Crime in India: Statistics. Ministry of Home Affairs,
J. Polit. Econ. 20 (4), 887–906. New Delhi.
Abbink, K., 2006. Laboratory experiments on corruption. In: Rose-Ackerman, S. (Ed.), Rose-Ackerman, S., 1975. The economics of corruption. J. Public Econ. 4 (2), 187–203.
International Handbook on the Economics of Corruption. Edward Elgar, Cheltenham. Rose-Ackerman, S., 1978. Corruption: A Study in Political Economy. Academic Press, New
Abbink, K., Hennig-Schmidt, H., 2006. Neutral versus loaded instructions in a bribery York.
experiment. Exp. Econ. 9 (2), 103–121. Rose-Ackerman, S., 2006. The International Handbook on the Economics of Corruption.
Abbink, K., Irlenbusch, B., Renner, E., 2002. An experimental bribery game. J. Law Econ. Edward Elgar, Cheltenham.
Org. 18 (2), 428–454. Rose-Ackerman, S., Soreide, T., 2011. The International Handbook on the Economics of
Abbink, K., Brandts, J., Herrmann, B., Orzen, H., 2010. Inter-group conflict and intra-group Corruption, Volume 2. Edward Elgar, Cheltenham.
punishment in an experimental contest game. Am. Econ. Rev. 100 (1), 420–447. Schickora, J., 2011. Bringing the four-eye-principle to the lab. Munich discussion paper.
Alatas, V., Cameron, L., Chaudhuri, A., Erkal, N., Gangadharan, L., 2009a. Gender, culture, and Serra, D., 2012. Combining top-down and bottom-up accountability: evidence from a
corruption: insights from an experimental analysis. South. Econ. J. 75 (3), 663–680. bribery experiment. J. Law Econ. Org. 28 (3), 569–587.
Alatas, V., Cameron, L., Chaudhuri, A., Erkal, N., Gangadharan, L., 2009b. Subject pool Serra, D., Wantchekon, L. (Eds.), 2012. New advances in experimental research on corrup-
effects in a corruption experiment: a comparison of Indonesian public servants and tion. Research in Experimental Economics, vol. 15.
Indonesian students. Exp. Econ. 12 (1), 113–132. Shleifer, A., Vishny, R., 1993. Corruption. Q. J. Econ. 108 (3), 599–617.
Amegashie, J., 2013. Consumers' Complaints, the nature of corruption and social welfare. Transparency International, 2010. Africa Education Watch: Good Governance Lessons for
Working Paper. Primary Education.
Apesteguia, J., Dufwenberg, M., Selten, R., 2007. Blowing the whistle. Economic Theory 31 Transparency International, 2012. Corruption Perceptions Index.
(1), 143–166. Transparency International, 2013. Global Corruption Barometer.
Banuri, S., Eckel, C., 2012. Experiments in culture and corruption: a review. Res. Exp. Econ. Tummala, K.K., 2009. Combating corruption: lessons out of India. Int. Public Manag. Rev.
15, 51–76. 10 (1), 34–58.
Banuri, S., Eckel, C., Wilson, R., 2010. Experimental bribery in context: comparing the U.S. van Rijckeghem, C., Weder, B., 2001. Bureaucratic corruption and the rate of temptation:
and Pakistan. Center for Behavioural and Experimental Economic Science. Working do wages in the civil service affect corruption, and by how much? J. Dev. Econ. 65 (2),
Paper #09-01. 307–331.