Does Corruption Affect Behavior in Ultimatum Games? Evidence from a Collective Action Pilot Study

The role of corruption in fomenting civil unrest has gained increasing attention recently. Does corruption engender action, or is it merely an excuse used to cover for contentious behavior? This research tests this question in a laboratory pilot study using a collective-action ultimatum game. The results show that participants that receive a ‘corrupt’ treatment protest more often than participants that receive a ‘not corrupt’ treatment.

Social movements, rebels, and terrorists often cite government corruption as a motivating factor for their behavior. In 2003, Nigerians took to the streets to protest massive electoral fraud following what many considered to be a rigged election (Human Rights Watch, 2004). Across North Africa and the Middle East, citizens banded together in 2011 and 2012 to challenge their governments during the Arab Spring. In 2015, following weeks of protest focused on contending President Nkurunziza’s attempts to circumvent constitutional term limits, a faction of the army attempted a coup d’état (Vircoulon, 2015).

But assessing the role of corruption in motivating large-scale political events is difficult, particularly when rebels stand to gain political power. Furthermore, corruption, by its nature a clandestine activity, is highly endogenous to flawed democratization, reduced trust in civil society, and institutional weakness. As such, corrupt places tend to be conflictual places. Teasing out the effects of perceptions of corruption are extraordinarily difficult.  This begs the question: does corruption actually motivate civil conflict, or is it simply an excuse to cover ulterior political motives?

In order to test this question, this research uses an ultimatum game that has been modified to allow a group of participants to collectively decide whether to accept or reject an offer made to them by an automated ‘government’. The results of the first trial of the experiment indicate that groups of participants that were exposed to the ‘not corrupt’ treatment were fifteen percentage points less likely to protest than those exposed to either the ‘corrupt’ treatment or control.

Corruption, Contention, and Conflict

The end of the Cold War brought about a renewed scholarly focus on issues of governance, particularly in developing nations. As such, corruption gained increased attention in both the policy and academic worlds. While initial scholarship focused largely on the relationship between corruption and economic development (Bardhan, 1997; Girling, 1997; Heidenheimer, Johnston, & LeVine, 1989; Johnston, 2005; Rock, 2009), more recent work has begun to expand the analysis to include corruption’s deleterious effects on social trust (Rothstein & Uslaner, 2011; Uslaner, 2007, 2008, 2013), as well as civil society responses to corruption (Johnston, 2012, 2014).

A third body of work focuses on the relationship between government corruption and conflict, though the relationship remains unclear. Le Billon (2003) argues that corruption is neither necessary nor sufficient in causing the onset of conflict. Fjelde (2009) finds that corruption may indeed reduce the likelihood of conflict, as elites can ‘buy off’ potential challengers. Neudorfer and Theuerkauf (2014), on the other hand, argue that corruption intensifies the likelihood of ethnic conflict due to the exacerbation of ethnic grievances. Teets and Chenoweth (2009) argue that corruption likely facilitates conflict and terror, but does not itself motivate it. Recent experimental work has endeavored to more clearly tease out the impact of corruption on behavior (De La O & Chong, 2012; De La O, 2013; Figueiredo et al., 2013; Serra & Wantchekon, 2012; Wantchekon, 2003); however, to the author’s knowledge, this work has not extended to cover collective contentious behavior.


Corruption, despite its surreptitious and complex nature, is a salient force in the lives of many citizens across the globe. Acts of financial corruption – kleptocracy and patronage – can halt economic development, exacerbate poverty, and create ethnic grievances. Acts of systemic corruption – electoral fraud – disenfranchises citizens. Significant research has demonstrated that, when faced with government malfeasance and deeply rooted inequalities, citizens act upon their grievances to challenge the government (Buhaug & Gleditsch, 2014; Cederman, Gleditsch, & Buhaug, 2013; Cederman, Weidmann, & Gleditsch, 2011; Gurr, 2011).

Following this logic, the primary argument of this article is that corruption is, in and of itself, a motivator of contentious political action. Because citizens perceive corrupt governments as illegitimate and untrustworthy, they should be more likely to protest perceived slights and deprivations. In regions such as Africa in which citizens worry overwhelmingly about issues of poverty and unemployment, corruption highlights economic plight by generating between-group disparities. This is particularly true, given the patrimonial nature in which not only economic resources, but economic opportunities are distributed. When corruption reigns, ‘losers’ not only lose today, but lose for the foreseeable future.

Hypothesis 1: holding all else equal, citizens will protest against governments perceived to be corrupt than against governments not perceived to be corrupt.

Research Design

In order to test this hypothesis, an experiment was designed to allow groups of participants to accept or protest government offers in a modified collective ultimatum game. Past research (Habyarimana, Humphreys, Posner, & Weinstein, 2004, 2007, 2009) has examined the well-known ‘collective-action problem’ (Lichbach, 1995; Olson, 1965). Drawing from this, a modified ultimatum game was developed that allows groups of up to five participants the opportunity to collectively decide how to respond to varying offers by a government regime. The game is structured as follows.

An ultimatum game software application was developed to run in a web-browser using the Python computing language. The game can accommodate up to five simultaneous players, each of which acts as a ‘citizen’ in a newly democratic regime[1]. In this game, the role of the ‘government regime’ (the part played by the ‘offerer’ in traditional ultimatum games) is controlled by the software application. Participants interact with the game via a computer or tablet device[2], and each participant is issues his or her own device.

At the beginning of the game, the participants are treated with one of three treatments explaining the rules of game. All treatments indicate that the participants are citizens in a newly democratic regime that, in its efforts to promote development, provides each citizen with a stipend of up to 100 ‘shillings’. At each round, the government will offer up to 100 shillings. Participants can either accept the offer or protest (in hopes of gaining the full 100 shillings). The manipulation revolves around a portion of the prompt that indicates whether the government is considered ‘not corrupt’ or ‘very corrupt’, or whether no indication of its corruption is given (the control).

Once participants signal that they have read the prompt, five successive rounds of the ultimatum game begin. In each round, the computer randomly offered between 0 and 100 shillings, at 25 shilling increments. Participants have two options: to accept the offer or to protest the offer in an attempt to earn the full 100 shillings. Failed protests result in payoffs of zero shillings. In order to attempt to address the issue of collective action, the user interface indicates how many other participants have decided to accept or protest the offer. For every participant that decides to protest, the likelihood of the success of that protest increases, and participants can view the probability of success. This attempts to model the cascades effect of contentious action in which the more citizens are protesting, the more likely it is for others to join (Hussain & Howard, 2013; Lohmann, 1994).

Each round ends after all participants have made a selection of whether to protest or accept. After the fifth round, the game terminates. Participants were paid a base participation fee of ten dollars, and then paid an additional amount based on their in-game behavior. The maximum additional amount possible was five dollars. This was modeled in order to encourage participants to consider the potential loss of their earnings upon a failed protest. In total, 46 undergraduate students were recruited, of which 41 (89.13 percent) showed up to participate, resulting in a total of 205 rounds of the ultimatum game being played. Each group arrived on the hour to participate, and no groups overlapped in arriving or departing. It is therefore reasonable to assume that SUTVA was not violated.


The results provide initial, though not conclusive, support for the hypothesis. On the whole, participants that were exposed to ‘not corrupt’ treatment were approximately 15 percentage points less likely to protest than those exposed to the ‘very corrupt’ treatment. This result is substantively significant, and a one-tailed t-test indicates the result to be significant at the 10 percent confidence level[3]. The substantive difference is large enough to assume that, given enough observations, the result would be significant at a higher level of confidence.

Table 1 – Between Group T-Test Comparisons




Not Corrupt


Not Corrupt

Group 1 µ 1.6333 1.5833 1.6333
Group 2 µ 1.5833 1.4588 1.4588
Upper CI -0.0988 -0.0150 0.0363
Lower CI 0.1988 0.2640 0.3127
2-tailed p value 0.5785 0.1416 0.0383
1-tailed p value 0.2983 0.0710 0.0192

Here, the results become more complicated. While the difference between the corrupt and not-corrupt treatments is significant at the ninety percent level, and the difference between the control and the not-corrupt treatments is significant at the 95 percent level[4], the difference between the control and the corruption treatment is both substantively small (five percentage points) and statistically insignificant. This indicates that, to participants, there was little theoretical difference between a newly democratic government and a corrupt newly democratic government. One explanation of the results would indicate that, to the average participant, newly democratizing regimes are synonymous with corrupt governance.

Figure 1 – Protest Behavior Across Treatments


A further examination of the data reveal a second possible explanation of this relationship. The average offer in the control and corrupt treatments are very similar (47.5 and 43 shillings, respectively), whereas the not-corrupt treatment averaged an offer of 55 shillings. However, this relationship is not entirely direct. While protest rates are highest in the control group, the lowest average offer is in the corrupt group. Therefore, while the higher average offer in the not-corrupt group may explain lower levels of protest, the average offer does not explain the relationship between the control and corruption groups.


Does corruption affect behavior in collective ultimatum games? In short, it is too soon to tell. While the initial data analysis indicates that levels of perceived corruption do matter, the sample is too small to make any conclusions. Furthermore, the likelihood that the large variation in average amounts is at least partially confounding the results. Future trials will ensure that all rounds offer the same average amount in order to remove this from consideration. Furthermore, future trials will likely remove the control group, leaving only the ‘very corrupt’ and ‘not corrupt’ treatments, mainly because of the scarcity of data.

[1] In this trial, the maximum number of participants was capped at three.

[2] Any modern computing device with a web browser will work. In this case, participants used Amazon Fire tablet computers.

[3] The p-value for a two-tailed t-test between corrupt and not-corrupt is 0.1416. Because the hypothesis is directional, however, a one-tailed test is more appropriate. This splits the p-value in half, resulting in a p-value of 0.0708.

[4] 0.03832 for the two-tailed t-test.


Bardhan, P. (1997). Corruption and Development: A Review of Issues. Journal of Economic Literature, 35(3), 1320–1346.

Buhaug, H., & Gleditsch, K. S. (2014). Contagion or Confusion ? Why Conflicts Cluster, 52(2), 215–233.

Cederman, L.-E., Gleditsch, K. S., & Buhaug, H. (2013). Inequality, Grievances, and Civil War. New York, NY: Cambridge University Press.

Cederman, L.-E., Weidmann, N. B., & Gleditsch, K. S. (2011). Horizontal Inequalities and Ethnonationalist Civil War: A Global Comparison. American Political Science Review, 105(03), 478–495.

De La O, A. L. (2013). Do Conditional Cash Transfers Affect Electoral Behavior? Evidence from a Randomized Experiment in Mexico. American Journal of Political Science, 57(1), 1–14.

De La O, A. L., & Chong, A. (2012). Exposing Corruption on Electoral Outcomes.

Figueiredo, M. F. P. De, Hidalgo, F. D., Collier, D., Collier, R. B., Figueiredo, J. De, O, A. D. La, … Connell, A. J. O. (2013). When Do Voters Punish Corrupt Politicians ? Experimental Evidence from Brazil, (312), 1–40.

Fjelde, H. (2009). Buying Peace? Oil Wealth, Corruption and Civil War, 1985-99. Journal of Peace Research, 46(2), 199–218.

Girling, J. L. S. (1997). Corruption, capitalism and democracy. Routledge studies in social and political thought 4.

Gurr, T. R. (2011). Why Men Rebel (40th Anniv). Boulder, CO: Paradigm Publishers.

Habyarimana, J., Humphreys, M., Posner, D. N., & Weinstein, J. M. (2004). Group Preferences or Group Strategies ? Untangling the Determinants of Successful Collective Action Among Ethnic and Gender Groups, (May), 7–8.

Habyarimana, J., Humphreys, M., Posner, D. N., & Weinstein, J. M. (2007). Why Does Ethnic Diversity Undermine Public Goods Provision? The American Political Science Review, 101(4), 709–725.

Habyarimana, J., Humphreys, M., Posner, D. N., & Weinstein, J. M. (2009). Coethnicity: Diversity and Dilemmas of Collective Action. New York, NY: Russell Sage Foundation.

Heidenheimer, A. J., Johnston, M., & LeVine, V. T. (1989). Political Corruption: A Handbook. Transaction Publishers.

Human Rights Watch. (2004). Nigeria’s 2003 Elections: The Unacknowledged Violence. Retrieved from

Hussain, M. M., & Howard, P. N. (2013). What Best Explains Successful Protest Cascades? ICTs and the Fuzzy Causes of the Arab Spring. International Studies Review, 15(1), 48–66.

Johnston, M. (2005). Syndromes of Corruption: Wealth, Power, and Democracy. New York, NY: Cambridge University Press.

Johnston, M. (2012). Building a Social Movement Against Corruption. Brown Journal of World Affairs, 18(2), 57–74. Retrieved from,shib&db=bth&AN=85090608&site=ehost-live&custid=s4121186

Johnston, M. (2014). Corruption, Contention, and Reform. New York, NY: Cambridge University Press.

Le Billon, P. (2003). Buying peace or fuelling war: the role of corruption in armed conflicts. Journal of International Development, 15(4), 413–426.

Lichbach, M. I. (1995). The Rebel’s Dilemma. Ann Arbor, MI: University of Michigan Press.

Lohmann, S. (1994). The Dynamics of Informational Cascades: The Monday Demonstrations in Leipzig, East Germany, 1989-94. World Politics, 47(1), 42–101.

Neudorfer, N. S., & Theuerkauf, U. G. (2014). Buying War Not Peace The Influence of Corruption on the Risk of Ethnic War. Comparative Political Studies, 0010414013516919.

Olson, M. (1965). The Logic of Collective Action. Cambridge, MA: Harvard University Press.

Rock, M. T. (2009). Corruption and Democracy. Journal of Development Studies, 45(1), 55–75.

Rothstein, B., & Uslaner, E. M. (2011). All for All: Equality, Corruption, and Social Trust. World Politics, 58(01), 41–72.

Serra, D., & Wantchekon, L. (2012). Experimental Research on Corruption: Introduction and Overview. In D. Serra & L. Wantchekon (Eds.), New Advances in Experimental Research on Corruption (pp. 1–12). Bingley, UK: Emerald Books.

Teets, J. C., & Chenoweth, E. (2009). To Bribe or ot Bomb: Do Corruption and Terrorism Go Together? In R. I. Rotberg (Ed.), Corruption, Global Security, and World Order (pp. 167 – 193). Washington, DC: Brookings Institution Press.

Uslaner, E. M. (2007). Corruption and the Inequality Trap in Africa (No. 69).

Uslaner, E. M. (2008). Corruption, Inequality, and the Rule of Law. New York, NY: Cambridge University Press.

Uslaner, E. M. (2013). Trust and corruption revisited: How and why trust and corruption shape each other. Quality and Quantity, 47(6), 3603–3608.

Vircoulon, T. (2015). Burundi’s Coup from Within. Retrieved January 6, 2016, from

Wantchekon, L. (2003). Clientelism and Voting Behavior: Evidence from a Field Experiment in Benin. World Politics, 55(3), 399–422.




You are a citizen of a newly formed country. This country is an emerging democracy, and has just completed a round of relatively fair elections. The president was elected by direct vote. Most election monitors concluded that this was a flawed, but fair election.


Though the country is underdeveloped, citizens are hopeful that the situation will improve. Due to a combination of taxes, natural resource exports, and humanitarian aid, the government is able to provide money to each of its citizens. This money is intended to help in the development of the country, including building better schools, roads, electricity, and other infrastructure.

In each turn, the government will offer you up to 100 Shillings. If you receive 100 Shillings, that means the government is giving you 100 percent of your funds. If you receive less than 100 Shillings, it likely means that the government is using some of your development funding for other projects, or keeping it for itself.




The country has a relatively equal distribution of mountains, rivers, lakes, and plains. English is the primary language. The economy is mostly agricultural, with some natural resource extraction and a growing manufacturing sector.


Many citizens consider the government to be very corrupt. People talk of government officials stealing state funds and putting the money into their private bank accounts. Others say that certain ethnic groups are favored and receive lots of public goods, while less favored groups receive almost none.


Most citizens agree that the government is not very corrupt. People generally believe that officials use state funds properly. Most agree that, even though the population is ethnically mixed, each group received generally equal distributions of public goods.

Descriptive Statistics

Table 2 – Gender Distribution Across Treatments

  Control Corrupt Not Corrupt
Male 4 5 9
Female 8 7 8


Table 3 – Protest Behavior Across Gender

  Accept Protest
Male 40 50
Female 53 62


Table 4 – Number of Rounds

  Number of Rounds
Control 30
Corrupt 25
Not Corrupt 30