Brief 13: Reducing Clientelism in Benin

The authors tested mechanisms aimed at reducing clientelism during elections in Benin. They found that interventions aimed at changing how citizens engage with candidates could actually help reduce clientelism.

Link to Full Study

Category: Corruption, Elections

Date of Publication: Friday, March 20, 2015

EGAP Researcher: Leonard Wantchekon

Other Authors: Thomas Fujiwara (2013 co-author)

Click to Download the Data

Geographical Region: Africa

Research Question:

Can presidential candidates run election campaigns in a way that sustainably reduces the role of clientelism in politics?

Preparer: Jasper Cooper



A former French colony, Benin transitioned from an authoritarian to a democratic system in 1989. Despite vibrant elections with widespread participation, Benin has suffered from years of poor economic governance and shortsighted policymaking. According to several researchers, international bodies, and civil society organizations, rampant clientelism has played a role in the creation of this unfortunate situation, in which high-turnout elections produce weak policy outcomes. Yet, while much has been written about how clientelistic practices work and how they hamper development, we still understand little about how they can be curtailed.

Wantchekon’s field experiments in Benin have sought to test the effectiveness of different anti-clientelistic policies, all of which could be feasibly implemented at a broad scale. Although each intervention has employed a slightly different design, the basic formula is the same: with the cooperation of the leading presidential candidates, the researchers randomly assign some villages to the ‘control’ group, which receives the typical clientelistic campaign rallies (featuring handouts of cash and gifts), and other villages to the ‘treatment’ group, which receives a non-clientelistic electoral campaign aimed at producing healthier voter-candidate relationships. Vote-buying practices such as giving gifts or cash are banned from interventions in the treatment group, and broadly beneficial policies (such as national healthcare reform) are promoted and discussed.

Research Design:

In his 2001 experiment, Wantchekon assigned 24 villages to receive one of two kinds of electoral campaigns: a ‘clientelistic’ campaign, with handouts of gifts and promises of material favors in exchange for votes; and a ‘programmatic’ campaign, in which candidates promised to implement broad policy programs that would benefit the nation as a whole. The researchers compared the share of votes obtained by candidates in the treated villages with their results in other villages in which they ran their election campaign as usual. The results of this study suggest that simply abandoning clientelistic practices and campaign messages in favor of broad policy programs is an unsustainable strategy for candidates. Candidates received less of the vote when they campaigned on policy programs and more of the vote when they used clientelistic practices. Despite this, voters who were better informed, female, or co-ethnics of the campaigners were significantly more likely to vote for candidates who had run on programmatic policy platforms.

Wantchekon and Fujiwara explored the latter finding in a second experiment that took place during the presidential elections of 2006. The research team hypothesized that the programmatic policy platforms used in the 2001 experiment may have been unsuccessful against clientelistic campaign strategies because voters were uninformed about the potential benefits of such policies. Therefore, the 2006 intervention engaged voters in town hall-style debates, seeking to increase the ‘demand’ for less-clientelistic policies. In this experiment, the research team organized large public meetings on the main problems facing Benin and how the candidate they were representing planned to resolve them. These meetings stood in stark contrast to the usual campaign rallies run in the control villages, which featured the usual handouts of gifts and cash, as well as a one-sided delivery of the candidate’s platform.

As with the first experiment, Fujiwara and Wantchekon measured how the treatment affected the share of the vote that the experimental candidate won at the village level, and the likelihood that a voter voted for the candidate, given their individual characteristics. The treatment had no significant effect on turnout or on the share of votes that candidates won. This suggests that voters will still turn out to vote even in the absence of direct cash benefits. However, the data tells a different story when one takes into account the dynamics of the election: when the experimental candidate had won the majority of votes in that village in the 2001 election, the town hall meetings decreased his vote share by up to eight percentage points; however, when the candidate had not been dominant in that village in 2001, the treatment increased his vote share by up to 17 percentage points. In other words, the town hall campaign treatment hurt incumbents and helped challengers.

One possible interpretation of this effect is that the anti-clientelistic treatment increases democratic competition: by informing voters about candidates’ policies and their consequences, the town hall meetings may have provided an alternative to the long-running vote-buying arrangements that had shored up votes for candidates in their ‘strongholds’ in previous years.

Wantchekon’s third and most recent experiment in 2011 sought to address these concerns, and provides the most compelling evidence for the sustainability of non-clientelistic campaign strategies. As with the 2006 experiment, the treatment featured town hall style meetings in which villagers were presented with and debated the policy proposals of the experimental candidates. The 2011 experiment featured the largest sample of the three experiments and can be considered as representative of Benin as a whole (with the exception of the capital, Cotonou).

Did Survey Respondent Vote for Experimental Candidate? (OLS estimates)

  Overall Opposition Candidates Incumbent Candidate
Vote took place in treated village 0.0599***(0.0206) 0.0864***(0.0275) -0.0101(0.0156)
N 4529 3285 1244

Did Survey Respondent Consider Experimental Candidate the ‘Best Candidate’? (OLS estimates)

  Overall Opposition Candidates Incumbent Candidate
Vote took place in treated village 0.0663***(0.0212) 0.0999***(0.0270) -0.0256(0.0227)
N 5113 3744 1369
Interpretation: When they ran on the treatment platform, all types of candidates received 6 percentage points more of the vote share as reported in representative surveys of voters in treatment and control villages. This effect is highly statistically significant. Opposition candidates received 9 percentage points more of the self-reported vote share when they ran on the treatment platform as opposed to the control. The incumbent candidate is estimated to have lost one percentage point of the vote share on the treatment platform, although this effect is not statistically distinguishable from zero effect.* p < 0.1, ** p < 0.05, *** p < 0.01Robust standard errors clustered at village level. Regression controls for regional-level fixed effects.

The results support many of the encouraging findings from the 2006 study. The treatment increased turnout by three percentage points, lending further weight to the notion that anti-clientelistic policies do not necessarily adversely impact democratic participation. Relative to the control, candidates running non-clientelistic town hall meetings received 6 percentage points more of the vote. As in the 2006 experiment, this effect was especially true for candidates campaigning outside of their strongholds: opposition candidates won nine percentage points more of the vote share when using the town hall meetings as a campaign strategy. These results strongly suggest that involving voters in deliberation around policy issues can provide a sustainable campaign alternative to clientelistic strategies, especially for candidates in challenger positions.

Policy Implications:
  • In contexts where clientelism is rife, simply abandoning vote-buying practices and clientelistic favors is unlikely to be a sustainable strategy for candidates, who will be at a competitive disadvantage. Any campaign strategy aimed at sustainably reducing clientelism should be competitive against clientelistic tactics.
  • Efforts to make non-clientelistic campaign strategies competitive should focus on the voters. This means not only offering programmatic alternatives to clientelistic arrangements, but also informing and engaging voters in debate over the relative merits of different policy platforms. Town hall style meetings are one effective strategy for fostering this engagement.
  • Consideration should be given to the different ways in which such reforms will impact political competitors. Participatory campaign strategies such as town hall debates are likely to benefit some candidates more than others, in particular those in a challenger position.
  • All efforts to discourage clientelism through election campaigns should attempt to produce net positive democratic outcomes. In particular, any intervention that targets vote-buying should ensure that decreases in participation are kept to a minimum or avoided. Evidence suggests that town hall meetings may even increase participation.
  • Parties or candidates in upcoming elections in the West African region, such as those running in the presidential elections in Nigeria in 2015, might consider implementing participatory, anti-clientelistic campaign strategies in districts where they occupy a challenger position.
  • Many important questions are raised by the electoral success of these anti-clientelistic campaign strategies. New experiments, for example, could employ factorial designs to analyze which kinds of campaign tactics are the most effective against clientelistic alternatives. But some questions could even be answered without the need for new experiments. For example, researchers could explore the long-term consequences of interventions such as Wantchekon’s through downstream analysis. Are policy outcomes in villages treated in previous experiments objectively better by some measure (i.e. number of projects completed, money spent)? Are voters more or less satisfied with such outcomes? How have their perceptions about vote-buying or electoral corruption evolved as a result of the treatment?