Measuring the Impact of Clientelism on Voter Behavior in Benin

Trabajo de campo:
Approximately 234,000 people in 21 villages
Línea de tiempo:
Grupo objetivo:
  • Voters
Resultado de interés:
  • Electoral participation
Tipo de intervención:
  • Information
Número de registro del AEA RCT Registry:
Socios Implementadores:

Voters in Benin had a preference for clientelist political platforms, but certain subsets of voters such as women, consumers of mass media, and members of social organizations were less receptive to clientelism.

Problema de política pública

Politicians often promise to redistribute resources to local constituencies in order to win their votes, even when nationwide programs may be more effective for social, political, and economic development. “Clientelism,” or transactions between politicians and select citizens in which material favors are offered in return for support at the polls, may diminish the resources dedicated to public goods like infrastructure or education and reduce political competition as incumbent politicians use the promise of government jobs and other benefits to gain an advantage over challengers.

Little is known about how clientelist politics affect voter behavior or the degree to which clientelism is effective at generating support among different types of voters. For example, it could be that women, who are less likely to receive government jobs, are less prone than men to support clientelist political platforms, and are therefore more likely to vote for platforms that promise public goods like vaccination campaigns. In that case, initiatives to increase women’s participation in the political process may improve the provision of public goods. Similarly, it is unknown whether policies to promote membership in social organizations would foster a sense of community and increase the demand for national public goods or would strengthen interest groups and their power to demand special benefits. More information on the determinants of voter behavior could inform policies that diminish the demand for clientelist politics and lay the groundwork for more focus on poverty reduction programs.

Contexto de la evaluación

In 1990, Benin's authoritarian regime collapsed after 18 years in power. A new constitution established a multiparty parliamentary system with one legislative assembly and a president elected by popular vote for up to two five-year terms.1 To be elected president, a candidate must win more than 50 percent of the national vote. If no candidate wins a majority, a second round election is held between the two leading candidates to determine a victor. As of 2001, all of Benin’s presidential elections were decided in the second round.

Clientelism is an important feature in African electoral politics. In Benin, state payroll consumed between 65 and 90 percent of the government budget and an estimated 50 percent of public service jobs were allocated to constituents in exchange for political support. This may explain why public service provision was particularly poor. At the time of this study, only 50 percent of the population had access to drinking water and 18 percent had access to basic healthcare despite the economy recording a 4.9 percent average annual growth rate for twelve years.

Detalles de la intervención

During the first round of the 2001 presidential election, researchers tested how clientelist political messages affected voting behavior of those exposed to the messages. Researchers collaborated with four leading political parties, FARD-Alafia, PSD, RB, and UDS and their candidates to develop “clientelist” and “public goods” versions of political messages based on existing party platforms. The study was limited to districts where a single party traditionally dominated the vote, in order to prevent the intervention from influencing election results. Once stronghold districts were identified, two of these districts were randomly selected for each of the four parties, although one district was later dropped from the study because it was won by a different candidate than expected.

One village from each district was randomly selected to receive only clientelist political messages, emphasizing specific projects transferring government resources to the region or village, while another was randomly chosen to receive only public goods messages, which emphasized public goods delivered through national programs. An additional village was selected in each district to serve as a comparison, where political campaigns were conducted as usual. In villages that were selected to receive the intervention, teams of party representatives and a research assistant delivered political messages by visiting influential local figures, attending sporting or cultural events, and organizing public meetings.

An endline survey was conducted 14 months after the election, to gather information on demographic characteristics, voter turnout, knowledge of candidates, and voter choices in the 1996 and 2001 elections.

Resultados y lecciones de la política pública

On average, candidates using public goods messages lost support from voters while candidates using purely clientelist messages mostly maintained their supporters. Voters in villages that were exposed solely to public goods messages were about 25 percent less likely to favor the dominant candidate than in comparison villages. Clientelist messages did not seem to alter voter behavior, likely because political messages in comparison villages were already clientelist in nature.

Gender: Women seemed to have stronger preferences for public goods platforms than men. In villages that were exposed to only public policy messages, women were significantly more likely than men to support the dominant candidate.

Ethnicity: Ethnic ties only influenced voting behavior in villages that were exposed to public goods messages. Voters in public goods villages that had ethnic ties to the dominant candidate were 15 percentage points more likely to support that candidate than voters in treatment villages who did not have ethnic ties to the dominant candidate. Voters sharing an ethnic group with the dominant candidate may have been more receptive to public goods messages because they believed their ethnic group would benefit more than others from a national development plan, or because their ethnic ties would give them a unique ability to monitor the implementation of public goods platforms, which are typically more difficult to monitor than the provision of clientelist goods.

Participation in national social networks: Voters who were well connected with networks outside their local villages, through access to media and membership in political parties, unions, and NGOs, were more likely to support the dominant candidates where they campaigned on public goods platforms. This may be because voters who listen to radio, watch TV, read newspapers and are members of national organizations have a better understanding of the problems faced by other regions, travel more, and have a greater sense of belonging to a national community. Furthermore, exposure to media may make it easier for constituents to monitor the implementation of national platforms.

Political Discussions: Voters who engaged in local political discussions were less receptive of public goods political messages, possibly because local discussions may reinforce local bonding and thus increase demands for redistribution toward the locality.

1. Wantchekon, Leonard. 2003. "Clientelism and Voting Behavior: Evidence from a Field Experiment in Benin." World Politics 55(2003): 399-422. 2. Vermeersch, Christel, and Leonard Wantchekon. "Information, Social Networks and the Demand for Public Goods: Experimental Evidence from Benin." in Accountability through Public Opinion: From Inertia to Public Action, edited by Sina Odugbemi and Taeku Lee, 12-135. Washington, DC: The World Bank, 2011.
“The World Factbook: Benin,” Central Intelligence Agency, accessed April 25, 2013,