Principal investigator: Eduarda Miller de Figueiredo
Article title: Vulnerability and Clientelism
Article authors: Gustavo J. Bobonis; Paul Gertler; Marco Gonzalez-Navarro; Simeon Nichter;
Location of the interventionSemi-arid region – Northeast Brazil.
Sample size: 1.308 families – 615 families in the treatment group and 693 families in the control group.
SectorEconomic Policy and Governance
Type of interventionConstruction of cisterns for collecting rainwater or for storing water that comes from tanker trucks.
Main variable of interestRequests for goods from politicians and vulnerability indices.
Assessment methodExperimental Evaluation (RCT)
Evaluation Context
Clientelism is the term used to describe the distribution of goods or services in exchange for political support, and is a very important reason why many politicians are elected (KICKEN 2011; STOKES ET AL, 2013). The literature also argues that clientelism exacerbates governmental allocative inefficiencies, in addition to harming the functioning of democratic institutions and leading to a low provision of public goods. Previous studies show evidence that people with low socioeconomic status are more likely to participate in clientelism.
The Northeast region of Brazil is known as a semi-arid zone with over 28 million inhabitants, disproportionately poor and living in rural areas. Furthermore, it is characterized by recurring droughts, with an average rainfall of only 57,2 cm in 2012 compared to 153,1 cm for the rest of the country. Previous studies have recognized these droughts as a key factor in the region's vulnerability.
The study claims that mayors and councilors in rural northeastern Brazil tend to favor citizens with whom they maintain ongoing clientelistic relationships, where material benefits are exchanged for political support. Therefore, the authors propose to conduct a randomized experiment to test the hypothesis that reducing the vulnerability of families decreases the demand for private goods from politicians.
Intervention Details
The intervention consisted of building cisterns in the homes of randomly selected families in the Northeast region, aiming to reduce their vulnerability. Thus, the experiment was applied in rural homes that lacked potable water or a cistern, had the physical space for cistern construction, and had roofs made of metal sheets or tiles to facilitate rainwater harvesting.
The cisterns are connected to the roof of the house by pipes and gutters, allowing for rainwater collection. They can store up to 16.000 liters of water, so in addition to being a technology for collecting water, they also serve to store water supplied by tanker trucks. Thus, the cisterns will reduce the vulnerability of families.
Methodology Details
To create the control and treatment groups, 40 municipalities were first randomly selected across the nine states of the semi-arid region, where each city had an average of 49 citizens. Proportional weights were used for the number of residences without access to piped water and cisterns, according to the Unified Registry. Then, also randomly, groups of neighboring rural families were selected, where up to six families were interviewed.
In these selected families, three phases of in-person research were conducted. The first phase took place in the last quarter of 2011 to collect data on pre-intervention domestic and individual characteristics. Immediately after the municipal elections of October 2012, the second phase was carried out, where new questions were asked and those from the first phase were repeated, making it possible to capture a possible effect of the treatment. Finally, the last data collection took place in the last quarter of 2013, aiming to capture effects during a non-election period.
From this sample, the authors found that 21,3% of respondents requested private goods from a mayoral or city council candidate during the 2012 election year, and in a non-election year, 8,6% of respondents made such requests. Of these, 7% of respondents had their requests fulfilled by candidates who were in office. Importantly, the requests were motivated by vital needs, with a quarter of them involving water.
The data from the 2012 municipal election were extracted from the Superior Electoral Court of Brazil (TSE), which publishes aggregated data from electronic voting machines, as well as their geographic locations.
As a marker of clientelism, it was observed whether the interviewee spoke at least monthly with a local politician before the 2012 election campaign, given that this is a hallmark of continuous exchange relationships. In addition, a marker was used for when the interviewee spoke with the local politician monthly before the election campaign and publicly declared their support for the candidate. These measures, as can be seen in the graph above, are balanced between the treatment and control groups, demonstrating that it will be possible to observe whether there is a relationship between clientelism and vulnerability.
Thus, a sample of 1.308 families was constructed, of which 615 families were randomly selected for the treatment group and 693 for the control group. The authors observed the intention-to-treat effects, that is, their analyses compared the treatment group – which received the cistern intervention – with the control group.
Results
For the families treated, an improvement was found in depressive symptoms, the food security index, and self-reported health status. Furthermore, the overall index that standardizes these last three reported components demonstrates a substantial reduction in vulnerability. Therefore, the results show that the cistern program reduced the vulnerability of the research participants.
The research also demonstrated that the intervention ultimately reduced the likelihood of citizens requesting private goods from politicians by 3 percentage points. And, when observed for those citizens who were in a clientelistic relationship, there was a reduction of 10,9 percentage points in requests.
Similarly, it was found that reducing vulnerability through the cistern program negatively impacts the performance of incumbent mayors seeking reelection, significantly reducing the votes they receive. Furthermore, it was verified that these reductions occurred both during the election period (2012) and in the post-election period (2013), suggesting that the effect is persistent and long-term in the relationship between citizens and politicians.
Thus, after presenting these results, the authors concluded that the intervention of the cisterns provided to the treatment group reduced the number of requests for private services or goods from local politicians by citizens. This occurred because the program mitigated the vulnerability of the population in the semi-arid northeast of Brazil.
Lessons in Public Policy
In light of what has been presented here, it is evident that vulnerability is a determining factor in clientelism. Where improving the livelihoods of citizens will reduce participation in exchanging benefits for political support.
Since clientelism exacerbates governmental allocative inefficiencies, as argued earlier, mitigating it could contribute to a reduction in these allocative inefficiencies on the part of the government, as well as improve the functioning of democratic institutions.
Reference
Bobonis, Gustavo J., Paul Gertler, Marco Gonzalez-Navarro, and Simeon Nichter. “Vulnerability and Clientelism.” NBER Working Paper No. 23589, July 2017.
