Reducing vulnerability, curbing clientelism: A case study in Brazil


Published 13.06.23

Reducing vulnerability can combat clientelism: Access to rain-fed water cisterns in Brazil decreased citizen requests and votes for incumbents

Clientelism: A global challenge

In many countries, democracy fails to represent the interests of poor and vulnerable citizens. Substantial research suggests that clientelism — the exchange of contingent benefits for political support — is an important reason why many elected politicians are neither accountable nor responsive to their constituencies. Voters frequently receive campaign handouts or ongoing benefits in direct exchange for their political support. Among its numerous consequences, many argue that clientelism exacerbates governmental allocative inefficiencies and undermines democratic institutions, leading to reduced political competition as well as the under-provision of public goods and social insurance (e.g. Baland and Robinson 2008, Anderson et al. 2015).

Clientelism is especially prevalent in low-income countries, and within nations, politicians disproportionately distribute clientelist benefits to poor citizens (Kitschelt 2011, Stokes et al. 2013). Although experts long believed that clientelism would diminish as countries modernised, it persists in much of the world. A cross-national survey suggests that clientelism exists in over 90% of countries, with “moderate” or “major” clientelist efforts in 74% of nations (Kitschelt 2013). In addition, 19% of survey respondents in Africa reported receiving offers of benefits in exchange for their votes (Afrobarometer 2021), as did 10% in Latin America (LAPOP 2018).

As argued by Nichter (2018), citizens play an important role in sustaining clientelism. When countries fail to provide sufficient social safety nets, citizens are often motivated by their vulnerability to participate in “relational clientelism” — ongoing exchange relationships that extend beyond election campaigns. Economic vulnerability is a concept that encompasses both poverty and risk, as both low average income and high uncertainty can reduce a citizen’s welfare (Ligon & Schechter 2003). Many of the world’s citizens continue to be exposed to numerous sources of uncertainty, including drought, illness, and unemployment.

Clientelism thus has important policy implications and raises two important, unexplored questions. First, is there in fact a causal link between economic vulnerability and citizens’ participation in ongoing clientelist relationships? And second, if vulnerability is indeed a cause of clientelism, what are the electoral consequences of reducing vulnerability? In order to explore these questions, we designed a randomised control trial to reduce household vulnerability. More specifically, our development intervention constructed residential water cisterns in drought-prone areas of Brazil.

Context of the evaluation

Our study focused on Brazil’s semi-arid zone, the vast majority of which is located in the country’s Northeast region. The zone spans over one million square kilometres, and its population of over 28 million residents is disproportionately low-income and rural. A fundamental source of vulnerability is the region’s exposure to recurring droughts. With limited financial resources and poor access to healthcare, the region’s 28 million residents have scant tools to protect themselves from economic and health shocks.

Local politicians in Brazil, including mayors and councillors, have substantial discretion over local public spending, and access to local services is often contingent on political support. Clientelism is a longstanding feature of Brazilian politics. In our study’s context, we found that voters were more likely to publicly declare support for politicians and more likely to request private help from politicians when they experienced droughts, suggesting that vulnerability increases clientelism in this region.

Details of the intervention

In partnership with the NGO Brazilian Semi-Arid Articulation (ASA), we conducted a randomised evaluation of a household water cistern construction programme to test the impact of economic vulnerability on clientelism. In particular, 615 households were randomly assigned to receive water cisterns for free; the other 693 households formed the comparison group and were assigned not to receive cisterns during our study period. ASA constructed the cisterns starting in January 2012, before Brazil’s October 2012 municipal elections.

The rain-fed water cisterns were intended as a strategy for low-income, rural households to cope with droughts by allowing them to collect and store rainfall from the roofs of their houses. The cisterns aimed to reduce households’ vulnerability to water shortages and their need to depend on politicians for water supplies.

We conducted two follow-up surveys after the cisterns were constructed. The first follow-up survey occurred immediately after the October 2012 municipal elections, shedding light on effects of reduced vulnerability during an election year. The second follow-up survey occurred one year later, revealing effects of reduced vulnerability in a non-election year.

Results and policy lessons

As shown in Bobonis et al. (2022), the water cistern programme reduced economic vulnerability, decreased clientelism, and negatively affected the performance of incumbent mayors vying for re-election.

Impact on Vulnerability: On an index of vulnerability including measures of depression, child food security, and self-reported health, households assigned to receive water cisterns were 0.13 standard deviations less vulnerable than households in the comparison group.

Impact on Clientelism: Citizens in the cisterns treatment group — that is, those in households randomly assigned to receive water cisterns — were significantly less likely to request private goods from politicians. The intervention reduced their likelihood of making such requests by 3.0 percentage points, a substantial decline of 17%. This finding is robust to excluding water requests, which were directly affected by cisterns. We also found that the cisterns treatment — a technology that reduces vulnerability — decreased citizens’ requests not only during the election campaign, but also during the year after the election. We found that effects of the cisterns treatment were fully concentrated among citizens who were likely to be in ongoing clientelist relationships: their requests fell by 10.9 percentage points, a remarkable 38% reduction in proportional terms. By contrast, we found no effect among citizens without such relationships.

Impact on Voting Behaviour: We investigated whether decreased vulnerability renders citizens less likely to vote for incumbents, who typically have more resources for clientelism. Because our randomised control trial was designed to reduce vulnerability at the household level, we were able to leverage extraordinarily granular data on voting outcomes. Our survey linked individual subjects in the cisterns experiment to their specific electronic voting machines in Brazil’s 2012 municipal elections. The cisterns treatment was estimated to decrease a citizen’s probability of voting for the incumbent mayor running for re-election by 10.1 percentage points. As with requests, electoral effects were concentrated among citizens likely to be in clientelist relationships. While not definitive, these results are consistent with the argument that reduced vulnerability makes citizens less beholden to incumbent politicians, in that they may be less reliant on clientelist relationships as a risk-coping mechanism.

Taken together, these results suggest that economic vulnerability contributes to clientelism in contexts where citizens do not have adequate resources or insurance to protect against risk. A broader policy implication is that clientelism may be perpetuated by inadequate welfare states. In many countries, welfare states are underdeveloped and truncated, reaching only a subset of the population. Relational clientelism frequently enables unprotected citizens to cope with adverse shocks. But even if poverty declines, clientelism may continue unabated until a welfare state emerges that adequately protects citizens from these shocks. Overall, our study points to reducing citizen vulnerability as an important tool to fight clientelism.


Afrobarometer (2021), Data based on Round 8 data collected from 2019-2021, Available at

Anderson, S, P Francois, and A Kotwal (2015), “Clientelism in Indian Villages.” American Economic Review 105(6): 1780–816.

Baland, J M, and J Robinson (2008), “Land and Power: Theory and Evidence from Chile.” American Economic Review 98(5): 1737–65.

Bobonis, G J, P Gertler, M Gonzalez-Navarro, and S Nichter (2022), “Vulnerability and Clientelism.” American Economic Review 112(11): 3627-3659.

Kitschelt, H (2011), Clientelistic Linkage Strategies: A Descriptive Exploration. Paper Prepared for the Workshop on Democratic Accountability Strategies, Duke University.

Kitschelt, H (2013), Democratic Accountability and Linkages Project. Durham, NC: Duke University.

Latin American Public Opinion Project (LAPOP) (2018), The AmericasBarometer. Data based on surveys collected from 2014-2018, Available at

Ligon, E, and L Schechter (2003), “Measuring Vulnerability.” Economic Journal 113(486): C95–C102.

Nichter, S (2018), Votes for Survival: Relational Clientelism in Latin America. Cambridge, UK: Cambridge University Press.

Stokes, S C, T Dunning, M Nazareno, and V Brusco (2013), Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics. New York: Cambridge University Press.