Skip to main content


Log in

The Most Vulnerable Poor: Clientelism Among Slum Dwellers

  • Published:
Studies in Comparative International Development Aims and scope Submit manuscript


Are slum dwellers more involved in clientelistic arrangements than other (urban poor) voters? While poverty is a key predictor of clientelism, some urban poor voters are more involved in clientelistic arrangements than others. Insecure tenure, lack of access to public resources, and location in areas exposed to environmental shocks increase the vulnerability of slum dwellers. This vulnerability is used by politicians and brokers, who politicize access to scarce resources, and thus make slum dweller more exposed to clientelism. The qualitative literature has long highlighted how clientelism provides a strategy for slum dwellers to cope with their vulnerability, but this population is often excluded from quantitative analyses of clientelism. Using survey data from Argentina and a matching technique that allows us to compare slum dwellers with similar non-slum dwellers, we find that there is indeed a higher prevalence of clientelism among the former. We use a survey experiment on monitoring and sanctions to show that this different exposure to clientelism is consequential. We find different responses across similarly poor slum dwellers and non-slum dwellers regarding the potential consequences of defecting from clientelistic arrangements. Our findings suggest that including slum dwellers in quantitative analyses would improve our understanding of clientelism.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Subscribe and save

Springer+ Basic
EUR 32.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or Ebook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others


  1. The association between poverty and clientelism has been explained by the marginal value of handouts, the scarcity of labor market opportunities, shorter time horizons of poorer voters, or dependence on political discretion to access scarce resources; all of which increase the current value of immediate assistance over uncertain policy promises for the future (e.g., Auyero 2001; Calvo and Murillo 2004; Holland 2017; Kitschelt and Wilkinson 2007a; Mares and Young 2016; Stokes et al. 2013; Weitz-Shapiro 2014).

  2. A slum is a “contiguous settlement that lacks one or more of the following five conditions: access to clean water, access to improved sanitation, sufficient living area that is not overcrowded, durable housing, and secure tenure” (UN-Habitat 2016, 57). See next section for more details.

  3. The slum we study appears in the municipal map as a large empty spot.

  4. Half of the requests for substitution of sampling points received in the 2016/17 Latin American Public Opinion Project (LAPOP), the biggest ongoing survey in the Americas, from local polling firms implementing the survey were for security reasons. Other reasons included abandoned locations, commercial areas, and areas inaccessible due to flooding (Personal communication with Noam Lupu, LAPOP Associate Director, July 3, 2018). The survey manual for the Afrobarometer, the biggest ongoing survey in Africa, in turn, states that: “In some cases, a few EAs [Census Enumeration Areas] may be so inaccessible or so dangerous that substitution becomes necessary” (Afrobarometer 2017, p. 34). Substitutions should not exceed 5% of EAs and should be done with other EAs with similar characteristics (except the reason that generated the substitution). In Argentina, we talked to three of the main national polling firms and they confirmed the difficulties in surveying slum dwellers, who are thereby excluded from samples.

  5. In fact, our pilot survey had to be postponed because the slum was flooded.

  6. See, for instance, Brusco et al. (2004), Calvo and Murillo (2013, 2019), and Stokes (2005) in Argentina; González-Ocantos et al. (2012; 2014), Holland and Palmer-Rubin (2015), and Schaffer and Baker (2015) in Latin America; and Jensen and Justesen (2014) and Kramon (2019) in Africa.

  7. On the 53(3) SCID special issue, Auerbach et al. (2018) discuss the multiple problems associated with the study of informal settlements around the developing world, suggesting that the difficulties we highlight here for the study of clientelism may apply to many other issues in other developing countries as well.

  8. Although our slum sample is not a representative sample of Argentinean slum dwellers, it provides an initial view into the political reality of this understudied population. The slum we study, however, is fairly representative in terms of its characteristics. See Table A2 in the Online Appendix and the next section.

  9. This report is based on mapping all cities with more than 10,000 inhabitants. Informal settlements are defined by the NGO TECHO (2016, p.12) as groups of at least 8 families in which more than half of the population does not have deeds certifying property rights over the land and no regular access to at least two of the most basic public services (running water, sewers, and/or electricity with an individual electric meter). Slums are a subtype of settlement characterized by high population density and irregular urban layout. These are the most common types of informal settlements in the area we study.

  10. A more recent official report identifies 4228 informal settlements in cities over 10,000 inhabitants (39% of them in Buenos Aires), and estimates that around 3.5 million people (around 9% of the population) live in informal settlements (ReNaBAP 2017).

  11. GBA refers to the 24 municipalities that are closer to the City of Buenos Aires, excluding the city itself. See a map of this area (Figure A1) in the Online Appendix.

  12. Table A2 in the Appendix summaries the characteristics of Argentinean informal settlements in comparison with the slum we surveyed.

  13. According to the priest from the slum where we conducted the survey, all his young parishioners use non-slum relatives’ addresses when applying for jobs.

  14. For the effect of droughts on vulnerability and clientelism, see also Bobonis et al. (2017).

  15. When the slum we study floods, for instance, brokers provide clothes, mattresses, and medicines to residents. A broker was reported to have rescued people with his own boat and another one to have delivered mattresses for 100 slum residents relocated to the church during a flood (personal communication with the slum priest).

  16. The debate about brokers’ capacity (and need) to monitor voters’ behavior is not settled. Clients may comply with the clientelistic agreement because they are afraid of punishment if they fail to deliver the requested political support (e.g., Brusco et al. 2004; Stokes 2005; Stokes et al. 2013; Weitz-Shapiro 2014). In this case, brokers’ capacity to monitor voting behavior—or making clients believe so, even with secret ballot (Chandra 2007; Kitschelt and Wilkinson 2007a)—becomes fundamental. Alternatively, brokers may monitor visible political support, such as turnout (Nichter 2008) or rally attendance (Stokes et al. 2013; Szwarcberg 2015) to evaluate the loyalty of their clients, or they can monitor collective behavior at the polling station (Cooperman 2019; Gingerich and Medina 2013; Rueda 2017). Monitoring, however, is not necessary if clients support brokers due to feelings of reciprocity (Finan and Schechter 2012; Lawson and Greene 2014; Scott 1972) or if voters perceive such support as part of their self-interest to maintain the flow of resources (Auerbach and Thachil 2018; Diaz-Cayeros et al. 2016; Oliveros 2021; Zarazaga 2014, 2015). See González-Ocantos and Oliveros (2019) and Hicken & Nathan (2020) for a discussion of this debate.

  17. After reviewing all publications on clientelism from 2008 to 2018, the only studies found by Hicken and Nathan (2020) showing concrete evidence of monitoring electoral behavior were historical or non-fully democratic cases.

  18. Whereas an alternative explanation would be that monitoring turnout is easier than monitoring vote choice in the slum, but not outside the slum, we find no theoretical argument to support this interpretation. Moreover, the slum dwellers in our sample reported voting in around 30 different schools outside the slum (the assignment of polling places is based both on the address of the voter and the first letter of his/her last name).

  19. Information on APES 2015 can be found on; more information on the slum survey can be found in the Appendix.

  20. To compensate for sample attrition, a refresh sample was drawn, selected according to the same procedures used for the first wave.

  21. Table A3 in the Appendix shows the sample representativeness for the APES survey and the comparison with our slum sample for age, education, and gender.

  22. We use the first wave because some of the questions analyzed here were not included on the second wave.

  23. This is an imperfect methodology since houses may contain more than one household—particularly in slums—but it is a reasonable, feasible solution to the challenge of drawing a representative sample from these communities.

  24. See Table A1 in the Appendix for more information. 35.7% of the population of GBA was poor in 2016, compared to 32.5% for the entire country (Source: with data from EPH–INDEC, 2016).

  25. The province of Buenos Aires has a population density of 50.8 people per km2 and Argentina, of 10.7 people per km2.

  26. Question wording is in the Appendix.

  27. See Kitschelt and Wilkinson (2007b, pp. 323–327) for a discussion.

  28. To protect the privacy of the responses, it is crucial to avoid lists that would result in respondents choosing none or all the items, generating “floor” or “ceiling” effects, respectively. To minimize ceiling effects, we included one-low prevalence activity (being a candidate); to minimize floor effects, we included two high-prevalence activities (saw campaign posters and saw campaign adds on TV and radio). The strategy seemed successful since very few of the respondents who received the control list reported either zero or four of the control items. To test the validity of the experiment, we used the method developed by Blair and Imai (2012), and we failed to reject the null hypothesis in the test (ict.test, rejection criteria of ≤ 0.05) for design effects. Table A8 in the Appendix reports the distribution of responses across groups for the list experiment estimates; the experiment wording is reported on pages 11–12 of the Appendix. For advice on designing list experiments, see Glynn (2013).

  29. Tables A4-A7 in the Appendix show the distribution across the different treatment conditions and balance on pre-treatment characteristics for both surveys.

  30. There were then 2 × 2 × 2 = 8 randomly assigned vignettes. For the national survey, randomization was programmed into the PDAs that the enumerators used to administrate the survey. For the slum survey, enumerators received printed questionnaires with the different vignettes.

  31. For an overview of this method, see Zubizarreta et al. (2014) and Zubizarreta and Keele (2017); for a discussion of its advantages, see Visconti and Zubizarreta (2018); for an application to economic voting, see Visconti (2017).

  32. We use the designmatch package available in CRAN (Zubizarreta and Kilcioglu 2016). To conduct the optimization, we use the Gurobi 9.0.0 solver.

  33. The details for each covariate are included in Table A9 in the Appendix. Covariates for the treatment assignment of the experiments were also included to make sure groups remained balanced.

  34. For advice on determining which covariates to include in the matching procedure, see Stuart (2010).

  35. Regarding missing values for covariates, we impute the median and generate binary variables for missingness. These variables indicating missing values are also included in the mean balance optimization.

  36. Standardized differences of greater than 0.25 generally indicate serious imbalance in covariates (see, for instance, Stuart 2010).

  37. To preserve the balance across groups in the survey experiments, the matching procedure includes the variables for treatment assignment.

  38. Full table can be found in the Appendix (Table A10).

  39. The percentage of non-slum residents in our matched sample reporting Self-reported clientelism is 5.56%, compared to 11.11% reported by the slum residents in the matched sample. The percentage of non-slum residents in the matched sample reporting Witnessed clientelism is 30.34%; compared to 51.71% reported by the slum residents.

  40. The 23 percentage point’s difference between slum respondents and GBA APES respondents (columns 2 and 3) is significant at the 99% level.

  41. The 15 points difference is, however, not significant—probably due to the low number of observations.


  • Afrobarometer. (2017). Round 7 Survey Manual.

  • Auerbach A. Clients and communities: the political economy of party network organization and development in India’s urban slums. World Polit. 2016;68(1):111–48.

    Article  Google Scholar 

  • Auerbach A, Thachil T. How Clients select brokers: competition and choice in India’s slums. Am PolitSci Rev. 2018;112(4):775–91.

    Article  Google Scholar 

  • Auerbach A, Thachil T. Cultivating Clients: reputation, responsiveness, and ethnic indifference in India’s slums. Am J PolitSci. 2020;64(3):471–87.

    Article  Google Scholar 

  • Auerbach A, LeBas A, Post AE, Weitz-Shapiro R. State, society, and informality in cities of the Global South. Stud Comp Int Dev. 2018;53(3):261–80.

    Article  Google Scholar 

  • Auyero J. Poor people’s politics: peronist survival networks and the legacy of Evita. Durham: Duke University Press; 2001.

    Google Scholar 

  • Blair G, Imai K. Statistical analysis of list experiments. Polit Anal. 2012;20(1):47–77.

    Article  Google Scholar 

  • Bobonis G, Gertler P, González-Navarro M, & Nichter S. (2017). Vulnerability and clientelism. Unpublished manuscript.

  • Brusco V, Nazareno M, Stokes S. Vote buying in Argentina. Lat Am Res Rev. 2004;39(2):66–88.

    Article  Google Scholar 

  • Calvo E, Murillo MV. Who delivers? Partisan clients in the Argentine electoral market. Am J PolitSci. 2004;48(4):742–57.

    Article  Google Scholar 

  • Calvo E, Murillo MV. When parties meet voters: assessing political linkages through partisan networks and distributive expectations in Argentina and Chile. Comp Pol Stud. 2013;46(7):851–82.

    Article  Google Scholar 

  • Calvo E, Murillo MV. Non-policy politics: richer voter, poorer voter, and the diversification of parties’ electoral offers. New York: Cambridge University Press; 2019.

    Book  Google Scholar 

  • Census. Censo Nacional de Población, Hogares y Viviendas 2010. República Argentina: Instituto Nacional de Estadística y Censos; 2010.

  • Chandra K. Counting heads: a theory of voter and elite behavior in patronage democracies. In: Kitschelt H, Wilkinson S, editors. Patrons, clients, and policies: patterns of democratic accountability and political competition. New York: Cambridge University Press; 2007. pp. 84–109.

  • Collier D. Squatters and oligarchs: authoritarian rule and policy change in Peru. Baltimore: The Johns Hopkins University Press; 1976.

  • Cooperman A. (2019). Trading favors: local politics and development in Brazil [PhD Dissertation].

  • Cornelius W. Politics and the migrant poor in Mexico city. Stanford: Stanford University Press; 1975.

  • Dawson HJ. Patronage from below: political unrest in an informal settlement in South Africa. Afr Aff. 2014;113(453):518–39.

    Article  Google Scholar 

  • De La O A. Crafting policies to end poverty in Latin America. New York: Cambridge University Press; 2015.

    Book  Google Scholar 

  • Departamento Judicial de San Martín. (2012). Homicidios dolosos 2011. Corte Suprema de la República Argentina.

  • Diaz-Cayeros A, Estevez F, Magaloni B. The political logic of poverty relief: electoral strategies and social policy in Mexico. New York: Cambridge University Press; 2016.

    Book  Google Scholar 

  • Finan F, Schechter L. Vote-buying and reciprocity. Econometrica. 2012;80(2):863–81.

    Article  Google Scholar 

  • Garay C. Social policy expansion in Latin America. New York: Cambridge University Press; 2016.

    Book  Google Scholar 

  • Gay R. Popular organization and democracy in Rio De Janeiro: a tale of two favelas. Philadelphia: Temple University Press; 1994.

  • Gingerich DW, Medina LF. The endurance and eclipse of the controlled vote. Econ Politics. 2013;25(3):453–80.

    Google Scholar 

  • Glynn AN. What can we learn with statistical truth serum? Design and analysis of the list experiment. Public Opin Q. 2013;77(S1):159–72.

    Article  Google Scholar 

  • González-Ocantos E, Kiewiet de Jonge C, Meléndez C, Osorio J, Nickerson DW. Vote buying and social desirability bias: experimental evidence from Nicaragua. Am J PolitSci. 2012;56(1):202–17.

    Article  Google Scholar 

  • González-Ocantos E, Kiewiet de Jonge C, Nickerson DW. The conditionality of vote-buying norms: experimental evidence from Latin America. Am J PolitSci. 2014;58(1):197–211.

    Article  Google Scholar 

  • González-Ocantos E, Oliveros V. Clientelism in Latin American politics. In: Prevost G, Vanden H, editors. The encyclopedia of Latin American politics. Oxford: Oxford University Press; 2019.

  • Hicken A, Nathan NL. Clientelism’s red herrings: dead ends and new directions in the study of non-programmatic politics. Annu Rev Polit Sci. 2020;23:277–94.

    Article  Google Scholar 

  • Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal. 2007;15(3):199–236.

    Article  Google Scholar 

  • Holland A. Forbearance as redistribution: the politics of informal welfare in Latin America. New York: Cambridge University Press; 2017.

    Book  Google Scholar 

  • Holland A, Palmer-Rubin B. Beyond the machine clientelist brokers and interest organizations in Latin America. Comp Pol Stud. 2015;48(9):1186–223.

    Article  Google Scholar 

  • IDB. Slum upgrading and housing in Latin America. Washington, D.C.: Inter-American Development Bank; 2016.

  • Jensen PS, Justesen MK. Poverty and vote buying: survey-based evidence from Africa. Elect Stud. 2014;33:220–32.

    Article  Google Scholar 

  • Kessler G, & Bruno M. (2018). Inseguridad y vulnerabilidad al delito. In J. I. Piovani & A. Salvia (Eds.), La argentina en el siglo XXI. Buenos Aires: Siglo XXI, 329–356.

  • Kitschelt H, Wilkinson S. Citizen-politician linkages: an introduction. In: Kitschelt H, Wilkinson S, editors. Patrons, clients, and policies: Patterns of democratic accountability and political competition. New York: Cambridge University Press; 2007a. p. 1–49.

    Chapter  Google Scholar 

  • Kitschelt H, Wilkinson SI. A research agenda for the study of citizen-politician linkages and democratic accountability. In: Kitschelt H, Wilkinson S, editors. Patrons, clients, and policies: Patterns of democratic accountability and political competition. New York: Cambridge University Press; 2007b. p. 322–43.

  • Kramon E. Ethnic group institutions and electoral clientelism. Party Polit. 2019;25(3):435–47.

    Article  Google Scholar 

  • Larreguy H, Marshall J, & Trucco L. (2018). Breaking clientelism or rewarding incumbents? Evidence from an urban titling program in Mexico. Unpublished Manuscript.

  • Lawson C, Greene K. Making clientelism work: how norms of reciprocity increase voter compliance. Comp Polit. 2014;47(1):61–85.

    Article  Google Scholar 

  • Ligon E, Schechter L. Measuring vulnerability. Econ J. 2003;113(486):C95–102.

    Article  Google Scholar 

  • Lupu N, Gervasoni C, Oliveros V, & Schiumerini L. (2015). Argentine panel election study.

  • Lupu N, Michelitch K. Advances in survey methods for the developing world. Annu Rev Polit Sci. 2018;21:195–214.

    Article  Google Scholar 

  • Lustig N, Pessino C. Social spending and income redistribution in Argentina during the 2000s: the increasing role of noncontributory pensions. Public Finance Rev. 2014;42(3):304–25.

    Article  Google Scholar 

  • Mares I, Young L. Buying, expropriating, and stealing votes. Annu Rev Polit Sci. 2016;19:267–88.

    Article  Google Scholar 

  • Nichter S. Vote buying or turnout buying? Machine politics and the secret ballot. Am PolitSci Rev. 2008;102(01):19–31.

    Article  Google Scholar 

  • Nichter S. Votes for survival: Relational clientelism in Latin America. New York: Cambridge University Press; 2018.

    Book  Google Scholar 

  • Oliveros V. Perceptions of ballot integrity and clientelism. In: Lupu N, Oliveros V, Schiumerini L, editors. Campaigns and voters in developing democracies: Argentina in comparative perspective. Ann Harbor: University of Michigan Press; 2019. p. 213–38.

  • Oliveros V. Working for the machine: patronage jobs and political services in Argentina. Comp Polit. 2021;53(3):381–402.

  • Paller JW. Informal institutions and personal rule in urban Ghana. Afr Stud Rev. 2014;57(3):123–42.

    Article  Google Scholar 

  • Post AE. Cities and politics in the developing world. Annu Rev Polit Sci. 2018;21(1):115–33.

    Article  Google Scholar 

  • ReNaBAP. Relevamiento Nacional de Barrios Populares. Informe General Período 08/2016 a 12/2017. 2017.

  • Rueda M. Small aggregates, big manipulation: vote buying enforcement and collective monitoring. Am J PolitSci. 2017;61(1):163–77.

    Article  Google Scholar 

  • Schaffer J, Baker A. Clientelism as persuasion-buying: evidence from Latin America. Comp Pol Stud. 2015;48(9):1093–126.

    Article  Google Scholar 

  • Scott JC. Patron-client politics and political change in Southeast Asia. Am PolitSci Rev. 1972;66(1):91–113.

    Article  Google Scholar 

  • Spater J, & Wibbels E. (2018). Social density, clientelism and vote banking. Unpublished Manuscript.

  • Stokes S. Cultures in conflict: social movements and the state in Peru. Berkeley: University of California Press; 1995.

    Book  Google Scholar 

  • Stokes S. Perverse accountability: a formal model of machine politics with evidence from Argentina. Am PolitSci Rev. 2005;99(3):315–25.

    Article  Google Scholar 

  • Stokes S, Dunning T, Nazareno M, Brusco V. Brokers, voters, and clientelism: the puzzle of distributive politics. New York: Cambridge University Press; 2013.

    Book  Google Scholar 

  • Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1–21.

  • Szwarcberg M. Mobilizing poor voters: machine politics, clientelism, and social networks in Argentina. New York: Cambridge University Press; 2015.

    Book  Google Scholar 

  • TECHO. (2016). Relevamiento de Asentamientos Informales 2016.

  • UN-Habitat. State of the world’s cities 2012/2013. 2013.

  • UN-Habitat. (2016). World cities report.

  • Visconti G. (2017). Economic perceptions and electoral choices: a design-based approach. Polit Sci Res Methods, 1–19.

  • Visconti G, Zubizarreta JR. Handling limited overlap in observational studies with cardinality matching. Observ Stud. 2018;4:217–49.

    Article  Google Scholar 

  • Weitz-Shapiro R. Curbing clientelism in Argentina: politics, poverty, and social policy. New York: Cambridge University Press; 2014.

    Book  Google Scholar 

  • Zarazaga R. Brokers beyond clientelism: a new perspective through the Argentine case. Latin Am PolitSoc. 2014;56(3):23–45.

    Article  Google Scholar 

  • Zarazaga R. Plugged in brokers: a model of vote-buying and access to resources. J Appl Econ. 2015;18(2):369–90.

    Article  Google Scholar 

  • Zubizarreta JR, Keele L. Optimal multilevel matching in clustered observational studies: a case study of the effectiveness of private schools under a large-scale voucher system. J Am Stat Assoc. 2017;112(518):547–60.

    Article  Google Scholar 

  • Zubizarreta JR, & Kilcioglu C. (2016). Designmatch: construction of optimally matched samples for randomized experiments and observational studies that are balanced by design. R Package Version 0.1.1.

  • Zubizarreta JR, Paredes RD, Rosenbaum PR. Matching for balance, pairing for heterogeneity in an observational study of the effectiveness of for-profit and not-for-profit high schools in Chile. Ann Appl Stat. 2014;8(1):204–31.

    Article  Google Scholar 

Download references


We thank Ernesto Calvo, Jessica Gottlieb, Noam Lupu, Raúl L. Madrid, Thomas Mustillo, Aníbal Pérez-Liñan, David Samuels, Luis Schiumerini, Julie Weaver, Rebecca Weitz-Shapiro, Liz Zechmeister, and colleagues at Tulane University for helpful suggestions on the design of the survey and comments on a previous version of this paper. Participants at seminars at University of Notre Dame and University of Chicago as well as LASA 2017, SAAP 2017, and APSA 2018 also provided valuable feedback. Giancarlo Visconti provided indispensable and generous methodological help. Mercedes Sidders and Gonzalo Elizondo helped us with the implementation of the survey. We also want to acknowledge the assistance of Ezra Spira-Cohen. The slum survey was funded by Centro de Investigación y Acción Social (CIAS, Argentina) and the Center for Inter-American Politics & Research (CIPR) at Tulane University. Part of this paper was completed while Virginia Oliveros was a Visiting Fellow at the Kellogg Institute for International Studies, University of Notre Dame.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Virginia Oliveros.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 1429 KB)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Murillo, M.V., Oliveros, V. & Zarazaga, R. The Most Vulnerable Poor: Clientelism Among Slum Dwellers. St Comp Int Dev 56, 343–363 (2021).

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: