What did motivate you and your co-authors to write on elites’ weak support for redistributive policies in the Global South? 

Matias: Until recently, the expansion of social programmes in the developing world, in particular in Latin America, contributed to the reduction of inequalities in countries like Brazil, Mexico, Argentina, and many others. Elites in corporations and ruling parties seemed to be converging towards embracing more social inclusion as part of a general strategy to promote stabler societies. From a political economy perspective, it makes sense that elites in such conflictive and extremely unequal societies would cooperate to reduce income disparities. By doing so they also reduce the threats and damages that distributive conflict generates for themselves. In recent years, however, we observed an important backlash against redistributive policies. This pushed us to think about the causes of such ambivalent policy preferences amongst elites. 

What are the paper’s main findings?

Matias: What we find that is particularly novel is that elites are guided by both economic rationales and cultural constraints when dealing with the consequences of inequality. Our results show that elites acknowledge the need for redistribution to the extent to which inequality harms their own interest. Nonetheless, elites also take into account other aspects which are less rooted in crude economic reasoning. In particular, they distinguish themselves from the poor in terms of rationality, portraying the latter as ignorant and irrational. Seeing the poor as irrational prompts them to reconsider the efficiency of social policies and, on average, sets them back from supporting redistribution. This occurs even when elites acknowledge the pertinence of redistribution in light of objective threats to their own wellbeing. This pattern proved to be very robust. We observe it across cases, and in both right-wing and left-wing political elites, as well as among economic elites. 

Why did you choose Brazil as the main focus of your analysis and South Africa and Uruguay to examine external validity?

Graziella: When we wrote the paper, we had been working together on the Brazilian case for quite some time. In 2013, when we were both based at the Federal University of Rio de Janeiro (UFRJ), me as a professor and Matias as a master student, we gained access to a set of in-depth interviews conducted with Brazilian elites in the late 1990s. The idea of working with their perceptions about the poor emerged inductively from our analysis of these interviews. We co-authored a first paper on how elites in Brazil categorise the average Brazilian, which was published in the journal Sociologia & Antropologia in 2015. At the same time, with my UFRJ colleague Elisa Reis, we were working on implementing a survey with elites in Brazil and Uruguay, which intended to replicate a study she had done in the early 1990s. We saw Uruguay as a good contrasting case to Brazil, as it was one of the least unequal countries in Latin America. We decided to include South Africa as, similarly to Brazil, it had historically been also one of the most unequal countries in the world. In addition, I already had some expertise on it from my work on racial inequalities in Brazil and South Africa. As we discuss in the paper, this proved to be a good selection of cases as elites in Brazil and South Africa showed much distrust of the poor (and average voters) while Uruguayan elites were much less critical towards their poorer well-off co-nationals.

Why are the perceptions of others important in shaping elites’ incentives and policy outputs?

Graziella: Because we define elites in terms of their power position in market and political institutions, their perceptions become relevant almost by definition. In fact, a growing political science literature has shown that elite preferences, especially those of economic elites, have an important impact on policy outcomes. Beyond the political science literature, our definition of perceptions is in close dialogue with recent debates on cultural sociology. Therefore, the idea is not simply to find whether elites see the poor as deserving or undeserving but also to analyse how these perceptions enable cultural processes that create incentives (or disincentives) for them to support (or reject) redistributive policies. For example, in the Brazilian interviews, elites commonly described poor Brazilians as deserving of help but also as ignorant and irrational voters. We found that their perceptions on ignorance played an important role in shaping their ambiguous support to social policies, not only because those perceptions encouraged their belief that these policies are unsustainable but also because they enabled the narrative that these policies will be used by ruling (and often undeserving) political elites to manipulate ignorant poor people.

How did you construct the index to measure elites’ support for redistributive policies?

Matias: Very few people in Brazil, Uruguay or South Africa would say that they oppose public education or public health. Following previous elite surveys, we asked participants about the desirability and the feasibility of policies in separate questions. These were simple yes or no questions for ten individual policies, from food stamps to free college. In order to create our index, we relied on the interaction between desirability and feasibility. The result is an index in which participants get the highest score only if they see all policies as both viable and feasible, but get the lowest score if they believe all policies to be infeasible, regardless of how many they deem desirable. Through this strategy we were hoping to address issues of social desirability bias in the statements about redistributive policy, and that was appreciated by the paper’s reviewers. 

Could you elaborate on the strategy of using both data collected from in-depth interviews and survey data?

Matias: Multi-method designs are about addressing different aspects of causal processes, estimating “causes of effects” and “effects of causes” in one research design. The difference between survey and interview data goes beyond sample size. We could have one thousand in-depth interviews and the data would still not allow us to make the estimations that we make with the survey. Likewise, we could submit survey participants to endless follow-up questions and we would still not get the richness of in-depth interviews. What differs both is the type of intervention that is made. During in-depth interviews, both the interviewer and the participant are fairly free to conduct the interview like a conversation. This is great for observing spontaneous associations, which are later coded. But we cannot presume that participants that did not express those associations would not make them or that they would disagree with them. The survey instrument, on the other hand, forces participants to react to the same set of questions with the same range of possible answers. By doing so it creates believable and simple counterfactual scenarios, which allow us to model their response and estimate the degree of association between certain types of responses in the wider population. We were very keen about making a randomised survey with elites. To the best of my knowledge, we were the firsts to conduct one. Replicating the survey was also of key importance. The summed pieces of data that we have tell a very compelling story about the role of culture in elite decision-making. 

Graziella: Data from interviews also have an inductive quality, which allows researchers to build theory out of empirical analysis. Meanwhile, the type of data generated by randomised surveys allows us to test the theory, or at least the logical implications of the theory. 

Are you planning further research building on your paper? 

Graziella: This fall we are actually starting a new four-year SNSF-funded project entitled “How Elites Shape Unequal Democracies: Perceptions of Redistribution in Brazil and South Africa”, hosted at the Albert Hirschman Centre on Democracy, which will follow up many of the questions we raise in the paper. So, we will be back with more results and papers soon. The plan now is to co-author a monograph extending this debate, and showing evidence from other cases too.

*  *  *

Full citation of the article:
López, Matias, Graziella Moraes Silva, Chana Teeger, and Pedro Marques. “Economic and Cultural Determinants of Elite Attitudes toward Redistribution.” Socio-Economic Review, mwaa015. Published electronically 1 May 2020. https://doi.org/10.1093/ser/mwaa015.

Interview by Guilherme Mateus Suedekum, PhD Candidate in International Economics; editing by Nathalie Tanner, Research Office.
Banner image: MJgraphics/Shutterstock.com.