Whom you observe in your daily life alters your willingness to tax the rich

Recent research suggests that the visibility of extreme wealth within a person’s social circle drives their support for economic redistribution but simultaneously fosters political polarization and personal dissatisfaction. A study published in PNAS Nexus combines computational modeling with online experiments to demonstrate how network structure creates specific biases in how individuals perceive inequality.

Economic inequality has expanded significantly across many industrial democracies in recent decades. This trend presents a difficult puzzle for political scientists and economists. Standard political theory posits that in a democracy where wealth is concentrated at the top, the majority of voters—who have below-average wealth—should vote for policies that redistribute resources more evenly.

This concept is known as the median voter theory. It predicts that as the gap between the rich and the middle class widens, the demand for redistribution should rise. However, this theoretical correction often fails to occur in the real world.

One leading explanation for this discrepancy is that individuals lack accurate information about the true state of the economy. Humans do not typically possess a bird’s-eye view of national wealth statistics. Instead, people rely on their immediate social environments to gauge their financial standing.

They look at their neighbors, colleagues, and friends to estimate how they compare to the general population. Because people tend to associate with others who share similar economic backgrounds, their reference samples are often skewed.

Free daily newsletter

When social networks are economically homogeneous, individuals struggle to perceive the true extent of inequality. Poor individuals surrounded by other poor individuals may overestimate their relative standing. Wealthy individuals surrounded by other wealthy individuals may underestimate how advantaged they are.

To investigate this phenomenon, Milena Tsvetkova from the London School of Economics and Political Science collaborated with Henrik Olsson and Mirta Galesic. Olsson and Galesic are affiliated with the Santa Fe Institute and the Complexity Science Hub Vienna. This team of computational social scientists hypothesized that these network structures fundamentally alter voting behavior. The researchers wanted to understand if changing who people see changes how they vote on tax policy.

The research team began their investigation by developing a computational model. This model simulated a population with an unequal distribution of resources, where a small minority held the majority of the wealth.

The researchers programmed digital agents to behave with a mix of self-interest and a distaste for unfairness. These agents wanted to increase their own wealth, but they also experienced a decline in utility when they possessed significantly less or significantly more than their neighbors.

In the simulation, agents observed a small subset of the population. Based on what they saw, they voted on a tax rate. The model assumed a direct democracy where the median vote determined the policy.

The collected taxes were then redistributed equally among all agents. The researchers manipulated the structure of the social networks to see how different observational paths changed the outcome. They tested conditions where agents mostly saw similar others, different others, or specifically the rich or the poor.

To validate the predictions of their computer model, the researchers recruited 1,440 human participants for a large-scale online experiment. They divided participants into groups of 24.

Within each group, nine participants were randomly assigned a “rich” status with a high initial score. The remaining 15 participants were assigned a “poor” status with a low initial score. The participants were informed that they were part of a larger group, but they could only see the scores of eight other individuals.

The experiment placed these groups into distinct network structures. In “segregated” networks, rich participants only observed other rich people, and poor participants only observed other poor people.

In “homophilous” networks, participants mostly saw others with similar scores. In “heterophilous” networks, participants mostly saw those with different scores. The researchers also created “rich visible” networks, where most connections were directed toward the wealthy members, and “poor visible” networks, where the poor were most prominent.

Participants engaged in three rounds of voting. They used a slider to select their preferred tax rate. After each round, they saw the results and could adjust their vote. The researchers tracked the median tax rate selected by the group and the level of disagreement, or polarization, among the voters. They also asked participants to rate their satisfaction with their scores and their perception of fairness.

The results from the experiment revealed that network structure serves as a powerful lever for collective economic decisions. In the segregated networks, the groups consistently voted for the lowest tax rates.

This occurred because the poor participants, seeing only other poor people, underestimated the potential benefits of redistribution. They did not realize how much wealth was available to be taxed. Simultaneously, the rich participants, seeing only other rich people, felt no pressure to share. This created a consensus for low taxes.

A different dynamic emerged in networks where the wealthy were highly visible. When poor participants were placed in a network where they frequently observed rich neighbors, they voted for significantly higher tax rates.

The visibility of extreme wealth clarified the economic disparity within the group. The poor correctly identified that a high tax rate would redistribute significant resources to them. As a result, the “rich visible” networks produced the highest levels of redistribution.

However, this shift toward higher taxation was accompanied by intense polarization. In the networks where the rich were visible, the voting gap between the rich and the poor was the widest. The poor participants radicalized, often demanding a 100 percent tax rate.

The rich participants generally refused to budge from their preference for low taxes. While the segregated groups achieved a peaceful consensus on low redistribution, the groups with visible wealth experienced deep political conflict.

The study also uncovered a paradox regarding individual happiness. The researchers found that poor participants in the segregated networks were the most satisfied with their outcomes. Even though they remained the poorest in absolute terms because they voted for low redistribution, they felt content. They were essentially shielded from the knowledge of their relative disadvantage.

On the other hand, poor participants in the “rich visible” networks reported the lowest levels of satisfaction. This dissatisfaction persisted even though they ended up with more money due to the higher tax rates they voted for.

The act of constantly comparing themselves to the visible wealthy induced a sense of relative deprivation. They were objectively better off financially, yet they felt worse about their situation and viewed the game as unfair.

The experiment also yielded unexpected demographic findings. Women who were assigned to the “poor” condition tended to vote for lower taxes than men in the same position. This finding contradicts general survey data, which usually indicates that women favor more redistribution than men.

The authors suggest this behavior might stem from a stronger tendency among the female participants to conform to external social norms. In the United States, where the participants were recruited, there is a strong cultural norm against high taxation.

The researchers note several limitations to their study. The experiment utilized a simplified economic model with a flat tax and equal redistribution. Real-world economies involve progressive taxation, complex welfare systems, and potential negative impacts on productivity if taxes are too high.

Additionally, the participants were US residents who likely brought their own strong political beliefs into the experiment. The data showed that many participants voted for tax rates close to the actual US effective tax rate, rather than the rate that would mathematically maximize their earnings in the game.

The study suggests that breaking down information silos can indeed increase political support for redistribution. When the poor see the rich, they want to tax them.

However, the authors highlight that this is not a cost-free solution. Strategies that increase the visibility of excessive wealth may successfully shift policy, but they risk exacerbating political polarization and reducing the subjective well-being of the less advantaged.

Future research could investigate how these dynamics play out in cultures with different baseline attitudes toward inequality. It would also be useful to explore how voluntary network formation influences these outcomes.

In the real world, people actively choose their social circles, potentially reinforcing the segregation that the researchers found leads to apathy. The study implies that while ignorance may be bliss for the disadvantaged, it also cements their economic status.

The study, “Social networks affect redistribution decisions and polarization,” was authored by Milena Tsvetkova, Henrik Olsson and Mirta Galesic.