Two people with identical income, exposed to the same data on inequality, vote differently. One supports redistribution and affirmative action. The other opposes it. This is not an anomaly to be explained by misinformation or ignorance. It is the predictable result of a profound difference in how each answers a single question: when someone succeeds, does that come at the expense of someone else?
This conviction, which economists call zero-sum thinking, predicts Americans’ political preferences better than their income level. This is what a survey conducted among 20,400 American residents in seven waves between 2020 and 2023 shows. The result challenges a central assumption in public debate: the idea that better informing citizens about inequality would change their preferences for economic policy.
The Essentials
- A survey of 20,400 American residents, conducted in seven waves between 2020 and 2023, shows that zero-sum thinking is strongly associated with political preferences, even after controlling for income, education, and party affiliation
- Stefanie Stantcheva, winner of the 2025 Clark Medal, has developed a survey methodology that measures the moral reasoning behind fiscal and redistributive attitudes
- Two people facing the same figures on inequality do not draw the same conclusions if they do not share the same vision of merit and chance
- The identified lever: targeting moral reasoning rather than facts alone about income inequality opens a space for compromise that data alone does not produce
- This line of research challenges the assumption that information is the main obstacle to fiscal and redistributive reform
The question is an old one. Economists, political philosophers, and electoral strategists have answered it in contradictory ways. For decades, an implicit consensus took hold in liberal and progressive circles: if citizens better understood inequality, they would support more policies to correct it. All you needed were the right figures. Better communication. Improved economic education.
This consensus is now being challenged by the data itself.
Believing in a Fixed Pie
Zero-sum thinking, in its economic formulation, designates the belief that total wealth is a fixed quantity. If someone becomes richer, it must be at the expense of another. Individual success is a claim on a limited common resource, not the creation of new value.
This vision is not irrational. In certain contexts, it corresponds to reality: agricultural land whose ownership changes hands, a unique position awarded to one candidate rather than another. But applied to the economy as a whole, it is empirically false. Economic growth creates net wealth. Innovation enriches producers without impoverishing their competitors fairly. An entrepreneur who launches a prosperous business does not steal its added value from someone else.
Yet surveys show that this belief remains widespread. And that it is stable over time. Observed across multiple generations in the same American families and transmitted through time, zero-sum thinking resembles less a political opinion than a deep moral disposition. Less a conclusion drawn from facts than a framework that determines which facts seem relevant.
This is where Stefanie Stantcheva’s work becomes decisive.
What Stantcheva Measures That Others Do Not
Stantcheva, professor at Harvard and winner of the 2025 Clark Medal from the American Economic Association, has built part of her research on a simple intuition: to understand why people support or reject redistribution, it is not enough to measure their income, their stated preferences, or their level of information. One must understand the moral reasoning that structures how they read inequality.
Her surveys do not simply ask the question “are you for or against such a policy?” They explore how respondents explain success and failure. Is it personal merit? Chance? Family inheritance? Discrimination? The answers to these moral questions predict political attitudes with a precision that classical socioeconomic variables do not allow one to achieve.
The survey of 20,400 American residents confirms this result robustly. Contrary to what intuitive reading might suggest, in the study, more zero-sum thinking is associated with greater support for redistribution and affirmative action: individuals who believe that the gains of some come at the expense of others are logically more inclined to want to correct this dynamic. Individuals with less zero-sum thinking, who may more readily believe in individual merit as a driver of success, are less favorable to these policies. The zero-sum thinking variable remains predictive after controlling for income, education, and party affiliation.
This is not a result about opinions. It is a result about the cognitive structure that produces these opinions.
Why Information Alone Is Not Enough
The political implications of this research are uncomfortable for a certain liberal tradition. It suggests that a significant part of political disagreement over redistribution, immigration, or affirmative action is not a disagreement about facts, but a disagreement about moral principles that give facts their meaning.
Showing two people the same data on intergenerational social mobility in the United States produces two different reactions depending on their moral framework. The one who believes in merit will see it as proof that the system works for those who work hard. The one who believes in the weight of chance and origin will see it as proof that the system is unjust. They are not wrong about the data. They read it through different lenses.
This observation must be considered in relation to abundant literature on economic beliefs and politics. Daron Acemoglu and Simon Johnson have shown how the distribution of power determines who benefits from technological progress. But their framework is interested in institutions that channel gains, not beliefs that legitimate or contest this distribution. Stantcheva goes down one level: she interrogates the moral substrate on which institutional preferences rest.
The question then becomes: if information does not change fundamental beliefs about merit and chance, what does change them?
What Genealogical Data Reveals About the Transmission of Beliefs
Perhaps the most striking aspect of this research is the least commented on: zero-sum thinking is transmitted between generations. Data covering multiple generations of American families show a significant correlation between the moral dispositions of ancestors and those of current descendants, after controlling for all socioeconomic variables.
This transmission is not entirely biological. Researchers identify plausible channels: family socialization, narratives transmitted about success and failure, experiences lived in particular economic contexts that marked a generation and influenced how it educated the next one. Families that went through periods of strong competition for scarce resources, or that belong to communities where the success of some was perceived as a direct threat to others, transmit this disposition more strongly.
This result has a direct implication for the American debate on political polarization. The divergence in visions of merit between Democrats and Republicans is not simply the product of media fragmentation or misinformation. It is rooted in long family histories, in community experiences sedimented over multiple generations. This makes the work of targeting information even more difficult than it appears.
The Tension Between Stantcheva and Classical Liberal Tradition
This line of research enters into direct tension with a central hypothesis of liberal political economy, which Tyler Cowen or Johan Norberg defend in different forms: if markets produce growth and citizens understand how markets work, support for pro-growth and pro-market policies should naturally expand.
Stantcheva does not say this hypothesis is false. She says it is incomplete. Understanding of economic mechanisms is filtered through prior moral beliefs about what is just and unjust. A citizen who believes that success is largely determined by merit can perfectly well understand market mechanisms and conclude that redistribution is illegitimate, not because he is poorly informed, but because his moral framework logically leads him there.
Conversely, a citizen convinced that success depends largely on chance and origin can support aggressive redistribution while perfectly understanding the potentially discouraging effects of high taxation, because these effects seem to him an acceptable price to correct an injustice he considers structural.
This is not ignorance. It is deep philosophical disagreement, and the data confirm it.
The tension here is not resolved by Stantcheva’s data. It is illuminated. We better understand why the debate over inequality in the United States remains blocked despite increasingly precise data on social mobility, income gaps, and the effects of fiscal policies. Data are not the problem. The disagreement concerns the principles that allow one to interpret what data mean morally. This is moreover an angle that the study of labor markets and social protection illustrates differently: institutions only function if actors trust them, and this trust is itself conditioned by beliefs about fairness.
The Identified Lever: Reason About Merit, Not Figures
The operational conclusion of the research is also its least developed part, which leaves open space. The researchers identify a plausible lever: if moral beliefs about merit and chance predict political preferences better than income, then targeting these beliefs directly, rather than improving factual information about inequality, opens a different space for compromise.
Concretely, this means several things. First, showing that success depends more widely than individuals think on factors beyond their control, zip code, family network, macroeconomic conditions at the time of labor market entry, without denying the reality of individual merit. This is not denying personal agency. It is broadening the framework within which it operates.
Next, building redistributive policies that explicitly acknowledge the tension between merit and chance, rather than setting aside one in favor of the other. Stantcheva’s work on taxation suggests that reforms presented as a correction to undeserved chance tend to obtain broader support than reforms presented as a levy on merit. The same policy, framed differently, meets different opposition.
Finally, acknowledging that intergenerational transmission of beliefs limits the speed of change. Moral dispositions about merit do not change after a well-designed information campaign. They evolve over decades, at the pace of lived experiences and family socialization.
This last point is important not to fall back into another naive optimism, this time about the capacity of “targeted moral reasoning” to quickly transform political preferences. Stantcheva opens a path. She does not promise a shortcut.
What This Says About the American Debate Over Affirmative Action and Immigration
The survey measures preferences on three specific areas: fiscal redistribution, affirmative action, and immigration. In all three cases, zero-sum thinking predicts positions better than income. The result is particularly strong on immigration, which is not surprising: the American debate on immigration is structured largely around the question of whether immigrants “take” jobs, social benefits, or resources from existing residents.
This zero-sum reading of immigration is empirically contested by most available economic studies, which show that immigration increases the size of the pie rather than redistributing existing shares. But if zero-sum belief is a stable moral disposition, transmitted over several generations, confrontation with data is not enough to modify it.
This is a point that Stantcheva’s research shares with work on democracy and opinion formation in fragmented media. Polarization over immigration is not principally an information problem. It is a problem of incompatible moral frameworks about what success means in an open economy.
The debate remains open on a question that Stantcheva’s data do not exhaust: are moral frameworks about merit themselves constructions shaped by identifiable economic and political interests? Tyler Cowen and the liberal market tradition would answer that these beliefs often reflect well-founded intuitions about how markets work. Stantcheva, by measuring their transmission across generations, suggests that they also reflect stories, fears, and experiences that have not always followed the same boundaries as rational economic arguments.
The question this research leaves open is this: if moral beliefs about merit are more stable than stated political preferences, which should be reformed first, policies or the narratives that make them acceptable?
Sources
-
Works by Stefanie Stantcheva — NBER / American Economic Review (IDEAS/RePec)
-
American Economic Association — Clark Medal 2025, Stefanie Stantcheva (aea.org)
-
Stantcheva, S. — “Understanding Tax Policy: How do People Reason?”, Quarterly Journal of Economics, 2021 (no link — available via NBER)
-
Stantcheva, S. — “Immigrants and the Making of America”, NBER Working Paper, 2020 (no link — available via NBER)
-
Acemoglu, D. & Johnson, S. — Power and Progress, PublicAffairs, 2023
-
Cowen, T. — The Complacent Class, St. Martin’s Press, 2017 (no link)
-
Chetty, R. et al. — data on intergenerational social mobility, Opportunity Insights (opportunityinsights.org)