The Plateau of Cognitive Ability Among Top Earners

The richest people are the smartest people, and people who reach the top of professional hierarchies do so because of cognitive ability. Both claims are widely held and both are, the recent research suggests, only partly true. A 2023 study by Marc Keuschnigg, Arnout van de Rijt, and Thijs Bol in the European Sociological Review reported that in Sweden, average cognitive ability stops rising with income above roughly €60,000 per year — and the very highest earners (top 1%) actually score slightly lower on cognitive tests than those earning just below them. A 2024 follow-up study using comparable data from Finland and Norway reached a strikingly different conclusion: in those countries, ability rises through the top of the income distribution, with the steepest slope at the very top. The two studies together complicate any simple story about merit, intelligence, and elite earnings.

The Swedish finding: a clear plateau

The Keuschnigg, van de Rijt, and Bol study analyzed Swedish administrative records covering 59,000 men who took compulsory military conscription cognitive tests. The conscription tests are well-validated measures of general cognitive ability, administered to essentially every Swedish man of conscription age, providing a near-population sample without the selection biases that plague survey-based cognitive testing.

The findings:

• Across most of the income distribution, ability and earnings are strongly related. Higher cognitive scores predict higher wages. This is not in dispute.
• Above approximately €60,000 per year, the relationship plateaus. Average cognitive ability stops rising with income at roughly +1 standard deviation above the population mean.
• The top 1% of earners actually score slightly lower in cognitive ability than those in income brackets just below them. This is the headline finding: at the very top, additional income is not associated with additional cognitive ability, and may be associated with slightly less.
• A similar but less pronounced pattern holds for occupational prestige. The plateau is not unique to income — it also appears, more weakly, when prestige rather than wage is the success metric.

The interpretation the authors offered was framed around stratification theory: when income or prestige is determined by cumulative advantage — family background, social networks, organizational politics, capital access — cognitive ability becomes less decisive at the top. The mechanisms that distinguish a €60,000 earner from a €50,000 earner may not be the same mechanisms that distinguish a €600,000 earner from a €60,000 earner, and the role of cognitive ability shrinks accordingly.

The Finnish-Norwegian counter-finding: a steepening tail

Bratsberg, Rogeberg, and Terviö’s 2024 paper in the same journal applied a comparable methodology to administrative data on over 350,000 men in each of Finland and Norway. Their findings differ markedly from the Swedish picture:

• The earnings-ability relationship is convex, not concave. The slope between cognitive ability and earnings rank is steepest at the top of the earnings distribution.
• Top earners differ substantially from median earners on cognitive ability. The top earnings percentile scores on average 1 standard deviation higher than median earners, while median earners score about 0.5 SD higher than the bottom percentile.
• Variability in cognitive ability shrinks at the top. Top earners have substantially less spread in cognitive scores than other income groups. The lowest cognitive scores are almost absent among top earners.
• Some high-scoring men have low earnings — high cognitive ability is necessary but not sufficient for top earnings. But low cognitive ability is essentially incompatible with reaching the top of the earnings distribution in these data.
• The Finland and Norway joint distributions are remarkably similar to each other, and both differ from the Swedish pattern.

The authors are explicit that this finding “differs markedly from the flat or declining slope recently reported for Sweden.” The two countries’ findings are not separate facts about separate phenomena. They are competing claims about the same phenomenon — the relationship between cognitive ability and elite earnings — and they cannot both be straightforwardly true in a unified theory.

How can both findings be right?

Several possibilities can reconcile the apparent contradiction:

• Genuine cross-country differences. Sweden, Finland, and Norway are similar in many ways but differ in tax structure, labor market regulation, housing affordability, and the structure of wealth. If elite earnings in Sweden depend disproportionately on inherited capital or business ownership while elite earnings in Finland and Norway depend more on labor income from cognitively demanding professions, the patterns could legitimately differ.
• Definition of “top earners.” The top 1% in Sweden has different absolute purchasing power and different occupational composition than the top 1% in Finland or Norway. The cognitive demands of the actual occupations dominating each country’s top percentile may differ.
• Methodological differences. Even when both studies use administrative data, the choice of cognitive measure, the income windows examined, and the modeling decisions can produce different summary patterns.
• Sampling and cohort differences. The Swedish and Nordic samples are not exactly comparable in age range, time period, or population coverage.

The honest interpretation is that the relationship between cognitive ability and top earnings is country-specific and structurally dependent. The “elite are smart” story is true in some places and not in others, and the conditions that produce one outcome rather than the other are not yet well-characterized.

The longer-term context: intelligence and economic success in general

The relationship between cognitive ability and economic outcomes is one of the most-studied questions in social science. Strenze’s 2007 meta-analysis in Intelligence pooled longitudinal studies and found that intelligence is consistently one of the strongest predictors of socioeconomic success — but that parental social class and educational attainment are comparable predictors, with substantial overlap among the three. No single factor dominates.

Zagorsky’s 2007 paper in the same journal, using U.S. NLSY79 data, found a more granular pattern: IQ predicts income, but does not reliably predict wealth, and does not protect against financial distress. A high-IQ individual is more likely to earn well but not noticeably more likely to accumulate net worth or avoid financial trouble. This finding — sometimes summarized as “smart people earn well but spend it” — is consistent with the Keuschnigg et al. plateau, in that it suggests cognitive ability operates on income generation more cleanly than on the broader package of economic success that elite status implies.

What this means for the merit-and-success debate

The findings have implications for several adjacent debates:

• Compensation and meritocracy. The Keuschnigg et al. plateau weakens any defense of extreme top-end compensation that rests on the cognitive-ability-as-merit framing. If top-1% earners are not more cognitively able than top-2%-to-5% earners, the cognitive-merit justification for the income gap between them does not hold.
• Selection into top occupations. The Bratsberg et al. Finnish-Norwegian finding is more consistent with a meritocratic selection narrative for those countries’ top occupations, with the caveat that “merit” is here narrowly operationalized as cognitive ability.
• Cumulative advantage. Both findings, in their differing ways, support the broader claim that cumulative-advantage processes — initial small differences amplified over time by network effects, capital access, opportunity bottlenecks — shape elite outcomes more than first-order cognitive differences. The Swedish data show this most starkly; the Nordic data show it more subtly.
• The ceiling on cognitive selection. Most occupational selection processes likely cannot reliably distinguish between candidates at +2 SD versus +3 SD on cognitive ability. Above a relatively high threshold, other factors take over not because intelligence stops mattering but because intelligence stops being a usable basis for selection.

What the findings do not establish

Several limits to interpretation are worth holding in mind:

• Both studies are men-only. The conscription tests provide measurement only for men. Whether parallel patterns hold for women in the same labor markets is a separate empirical question.
• Cognitive ability is not the only candidate predictor. Personality (particularly conscientiousness), motivation, ambition, social skills, and what is sometimes labeled “emotional intelligence” all matter for occupational outcomes and are not measured by conscription cognitive tests. The plateau in cognitive ability does not imply a plateau in all psychological correlates of top earnings.
• Cross-sectional snapshots, not life trajectories. Both studies measure cognitive ability at conscription age (around 18) and earnings later in life. They do not capture how cognitive ability and earnings co-evolve over the life course.
• Population-level patterns, not individual outcomes. The plateau means that a randomly selected top earner is not more cognitively able on average than a randomly selected upper-middle earner. It does not mean that any given top earner is not exceptionally able.
• Country-specific findings. The Swedish, Finnish, and Norwegian patterns may not generalize to less egalitarian economies or to countries with different educational and occupational structures.

Frequently Asked Questions

Are top earners the smartest people in society?

It depends on the country and on how “top earner” is defined. In Sweden, the relationship between income and cognitive ability flattens above about €60,000/year and the top 1% scores slightly lower than the income groups just below them. In Finland and Norway, the relationship continues to climb at the top. There is no single universal answer.

Why does Sweden differ from Finland and Norway?

The reasons are not yet well-established. Possibilities include differences in how top incomes are generated (capital vs. labor), differences in tax and labor-market structure, methodological differences between the studies, and structural differences in the occupational composition of the top of each income distribution.

Does cognitive ability still matter for income?

Yes, substantially, across most of the income distribution. The plateau in Sweden and the steeper-at-top pattern in Finland and Norway both apply specifically to the very top of the distribution. For most of the population, higher cognitive ability is associated with higher earnings.

Does this mean rich people don’t deserve their wealth?

The findings speak to the relationship between one specific aspect of merit (cognitive ability) and one specific outcome (earned income). They do not address other dimensions of merit (effort, risk-taking, persistence, ethical conduct) or other forms of success (creative output, social impact, well-being). What they do challenge is any defense of extreme top-end compensation that rests narrowly on cognitive-ability-as-merit.

What about IQ and wealth, as opposed to income?

Zagorsky’s 2007 study found that IQ predicts income better than it predicts wealth, and does not protect against financial distress. The cognitive-ability story for income is cleaner than the story for accumulated net worth.

Are these findings about men only?

Yes — both the Swedish and the Finnish-Norwegian studies use compulsory male military conscription cognitive testing data, which is available only for men. Whether parallel patterns hold for women is a separate question that requires different data sources.

What does this mean for elite hiring decisions?

The findings suggest that, above a relatively high cognitive threshold, additional cognitive selection has diminishing returns — and in some labor markets may not be associated with marginal earnings at all. Hiring practices that emphasize cognitive ability beyond a certain level may be picking up signal that does not translate into the higher-end performance differences the practices are intended to capture.