“That study does not mean what you think it does”

This study has been going viral: The plateauing of cognitive ability among top earners.

People were giving their theories as to why it was true. It went viral in part because it appeals to our idealization fairness, that superior cognitive ability must come at some cost, such as superior earning potential. Or it gives hope to people who are not that smart. Or we want to believe that the people who are at the top are not much better than everyone else. But few stepped back to consider that maybe the study does not actually prove what the hype or headlines would seem to suggest.

We draw on Swedish register data containing measures of cognitive ability and labour-market success for 59,000 men who took a compulsory military conscription test. Strikingly, we find that the relationship between ability and wage is strong overall, yet above €60,000 per year ability plateaus at a modest level of +1 standard deviation. The top 1 per cent even score slightly worse on cognitive ability than those in the income strata right below them. We observe a similar but less pronounced plateauing of ability at high occupational prestige.

Some major caveats:

The economic conditions and labor market of Sweden does not readily generalize to the US. We would expect in Sweden for earnings to be capped or lower overall compared to the US, and less competitive too. This could possibly create a disincentive for the smartest of people to try to earn more money, hence a point of diminishing returns at relatively low thresholds of high intelligence. It’s possible Sweden’s smartest people emigrate to the US to earn more, such as in Silicon Valley.

Because the threshold for high income is low in the study, as is intelligence, it means tail effects are ignored.

Figure A2 shows that the percentage of those scoring 9 in cognitive ability is highest among earners of relatively high wages (between SEK500,000 and SEK800,000). With 23.2 per cent it is highest in the 97th wage percentile. But among the top 3 percentiles, the percentage of top scorers

SEK800,000 = $73,000. Indeed, this a low ceiling for wealth. Even adjusting for inflation, it maxes out at around $140,000/year in 2012 dollars.

This is only equivalent to the top 15-20% for income in the US, which is not that high.

Cognitive tests for conscription typically have a low ceiling, at around an IQ of 130 or so. The AFQT for example has an effective ceiling of around 135 IQ (a 99-percentile score). Of course, the AFQT is not an official IQ test, but the scores tend to be highly correlated.

Survivorship bias is another important variable and can skew the results to favor people who are not exceptionally smart. Average-IQ people benefit from lottery-like systems of wealth, like podcasting, entrepreneurship, or YouTube, but these also have very high failure rates and low median earnings. Because the top podcasters and YouTube people are not that smart, one may be under the misapprehension that high IQ and wealth are not that highly correlated, but this ignores all the failures. For more meritocratic forms of wealth, like tech jobs or Substack success, IQ is much more important and the role of survivorship bias goes away. It’s like, “The top 50 podcasters or YouTube people who earn $1 million/year don’t have college degrees; ergo IQ is not that important,” but this ignores the 50,000 who failed or earn much less.

From the study:

Our analysis includes men who joined the labour force between 1991 and 2003 (median 1993). In total, 670,203 men, aged 18–60, entered the labour market in this period to be fully employed for at least 1 year. Cognitive-ability scores are available only for Swedish men who had the obligation to undergo military conscription. We subset on men who took a compulsory conscription test at age 18–19 during 1971–1977 or 1980–1999, years in which ≥90 per cent of each cohort enlisted (94 per cent on average).

The above study, like most studies on income/wealth vs. IQ, is based on old data. Recent economic conditions in the US, since 2008 and especially since 2020, have made IQ more important for the highest-paying of jobs, like in tech or finance. 2+ decades ago, such jobs either didn’t exist or paid much worse (even adjusted for inflation). Mid-6-figure tech jobs in the Bay Area have a much higher cognitive requirement or threshold and are much more competitive than the jobs analyzed in such surveys.

Large companies in the US have much more pre-employment screening compared to other countries and compared to in the past, especially for high-paying jobs, such as phone interviews, automated resume screening, and technical interviews. Many of these things did not exist 30 years ago, at least not in Sweden, and act as cognitive filters. Decades ago it was easer to maybe bullshit your way into a good job or use your old college connections, but much harder now. Save for some DEI, employers are obsessed with not seeming biased or showing favoritism for fear of legal retaliation or bad press. Thus, employers rely heavily on impartial, dispassionate screening, and employees tend to be very overqualified, whereas in the past they were maybe less qualified and more of the training was on-site. A greater pool of talent, especially overseas, means US companies do not need to invest as heavily in employee training and can choose from a much larger pool of candidates who are already well-qualified.

Overall, 30-year-old data from Sweden has near zero applicability to the post-Covid US economic landscape, which is much more competitive , especially at the tails. I would take this study with a big gain of salt.