Results

3.1Results: Socioeconomic Ranks of Nobel Laureate Childhoods, 1901–2023

The analysis dataset consists of 714 winners from 44 countries over a period of 123 years. About 4% of winners are women, 35% grew up in the U.S., and an additional 12% in the United Kingdom.1

Figure 1 plots the distribution of the father ranks of the laureates. 68% of winners have fathers above the 90ᵗʰ income percentile in their birth country, 71% have fathers above the 90ᵗʰ education percentile, and 84% above the 90ᵗʰ global income percentile. In all three cases, more than half of winners are from the top 5%. The mean laureate has a father at the 87th income percentile, the 89th education percentile, and the 94th global income percentile.

Panel A of Appendix Figure A3 shows the ten most common occupations held by the fathers of laureates, along with the occupation's census population share. Business owners, professors, physicians, and scientists feature most prominently. Relative to their occupational shares, professors are overrepresented by a factor of over 1000, physicians and engineers by 20 and 97 respectively, and business owners by a factor of 2.5.2

We next examine how the relative childhood status of laureates has changed over time. Figure 2 shows binned scatterplots of average laureate ranks, along with linear estimates of the time trend.3 Table 1 shows regression estimates of the time trend in average ranks and in indicators for families from the bottom 80% and 90%. Every 100 years, on average, there is a 6 rank point (p<0.01) decline in the average national income and education ranks of laureates' fathers. These changes are not driven by low-rank outliers; the share of laureates from the national bottom 80% and 90% has increased by 60–200% over the sample period, depending on the measure.4 Present-day laureates are still overwhelmingly from high-SES families, but they draw from a broader range of socioeconomic ranks than they did a century earlier.5

We can understand these trends better by looking into laureate production by occupation group. The sample period is characterized by a substantial growth of the middle class in high-income countries; the share of people in white-collar professional and managerial jobs grew from about 8% in 1850 to about 18% in 1950. If the relative advantage of parents in each occupation was fixed, we would expect to see the prize share of those occupations rise over the sample period; instead, it falls from 84% to 73% (Appendix Table 7).6 The talent pool with access to scientific opportunity is thus expanding on two margins. First, more children grow up with parents in high-opportunity professional occupations. Second, children of parents outside those occupations are gaining ground. Explore how the income and education rank of each occupation changed over timeA14.

Disaggregating further, children of technical professionals in the 1850s cohorts received 14.6 times more prizes than their population share would predict; this relative representation ratio fell to 7.8 by the 1950s. The ratio for children of managerial professionals (where we include business owners) fell from 7.4 to only 1.7 over the same period. Farmers gained the most in relative terms. Their population share fell by a factor of four while their prize share doubled; they are still underrepresented in the 1950 birth cohort with 9% of the population and 4% of prizes. The least skilled workers (service, farm, and other laborers) remain mostly excluded; they represented over 18% of the working population in the 1950 parent cohorts, but their children received less than 1% of Nobel Prizes.

The global income distribution of laureate backgrounds has also broadened, but less so than the national rank distribution and with marginal statistical significance (Column 3 of Table 1 and Panel C of Figure 2). The average global rank has fallen by two rank points over the 123 years since 1901 (p=0.06). The share of laureates from the global bottom 80% and bottom 90% rose by 50–80% over the sample period, but less so than the national measures and without statistical significance. The estimates suggest that there could be a substantial number of individuals with high latent scientific talent who grow up in contexts that prevent them from achieving their potential and sharing it with humanity, and that little progress has been made to close the cross-country component of this gap.7

3.2Results: Heterogeneity by Gender, Field and Region

3.2.1Results: Inequality by Gender

Female laureates came from more educated families than male laureates, but the difference estimates are imprecise, given there have been only 28 female winners in the sciences in the 123 years of the Nobel Prize's history (Table 2, Panel A). The average female laureate has a national education rank that is 4.6 rank points higher than the average male laureate (p=0.05), and is more likely to be from a top-10% or top-20% education family (p=0.13 and p=0.06 respectively). The national income estimates point in the same direction but are not statistically significant.8 On the global income rank measures, women's ranks are very similar to men's.

The raw share of female laureates (4%) itself indicates that women have faced greater barriers to success in the sciences. Our results further suggest that high socioeconomic status can counteract some of these barriers, but it means that elite women in the sciences are drawn from an even smaller socioeconomic pool than men.

3.2.2Results: Inequality by Field

Table 2B shows differences in the childhood socioeconomic status of Nobel laureates by prize category. The average ranks and top 10%/20% indicators are similar across fields.

3.2.3Results: Inequality by World Region

Table 2C shows the regional pattern of winners' ranks. Laureates from the U.S. tend to come from less elite backgrounds than winners from Europe and the rest of the world, suggesting more equal access to opportunity in the sciences in the United States. 39% of U.S. laureates come from the bottom 90% of their national income distributions, compared with 27% in the rest of the world, with similar differences for other measures. The regional differences are highly statistically significant. Appendix Table A8 shows regression estimates from just the United States against the rest of the world.9

3.2.4Heterogeneity in Time Trends

In Appendix Table A9, we examine whether the time trends vary by the subgroups examined here. The point estimates suggest some heterogeneity: access to scientific opportunity is rising the most for women; in chemistry and physiology; and in Western Europe. But these differences are imprecise and joint tests of equal trends across subgroups are for the most part not statistically significant. You can explore all of these subgroup trends — across SES measure, outcome, field, gender, and region — interactively in the Extended Results explorer.

3.3Results: Father Occupation and Scientific Field

We next test whether father occupation predicts the laureate's direction of scientific research. We classified occupations to a Nobel discipline (chemistry, physics, physiology or economics) if workers in that occupation require substantive engagement with the scientific knowledge of that discipline. For professors, this was straightforward based on their field. We additionally classified doctors and surgeons to physiology, chemists and chemical engineers to chemistry, other engineers to physics, and economists to economics.

We then created a secondary classification for fields which have topical relations to the Nobel fields but do not involve a shared technical toolkit and are less directly related to academic work in the field. In this categorization, for example, we assigned actuaries and business owners to economics, veterinarians to physiology, and mechanics and repairmen to physics. Most occupations remained unclassified as they do not have a clear link to any specific Nobel field.10

Our aim is to test whether laureate children are more likely to be in the parent-associated field than random chance alone would predict. We use a randomization test: we take the sample of fathers whose occupations can be mapped to a Nobel field and we randomly assign their children to scientific fields based on the full sample distribution of laureate fields.11 We then compare the observed own-field share with the distribution of own-field shares generated by the randomization test. Because the economics prize was only awarded in recent years, we run this test for two samples: (i) the sample of chemistry, physiology, and physics-associated parent-child pairs from 1901–2023 (N=172); and (ii) the sample including economics-associated parents since the first econ prize, 1969–2023 (N=143).

Panels A and B of Figure 3 show the distribution of parent-child field match shares from the randomization test and the empirical realization. The empirical values are in the right tails of the simulated distributions: less than 1% of the randomization values at least as large as the observed value for the non-economics sample, and 3% for the economics sample. Depending on the sample, laureates are about 25–30% more likely to receive their award in the parent-associated field than would be predicted from chance alone.

Table 3 summarizes these results, as well as results for three other samples. Splitting the sample above into professor and non-professor parents, we find that children are somewhat more likely to be in parent-associated fields when parents are professors (36–40%) than when parents are not professors (22–26%).1213

In short, laureates' fields are correlated with their fathers' occupations, and the correlation is stronger as the occupational knowledge base becomes closer to the academic research. This suggests that parent income and education raise the probability of producing Nobel laureate children at least in part through a channel other than undifferentiated access to education or opportunity.14