This chapter examines how disparities in 21st-century skills contribute to disparities in labour market outcomes in OECD countries. Using data from the 2023 Survey of Adult Skills, it analyses the extent to which disparities in employment, earnings and job satisfaction between men and women, adults with different socio-economic and immigrant backgrounds, and those who grew up in cities and rural areas reflect differences in skills and educational attainment. The chapter further considers how occupational and sectoral segregation in the labour market explains disparities in outcomes. For example, workers from socio-economically disadvantaged backgrounds are less likely to work in roles with large growth prospects in the coming decade and that are experiencing rapid changes in skills demands. Finally, this chapter quantifies the transmission of educational and occupational advantage and the role of skills in promoting upward educational and occupational mobility.
OECD Skills Outlook 2025
Abstract
Disparities in employment and earnings linked to individuals’ socio‑demographic profile are widespread across OECD countries. They largely reflect disparities in lifelong learning opportunities, as well as the skills individuals gain over time and the type of jobs they are employed in. Key findings include:
Socio-economic disparities:
Socio-economic disparities in employment prospects are largely explained by disparities across groups in educational attainment and skills. Adults from tertiary-educated families are 2.7 percentage points more likely to be employed than those whose parents did not complete tertiary education, despite having similar socio-demographic characteristics. Similarly, adults with parents who worked in high-status occupations, such as managerial or professional occupations, are 3.4 percentage points more likely to be employed than those whose parents worked in low-status occupations with similar socio-demographics. These differences are largely explained by differences between the groups in individuals’ own educational attainment, skills and engagement in lifelong learning.
Close to three-quarters of socio-economic disparities in wages can be explained by disparities across groups in learning trajectories and core 21st-century skills. Adults from tertiary-educated families or with parents who worked in high-status occupations earn 11-13% more per hour than similar individuals without tertiary-educated parents or whose parents worked in low-status occupations. When controlling for individuals’ learning trajectories, skills, engagement in lifelong learning, these differences are reduced to 4% for parental occupation and disappear for parental education.
Over a third of working-age people have experienced absolute upward educational and occupational mobility compared to the previous generation, and among today’s young people, expectations of upward educational and occupational mobility remain high. However, whereas only 12% of working age adults have a lower level of education than their parents, as many as 36% work in an occupation with a lower social status than the occupation of their parents. Among 15-year-old students, only 20% expect to achieve a lower level of educational attainment than their parents and 32% expect to work in a lower-status occupation.
Adults with socio-economically advantaged backgrounds are more likely to work in jobs that are experiencing rapid skills changes and growing demand, and that are commanding high relative wages, whereas adults with socio-economically disadvantaged backgrounds are more likely to work in jobs that are experiencing low levels of skills changes and declining demand, and commanding low relative wages.
Gender disparities:
Gender disparities in employment are large and remain wide even when comparing men and women who have similar educational qualifications and skills levels. When comparing men and women with similar socio-demographic characteristics, men’s employment rate exceeds women’s by 7.9 percentage points. This gap remains relatively similar when comparing men and women with similar educational qualifications, skills and engagement in lifelong learning.
Disparities in educational qualifications and skills do not explain gender disparities in wages. The gender gap in wages is 14% when men and women with similar socio‑demographic characteristics and personal circumstances are compared. This gap is 16% when further accounting for differences in learning trajectories, skills and engagement in lifelong learning.
Occupations and industries with a prevalence of male workers tend to command higher wages than occupations with a prevalence of female workers, even when occupations with similar skills requirements are compared. Women tend to work in human-centric expansion roles and routine retreat roles, whereas men are more likely to work in transformative growth roles and disrupted declining roles, as well as routine retreat roles.
Residential context:
Employment disparities between adults who grew up in different environments during childhood are small; however, adults who grew up in cities earn significantly more than those who grew up in villages. This difference reflects, to a large extent, differences between the two groups in educational attainment and information-processing skills. Adults who grew up in cities have an employment rate that is 0.89 percentage points lower than those with similar socio-demographic characteristics, skills and engagement in lifelong learning who were raised in villages. Those who grew up in cities earn 7% more per hour than those from villages, when comparing adults with similar socio-demographic characteristics. This earnings gap decreases to 3% after accounting for differences in educational qualifications and core 21st-century skills.
Immigrant background:
Employment disparities between adults without an immigrant background and the children of immigrants are small and almost entirely explained by differences in educational attainment and skills between the two groups. Children of immigrants – defined as those born in the country of residence to at least one foreign born parent or those who migrated before the age of 18 – have a 2.2 percentage point lower employment rate than their non-immigrant counterparts with similar socio-demographic characteristics. Accounting for differences in additional educational attainment and information-processing skills results in a difference that is not statistically significant. These differences vary greatly across countries, which is a reflection of differences in characteristics of the children of immigrants in different contexts that cannot be fully accounted for, as well as differences in policies and practices in host countries.
In the knowledge-driven economies of the 21st century, the role of individual skills in determining labour market outcomes has become increasingly critical. As globalisation, technological change, demographic shifts and the green transition reshape the nature of work, understanding how information-processing and socio-emotional skills influence individuals’ labour market trajectories is essential for ensuring individual well-being and promoting economic growth at the economy level (Heckman and Kautz, 2012[1]; OECD, 2019[2]). Research consistently indicates that information-processing skills are significantly related to higher labour market earnings, success in the job market, societal participation and overall economic growth. Estimated returns tend to be largest for numeracy and literacy skills and smaller for problem-solving skills, although the relative importance of different skills dimensions varies across countries (OECD, 2024[3]). As a result, the disparities in 21st-century skills identified in Chapters 2 and 3 could limit opportunities for labour market integration among many socio-demographic groups, potentially increasing skills shortages and limiting opportunities for social mobility.
This chapter considers how skills and education relate to employment outcomes, and if the disparities in skills and education described in Chapter 3 contribute to disparities in employment status, earnings and job satisfaction, as well as the allocation of workers in occupations and industries with different labour market prospects and skills requirements. Specifically, it examines how these disparities interact with individual socio-demographic characteristics such as gender, parental education and occupation, immigrant background, and childhood residential context.
Research has long documented that socio-demographic characteristics such as gender, parental background and educational attainment are significant predictors of occupational trajectories (Blanden, Gregg and Machin, 2005[4]). However, recent work, made possible by the availability of direct measures of skills, has enabled analyses of the determinants of employment outcomes to incorporate the role played by skills. Literacy and numeracy, the information-processing skills measured in the first cycle of the Survey of Adult Skills, have been shown to be critical predictors of employability and wage levels (Hanushek et al., 2015[5]; OECD, 2016[6]). The Survey of Adult Skills further reveals that higher proficiency in these domains is consistently associated with greater labour force participation and higher wages across countries (OECD, 2024[3]; 2019[2]). Moreover, the ability to solve problems in technology-rich environments has gained prominence as workplaces demand more flexible, innovative thinking (Autor, 2015[7]).
Social and emotional skills, which in this report refer to the Big Five personality traits of extraversion, emotional stability, agreeableness, conscientiousness and open-mindedness, are also increasingly considered as critical determinants of labour market success. In particular, conscientiousness, emotional stability and agreeableness have been linked to higher employment probabilities and earnings (Almlund et al., 2011[8]; Heckman, Stixrud and Urzua, 2006[9]). These skills are also associated with job performance, co‑operation and resilience, all of which are valued in work environments (Soto, 2019[10]). Similarly, delayed gratification, an indicator of self-regulation, has been shown to predict long-term economic outcomes (Duckworth and Seligman, 2005[11]; Mischel, Shoda and Rodriguez, 1989[12]), suggesting that time preferences and impulse control may partly mediate how initial disparities translate into disparities in long-term outcomes.
These skills are not randomly distributed across the population. For example, children with socio‑economically advantaged backgrounds (i.e. those with tertiary-educated parents or parents in high-status occupations) tend to score higher on both information-processing skills (such as literacy and numeracy) and social and emotional skills, reflecting differences in early life experiences, parental investments and educational opportunities (Heckman and Mosso, 2014[13]). However, individuals with an immigrant background and those raised in rural or with socio-economically disadvantaged backgrounds face systemic barriers that may limit skills development and labour market access (Dustmann and Glitz, 2011[14]; Oded, 2011[15]). As such, skills disparities can perpetuate intergenerational inequalities and reinforce structural divides within society.
This chapter builds on and complements a growing body of OECD analysis on the drivers of disparities in labour market outcomes. It extends insights on employment and earnings differences by parental education, which highlights how family background continues to shape wages across countries, by broadening the lens to consider a wider set of socio-demographic characteristics (Causa, Forthcoming[16]). It also connects with analyses documenting the role of unequal opportunities in perpetuating income inequality (OECD, 2025[17]) and of socio-economic and gender disparities in young people’s aspirations (OECD, 2025[18]). In addition, it complements the OECD’s recent stocktaking on gender gaps in education, employment, leadership, health and pay across OECD and European Union (EU) countries (OECD, 2025[19]). Finally, It draws on recent OECD work on job creation and local economic development, which highlights how place-based policies and local labour market dynamics can amplify or mitigate social and demographic disparities in labour market outcomes (OECD, 2023[20]; OECD, 2024[21]).
Educational qualifications and information-processing skills are highly predictive of employment status. Employment is highest among those with at least a bachelor’s degree, who have around a 18 percentage point higher employment rate than similar adults without an upper secondary qualification (Table 4.1, Model 2). Short-cycle tertiary graduates and upper secondary graduates also have higher employment rates than adults without an upper secondary qualification. Controlling for adults’ education and 21st-century skills at the same time isolates each factor’s relationship with labour market outcomes while holding the other factor constant. Results are similar when controls for social and emotional skills are introduced (Table 4.1, Model 3).
Differences in adults’ educational attainment and skills explain a large proportion of the differences in employment rates observed based on age, socio-economic background (parental education and parental education), and whether people grew up in cities or villages. After accounting for other socio-demographic characteristics, the gap between those with tertiary-educated and non-tertiary educated parents is 2.7 percentage points, and 3.4 percentage points between those with high-status and low-status parents (Table 4.1, Model 1). However, there is no significant gap when differences between the two groups in adults’ educational attainment and both information-processing and social and emotional skills are considered (Table 4.1, Model 3), or whether other differences in individual-level characteristics (Table 4.1, Model 4) and possible non-linearities in skills are considered (Table 4.1, Model 5). Parental education, parental occupation and childhood residential context are positively associated, because tertiary-educated parents tend to work in higher-status occupations and live in urban areas. Therefore, disparities linked to socio-economic background often compound and accumulate.
There are between-country differences in disparities related to the likelihood of employment across socio-demographic groups, and in the role that adults’ educational attainment, information-processing skills and social and emotional skills play in shaping such disparities. Figure 4.1 indicates that the variation in between-country differences in employment related to socio-economic background is smaller than the variation related to age, immigrant background and gender. In Italy, disparities by parental education are as high as 14 percentage points, reflecting that adults with tertiary-educated parents are more likely to be employed (Figure 4.1, basic adjusted model). After accounting for adults’ educational attainment and skills, this difference reduces to 5 percentage points. In contrast, in Chile, employment differences increase from 3 to 5 percentage points – favouring those adults whose parents do not have tertiary education (Figure 4.1, basic and fully adjusted model).
Differences in the likelihood of employment between men and women are similar irrespective of whether differences between men and women in educational attainment and information-processing skills are considered or not, being 7.9 percentage points in both model estimates (Table 4.1, Models 1 and 2). The large difference in employment between men and women who hold similar educational qualifications and who have similar levels of skills proficiency could reflect differences in how men and women engage with and benefit from formal education, or how societal norms and gender roles shape their employment prospects (e.g. women often exit the labour market to care for young children or elderly relatives). When the broad set of social and emotional skills is added, the gender gap in employment remains relatively unaffected at 7.1 percentage points (Table 4.1, Model 3).
Change in likelihood of being employed in percentage points, OECD average
Note: Adults aged 30-65, excludes students and retired individuals. The figure reports coefficient estimates from a linear model for the probability of being employed. Coefficients in bold are statistically significant at the 5% level. Additional individual-level characteristics include: number of children, health, employment status of partner, and current residential context (urban-rural). Estimates in models (3), (4) and (5) exclude Japan and the United States because no information on social and emotional skills was collected in these countries. These countries are also excluded from the OECD average in all models presented in this table for comparability but country specific estimates for these countries are provided alongside those of other countries in Table 4.A.1.1 in Annex 4.A. Respondents' educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 0, 1, 2), upper secondary and post-secondary general (gen. ISCED 3 short, gen. ISCED 3 access 3, gen. ISCED 3 access 3/4, gen. ISCED 3 access 5/6/7), upper secondary and post-secondary vocational (voc. ISCED 3 short, voc. ISCED 3 access 3, voc. ISCED 3 access 3/4, voc. ISCED 3 access 5/6/7), short-cycle tertiary education (ISCED 5 nfs [not further specified], gen ISCED 5, voc. ISCED 5), and bachelor’s or equivalent or above (ISCED 6, 7, 8). Parental education (at respondents’ age 14) is based on the International Standard Classification of Education (ISCED) 2011 and distinguishes between adults with at least one tertiary-educated parent (ISCED 2011 5, 6, 7 and 8) and those with no tertiary-educated parent. Parental occupation (at respondents’ age 14) is based on the International Classification of Occupations (ISCO) and grouped into high-status: managers, professionals, and technicians and associate professionals (ISCO 1-3); and low-status: clerical support workers; service and sales workers; skilled agricultural, forestry and fishery workers; craft and related trades workers; plant and machine operators, and assemblers; and elementary occupations (ISCO 4-9). Childhood residential context (at respondents’ age 14) refers to whether the respondent grew up in villages, towns or cities. Groups by immigrant background distinguish between children of immigrants, immigrants and non-immigrants. Children of immigrants were born in the country in which they currently reside, but their parents were not, or they were born in a different country and moved to their current country of residence before the age of 18. Immigrants are defined as those who migrated to their current country of residence at age 18 or older. Non-immigrants were born in their current country of residence, as were their parents.
*Differences by immigrant background vary greatly across countries because of differences in the size of immigrant groups as well as composition and context. Readers are therefore encouraged to consult country-specific results in Table 4.A.1.1 in Annex 4.A.
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
Change in probability of being employed by socio-demographic group: basic and fully adjusted country-specific regression coefficients
Note: Adults aged 30-65, excludes students and retired individuals. The figure reports coefficients for the following socio-demographic groups – gender, age, parental education, parental occupation, immigrant background and childhood residential context – based on separate country estimates. Coefficients from the basic adjusted model follow the specification of Model 2 in Table 4.1, while those from the fully adjusted model follow the specification of Model 5 in Table 4.1. The fully adjusted model additionally accounts for variables that may not be available for all countries, which is why for some countries, only estimates in the basic adjusted model are available. See the note for Table 4.1 for the definitions of groups based on parental occupation, parental education, immigrant background and childhood residential context. Country-specific results including standard errors are provided in Table 4.A.1.1 in Annex 4.A. For immigrant background, the figure only indicates a country if at least 200 adults for each group are part of the final PIAAC sample. Country-specific results are provided in Table 4.A.1.1 in Annex 4.A.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
Average results mask large between-country differences in the size of the gender gap in employment. For example, in Italy, the gender gap in employment is 27 percentage points, whereas in Finland the gender gap (0.8 percentage points, favouring women) is insignificant (Figure 4.1, basic adjusted model). However, country-specific gender gaps evolve differently after accounting for differences in educational attainment and skills. For example, in the Czech Republic (hereafter ‘Czechia’), the gender gap decreases from 14 to 11 percentage points (Annex Table 4.A.1). By contrast, in Israel, the gender gap in employment rises from 7 to 10 percentage points when men and women with similar educational attainment and skills are compared.
Adults without an immigrant background are more likely to be employed than immigrants and the children of immigrants, although differences between groups disappear once individuals with similar educational attainment and skills are compared. The employment rate of adults without an immigrant background is 4.3 percentage points higher than that of immigrants and 2.2 percentage points higher than that of the children of immigrants (Table 4.1, Model 1). These differences reflect variations in educational attainment and information-processing skills across groups: there are no differences in the likelihood of employment between these groups when adults with similar educational attainment and information-processing skills are compared (Table 4.1, Models 2 to 5). This suggests that disparities in education and information-processing skills development opportunities are the primary reason for the lower likelihood of employment among immigrants and the children of immigrants compared to adults without an immigrant background (Table 4.1, Model 2).
Estimates suggest that there are no employment differences between adults raised in villages rather than towns or cities after taking into account other socio-demographic characteristics (Table 4.1, Model 1). In fact, when comparing individuals with a similar level of educational attainment and skills, those raised in villages have a marginally higher likelihood of being employed than those raised in cities or towns, with a difference of approximately 1 percentage point (Table 4.1, Model 3).
It is important to recognise non-linearities related to the impact of 21st-century skills when analysing the data. The empirical analyses described so far rely on the assumption of linearity: each additional standard deviation (SD) increase in 21st-century skills is assumed to raise the probability of employment by a constant amount. Relaxing these assumptions reveals diminishing marginal returns to numeracy and broadly linear returns to literacy and adaptive problem solving (Table 4.1, Model 5). The marginal effect of increases in numeracy skills falls to approximately zero once an individual is about half a SD above the population mean, after which further improvements no longer increase employment probabilities. In other words, stronger numeracy skills are associated with improved employment prospects only up to a point, beyond which further improvements do not make it easier to find work. By contrast, better literacy and the ability to problem solve keep paying off at a steady rate: the better these skills, the better the employment prospects. The association between extraversion, conscientiousness and emotional stability and the likelihood of employment is also non-linear.
Recognising non-linearities has two key implications: first, it strengthens the efficiency case for targeted skill investments, and second, it cautions against extrapolating average marginal effects to the entire population when designing policy or projecting labour market impacts. Programmes that support adults as they improve their skills from low to intermediate levels of proficiency generate larger employment dividends than those that strengthen the skills of workers who already have strong numeracy skills. If policy makers fail to account for diminishing returns, there is a risk of overstating the benefits of incremental upskilling among high-skilled groups and understating the employment payoff of basic skills remediation for adults with low levels of numeracy proficiency.
Educational attainment and information-processing skills are positively associated with workers’ productivity, and various studies have confirmed the independent effects of information-processing skills and educational qualifications on wages (Araki, 2020[24]; OECD, 2024[3]; Hanushek et al., 2015[5]). Further evidence from the 2023 Survey of Adult Skills suggests that the effect of years of education on earnings is greater than the effect of information-processing skills (OECD, 2024[23]). This is possibly because the number of years spent in education captures a wider range of skill acquisition processes, including social and emotional skills, and may hold a signalling value for employers. Results presented in Table 4.2 allow for the identification of between-group differences in hourly wages and the extent to which such differences reflect differences between socio-demographic groups in educational attainment, 21st-century skills, and the distribution of individuals across different occupations and sectors.
The analysis presented complements Causa (Forthcoming[16]), which examines wage disparities by parental education in greater detail – explicitly accounting for differences in employment probabilities and recognising that selection into employment itself is influenced by background characteristics. By contrast, the estimates in this report reflect differences in earnings among those who are employed, offering a descriptive overview of disparities without correcting for differential labour-market participation. The two approaches are thus complementary: Causa (Forthcoming[16]) focuses on intergenerational differences in wages by parental education, whereas this report provides a broader mapping of disparities across a range of socio-demographic groups and considers whether these may stem from unequal access to lifelong learning opportunities. Taken together, the two perspectives enrich the understanding for policy makers of how family background shapes labour-market outcomes, both through access to employment and through differences in the returns to education and skills once in work.
On average across OECD countries, men earn around 14% more per hour than women (Table 4.2, Model 1). The gender gap increases to 17% after additionally accounting for differences in learning trajectories and information-processing skills (Table 4.2, Model 2). However it is 16% when also accounting for social and emotional skills (Table 4.2, Model 3) and 14% when field-of-study is additionally controlled for (Annex Table 4.A.1). Accounting for social and emotional skills and individual level characteristics, such as health, presence of children and current residential context does not change estimates (Table 4.2, Model 4), whereas differences in earnings between men and women are 13% once controls for adults’ occupation and industry, experience, whether individuals work full or part-time are taken into account (Table 4.2, Model 5).
Differences in earnings between men and women vary considerably across countries. For example, they are highest in Japan, where men earn 39% more than women, and are smallest in the Flemish Region (Belgium), where the gender gap in earnings corresponds to 6% (Figure 4.2, basic adjusted model). The extent to which these gaps change after accounting for men’s and women’s educational trajectories, skills and participation in lifelong learning also varies across countries. For example, in Israel, the gender gap is considerably more pronounced, increasing from 18% to 26% when considering these additional factors, while in Norway, the gender gap remains constant at 11% (Annex Table 4.A.1). In Czechia, a large share of gender differences are explained by these factors, with the gender gap reducing from 21% to 17% once taking into account these additional factors (Annex Table 4.A.1).
Between-country differences suggest that even though the gender wage gap is a global phenomenon, and factors driving the gender wage gap are similar across countries, the importance of certain factors differ across countries. Therefore, certain policy responses may be more effective at translating into reductions in the gender wage gap in some countries than in others, suggesting the continued need for country- and context-specific policy responses.
Individuals with socio-economically advantaged backgrounds earn between 11-13% more per hour than individuals with socio-economically disadvantaged backgrounds, a difference that is slightly smaller compared to the disparity observed between men and women (Table 4.2, Model 1). Contrary to the gender pay gap, 70% of the difference between adults with high-status and low-status parents can be explained by additionally accounting for differences between the two groups in educational trajectories, skills and engagement in lifelong learning, reducing the earnings gap from 13% to 4% (Table 4.2, Model 3). In the case of parental education, the entire difference can be explained by the additional accounted differences between the two groups (Table 4.2, Model 3).
Socio-economic disparities in wages vary across countries. The wage difference between individuals with and without tertiary-educated parents is highest in Israel – where adults without tertiary-educated parents earn 21% less than their more advantaged counterparts – and smallest in France where it is 2% (Figure 4.2, basic adjusted model). However, in the fully adjusted model (after accounting for differences in educational trajectories, skills and participation in lifelong learning) differences in Israel are 7% and in France, adults with non-tertiary educated parents earn 5% more than those with tertiary educated parents.
Percentage difference in hourly earnings in purchasing power parity (PPP)-adjusted 2022 USD, OECD average
Note: Adults aged 30-65. The table reports coefficient estimates from a linear regression with log earnings as dependent variable. Earnings below the 1st percentile and above the 99th percentile are excluded to remove outliers. Earnings include bonuses and earnings by self-employed individuals. Coefficients in bold are statistically significant at the 5% level. Additional individual-level characteristics: of children, health, employment status of partner, and current residential context. Job and firm-related characteristics include: experience, firm size, part-time, public sector. Estimates in models (4) and (5) exclude Japan and the United States, these countries are excluded from the OECD average in all models presented in this table. Korea is not included in this table. See the note for Table 4.1 for the definitions of groups based on parental occupation, parental education, immigrant background, childhood residential context and respondents’ educational attainment.
Country-specific results are provided in Table 4.A.1.2 in Annex 4.A.
*Differences by immigrant background vary greatly across countries because of differences in the size of immigrant groups as well as composition and context. Readers are therefore encouraged to consult country-specific results in Table 4.A.1.2 in Annex 4.A.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
Differences in earnings between individuals with and without an immigrant background are large. Immigrants – those who migrated to their country of residence at age 18 or older – earn, on average, 14% less than individuals without an immigrant background (Table 4.2, Model 1), after accounting for socio-demographic characteristics including parental education and occupation, gender, age and childhood residential context. This earnings disparity decreases to around 8% when accounting for differences between the two groups in educational trajectories, skills and engagement in lifelong learning (Table 4.2, Model 3). When differences in occupation and industry factors are also considered, the gap further narrows to 4% (Table 4.2, Model 5).
Children of immigrants – those who were either born in the country in which they reside, but their parents were not, or who moved to the country of residence before the age of 18 – do not suffer a wage penalty compared to individuals without an immigrant background, on average, after accounting for other socio-demographic characteristics, educational attainment, skills, and additional individual-level characteristics (Table 4.2, Models 1-4). However, accounting for job and firm-level and occupation and industry-related characteristics, children of immigrants, on average, enjoy a wage premium of 2% compared to individuals without an immigrant background (Table 4.2, Models 5). However, there are large differences across countries, which reflect the variations in country of birth and socio-demographic characteristics of different immigrant communities, as well as differences in integration trajectories and labour market opportunities for individuals with different linguistic and cultural backgrounds (Annex Table 4.A.1).
Individuals who grew up in cities earn 7% more than individuals who grew up in villages, on average across countries, even when accounting for other socio-demographic characteristics (Table 4.2, Model 1). However, this gap also reflects differences between the two groups in educational trajectories, skills, engagement in lifelong learning and occupational stratification, as individuals who grew up in villages and cities have a similar wage after accounting for such factors (Table 4.2, Models 5 to 6).
These disparities also vary across countries. For example, in the United States, wage disparities between individuals who grew up in cities and those who grew up in villages are smallest (-4%) – favouring those who grew up in villages. Disparities are widest in Chile (23%) – favouring those who grew up in cities (Figure 4.2, basic adjusted model). However, such variations reflect differences between the two groups in educational trajectories, skills and engagement in lifelong learning. After accounting for these factors, the wage gap in Chile reduces from 23% – in favour of those who grew up in cities – to 3% in favour of those who grew up in villages (Figure 4.2, fully adjusted model).
Disparities in educational attainment shape earnings disparities between different groups. As explored in detail in Chapter 3, there are large differences across socio-demographic groups in educational trajectories, particularly in relation to socio-economic background; these trajectories, in turn, shape skills development. As Table 4.2 shows, adults with an upper secondary or a post-secondary qualification (regardless of a vocational or a general orientation) earn around 7-8% more than those without such a qualification (Table 4.2, Model 2). Adults with short-cycle tertiary qualifications earn 18% more, while those with a bachelor’s degree or higher earn 35% more. Most of these wage differentials reflect occupational differences, as jobs typically requiring tertiary qualifications – such as professional and associate professional occupations – tend to offer higher than average wages. For example, the earnings premium associated with having a bachelor’s degree or higher, compared to not having an upper secondary qualification, decreases from 35% to 19% when accounting for differences in wages across occupations and industries (Table 4.2, Model 5). Although it is impossible with cross-sectional data to partition the relative contribution of educational attainment and skills, Table 4.2 suggests that numeracy skills in particular are valued in addition to and beyond educational qualifications. Specifically, a one SD increase in numeracy is associated with a 5%1 increase in wages, when comparing individuals with similar socio-demographic characteristics, educational attainment, engagement in lifelong learning and working in similar occupations and industries (Table 4.2, Model 5).
Wage premiums differ by industry, occupation and across countries. For example, individuals employed in industries such as information and communications, financial and insurance activities, and those working in real estate can expect to earn around 19% more than similar individuals working in administrative and support service activities (Annex Table 4.A.1). By contrast, adults working in education can expect to earn around 5% less than otherwise similar individuals.
Workers with strong numeracy skills reap disproportionately large gains from further upskilling, with wage returns to numeracy especially high among those with the highest numeracy skills. Although the association between employment and certain 21st-century skills is non-linear, the results presented in Model 6 of Table 4.2 suggest that the wage returns to literacy and adaptive problem solving are broadly linear. These findings underline the need for policies that expand access to lifelong learning and training opportunities, particularly for individuals with low levels of proficiency in numeracy, ensuring that financial constraints, time limitations and employment insecurity do not hinder participation. Strengthening adult learning systems and targeted upskilling programmes can help make skill development more inclusive and support wage progression across the workforce.
Percentage difference in hourly earnings in PPP-adjusted 2022 USD by socio-demographic group basic and fully adjusted country-specific regression coefficients
Note: Adults aged 30-65. The figure reports coefficients for the certain socio-demographic groups – gender, age, parental education, parental occupation, immigrant background and childhood residential context – based on separate country estimates. Coefficients from the basic adjusted model follow the specification of Model 1 in Table 4.2, while those from the fully adjusted model follow the specification of Model 4 in Table 4.2. The fully adjusted model additionally accounts for variables that may not be available for all countries, which is why for some countries, only estimates in the basic adjusted model are available. See the note for Table 4.1 for the definitions of groups based on parental occupation, parental education, immigrant background and childhood residential context. For immigrant background, the figure only indicates a country if at least 200 adults for each group are part of the final PIAAC sample. Country-specific results are provided in Table 4.A.1.2 in Annex 4.A.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
A critical policy question for the design of lifelong learning strategies is whether investing in a broad mix of skills yields bigger returns than investing in specific skills, and whether the returns from developing a specific skill differ according to a learner’s existing skills set. If the wage returns to skills are additive, each skill should command roughly the same wage premium no matter which other attributes a worker already has. In this context, modular training – allowing people to engage in upskilling or reskilling in their weakest area or in skills that are most in demand – can be an efficient way to organise training. However, if the returns to skills are multiplicative, the value of any skill depends on which other skills it is combined with and the broader skills set of workers. For example, increasing numeracy skills might pay off only when accompanied by a capacity to delay gratification, while the advantages linked to behavioural tendencies associated with high scores on conscientiousness may accrue primarily to workers who already possess strong adaptive problem solving skills. In such circumstances, bundled interventions would likely be most efficient because the lack of complementary skills might sharply reduce the returns on any single skills investment.
Employers appear to value individual skills and behavioural tendencies but do not reward particular combinations of skills. In order to consider the relative merit of interventions promoting targeted standalone upskilling or comprehensive skill bundles when designing education, training and lifelong learning strategies, Table 4.3 presents the results of nine augmented wage‐equation specifications of Model 4 in Table 4.2. Each equation is built around one “focal” skill (listed as rows). Diagonal entries refer to wage premiums estimated for each focal skill, whereas off-diagonal entries refer to interaction terms between the focal skill and each of the remaining eight 21st-century skills. Statistically significant coefficients are marked in bold. The findings in Table 4.3 suggest that the wage returns to skills are largely independent of a worker’s broader skills profile, when comparing individuals with similar socio‑demographic characteristics, educational attainment, health status, work experience, number and age of their children, and who work in similar occupations and industries. Table 4.3 points to an additive rather than a multiplicative skills model: most interaction terms are not statistically significant and are quantitatively small. Based on this evidence, it appears that employers value numeracy, literacy, adaptive problem solving and specific behavioural tendencies, but they do not appear to reward particular combinations of skills. Put simply, having two strengths is worth the sum of their separate contributions, not a premium above this baseline.
Percentage change in hourly earnings in PPP-adjusted 2022 USD, OECD average
Note: Adults aged 30-65. Main effects are displayed along the diagonal from upper left to bottom right. Coefficients in bold are statistically significant at the 5% level. Estimates exclude Japan, Korea and the United States. Results are adjusted for differences in gender, age, parental education, parental occupation, childhood residential context, immigrant background, educational attainment, skills, occupation, industry and establishment (firm) size. The model also accounts for employment status of partner, subjective health status, number of children, number of years of experience, whether respondents were volunteering, whether they participated in non-formal adult education and training, part-time employment, and whether working in the public sector. Earnings include bonuses and earnings by self-employed individuals. See the note for Table 4.1 for the definitions of groups based on parental occupation.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
Understanding whether earnings returns to acquiring new skills vary across socio-demographic groups is critical for assessing the efficiency of investments in skills policies and the extent to which such policies should be accompanied by additional policy efforts to ensure different groups enjoy similar opportunities for personal well-being. For example, if wage returns to skills diverge for women and men, for mature and young adults, or for adults with different socio-economic backgrounds, similar training investments could yield systematically different changes in life chances, thereby perpetuating existing wage gaps even when access to training is equalised. Measuring such heterogeneities can help to identify whether wage disparities stem primarily from unequal skills acquisition, unequal valuation of those skills in the labour market, or a combination of both. Without this information, policy interventions risk treating symptoms rather than causes – either over-subsidising training that yields limited pay-offs for some groups, or neglecting complementary measures (e.g. anti-discrimination enforcement, certification standardisation) that ensure that skills translate into equivalent labour market rewards.
Earnings returns to literacy are only marginally larger among women than men whereas the earnings returns to numeracy are stronger for men than women. Similarly, the earnings returns to numeracy skills are larger for individuals whose parents worked in high- rather than low-status occupations, as shown in Table 4.4. This finding shifts the policy focus from differential valuation to differential access. As employers broadly value the skills of men and women equally, the emphasis should instead be on ensuring equitable access to learning opportunities for both men and women to acquire such skills. As outlined in Chapter 3, access to better formal learning environments is at least partially determined by an individual’s socio‑economic background (Domina, 2005[25]). Box 3.5 in Chapter 3 provides policy examples from across the OECD illustrating approaches to broadening access to lifelong learning opportunities to reduce socio-economic skills disparities.
Given that equivalent skills profiles command similar wage premiums, the wage and employment gaps observed appear to originate primarily from disparities in opportunities to develop, signal or deploy those skills rather than from employers’ wage-setting practices. This implies that closing participation gaps in high-quality training – especially in digital and science, technology, engineering and mathematics (STEM)-related domains – could translate more directly into narrowing wage disparities. It also underscores the importance of lowering non-monetary barriers to upskilling, as identified in Chapter 3, and ensuring that credentials gained by under-represented groups are portable and transparently recognised. Examples of relevant policies include the Irish Higher Education Access Route, which facilitates entry into higher education for students from disadvantaged backgrounds by adjusting entry requirements and reserving university spots for students with socio-economically disadvantaged backgrounds (see Chapter 3, Box 3.5.). Similarly, there are initiatives promoting girls’ and women’s participation in STEM courses, such as France’s “Girls and Maths” plan which couples targets for girls’ enrolment in specialised mathematics courses with investment in different resources throughout the duration of formal education (see Chapter 3, Box 3.3).
Percentage change in hourly earnings in PPP-adjusted 2022 USD, OECD average
Note: Adults aged 30-65. Coefficients in bold are statistically significant at 5% level. Estimates exclude Japan, Korea and the United States. Results in column (1) display the percentage change earnings for men only, column (2) for women only, column (3) for high-status parental occupations and column (4) for low-status parental occupations. All columns adjust for differences in gender – except columns (1) and (2) – age, parental education, parental occupation – except columns (3) and (4) – childhood residential context, immigrant background, educational attainment, skills, occupation, industry and establishment (firm) size. All columns also account for employment status of partner, subjective health status, number of children, number of years of experience, part-time employment, whether working in the public sector, whether respondents were volunteering, and whether they participated in non-formal adult education and training. Earnings include bonuses and earnings by self-employed individuals. See the note for Table 4.1 for the definitions of groups based on parental education.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
While wages play a crucial role in influencing employment choices, job satisfaction profoundly impacts individuals' overall happiness, health, productivity and personal growth, and expectations of satisfaction drive career decisions (Faragher, Cass and Cooper, 2005[26]; Judge et al., 2001[27]). Higher job satisfaction is correlated with improved mental health outcomes, reduced stress, enhanced self-esteem and stronger interpersonal relationships, making it integral to achieving broader societal goals related to personal and communal well-being (Helliwell and Huang, 2010[28]). Workers who are satisfied in their job are less likely to leave their job and to invest effort in what they do, increasing productivity and promoting growth. Job satisfaction is also strongly influenced by workplace characteristics such as autonomy, job security and opportunities for participation in learning opportunities, which are central components of job quality. Enhancing these aspects can simultaneously improve worker well-being and organisational performance, reinforcing the link between satisfaction and productivity. Examining disparities in job satisfaction and whether such disparities arise because of differences in skills and/or educational attainment allows for the creation of adequate support mechanisms, training and credential support if skills are the constraint, or the promotion of better workplace practices if other factors are at play. Tracking disparities in job satisfaction ensures that investments in employment and education translate into high-quality, sustainable jobs for all groups, improving the efficiency of public spending by reducing turnover and underemployment.
Educational trajectories and skills development could potentially mediate or exacerbate disparities in job satisfaction, as individuals with higher educational attainment and advanced 21st‑century skills are generally more likely to secure employment offering intrinsic rewards, autonomy and opportunities for advancement, thus enhancing their job satisfaction (Autor and Handel, 2013[29]; Oreopoulos and Salvanes, 2011[30]). Investigating whether differential access to education and skills formation contributes significantly to observed disparities can support targeted policy interventions aimed at promoting inclusive workplace environments.
On average, job satisfaction differs little depending on individuals’ educational attainment or engagement in lifelong learning and skills. However, exceptions exist for adults holding a bachelor's degree or higher, who, all else being equal, tend to report higher job satisfaction. The link between educational attainment, skills and job satisfaction varies notably across countries. For example, after controlling for other socio-demographic characteristics, adults in Korea with a bachelor's degree or higher are 12 percentage points more likely to report high job satisfaction compared to those who did not complete upper secondary education (Annex Table 4.A.1). Conversely, in New Zealand the opposite relationship emerges, with adults with a bachelor's degree or higher significantly less likely to report high job satisfaction (18 percentage points).
Change in probability of reporting high job satisfaction (being satisfied or extremely satisfied with the job) (in percentage points), OECD average
Note: Adults aged 30-65. Coefficients in bold are statistically significant at the 5% level based on a linear probability model. Additional individual-level characteristics: of children, health, employment status of partner, experience, part-time, public sector and current residential context. Job and firm-related characteristics include: experience, firm size, part-time, public sector, permanent contract. Survey question used to measure job-satisfaction: “All things considered, how satisfied are you with your current work?”. High job satisfaction comprises the answer categories “Extremely satisfied” and “Satisfied”, while low job satisfaction comprises the answer categories “Neither satisfied nor dissatisfied”, “Dissatisfied” and “Extremely dissatisfied”. See the note for Table 4.1 for the definitions of groups based on parental occupation, parental education, immigrant background, childhood residential context and respondents’ educational attainment. Country-specific results are provided in Table 4.A.1.3 in Annex 4.A.
*Differences by immigrant background vary greatly across countries because of differences in the size of immigrant groups as well as composition and context. Readers are therefore encouraged to consult country-specific results in Table 4.A.1.3 in Annex 4.A.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
Levels of job satisfaction differ little depending on individuals’ socio-demographic characteristics. Individuals from advantaged and disadvantaged backgrounds, adults with and without immigrant backgrounds, as well as younger and older adults, report similar levels of job satisfaction (Table 4.5). Differences linked to the gender, residential context during childhood are statistically significant, although quantitatively small.
Countries differ with respect to the magnitude of disparities in job satisfaction and the extent to which educational attainment, skills, lifelong learning participation and occupational sorting influence these disparities. After only accounting for other socio-economic characteristics, in Germany, men are 4.4 percentage points more likely than women to report high job satisfaction, whereas in Latvia, women are 2.6 percentage points more likely than men to report high job satisfaction (Annex Table 4.A.1). However, when comparing men and women with similar educational attainment, skills, lifelong learning participation, occupation and industry, the gender gap disappears in Germany (reducing to 0.7 percentage points) and increases in Latvia (increasing to 8.1 percentage points) (Annex Table 4.A.1).
There are also differences in job satisfaction when considering parental education. As shown in Figure 4.3 (basic adjusted model), in Germany, adults whose parents are tertiary educated are 5.8 percentage points more likely to report high job satisfaction compared to adults whose parents are not tertiary educated. Conversely, in Sweden, adults whose parents are not tertiary educated are almost 2.4 percentage points more likely to report high job satisfaction than adults whose parents are tertiary educated – although this difference is not significant. Accounting for differences in adults' educational attainment, skills, lifelong learning participation and occupational sorting does not reduce disparities in Germany substantially (reducing to 4.12 percentage points), but significantly accentuates them in Sweden (increasing to 3.9 percentage points).
Change in probability of reporting high job satisfaction (being satisfied or extremely satisfied with the job) by socio-demographic group: basic and fully adjusted country-specific regression coefficients
Note: Adults aged 30-65. The figure reports coefficients for certain socio-demographic groups – gender, age, parental education, parental occupation, immigrant background and childhood residential context – based on separate country estimates. Coefficients from the basic adjusted model follow the specification of Model 1 in Table 4.5, while those from the fully adjusted model follow the specification of Model 4 in Table 4.5. The fully adjusted model additionally accounts for variables which may not be available for all countries, which is why for some countries, only estimates in the basic adjusted model are available. Survey question used to measure job-satisfaction: “All things considered, how satisfied are you with your current work?”. High job satisfaction comprises the answer categories “Extremely satisfied” and “Satisfied”, while low job satisfaction comprises the answer categories “Neither satisfied nor dissatisfied”, “Dissatisfied” and “Extremely dissatisfied”. See the note for Table 4.1 for the definitions of groups based on parental occupation. parental education, immigrant background and childhood residential context. For immigrant background, the figure only indicates a country if at least 200 adults for each group are part of the final PIAAC sample. Country-specific results are provided in Table 4.A.1.3 in Annex 4.A.
Source: Calculations based on OECD (2024[23]), Survey of Adult Skills (PIAAC) 2nd cycle database, www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html.
The previous section highlighted substantial wage disparities across different occupations and industries which, in turn, contribute to disparities in labour market outcomes – particularly between men and women, and between adults from socio-economically advantaged and disadvantaged backgrounds – even when they possess similar levels of information-processing skills and educational attainment. Occupational and sectoral clustering – where certain groups are disproportionately represented in particular occupations – emerges as a key driver of these disparities.
Occupations may function as meaningful skills ecosystems for three distinct reasons: 1) people with similar skills may be attracted by the same job characteristics and requirements; 2) employers may select candidates with specific skills for distinct roles, as differences in production processes require specific skill sets; and 3) jobs shape workers over time: through daily practice, individuals performing the same tasks tend to develop similar skills and behavioural traits.
While occupations are commonly perceived as requiring distinct personality traits and skills, in reality there is considerable overlap in the underlying skills needed across many roles (Anni, Vainik and Mõttus, 2024[31]); however, little evidence exists on between and within variations in skills across occupations – see (Anni, Vainik and Mõttus, 2024[31]; Wolfram, 2023[32]) for important exceptions. This suggests that prevailing beliefs and stereotypes about the suitability of specific occupations for particular demographic groups may inadvertently influence career choices and employment opportunities. Such biases can lead to the systemic clustering of different groups by occupation, limiting individuals’ labour market opportunities and reinforcing wage inequalities. Understanding and addressing the drivers of this clustering is therefore essential to reducing disparities and promoting fairer and more inclusive labour market outcomes.
4.5.1. Occupational roles
As structural shifts (such as technological change, demographic shifts and the green transition) and shocks (such as geopolitical crises) reshape economies, understanding the skills requirements of different occupations is essential to identify areas at risk of labour shortages. For career orientation programmes and skills-first initiatives (OECD, 2025[33]) to be effective, it is equally important to map the skills requirements of occupations with precision.
There is limited systematic analysis of differences in skills distribution – including personality traits and social and emotional skills – across occupations, with most such studies focusing on specific occupations or small samples (Booth et al., 2015[34]; Furnham, 2017[35]; Lan, Wong and Zeng, 2021[36]). However, a few comprehensive studies have systematically examined personality traits across a wide range of occupations (Anni, Vainik and Mõttus, 2024[31]; Törnroos, Jokela and Hakulinen, 2019[37]; Wolfram, 2023[32]). Insight into how personality traits align with occupational requirements can support effective career planning, recruitment and career coaching strategies.
This section examines the skill profiles of 458 occupations – defined at the four-digit International Classification of Occupations (ISCO) level – in information-processing skills, willingness to delay gratification, and social and emotional skills. The analysis draws on pooled data from the 2023 Survey of Adult Skills, with each country and economy weighted according to its population of 16-65 year-olds.2
The analysis breaks down the total variance in each skill into two components: 1) a between-occupation component, which captures the extent to which individuals in different jobs have different skills; and 2) a within-occupation component, which reflects the variation in skills among workers performing the same job. For information-processing skills, the between-occupation component accounts for 26‑29% of the overall variability observed across participating countries (Table 4.6, Column 1). This means that, on average, people who perform different jobs tend to differ markedly in these skills: about one-quarter of the skills gap between two randomly selected individuals in different jobs comes from the fact that their jobs tend to attract (or require) different average proficiency levels. However, the remaining three-quarters of the variation reflects random individual level differences, even among workers in the same job. The degree to which occupation explains skills gaps increases once socio-demographic characteristics are considered. When controlling for these factors, the between-occupation component accounts for 33‑35% of the overall variability in information-processing skills (Table 4.6, Column 2).
For delayed gratification and social and emotional skills, the pattern is markedly different. The between-occupation component accounts for 11% of the variation in willingness to delay gratification and between 3% and 8% in social and emotional skills (Table 4.6, Column 1). When socio-demographic characteristics are taken into account, the between-occupation component accounts for 13% in willingness to delay gratification and between 5% and 9% in social and emotional skills (Table 4.6, Column 2). Thus, while occupations form distinctive skills ecosystems for literacy, numeracy and adaptive problem solving, they are far less differentiated with respect to social and emotional skills. People performing the same job tend to resemble each other to a greater extent in terms of numeracy, literacy and adaptive problem solving than social and emotional skills. Most of the dispersion in social and emotional skills and in delayed gratification occurs within rather than between jobs. This implies that workers with very different behavioural tendencies are attracted to the same job, but also that employers see the benefits of a wide range of individuals with different social and emotional skills bringing value to a job. Furthermore, whereas information-processing skills appear to be important in shaping who can successfully perform certain tasks, a wide variety of individuals with different behavioural tendencies can support the same work processes.
Variance proportions (adjusted R2) of the skills accounted for by occupations
Note: Occupations are coded based on the International Standard Classification of Education (ISCED) (four-digit level). Column 1: only accounts for occupations. Column 2: accounts for occupations and additional for other socio-demographic characteristics include gender, age, age squared, parental education, parental occupation, childhood residential context and immigrant background. Country fixed effects are included in all models with literacy, numeracy, adaptive problem solving and delayed gratification. They are not included in models with social and emotional skills as these are within-country centred (see Comparison between groups and countries across different skills in the Reader’s Guide).
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
These results suggest that there is a high degree of overlap in the set of occupations that require strong information-processing skills and those that do not, reflecting the high correlation between information-processing skills and the fact that certain occupations may require individuals to possess high proficiency in all such skills. Table 4.7 displays occupations with the lowest and highest mean scores in literacy, numeracy and adaptive problem solving illustrate differences for the remaining 21st-century skills analysed in this report. Occupations requiring the highest proficiency in these skills include technical occupations requiring attention to detail and strong information-processing capacity, such as mathematicians, actuaries and statisticians; town and traffic planners; and geologists and geophysicists. By contrast, occupations with the lowest mean scores in literacy, numeracy and adaptive problem solving tend to be manual occupations that do not require workers to perform complex information-processing tasks, such as poultry producers, concrete placers, concrete finishers and related workers, and laundry machine operators.
Note: The table shows the five occupations with the highest (top) and lowest (bottom) standardised mean scores in literacy, numeracy and adaptive problem solving. Occupations are consistently colour-coded. Tables 4.A.2.1, 4.A.2.2 and 4.A.2.3 in Annex 4.A provide detailed information on differences in occupations for literacy, numeracy and adaptive problem solving.
Source: Based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
Levels of willingness to delay gratification and social and emotional skills vary considerably more than information-processing skills within occupations (Tables 4.A.2.4 to 4.A.2.9 in Annex 4.A). At the same time, on average there are some differences in the social and emotional skills profile of workers in different occupations. For example, adults working as traditional and complementary medicine professionals, religious professionals, domestic housekeepers, health services managers and life science technicians report the highest willingness to delay gratification, and adults working as aircraft pilots and related associate professionals, ships’ engineers, non-commissioned armed forces officers, database and network professionals, and paper products machine operators report the highest levels of emotional stability. Workers in artistic occupations report behavioural tendencies associated with the highest levels of openness to experience. These include creative and performing artists; musicians, singers and composers; visual artists; and arts and music teachers.
4.5.2. Changes in the mix of occupations and their skills requirements
Technological advances such as the rapid adoption of artificial intelligence (AI) tools and applications, demographic shifts, policies to adapt to climate change and mitigation policies to reach net-zero commitments, alongside geopolitical instability and evolving consumer preferences, are reshaping labour markets across OECD economies. These changes are significantly altering labour markets in two distinct but mutually reinforcing ways: 1) they are changing the mix of occupations that are in demand; and 2) they are changing the skills required within each occupation.
Men and women, and adults with different socio-economic backgrounds, are not equally likely to work in the same set of occupations. As a result, patterns of clustering of different groups by occupation could significantly influence the future labour market risks and opportunities faced by different groups. A clear view of where employment is growing or shrinking, and how task content is changing, is therefore essential for designing upskilling and reskilling initiatives that cushion disruption, promote inclusive growth and narrow disparities in life chances through labour market integration.
By identifying groups most at risk of job displacement, and who thus require employment transitions, or skills obsolescence due to rapid job specific shifts, this section provides evidence that policy makers can use to direct efforts and resources more efficiently, enabling them to prioritise training initiatives that are aligned with the needs of different populations.
Occupations can be categorised according to four dimensions to enable the identification of the labour market vulnerabilities and opportunities faced by workers in different occupations, and to define targeted reskilling and upskilling programmes for workers who may expect to see their job disappear in the near future because of structural changes, or who may expect their existing job to undergo extensive transformations. These four dimensions are:
1. Employment projection: This dimension involves distinguishing between occupations projected to experience growth or contraction in total employment over the next decade, enabling the identification of job roles likely to expand or decline in the future given current expectations of labour market dynamics. This takes into account advances in AI and digital technologies, adaptation and mitigation efforts due to climate change and demographic shifts, and geopolitical uncertainty. The employment projection measure used in this section reflects projected occupational-level employment growth between 2023 and 2033 (U.S. Bureau of Labor Statistics, 2025[38]).
2. Skills evolution: This dimension involves differentiating occupations based on whether the skills required to perform a particular occupation have remained stable or have undergone rapid transformations. This enables the identification of roles in which adaptability and continuous learning are essential because of rapid transformation in key tasks performed by workers in such jobs. Lightcast’s Skills Disruption Index (Lightcast, 2025[39]) is used in this section to reflect occupational-level skills evolution. This index was derived using information available in more than 2.5 billion online job postings on job boards, newspapers and employer websites between 2021 and 2024. The index ranges between 0 and 100, with higher scores indicating a greater degree of skills change.
3. Earnings: This relates to the wage level commanded by workers. Earnings information in this section reflects gross hourly wages adjusted for purchasing power parity (PPP) and is expressed in 2022 USD of workers surveyed in the 2023 Survey of Adult Skills.
4. Size: This relates to the number of workers and in this section reflects the number of people in specific occupations in the 2023 Survey of Adults Skills.
Figure 4.4 identifies occupations according to the degree of skills evolution they experienced between 2021 and 2024, and employment projections. Occupations in the top right quadrant combine rapid skills shifts and robust employment projections. Occupations in the bottom right quadrant combine rapid skills shifts and declining employment projections. Occupations in the top left quadrant combine relatively stable skills requirements and robust employment projections. Occupations in the bottom left quadrant combine relatively stable skills requirements and declining employment projections. For each occupation, Figure 4.4 further indicates current employment – as indicated by the size of the bubble – and earnings – as indicated by the colour of the bubble.
Projected employment change between 2023 and 2033 and skills disruption across countries participating in the 2023 Survey of Adult Skills, by occupation
Note: The Skills Disruption Index measures how employer skills requirements have evolved across different occupations between 2021 and 2024. It ranges from 0 to 100, where a higher score indicates a greater degree of skill change. Projected employment change was calculated after converting the occupational classification into ISCO-08 (from OES2023 via SOC2010) using the crosswalk tables provided by the U.S. Bureau of Labor Statistics.
Source: Calculations based on Lightcast (2025[39]), The Speed of Skill Change, https://lightcast.io/resources/research/speed-of-skill-change; OECD (2024[22]); PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html; and U.S. Bureau of Labor Statistics (2025[38]), Table 1.2 Occupational projections, 2023-2033, and worker characteristics, 2023 (Numbers in thousands), www.bls.gov/emp/tables/occupational-projections-and-characteristics.htm.
Occupations can be categorised into four groups:
1. Transformative growth roles. These roles combine rapid skills obsolescence with robust employment prospects. Occupations in the top right quadrant of Figure 4.4 are those undergoing intensive skill changes but that are, on average, anticipated to grow in demand in the future. Specialist professions such as mathematicians, actuaries and statisticians combine very high disruption scores and high expected expansion in overall demand. Most high-wage occupations are in this group, and workers in these occupations can expect to have to regularly update their skills to be able to continue to perform in their job. However, because the job itself is expected to increase in demand, workers can also expect their employers to have a strong incentive to support or facilitate such training either directly (by providing this in house) or indirectly (by allowing their workers to invest time and resources in upskilling and reskilling), with learning opportunities ranging from paid micro-credential courses to in-house academies. Demand for many occupations in this group is fuelled by the increased datafication of business processes, regulatory scrutiny of risk and the diffusion of AI tools.
2. Human-centric expansion roles. These roles combine relatively stable task profiles with growing employment prospects; examples include the occupations in the top left quadrant of Figure 4.4. Occupations with high interpersonal or manual components – physicians, nurses, electricians and other green-infrastructure trades – face incremental technological change and rising demand driven by population ageing, the green transition and adaptation efforts to reduce the impact of climate change in infrastructure. Although on-average wage levels are markedly lower than for occupations in the transformative growth group, some occupations in this group enjoy very high wages, for example airline pilots and medical doctors. Upskilling for workers in these occupations generally focuses on deepening professional practice rather than the rapid integration of new tasks and processes.
3. Disrupted declining role. These roles combine rapid skills obsolescence with declining employment prospects; examples can be seen in the bottom right quadrant of Figure 4.4. Roles such as application programmers illustrate how fast-moving technological changes due to the emergence of generative AI can erode demand even while rapidly transforming task content. This is because low-code platforms and generative AI reduce the need for routine coding but raise the premium on system architecture and cyber-security.
4. Routine retreat roles. These roles combine relatively stable task profiles with declining employment prospects; examples include occupations in the bottom left-hand quadrant of Figure 4.4 such as telephone operators, data entry clerks and contact-centre salespersons who perform routine, codifiable tasks that are easy to automate or offshore. As firms operating in markets that employ these workers have little incentive and/or capacity to invest in employees who are operating in shrinking functions that do not require immediate upskilling, external public support is needed to support these workers’ transitions paths. Given the possible geographic concentration of workers in many of these roles, training provision and other active labour market policies may have to be coupled with local regeneration efforts to create viable bridges to help workers move to in-demand occupations and prevent their long-term detachment from the labour market.
Notably, the cluster enjoying both rapid skills changes and strong employment growth includes many high-paying occupations, whereas many lower-wage roles are in quadrants where immediate change is slower. This may reduce the short-term pressure for training workers in these occupations, but it also significantly limits their exposure to new skills and dispositions that would support their mobility to higher paying and in-demand occupations. Historical evidence suggests that when disruptive change eventually reaches a broad spectrum of occupations, workers without recent learning experience suffer deeper displacement and longer spells out of work (Schmidpeter and Winter-Ebmer, 2021[40]).
What ties these four occupational groupings together is the need for a risk-weighted incentive architecture for skills investments – namely, a system that encourages workers to invest in their skills while remaining mindful of the expected returns on those investments. By tailoring incentives to the distinct trajectories revealed in Figure 4.4, policy can nudge both workers and employers towards the continuous learning that resilient labour markets demand but that, as detailed in Chapter 3, remains out of reach for many. Broadening accessibility may take many different forms, for example, programmes can target lifelong learning specifically while giving individuals agency over their reskilling trajectory, such as France’s Personal Training Account (compte personnel de formation), which gives adults access to vetted training via a secure online platform. On the other hand, broader investments in spaces or initiatives where learning can happen, such as Mexico’s UTOPÍAS initiative, which provides funding for repurposing neglected spaces into free-access centres offering educational, cultural, health and recreational services in Iztapalapa, Mexico City, can also promote lifelong learning (see Chapter 3, Box 3.10.).
How occupational transformations impact skills disparities and training investment
This section examines the distribution of women and adults with socio-economically disadvantaged backgrounds into occupations characterised by different occupational prospects and changing skills requirements. By understanding this distribution, it is possible to determine the different needs of certain groups to engage in lifelong learning, alongside different incentives and opportunities to train. Establishing the scale of labour market occupational clustering is necessary for assessing whether current training and career-guidance policies are aligned with the segments of the workforce most likely to require support when future adjustments occur. It can also help with understanding who can already expect to enjoy better access to employer-provided training, formal credentials and professional networks.
Socio-economic disparities
Individuals with socio-economically advantaged backgrounds are more likely to be in transformative growth roles, which are higher paying occupations that have experienced some of the largest skills changes in recent years and are projected to increase in demand in the medium term. For instance, mathematicians, actuaries and statisticians, which are among the highest-earning occupations, have a very large share of current workers who have at least one parent educated at the tertiary level (79%) (Figure 4.5). These occupations have also experienced a large degree of skills change in recent years (Disruption Index of 78) and are projected to experience robust employment prospects (employment increase of 12% by 2033). Many transformative growth roles are also technical in nature – such as software developers, electrotechnology engineers, physicists and astronomers – and current workers are predominantly individuals with tertiary-educated parents (Figure 4.6). This is because many transformative growth roles typically require tertiary educational qualifications and, as illustrated throughout this report, individuals with tertiary-educated parents are considerably more likely than those without tertiary-educated parents to pursue and complete a tertiary degree themselves.
Share of workers with tertiary-educated parents and skills disruption across countries participating in the 2023 Survey of Adult Skills, by occupation
Note: The Skills Disruption Index measures how employer skill requirements have evolved across different occupations. It ranges from 0 to 100, where a higher score indicates a greater degree of skill change. See the note for Table 4.1 for the definitions of groups based on parental education.
Source: Calculations based on Lightcast (2025[39]), The Speed of Skill Change, https://lightcast.io/resources/research/speed-of-skill-change and OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
Most occupations with only a minority of current workers with advantaged socio-economic backgrounds command low wages and entail routine retreat roles, meaning that they combine relatively stable task profiles and declining employment prospects. For example, only around 2% of hand launderers and pressers have at least one tertiary-educated parent, an occupation with a projected employment decline of 12%. These workers can expect to earn below-average hourly wages, face a high risk of job displacement and will experience little to no skills change in their current occupation.
Share of workers with tertiary-educated parents and projected change in employment across countries participating in the 2023 OECD Survey of Adult Skills, by occupation
Note: Projected employment change was calculated after converting the occupational classification into ISCO-08 (from OES2023 via SOC2010), using the crosswalk tables provided by the U.S. Bureau of Labor Statistics. See the note for Table 4.1 for the definitions of groups based on parental education.
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html and U.S. Bureau of Labor Statistics (2025[38]), Table 1.2 Occupational projections, 2023-2033, and worker characteristics, 2023 (Numbers in thousands), www.bls.gov/emp/tables/occupational-projections-and-characteristics.htm.
Gender disparities
Gender differences are less pronounced than differences by parental occupation; however, Figure 4.7 reveals some key differences in the types of occupations men and women work in. Women are more likely to work in occupations that are experiencing low to medium skills changes (the left part of Figure 4.7) whereas men are more likely than women to work in jobs with both very high and very low changing skills requirements. Many occupations in the left side of Figure 4.7, i.e. routine retreat roles and human-centric expansion roles, are women-majority or men-majority occupations, i.e. occupations in which the majority of workers are of the same gender. These occupations command below average wages. Data entry clerks and telephone switchboard operators are examples of women-majority routine retreat roles occupations, whereas physiotherapy technicians and assistants are examples of women-majority human-centric expansion roles. Metal working machine tool setters and operators are examples of men-majority routine retreat roles occupations, whereas building and related electricians are examples of men-majority human-centric expansion roles. By contrast, men are more likely than women to work in transformative growth roles, occupations in the top right quadrant of Figure 4.7, although, on average, the transformative growth roles tend to have a less skewed gender composition than routine retreat roles.
Projected employment growth 2023-2033 and skills disruption across countries participating in the 2023 Survey of Adult Skills, by occupation
Note: The Skills Disruption Index measures how employer skill requirements have evolved across different occupations between 2021 and 2024. It ranges from 0 to 100, where a higher score indicates a greater degree of skill change. Projected employment change was calculated after converting the occupational classification into ISCO-08 (from OES2023 via SOC2010) using the crosswalk tables provided by the U.S. Bureau of Labor Statistics.
Source: Calculations based on U.S. Bureau of Labor Statistics (2025[38]), Table 1.2 Occupational projections, 2023-2033, and worker characteristics, 2023 (Numbers in thousands), www.bls.gov/emp/tables/occupational-projections-and-characteristics.htm; Lightcast (2025[39]), The Speed of Skill Change, https://lightcast.io/resources/research/speed-of-skill-change; and OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
The clustering of women and men into distinct occupations matters for three reasons: 1) wage premiums; 2) reskilling opportunities; and 3) labour shortages. First, wage premiums accrue disproportionately to roles where skills requirements are evolving quickly and for which demand is growing but supply is not growing at the same pace. Women appear to be under-represented in these transformative growth roles and over-represented in routine retreat roles. If women remain clustered in low-disruption jobs, the gender pay gap is liable to widen even if overall employment rates were to converge. Second, most large-scale reskilling initiatives – particularly those focused on digital, data or green competences – are designed and target transformative growth roles (Frey, Alajääskö and Thomas, 2024[41]; OECD, 2024[42]; 2021[43]). Unless such programmes explicitly reach out to women, they risk channelling public resources towards male workers and bypassing many women whose jobs face stagnation or obsolescence rather than transformation. Third, labour shortages are already acute in several of the high-growth occupations where women are least represented. Persisting stratified employment patterns therefore constrain the effective supply of talent and threaten to dampen productivity gains across the OECD.
From disruption to opportunity: Pathways for workforce adaptation
Analysing the adaptive problem solving skills of individuals working in occupations experiencing rapid skills change, or who will most likely have to transition to new roles because of changing structural factors and shrinking opportunities, can provide policy makers with indications of where support and initiatives will need to be targeted. Encouragingly, Figure 4.8 suggests that adults currently working in roles with above-average changes in skills requirements have higher than average proficiency in adaptive problem solving, which means that although these individuals are likely to need to invest in significant upskilling and reskilling to be able to continue to perform tasks as their role evolves, they will have the skills to manage and adapt to the transition and thrive in a future role.
Mean proficiency in adaptive problem solving and Skills Disruption Index by occupation across countries participating in the 2023 OECD Survey of Adult Skills
Note: The Skills Disruption Index measures how employer skill requirements have evolved across different occupations. It ranges from 0 to 100, where a higher score indicates a greater degree of skill change.
Source: Calculations based on Lightcast (2025[39]), The Speed of Skill Change, https://lightcast.io/resources/research/speed-of-skill-change and OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
However, individuals who work in occupations with declining projected demand typically have low proficiency in adaptive problem solving, as indicated in Figure 4.9. Because structural adjustments may require some (or many) of these individuals to transition into different roles, low proficiency in adaptive problem solving skills may constrain their upskilling and reskilling efforts. Without targeted support, these adults risk a double penalty of job loss or wage erosion, combined with a steeper learning curve when trying to retrain.
Mean proficiency in adaptive problem solving and projected change in employment by occupation across countries participating in the 2023 OECD Survey of Adult Skills
Note: Projected employment change was calculated after converting the occupational classification into ISCO-08 (from OES2023 via SOC2010) using the crosswalk tables provided by the U.S. Bureau of Labor Statistics.
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html and U.S. Bureau of Labor Statistics (2025[38]), Table 1.2 Occupational projections, 2023-2033, and worker characteristics, 2023 (Numbers in thousands), www.bls.gov/emp/tables/occupational-projections-and-characteristics.htm.
Overall, these results indicate a stark mismatch between where labour demand is growing and where many women and adults with socio-economically disadvantaged backgrounds are currently employed. Many occupations that mostly employ adults from disadvantaged socio-economic backgrounds are in routine retreat or human-centric roles. Whereas human-centric roles are increasing in demand – because of the demographic transition and the limited impact AI so far has on these roles – routine retreat roles are shrinking – likely due to automation and structural adjustments to meet net-zero targets. These are roles that command low-to-modest wages and, because they are not directly impacted by rapid skills obsolescence, are likely to receive limited employer investment in training. Meanwhile, transformative and disrupted declining roles – those most affected by AI – command higher pay but remain men-majority and socially selective.
Employers in industries relying on transformative roles have a direct incentive to invest in training their workforce because of rapidly changing skills requirements. At the same time, because of increased demand for such roles, they have an incentive to provide training that is as company specific as possible to reduce the likelihood that workers will move to other companies. Public policy should focus on the portability and recognition of micro-credentials so that advanced skills diffuse across firms and regions rather than remaining siloed in specific companied. Disrupted declining roles, many of which are in “white-collar” occupations, need early-warning reskilling programmes to help them transition into emerging AI-complementary occupations and become transformative roles. Human-centric growth roles – especially in health and care – call for expanded apprenticeship places and faster foreign-qualification recognition that attracts under-represented groups in order to ease shortages (OECD, 2023[44]). Finally, routine retreat roles require a combined strategy of local regeneration and transition stipends that finance full-time retraining, as incremental upskilling will not offset looming job losses.
Low adaptive problem solving proficiency among workers in declining occupations highlights the need for scaffolded learning pathways, and could partly explain the low willingness of some workers to engage in adult learning, as identified in Chapter 3. Short, confidence-building modules on how to deal with change that blends training on basic ICT use, problem solving and peer mentoring can act as an “on-ramp” before more demanding technical courses are introduced.
This section quantifies the transmission of occupational and educational advantage across generations and maps differences across countries. It identifies countries where individuals attain educational qualifications and occupational status that are equivalent, exceed or fall short of those of their parents. Because young people’s expectations reflect perceived opportunity structures and can predict future distributions of attainment, this section combines an analysis of historical patterns of observed intergenerational transmission of educational attainment and occupational status in adult populations with an analysis of patterns of expectations among school-aged children. If teenagers have expectations not aligned with available educational pathways or labour market conditions, disparities risk becoming entrenched or even widening further. The integrated examination of realised transmission and stated expectations provides policy makers with detailed evidence for targeted interventions and supports the design of measures that promote opportunities, calibrate expectations to realistic educational and labour market pathways, and promote inclusive economic growth across OECD countries.
The analysis in this section complements ongoing OECD work on social mobility and opportunity. For example, recent evidence documents sizeable earnings disparities by parental educational attainment, as well as the central role family background plays in shaping labour market outcomes (Causa, Forthcoming[16]). The section also provides a micro-to-macro bridge to new estimates of inequality of opportunity, indicating that at least one-quarter of total inequality in market income reflects circumstances beyond individuals’ control, and that parental socio-economic background is often the single largest contributor to inequality of opportunity (OECD, 2025[17]). Finally, by linking adult outcomes to the formation of expectations, this section complements ongoing efforts to map the extent to which teenagers’ career and educational expectations reflect their socio-economic background, frequently more so than academic results. These expectations are increasingly concentrated in a narrow set of high-status occupations, yielding misalignments with labour market demand and unequal engagement in career development activities (OECD, 2025[18]).
The opportunities individuals have to match or exceed the educational qualifications or occupational status of the previous generation are strongly influenced by the timing and scope of structural transformations. Productivity growth and rising living standards typically lead to increases in educational attainment or the prevalence of workers in services. However, ceiling effects in tertiary enrolment and occupational structures constrain further upward movement in education and occupation. Countries that did not expand higher education or shift towards service-sector employment until the latter part of the 20th century therefore have greater opportunities to fare well on measures of intergenerational transmission of education or occupation, because adults in these countries have considerably greater chances of surpassing their lower-than-average parental educational attainment or occupational status. In countries where these changes occurred earlier, previous cohorts already occupy much of the upper distribution, thereby limiting additional advancement (OECD, 2018[45]).
Inadequate social mobility imposes an economic cost, with limited upward mobility resulting in the underutilisation of talent as individuals with the potential to excel in education or in occupations that require advanced proficiency in information-processing skills (such as professional and managerial occupations) are held back because of their background. Credit constraints, informational deficits and insufficient family resources reduce investment in education and entrepreneurship, depressing aggregate productivity. Conversely, restricted mobility at the top can entrench economic rents and facilitate opportunity hoarding, distorting resource allocation (OECD, 2018[45]).
4.6.1. Intergenerational occupational (im)mobility
Figure 4.10 illustrates the link between the social status of the occupation of adults who participated in the 2023 Survey of Adult Skills and the social status of their parents, as measured through the occupation of the parent with the higher occupational status within each household, alongside the average numeracy proficiency of those who experienced different occupational transitions. Nine occupational categories are featured, ranging from those with higher to those with lower social status: 1) managers; 2) professionals; 3) technicians and associate professionals; 4) clerical support workers; 5) service and sales workers; 6) skilled agricultural, skilled agricultural, forestry and fishery workers; 7) craft and related trades workers; 8) plant and machine operators and assemblers; and 9) elementary occupations. The width of the connectors between occupational groups in Figure 4.10 represents the size of the group, transitioning from a given parental occupational group to own occupational group, while the colour gradient – from yellow (low performance) to orange to purple (high performance) – captures variations in numeracy proficiency, which is the skill with the highest labour market returns.
Figure 4.10 reveals marked generational shifts in the structure of the labour market, with the relative size of different occupational groups differing between the “parent generation” and the generation of 30-65 year-olds who participated in the 2023 Survey of Adult Skills. The occupation reported for the parent generation reflects the status of the occupation held by the parent with the highest social status, used as an indicator of the household’s overall social standing. This approach means that the parent generation’s occupational profile is biased towards higher-status jobs, since households with one higher-status and one lower-status parent are represented by the higher one. Seventeen per cent of households in the parent generation were headed by a parent who worked as a professional and up to 12% were headed by a skilled agricultural, forestry and fishery worker, whereas in the generation of 30-65 year-olds, 20% worked as professionals and only 1% worked as skilled agricultural, forestry, and fishery workers.
Figure 4.10 suggests a strong degree of intergenerational transmission of occupational status in OECD countries: as many as 52% of adults who had at least one parent working in a professional occupation when they were children went on to work as professionals or managers as adults. By contrast, only around 13% of adults whose parents worked in elementary occupations when they were children went on to work as professionals or managers in adulthood. Similarly, as many as 45% of professionals had parents who worked as either professionals or managers and only 16% had parents who worked as craft and related trades workers, plant and machine operators and assemblers, or in elementary occupations.
Numeracy proficiency is, on average, higher among individuals whose parents worked in high-status occupations such as managers, professionals, technicians and associate professionals and lower among individuals whose parents worked in low-status occupations, as clearly indicated in Figure 4.10. For example, the average numeracy proficiency of adults whose parents worked as professionals is 0.4 SD and the average numeracy proficiency of adults whose parents worked in elementary occupations is ‑0.5 SD. At the same time, on average across OECD countries, adults with higher numeracy skills are more likely than adults with lower numeracy skills to maintain the high-status occupation of their parents or to transition into higher status occupations than their parents. For example, whereas the average numeracy proficiency of adults whose parents worked as professionals and went on to become professionals themselves was 0.7 SD, the average numeracy proficiency of adults whose parents worked as professionals and went on to work as craft and related trades workers was 0.0 SD. Similarly, the average numeracy proficiency of adults whose parents worked in elementary occupations and went on to become professionals was 0.1 SD, whereas the average numeracy proficiency of adults whose parents worked in elementary occupations and went on to work as craft and related trades workers was -0.7 SD.
Percentage of respondents’ own occupation, parental occupation and numeracy proficiency, OECD average
Note: Adults aged 30-65. The width of the parental and own occupational groups (nodes) reflects respective sizes. The width of the connectors between the nodes represents the size of the occupational group flowing from parental occupation (left) to own occupation (right). Colour coding – yellow (low performance) to orange to purple (high performance) – is based on a standardised average numeracy score for each connector.
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
The analysis presented highlights significant occupational stability across generations in OECD countries, with higher-status occupations (managers, professionals, technicians and associate professionals) displaying strong intergenerational persistence. Individuals in these occupations often come from similar backgrounds and tend to have relatively high numeracy skills. While upward mobility from lower-status occupations to higher-status roles occurs, it is generally limited to those with strong numeracy skills. Conversely, middle- and lower-status occupations often exhibit occupational persistence, particularly among skilled agricultural, forestry and fishery workers. However, downward mobility is present across all occupational groups, with individuals moving from higher-status occupations to lower-status roles often showing relatively lower numeracy performance. These findings suggest that although skills are a key determinant of mobility, substantial occupational inheritance persists (OECD, 2025[19]).
Taken together, the evidence presented points to a labour market hierarchy that is both persistent and selective. At the top of the social hierarchy, more than half of managers and professionals have parents who worked in the same occupations, and those who maintain such high-status positions generally have above-average numeracy proficiency. At the bottom, adults whose parents worked in elementary and routine-manual occupations rarely work in professional and managerial occupations, and those who do typically have above-average numeracy skills. These results suggest that the chances of upward mobility hinge on individuals having opportunities to develop strong information-processing skills. However, as Chapter 3 of this report details, there are large disparities in such opportunities by socio-economic background.
Contrary to what can be observed among adult workers, a large share of young people with parents working in low-status occupations expect to go on to work as managers or professionals by age 30. For example, Figure 4.11 indicates that as many as 54% of young people with parents working in craft and related trades and 53% of young people with parents working in service and sales expect to work as managers or professions by age 30. Moreover, as many as 65% of young people with professional parents and 61% whose parents are managers expect to work as managers or professionals.
Young people who expect to be upwardly mobile generally have the highest levels of mathematics proficiency among young people with a similar parental background, as indicated in Figure 4.11. For example, the mean mathematics achievement of young people whose parents work as professionals and who expect to work as professionals by age 30 is 0.7 SD. By contrast, the mathematics achievement of young people whose parents work as professionals but who expect to work in service and sales is 0.0 SD. Similarly, the mean mathematics achievement of young people whose parents work in sales and service and who expect to work as professionals by age 30 is 0.1 SD, whereas the mathematics achievement of young people whose parents work in service and sales and who expect to work in service and sales themselves is -0.5 SD. Even among young people whose parents work in elementary occupations, those who expect to work as professionals have a higher relative achievement compared to their peers who expect to work in less prestigious occupations by age 30, such as service and sales. (-0.1 SD and -0.6 SD, respectively).
Crucially, the strong social gradient in mean levels of mathematics achievement means that the mathematics achievement of those who expect to work in high-status occupations closely maps parental circumstances. For example, among the 53% of young people who expect to work as professionals, mathematics achievement is highest among those whose parents are themselves professionals and lowest among those whose parents work in elementary occupations.
These results suggest a large gap between the expectations of many teenagers and the academic proficiency typically associated with the occupations they hope to enter. Many teenagers from lower-status households who have high levels of achievement compared to their similarly disadvantaged peers expect to work in professional or managerial roles. However, their average mathematics proficiency generally remains well below the proficiency of peers from higher-status households who hold similar expectations. Educational segregation and norm-referenced grading systems likely shape the perceptions of the opportunities young people have to thrive in roles that typically require strong information-processing skills. Nonetheless, without additional academic support and clearer information on pathways, these students may encounter difficulties meeting entry requirements for tertiary study or employment in occupations that require strong information-processing skills.
Percentage of 15-year-olds by parental occupation and expected occupation by age 30, and mathematics proficiency, OECD average
Note: The width of the parental and student occupational groups (nodes) reflects respective sizes. The width of the connectors between the nodes represents the size of the student group flowing from parental occupation (left) to expected student occupation (right). Colour coding – yellow (low performance) to orange to purple (high performance) – is based on average mathematics score for each connector.
Source: Calculations based on OECD (2022[46]), PISA 2022 database, www.oecd.org/en/data/datasets/pisa-2022-database.html.
Box 4.1 explores how policy making can promote occupational mobility through a skills-first approach, demonstrating examples across OECD countries of where initiatives have been implemented to drive the development of skills.
A skills-first approach to recruitment and human resource management can promote occupational mobility by reducing reliance on formal educational qualifications, which often reflect individuals’ socio-demographic background more than skills. By focusing on what people can do rather than how skills were acquired, these practices can improve transparency between jobseekers and employers, helping firms identify needed competencies and enabling workers to better showcase their skills. Examples include name-blind recruitment, skills assessments and ongoing professional development, all of which broaden access to jobs that have traditionally been limited by inherited networks or institutional prestige.
Such practices have implications for all stakeholders: employers may need to rethink recruitment and HR strategies, education providers may adapt curricula and certification, and individuals must find new ways to demonstrate their skills. Addressing hiring bias and improving communication about skills is especially relevant for groups whose qualifications do not fully signal their abilities. Skills-first approaches are gaining ground across OECD countries, although uptake is uneven. In some, such as the United States and Lithuania, employer demand for skills-based hiring aligns with strong jobseeker interest. Elsewhere, mismatches persist: in Luxembourg and Greece, employers increasingly post skills-based vacancies, but jobseeker interest is low, while in the Netherlands and Sweden, interest is high but few postings emphasise skills (Figure 4.12).
Governments play a crucial role in scaling up adoption. With public employment accounting for around 20% of jobs across OECD countries, governments can lead by example. For instance, several US states have already removed degree requirements for many public roles. They can also steer a skills-first culture through strategies such as the EU’s Union of Skills initiative, which invests in lifelong learning, reskilling and tools such as European Skills, Competences, Qualifications and Occupations (ESCO)1 to support job matching. They can also strengthen HRM capacity by equipping professionals with the skills to use new tools, including AI-enabled systems piloted in France and the United Kingdom to support recruitment and workforce planning. Finally, governments can provide support structures for individuals, such as the Flemish Region’s Learning and Career Account, which consolidates financial incentives and training opportunities into a single digital platform to encourage lifelong learning and smoother career transitions.
Note: The level of skills-first prevalence is on the horizontal axis. This metric refers to the extent to which individuals are inclined to apply – or consider applying – to job postings that specifically list required skills instead of those that do not mention particular skills. Categories are defined based on the standard deviation from the mean: a higher level of skills-first prevalence suggests that jobseekers respond more to postings that outline precise skills, reflecting a labour market where specific skills are prioritised over general qualifications or degrees. Conversely, a lower level indicates a continued reliance on traditional qualifications in the job application process. The percentage of online job postings with skills listed is on the vertical axis. This metric is used to rank countries in each level of skills-first prevalence.
Source: OECD (2025[33]), Empowering the Workforce in the Context of a Skills-First Approach, https://doi.org/10.1787/345b6528-en.
1. The European Skills, Competences, Qualifications and Occupations (ESCO) skills taxonomy acts as a “dictionary” for different stakeholders on education and training topics; describing, identifying and classifying professional occupations and skills relevant for the EU labour market and for stakeholders in the fields of education and training.
Source: OECD (2025[33]), Empowering the Workforce in the Context of a Skills-First Approach, https://doi.org/10.1787/345b6528-en.
4.6.2. How intergenerational occupational and educational (im)mobility compare
In recent decades, levels of educational attainment have generally increased markedly, resulting in a large number of adults having completed a higher educational qualification compared to their parents. On average across OECD countries, half of 30-65 year-olds have attained the same level of education as their parents, 37% have attained a higher level and 12% a lower level (Figure 4.13). Korea displays the highest level of upward educational mobility (i.e. the largest share of individuals with a higher educational level than their parents) and Czechia the lowest (64% and 19%, respectively). Conversely, Sweden exhibits the highest rate of downward intergenerational educational mobility, measured as the proportion of individuals with a lower qualification than their parents, while Singapore records the lowest (22% and 2%, respectively).
Share of adults with lower, the same and higher levels of educational attainment than their parent(s)
Note: Adults aged 30-65. See the note for Table 4.1 for the definitions of groups based on parental occupation and respondents’ educational attainment.
Countries are ranked in descending order based on the percentage of individuals with lower educational level than their parent(s).
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html.
Young people’s expectations of educational mobility are more conservative than actual intergenerational mobility in educational attainment observed among 30-65 year-olds, as findings from the 2022 Programme for International Student Assessment (PISA) show (Figure 4.14). On average across OECD countries, 17% of 15-year-old students expect to attain a higher level of education than their parents, almost 20% expect to attain a lower level and 63% expect to attain the same level. Portugal has the highest expected upward educational mobility (i.e. the largest share of students with higher expected educational level than their parents), while Denmark has the lowest (33% and 6%, respectively). Chile has the lowest expected downward educational mobility, and Denmark has the highest (5% and 35%, respectively).
Share of students whose educational expectations are lower, the same or higher than the educational level of their parent(s)
Note: Students aged 15 years old. In 2022, students participating in the Programme for International Student Assessment (PISA) were asked to report their parents’ educational attainment as well as the highest educational qualification they expected to obtain.
Countries are ranked in descending order based on the percentage of students with higher expected educational level than parent(s).
Source: Calculations based on OECD (2022[46]), PISA 2022 Database, www.oecd.org/en/data/datasets/pisa-2022-database.html.
On average across participating OECD countries, 36% of adults aged 30-65 work in a lower occupational group than their parents, 22% work in the same group and 42% in a higher occupational group (Figure 4.15, Panel A). The figure highlights substantial differences across countries. For example, in Singapore, 56% of adults work in a higher occupational group than their parents, compared to only 34% in Czechia. Figure 4.15 (Panel B) also shows that, on average across participating OECD countries, 38% of 15-year-olds expect to work in a higher occupational group than their parents, 30% expect to work in the same group and 32% in a lower group. In line with the actual occupational comparisons of adults and their parents (Figure 4.15 Panel B), students in Chile report the highest expectations of upward occupational mobility, (62%), while students in Denmark report the lowest (22%).
Panel A: Share of adults with lower, same and higher ISCO group than their parents. Panel B: Share of students who report expecting to work in a lower, same and higher ISCO group than their parents
Note: Panel A: adults aged 30-65. Shows the percentage of adults with lower, same and higher ISCO group (1-digit level) than their parents. Panel B: shows the percentage of 15-year-old students who expect to work by age 30 in a lower, same and higher ISCO group (1-digit level) than their parents.
Countries are ranked in descending order of the share of adults (students) with downward occupational mobility (expectations).
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html and OECD (2022[46]), PISA 2022 database, www.oecd.org/en/data/datasets/pisa-2022-database.html.
For each country, Figure 4.16 presents two indicators. The first shows the share of adults who have attained the same or a higher level of education compared to their parents but who are employed in an occupation with a lower social status than that of their parents. The second shows the share of 15-year-old students who expect to achieve the same or a higher level of education compared to their parents but, by the age of 30, anticipate working in an occupation with a lower social status than their parents. On average, 28% of adults have experienced downward occupational mobility despite experiencing stable or upward educational mobility, while 24% of young people expect to experience such downward mobility. Among adults, the divergence between educational and occupational outcomes is particularly pronounced in Czechia, where as many as 37% have experienced downward occupational mobility despite stable or upward educational mobility. By contrast, Singapore has the lowest share at 22%. However, young people in Singapore are especially pessimistic about their future, with as many as 36% of 15-year-old students expecting to experience downward occupational mobility despite stable or upward educational mobility. In Chile, only 14% hold this expectation.
Note: Percentage of adults aged 30-65 with same or higher educational attainment but lower occupational status (ISCO group [1-digit level]) than their parents. Percentage of 15-year old students who expect same or higher educational attainment but lower ISCO group (1-digit level) than their parents. In 2022, students participating in PISA were asked to report their parents’ educational attainment, occupation as well as the highest educational qualification and occupation they expected to obtain.
Countries are ranked in descending order based on the percentage of adults with same or higher educational attainment but lower occupational status than their parents.
Source: Calculations based on OECD (2024[22]), PIAAC data and methodology, www.oecd.org/en/about/programmes/piaac/piaac-data.html and OECD (2022[46]), PISA 2022 database, www.oecd.org/en/data/datasets/pisa-2022-database.html.
As educational attainment has risen over recent decades, most adults today hold qualifications that are at least as high as, and often higher than, those of their parents. However, for many, these gains in education have not translated into corresponding upward mobility in terms of occupational status. In many economies, the supply of well-qualified individuals has outpaced the creation of higher-status jobs capable of absorbing them. The result is a widening gap between educational attainment and occupational outcomes. This dynamic may help to explain growing feelings of disillusionment and frustration in many OECD countries: although families have invested heavily in education, improved qualifications do not reliably open the door to managerial or professional occupations that confer higher social status.
Looking ahead, the rapid diffusion of AI may further reshape the relationship between education and occupational mobility. On the one hand, AI could deepen existing mismatches by automating routine cognitive and technical tasks, reducing the number of mid-level professional roles traditionally accessible to well-qualified workers. On the other hand, it may create new opportunities for workers with advanced digital, analytical and creative skills, particularly in sectors that complement rather than compete with AI. The overall impact will depend, at least in part, on how effectively education and training systems adapt to evolving skill demands and how labour market institutions support workers through these transitions.
References
[8] Almlund, M. et al. (2011), “Personality psychology and economics”, Handbook of the Economics of Education, Vol. 4, pp. 1-181, https://doi.org/10.1016/b978-0-444-53444-6.00001-8.
[31] Anni, K., U. Vainik and R. Mõttus (2024), “Personality profiles of 263 occupations”, Journal of Applied Psychology, Vol. 110/4, pp. 481-511, https://doi.org/10.1037/apl0001249.
[24] Araki, S. (2020), “Educational expansion, skills diffusion, and the economic value of credentials and skills”, American Sociological Review, Vol. 85/1, pp. 128-175, https://doi.org/10.1177/0003122419897873.
[7] Autor, D. (2015), “Why are there still so many jobs? The history and future of workplace automation”, Journal of Economic Perspectives, Vol. 29/3, pp. 3-30, https://doi.org/10.1257/jep.29.3.3.
[29] Autor, D. and M. Handel (2013), “Putting tasks to the test: Human capital, job tasks, and wages”, Journal of Labor Economics, Vol. 31/S1, pp. S59-S96, https://doi.org/10.1086/669332.
[4] Blanden, J., P. Gregg and S. Machin (2005), Intergenerational Mobility in Europe and North America, Centre for Economic Performance, http://www.suttontrust.com/wp-content/uploads/2020/01/IntergenerationalMobility.pdf.
[34] Booth, T. et al. (2015), “Distinguishing CEOs from top level management: A profile analysis of individual differences, career paths and demographics”, Journal of Business and Psychology, Vol. 31/2, pp. 205-216, https://doi.org/10.1007/s10869-015-9416-7.
[16] Causa (Forthcoming), Does the apple fall far from the tree? New evidence on intergenerational social mobility across OECD countries.
[25] Domina, T. (2005), “Leveling the home advantage: Assessing the effectiveness of parental involvement in elementary school”, Sociology of Education, Vol. 78/3, pp. 233-249, https://doi.org/10.1177/003804070507800303.
[11] Duckworth, A. and M. Seligman (2005), “Self-discipline outdoes IQ in predicting academic performance of adolescents”, Psychological Science, Vol. 16/12, pp. 939-944, https://doi.org/10.1111/j.1467-9280.2005.01641.x.
[14] Dustmann, C. and A. Glitz (2011), “Migration and education”, Handbook of the Economics of Education, Vol. 4, pp. 327-439, https://doi.org/10.1016/b978-0-444-53444-6.00004-3.
[26] Faragher, E., M. Cass and C. Cooper (2005), “The relationship between job satisfaction and health: A meta-analysis”, Occupational and Environmental Medicine, Vol. 62/2, pp. 105-112, https://doi.org/10.1136/oem.2002.006734.
[41] Frey, V., L. Alajääskö and J. Thomas (2024), “Gendered perceptions of social protection across OECD countries”, OECD Social, Employment and Migration Working Papers, No. 311, OECD Publishing, Paris, https://doi.org/10.1787/f3e002c2-en.
[35] Furnham, A. (2017), “Personality differences in managers who have, and have not, worked abroad”, European Management Journal, Vol. 35/1, pp. 39-45, https://doi.org/10.1016/j.emj.2016.10.002.
[5] Hanushek, E. et al. (2015), “Returns to skills around the world: Evidence from PIAAC”, European Economic Review, Vol. 73, pp. 103-130, https://doi.org/10.1016/j.euroecorev.2014.10.006.
[1] Heckman, J. and T. Kautz (2012), “Hard evidence on soft skills”, Labour Economics, Vol. 19/4, pp. 451-464, https://doi.org/10.1016/j.labeco.2012.05.014.
[13] Heckman, J. and S. Mosso (2014), “The economics of human development and social mobility”, Annual Review of Economics, Vol. 6/1, pp. 689-733, https://doi.org/10.1146/annurev-economics-080213-040753.
[9] Heckman, J., J. Stixrud and S. Urzua (2006), “The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior”, Journal of Labor Economics, Vol. 24/3, pp. 411-482, https://doi.org/10.1086/504455.
[28] Helliwell, J. and H. Huang (2010), “How’s the job? Well-being and social capital in the workplace”, ILR Review, Vol. 63/2, pp. 205-227, https://doi.org/10.1177/001979391006300202.
[27] Judge, T. et al. (2001), “The job satisfaction–job performance relationship: A qualitative and quantitative review”, Psychological Bulletin, Vol. 127/3, pp. 376-407, https://doi.org/10.1037/0033-2909.127.3.376.
[36] Lan, J., C. Wong and G. Zeng (2021), “Personality profiles for hospitality employees: Impact on job performance and satisfaction”, International Journal of Hospitality Management, Vol. 98, p. 103018, https://doi.org/10.1016/j.ijhm.2021.103018.
[39] Lightcast (2025), The Speed of Skill Change, Lightcast, https://lightcast.io/resources/research/speed-of-skill-change (accessed on 18 March 2025).
[12] Mischel, W., Y. Shoda and M. Rodriguez (1989), “Delay of gratification in children”, Science, Vol. 244/4907, pp. 933-938, https://doi.org/10.1126/science.2658056.
[15] Oded, G. (2011), “Inequality, human capital formation, and the process of development”, Handbook of the Economics of Education, Vol. 4, pp. 441-493, https://doi.org/10.1016/b978-0-444-53444-6.00005-5.
[33] OECD (2025), Empowering the Workforce in the Context of a Skills-First Approach, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/345b6528-en.
[19] OECD (2025), Gender Equality in a Changing World: Taking Stock and Moving Forward, Gender Equality at Work, OECD Publishing, Paris, https://doi.org/10.1787/e808086f-en.
[18] OECD (2025), The State of Global Teenage Career Preparation, OECD Publishing, Paris, https://doi.org/10.1787/d5f8e3f2-en.
[17] OECD (2025), To Have and Have Not – How to Bridge the Gap in Opportunities, OECD Publishing, Paris, https://doi.org/10.1787/dec143ad-en.
[3] OECD (2024), Do Adults Have the Skills They Need to Thrive in a Changing World?: Survey of Adult Skills 2023, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/b263dc5d-en.
[21] OECD (2024), Job Creation and Local Economic Development 2024: The Geography of Generative AI, OECD Publishing, Paris, https://doi.org/10.1787/83325127-en.
[42] OECD (2024), OECD Employment Outlook 2024: The Net-Zero Transition and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/ac8b3538-en.
[22] OECD (2024), PIAAC data and methodology, OECD, Paris, https://www.oecd.org/en/about/programmes/piaac/piaac-data.html (accessed on 10 December 2024).
[23] OECD (2024), Survey of Adult Skills (PIAAC) 2nd cycle database, OECD, Paris, https://www.oecd.org/en/data/datasets/piaac-2nd-cycle-database.html (accessed on 10 December 2024).
[44] OECD (2023), Beyond Applause? Improving Working Conditions in Long-Term Care, OECD Publishing, Paris, https://doi.org/10.1787/27d33ab3-en.
[20] OECD (2023), Job Creation and Local Economic Development 2023: Bridging the Great Green Divide, OECD Publishing, Paris, https://doi.org/10.1787/21db61c1-en.
[46] OECD (2022), PISA 2022 Database, OECD, Paris, https://www.oecd.org/en/data/datasets/pisa-2022-database.html (accessed on 5 November 2024).
[43] OECD (2021), “The inequalities-environment nexus: Towards a people-centred green transition”, OECD Green Growth Papers, No. 2021/01, OECD Publishing, Paris, https://doi.org/10.1787/ca9d8479-en.
[2] OECD (2019), Skills Matter: Additional Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/1f029d8f-en.
[45] OECD (2018), A Broken Social Elevator? How to Promote Social Mobility, OECD Publishing, Paris, https://doi.org/10.1787/9789264301085-en.
[6] OECD (2016), Skills Matter: Further Results from the Survey of Adult Skills, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/9789264258051-en.
[30] Oreopoulos, P. and K. Salvanes (2011), “Priceless: The nonpecuniary benefits of schooling”, Journal of Economic Perspectives, Vol. 25/1, pp. 159-184, https://doi.org/10.1257/jep.25.1.159.
[40] Schmidpeter, B. and R. Winter-Ebmer (2021), “Automation, unemployment, and the role of labor market training”, European Economic Review, Vol. 137, p. 103808, https://doi.org/10.1016/j.euroecorev.2021.103808.
[10] Soto, C. (2019), “How replicable are links between personality traits and consequential life outcomes? The life outcomes of personality replication project”, Psychological Science, Vol. 30/5, pp. 711-727, https://doi.org/10.1177/0956797619831612.
[37] Törnroos, M., M. Jokela and C. Hakulinen (2019), “The relationship between personality and job satisfaction across occupations”, Personality and Individual Differences, Vol. 145, pp. 82-88, https://doi.org/10.1016/j.paid.2019.03.027.
[38] U.S. Bureau of Labor Statistics (2025), Table 1.2 Occupational projections, 2023—2033, and worker characteristics, 2023 (Numbers in thousands), https://www.bls.gov/emp/tables/occupational-projections-and-characteristics.htm (accessed on 30 May 2025).
[32] Wolfram, T. (2023), “(Not just) Intelligence stratifies the occupational hierarchy: Ranking 360 professions by IQ and non-cognitive traits”, Intelligence, Vol. 98, p. 101755, https://doi.org/10.1016/j.intell.2023.101755.
← 1. The 2023 Survey of Adult Skills International Report (OECD, 2024[3]) noted that an increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. This estimate accounts for years of education, age, gender, immigrant background, parental education, whether one lives with a partner or has children, and work experience. This estimate is also shown in Table 4.2, Model (2) in this chapter – as it is most similar in terms of control variables in the regression model.
← 2. Whereas in other parts of the report the evidence presented reflects the average of country specific associations, in these sets of analyses the reference unit is the overall population of workers and the extent to which they are clustered in different occupations. To correctly account for differences in the probability of being sampled in different countries and population groups, weights are used when estimating a pooled model.