The 2023 Survey of Adult Skills enables, for the first time, an analysis of how skills-use has evolved over time. This chapter examines how the use of information-processing and generic skills in the workplace has changed over the past decade, and how these patterns differ between countries and occupations. The analysis pays particular attention to two groups traditionally lagging behind in skills use – low-qualified workers and women – assessing whether they have caught up over the past decade. Understanding these dynamics is essential not only for advancing equity in the labour market but also for ensuring that available talent is fully utilised.
How Workers Use, or Don’t Use, their Skills in the Workplace
Abstract
The evolution of skills use at work is highly heterogenous
Over the past decade, the use of information-processing skills has increased markedly and consistently across countries, while the use of generic skills has remained broadly stable. The use of reading, writing, numeracy, ICT and problem-solving skills has expanded significantly, with particularly strong gains in writing and ICT use. By contrast, the use of task discretion, learning at work, influencing and co‑operative skills has changed little over time on average, with significant differences in the change across countries.
ICT use has risen especially among workers in low- and medium-skilled occupations, indicating that digital demands are now pervasive across the labour market. In contrast, the use of physical and dexterity skills has declined, pointing to a continued shift towards the automation of physical tasks.
The use of skills at work has increased substantially among low-qualified adults, narrowing the gap with higher-qualified workers in most domains, particularly in numeracy, ICT and influencing skills. This convergence is driven mainly by within-occupation increases in the use of these skills rather than changes in the distribution of low-qualified by occupation, suggesting that job upgrading and greater skills use are occurring across the qualification spectrum.
There is limited evidence of overall convergence between men and women in the use of skills at work over the past decade. Some progress is observed in the use of co‑operative skills, yet declines in domains such as numeracy and ICT point to persistent gender disparities. Gender equality in skills use remains closely linked to national labour market structures.
Understanding how the use of skills in the workplace evolves over time is essential for assessing how labour markets adapt to technological progress, organisational change and shifting patterns of labour demand. The 2023 Survey of Adult Skills provides, for the first time, comparable measures of skills use collected a decade apart, offering a unique opportunity to analyse how workers’ tasks and skill requirements have evolved across countries and occupations. Studying the evolution of skills use is particularly relevant in the context of rapid digitalisation, automation and the diffusion of new forms of work organisation. These forces have transformed not only the composition of employment but also the nature of tasks within occupations, potentially increasing the demand for analytical, digital and collaborative skills while reducing the reliance on routine and physical activities. From a policy perspective, trends in skills use can help identify which groups and sectors are advancing – and which are lagging – to inform strategies to promote inclusive digital transformation, support workplace learning, and strengthen the alignment between skills supply and demand.
The comparison of skills-use measures between the two cycles of the Survey of Adult Skills is made possible by careful survey design, which ensures that results from the second cycle are comparable with those from the first (details on the few methodological differences between the two cycles of the Survey of Adult Skills can be found in Box 3.2 of OECD (2024[1])). In particular, the variables underpinning the indicators of skills use employed in this report show consistency between cycles. Although a limited number of items differ between the questionnaires, considerable effort has been made to maintain continuity in the underlying latent concept of skills use that these variables are intended to capture (OECD, 2021[2]; OECD, 2025[3]). The exact variables used in each cycle to construct the skills use measures of this report are provided in the Annex.1
A total of 27 countries and economies participated in both survey cycles. Of the 31 countries and economies included in Cycle 2 and discussed in earlier chapters, only Croatia, Latvia, Portugal and Switzerland did not take part in Cycle 1 and are therefore excluded from the present analysis. For ease of presentation, this report often refers to changes occurring “over the past decade”. It should, however, be noted that the first cycle was implemented in three rounds: Round 1 in 2011/12, Round 2 in 2014/15 and Round 3 in 2017. As a result, the time elapsed between the two observations varies across countries. Although the United States was the only country to participate in two rounds of Cycle 1, the analysis of this chapter relies solely on data from Round 1. This choice aligns the US data with the reference period for the majority of the other countries, as 21 countries of the 27 that participated to both cycles also participated in the 2011/12 round.
Figure 3.1 and Figure 3.2 present a comparative overview of changes in the use of information-processing and generic skills at work between the two cycles of the Survey of Adult Skills. The evidence points to a larger and more consistent increase in the use of information-processing skills than in generic skills. The use of generic skills such as task discretion, learning at work, influencing, co‑operative skills and physical abilities has on average remained relatively stable over the past decade, with changes typically ranging between ‑0.5 and +0.5 on the five‑point frequency scale.
The evolution of the use of these skills is, however, highly heterogeneous across countries and in several cases the direction of change may be a cause for concern. For example, opportunities for learning at work expanded in Ireland and the Netherlands, yet declined in Italy, the Slovak Republic and Poland. Declines in this area may indicate a weakening of workplace learning systems that are essential for ensuring that workers remain adaptable, particularly in environments undergoing technological or organisational change. Similarly, France and Norway experienced an increase in the use of co‑operation skills in the workplace, whereas several Central and Eastern European countries reported reductions. Lower use of co‑operative skills may suggest work environments where teamwork and participatory approaches are becoming less central, with potential implications for job quality, but also innovation and firms’ performance.
By contrast, the use of information-processing skills shows a much clearer upward trajectory across nearly all countries. The use of reading, writing, numeracy, ICT and problem-solving skills has seen widespread gains, although the magnitude differs. Writing and ICT skills use exhibit significant increases, with the latter standing out as one of the most dynamic areas of change. The increase in ICT use is particularly pronounced in Spain, Chile and Ireland, pointing to a strong intensification of digital tool use in the workplace (see Box 3.1 for more analysis on the evolution of the use of ICT skills in the last decade). The use of reading and numeracy skills also increased, albeit more moderately, with countries such as Lithuania, Italy, France and Ireland reporting above‑average increases. The use of problem solving in the workplace presents a more nuanced picture. While some countries such as Ireland, New Zealand and Estonia record large increases, others including Italy and Poland display small declines. This suggests that although workplaces are becoming more digital and data‑intensive, the extent to which they require employees to engage in complex, non-routine problem solving is uneven.
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
Studying the correlation between changes in ICT skills use at work and internet use between 2012 and 2023 is particularly interesting because it captures a period of rapid digital transformation, marked by the diffusion of smartphones, the rise of platform economies, and the shift to remote work and online services accelerated by the COVID‑19 pandemic. ICT skills are not only a prerequisite for meaningful internet use but also a determinant of how individuals benefit from digital technologies, whether in accessing information, improving employability, or engaging socially and politically. At the same time, greater internet use can reinforce skill acquisition through practice, exposure, and online learning opportunities. Analysing how these two dimensions evolved together over the past decade provides insights into digital inclusion, inequality, and the extent to which technological change has translated into broader participation in the digital economy.
Figure 3.3 illustrates a positive correlation between changes in the use of ICT skills at work between the two cycles of the Survey of Adult Skills and changes in the share of individuals using the internet over the same period. Countries that experienced larger increases in internet use, such as Spain, Italy, and Poland, also tended to record more pronounced growth in the use of ICT skills at work, reflecting the spreading of technology at work and in society. Conversely, countries like Sweden, Norway, Finland, and the United States, where internet use was already high in 2012, show only marginal increases in both internet use and the use of ICT skills at work, consistent with a saturation effect. These patterns highlight convergence, as late adopters catch up.
Note: The size of the bubble represents the share of individuals using the internet in 2012. For consistency, the country name is reported for Belgium and the United Kingdom, even if the Survey of Adult Skills is conducted only at subnational level – namely in the Flemish region and in England, respectively.
Source: 2023, 2018, 2025, 2012 Survey of Adult Skills; OECD.Stat – ICT Access and Usage by Individuals (2012 and 2023).
To better understand the drivers of the change in skills use at work over the past decade, a Blinder-Oaxaca decomposition is applied (Blinder, 1973[4]; Oaxaca, 1973[5]; Elder, Goddeeris and Haider, 2010[6]). This technique separates observed changes into two components. The “between-occupation” component captures the share of the change attributable to shifts in the composition of the workforce and job characteristics, such as education, occupation, industry, firm size, or contract type. The “within-occupation” component reflects differences in the way these same characteristics are associated with skills use between cycles. In this context, it measures how the relationship between individual and job attributes and the use of specific skills has evolved over time, once compositional changes are held constant. The technique is particularly useful for distinguishing between changes arising from who is employed in the economy and in which jobs and the changes linked to how work is organised and tasks are performed.
The results in Figure 3.4 indicate that the majority of the change in skills use between cycles of the Survey of Adult Skills is due to changes in skills use within occupations. This pattern suggests that changes within jobs – the way jobs are organised or the adoption of new technologies – have influenced changes in skills use more than structural changes in workforce composition. For instance, the large “within-occupation” increase in ICT use points to deeper technological integration within occupations that previously made limited use of digital tools. Similarly, the rise in writing, numeracy and problem-solving skills is only partially accounted for by observable factors such as higher education levels, implying that the content of jobs themselves has changed to require more of these competences. Similarly, the use of dexterity skills at work shows a pronounced decline over the past decade dominated by the unexplained component.
The limited contribution of the “between-occupation” component in most skill domains suggests that shifts in education, age, or occupational composition have played a secondary role in explaining the evolution of skills use. Instead, changes in technology, management practices, and work organisation have altered the underlying production of tasks. This evidence points to a labour market in which the way work is performed has evolved faster than the characteristics of the workforce itself, reinforcing the need for adult learning policies aimed at helping workers adapt to changes in the skill requirements of their own jobs more than to transition to new jobs.
Note: The Figure presents results of a Blinder-Oaxaca decomposition of the change in skills use between PIAAC cycles. Each bar represents a separate regression. Controls include gender, age, age squared, educational attainment, literacy proficiency, occupation, industry, firm size, permanent contract, public employment, and country fixed effects. All coefficients are statistically significant (at the 1% significance level), except those of co‑operative skills, which are not significant.
Source: 2023, 2018, 2025, 2012 Survey of Adult Skills.
Tracking how the use of skills at work evolves by occupation offers additional insights into shifting job requirements and the reorganisation of tasks within the labour market. Figure 3.5 shows that, for all occupation groups, the largest increases in skills use are observed in ICT, writing, and numeracy. While these trends highlight the pervasive integration of digital tools and communication tasks, the increase is particularly marked among workers in low- and medium-skilled occupations, suggesting that technological diffusion is no longer limited to high-skilled roles but has become a defining feature of work more broadly. Reading, influencing, and task discretion have also grown in all occupations, though to a lesser extent. By contrast, the use of physical skills and especially dexterity has declined substantially, most notably among high-skilled occupations, pointing to a continued shift away from manual requirements towards knowledge‑intensive and cognitively demanding tasks.
Note: Skill level is derived from ISCO 1‑digit occupation codes: high skilled (1‑3), moderately skilled (4‑8), low skilled (9). Armed forces (0) are excluded. Skills are sorted by largest change for moderate‑skilled occupations.
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
Further examination of changes by occupation reveals additional heterogeneity (Figure 3.6). For example, although ICT use increased among all workers between 2012 and 2023, the increase has not been uniform. Skilled agricultural workers, such as livestock producers and forestry workers, recorded an increase twice as large as that of managers (1.5 compared with 0.7 points on the five‑point scale used to measure skills use). Differences also emerge within the same skill group. Among medium-skilled occupations, clerks, including receptionists and bank tellers, experienced only a negligible increase in numeracy use (0.1 points), whereas craft and trades workers, such as plumbers and carpenters, saw a much larger rise of 0.7 points. Even for skills whose use has declined, the extent of the reduction differs between occupations. Craft and trades workers experienced almost no decline in the use of dexterity skills, while clerks registered a fall by 0.8 points.
These findings underline the uneven pace of change in skills demand among occupations. They point to the need for targeted training and upskilling policies that account for such variation. For workers in lower- and medium-skilled occupations, the strong rise in ICT and numeracy requirements highlights the importance of widening access to digital and foundational skills training. At the same time, the sharp reduction in dexterity-based work suggests that policies should anticipate structural shifts in manual occupations and prepare workers for increasingly knowledge‑oriented tasks in their own jobs.
Source: 2024, 2018, 2015, 2023 Survey of Adult Skills.
Figure 3.7 summarises the evolution of skills use at work for low- and high-qualified adults between the first and second cycles of the Survey of Adult Skills. The estimates derive from OLS regressions controlling for gender, age, immigration background, literacy proficiency, occupation, industry, firm size, employment status, contract type, public-sector employment and country fixed effects. In nearly all skill domains, the results point to an increase in the frequency with which adults report using their skills at work that is stronger among low-qualified workers. In Cycle 1, these workers reported substantially lower frequency of skills use compared with high-qualified adults, but the gap has narrowed in almost all areas by Cycle 2. The increase is particularly pronounced for the use of numeracy, ICT, and influencing skills, which are core to work in more complex work environments.
These patterns indicate a process of convergence in skills use between qualification groups and are consistent with earlier evidence in this chapter showing that workers in low- and medium-skilled occupations are increasingly engaging in tasks requiring higher levels of ICT, numeracy, reading, influencing and problem-solving skills. The results therefore confirm that the process of job upgrading is not confined to high-skilled occupations but extends across the qualification spectrum, reflecting an overall shift in the nature of work towards greater use of a wide range of skills.
Predicted skills use at work by PIAAC cycle
Note: The Figure represents the marginal effects of OLS regressions controlling for gender, age, age squared, immigration background, literacy proficiency, occupation, industry, firm size, full-time work, permanent contract, public employment, and country fixed effects. Low-qualified workers are defined as those with up to lower secondary education, while the rest are defined as high-qualified workers. All coefficients are statistically significant (at the 1% significance level).
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
To understand the mechanisms driving the fact that low-qualified workers have experienced a stronger rise in the use of skills at work than their high-qualified counterparts, it is important to distinguish between two possible explanations. The first relates to compositional shifts: low-qualified workers may now be employed in different types of occupations that are inherently more skill-intensive. The second involves within-occupation adjustments, in which the content of existing jobs has evolved and become more demanding in terms of skill requirements. Formally, changes in the average skills use of low-qualified workers can be decomposed into these two components:
Equation 3.1.
where is the average skills use in occupation j in cycle c, is the employment share of low-qualified workers in occupation j in cycle c. The first term in the decomposition captures within-occupation changes, reflecting the evolution of skill requirements within occupations over time, whereas the second term captures between-occupation reallocation, indicating whether workers have moved into occupations that use skills more intensively.2 This analytical framework follows well-established approaches in the labour economics literature, including the shift-share decomposition methods proposed by Freeman, Ganguli and Handel (2020[7]) and by Caunedo, Keller and Shin (2023[8]), which have been used to examine changes in task content and job structure over time. Both studies highlight that much of the evolution in the nature of work stems from transformations within occupations rather than large‑scale shifts between them.
Figure 3.8 summarises the results of this decomposition for low-qualified workers. For each domain of skills use, the total change in predicted skills utilisation between cycles of the Survey of Adult Skills is separated into within-occupation and between-occupation components. The figure shows that – in line with the recent economic research cited above – the bulk of the increase in skills use among low-qualified workers arises from within-occupation changes. In other words, most of the observed catching-up effect is explained by jobs themselves becoming more skill-intensive, rather than by low-qualified workers moving into new, more demanding occupations. The within-occupation component is positive and substantial in almost all skill domains, particularly for the use of ICT skills, learning at work, co‑operative skills, and self-organisation, which are critical to adapt to technological and organisational changes in the workplace (Box 3.2 provides additional insights on which specific ICT skills have increased in use at work for the low qualified). By contrast, the contribution of between-occupation reallocation is relatively modest and, for some skills, even negative.
Note: The Figure restricts the analysis to low-qualified workers, defined as those with up to lower secondary education.
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
The case of the use of ICT skills is particularly interesting, as, taken together, the evidence presented in Figure 3.7 and Figure 3.8 indicates that: 1) the skills displaying the greatest increase in workplace use between cycles of the Survey of Adult Skills for low-qualified workers are those related to ICT, and 2) almost all of this growth stems from the fact that existing jobs held by low-qualified workers have become more demanding in terms of the use of ICT skills.
To illustrate this point more precisely, the composite index of ICT skills use can be disaggregated to examine the evolution of its underlying components over the past decade for a subset of detailed occupations. Figure 3.9 presents data on the frequency of use of selected ICT skills in (2‑digit ISCO) occupations where, in 2012, more than one‑third of workers were low-qualified. While in Cycle 1 of PIAAC workers in these occupations reported never using digital tools for communication, by 2023 nearly all of them used applications such as email or internet-based calls at least once per week. The increase is especially marked among street and related sales and service workers, whose use of digital communication tools shifted from “Never” to almost “Every day”. A similar pattern emerges for the use of the internet to access information, which has become substantially more frequent across all low-qualified occupational groups. Remarkably, even the use of standard software, such as spreadsheets and word processors, has risen considerably. For example, agricultural, forestry, and fishery labourers – who in 2012 reported almost no engagement with such tools – now use electronic documents, spreadsheets, or presentation software at least once per month.
Note: Caution is required to interpret these trends, as variables are differently phrased in the Cycle 1 and Cycle 2 PIAAC questionnaires – see the Annex for the exact variables used in each cycle.
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
Chapter 1 of this report found substantial disparities between men and women in the frequency of skills use at work (see, for instance, Figure 1.7). It is therefore interesting to assess whether there is evidence of a catching-up between the two cycles of the Survey of Adult Skills. Figure 3.10 summarises OLS coefficients from regressions of skills use on a gender dummy interacted with a dummy indicating the reference cycle. The blue bars show the baseline gender gap in the earlier cycle and the grey bars report the change between cycles. Two results stand out. First, for most information-processing skills including reading, numeracy, writing, and ICT the change coefficients are negative and statistically significant. The use of numeracy and ICT skills exhibits the largest deterioration in relative terms with the change coefficient implying a widening of the gap rather than convergence. The use of reading and writing also display further negative movement. By contrast, most of the other skills use present no statistically significant trend across cycles. This indicates not only no systematic catching up of women, but actually a substantial deterioration of the gender gap in skills use at work over time. Only two domains present a significant positive change over time. Notably, use of co‑operative skills shows a positive and relatively large change coefficient indicating that the within-occupation gender gap in reported use of co‑operative skills has declined between cycles. Physical skills also register a positive change coefficient, although the baseline disadvantage for women is large and not fully offset by the change.
These findings suggest that reducing the gender gap in skills use remains a priority requiring workplace‑level interventions that address how tasks are allocated and that increase the degree of autonomy and discretion available to women. Complementary measures should strengthen the use of numeracy and ICT skills in everyday job contexts, as well as promote management practices that systematically review within-firm task assignment.
OLS coefficients
Note: The Figure represents the coefficient of OLS regressions of skills use on gender interacted with a PIAAC cycle dummy. Coefficients are adjusted for age, age squared, immigration background, educational attainment, literacy proficiency, occupation, industry, firm size, full-time work, permanent contract, public employment, and country fixed effects. Each bar represents a separate regression. Shaded bars represent statistically not significant results, while all the other coefficients are statistically significant (at least at the 10% significance level).
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
Cross-country comparisons provide a better understanding of how the gender gap in skills use evolves over time. Figure 3.11 focusses on two domains where the gender gap in skills use was found to be most pronounced and widening between cycles of the Survey of Adult Skills, namely numeracy and ICT skills. Overall, the chart shows that the size of the gap differs considerably between countries, suggesting that national labour market structures, occupational segregation patterns and policy contexts may play a significant role.
Panel A shows that the gender gap in the use of numeracy skills remains sizeable in many countries. In the first cycle of the Survey of Adult Skills, women used numeracy-related skills less frequently than men in almost all countries, with particularly large negative coefficients in Japan and in some of the Nordic countries, like Finland, Norway and Denmark. Over the following decade, the situation generally worsened. The grey bars that represent changes between the two cycles are negative in a majority of cases, suggesting that women’s relative use of numeracy skills declined further. However, there are exceptions. In Japan and Finland, where the initial gap was among the largest, the coefficients for change are close to zero, indicating that the gap did not worsen. This relative stability contrasts with the general pattern of widening gaps observed elsewhere.
In the use of ICT skills (Panel B), many countries displayed only a modest or statistically insignificant gender gap in 2012. The pattern over time suggests that the gap has emerged or widened in the most recent cycle. The change coefficients are negative for most countries, indicating that men have increased their use of ICT skills at work relative to women. The emergence of this new gender gap may reflect structural changes in digitalisation and occupational upgrading, where male‑dominated roles (even with broader sectors and occupations) have been more successful in integrating ICT-intensive work tasks. Nonetheless, a few countries stand out for showing little deterioration or some improvement, notably Japan and Norway, which had large gender differences in ICT use in 2012 but have not experienced further divergence.
OLS coefficients
Note: The Figure represents the coefficient of OLS regressions of skills use on gender interacted with a PIAAC cycle dummy. Coefficients are adjusted for age, age squared, immigration background, educational attainment, literacy proficiency, occupation, industry, firm size, full-time work, permanent contract, public employment, and country fixed effects. Each bar represents a separate regression. Shaded bars represent statistically not significant results, while all the other coefficients are statistically significant (at least at the 10% significance level). Countries are sorted in descending order according to the magnitude of the change in skills use between 2012 and 2023.
Source: 2023, 2018, 2015, 2023 Survey of Adult Skills.
Figure 3.12 provides further insight into the mechanisms underpinning the widening gender gap in numeracy skills use by examining how patterns differ between qualification levels, occupations, and industries. The predicted values shown are obtained from OLS regressions controlling for individual and job characteristics, isolating how numeracy skills use evolves over time by gender within comparable groups. In all panels, men exhibit steeper increases in numeracy use between 2012 and 2023, controlling for a set of individual and job characteristics, confirming that men have benefited more from changes in work organisation and technological content of tasks.
Panel A focusses on qualification levels and reveals a substantial divergence. In 2012, low-qualified men and women used numeracy skills at a similar level. Over the following decade, however, the trajectory of low-qualified men shows a pronounced upward slope, almost converging with that of high-qualified men by 2023. Low-qualified women have also increased their numeracy use, reaching the level of high-qualified women, but the latter group remains substantially below men with similar educational attainment. This pattern suggests that even when educational disparities narrow, gendered patterns of task allocation and occupational segregation continue to limit the extent to which women apply numeracy skills at work. These results are consistent with previous research indicating that women’s underuse of numeracy skills even when controlling for proficiency levels and educational attainment (OECD, 2019[9]).
Panel B compares trends by occupational skill level. Here, the evolution among workers in unskilled occupations is particularly striking. While men in unskilled occupations started from lower levels of numeracy use in the first cycle, their numeracy use has increased much more than that of women in similar occupations. By 2023, men in unskilled occupations report using numeracy skills almost as frequently as skilled women, underscoring the gendered dimension of task allocation even within low-skill occupations. The gap among skilled workers remains large and has not narrowed over time. These results reinforce the idea that the workplace environment and job content – not only education – play a decisive role in explaining gender disparities in skills utilisation.
Panel C, which disaggregates by industry, provides one of the most revealing findings. The analysis distinguishes skilled industries – those with more than 60% of tertiary-educated workers, such as ICT, finance and education – from low-skilled ones. In 2012, men in low-skilled industries used numeracy skills at levels similar to those of women in skilled industries. Over the subsequent decade, however, male workers in low-skilled industries experienced a sharp increase in numeracy use, while women in skilled industries saw no gain. This divergence highlights that even in sectors characterised by high levels of education, women’s opportunities to apply numeracy skills have not expanded. The evidence points to persistent gender segregation within industries, where female workers are concentrated in roles with less analytical content and limited exposure to numeracy-intensive tasks.
Overall, these findings suggest that addressing gender disparities in skills use requires policy interventions that go beyond improving women’s skills and focus instead on workplace practices, job design, and career progression pathways that enable women to apply these skills throughout their working lives.
Note: The Figure represents the marginal effects of OLS regressions controlling for age, age squared, immigration background, literacy proficiency, educational attainment (not in Panel A), occupation (not in Panel B), industry (not in Panel C), firm size, full-time work, permanent contract, public employment, and country fixed effects. High-qualified workers in Panel A are defined as those with a post-secondary or tertiary education. Skilled occupations in Panel B are based on ISCO 1‑digit codes, and, as standard in the literature, include “Managers”, “Professionals”, and “Technicians”. Skilled industries in Panel C have been selected as those industries at ISIC 1‑digit level that have a share of tertiary-educated workers in PIAAC higher than 60%; namely, “Info & communication”, “Financial & insurance”, “Professional, scientific & technical”, “Education”, and “Extraterritorial”. All coefficients are statistically significant (at the 1% significance level).
Source: 2023, 2018, 2015, 2012 Survey of Adult Skills.
References
[4] Blinder, A. (1973), “Wage Discrimination: Reduced Form and Structural Estimates”, The Journal of Human Resources, Vol. 8/4, p. 436, https://doi.org/10.2307/144855.
[8] Caunedo, J., E. Keller and Y. Shin (2023), “Technology and the Task Content of Jobs across the Development Spectrum”, The World Bank Economic Review, Vol. 37/3, pp. 479-493, https://doi.org/10.1093/WBER/LHAD015.
[6] Elder, T., J. Goddeeris and S. Haider (2010), “Unexplained gaps and Oaxaca–Blinder decompositions”, Labour Economics, Vol. 17/1, pp. 284-290, https://doi.org/10.1016/J.LABECO.2009.11.002.
[7] Freeman, R., I. Ganguli and M. Handel (2020), “Within-Occupation Changes Dominate Changes in What Workers Do: A Shift-Share Decomposition, 2005–2015”, AEA Papers and Proceedings, Vol. 110, pp. 394-99, https://doi.org/10.1257/PANDP.20201005.
[5] Oaxaca, R. (1973), “Male-Female Wage Differentials in Urban Labor Markets”, International Economic Review, Vol. 14/3, p. 693, https://doi.org/10.2307/2525981.
[3] OECD (2025), Survey of Adult Skills 2023 Technical Report, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/80d9f692-en.
[1] 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.
[2] OECD (2021), The Assessment Frameworks for Cycle 2 of the Programme for the International Assessment of Adult Competencies, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/4bc2342d-en.
[9] OECD (2019), OECD Skills Outlook 2019: Thriving in a Digital World, OECD Publishing, Paris, https://doi.org/10.1787/df80bc12-en.
← 1. Despite efforts to make the composite indices of skills use comparable over time, caution is still needed when interpreting these trends, as some variables may be phrased differently in the Cycle 1 and Cycle 2 PIAAC questionnaires.
← 2. Note that this technique – typically called “shift-share decomposition” – is conceptually similar to the Blinder-Oaxaca decomposition used to construct Figure 3.4, in that it separates total change into composition and within-category effects. However, it is not the same method in the strict econometric sense, because it does not involve regression coefficients, but it uses observed averages to explain changes over time within a single group (in this case, low-qualified workers).