Mapping Skill Shortages in OECD Countries

1. Introduction

Abstract

Persistent skill imbalances are a key challenge for OECD labour markets, with important implications for productivity, innovation and inclusive growth. This chapter examines the causes and consequences of labour and skill imbalances, highlighting the impact of current structural changes on the labour market. It defines key concepts underlying the OECD Skills for Jobs Database, including skill shortages, surpluses, mismatches and gaps, and discusses how labour market imbalances can be measured.

Although labour shortages have eased since their peak in 2022, labour market tightness remains above levels observed during most of the pre‑pandemic decade (OECD, 2026[1]). The persistence of shortages despite improving labour market conditions points to structural factors affecting labour supply and demand, rather than cyclical pressures.

Population ageing is one such factor. In regions with the highest old-age dependency ratios, shrinking workforces could increase labour shortages by around 15% over the next 15 years (OECD, 2026[1]). At the same time, shortages are not only a question of the number of available workers, but also whether workers have the skills that employers need. Digitalisation, the diffusion of artificial intelligence (AI) and the transition to a low-carbon economy are changing skill requirements for many occupations (OECD, 2023[2]; OECD, 2024[3]; OECD, 2024[4]; Cedefop and UNESCO-UNEVOC., 2025[5]; OECD, 2025[6]). These changes may contribute to growing skill mismatches if workers do not acquire the new skills demanded by employers. Existing evidence suggests that such mismatches are already widespread. According to the Programme for the International Assessment of Adult Competencies (PIAAC) Survey of Adult Skills 2023, around 9% of workers report being underqualified for their job, while 10% report being under-skilled (OECD, 2024[4]).

Labour shortages are also not limited to highly skilled occupations. Many occupations requiring medium-level, occupation-specific skills, particularly in construction, transport and personal services, continue to face persistent recruitment difficulties (European Labour Authority, 2026[7]). In some cases, these shortages may reflect lower job quality rather than a lack of available workers, as demanding working conditions, non-standard hours and limited flexibility can make these occupations less attractive. The COVID‑19 pandemic further accelerated a shift in worker preferences towards greater flexibility and work-life balance, making it even harder to attract workers to some occupations (Causa et al., 2022[8]; Causa et al., 2025[9]; Knez and Arpaia, 2026[10]).

As labour and skill shortages increasingly reflect structural rather than cyclical factors, developing policies to reduce them has become a policy priority. Effective responses depend on identifying these imbalances and determining whether they stem from demographic change, skill mismatches, labour market frictions or job quality challenges. The following section reviews the main drivers of labour and skill imbalances and discusses their implications for workers, firms and economic performance.

1.1. Causes and consequences of skill imbalances

In standard textbook models of the labour market, imbalances between labour demand and labour supply are resolved through wage adjustments and worker mobility (Brave et al., 2026[11]). When labour demand exceeds labour supply, firms raise wages to attract additional workers. Conversely, when labour supply exceeds labour demand, competitive pressures compel job seekers to accept lower wages and induce firms to increase hiring, thereby restoring equilibrium.

In practice, quick adjustment is prevented by various labour market frictions. For example, imperfect information leads students and workers to form biased expectations about future skill demand, while firms cannot fully observe worker productivity prior to hiring (Spence, 1978[12]; Mortensen and Pissarides, 1994[13]; Wiswall and Zafar, 2015[14]). At the same time, mobility and reskilling are constrained by education, training and job‑search costs (Becker, 1964[15]; Pissarides, 2000[16]; Lochner and Monge-Naranjo, 2012[17]; Lochner and Monge-Naranjo, 2016[18]). In addition, geographical mobility is constrained by housing affordability, which can prevent workers from relocating to areas where labour demand is strongest (Cavalleri, Luu and Causa, 2021[19]). Wage adjustment is further slowed by wage rigidities, as wages do not easily fall or rise in response to changing labour market conditions (Shapiro and Stiglitz, 1984[20]; Akerlof and Yellen, 1990[21]).

In addition to these frictions, structural forces affect both the level and the composition of the workforce. By 2060, the working‑age population is projected to decline by around 8% on average in the OECD, and by more than 30% in a quarter of the Member countries, putting downward pressure on labour supply (OECD, 2025[6]). At the same time, digitalisation, the rise of AI and decarbonisation are reshaping the demand for skills across occupations (OECD, 2023[2]; OECD, 2024[3]; OECD, 2024[4]; Cedefop and UNESCO-UNEVOC., 2025[5]; OECD, 2025[6]). This combination of declining labour supply and shifting skill demand further increases the risk of labour and skill shortages.

Skill shortages have significant implications at the individual, firm and national levels. At the individual level, skill mismatches are associated with lower wages and job satisfaction, as well as reduced employment prospects (Wolbers, 2003[22]; Béduwé and Giret, 2011[23]; Quintini, 2011[24]; Badillo-Amador and Vila, 2013[25]; Montt, 2015[26]; Kim and Choi, 2018[27]). At the national level, skill shortages undermine competitiveness and constrain economic growth, especially when concentrated in innovation-intensive sectors such as information technology and healthcare (Mavromaras, McGuinness and Wooden, 2007[28]; McGowan and Andrews, 2017[29]).

At the firm level, skill gaps and shortages depress productivity, hinder innovation and profitability and raise employee turnover (Tang and Wang, 2005[30]; OECD, 2012[31]; Maynard and Parfyonova, 2013[32]; Brunello and Wruuck, 2021[33]; Fanti, Guarascio and Tubiana, 2021[34]; Horbach and Rammer, 2022[35]). However, the effects are not uniform. Young and small firms are more likely to report recruitment difficulties (European Commission, 2023[36]; Filippucci, Laengle and Marcolin, 2025[37]), which may partly reflect lower wages and job quality compared to larger, more established firms (Brown and Medoff, 1989[38]; Berlingieri, Calligaris and Criscuolo, 2018[39]; Calvino et al., 2025[40]). The pattern is similar across the productivity distribution: in low-productivity firms, shortages are often linked to wage and working condition constraints, whereas in high-productivity firms they more often reflect a scarcity of specialised skills (Bergeaud, Cette and Stary, 2022[41]; Camussi et al., 2025[42]; Filippucci, Laengle and Marcolin, 2025[37]).

In some cases, shortages can also trigger positive responses, prompting investment in physical capital through automation or in human capital through training (Acemoglu, 2010[43]; Danzer, Feuerbaum and Gaessler, 2024[44]; Dueholm, Kalyani and Ozkan, 2024[45]; OECD, 2024[46]; Ugurlu and Aboobaker, 2025[47]). However, the recent period of intense labour shortages did not have a significant impact on investment, which remains below the pre‑pandemic levels in many economies (OECD, 2025[48]).

1.2. Definitions and measures of skill imbalances

Skill imbalances are generally defined as imbalances between the supply of and demand for workers, or specific skills, at given wages and working conditions (Veneri, 1999[49]; Barnow, Trutko and Schede Piatak, 2013[50]; Eurofound, 2024[51]; Filippucci, Laengle and Marcolin, 2025[37]). Surpluses occur when labour supply exceeds labour demand, while shortages arise when demand exceeds supply. Furthermore, shortages may lead to skill mismatches, as employers unable to find the required skills may have to settle for underqualified candidates (OECD, 2017[52]; Brunello and Wruuck, 2021[33]; Filippucci, Laengle and Marcolin, 2025[37]). Such mismatches can persist even when aggregate labour supply and demand appear broadly balanced.

Measuring skill imbalances remains a major conceptual and empirical challenge, as skills are not directly observable and must therefore be proxied (see Box 1.1). Traditionally, this has been done using unidimensional measures such as occupation or educational attainment, reflecting both the state of economic theory and data availability. More recently, the literature has moved towards multidimensional measures of skills, distinguishing between cognitive, personality and social skills – for a review, see Woessmann (2025[53]).

Box 1.1. The evolution of the concept of skills in economics

Skills were brought to the forefront of economic analysis by human capital theory, which posits that investments in education and training help people acquire skills and knowledge that increase productivity and earnings (Schultz, 1961[54]; Becker, 1964[15]). However, in the absence of direct measures, skills were typically proxied by educational attainment (Mincer, 1974[55]; Griliches, 1977[56]; Card, 1999[57]; Heckman, Lochner and Todd, 2006[58]).

In the 1990s, the concept of skill-biased technological change was used to explain the increasing demand for more skilled workers, with wages reflecting their relative scarcity when the expansion of education lags behind (Tinbergen, 1974[59]; Katz and Murphy, 1992[60]; Goldin and Katz, 2018[61]). Despite this stronger emphasis on skills, empirical analyses continued to rely primarily on educational attainment or occupations as a proxy for skills (Katz and Murphy, 1992[60]; Berman, Bound and Griliches, 1994[62]; Autor, 2020[63]).

In the 2000s, employment growth was observed at both the high- and lower-skilled ends of the occupational distribution, while middle‑skilled jobs declined (Autor, Katz and Kearney, 2006[64]; Goos and Manning, 2007[65]). This job polarisation has been linked to the routinisation hypothesis, which posits that technological change does not affect broad groups of high- and lower-skilled workers uniformly (Autor, Levy and Murnane, 2003[66]). Rather, it operates at the level of tasks, substituting for routine activities while complementing non-routine ones. An important conceptual refinement in this literature is the distinction between skills and tasks, where tasks refer to units of work activity, while skills refer to workers’ capabilities to perform them (Autor, Levy and Murnane, 2003[66]; Acemoglu and Autor, 2011[67]). Empirically, however, task measures are typically constructed at the occupational level, implicitly treating occupations as bundles of similar tasks.

More recently, existing models of education and technology failed to explain why the demand for college‑educated labour continued to increase even as the return to cognitive skills declined after 2000 (Castex and Kogan Dechter, 2014[68]; Edin et al., 2022[69]). This prompted a shift towards multidimensional conceptions of human capital, recognising that labour market outcomes depend on combinations of cognitive and social skills rather than a single skill dimension (Deming, 2017[70]; Deming, 2023[71]).

Unidimensional proxies (occupations and educational attainment) of skills have several advantages. People acquire skills through formal education and workplace training that are often oriented toward particular occupations, while firms use occupational categories to communicate job requirements and evaluate potential applicants (Levenson and Zoghi, 2007[72]). Moreover, occupational classifications are widely used in statistical systems and administrative records, providing researchers with a standardised and readily observable measure that is available across datasets, countries and time periods.

However, unidimensional proxies are not unproblematic (Levenson and Zoghi, 2007[72]; Labussière and Bol, 2026[73]). A growing body of literature questions the assumption of occupations as bundles of skills. First, skill requirements differ widely within occupations (Yamaguchi, 2012[74]; Labussière and Bol, 2026[73]). For example, building a house requires more complex skills than installing a bookshelf, although both people doing the work could be classified as carpenters. This within-occupation skill variation persists even when using detailed occupational categories (Autor and Handel, 2013[75]; Cassidy, 2017[76]; Freeman, Ganguli and Handel, 2020[77]; Martin-Caughey, 2021[78]). Second, workers transfer skills from one occupation to another, which implies that skills overlap across different occupations (Poletaev and Robinson, 2008[79]; Gathmann and Schönberg, 2010[80]; Cheng and Park, 2020[81]).

Educational attainment faces similar limitations as a proxy for skills. First, individuals with the same level of education can possess very different skill levels. This is because educational qualifications do not capture differences in the quality of education and the skills actually acquired through schooling (Hanushek and Woessmann, 2008[82]; Woessmann, 2025[53]). In addition, skills are developed not only through formal education but also through workplace experience, training and other forms of non-formal learning that are not reflected in educational qualifications (Heisig and Solga, 2017[83]).

In response to these limitations, a growing literature measures skills directly rather than relying on occupations or educational attainment as proxies. This work conceptualises skills as multidimensional: basic skills, such as cognitive, personality and social skills, are used to acquire applied skills, which in turn are used to perform work tasks (Woessmann, 2025[53]). The expansion of this research stream has been facilitated by the increasing availability of large‑scale data sources, including job postings and online professional profiles, which offer detailed information on skill needs and workers’ skill portfolios.

While multidimensional measures provide a more detailed representation of skills, they also require researchers to make choices about which skills to measure and how to measure them, which can affect empirical findings. For example, cognitive skills are better captured by extensive domain-specific assessments, such as the PIAAC Survey of Adult Skills, than by brief IQ tests, which may miss skills that develop over the life course (Hanushek et al., 2015[84]). Thus, although multidimensional measures address some limitations of occupational and educational proxies, they do not eliminate measurement issues but relocate them to the skill level.

Given these measurement challenges, the approach for the Skills for Jobs Indicator combines the advantages of occupational classifications with detailed information on skill requirements. Rather than treating occupations as skills, we use occupations as containers of skill requirements and link them to O*NET data on the skills, knowledge and abilities associated with each occupation. This approach retains the comparability and broad availability of occupational data while allowing for a more nuanced characterisation of the skills associated with different jobs. Given the diversity of definitions used in the literature, Box 1.2 summarises the key concepts and terminology used throughout this report.

Box 1.2. Terminology

Jobs

A job is a set of tasks performed by a particular worker in the expectation of economic remuneration (Hauser and Warren, 1997[85]).

Occupations

An occupation is a category used to group and classify similar jobs (Hauser and Warren, 1997[85]). There are different ways of classifying occupations into comparable categories, the most common one being the International Standard Classification of Occupations (ISCO), which has been developed by the International Labour Organization (ILO, 2012[86]).

Skills

Skills are broadly defined as knowledge, skills, abilities and competences, and, where relevant, education (OECD, 2017[52]). Following O*NET’s definition, knowledge refers to organised sets of factual and procedural information within specific domains; skills are developed capacities that facilitate learning and the performance of tasks across occupations; and abilities are enduring individual attributes that influence performance (Boese et al., 2001[87]).

Imbalances

Skill shortages occur when employers are unable to recruit workers with the required skills at prevailing wages and working conditions in a specific location and time (Veneri, 1999[49]; Barnow, Trutko and Schede Piatak, 2013[50]; Eurofound, 2024[51]; Filippucci, Laengle and Marcolin, 2025[37]). They can be described as an inadequate supply of workers in high-demand occupations or of the skills needed to perform tasks associated with those occupations.

Skill surpluses occur when the supply of a skill exceeds demand, typically during periods of high unemployment or as a result of a misalignment between education choices and labour market demand (OECD, 2017[52]).

Skill mismatches arise when workers’ skills do not align with employers’ requirements. They can persist even in otherwise balanced labour markets and are typically measured as discrepancies between workers’ educational attainment or field of study and job requirements, as well as through self-reported assessments of whether workers’ skills match job requirements.

Skill gaps occur when employees lack the skills required for their current job, according to their employers (OECD, 2024[46]). Unlike skill shortages, which relate to recruitment difficulties, skill gaps are identified at the firm level and concern the skills of existing employees.