OECD Employment Outlook 2026

Geographic Disparities in Jobs and Incomes

Report

OECD Employment Outlook

7 July 2026

Available in:

English
français

Download PDF

1. From resilience to risk: Employment and wages under pressure

Copy link to 1. From resilience to risk: Employment and wages under pressure

Abstract

OECD labour markets have continued to show resilience over the past year, with employment and labour force participation remaining at historic highs. However, there have been new signs of weakening, with a slight rise in unemployment, a slowdown in the growth in employment and labour force participation, and a continued easing of labour market tightness – even though structural labour shortages persist. In addition, even before the recent surge in energy prices, the real wage recovery was slowing down, despite real wages remaining below the levels seen in early 2021 – just before the post-pandemic inflation surge – in one‑third of OECD countries. As real wage growth slows down, the contributions of profits and wages to domestic price pressures tend to stabilise to patterns similar to those prevailing before the COVID‑19 pandemic.

In Brief

Copy link to In Brief

OECD labour markets have continued to show resilience, but new signs of weakening emerged over the past year, with rising unemployment, a slowdown in employment and labour force participation growth, a further easing of labour shortages, and a deceleration of the real wage recovery.

The latest labour market data available at the time of writing suggest that:

Employment and labour force participation rates remain at historic highs, while OECD unemployment remains low. The OECD unemployment rate was 4.9% in May 2026, the average employment rate in the OECD reached 72.1% in Q1 2026, while the average labour force participation rate rose to 76.7%.
However, there are further signs of weakening in labour markets, as unemployment is rising slightly, employment trends are flattening and growth in labour force participation is slowing down. In particular, the median unemployment rate in the OECD increased from 5.4% in May 2025 to 5.7% in May 2026.
Many young entrants to the labour market have become more likely to be unemployed. The unemployment gap between young college graduates and the working age population has been widening since before the pandemic in all countries analysed. Young entrants without a graduate degree have also recently seen their unemployment gap increase in a few countries. So far, the role of recent advances in large language models (LLM) in explaining the difficulties facing young people appears to be limited. Instead, young labour market entrants may have been particularly vulnerable to the recent weakening of labour markets, as well as long-term changes in technology and skill needs.
Labour shortages continued to ease, but structural shortages remain. While labour market tightness is now lower than at the onset of the COVID‑19 pandemic in many countries, it remains higher than during most of the pre‑pandemic decade and, in particular, than during comparable stages of the business cycle.
Labour productivity growth is slowing down. In the euro area, labour hoarding (among other factors) has supported employment in spite of weak economic growth, and this has mechanically reduced labour productivity growth. Nonetheless, sectors that play a central role in AI adoption have boosted productivity performance in the United States.
Even before the recent surge in energy prices, the real wage recovery was slowing down, while real wages remained below the levels seen in early 2021 (just before the post-pandemic inflation surge) in one‑third of OECD countries. Annual real wage growth was positive in virtually all OECD countries in Q1 2026, but it was lower than one year earlier in two‑thirds of them. It was 2.2% in Q1 2026 on average across countries, compared with 2.7% in Q1 2025.
Real statutory minimum wages remain above January 2021 levels and continue to rise. In April 2026, the real minimum wage was higher than a year earlier in a majority of the 30 OECD countries that have a national statutory minimum wage and was higher than in January 2021 in virtually all of them.
As a result, the wages of the lowest-paid workers have proved more resilient to inflation surge than median wages. Statutory minimum wages have risen more than median wages since 2021 (i.e. Kaitz indexes have increased), and real wages have been more resilient in low wage sectors in some countries, reflecting a compression of the wage distribution at the bottom.
As real wage growth slows down, the profit – wage “catch-up” phase may be shifting towards a steadier pattern similar to that seen before the pandemic. Unit labour costs continued to rise faster than unit profits between Q1 2025 and Q1 2026 in most OECD countries, but there are signs that the catching-up of real wages – following the disproportionate contribution of unit profits to the inflation surge in 2021‑2022 – could be slowing down, with their relative contribution to domestic price pressures stabilising around pre‑pandemic levels in some countries.

Looking ahead, labour market policies (e.g. unemployment insurance, active labour market policies, collective bargaining) will play a key role in addressing the labour market impacts of persistent economic uncertainty and elevated energy costs. In particular, in a context of stabilising wage pressure and rising inflation, wage setting policies, collective bargaining systems and social dialogue have a role to play in ensuring that the cost of inflation is shared fairly between workers and employers. Addressing structural labour shortages will also require co‑ordinated efforts across various areas of labour market policy.

Introduction

Copy link to Introduction

OECD economies suffered major shocks in the wake of the COVID‑19 crisis and Russia’s war against Ukraine, and OECD labour markets proved resilient to the negative consequences of these shocks, with high employment, low unemployment and substantial labour shortages in many sectors of the economy. Several factors may have contributed to these strong employment outcomes. In particular, surging inflation kept real wages relatively low, labour hoarding became widespread in many countries, and workers’ preferences shifted toward fewer working hours (OECD, 2025[1]; 2024[2]; 2023[3]; 2022[4]; ECB, 2024[5]).

However, this decoupling between economic activity and labour market performances showed some first signs of ending in 2023, with labour shortages declining, employment growth slowing down, and the downward trend in unemployment coming to halt (albeit with historically low unemployment rates) in many OECD countries (OECD, 2024[2]; 2025[1]).

This chapter reviews recent developments by presenting the latest trends observed in the labour market and wages across OECD countries. It is organised as follows: Section 1.1 reviews recent labour market developments; Section 1.2 reports on recent wage developments, including updated information on statutory minimum wages and negotiated wages; Section 1.3 highlights the particular difficulties faced by young labour market entrants and the potential role of Artificial Intelligence (AI) in explaining these difficulties; and Section 1.4 concludes.

1.1. Labour markets have remained resilient but there have been further signs of weakening

Copy link to 1.1. Labour markets have remained resilient but there have been further signs of weakening

In Q1 2026, at the onset of the recent surge in energy prices, GDP growth in OECD countries remained resilient (Figure 1.1). However, the evolving conflict in the Middle East is testing the resilience of the global economy. The duration and extent of the conflict remain uncertain, though at the time of finalising this publication a negotiated settlement is in place. Even after the conflict ends, the economic effects are likely to be felt for some time given the months it will take to restore damaged infrastructure and transport routes and deliver products around the world (OECD, 2026[6]).

Figure 1.1. GDP growth remained resilient in the first quarter of 2026
Copy link to Figure 1.1. GDP growth remained resilient in the first quarter of 2026

Real GDP indexed to 100 in Q1 2021, seasonally adjusted, selected OECD countries

1.1.1. Unemployment remains low but is on the rise

Unemployment remains low in many OECD countries but is rising in about two‑thirds of them (Figure 1.2, Panel A). In particular, the unemployment rate was more than 0.2 percentage points (p.p.) higher in Q1 2026 than a year earlier in many European countries1 and Chile (Panel A). While the median unemployment rate has increased from 5.4% in Q1 2025 to 5.7% in Q1 2026, the overall OECD unemployment rate has risen only slightly, from 4.9% to 5%. This recent increase in unemployment mainly affects young and prime age workers (Annex Figure 1.A.1). Young workers have faced specific labour market difficulties in recent years, which are discussed in detail in Section 1.3.

Still, unemployment has declined substantially in some countries over the past year. In Q1 2026, the annual decline in the unemployment rate was greater than 0.2 p.p. in two Baltic states (Estonia and Latvia), three Southern European countries (Italy, Portugal and Spain), Colombia, Costa Rica and Sweden.

Figure 1.2. Unemployment rates remain low but are on the rise
Copy link to Figure 1.2. Unemployment rates remain low but are on the rise

Unemployment rate (percentage of labour force), seasonally adjusted

1.1.2. Growth in employment and labour force participation is slowing down

Employment and labour force participation rates remain at historic highs – 72.1% and 76.7% respectively in Q1 2026 on average in the OECD, but their growth is slowing down. On average in the OECD, employment trends have flattened. Labour force participation rates rose by 0.1 p.p. between Q1 2025 and Q1 2026, compared with 0.2 p.p. between Q1 2024 and Q1 2025 (Figure 1.3, Panel A). Employment rates decreased markedly over the past year in Iceland, Ireland and Luxembourg. By contrast, employment rate growth accelerated dramatically in Estonia, Latvia and Poland.

Over the past two years, employment rate growth has been strongest in some Latin American countries (Colombia and Costa Rica) and Southern European countries (Italy, Greece, Portugal and Spain), as well as in Japan and Lithuania. The growth in the labour force participation rate has also been relatively strong in all these countries, except for Costa Rica, Italy and Spain (Figure 1.3, Panel B).

In most OECD countries, the gender gap in employment and labour participation rates have narrowed over the past year (Annex Figure 1.A.2, Panels A and B), continuing a trend observed throughout the recovery from the COVID‑19 crisis (OECD, 2024[2]).

On average across OECD countries, approximately one‑third of the increase in the employment rate between Q1 2024 and Q1 2026 is due to the increase in the employment rate of prime age workers (25‑54), and two‑thirds to the increase in the employment rate of older workers (55‑64) (Annex Figure 1.A.1, Panel B). By contrast, the employment rate of younger workers (15‑24) fell over the period. The increase in the employment rate for older workers is part of a trend that has been ongoing for two decades (OECD, 2025[1]). Similarly, the increase in the labour force participation rate between Q1 2024 and Q1 2026 is solely due to prime age and older workers, in roughly equal shares on average across OECD countries (Panel C).

In some countries, the increase in employment and/or rising labour force participation rates has been largely driven by growing employment and/or labour force participation rates among migrants. This is the case in Austria, Belgium, Denmark, Germany, Greece, Latvia, Norway, the Netherlands and Sweden (Annex Figure 1.A.3). Nonetheless, the gaps between both the average employment and unemployment rates of migrants and native‑born people in the OECD widened in 2024, following a strong post-pandemic recovery in migrants’ labour market outcomes in 2022‑2023 – in most countries, migrants fare worse than native‑born people in terms of employment and unemployment, even though they tend to participate more in the labour market (OECD, 2025[7]).

Figure 1.3. Growth in employment and labour force participation rates is slowing down
Copy link to Figure 1.3. Growth in employment and labour force participation rates is slowing down

p.p. change among the working age population (persons aged 15‑64), seasonally adjusted data

1.1.3. Labour markets continued to ease but structural shortages remain

Labour markets have eased further over the past year in most of the 27 OECD countries analysed in Figure 1.5 Panel A. Labour market tightness (measured by the number of vacancies per unemployed person)2 was higher in Q1 2026 than one year earlier in only five countries: Estonia, Poland, Portugal, Slovenia and Spain. With the exception of Slovenia, these countries are among the one‑third of OECD countries where unemployment has declined over the past year (Figure 1.2).

In addition, in Q1 2026, labour market tightness was close to or below pre‑COVID‑19 levels in most of the countries analysed. It was also well below the post-COVID‑19 peak in all countries where there was one, with the exception of Israel. In the EU, in Q1 2026, the share of firms citing lack of labour as one of the factors limiting production is also close or below the pre‑pandemic records of 2019: one in six firms in industry and one in five firms in services mentioned to be limited by lack of labour.3 In this context of decreasing recruitment difficulties, labour hoarding behaviours have continued to decline (see Box 1.1).

Box 1.1. Labour hoarding in the EU has been declining but remains above pre‑COVID‑19 levels

Copy link to Box 1.1. Labour hoarding in the EU has been declining but remains above pre‑COVID‑19 levels

Since the pandemic, employment has been supported by strong labour hoarding in many OECD countries, including in EU countries (ECB, 2024[5]), the United Kingdom¹ and the United States² – in other words, firms have preferred to adjust to negative shocks by retaining their staff while reducing working hours rather than cutting jobs. This has accentuated the decoupling between employment levels and economic activity.

Between the onset of the pandemic (Q1 2020) and early 2023, on average over the 20 OECD EU countries analysed, labour hoarding behaviour (measured by the share of firms expecting their output to decrease but their employment to increase or remain unchanged over the next 3 months) reached historic highs, overall and in most sectors (the construction sector being an exception, with persistently high labour hoarding) (Figure 1.4, Panel A). Labour hoarding was initially (2020-2021) encouraged by the widespread use of job retention schemes and other exceptional measures aimed at supporting employment during the pandemic. Then, in a second phase (2022-2023), high profit margins and persistent labour shortages increased the expected costs of rehiring. In this context, Europe’s high exposure to rising gas prices and the uncertainty surrounding Russia’s war against Ukraine led many firms to hoard labour while waiting for the economic situation to improve (ECB, 2025[8]; ECB, 2024[5]; OECD, 2021[9]). This rise in labour hoarding was particularly marked in the trade sector.

Since then, labour hoarding has declined (Figure 1.4, Panel A), as economies have adjusted to the war in Ukraine, and profits (see Section 1.2) and labour shortages (Figure 1.5) have moved closer to pre‑pandemic levels. In Q1 2026, labour hoarding was lower than one year earlier in two‑thirds of the countries analysed (Panel B).

Nevertheless, labour hoarding also remains above pre‑pandemic levels in two‑thirds of the countries analysed (Figure 1.4, Panel B), and by more than 10% in about half of them, as economic uncertainty remains high and real wages have not caught up with the post-pandemic inflation surge (see Section 1.2). In addition, shifts in worker preferences towards shorter hours may have increased rehiring costs, further fuelling labour hoarding (OECD, 2025[1]).³

In the future, labour hoarding could remain substantially higher than before the pandemic, as real wages could remain below pre‑pandemic levels in a number of countries. Elevated energy costs in particular may weaken labour markets and raise inflation, at least in the short run. Several other factors could increase the expected costs of rehiring: demographic ageing will impose workforce reductions (OECD, 2025[1]) – thereby increasing labour market tightness –, the digital transformation (AI) and the move toward carbon neutrality may accentuate skill mismatches (OECD, 2023[3]; OECD, 2024[2]), and worker preferences may continue to shift toward fewer working hours.¹

Figure 1.4. Labour hoarding in the EU has been declining but remains above pre‑COVID‑19 levels
Copy link to Figure 1.4. Labour hoarding in the EU has been declining but remains above pre‑COVID‑19 levels

Percentage of firms which expect their output to decrease but their employment to increase or remain unchanged over the next three months, seasonally adjusted data

Figure 1.5. Labour market tightness is close to or below pre‑COVID‑19 levels in most countries but remains high
Copy link to Figure 1.5. Labour market tightness is close to or below pre‑COVID‑19 levels in most countries but remains high

Difference in the vacancies-to‑unemployed ratio relative to its Q4 2019 level

However, most OECD labour markets remain tight when viewed over a longer time horizon. While, on average across the countries analysed, labour market tightness was lower in Q1 2026 than at the onset of the pandemic, it was higher in Q1 2026 than during most of the pre‑pandemic decade and, in particular, than during comparable stages of the business cycle (Figure 1.5, Panels B and C). In fact, average labour market tightness had been on the rise for several years when the pandemic hit. A number of structural factors, which are still at play, contribute to explaining this trend, including population ageing, the digital transformation, the move toward carbon neutrality and poor job quality in certain sectors (Dorville, Filippucci and Marcolin, 2025[10]; Filippucci, Laengle and Marcolin, 2025[11]; OECD, 2025[1]; Knez and Arpaia, 2026[12]).

To capture labour market tightness at industry level, Figure 1.6 provides an overview of the industry-specific vacancy rates (i.e. the fraction of all jobs available in the industry that are unfilled and for which employers report that they are actively trying to recruit) in Australia, the euro area and the United States.

The loosening of the labour market over the past year has been particularly pronounced in some sectors, including business services (Australia and the euro area), transportation and storage (Australia and the euro area), information and communication (the euro area and the United States), real estate (Australia and the United States), and education (the United States) (Figure 1.6).

In Q1 2026, in both the euro area and the United States, labour market tightness was close to or below pre‑COVID‑19 levels in most sectors, including some where the peak in tightness had been particularly high, such as accommodation and food service activities or business services (Figure 1.6).4 The sectors where labour market tightness remained highest compared with pre‑crisis levels are (i) in the euro area: arts and entertainment, and manufacturing; (ii) in the United States: finance and insurance, and manufacturing. In Australia, however, labour market tightness remained substantially above pre‑COVID‑19 levels in most sectors, particularly those where peak tightness had been among the highest (Figure 1.6): accommodation and food service activities, manufacturing, and real estate.

In the three areas analysed, most sectors remain tight when looking at a longer time horizon (Figure 1.6 and Annex Figure 1.A.4). In Q1 2026, even in the euro area and the United States, labour market tightness remained higher than in Q1 2015 in most sectors, particularly in those where it had however returned to pre‑COVID‑19 levels, such as human health and social work activities; or, in the euro area, construction, and transportation and storage. Interestingly, in no country there appears to be a correlation between the typical level of pay of the sector and the difference between current tightness and Q1 2015 levels, suggesting that structural labour shortages occur both in high and low-skill jobs.

Figure 1.6. Labour market tightness is close to pre‑COVID 19 levels in most sectors in the euro area and the United States
Copy link to Figure 1.6. Labour market tightness is close to pre‑COVID 19 levels in most sectors in the euro area and the United States

p.p. difference in job vacancy rates by industry relative to their Q4 2019 levels

1.1.4. The reduction in annual hours worked per worker has slowed down

Over the past two years, the reduction in total hours worked per worker and per year has decelerated compared to its long-term trend (Figure 1.7). On average across the OECD countries for which data are available, the annualised reduction in hours worked was around 0.24% during the period 2023-2025, compared with 0.27% during the period 1995-2019 and 0.33% during the period 2019-2023. The strongest slowdown (or acceleration) in the decline (growth) in working hours between these two periods was observed in the Baltic states (Estonia, Latvia and Lithuania), five Central European countries (Austria, Czechia, Hungary, the Slovak Republic and Slovenia), Ireland and the United Kingdom. In fact, annualised growth in working hours turned positive in most of these countries.

Figure 1.7. .The reduction in annual hours worked per worker has slowed down
Copy link to Figure 1.7. .The reduction in annual hours worked per worker has slowed down

Percentage change in average hours worked per worker and per year

The decline in hours worked can be explained by several factors, including productivity gains, changes in workers’ preferences towards fewer overtime hours or part-time jobs, changes in working time regulations or collective agreements (OECD, 2021[9]), but also compositional shifts towards jobs requiring fewer working hours. Evidence from Europe suggests that worker preferences are at play rather than regulatory changes or compositional effects (Astinova et al., 2024[13]; ECB, 2021[14]): while the long-term downward trend in hours worked reflects preferences for part-time employment, the recent decline observed since the end of 2021 appears to be due to a decrease in long working hours and overtime among full-time workers.5

The recent slowdown in the decline in annual working hours per worker has not been accompanied by a deceleration in the rise (or an acceleration in the decline) of part-time or temporary employment. Instead, on average across the countries analysed, the incidence of part-time and temporary employment increased more (or decreased less) between 2023 and 2025 than between 2019 and 2023 (Annex Figure 1.A.5, Panel B), possibly in part due to the loosening of labour markets – firms had less need to attract workers by offering them full-time open-ended contracts.

1.1.5. Labour productivity growth is slowing down

Since the onset of the COVID‑19 crisis, labour productivity growth, measured in terms of real gross domestic product (GDP) per hour worked, has been substantially below its pre‑COVID‑19 trend on average across OECD countries: its average annualised growth rate was 1% over the period 2019-2025, compared with 2% over the period 1995-2019 (Figure 1.8).

Figure 1.8. Labour productivity growth is slowing down
Copy link to Figure 1.8. Labour productivity growth is slowing down

Annualised percentage change in real GDP per hours worked

The recent slowdown in productivity growth has been particularly marked in the euro area (Figure 1.8), where long-term productivity growth (1995-2019) was already relatively weak. As a result, annualised labour productivity growth in the euro area was far below the OECD median between 2019 and 2025 (Figure 1.8 and Annex Figure 1.A.6). This can be explained by the more pronounced effects of higher gas prices and uncertainty related to Russia’s war against Ukraine, as well as a tendency to hoard labour in European countries (ECB (2024[15]) and Box 1.1).

By contrast, labour productivity growth has been relatively high since the onset of the COVID‑19 crisis in two Latin American countries (Colombia and Mexico), two Central and Eastern European countries (Poland and the Slovak Republic), two Baltic states (Latvia and Lithuania), as well as in Korea and the United States (Figure 1.8 and Annex Figure 1.A.6). In all these countries, except for Mexico, productivity growth has been relatively high since at least the mid‑1990s. But productivity growth has even exceeded its long-term dynamic trend in Poland and the United States. In the United States, labour productivity growth registered a small uptick above its long-term trend of 1.7‑1.8% over 2023-2025 (Figure 1.8), driven by a number of sectors that play a central role in the adoption of AI technologies and the transition to online retail (e.g. computer systems, data processing, online retail)6 (Hobijn et al., 2025[16]). This suggests that AI may have played some role in the United States’ strong productivity performance over recent years.

1.1.6. Labour markets are highly uncertain

In the near future, economic activity and, as a result, labour markets, will be largely shaped by the evolution of the energy crisis, the time taken to achieve a lasting settlement to the conflict in the Middle East and the resulting policy responses (OECD, 2026[6]).

Assuming the disruptions from the conflict are sizeable but limited to a relatively short period of time, and an otherwise solid underlying momentum in the global economy – with output boosted by strong AI-related investment, production and trade, supportive financial and fiscal conditions, and lower US effective tariff rates on imports than originally announced – global growth is projected to soften considerably in the second quarter of 2026, before picking up gradually thereafter (OECD, 2026[6]). Consistent with these developments, employment growth is expected to be relatively subdued in 2026 before edging up in 2027, while the median unemployment rate across OECD economies is projected to rise further in 2026 before falling back around its Q4 2025 level at the end of 2027 (Figure 1.9).

These projections should nonetheless be treated with caution, given the exceptionally uncertain situation and the extent to which economic prospects depend on achieving a durable resolution to the conflict (OECD, 2026[6]). Failing to secure a peace agreement would lead to substantially worse economic and labour market outcomes and could have adverse impacts on a wider range of countries or industries.

Figure 1.9. OECD labour markets are expected to slow down temporarily in 2026 if the conflict in the Middle East remains of limited duration
Copy link to Figure 1.9. OECD labour markets are expected to slow down temporarily in 2026 if the conflict in the Middle East remains of limited duration

In response to practices by its trading partners that the United States considered to be unfair and discriminatory, the country made a historic turn in its import tariff policy, shifting from targeted protection on selected goods toward broader and higher tariff regimes affecting multiple products (e.g. cars, car parts, steel and aluminium products, but also energy and oil) and most trading partners.7 2025 marks the year of most significant tariff increases. In the absence of a causal analysis, the assessment of the impact of 2025 tariff rates must rely on structural modelling and empirical analyses of previous (smaller) tariff increases, which makes it somewhat speculative. Moreover, the 2025 tariff rate hike was partially reverted following a decision of the US Supreme Court in February 2026,8 and uncertainty remains about future tariff rates.

The overall impact of the 2025 tariff rates on employment in the United States is expected be slightly negative, once inflation, higher input costs and possible foreign retaliation are accounted for. Structural modelling points to moderate negative net employment effects in the United States alongside composition shifts (Ai et al., 2025[17]; OECD, 2025[18]): tariffs may increase manufacturing employment, but at the cost of employment declines in other sectors (e.g. service and agriculture), with negative effects overall (Rodríguez-Clare, Ulate and Vasquez, 2025[19]).9 Empirical analyses of 2018-2019 tariff implementations support these conclusions: they find that import protection gains were outweighed by rising input costs and retaliation (Autor et al., 2024[20]; Smarzynska Javorcik et al., 2023[21]), even within the (protected) manufacturing sector (Flaaen and Pierce, 2019[22]). Choi and Hong (2025[23]) complement these findings by examining the 2018 safeguard measures on washing machines. They show that the new production jobs created were not in US major domestic manufacturers, but rather in multinationals which relocated production.

Similarly, the detrimental employment effects of 2025 tariffs on OECD partner countries should remain moderate overall (OECD, 2025[18]). Nonetheless, in some countries (e.g. Canada or EU countries), substantial impacts can be expected in the manufacturing sector – which is particularly exposed to tariffs (Galgóczi, Watt and Mehtap, 2025[24]; Statistics Canada, 2025[25]). In addition, besides lowering exports and raising input costs (Salotti et al., 2019[26]), US tariffs may redirect tariff-hit Chinese exports towards the euro area, a phenomenon observed following the 2018 tariff wave (ECB, 2025[8]). This could intensify competition in vehicles and chemicals and put further pressure on manufacturing jobs in Europe (ECB, 2025[27]).

Uncertainty in US tariff policy remains high (OECD, 2025[28]), which can exert negative effects on employment independently of the actual tariff rates applied (Javorcik et al., 2019[29]). Ongoing legal challenges and negotiations, and the risk of new tariffs on currently exempt items might discourage investment, disrupt production planning, and delay hiring decisions, particularly in sectors integrated into global value chains. Over time, this policy volatility might weaken business confidence and dampen labour demand even in the case of moderate or declining tariff levels.

1.2. The real wage recovery was already slowing down before the recent surge in energy prices

Copy link to 1.2. The real wage recovery was already slowing down before the recent surge in energy prices

Inflation (as measured based on the Consumer Price Index) was on a downward path until the recent surge in energy prices hit OECD economies. In Q1 2026, it was lower than one year earlier in a majority of OECD countries (Figure 1.10, Panel A), and headline inflation had decreased from 4.3% in Q1 2025 to 3.6% in Q1 2026. Energy inflation was also lower in Q1 2026 than one year earlier (Panel B); in other words, the effects of the conflict in the Middle East were still not reflected in prices in Q1 2026. The expected rise in energy inflation will weigh more heavily on low-income households, which tend to spend a larger share of their budget on these products (Caisl et al., 2023[30]).

As a result, headline inflation was a long way from its 2022 peak at the onset of the latest surge in energy prices. In Q1 2026, inflation in the OECD area (3.6%) was less than half its Q3 2022 level of 10.4% (OECD, 2025[1]), but still above 2%, which is the central bank target for many OECD countries (e.g. the euro area, the United Kingdom and the United States). Inflation remained above this target in 25 OECD countries – 31% in Türkiye and above 4% in three other OECD countries.

Figure 1.10. Inflation was on a downward path until the recent surge in energy prices
Copy link to Figure 1.10. Inflation was on a downward path until the recent surge in energy prices

Inflation defined as annual percentage change in the consumer price index (CPI), Q1 2026

1.2.1. Real wage growth was slowing down in early 2026, while there was still room for catching up in one‑third of OECD countries

According to the latest data available, annual real wage growth was positive in virtually all OECD countries in Q1 2026 – before the recent surge in energy prices –, but was slowing down (Figure 1.11, Panel A), in spite of decreasing inflation.10 The average annual growth in real wages across the 37 countries for which data are available was 2.2% in Q1 2026, compared with 2.7% in Q1 2025, and annual real wage growth in Q1 2026 was lower than a year earlier in two‑thirds of these countries. In addition, recent data on wages advertised in online vacancies for nine countries also indicate a further slowdown in real wage growth in recent months (Box 1.2), as the conflict in the Middle East began to affect prices.

Despite persistent annual growth, in Q1 2026, real wages remained below their Q1 202111 levels (pre‑dating the post-pandemic inflation surge) in about one‑third (13) of the 37 countries analysed (Figure 1.11, Panel C). Real wages were more than 2% below Q1 2021 levels in a quarter of countries: Australia, Czechia, Denmark,12 Italy, New Zealand and Sweden. It should be noted that wage recovery is slowing down in most of these countries, with annual real wage growth in Q1 2026 being lower than a year earlier – wage growth accelerated only in Czechia and Sweden (Panel A). Nevertheless, real wages have regained some of the lost ground in virtually all OECD countries – real wages are near the trough of the cost-of-living crisis only in New Zealand and Australia.

Several factors could explain the slowdown in real wage growth between Q1 2025 and Q1 2026. The slowdown in labour productivity and the easing of labour market tightness, as discussed in the previous section, may have moderated wage growth, particularly for new hires. On the institutional side, negotiated wage claims have been tempered in some countries (see Box 1.3 and European Commission (2025[31])), while geopolitical and trade tensions have maintained a climate of high economic uncertainty.13 In the future, geopolitical uncertainties and a time-limited increase in energy costs may significantly weaken labour markets while exerting further upward pressure on inflation, which likely will depress wages (Figure 1.12) (OECD, 2026[6]).

Figure 1.11. Real wage growth is slowing down while real wages remain below Q1 2021 levels in one‑third of OECD countries
Copy link to Figure 1.11. Real wage growth is slowing down while real wages remain below Q1 2021 levels in one‑third of OECD countries

Figure 1.12. Real wages are expected to weaken
Copy link to Figure 1.12. Real wages are expected to weaken

Percentage change relative to Q1 2026 in projected real compensation rate

Box 1.2. Online vacancies from selected OECD countries point to a recent slowdown in the growth of real posted wages

Copy link to Box 1.2. Online vacancies from selected OECD countries point to a recent slowdown in the growth of real posted wages

Data on advertised wages from job postings on the online platform Indeed¹ indicate a slowdown in the growth of real posted wages between November 2025 and April 2026 in most countries for which data are available (Figure 1.13). In countries where real wage growth is decreasing (France, Germany, Ireland, Italy and the United States), this is due to higher inflation in April 2026, sometimes combined with declining nominal wage growth (France, Spain and the United States). The euro area as a whole has also experienced both declining nominal wage growth and higher inflation in April 2026.

In Canada, the Netherlands, Spain and the United Kingdom, the effects of the conflict in the Middle East had remained limited up to April 2026. In the Netherlands and the United Kingdom, nominal wage growth was following a fall in inflation between November 2025 and April 2026, so real wage growth remained virtually constant. In Canada and Spain, both inflation and nominal wage growth remained stable over the period.

Figure 1.13. Posted wages point to a recent slowdown in real wage growth
Copy link to Figure 1.13. Posted wages point to a recent slowdown in real wage growth

Year-on-year percentage change, three‑month moving averages, from November 2025 to April 2026

Box 1.3. Real negotiated wage growth in selected OECD countries is slowing down, but there is still room for catching up

Copy link to Box 1.3. Real negotiated wage growth in selected OECD countries is slowing down, but there is still room for catching up

Real negotiated wages (i.e. wages defined in collective agreements, as opposed to those actually paid to workers) continued to increase, albeit at a slower pace than a year earlier (Figure 1.14). Growth in real negotiated wages was positive between Q1 2025 and Q1 2026 in the 13 countries analysed as well as in the euro area as a whole, although generally lower than between Q1 2024 and Q1 2025 – only Canada, Finland and Sweden recorded a marked acceleration.

Nevertheless, in all countries except Denmark and the Netherlands, where they have fully recovered, and Mexico, where they did not fall with the inflation surge (OECD, 2025[1]), real negotiated wages remain below the levels observed just before the post-COVID‑19 inflation surge (Figure 1.14). Still, in Q1 2026, real negotiated wages were less than 2% below their Q1 2021 levels in about one‑third the countries analysed (Austria, Belgium, France and the United States).

The dynamics of real negotiated wages reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the time lag between the completion of negotiations and actual wage revisions, the infrequent use of automatic inflation indexation, and workers’ bargaining power (OECD, 2023[3]). Between the start of the cost-of-living crisis and Q1 2026, as bargaining rounds multiplied and affected a growing number of workers, negotiated wages regained more of the ground lost during the inflation surge.

Looking ahead, the European Central Bank’s (ECB) wage tracker forecasts annual nominal negotiated wage growth to stabilise at around 2.6% in Europe by the end of 2026, below the levels seen in 2025 and in the first quarter of 2026.¹

Figure 1.14. Real negotiated wages continued to increase but remain below Q1 2021 levels
Copy link to Figure 1.14. Real negotiated wages continued to increase but remain below Q1 2021 levels

Percentage change in real negotiated wages (i.e. resulting from collective agreements) since Q1 2021

1.2.2. The wages of the lowest-paid workers have proved more resilient than median wages to the inflation surge

Real minimum wages remain above January 2021 levels but their growth is slowing down

Real minimum wages increased year-on-year but the rate of growth is slowing down. In April 2026, their annual growth rate was positive in nearly two‑thirds of the 30 OECD countries that have a national minimum wage in place (Figure 1.15, Panel A). However, real minimum wages increased less between April 2025 and April 2026 than between April 2024 and April 2025 in about two‑thirds of countries; and their median annual growth rate was 1.4% in April 2026, compared with 2.3% in April 2025. The real minimum wage decreased between April 2025 and April 2026 in 11 countries: Australia, Belgium, Canada, France, Greece, Luxembourg, New Zealand, Poland, Spain, Türkiye and the United States.14

The real statutory minimum wage was also higher in April 2026 than in April 2021 in virtually all OECD countries with a national statutory minimum wage (Figure 1.15, Panel B). The real minimum wage was 11% higher in April 2026 than in April 2021 on average across these 30 countries, and its median increase, unaffected by outliers (e.g. Mexico), was 9%. The real minimum wage was lower in April 2026 than in January 2021 in only two countries – Canada and the United States.

During the post-pandemic inflation surge and the ensuing cost of living crisis, national minimum wages were able to keep pace with inflation thanks to automatic or discretionary increases in the nominal minimum wage, introduced by countries to support the lowest earners (OECD, 2023[3]). By contrast, negotiated minimum wages have been less dynamic in countries without a national minimum wage (see Box 1.4). During 2022, the real gains resulting from automatic or discretionary increases in the nominal minimum wage quickly faded away on average across countries, as inflation continued to rise (Figure 1.15, Panel C). In 2023 and 2024, larger minimum wage adjustments and moderate inflation allowed the real minimum wage to catch up (in 2023) and exceed (in 2024) its January 2021 level on average across countries. During 2025 and early 2026 (before the surge in energy prices), still moderate inflation combined with minimum wage adjustments similar to those in 2024 on average led to a further increase in the average real minimum wage. In April 2026, the impact of the new surge in energy prices on real minimum wages has become visible (‑1.5% since the beginning of the year on average across countries), although rapid adjustment mechanisms in many countries can be expected to restore purchasing power in the future months. Nevertheless, minimum wage trends should be carefully monitored given the high level of geopolitical uncertainty.

Box 1.4. Minimum wages have grown by less in countries without a national minimum wage

Copy link to Box 1.4. Minimum wages have grown by less in countries without a national minimum wage

In countries without a national statutory minimum wage, wage floors are predominantly established through collective agreements. Eurofound has analysed minimum wages negotiated in ten representative low-paid occupations (such as cleaners, waiters, and cooks) in six European countries without a national minimum wage – i.e. Austria, Denmark, Finland, Italy, Norway and Sweden (Eurofound, 2025[36]).

Between January 2020 and January 2025, negotiated minimum wages decreased in real terms in most of the occupations analysed. This contrasts with the increases in the real minimum wage observed in virtually all OECD countries with a national statutory minimum wage during this period (OECD, 2025[37]; OECD, 2024[2]). There is some variation across countries: for example, real negotiated minimum wages in Italy have fallen increasingly behind those of the five other countries, while Austria recorded the strongest real growth.

Yet, several post-2022 collective agreements have incorporated specific mechanisms to protect low-wage workers. In Sweden, 2023 and 2024 collective agreements ensured that wages below a certain threshold received a higher percentage increase than wages above that threshold. Similarly, under 2024 collective agreements in Norway, workers in low-wage sectors often received wage increases above the general increase, and in several sectors (e.g. retail, and hotels and restaurants), lowest-paid employees received a wage increase above the sector’s general increase.

Figure 1.15. Real minimum wages remain above January 2021 levels and continue to rise
Copy link to Figure 1.15. Real minimum wages remain above January 2021 levels and continue to rise

Statutory minimum wages have increased more than median wages since 2021

In most OECD countries, minimum wages increased by more than median wages between 2021 and 2025, reflecting a compression of the wage distribution at the bottom. Figure 1.16 illustrates the changes in the minimum wage‑to-median wage ratio (or Kaitz index) during this period. There was an increase in 20 of the 30 countries analysed. On average, the Kaitz index rose from 54.9% in 2021 to 55.6% in 2025. Mexico exhibits by far the largest increase in the Kaitz index, reflecting the dramatic rise in the minimum wage illustrated in Figure 1.15, Panel B. In some countries, this wage compression at the bottom of the distribution led to a substantial increase in the share of employees earning the minimum wage, as the latter caught up with slightly higher wages (Box 1.5). Türkiye is an exception, with a substantial drop in the Kaitz index, in line with very strong real growth of average wages since 2021 (Figure 1.11).

This trend toward wage compression at the bottom of the distribution can be explained in part by the fact that wages negotiated in collective agreements generally take longer than the minimum wage to keep pace with inflation (OECD, 2023[3]), particularly in a context of low productivity growth (OECD, 2024[38]) and a long-term trend toward a reduction in the scope of collective agreements (OECD, 2019[39]). Further wage compression should be observed in 2026, as inflation fuelled by rising energy costs is expected to lead to substantial minimum wage increases while impacts on negotiated wages can be expected to take time.

Between 2021 and 2025, as minimum wages moved closer to median wages, minimum wage workers became less at risk of in-work poverty – i.e. their net disposable income increased relative to the poverty line (set at 50% of median net disposable income) (Figure 1.17).15 This is true for the majority of OECD countries analysed, as well as for the average across countries. The same remains valid when looking at the evolution of real income relative to an invariant “pre‑inflation” poverty line (anchored at its 2021 level, i.e. 50% of the 2021 median income), albeit to a slightly lesser extent, in line with the overall increase in real minimum wages over the period.

However, in a number of countries, particularly those where real wages remain close to or below pre‑pandemic levels, workers with middle and lower-middle earnings, who did not directly benefit from minimum wage increases, saw their economic security deteriorate (European Commission, 2025[31]). In addition, by narrowing pay differentials, wage compression can reduce early-career progression prospects, especially for workers with weaker bargaining power, by lowering the returns to incremental skill acquisition and job mobility within the low-wage segment. For the tax-and-benefit systems, compression may mechanically increase the caseload and budgetary cost of in-work supports and bottom-targeted transfers if more workers cluster near eligibility thresholds. A larger number of workers eligible for targeted transfers could also further weaken workers’ incentives to accept slightly higher-paid jobs or invest in short training spells. Indeed, in some countries the tax and benefit system is susceptible to generate low-wage traps mechanisms, i.e. gross wage gains at the bottom of the wage distribution translate into limited net income increases once taxes and social contributions are applied and means-tested benefits are withdrawn (as is the case, for example, in France, Germany or the United Kingdom, see Barreto et al. (2025[40])).

Figure 1.16. Statutory minimum wages have increased more than median wages
Copy link to Figure 1.16. Statutory minimum wages have increased more than median wages

Gross minimum wage as a percentage of gross median wage of full-time workers

Figure 1.17. In-work poverty risk has decreased for minimum wage workers
Copy link to Figure 1.17. In-work poverty risk has decreased for minimum wage workers

Actual and anchored poverty gaps for selected minimum-wage household types

Box 1.5. The share of minimum wage workers evolved along different trends across countries during the cost-of-living crisis

Copy link to Box 1.5. The share of minimum wage workers evolved along different trends across countries during the cost-of-living crisis

Minimum wage hikes during the cost-of-living crisis could have resulted in an increase in the share of workers earning the minimum wage, if the minimum wage had caught up with wages higher up in the distribution. This phenomenon would be more pronounced in countries where wages close to the minimum wage (i) cover a large number of workers or (ii) benefit from only limited spillover effects from minimum wage increases.

However, among the 10 countries analysed, minimum wage increases between 2021 and 2024 were associated with a steady increase in the share of minimum wage workers in only two of them: France and Japan (Figure 1.18). By contrast, in six other countries (Ireland, the Netherlands, New Zealand, Spain, the United Kingdom and the United States), the share of minimum wage workers follows a U-shape curve over time (Figure 1.18, Panel A). It initially declined, mainly in 2022, possibly because many minimum wage workers were able to negotiate higher wages in response to soaring inflation, because of widespread labour shortages at the bottom of the wage distribution in the wake of the pandemic and during the cost-of-living crisis (Autor, Dube and McGrew, 2023[41]). The share of minimum wage earners then increased in 2023 – 2024, sometimes to levels above those of 2021 (e.g. in Ireland, Spain and the United Kingdom), as the minimum wage rose significantly, possibly regaining some of the ground lost to market wages. Finally, in Lithuania and Canada, the share of workers at the minimum wage did not increase. In Canada, this pattern likely occurs because market wages increased together with the minimum wage – median wages increased at the same pace as the minimum wage (see Figure 1.16 above). In Lithuania, persistent labour shortages in low-skill sectors, likely due to high outmigration of low-skilled workers (OECD, 2025[1]), may have conferred larger bargaining power to workers at the bottom of the distribution.

Figure 1.18. The share of minimum wage workers evolved along different trends across countries during the cost-of-living crisis
Copy link to Figure 1.18. The share of minimum wage workers evolved along different trends across countries during the cost-of-living crisis

Real wages have been more resilient to the post-pandemic inflation surge in low-wage sectors in some countries

In addition to the minimum wage, other factors may have influenced wage compression at the bottom of the distribution in recent years. In the United States, for example, it was the tightening of the labour market rather than minimum wage increases that enabled some low-wage workers to benefit from substantial wage increases (Autor, Dube and McGrew, 2023[41]).

In fact, since the onset of the post-pandemic inflation surge (i.e. since Q1 2021), there has been a clear trend towards wage compression between sectors in the United States, a phenomenon that has not been observed in Australia or the euro area (Figure 1.19). In the United States, accommodation and food service activities is the sector with the highest wage growth over the period Q1 2021 – Q1 2026, while the largest declines in real wages are all in sectors where wages are relatively high: finance, information and communication, education, business services, and construction. The situation is less clear-cut in some other OECD countries (Annex Figure 1.A.8): there are some trends toward wage compression between sectors in Chile, Mexico and Poland but, except in Chile, they are significantly less pronounced than in the United States.

Wage compression between sectors in the United States has ceased if we focus on the latest period Q1 2025 – Q1 2026 (Figure 1.19). The only two sectors where real wages decreased are low-wage sectors (accommodation and food service activities and administrative and support services), while finance exhibits the strongest wage growth.

Figure 1.19. There has been a clear trend towards wage compression between sectors since Q1 2021 in the United States but not in Australia or the euro area
Copy link to Figure 1.19. There has been a clear trend towards wage compression between sectors since Q1 2021 in the United States but not in Australia or the euro area

Percentage change in real hourly wages since Q1 2021

1.2.3. The catching up of real wages relative to unit profits was already slowing down before the recent surge in energy prices

Unit labour costs16 (i.e. nominal compensation of employees divided by real GDP) rose more than unit profits (i.e. gross operating surplus divided by real GDP) between Q1 2025 and Q1 2026 in most OECD countries (Figure 1.20), continuing a trend that began around Q1 2023 (OECD, 2024[2]). While unit profits remained virtually unchanged in the majority of countries between Q1 2024 and Q1 2026 – the magnitude of the variation was greater than 5% in only one‑third of the 29 countries analysed –, unit labour costs increased significantly in many countries – by more than 5% in about three‑quarters of countries.

These developments, most of which pre‑date the recent surge in energy prices, still reflect the catching-up of purchasing power by wages and should not be interpreted as a warning sign of price‑wage spirals – i.e. a sustained acceleration in nominal wage growth passing through to domestic prices beyond its effects on productivity –, as (i) wages had not fully caught up with the inflation surge in half of OECD countries (see previous section), which suggests that, in these countries, unit profits continue to lose the ground gained during the inflation surge;17 and (ii) wage pressures are expected to stabilise at moderate levels for the remainder of 2026.18

Besides, there are signs that the catching-up of real wages might be slowing down. On average across countries, the increase in unit labour cost was slightly lower between Q1 2025 and Q1 2026 compared with the year before, while that of unit profits was slightly higher (see Figure 1.20, where the increase in the past year is represented by the difference between the top of the bar and the diamond). In addition, the relative contribution of unit labour costs to domestic price pressures has stabilised around pre‑pandemic levels in the euro area (Figure 1.21), while it is even decreasing below these levels in the United States.19,20

Figure 1.20. Profits continue to buffer rising labour costs
Copy link to Figure 1.20. Profits continue to buffer rising labour costs

Percentage change since Q1 2024, seasonally adjusted data

Figure 1.21. The contribution of unit profits to domestic price pressures has stabilised around pre‑pandemic levels in the euro area and is increasing in the United States
Copy link to Figure 1.21. The contribution of unit profits to domestic price pressures has stabilised around pre‑pandemic levels in the euro area and is increasing in the United States

Contribution to the GDP deflator, year-on-year percentage changes, seasonally adjusted data

1.3. Young labour market entrants face particular labour market difficulties

Copy link to 1.3. Young labour market entrants face particular labour market difficulties

Young labour market entrants have been increasingly exposed to unemployment compared to the rest of the working age population, following trends that vary according to their education level. While the unemployment rate of young college graduates has been rising relative to that of the overall population in all countries analysed since before the COVID‑19 pandemic, young entrants without a graduate degree have seen a significant increase in their unemployment gap relative to the overall population more recently in Canada and the United States,21 but not in Australia and the euro area (Figure 1.22).

Figure 1.22. Young labour market entrants have been increasingly exposed to unemployment compared to the rest of the working age population
Copy link to Figure 1.22. Young labour market entrants have been increasingly exposed to unemployment compared to the rest of the working age population

1.3.1. So far, the role of LLMs in explaining the difficulties encountered by young people entering the labour market appears to be limited

Recent advances in generative AI and Large Language Models (LLMs) may contribute to explaining the particular difficulties faced by young labour market entrants if these technologies tend to replace human labour.22 To date, evidence on this subject is mixed and focusses mainly on the United States.23 Some studies find no significant employment effects of generative AI (Chen and Stratton, 2026[43]; Gimbel et al., 2025[44]; Humlum and Vestergaard, 2025[45]), while others suggest that generative AI may indeed have a disproportionate impact on the employment prospects of early-career workers (with no evidence that this impact is particularly concentrated among college graduates) (Azar, Gine and Sanz-Espín, 2025[46]; Brynjolfsson, Chandar and Chen, 2025[47]; Klein Teeselink, 2025[48]; Lichtinger and Hosseini Maasoum, 2025[49]; Liu, Wang and Yu, 2025[50]; Lodefalk et al., 2026[51]). Nevertheless, some of these results might be biased by the greater sensitivity of occupations exposed to generative AI to macroeconomic developments, as these occupations are overrepresented in certain sectors that are particularly sensitive to capital cost and general economic uncertainty (e.g. information, finance and insurance, or professional and technical services), so that hiring is quickly reduced in these sectors when macroeconomic conditions worsen. Thus, what may seem an effect of the diffusion of LLMs is in fact simply the result of worsening macroeconomic environment (Curto Millet and Iscenko, 2026[52]; Frank et al., 2026[53]).

Data from Australia, Canada, the European Union and the United States suggest that the role of LLMs in explaining the unemployment gap of young labour market entrants remains limited, both for those with and without a graduate degree. Indeed, the data indicate divergent trends between, on the one hand, AI hiring (used as a proxy for AI use by firms), and, on the other hand, (i) unemployment gaps of young labour market entrants or (ii) online job postings in occupations exposed to AI language modelling capabilities24 (which will be referred to as “high LLM-exposure occupations”).

With regard to college graduates, there has been an upward trend in the unemployment gap since before the pandemic (Figure 1.22, Panel A), and therefore well before firms began turning to LLMs. In addition, trends in AI hiring from LinkedIn25 (Figure 1.23, Panel A) suggest that firms began using intensively LLMs from mid-2023 to early 2024 (depending on the country),26 with a dramatic increase in LLM use since then.27 As no turning point in the unemployment gap trend for college graduates is visible around those dates (Figure 1.22, Panel A) in any of the countries analysed, it can be concluded that LLMs diffusion is unlikely to be the main factor behind the increase in the unemployment gap for college graduates.

Figure 1.23. There have been divergent trends between AI hiring, and online job postings in high LLM-exposure occupations relative to low-LLM exposure occupations
Copy link to Figure 1.23. There have been divergent trends between AI hiring, and online job postings in high LLM-exposure occupations relative to low-LLM exposure occupations

A similar argument can be made for young labour market entrants without a college degree (Figure 1.22, Panel B). While the unemployment gap for this group in the United States – which started increasing around early 2024 – could potentially have been caused by LLMs diffusion, this is not the case in Australia – where the unemployment gap started declining in mid-2024 – nor in Canada or the euro area – where it has remained virtually flat since at least early 2024.

In addition, in all countries analysed, trends in online job postings (from Lightcast data) in high-LLM exposure occupations relative to low-LLM exposure occupations exhibits no break at the end of 2023 – which, as discussed, marks the start of intensive use of LLMs (Figure 1.23, Panel B).28 By contrast, job postings in LLM-exposed occupations appear, as expected (see the discussion above), more sensitive to macroeconomic shocks: posting went down during the COVID‑19 pandemic, improved in the ensuing recovery, and contracted again during the subsequent inflation shock, logistic disruption and monetary tightening. These observations remain unchanged when focussing on junior-level job postings in the United-States (Box 1.6).

The conclusion that LLMs diffusion has not significantly impacted the employment prospects of young entrants remain preliminary, and LLMs may well have a stronger impact in the future, as technological advances and the use of LLMs by firms develops. These broad observations may also mask substantial impacts in specific occupations or sectors, and further monitoring of the effects of LLMs advances on the labour market and the resulting skill needs is necessary.

Box 1.6. Online job postings for junior positions in the United States have not declined disproportionately in LLM-exposed occupations when firms began using LLMs

Copy link to Box 1.6. Online job postings for junior positions in the United States have not declined disproportionately in LLM-exposed occupations when firms began using LLMs

Lightcast data for the United States include a job seniority indicator that classifies job postings based on explicit mentions in the job title or role description indicating that the position targets a junior or senior candidate (e.g. specific words such as “junior”, “senior”, “entry-level”, “lead”, “principal”, as well as structured phrases related to seniority).

As with job postings overall, the end of 2023 – which marks the start of LLM use according to job postings and hiring data (Figure 1.24, Panel A) – does not correspond to either the start or a sustained acceleration of a reduction in junior postings in high-LLM exposure occupations relative to low-LLM exposure occupations (Figure 1.24, Panel B). Junior job postings in LLM-exposed occupations have, however, been more sensitive to macroeconomic shocks, and to the pandemic in particular, because, as noted above, these occupations are also the most exposed to these shocks (e.g. information, finance and insurance, or professional and technical services) (Curto Millet and Iscenko, 2026[52]), and it has been shown that, during downturns, employers reduce especially hiring in entry-level jobs which require little experience – see e.g. Raaum and Røed (2006[56]) and Brunner and Kuhn (2013[57]).

Figure 1.24. There have been divergent trends between AI hirings, and junior job postings in high LLM-exposure occupations relative to low-LLM exposure occupations in the United States
Copy link to Figure 1.24. There have been divergent trends between AI hirings, and<em> junior</em> job postings in high LLM-exposure occupations relative to low-LLM exposure occupations in the United States

1.3.2. Young labour market entrants may have been particularly vulnerable to the recent weakening of labour markets and long-term changes in technology and skill needs

Although resilient, OECD labour markets have been showing some signs of weakening since 2023 (OECD, 2024[2]; 2025[1]), which may have been felt more strongly by young labour market entrants (Curto Millet and Iscenko, 2026[52]). Young entrants are indeed more likely to be looking for a job or working on a fixed-term contract, which are the first to be terminated when the labour market deteriorates. In the United States, for example, hiring has fallen to its lowest level in a decade,29 preventing many young individuals from entering employment. Specific sectoral problems have added to these difficulties, such as job cuts in the high-tech, information, and business services sectors, following excessive hiring during the pandemic.30^,31 Working conditions may also appear less favourable than expected – lengthening the time it takes to find a job that meets expectations – as, even before the recent surge in energy prices, real wages remained below their 2021 levels in many countries (Section 1.2), and some of the flexibility gained from the remote revolution has been rolled back.32 Country-specific factors may also have played a role, such as the record increase in non-permanent residents in Canada in 2023 that led to a disproportionate rise in the young working-age population.33

On the longer-run, structural changes have been particularly unfavourable to graduate entering the labour market, which may explain the long-term upward trends in unemployment for this group compared to the rest of the working age population. Several studies show that the rapid growth of digital technologies has been fuelling a wave of “digital offshoring” or “telemigration” (see Broecke (2024[58]) for a review) In response to rapidly evolving skill needs, hiring models are also increasingly relying on demonstrated skills rather than formal credentials (OECD, 2025[59]) – see also Chapter 4. From a demographic perspective, longer working lives may have reduced hiring in management occupations relative to other occupations by increasing the proportion of older workers, who tend to be over-represented in these occupations (Aspen Economic Strategy Group, 2026[60]). Some of these trends may have been fostered by the COVID‑19 pandemic – which has boosted online leaning and digital offshoring – and recent labour shortages – which may have encouraged employers to offshore and relax degree requirements.

1.4. Concluding remarks

Copy link to 1.4. Concluding remarks

OECD labour markets have remained resilient. However, they have shown further signs of weakening, i.e. rising unemployment, further slowdowns in the growth in employment and labour force participation, a continuation of the decline in labour shortages, and a slowdown in real wage recovery. Young individuals entering the labour market may have been particularly vulnerable to this recent labour market slowdown in some countries.

Several factors – both cyclical and structural – could explain these developments. The catching up of real wages has brought real labour costs back to their pre‑pandemic levels in many – albeit not all – countries, the reduction in labour hoarding may have led to an increase in separations, the reduction in working hours has slowed down, and the use of generative AI by firms may have begun to automate certain jobs – although the data show little evidence of this to date. These factors may also be linked to the developments in labour productivity observed since the onset of the pandemic. In particular, labour hoarding has mechanically lowered labour productivity in EU countries, while certain sectors that play a central role in AI adoption account for the strong productivity performances of the United States over the past years.

Despite signs of weakening labour markets, structural labour shortages persist. While labour market tightness is now less pronounced than at the onset of the pandemic in many countries, it remains higher than during most of the pre‑pandemic decade and, in particular, than during comparable stages of the business cycle. A number of factors could contribute to this trend, including population ageing, the digital transformation, the energy transition, and poor job quality in some sectors.

Addressing these shortages requires co‑ordinated policy efforts across various areas of labour market policy (Filippucci, Laengle and Marcolin, 2025[11]), and regional variations in shortages in some countries require particular attention to local implementation. Encouraging the participation of underrepresented groups in the labour market – i.e. older people (e.g. by removing mandatory retirement, see Chapter 6), women, or migrants – could increase the supply of labour (OECD, 2025[1]). Streamlining occupational licenses and non-compete agreements (see Chapter 5) could promote labour mobility and matching efficiency. Education, reskilling and upskilling policies (Chapter 4) have a role to play in developing the skills needed for the digital transformation (OECD, 2023[3]) and the energy transition (OECD, 2024[2]), while ensuring that older workers continue to have meaningful employment in a context of demographic ageing (OECD, 2025[1]). Minimum wage and collective bargaining systems should also ensure that work offers attractive working conditions.

Labour market policies (e.g. unemployment benefits, active labour market policies, collective bargaining) will also remain essential in addressing the labour market impacts of persistent economic and geopolitical uncertainty and elevated energy costs. In particular, in a context of stabilising wage pressure and rising inflation, wage setting institutions – minimum wage and collective bargaining – have a role to play in ensuring that the cost of inflation is fairly distributed between workers and employers. Collective bargaining and social dialogue should also ensure that low-wage workers whose pay exceeds the minimum wage – and who therefore do not directly benefit from minimum wage increases – are protected against inflation. Collective bargaining can help identify solutions tailored to sectors and firms’ varying capacities to sustain further wage increases. In particular, the timing of wage negotiations matters, and could be adjusted through the frequency of bargaining rounds, early wage negotiations, bridge agreements and one‑off payments. Wage‑setting institutions could play an even bigger role than they did during the post-pandemic inflationary episode, when more dynamic labour markets were providing stronger support for market wages.

References

[33] Adrjan, P. et al. (2021), “Will it stay or will it go? Analysing developments in telework during COVID-19 using online job postings data”, OECD Productivity Working Papers, No. 30, OECD Publishing, Paris, https://doi.org/10.1787/aed3816e-en.

[32] Adrjan, P. and R. Lydon (2024), “What Do Wages in Online Job Postings Tell us About Wage Growth?”, in Big Data Applications in Labor Economics, Part A, Emerald Publishing Limited, https://doi.org/10.1108/s0147-91212024000052a021.

[17] Ai, J. et al. (2025), Economic Impacts of the 2025 Trade Wars, Elsevier BV, https://doi.org/10.2139/ssrn.5262696.

[60] Aspen Economic Strategy Group (2026), Demographic Headwinds: The Economic Consequences of Lower Birth Rates and Longer Lives.

[13] Astinova, D. et al. (2024), “Dissecting the Decline in Average Hours Worked in Europe”, IMF Working Papers, No. 24/2, IMF.

[20] Autor, D. et al. (2024), Help for the Heartland? The Employment and Electoral Effects of the Trump Tariffs in the United States, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w32082.

[41] Autor, D., A. Dube and A. McGrew (2023), The Unexpected Compression: Competition at Work in the Low Wage Labor Market, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w31010.

[46] Azar, J., M. Gine and J. Sanz-Espín (2025), AI Is Already Eroding Wages: Quasi-Experimental Evidence From Occupational Exposure, Elsevier BV, https://doi.org/10.2139/ssrn.5842084.

[40] Barreto, C. et al. (2025), “Low-wage employment in France: A cross-country perspective”, OECD Social, Employment and Migration Working Papers, No. 313, OECD Publishing, Paris, https://doi.org/10.1787/82539f44-en.

[34] Bellatin, A. and G. Galassi (2022), “What COVID-19 May Leave Behind: Technology-Related Job Postings in Canada”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4114783.

[42] Bozio, A. and E. Wasmer (2024), Les politiques d’exonérations de cotisations sociales : une inflexion nécessaire.

[58] Broecke, S. (2024), “Offshoring, Reshoring, and the Evolving Geography of Jobs: A Scoping Paper”, OECD Social, Employment and Migration Working Papers, No. 308, OECD Publishing, Paris, https://doi.org/10.1787/adc9a9d5-en.

[57] Brunner, B. and A. Kuhn (2013), “The impact of labor market entry conditions on initial job assignment and wages”, Journal of Population Economics, Vol. 27/3, pp. 705-738, https://doi.org/10.1007/s00148-013-0494-4.

[47] Brynjolfsson, E., B. Chandar and R. Chen (2025), Canaries in the coal mine? six facts about the recent employment effects of artificial intelligence, Stanford Digital Economy Lab.

[30] Caisl, J. et al. (2023), “The uneven impact of high inflation”, OECD Papers on Well-being and Inequalities, No. 18, OECD Publishing, Paris, https://doi.org/10.1787/59e2b8ae-en.

[43] Chen, F. and J. Stratton (2026), “Artificial Intelligence in the Firm”.

[23] Choi, J. and J. Hong (2025), Does Protectionism Save Domestic Jobs?: New Evidence from Tariffs on Washing Machines, Elsevier BV, https://doi.org/10.2139/ssrn.5037371.

[35] Ciminelli, G. et al. (2024), “Occupational reallocation and mismatch in the wake of the Covid-19 pandemic: Cross-country evidence from an online job site”, OECD Productivity Working Papers, No. 35, OECD Publishing, Paris, https://doi.org/10.1787/128b92aa-en.

[52] Curto Millet, F. and Z. Iscenko (2026), Looking for the Ladder: Is AI Impacting Entry-Level Jobs?, Economic Innovation Group.

[10] Dorville, Y., F. Filippucci and L. Marcolin (2025), “How do structural trends affect labour market shortages and mismatch?”, OECD Productivity Working Papers, No. 38, OECD Publishing, Paris, https://doi.org/10.1787/acfb5c31-en.

[8] ECB (2025), “Economic Bulletin Issue 3, 2025”, European Central Bank.

[27] ECB (2025), “Economic Bulletin Issue 5, 2025”, European Central Bank.

[15] ECB (2024), “Economic Bulletin Issue 6, 2024”, European Central Bank.

[5] ECB (2024), “Economic Bulletin Issue 8, 2024”, European Central Bank.

[14] ECB (2021), “Economic Bulletin Issue 6, 2021”, European Central Bank.

[36] Eurofound (2025), Minimum wages in 2025: Annual review.

[31] European Commission (2025), Labour market and wage developments in Europe – Annual review 2025, Publications Office of the European Union, https://data.europa.eu/doi/10.2767/1810636.

[54] Felten, E., M. Raj and R. Seamans (2023), “How will Language Modelers like ChatGPT Affect Occupations and Industries?”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4375268.

[11] Filippucci, F., K. Laengle and L. Marcolin (2025), “The firm side of labour shortages: Five new facts from the GFP employer survey”, OECD Productivity Working Papers, No. 39, OECD Publishing, Paris, https://doi.org/10.1787/aaea297f-en.

[22] Flaaen, A. and J. Pierce (2019), “Disentangling the Effects of the 2018-2019 Tariffs on a Globally Connected U.S. Manufacturing Sector”, Finance and Economics Discussion Series, Vol. 2019.0/86, https://doi.org/10.17016/feds.2019.086.

[53] Frank, M. et al. (2026), “AI-exposed jobs deteriorated before ChatGPT”.

[24] Galgóczi, B., A. Watt and A. Mehtap (2025), Trump on the trade and tariff warpath – how badly might the EU be affected?, ETUI.

[44] Gimbel, M. et al. (2025), Evaluating the Impact of AI on the Labor Market: Current State of Affairs, The Budget Lab, https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-current-state-affairs.

[61] Gopinath, G. and B. Neiman (2026), The Incidence of Tariffs: Rates and Reality, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w34620.

[16] Hobijn, B. et al. (2025), “Quarterly Industry-Level Labor Productivity Data for the U.S”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.5119318.

[45] Humlum, A. and E. Vestergaard (2025), Large Language Models, Small Labor Market Effects, Elsevier BV, https://doi.org/10.2139/ssrn.5219933.

[29] Javorcik, B. et al. (2019), The Brexit Vote and Labour Demand: Evidence from Online Job Postings, University of Oxford.

[48] Klein Teeselink, B. (2025), Generative AI and Labor Market Outcomes: Evidence from the United Kingdom, Elsevier BV, https://doi.org/10.2139/ssrn.5516798.

[12] Knez, K. and A. Arpaia (2026), “The dual nature of labour shortages – A cause of weak productivity growth and a consequence of poor working conditions?”, Directorate-General for Employment, Social Affairs and Inclusion Working Paper, No. 2026/01, Publications Office of the European Union, https://data.europa.eu/doi/10.2767/7084188.

[49] Lichtinger, G. and S. Hosseini Maasoum (2025), Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data, Elsevier BV, https://doi.org/10.2139/ssrn.5425555.

[50] Liu, Y., H. Wang and S. Yu (2025), Labor Demand in the Age of Generative AI : Early Evidence from the U.S. Job Posting Data (English), http://documents.worldbank.org/curated/en/099827011182513988.

[51] Lodefalk, M. et al. (2026), Same Storm, Different Boats: Generative AI and the Age Gradient in Hiring, Örebro University, School of Business.

[62] Ng, H. (2024), “Why is it so hard to find a job now? Enter Ghost Jobs”.

[55] Nurski, L. (2025), Occupational Exposure Database (Version 1.0) [Data set], AI World, https://aiworld.eu/occupation.

[6] OECD (2026), OECD Economic Outlook, Volume 2026 Issue 1: Under Pressure, OECD Publishing, Paris, https://doi.org/10.1787/2d1956f0-en.

[59] 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.

[7] OECD (2025), International Migration Outlook 2025, OECD Publishing, Paris, https://doi.org/10.1787/ae26c893-en.

[28] OECD (2025), OECD Economic Outlook, Interim Report September 2025: Finding the Right Balance in Uncertain Times, OECD Publishing, Paris, https://doi.org/10.1787/67b10c01-en.

[18] OECD (2025), OECD Economic Outlook, Volume 2025 Issue 1: Tackling Uncertainty, Reviving Growth, OECD Publishing, Paris, https://doi.org/10.1787/83363382-en.

[1] OECD (2025), OECD Employment Outlook 2025: Can We Get Through the Demographic Crunch?, OECD Publishing, Paris, https://doi.org/10.1787/194a947b-en.

[37] OECD (2025), “Real wages continue to recover: The OECD wage bulletin”, OECD Publishing, Paris, https://doi.org/10.1787/8f8ec0e4-en.

[38] OECD (2024), OECD Compendium of Productivity Indicators 2024, OECD Publishing, Paris, https://doi.org/10.1787/b96cd88a-en.

[2] OECD (2024), OECD Employment Outlook 2024: The Net-Zero Transition and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/ac8b3538-en.

[3] OECD (2023), OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/08785bba-en.

[4] OECD (2022), OECD Employment Outlook 2022: Building Back More Inclusive Labour Markets, OECD Publishing, Paris, https://doi.org/10.1787/1bb305a6-en.

[9] OECD (2021), OECD Employment Outlook 2021: Navigating the COVID-19 Crisis and Recovery, OECD Publishing, Paris, https://doi.org/10.1787/5a700c4b-en.

[39] OECD (2019), Negotiating Our Way Up: Collective Bargaining in a Changing World of Work, OECD Publishing, Paris, https://doi.org/10.1787/1fd2da34-en.

[56] Raaum, O. and K. Røed (2006), “Do Business Cycle Conditions at the Time of Labor Market Entry Affect Future Employment Prospects?”, Review of Economics and Statistics, Vol. 88/2, pp. 193-210, https://doi.org/10.1162/rest.88.2.193.

[19] Rodríguez-Clare, A., M. Ulate and J. Vasquez (2025), “The 2025 Trade War: Dynamic Impacts Across U.S. States and the Global Economy”, Federal Reserve Bank of San Francisco, Working Paper Series, Vol. 2010/09, pp. 01-48, https://doi.org/10.24148/wp2025-09.

[26] Salotti, S. et al. (2019), Macroeconomic effects of US tariffs on steel and aluminium – Who would pay the bill?, Publications Office of the European Union, https://data.europa.eu/doi/10.2760/387494.

[21] Smarzynska Javorcik, B. et al. (2023), “Did the 2018 Trade War Improve Job Opportunities for US Workers?”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4327049.

[25] Statistics Canada (2025), Impact of tariffs on businesses in Canada: Expectations and strategic responses, second quarter of 2025.

Annex 1.A. Additional results

Copy link to Annex 1.A. Additional results

Annex Figure 1.A.1. Employment rates have increased among prime aged and older individuals, but not among youth
Copy link to Annex Figure 1.A.1. Employment rates have increased among prime aged and older individuals, but not among youth

Decomposition of the overall change in the unemployment rate, employment rate and labour force participation rate (persons aged 15‑64) by age group between Q1 2024 and Q1 2026

Annex Figure 1.A.2. Gender employment and participation gaps have narrowed
Copy link to Annex Figure 1.A.2. Gender employment and participation gaps have narrowed

Male‑to-female differences in employment and labour-force participation rates (p.p.), persons aged 15‑64, seasonally adjusted

Annex Figure 1.A.3. The share of working-age migrants in employment or participating in the labour market has increased in some countries
Copy link to Annex Figure 1.A.3. The share of working-age migrants in employment or participating in the labour market has increased in some countries

Decomposition of the overall change in the employment rate and labour force participation rate (persons aged 15‑64) by country of birth between Q1 2024 and Q1 2026

Annex Figure 1.A.4. Labour market tightness is close to pre‑COVID 19 levels in most sectors in the euro area and the United States
Copy link to Annex Figure 1.A.4. Labour market tightness is close to pre‑COVID 19 levels in most sectors in the euro area and the United States

p.p. difference in job vacancy rates by industry relative to their Q4 2019 levels

Annex Figure 1.A.5. Incidence of part-time workers and temporary workers
Copy link to Annex Figure 1.A.5. Incidence of part-time workers and temporary workers

Annualised percentage change, 2019-2023 and 2023-2025

Annex Figure 1.A.6. Labour productivity growth has been diverging between OECD countries
Copy link to Annex Figure 1.A.6. Labour productivity growth has been diverging between OECD countries

Annualised percentage change in real GDP per hours worked, 1995-2019 and 2019‑2025

Annex Figure 1.A.7. Nominal and real wage index by country since Q4 2019
Copy link to Annex Figure 1.A.7. Nominal and real wage index by country since Q4 2019

Index base 100 in Q4 2019

Annex Figure 1.A.8. Change in real wage by industry in some non-euro area OECD countries
Copy link to Annex Figure 1.A.8. Change in real wage by industry in some non-euro area OECD countries

Percentage change in real hourly wages since Q1 2021

Annex Figure 1.A.9. Unemployment rates among young labour market entrants with and without a college degree
Copy link to Annex Figure 1.A.9. Unemployment rates among young labour market entrants with and without a college degree

Annex Figure 1.A.10. Trends in AI online job postings
Copy link to Annex Figure 1.A.10. Trends in AI online job postings

Share of job postings on Indeed containing Artificial Intelligence (AI) terms in percentage

Annex Figure 1.A.11. Relative index of new job postings in high- versus low-LLM exposure occupations in European countries
Copy link to Annex Figure 1.A.11. Relative index of new job postings in high- versus low-LLM exposure occupations in European countries

Ratio of the index of new online job postings in high LLM-exposed occupations to the index of new online job postings in low LLM-exposed occupations, four‑quarter moving averages, base 100 in Q4 2019, European countries

Notes

Copy link to Notes

← 1. This is the case in Austria, Czechia, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Lithuania, Luxembourg, Norway, the Slovak Republic, Switzerland and the United Kingdom.

← 2. It should be noted that vacancies might be overestimated, as employers tend to advertise job openings without intending to fill them, especially in specialised industries and in larger firms (“ghost jobs”) (Ng, 2024[62]).

← 3. https://ec.europa.eu/eurostat/databrowser/view/ei_bsse_q_r2__custom_16762528/default/table and https://ec.europa.eu/eurostat/databrowser/view/ei_bsin_q_r2__custom_16762125/default/table.

← 4. Nonetheless, the share of firms citing lack of labour as one of the factors limiting production are still close to peak (and much higher than before the pandemic) in the food service sectors in the EU (https://economy-finance.ec.europa.eu/economic-forecast-and-surveys/business-and-consumer-surveys/download-business-and-consumer-survey-data/subsector-data_en).

← 5. https://www.ecb.europa.eu/press/key/date/2025/html/ecb.sp250823~2f476a9a97.en.html.

← 6. It should be noted that significant production of intangible assets in these sectors could distort productivity measures.

← 7. https://www.piie.com/research/piie-charts/2019/us-china-trade-war-tariffs-date-chart?.

← 8. 24‑1 287 Learning Resources, Inc. v. Trump, 607 US _ (2026).

← 9. Recent evidence however highlights that the effective tariff rates are significantly lower than those announced, once exemptions and tariff evasion are taken into account (Gopinath and Neiman, 2026[61]).

← 10. Most of the data used in this section refer to the “wages and salaries” component of the Labour Cost Index (i.e. excluding employer’s social security contributions) produced by Eurostat – or similar measure for non-European countries (see notes to the figures for the details on the countries for which different wage measures have been used). In addition to separating wages from other labour cost components, these indicators have two main advantages relative to measures of compensation per hour worked derived from National Accounts. First, they are generally constructed to follow the evolution of hourly nominal wages for a constant industry structure, therefore minimising the potential impact of compositional changes on aggregate wage dynamics. Second, they are available at a more detailed sectoral breakdown than measures of compensation of employees from National Accounts, allowing the analysis on wage dynamics by industry of different pay levels of the next section.

← 11. Examining wage growth since Q4 2019 instead of Q1 2021 reveals positive wage changes in a greater number of countries. However, using Q1 2021 as the reference quarter is more relevant than pre‑COVID‑19 data (Q4 2019) for analysing the wage recovery from the inflation surge. Nominal wage index series since Q4 2019 show that the artificial increase in wages induced by composition effects during the pandemic had subsided by Q1 2021 in virtually all OECD countries (Annex Figure 1.A.7).

← 12. By contrast, the Danish standardised index of average earnings was only 1.3% above pre‑inflation levels in Q1 2026. This is because the labour cost index used in Figure 1.11 refers to a constant industry structure and covers all employees, while the standardised index of average earnings is adjusted for both the industrial and occupational structure and does not include employees in private‑sector firms with less than 10 employees.

← 13. Compositional shifts in the sectoral composition of the workforce are adjusted for in most countries in Figure 1.11.

← 14. In the United States the nominal federal minimum wage has remained unchanged since 2009, but state level minimum wages have risen frequently in recent times, so that the employment-weighted average of state level real minimum wages has been much been more dynamic than the federal real minimum wage.

← 15. The actual poverty gap (Figure 1.17) measures the p.p. difference between net equivalised disposable income expressed as a percentage of median equivalised disposable income, and 50% (the poverty line is defined as 50% of median equivalised disposable income). It is therefore a measure of relative poverty, reflecting the shape of the bottom of the income distribution in a given year. In contrast, by anchoring the poverty line at its 2021 level and focussing on real income, the anchored poverty line better captures developments in absolute poverty.

← 16. To enable a comparison of the dynamics between labour costs and profits, this section uses indicators from the National Accounts (see note to Figure 1.20). Using the income approach, nominal GDP can be decomposed as $P Y = N C E + G O S + T A X N$ where $P$ is the GDP deflator, $Y$ is real GDP, $N C E$ is nominal compensation of employees, $G O S$ is gross operating surplus, and $T A X N$ is nominal taxes. Unit labour costs and profits are derived by dividing the two relevant GDP components by real GDP. Unit labour costs increase when growth in compensation exceeds growth in real GDP or, dividing these two components by hours worked, when growth in compensation per hour worked exceeds growth in labour productivity. This measure of unit labour costs differs in some important respects from the measure of hourly wages based on the “wages and salaries” component of the labour cost index used in the previous sections (see note 7). Most notably, unit labour costs include employer’s social security contributions and do not control for changes in the sectoral composition of the economy.

← 17. Unit profits made an unusually large contribution to domestic price pressures during the 2021‑2022 inflation surge (OECD, 2023[3]).

← 18. https://www.ecb.europa.eu/press/pr/date/2026/html/ecb.pr260617~79dfc49802.en.html.

← 19. In Australia, however, the contribution of unit profits to domestic price pressures has remained at historically low levels since Q2 2023, after a sharp drop in that quarter.

← 20. It should be noted that this decomposition of domestic price pressures provides only a partial picture of real wage catch-up, as they exclude the distribution of productivity gains between labour and profits.

← 21. In absolute term, there has been an upward trend in the unemployment rate for both graduate and non-graduate young entrants in these three countries in recent years (Annex Figure 1.A.9).

← 22. For example, LLMs could replace simpler cognitive tasks, which are widespread within entry jobs, while more sophisticated expert tasks could be more difficult to automatise. Even if it does not automate more intensively entry-level jobs, generative AI could have a disproportionate impact on youth unemployment because any general negative employment shock is likely to have a stronger impact on young people by making it more difficult for them to enter the labour market.

← 23. Humlum and Vestergaard (2025[45]) focusses on Denmark and Klein Teeselink (2025[48]) on the United Kingdom.

← 24. Occupational exposure to language models is an index developed by Felten, Raj and Seamans (2023[54]) and made available in ISCO classification by Nurski (2025[55]). It measures the extent to which occupations are exposed to advances in AI language modelling capabilities.

← 25. A LinkedIn member is considered AI talent if they have explicitly added at least two AI skills to their profile and/or they are or have been employed in an AI job. LinkedIn categorises AI skills into 2 mutually exclusive groups: “AI Engineering” and “AI Literacy” skills, where, broadly, AI Engineering skills refer to the technical expertise and practical competencies required to design, develop, deploy, and maintain artificial intelligence systems, and AI Literacy skills refer to the knowledge, abilities, and critical thinking competencies needed to understand, evaluate, and effectively interact with AI technologies. The top skills that comprise the “AI engineering” skill grouping are machine learning, natural language processing, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, and OpenCV, among others. The skills that comprise the “AI literacy” grouping are prompt engineering, GPT‑3, GPT‑4, stable diffusion, ChatGPT, Github copilot, generative art, Dall-E, Midjourney and Google Bard. An AI job is an occupation that requires AI skills to perform the job (https://oecd.ai/en/linkedin).

← 26. The time lag between the release of ChatGPT (November 2022) and its use by firms might be explained by the later release dates of the OpenAI API – a prerequisite for building custom applications – and ChatGPT Enterprise – offering data privacy and security assurances for corporations (Curto Millet and Iscenko, 2026[52]). The former was released in March 2023, and the latter in August 2023.

← 27. These trends are confirmed by AI job postings from the online platform Indeed (Annex Figure 1.A.10). AI-related postings in Indeed data are identified using keyword searches covering both AI occupations (e.g. “Artificial Intelligence”, “Machine Learning”, “Data Science”) and generative AI terms (e.g. “Generative AI”, “Large Language Models”, “ChatGPT”).

← 28. Annex Figure 1.A.11 shows that this finding holds for the vast majority of the 15 EU countries analysed taken separately.

← 29. The hiring rate fell from about 4.5% at its 2022 peak to less than 3.5% in 2025 (https://www.apricitas.io/p/the-no-hire-economy).

← 30. https://news.crunchbase.com/startups/tech-layoffs/.

← 31. https://www.economist.com/finance-and-economics/2025/11/06/dont-blame-ai-for-your-job-woes.

← 32. LinkedIn global data for 2022-2023 shows that the number of applications for remote positions was increasing, while the number of remote job postings was declining (https://business.linkedin.com/content/dam/me/business/en-us/talent-solutions/resources/pdfs/future-of-recruiting-2024.pdf).

← 33. https://www150.statcan.gc.ca/n1/daily-quotidien/241217/dq241217c-eng.htm and https://www150.statcan.‌‌‌gc.ca/n1/daily-quotidien/250924/dq250924a-eng.htm.‌‌

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & events

News & events

Featured events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment