OECD Employment Outlook 2023

United States

Labour markets have been resilient despite the significant slow-down in economic activity

The labour market recovery from the COVID‑19 recession has been strong, but lost momentum in 2022 and early 2023 in the context of the economic slowdown. However, employment and unemployment have held their ground, and job vacancy rates remain high in most countries, despite some signs of easing. By May 2023, the OECD unemployment rate had fallen to 4.8%, a level not seen in decades.

• In the United States, strong labour demand has persisted despite the economic slowdown. The unemployment rate fell to 3.4% in January 2023 before rising slightly to 3.6% in June 2023. Concerns over labour supply, which dipped during the pandemic, have been alleviated as well. Recent trends indicate that labour force participation for persons aged 15‑64 is slightly above its pre‑pandemic level at 74.7% as of June 2023 (from 74.4% in December 2019). Buoyant job openings have also moderated in past months, signaling better alignment between labour demand and supply.

• To attract the best talent, employers increasingly offer discretionary benefits. Between December 2019 and December 2022, US online job advertisements contained increased mentions of health-related benefits, retirement schemes (e.g. 401(k) retirement accounts), paid time off and tuition assistance. Such increases were not unique to industries that tend to offer high-paying jobs, such as finance and insurance, but were also seen in industries that tend to offer low-paying jobs, such as accommodation and food services and administrative and support services.

• Real GDP growth in the United States is projected to slow as monetary policy further tightens in response to inflationary pressures. As demand slows, employment is expected to fall and the unemployment rate is projected to gradually rise to 4.5% by Q4 2024 (OECD Economic Outlook 2023).

• In recent labour market policy developments, the Inflation Reduction Act (IRA) of 2022 increased public expenditure for state‑based programs to train and educate workers spcialised in energy efficiency. Included in the IRA is over USD 200 million in state grants to fund training programs for the installation energy efficiency enhancements and conducting energy audits in commercial and residential buildings.

Figure 1. Unemployment rates remain low across the OECD

Unemployment rate (percentage of labour force), seasonally adjusted

Note: Latest month available refers to March 2023 for the United Kingdom; and June 2023 for Canada and the United States.

Source: OECD (2023), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 11 July 2023).

Real wages are falling amid a cost-of-living crisis

Although showing signs of slowing, inflation has reached unprecedented levels in many countries, which were not matched by nominal wage growth. Consequently, real wages have fallen in virtually every OECD country. On average, real wages were down 3.8% in Q1 2023 year-on-year among the 34 OECD countries with data available. The loss of purchasing power is particularly challenging for low-income households who have less capacity to deal with increases in prices through savings or borrowing.

• In the United States, real wages declined more modestly compared to the OECD average, by 0.7% between Q1 2022 and Q1 2023 compared to 3.8% over the same horizon across the OECD.

• The inflation squeeze makes it more difficult for households to meet usual expenses, particularly Black and Latino households and low-income families. For families with annual incomes of less than USD 50 000, about 60% of Black and Latino families reported difficulty in meeting expenses compared to nearly 52% of low-income white households. However, data show that workers with low education or in low-pay occupations were the only ones that avoided a real wage contraction between Q4 2019 and Q4 2022.

• With the unemployment rate historically low, nominal wage growth remains strong. Average hourly earnings were up 4.4% year-on-year in June 2023, even with the year-on-year growth rate seen in April 2023.

Profits increased more than labour costs in many countries

In most countries, profits have increased more than labour costs, making a large contribution to price pressures, and leading to a fall in the labour share of income. Data from Europe and Australia indicate that this is not limited to the energy sector. This indicates room for profits to absorb some future wage increases, but this might be less the case in small and medium firms that are more squeezed by other cost increases.

• In the United States, profits increased more than labour costs as the economy reopened from the pandemic, driven by supply shortages and consumers’ eagerness to spend. However, more recently, the pace of labour cost increases has picked up to match profit increases. Between Q4 2019 and Q1 2023, unit labour costs increased by 14.2% and unit profits by 13.5% (compared to 15.6% and 21.0%, respectively, across OECD countries). This near parity may foreshadow smaller wage gains over the coming quarters as firms seek to maintain profit margins.

Figure 2. Profits have increased more than labour costs in many countries

Percentage changes, seasonally adjusted, from Q4 2019 to Q4 2022

Note: OECD is an unweighted average of 29 OECD Member States (excluding Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus by real GDP, respectively. Unit labour costs and unit profits are components of the GDP deflator, which captures the price changes of all goods and services produced in an economy.

Source: OECD Employment Outlook 2023, Chapter 1.

Minimum wages and collective bargaining can support low-paid workers

Minimum wages and collective bargaining can help mitigate losses in purchasing power and ensure a fair distribution of the cost of inflation between firms and workers, while avoiding a price‑wage spiral. On average across the OECD, statutory minimum wages have increased significantly over the last two years, and this has allowed real minimum wages to keep up with inflation better than average wages. The evidence suggests that there is room for profits to absorb further increases in wages, at least for low-paid workers. Governments can also provide direct support through the tax and benefit system to raise the net income of low-income households.

• In the United States, as of May 2023 the average nominal minimum wage in effect in the various states grew 16.7% relative to December 2023 while the real weighted minimum wage was largely unchanged over the same period. While the federal minimum wage has not been increased since 2009, many states and localities have raised their own minimum wages.

Figure 3. Minimum wages have kept pace with inflation

Cumulative percentage change in nominal and real minimum wages since December 2020, May 2023

Note: OECD is the unweighted average of 30 OECD countries with a statutory minimum wage. W: employee‑weighted subnational minimum wages. Fed: Federal minimum wage.

Source: OECD Employment Outlook 2023, Chapter 1.

AI has so far helped high-skilled workers perform their jobs rather than displacing them, but employment effects could take time to materialise

So far, there is little evidence of decreased labour demand due to AI. High-skilled occupations have been the most exposed to recent progress in AI, but they have also experienced employment gains relative to lower-skilled workers. However, AI adoption is still relatively low and the technology is evolving rapidly including recent advances in generative AI. Any negative employment effects may therefore take time to materialise.

• AI appears to complement the skills of more exposed, high-skill occupations. When considering all automation technologies including AI, the occupations at the highest risk of automation are typically lower-skilled and held by younger workers. 21.2% of employment in the United States is in occupations at highest risk of automation compared to an OECD average of 27%. This reflects the fact that AI is just one of many automation technologies (ICT, robotics etc.).

• AI regulation has been advancing at both the federal and state levels, with generative AI attracting specific calls for action. In Congress, Senate Majority Leader Chuck Schumer launched the “SAFE Innovation Framework” in June 2023, which outlines principles describing what AI policies should achieve and a legislative approach for translating the framework into action. At the presidential level, the Biden Administration introduced a blueprint for an AI Bill of Rights in October 2022, which calls for developers to ensure “safe and effective” systems that do not discriminate or violate expectations of privacy and that explain when a user is engaging with an automated system.

• Ten states have enacted AI legislation while proposals are underway in thirteen. Some enacted measures restrict the use of AI in hiring processes (Illinois, Maryland, New York City) while others ensure data protection in contexts of automated decision-making (California, Connecticut, Colorado, Indiana, Montana, Tennessee, Virginia). Proposed legislation calls for algorithmic impact assessments (California, Connecticut, the District of Columbia, Indiana, Kentucky, New York, Vermont and Washington), data protection (Oregon), restricted use of automated decision systems (South Carolina) and the formation of a state‑level Artifical Intelligence Registry (Pennslyvania).

• On generative AI, a US-EU Joint Statement of the Trade and Technology Council in May 2023 announced that it would be added to work on the Roadmap for Trustworthy AI and Risk Management.

Figure 4. Although AI exposure has not (yet) led to decreased labour demand, certain groups remain at risk

Share of employment in occupations at the highest risk of automation by country, 2019

Note: The figure shows the employment shares of the occupations at highest risk of occupations, i.e. top quartile occupations when ranked according to their automation index. The results are based on a survey of experts who evaluated the degree of automatability for 98 skills and abilities. The risk of automation measure is then computed by occupation as the average rating for each skill or ability used in the occupation across all expert responses weighted by the skills or abilities’ importance in the occupation as rated by O*NET.

Source: Lassebie and Quintini (2022), “What skills and abilities can automation technologies replicate and what does it mean for workers?: New evidence”, https://doi.org/10.1787/646aad77-en, based on OECD Expert Survey on Skills and Abilities Automatability and O*NET.