Labour income is the most direct mechanism through which the benefits of productivity gains, and thus economic growth, are transferred to workers. Indeed, employers’ ability to raise wages and other forms of labour income depends on increases in labour productivity, highlighting the welfare implications of productivity growth and its role as the main driver of long-term living standards. However, the empirical evidence points to a decoupling of labour productivity growth from real labour income growth in the majority (two thirds) of OECD countries since the mid-1990s. The decline in the labour income share (i.e. the share of labour income in gross value added) observed in these countries precisely reflects the decoupling between real labour income and labour productivity growth.
OECD Compendium of Productivity Indicators 2023
7. Labour income and productivity
Context
Labour income shares and the decoupling between labour income and productivity
The decline in labour income shares observed since the mid-1990s in the majority of OECD countries can be reformulated as a decoupling between labour productivity and real labour income growth, when nominal labour income is adjusted for inflation using the same price index applied to deflate value added (and hence productivity).
The developments in the total economy labour income share may be partly driven by specific industries for which there are significant conceptual and measurement issues. For example, the value added of the real-estate sector includes all (actual and imputed) housing rents in an economy, whereas the corresponding labour income is only related to the people working in the real-estate sector. Therefore, the labour share in the real-estate sector is well below the labour share of the total economy and does not reflect the labour market mechanisms connecting labour income to productivity. Moreover, housing rent developments can induce large fluctuations in total-economy labour shares when the real-estate sector constitutes a large component of overall economic activity. Similarly, for countries with large primary (i.e. agricultural or mining) sectors, developments in total-economy labour shares may be largely driven by fluctuations in commodity prices. For example, when commodity prices increase, aggregate profits rise without commensurate increases in aggregate wages. Lastly, national accounting conventions in the non-market sector (e.g. education, health, and public administration) may bias labour share developments. Indeed, value added in the non-market sector is measured as the sum of labour compensation and capital consumption, which artificially limits variation in its labour share over time. For these reasons, it is useful to look at labour income share developments and at the decoupling between labour income and productivity after excluding primary, real estate, and non-market sectors (Schwellnus et al., 2017).
The impact on material well-being of the decoupling between labour income and productivity is further exacerbated by the widespread slowdown in productivity growth, and in some countries even more so when real labour income is adjusted for inflation using the consumer price index (CPI). This is because inflation based on value added or consumer prices can differ significantly, reflecting the effect of terms of trade. Indeed, the value-added deflator reflects movements in the prices of goods and services domestically produced, whereas the CPI captures movements in the prices of goods and services consumed within the economy, either imported or domestically produced.
Note that labour income shares and comparisons between average labour income and productivity developments in this chapter do not account for labour income inequalities across workers. The majority of OECD countries have experienced a further dissociating between median and average labour income since the mid-1990s, which is related to disproportionate labour income growth at the top of the income distribution (Bivens and Mishels, 2015; Schwellnus et al., 2017).
Key findings
Real average labour income has failed to keep up with labour productivity growth since the mid-1990s in around two thirds of OECD countries. This has occurred in addition to the widespread slowdown in labour productivity growth observed over the past decades, which has further undermined the increase in real average labour income.
The exclusion of the primary, real estate, and non-market sectors, where labour share developments are largely driven by commodity and asset price developments and national accounting conventions, reduces the decline in labour income shares on average across G7 and OECD countries since the mid-1990s. Looking at individual country results, this finding holds true for more than half of the countries for which data are available.
In around half of the countries for which data are available, the decoupling of real labour income growth from productivity growth is further exacerbated when real labour compensation is adjusted for inflation using the consumer price index (CPI) – i.e. from a consumer/worker perspective.
Large swings in output, hours worked and labour compensation in 2020 and 2021 reduced the decoupling between growth in real average labour income and labour productivity growth in some countries, including. Japan, Luxembourg, the Netherlands, Portugal and Spain, and intensified it in others, including Germany and the United States.
Indicators
How to read the indicators
Labour productivity in this chapter is defined as the ratio of real value added at factor cost to total hours worked, while average labour income is defined as the ratio of total labour compensation to total hours worked. Total labour compensation is computed as the sum of the compensation of employees and self-employed workers. The labour compensation received by employees includes remunerations in cash and in kind and employees’ and employers’ social contributions. It is readily available in the national accounts. However, the labour income received by self-employed is recorded in national accounts as mixed income, which bundles both their labour and capital income. Therefore, the labour compensation received by self-employed needs to be imputed. From a theoretical point of view, the best procedure would be to make this imputation based on micro-level information on the socio-economic status of self-employed workers, but the corresponding data are not readily available for most countries. Therefore, following Schwellnus et al. (2017), it is assumed that the hourly compensation of self-employed workers is equal to the hourly compensation received by employees at the level of each individual industry. For a few countries, hourly compensation received by employees by industry is not available. In such cases, compensation per employee is used.
The total business economy excluding primary and real estate activities includes the ISIC Rev.4 industry codes C to N, excluding L, plus R and S. However, for Israel, Japan, Korea, New Zealand the data includes the ISIC Rev.4 industry codes B to N, excluding L, plus R to U. For Switzerland, in the absence of information by industry, total labour compensation is compiled using compensation of employees, and hours worked for the total economy.
Data sources
OECD National Accounts Statistics (database), http://dx.doi.org/10.1787/na-data-en.
OECD Productivity Statistics (database), http://dx.doi.org/10.1787/pdtvy-data-en.
OECD STAN Structural Analysis Statistics (database), https://doi.org/10.1787/data-00649-en.
References and further reading
Bivens J. and L. Mishels (2015), “Understanding the Historic Divergence Between Productivity and a Typical Worker’s Pay”, EPI Briefing Papers, No. 406, Economic Policy Institute, Washington.
Cho, T., S. Hwang and P. Schreyer (2017), "Has the Labour Share Declined? It Depends", OECD Statistics Working Papers, No. 2017/01, OECD Publishing, Paris, http://dx.doi.org/10.1787/2dcfc715-en.
Schwellnus, C., A. Kappeler and P. Pionnier (2017), "Decoupling of wages from productivity: Macro-level facts", OECD Economics Department Working Papers, No. 1373, OECD Publishing, Paris, http://dx.doi.org/10.1787/d4764493-en.