OECD Compendium of Productivity Indicators 2023

4. Industry contributions to aggregate labour productivity growth

Context and key findings

Industries differ from each other in terms of their labour productivity level (defined as the ratio of value added to hours worked) and labour productivity growth. Such differences relate, for instance, to the intensity with which industries use skilled labour and capital in their production process, their capacity to innovate, and their exposure to competition and knowledge sharing through their participation in international trade and global value chains.

Decomposing aggregate labour productivity growth into the effects of within-industry labour productivity growth and reallocations between industries with different productivity levels and growth rates helps to better understand the drivers of overall labour productivity growth, especially in 2020-2021.

Key findings

• Under normal economic circumstances, reallocations across industries only play a limited role in explaining aggregate labour productivity growth. Overall reallocations are the sum of reallocations of hours worked between industries with different productivity levels (static reallocation effect) and between industries with different productivity growth rates (dynamic reallocation effect). The static effect usually dominates the dynamic effect.

• A more significant and positive reallocation effect on labour productivity growth is usually observed during recessions. The COVID-19 recession was no exception and the scale of the overall reallocation effect in 2020 was unprecedented. It largely contributed to the rebound in aggregate labour productivity growth that was observed in most countries in 2020. Nevertheless, early evidence from the few countries for which detailed industry data are available for 2021 shows that the effect of reallocations observed in 2020 was largely related to temporary disruptions brought by the pandemic.

• In the future, major structural changes induced by the digitalisation and the decarbonisation of economies may lead to more significant reallocation effects across industries.

Indicators

 

Compare

Reallocations across industries and within-industry productivity developments

Except during recessions, reallocations across industries tend to play a limited role in aggregate labour productivity developments

• The limited overall reallocation effect during normal economic circumstances is often the result of industry contributions with opposite signs (see Figure 4.2 below). For example, the opposite effects of declining hours worked in agriculture and manufacturing are related to the fact that agriculture has a below-average and manufacturing an above-average productivity level in most countries. These two effects tend to compensate each other.

• The unusually large reallocation effect observed in 2020 was largely due to the lockdown of contact-intensive industries such as accommodation and personal services or travel bans affecting the transport sector during the pandemic. These industries tend to have below-average labour productivity levels, explaining why the sharper decline in hours worked in these industries than in other parts of the economy acted as an overall boost to productivity. As a result, the overall reallocation effect was the main source of aggregate labour productivity growth in 2020 in most countries for which data is available. It was larger in countries that implemented more stringent lockdown measures in 2020.

• Less contact-intensive industries experienced a lower decline in hours worked, which had an heterogenous impact on aggregate labour productivity growth in 2020. For example, the limited decline in hours worked in low-productivity sectors such as agriculture and construction acted as a drag on aggregate labour productivity growth in some countries (e.g. France, Greece and the United States). By contrast, the resilience of hours worked in high-productivity industries such as information and communication, and finance and insurance, likely related to the rapid development of teleworking in these sectors, boosted aggregate labour productivity growth.

• Early evidence from the few countries for which detailed industry data are available for 2021 shows that the effect of reallocations observed in 2020 was largely related to temporary disruptions brought by the pandemic. In most countries, the allocation across sectors started to revert to its pre-pandemic level in 2021. The large positive effect on aggregate productivity of the shutdown of low-productive sectors such as accommodation and restauration in 2020 vanished or even turned negative in 2021, depending on the duration of lockdowns. In Austria, Italy and Slovenia, the overall reallocation effect went back to small values, while in the United States it reached a low negative value that nearly cancelled the large positive effect that was observed in 2020.

• Considering the rapid increase in job quit rates following the first lockdowns in different countries, a phenomenon coined as the “Great Resignation”, it was initially thought that the pandemic may be inducing a durable change in worker preferences for certain types of jobs. Nevertheless, recent empirical evidence shows that quit rates were highest in industries and occupations that also had the highest job growth (Hobijn 2022). These industries and occupations are those that were the hardest hit by lockdown measures during the pandemic, e.g. accommodation and food services. Moreover, only a small share of job quitters in the United States seems to be changing industry of employment or occupation, which goes against the initial view that the wave of resignations is largely driven by workers who are reconsidering their career choices. Additional data for 2021 and the subsequent years will allow to fully assess the effect of reallocations induced by the pandemic in the long term.

• In the future, major structural changes induced by the digitalisation and the decarbonisation of economies may lead to more significant reallocation effects across industries.

Indicators

 

Compare

Over the past decades, within-industry developments explain most of aggregate labour productivity growth

• Within-industry developments explain most of the aggregate labour productivity growth in OECD countries, including the surge in labour productivity growth from the mid-1990s until the mid-2000s and the slowdown in labour productivity growth since then. The overall contribution of within-industry developments can be further broken down into industry contributions (see Figure 4.3 below).

• As explained by Jorgenson et al. (2008) and Cette et al. (2016), Information and Communication Technology (ICT) is behind the surge in aggregate labour productivity growth at the turn of the 21st century. This includes productivity gains in ICT-producing sectors (manufacturing of computers, electronic and electrical equipment, and provision of ICT services), as well as the increased use of ICT by some services sectors (e.g. trade; finance and insurance; and business services). The surge in ICT-related labour productivity gains that started in the mid-1990s was more significant and durable in the United States than in Europe and Japan, partly due to higher ICT investments and, more importantly, to higher MFP gains in ICT-using industries (Timmer et al., 2011; Gordon and Sayed, 2020). It is important to note that labour productivity growth reflects both capital accumulation and MFP growth (see the section Labour productivity growth: contributions from capital and multifactor productivity in Chapter 3).

• The years following the ICT-related productivity boom have seen a decline in aggregate labour productivity growth. In the United States, it can be largely attributed to a decline in the contribution of ICT-producing and ICT-using industries (e.g. trade, finance and insurance, and business services) to within-industry labour productivity growth from the mid-2000s onwards. Among G7 countries, a similar phenomenon took place in the United Kingdom, with an even larger role played by finance. In France and Germany, the contribution of ICT-producing and ICT-using industries is less visible, and the decline in within-industry labour productivity growth between the early 2000s and the 2010s seems to be more related to manufacturing. Labour productivity developments within the manufacturing sector also largely contributed to the decline in aggregate labour productivity growth from the early 2000s onwards in Finland and Mexico.

• The within-industry contribution to labour productivity growth in 2020 declined in most countries for which data is available, as compared to the previous decade (2010-2019). It even turned negative in the Czech Republic, Denmark, Finland, France, Germany, Greece, Mexico, and the Netherlands. By contrast, this contribution was higher in 2020 than over 2010-2019 in Austria, Italy, and the United States. In most countries, transport, accommodation and personal services contributed negatively to within-industry productivity growth in 2020, while trade, ICT, finance and insurance, and business services contributed positively. Countries mainly differed in the relative magnitude of these negative and positive contributions.

• In the few countries for which detailed industry data are available for 2021, the within-industry contribution to labour productivity growth usually had an opposite sign in 2021 and 2020, thus bringing back the average contribution over 2019-2021 closer to the value recorded over 2010-2019. The only exception is the United States, where this contribution is both higher in 2020 than over 2010-2019, and higher in 2021 than in 2020. In the United States, all negative contributions turned positive and most positive contributions were maintained or even increased between 2020 and 2021.

• While aggregate macroeconomic data sourced from national accounts are insufficient to fully understand the unusual evolutions of within-industry contributions to labour productivity growth in 2020, they are consistent with a temporary reduction in hours worked during the pandemic, driven by furlough programmes or labour hoarding decisions by firms which were confronted with temporary declines in output. Large-scale firm closures in 2020 are unlikely to explain within-industry developments, as bankruptcies declined at the start of the pandemic (OECD, 2021). Nevertheless, firm-level data would be necessary to go beyond the evidence provided by national accounts data and fully understand within-industry labour productivity developments.

Indicators

 

Compare

How to read the indicators

The decomposition of aggregate labour productivity growth that is used in this chapter includes three main terms, each of them corresponding to a sum of industry contributions:

• A within-industry effect, where labour productivity growth in each industry is weighted by the industry share in total value added in year t-1.

• A static reallocation effect, accounting for changes between t-1 and t in the share of total hours worked of industries with different productivity levels. Industries with an increasing share in total hours worked contribute positively to aggregate labour productivity growth if they have an above-average labour productivity level$.$

• A dynamic reallocation effect, accounting for changes between t-1 and t in the share of total hours worked of industries with different productivity growth rates. An increase in the total hours worked share of industries with positive productivity growth has a positive effect on aggregate labour productivity growth. This effect is all the more significant that the industry value added is high.

The sum of the two reallocation effects is referred to as overall reallocation effect.

For additional information on this decomposition of aggregate labour productivity growth, see the methodological note.

This chapter focuses on a subset of the total economy that excludes real-estate, public administration and defense, education, and health activities (i.e. total economy less industries L, O, P, Q in the ISIC rev. 4 classification). Real-estate activities are excluded because their value added is largely imputed (it includes the value of both actual and imputed housing rents in the economy) and disproportionate as compared to the corresponding work force in national accounts (mostly real-estate agents and employees of notary offices are attached to the real-estate industry in national accounts). Public administration and defense, education and health services are excluded because they are largely non-market. Hence, their output value is measured as the sum of input costs, and in several countries their output volume is measured by deflating input costs, thus conventionally excluding any productivity gains.

For most countries, the above decompositions of aggregate labour productivity growth rely on breakdowns of value added and hours worked into economic sectors at the 2-digit level of the NAICS 2017 classification (for Canada, Mexico, and the United States) or the ISIC rev. 4 classification. Due to data limitations, the decompositions for France, Germany, Italy, and the United Kingdom rely on a mix of 1-digit and 2-digit level data corresponding to the A38 level of the ISIC rev.4 classification, and the decomposition for Australia relies on 1-digit level data according to the ANZSIC 2006 classification. This corresponds to between 20 and 64 industries, depending on data availability in different countries. Except for a few cases, this is the most granular industry data publicly available in national accounts, but it might not be sufficient to fully capture the heterogeneity of economic activities. Therefore, it cannot be excluded that part of the within-industry effects presented above actually correspond to resource reallocations between firms or economic activities belonging to the same 2-digit industry. A complete assessment of the contribution of reallocations and business dynamism (entries and exits of firms) to aggregate labour productivity growth would require firm-level data.

Even though a more granular breakdown is used for all calculations, the following breakdown by industry is used to visualise the contributions of reallocation and within-industry effects to aggregate labour productivity growth in the figures included in this Chapter:

• Agriculture and mining: industries A and B in the ISIC rev. 4 classification; industries 11 and 21 in the NAICS 2017 classification

• Manufacturing and utilities, excluding manufacturing of ICT: industries C, D and E except C26-27 in the ISIC rev. 4 classification; industries 22 and 31-33 except 3361MV and 3364OT in the NAICS 2017classification

• Construction: industry F in the ISIC rev. 4 classification; industries 23 in the NAICS 2017 classification

• Trade: industry G in the ISIC rev. 4 classification; industries 42 and 44RT in the NAICS 2017 classification

• Transport, accommodation, and personal services: industries H, I and R to U in the ISIC rev. 4 classification; industries 48TW, 71, 72 and 81 in the NAICS 2017 classification

• ICT production: industries C26-27 and J in the ISIC rev. 4 classification; industries 3361MV, 3364OT and 51 in the NAICS 2017 classification. This includes the manufacturing of computers, electronic and electrical equipment, as well as the production of information and communication services.

• Finance and insurance: industry K in the ISIC rev. 4 classification; industry 52 in the NAICS 2017 classification

• Business services: industries M and N in the ISIC rev. 4 classification; industries 54 to 56 in the NAICS 2017 classification.

It should be kept in mind that macroeconomic data for recent years are subject to revisions, especially in the years 2020 and 2021 covering the COVID-19 pandemic.

Data sources

OECD National Accounts Statistics (database), http://dx.doi.org/10.1787/na-data-en.

References and further reading

Cette G., J.G. Fernald and B. Mojon (2016), “The Pre-Great Recession Slowdown in Productivity”, European Economic Review, Vol. 88, pp. 3-20.

Gordon R.J. and H. Sayed (2020), “Transatlantic Technologies: The Role of ICT in the Evolution of U.S. and European Productivity Growth”, NBER Working Paper No. 27425.

Hobijn B. (2022), ““Great Resignations” are Common during Fast Recoveries”, FRBSF Economic Letter No. 2022-08. https://www.frbsf.org/economic-research/publications/economic-letter/2022/april/great-resignations-are-common-during-fast-recoveries/

Jorgenson D.W., M.S. Ho and K.J. Stiroh (2008), “A Retrospective Look at the U.S. Productivity Growth Resurgence”, Journal of Economic Perspectives, Vol. 22(1), pp. 3-24

OECD (2021), OECD SME and Entrepreneurship Outlook 2021, OECD Publishing, Paris, https://doi.org/10.1787/97a5bbfe-en.

Timmer M.P., R. Inklaar, M. O’Mahony and B. van Ark (2011), “Productivity and Economic Growth in Europe: A comparative Industry Perspective”, International Productivity Monitor, Vol. 21, pp. 3-23.