Is there really a global decline in the (non-housing) labour share?

Germán Gutiérrez and Sophie Piton

Much has been written on the global decline of the corporate labour share (defined as the share of corporate value added going to wages, salaries and benefits). The IMF and OECD worry about this trend, linking it to decreasing wages and rising inequality. And economists are hard at work looking for an explanation: prominent hypotheses range from automation and ‘superstar’ firms to offshoring. But is there really a global decline in the non-housing/business labour share? Not if you properly exclude housing income and account for self-employment, as described in a recent Staff Working Paper. Adjusting for housing and self-employment, labour shares have remained stable across most advanced economies except in the US, where the labour share still declines by 6% since 1980 (Figure 1).

Figure 1: Gross domestic labour share, Europe and United States, 1970-2015

Source: Gutiérrez & Piton (2019).

Measuring the labour share: harder than it looks

In principle, measuring the labour share should be straightforward. In practice, however, it is harder than it looks. Total economy labour shares require a difficult estimation of wages for the self-employed. The series are also affected by rising capital depreciation; and conflate two very different dynamics: those of the housing and non-housing (i.e., business) sectors (see Rognlie, 2015).

In response to these challenges, researchers have focused on corporate sector labour shares as a “common way to deal with… measurement difficulties… including ambiguity in the labour/capital split of mixed income [i.e., self-employment], as well as the crucial role of housing.” Matthew Rognlie summarizes the prevailing view, calling the net labour share of the corporate sector “the single best measure” of the labour share [for the US].

But is the corporate labour share truly free of these measurement challenges? Only in the US, where the corporate sector includes only legally organized corporations required to file corporate tax returns. Outside the US, most countries follow the 2008 system of national accounts (SNA), which includes “all units engaged in market production that act independently of their owners” in the corporate sector. This includes legally constituted corporations (as in the US), in addition to cooperatives, limited liability partnerships, notional resident units and quasi-corporations (i.e., unincorporated enterprises owned by households or governments for which detailed accounts are available). The broader definition of corporations reintroduces two old measurement challenges – but this time for the corporate sector.

First, notional resident units and quasi-corporations often own and operate housing. In fact, nearly 20% of the capital stock of non-financial corporations in Europe is in housing, reaching 30% for France. The US is the outlier, with only a 1% housing share. The corresponding rental income is included in corporate sector value added, which introduces a downward bias in corporate labour shares for the same reasons emphasized by Rognlie: housing has a low labor share relative to the corporate sector (~6% vs. ~66%, respectively), so its growth pushes the aggregate labor share downward. This is important for understanding the distribution of income in the economy, but does not speak to the allocation of business output between owners and workers.

Second, partnerships, cooperatives and quasi-corporations often include self-employed workers. This is the case in the UK, where there is an increasing number of `incorporated’ self-employed as documented by the UK’s Office for National Statistics. This is also true in countries with a vast network of small and medium enterprises, such as Italy and Germany. In Italy, for example, quasi-corporations include unlimited partnerships and sole proprietorship provided they have more than five employees and produce a complete set of accounts. This includes artisans, farmers, small businessmen and other self-employed persons. As a result, 15% of hours worked in the Italian corporate sector are by self-employed workers, and one-fourth of total hours worked by self-employed are in the corporate sector. Labour compensation for these self-employed workers is included in corporate gross operating surplus (GOS), just as wages for self-employed workers in the household sector are included in household mixed income. Mixed income rears its ugly head again – albeit in a different sector and with a different name. An assumption is again needed to separate labour from capital income, else the corporate labour share is underestimated.

Two `harmonized’ measures of the non-housing labour share

Both measurement challenges have become more important over time, biasing corporate labour shares downwards. In the recent Staff Working Paper, we propose two methods to correct for these measurement challenges and obtain harmonized non-housing labour share series. The first measure (solid red line in Figure 2) focuses on market activities while excluding all real estate activities –from both wages and value added– using industry accounts. This measure covers corporate and noncorporate activity and fully controls for housing. But it has two limitations: (i) it `over-controls’ for real estate by excluding commercial as well as residential real estate and (ii) it still relies on imputed wages for the self-employed, which are difficult to estimate. Our second measure (dotted red line in Figure 2) mitigates these limitations by focusing on the corporate sector. The corporate sector excludes self-employment in some but not all countries. We use national account data to estimate the contribution of housing to corporate value added and estimate wages for the self-employed where possible and relevant. Data limitations restrict the sample for the second measure but where available, it behaves similar to the labour share excluding real estate.

Figure 2 compares our harmonized measures of the labour share against the raw corporate sector labour shares from Karabarbounis and Neiman (2014) (extended using OECD data). Black lines report raw series and red lines report harmonized ones.

Consistent with the exclusion of housing and self-employment from the US corporate sector, all measures behave similarly in the US. They exhibit a ~6% decline from 1980 to 2015, concentrated in the post-2000 period. This confirms the well-known decline of the US labour share.

By contrast, the series evolve quite differently outside the US. Consistent with the inclusion of housing services and self-employment in non-US corporate sectors, the harmonized series are much larger and far more stable than the raw ones. The adjusted corporate sector series –available only since 1995 – behave like the adjusted business series. The levels differ due to different weights/industry mixes but the trends are quite consistent. The adjusted business series are higher in 2015 than in 1970 in all major European economies except for France, where our data starts at the historical peak. It is also higher for the EU4, EU14 and EU27 aggregates (unreported); and only slightly lower for the Global series excluding the US.

The paper then looks within regions, across industries, to show that manufacturing explains most of the decline in the US labour share. While manufacturing as a share of value added declined in both the US and Europe (by ~10% since 1977), the US manufacturing labour share fell by 21%, compared to only 1.5% in Europe.

Revisiting the role of technological change on labour shares

Our results challenge the common wisdom of a global decline in the non-housing labour share. They also cast doubt on most technological explanations for these trends: including declining capital prices, automation, import competition and intangibles – which, at least as emphasized so far, should have similar effects across countries and industries.

Determining why the labour share declined in US manufacturing, but not elsewhere is an important area of future research. Perhaps declining competition has led to rising profits in selected US industries. Or perhaps the mechanisms emphasized so far had different outcomes across regions and industries.

 Figure 2: Global gross domestic labour share, 1950-2015, in %

Source: Gutiérrez & Piton (2019).

Germán Gutiérrez works at the NYU Stern School of Business and Sophie Piton works the Bank’s Structural Economics Division.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied.Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.


3 thoughts on “Is there really a global decline in the (non-housing) labour share?

  1. Excellent, intriguing analysis. Two things:

    1 How does this compare to the productivity-wage gap — that is, the generally accepted theory that output per worker has been rising since the 1970s but median wages adjusted for inflation have been relatively stagnant (

    2 If this analysis does not disprove that “gap”, is it because of the difference between mean and median wages? Piketty and others have discussed the dramatic rise (esp. in US and UK but also in other developed economies) of “super-managers” who earn obscene salaries which inflate the average wage numbers. If that is the case, the UK and the US would look even worse in the above graphs, while France might look about the same. Thoughts?

    – International Politics & Economics Master’s Student, Paris

  2. I think your supposition in point 2 (following from point 1) is spot on. Conceptually, the productivity wage gap can arise with a constant labour share if there is more wage inequality. In that case, more of the productivity gains are still going to Labor but not to the Labour in the middle.

    For the UK case, Van Reenen is good on this – short version ( and long version (

  3. Excellent analysis from Van Reenen, thank you @Jeevun for the reference. Interesting to think about the nuances of this argument; is the “productivity-wage gap” the right way to talk about inequality?

Comments are closed.