The share of US gross value added in the nonfarm business sector that was paid out to workers as wage (or self-employment) income remained stable between 63% and 65% for more than a century. Kaldor (1961) tagged this constancy of the labour share as the first of his famous stylised facts about economic growth. Earlier, Keynes (1939: 48) had cited the “stability of the proportion of the national dividend accruing to labour, irrespective apparently of the level of output as a whole and of the phase of the trade cycle” as being “one of the most surprising, yet best-established, facts in the whole range of economic statistics”. And then, in about 2000, or perhaps a bit earlier, the share began to drop. It reached a low of about 56% in 2013, before recovering slightly to about 58% in 2020. A fall of similar magnitude has been observed in many other countries (Karabarbounis and Neiman 2014). The obvious question is: why?
This question has fascinated economists in recent years. Searching the Google Scholar website for the joint appearance of the phrase ‘labour share’ and the word ‘decline’ generates a list of more than 12,000 books, articles, and working papers in the last decade alone. Despite this surfeit of attention, or perhaps as a result of it, the profession is not close to a consensus answer. As we have outlined in our review article (Grossman and Oberfield 2022), the various attributions that have been offered by the literature seem to ‘explain’ the phenomenon many times over.
Perhaps the explanation lies, at least in part, in measurement issues. Should we measure the labour share in gross income that includes depreciation of the capital stock, or its share in net income that deducts depreciation? The latter might be more appropriate for gauging aggregate welfare (Hulton 1992) since accelerating capital depreciation reduces what is available to produce future consumption. Although Karabarbounis and Neiman (2014) argue that the labour share has declined markedly since 1975 relative to either measure of income, Bridgman (2018) and Rognlie (2015) counter that the labour share in net income remains within a range that have been observed over a longer history. Further, as spending on computer software and other forms of intellectual property (IP) has grown, the classification of such spending as either investment in durable property or as outlays for intermediate inputs affects the denominator in a labour-share calculation. Koh et al. (2020) attribute much of the decline in the reported labour share to a recent Bureau of Economic Analysis reclassification of intellectual property outlays from non-durable inputs to durable capital. Still other measurement issues arise from the methods used to allocate self-employment and entrepreneurial income between labour and capital (Elsby et al. 2013) and from the treatment of imputed rent on owner-occupied housing, which as Rognlie (2015) points out, can be considered to be a return to capital that accrues mostly to workers and their families. Finally, Barkai (2020) and others have emphasised the growth of ‘factorless income’, that is, typically profit income that accrues to neither labour nor capital.
Moving beyond measurement, many economic shocks have been nominated to possibly explain a decline in the labour share. Capital-biased technical change is one that is frequently mentioned. When technological change augments labour - as seemed to be roughly the case for many decades - income shares are not affected because the initial change in factor shares is offset by the subsequent wage increase. But recent advances in information and communications technology (ICT) and in artificial intelligence (AI) might be more capital-biased than past forms of technological progress. Moreover, the development of new forms of capital that entail substantial fixed costs may tilt industry competition in favour of ‘superstar’ firms (Autor et al. 2020) that distribute a smaller share of their revenues as wages than do smaller firms.
One new form of capital that has received considerable attention in the literature is the industrial robot. Whereas older forms of capital, like equipment and structures, required human operators and thus complemented at least some forms of labour input, robots are often seen as substituting for labour in the performance of a range of manufacturing tasks. Acemoglu and Restrepo (2020) provide evidence that firms and industries that adopt robots tend to reduce their labour demand per unit of output. Yet, as Humlum (2010) and Bonfiglioli et al. (2020) have found, such firms see especially large productivity increases that allow them to grow their market shares. The expansion of these firms offsets the direct effect on their demand for labour.
Globalisation, and especially the extraordinarily rapid emergence of China as a global player in manufacturing, has often been cited as a leading cause of a decline in the labour share in more developed economies. Firms in the advanced countries increasingly offshore to low wage countries, and especially to China, the performance of manufacturing tasks that previously paid a healthy share of cost to domestic workers. Elsby et al. (2013) find that industries that realised especially large increases in their imports were industries in which the labour share fell by more (see also vom Lehn 2018). Castro Vincenzi and Kleinman (2020) point to a different mechanism by which the astronomical growth rates in China might have responsible for a decline in labour shares elsewhere. They argue that China’s growth contributed to booming prices of traded raw materials and show that the decline in the US labour share was concentrated in industries that rely heavily on raw materials.
A third suspect frequently indicted for part of the decline in the labour share is the increase in industry concentration and the associated rise in the exercise of market power by ever-larger firms. De Loecker et al. (2020) have championed the argument that markups have risen substantially in the US in recent years.1 Rising markups can go hand-in-hand with rising profit shares, leaving a shrinking share for factors of production. De Loecker et al. (2020) conclude that the rise in markups that they estimate can account for a substantial part of the fall in the labour share.
Still further possible explanations for the decline in the labour share can be found in the recent literature. Farber et al. (2018) document a positive correlation between state-level labour shares since 1929 and state union membership rates, after controlling for time and state fixed effects and a variety of other controls. They infer that the steadily falling unionisation rates in the US, and the associated decline in workers’ exercise of monopoly power in labour markets, may be responsible for some of the fall in the labour share. Relatedly, Arnold (2020), Azar et al. (2020a, 2020b), and Berger et al. (2021) suggest that increased concentration of demand in local labour markets can tilt wage bargaining in favour of firms. Glover and Short (2020) and Acemoglu and Restrepo (2021) describe possible links between the aging of the population in many countries and the decline in the labour share, while Grossman et al. (2021) point to an endogenous rise in educational attainment together with capital-skill complementarity as another possible mechanism.
Much of the empirical research that examines these alternative explanations has focused on cross sections of industries, firms, or commuting zones. Researchers have investigated whether some possible explanatory variable (e.g. robot use or imports from China) has been positively or negatively correlated with the changes in labour shares in a relevant cross section. That is, the labour share may have declined more in firms that use more robots or in commuting zones that import more from China. This approach is useful because it is often a credible way to assess testable prediction of candidate mechanisms. Nevertheless, as we argue in our article, this empirical approach suffers from a serious shortcoming.2 In the cross-section, firms and industries typically face the same (or closely related) wages and technologies. To the extent that shocks to labour demand induce changes in wages, or that changes in factor prices induce an endogenous technological response, the cross-section evidence will miss the offsetting, equilibrating forces.
Finally, we speculate in the final section of our review article about why the literature seems to over-account so substantially for the phenomenon it seeks to explain. First, we point to the focus on proximate causes and the almost ubiquitous failure to identify fundamental causes. In consequence, many authors present different sides of the same coin. A fundamental cause is a shock that can plausibly be taken as exogenous to the macroeconomic history under investigation. Exogeneity is always difficult to verify, and it is relatively easy to develop stories that link most variables in economic history. But if proximate causes are identified in place of fundamental ones, interpretation becomes difficult. Even if the mechanisms that authors suggest are in fact active, it becomes difficult to gauge what part of the effect estimated in one study has already been accounted for by others. For example, technological progress that reduces the cost of ICT capital might induce substitution that depresses the labour share while at the same time bolstering the market power of large firms that have greater incentive to make these investments. Then a study that measures the effects of automation and another that measures the effects of increasing concentration may capture twice the effects of the technological progress that is actually the original shock. Or perhaps the technological progress itself represents and endogenous response to changes in the demographic and educational composition of the labour force.
Second, the myriad of shocks that researchers have uncovered that affect labour shares suggest to us the operation of some mechanism that exerts a countervailing force. Two such stabilising mechanisms that have been proposed in the literature have been that (i) firms may respond by attempting to devise production techniques that conserve on the use of inputs that are relatively scarce (Acemoglu 2003); and (ii) workers may accumulate skills or enter occupations that have become relatively scarce (Grossman et al. 2017). While there is some evidence for both of these types of mechanism, the processes of invention and skill accumulation operate slowly and their impact is common across cross-sectional units, making them difficult to observe directly in cross-section regressions or relatively short time series. When assessing whether the decline in the labour share is likely to persist or be reversed, we suggest that the equilibrating forces are powerful and likely account for the earlier long periods of stability in factor shares. These equilibrating forces may take a while to work themselves out, but the very recent, modest recovery in the labour share suggests to us that the post-2000 record should not be viewed too pessimistically.
Finally, we fear selection and publication bias in economic research. If some forces have pushed the labour share downward while others have buffeted this decline, the former are much more likely to be pursued by researchers and pass muster with referees and editors eager to explain the labour share decline.
References
Acemoglu, D (2003), “Labor- and capital-augmenting technical change”, Journal of the European Economic Association 1(1): 1-37.
Acemoglu, D and P Restrepo (2020), “Robots and jobs: Evidence from US labor markets”, Journal of Political Economy 128(6): 2188-2244.
Acemoglu, D and P Restrepo (2021), “Demographics and automation”, The Review of Economic Studies.
Aghion, P, A Bergeaud, T Boppart, P J Klenow and H Li (2019), “A theory of falling growth and rising rents”, NBER Working Paper 26448.
Arnold, D (2020), “Mergers and acquisitions, local labor market concentration, and worker outcomes”, Unpublished manuscript, Princeton University.
Autor, D, D Dorn and G H Hanson (2013), “The China syndrome: Local labor market effects of import competition in the United States”, American Economic Review 103(6): 2121-2168.
Autor, D, D Dorn, L F Katz, C Patterson and J Van Reenen (2020), “The Fall of the Labor Share and the Rise of Superstar Firms”, The Quarterly Journal of Economics 135(2): 645-709.
Azar, J, I Marinescu and M Steinbaum (2020a), “Labor market concentration”, Journal of Human Resources.
Azar, J, I Marinescu, M Steinbaum and B Taska (2020b), “Concentration in US labor markets: Evidence from online vacancy data”, Labour Economics 101886.
Barkai, S (2020), “Declining labor and capital shares”, Journal of Finance 75(5): 2421-63.
Berger, D W, K F Herkenhoff and S Mongey (2019), “Labor market power”, NBER Working Paper 25719.
Bonfiglioli, A, R Crinò, H Fadinger and G Gancia (2020), “Robot imports and firm-level outcomes”, CEPR Discussion Paper 14593.
Bridgman, B (2018), “Is labor’s loss capital’s gain? Gross versus net labor shares”, Macroeconomic Dynamics 22(8): 2070-87.
Castro Vincenzi, J M and B Kleinman (2020), “Intermediate input prices and the labor share”, Unpublished manuscript, Princeton University.
De Loecker, J, J Eeckhout and G Unger (2020), “The rise of market power and the macroeconomic implications”, Quarterly Journal of Economics 135(2): 561-644.
Doraszelski, U and J Jaumandreu (2019), “Using cost minimization to estimate markups”, Unpublished manuscript, Boston University.
Elsby, M W L, B Hobijn and A Sahin (2013), “The decline of the U.S. labor share”, Brookings Papers on Economic Activity 2013(2): 1-63.
Farber, H S, D Herbst, I Kuziemko and S Naidu (2018), “Unions and inequality over the twentieth century: New evidence from survey data”, NBER Working Paper 24587.
Glover, A and J Short (2020), “Can capital deepening explain the global decline in labor’s share?”, Review of Economic Dynamics 35: 35-53
Grossman, G M and E Oberfield (2022), “The Elusive Explanation for the Declining Labor Share”, Annual Review of Economics, forthcoming.
Grossman, G M, E Helpman, E Oberfield and T Sampson (2017), “Balanced growth despite Uzawa”, American Economic Review 107(4): 1293-1312.
Grossman, G M, E Helpman, E Oberfield and T Sampson (2021), “Endogenous education and long-run factor shares”, American Economic Review: Insights 3(2): 215-32.
Hulten, C R (1992), “Growth accounting when technical change is embodied in capital”, American Economic Review 82(4): 964-80.
Humlum, A (2010), “Robot adoption and labor market dynamics”, Unpublished manuscript, Princeton University.
Kaldor, N (1961), “Capital accumulation and economic growth”, In D C Hague (ed.), The Theory of Capital: Proceedings of a Conference Held by the International Economic Association, pp. 177-222, London: Palgrave Macmillan.
Karabarbounis, L and B Neiman (2014), “The global decline of the labor share”, Quarterly Journal of Economics 129(1): 61-103.
Keynes, J M (1939), “Relative movements of real wages and output”, The Economic Journal 49(193): 34-51.
Koh, D, R Santaeulàlia-Llopis and Y Zheng (2020), “Labor share decline and intellectual property products capital”, Econometrica 88(6): 2609-28.
Nakamura, E and J Steinsson (2018), “Identification in macroeconomics”, Journal of Economic Perspectives 32(3): 59-86.
Raval, D (2020), “Testing the production approach to markup estimation”, Unpublished manuscript, Federal Trade Commission.
Rognlie, M (2015), “Deciphering the fall and rise in the net capital share: Accumulation or scarcity?”, Brookings Papers on Economic Activity 2015(1): 1-69.
Traina, J (2018), “Is aggregate market power increasing? Production trends using financial statements”, Unpublished manuscript, University of Chicago.
vom Lehn, C (2018), “Understanding the decline in the U.S. labor share: Evidence from occupational tasks”, European Economic Review 108: 191-220.
Endnotes
1 Their claims have not gone unchallenged, however (e.g. Traina 2018, Doraszelski and Jaumandreu 2019, Raval 2020).
2 See also Nakamura and Steinsson (2018), who discuss more broadly the use of cross-sectional methods for understanding aggregate effects of macroeconomic shocks.
