The role of the wage-productivity gap in economic activity

Abstract

This paper argues that wages lagging behind productivity is a long-run structural phenomenon due to the interplay of wage dynamics and productivity growth. We call this interplay frictional growth, a term that can only be nullified in the utopian case of zero growth and/or no dynamics. In that vein, we challenge the prevailing view of the neutrality of the labour income share and investigate its impact on the evolution of employment. We thus estimate wage setting and labour demand equation systems – for France, Germany, Italy, Japan, Spain, the UK, and the US over the 1960–2008 period – and find that the labour share is negatively associated with employment even when the conventional assumption of a unitary long-run elasticity of wages with respect to productivity holds. Acknowledging the presence of the wage-productivity gap in both the short and long run, this work stands as the building block for assessing the effect of the falling labour share on economic activity. As recent work has shown that the widening wage gap is also an important factor prompting inequality, it can be argued that by supporting employment the falling labour share ‘sweetens’ the impact of rising income inequality, and, as such, deserves the attention of policy makers.

Keywords:

• labour income share
• employment
• frictional growth

JEL Classifications:

• E24
• E25
• O47

Acknowledgements

We greatly appreciate the insightful comments of an anonymous referee. We acknowledge financial support from the Spanish Ministry of Economy and Competitiveness through grant ECO2012-13081.

Notes

²³ Clearly, in a two-factor model the profit share is one minus the wage share: Thus, in a competitive economy (Π = 0) production is distributed according to the marginal products of its factors. Nevertheless, it is worth noting that the labour share will not remain constant without assuming a CD production function. For example, Manning (1993, 104), in the context of maximising a Nash bargain, uses a CES technology and derives a time-varying share of labour in output.

1. Further, note the intricate association of the labour share to the (real) unit labour cost; the latter is defined as the average (real) cost of labour per unit of real output.

2. See, for example, Ball and Mankiw (2002), Ball and Moffitt (2002), and Hatton (2007). It should also be noted that the concern of the contemporary literature has been with the falling labour share, i.e. the real wage growth lagging productivity growth. See, inter alia, Stockhammer (2009, 2013), Judzik and Sala (2013), and Magdoff and Foster (2013).

3. For example, Stockhammer, Onaran, and Ederer (2009) estimate consumption, investment, exports and imports equations for the Euro area (treated as one unit) over the 1962–2005 period and argue that wage moderation, defined as real wage growth below productivity growth, is unlikely to stimulate employment. However, their finding that demand is wage-led in the Euro area does not extend to the individual member states. Hein and Vogel (2008) analyse the relationship between the functional income distribution and economic growth from 1960 to 2005 via single-equation estimation of the components of aggregate demand (consumption, investment, net exports) and find that growth in Austria and the Netherlands (as small open economies) has been profit-led, whereas France, Germany, the UK and the US (as larger and less open economies) have been wage-led. Nevertheless, in their review of the empirical literature, the authors note that the evidence with respect to the long-run developments in major OECD countries is rather inconclusive.

4. Unsurprisingly, as wages have been lagging further behind productivity, spending can keep up with output only by the expansion of consumer debt – for example, the ratio of household debt to GDP in the US jumped from 75% in 2000 to 104% in 2007 (BIS 2009).

5. For an investigation of the causes of this fall see, for example, Bentolila and Saint-Paul (2003), IMF (2007), Stockhammer (2009), and Milberg and Winkler (2010).

6. This point of view is in line with the work of Layard, Nickell and Jackman (1991) (LNJ). The LNJ model uses a Cobb-Douglas framework that, as we show in the Appendix, implies a constant labour share that is neutral to employment.

7. Checchi and García-Peñalosa (2010) and Karanassou and Sala (2012).

8. For a survey of the wage curve literature vis-à-vis the Phillips curve one see Montuenga-Gómez and Ramos-Parreño (2005).

9. The static labour demand equation associated with a typical Cobb-Douglas production function is analysed in the Appendix.

10. We experimented with this variable in Germany, Spain, the UK, and the US, using the DAX, IGBM (Índice General de la Bolsa de Madrid), FTSE, and S&P 500 indices, respectively.

11. http://ec.europa.eu/economy_finance/ameco.

12. Furthermore, the fitted values track the actual data very well. To save space the plots of the fitted versus actual values are available upon request from the authors.

13. We view this negative trend as capturing some idiosyncratic features of the French economy for which data are not available – possibly related to issues dealing with the importance of public sector employment.

14. Among other studies, Pissarides (1991) finds that the real interest rate and M3 have significant employment effects in the Australian labour market over the 1966–86 period.

15. The influence of aggregate demand and input price factors on employment is also a feature of the labour demand function in Manning (1993, equation (2(2) )).

16. For all countries in our sample the number of jobs has generally grown across decades. There are only three exceptions. Employment decreased in Spain in the 1970s by 3.9%, Italy in the 1990s by 0.7%, and Japan in the 2000s by 0.9%.

17. Blanchard (2009, 220) correctly points out that “the use of ‘shocks’ is fraught with philosophical, but also with practical, difficulties: technological shocks, animal spirits, changes in perceived uncertainty, etc. all have deeper causes, which themselves have even deeper causes, and so on”.

18. The purpose of this selection is mainly to conserve space. The complete set of plots is available upon request from the authors.

19. The employment elasticity reported in Table 4 is computed by the difference between the log values of the red and blue lines in 1990 as a ratio of the difference between the log values of the labour share at the start and the end of the decade:

20. Following the 1984 labour market reform, the increase in employment is associated with the boom in temporary job contracts.

21. See, for example, Wolff and Zacharias (2006), Piketty and Saez (2006), the Report of the National Equality Panel: An Anatomy of Economic Inequality in the UK, January 2010, The High Pay Commission (2011), and Atkinson, Piketty, and Saez (2011).

22. To use the term of Ball and Mankiw (2002, 130).