Remeasuring labour's share

Abstract

Krueger (1999) provides a measure of ‘raw’ labour's share for the US post-war economy based on Mincerian regressions. He finds that raw labour's share fell by over 8 percentage points from 1959 to 1996. We provide an alternative estimate using direct observations on the wage rates of raw labour units, i.e. those with 8 years of education or less; aged 16–18 years. Our measure of raw labour's share is considerably higher on average than Krueger's. Furthermore, our measure rises during the later part of the sample and is over 22% by 1996.

Keywords:

• labour's share
• factor shares
• development
• biased technical change
• capital intensity
• human capital

JEL Classification:

• J30
• O15
• E25

Acknowledgements

We thank John Seater for providing very helpful suggestions based on a previous draft. We thank Mun Ho for supplementing the data that were publicly available to us and for commenting on some features of that data. All errors are our own.

Notes

¹ Acemoglu et al. (2001) link skill-biased technical change and deunionization as complementary factors in both US and UK rising inequality. Katz and Autor (1999), Goldin and Katz (2001) and Acemoglu (2002) review the relevant literature on weakening labour institutions.

² This is based on a conventionally estimated NIPA labour's share of 76.1% in 1996 (Krueger, 1999, table 1).

³ The rise in the return to education has been attributed to skill-biased technical change that outpaced a growing supply of educated labour (Lawrence Katz and Kevin Murphy, 1992) and was perhaps even induced by that growing supply (Acemoglu, 2002).

⁴ See Ho and Jorgenson (1999, section 2 and appendix A) for details of the methodology and data.

⁵ Most of the data we use are available at Dale Jorgenson's website, http://scholar.harvard.edu/jorgenson/data, and Mun Ho kindly provided additional data from the labour matrices. We choose to begin with male workers to avoid issues of gender discrimination.

⁶ The average is only slightly lower – just under 0.20 – if we consider only 1959 through 1996, which encompasses the Krueger points displayed in the figure. As well, Krueger (1999, table 2) also reports a value for 1939 which is the lowest of all his reported values: just under 0.10.

⁷ All estimated trends reported below are likewise significant.

⁸ Mun Ho helpfully provided these numbers in discussion of the volatility of the series.

⁹ Surprisingly, it is consistently lower than our baseline measure until 1965. One plausible explanation is that many such workers are still in school; their earnings largely represent after school/weekend jobs with relatively low wages.