The productivity challenge. What to expect from better-quality labour and capital inputs?

ABSTRACT

The aim of this article is to develop and implement an analytical framework assessing whether better-quality inputs, via a rise of TFP, could compensate an ageing-induced slowing of economic growth. Here ‘better-quality’ means more educated and older/more experienced workforces; and also better-quality capital proxied by its ICT content. Economic theory predicts that these trends should raise TFP. To assess these predictions, we use EU-KLEMS data, with information on the age/education mix of the workforce, as well as the importance on ICT in total capital, for 34 industries within 16 OECD countries, between 1970 and 2005. We generalize the Hellerstein–Neumark labour-quality index method to simultaneously capture workers’ age/experience or education contribution to TFP growth, alongside that of ICT. The conclusion of the article is that the quality of inputs matters for TFP. We find robust microeconometric evidence that better-educated and older/more experienced workers are more productive than their less-educated and younger/less-experienced peers. Also, ICT capital turns out to be more productive than other forms of capital. And when used in a growth accounting exercise covering the 1995–2005 period, these estimates suggest that up to 40% of the recorded TFP growth could be ascribed to the rising quality of inputs.

KEYWORDS:

• TFP growth
• ageing
• input quality
• ICT

JEL CLASSIFICATION:

• J11
• J24
• D24
• O30

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ In Y/P = Y/L*L/P, where Y is total output, P is total population, L is labour force, population ageing means that L/P goes down; potentially causing a reduction of output per head (Y/P).

² Macroeconomists estimate that the average annual per capita growth in Belgium may not exceed 0.5% per year until at least 2040 (Heylen et al. 2016), and it may stay below 1% until almost 2060. Demographic ageing² is by far the most influential cause of low growth. A strongly rising dependency ratio due to the retirement of the baby boom generation, increasing longevity and (to a lesser extent) a temporary fall in the population at working age, implies that the output of fewer workers must be shared with more inactive people. Arithmetically this could drag down annual per capita growth by about 1% point between 2010 and 2040

³ In Y/P = Y/L*L/P, population ageing (L/P↘) can be compensated by higher labour productivity (Y/L↗). Cutler et al. (1990) even posit that L/P↘ could boost labour productivity gains, arguing that ‘scarcity is the mother of invention’. This scarcity view assumes that in a situation of relatively slow population growth, there is an acceleration, on a per capita basis, in human capital accumulation. In their cross-national analysis of 29 non-OPEC countries for the period 1960–1985, Cutler et al. estimate that a decline in the annual labour force growth rate of 1 percentage point is associated with about a 0.5 percentage point increase in (labour) productivity growth.

⁴ Total Factor Productivity (TFP) is the portion of output (value-added) not explained by the amount of inputs used in production. As such, its level depends on how efficiently and intensely inputs are utilized in production, but also on the quality of these inputs.

⁵ That is induced by demographic/population ageing but should not be confounded with it.

⁶ The magnitude of productivity gains potentially generated by ICT/digitalization (big data, internet-of-things…) remains debated among economists. Could it be that robots, computers, e-platforms are about to generate a rise of labour productivity of a magnitude recorded in the wake of the two previous industrial revolutions (IR) that is, IR1 (steam, railroads) from 1750 to 1830 and IR2 (electricity, internal combustion engine, running water, indoor toilets, communications, entertainment, chemicals, petroleum) from 1870 to 1900? Brynjolfsson and Mc Afee (2014), ​​​​​strongly believe that we are about the embark in IR3. The key idea is that rapid growth in computation and artificial intelligence will cross some threshold after which productivity will accelerate sharply, as an ever-accelerating pace of improvements cascade through the economy.

⁷ Old Dependency Ratio = P⁶⁵⁺/ P^20-64. Note, more generally, that Y/P = Y/L*L/P can be rewritten as Y/P = Y/L*1/(1 + D) where D is the total dependency ratio (i.e. old + young).

⁸ With the aim of assessing the employability of different categories individuals, by comparing (labour) productivity to wage profiles (e.g. Vandenberghe 2011, 2013; Cataldi, Kampelmann, and Rycx 2011).

⁹ http://www.euklems.net/

¹⁰ The term αk_it + βl_it captures the contribution of capital deepening and (dis)economies of scale. Ignoring the latter (i.e. assuming α + β = 1) we have 1 + α(k_it-l_it) = 1+αln(K_it/L_it)

¹¹ With LP, coefficients ß, η^j (i.e. labour input coefficients) are identified at stage 1.

¹² OLS estimates, for example.

¹³ That can thus be used in a GMM analysis.

¹⁴ The data series are publicly available on http://www.euklems.net/euk08i.shtml#top

¹⁵ In EU-KLEMS, capital service input has been measured in a standard way, using harmonised depreciation rates and common rules to deal with a variety of practical problems, such as weighting and rental rates. Importantly, capital input is measured as capital services, rather than stocks.

¹⁶ Respectively ISCED<3, b:ISCED3-5, c: ISCED6+; where ISCED<3, b:ISCED3-5, c: ISCED6+; where ISCED stands for ISCED: International Standard Classification of Education: level 0 – Early childhood education; level 1 – Primary education; level 2 – Lower secondary education; level 3 – Upper secondary education; level 4 – Post-secondary non-tertiary education; level 5 – Short-cycle tertiary education; level 6 – Bachelor’s or equivalent level; level 7 – Master’s or equivalent level; level 8 – Doctoral or equivalent.

¹⁷ As a robustness check, we replicated the analysis using 3-year intervals. Results are very like those reported in Table 2.

¹⁸ Our variables consist of first-differenced logs, that is, approx. growth rates.

¹⁹ The worker sample underpinning EU-KLEMS might not be representative of the entire population of older individuals aged 50 and more. This means that there is a risk of a selection bias, due to early ejection from the workforce of less productive/motivated older workers. To the extent that this selection bias is an issue, we could view our estimated coefficients for older workers’ productivity as upper boundaries.

²⁰ LP, ACF deliver coefficients that are very similar to those obtained using OLS. They deliver simulation/growth accounting results that are qualitatively very similar to those exposed hereafter.

²¹ In a nutshell, our simulations are driven by two things: the magnitude of the estimated coefficients and that of the changes in the input share. They thus reflect the combined effect of the two dimensions.

²² Labour market history, such as experience and firm and industry tenure, as well as general human capital measures such as schooling and gender.

²³ A relatively small marginal productivity premium for ICT, combined to not-so-large rises of the share of ICT into total capital.

²⁴ In reference to Robert Solow’s 1987 quip: ‘You can see the computer age everywhere but in the productivity statistics’.