Which skills contribute most to absorptive capacity, innovation and productivity performance? Evidence from the US and Western Europe

ABSTRACT

Skills are widely recognised as central to absorptive capacity, that is, firms’ ability to identify and make effective use of knowledge, ideas and technologies that are generated elsewhere. But identification of the specific levels of education and skills that contribute most to the development of absorptive capacity is often hampered by the use of skill measures as proxies for absorptive capacity itself. Drawing on a cross-country industry-level dataset, we retain separate measures of key components of absorptive capacity, namely, skills, R&D investments and openness to foreign trade and investment. We then estimate a system of structural equations in order to evaluate the extent to which different levels of skill contribute to innovative output (measured by growth in patenting) and subsequently to growth in productivity. We find important roles for both high-level skills and upper intermediate (technician-level) skills in converting the knowledge sourcing opportunities provided by openness into innovative output. In final stages of production (making use of innovative output), productivity growth in countries near to the technological frontier is enhanced not just by high-level and upper intermediate skills but also by the skills of the workforce as a whole.

KEYWORDS:

• Productivity
• innovation
• skills
• absorptive capacity

Acknowledgements

We are grateful to the UK Economic and Social Research Council and the Centre for Research on Learning and Life Chances (LLAKES), UCL Institute of Education, London, for their financial support. We also thank seminar participants in London and the 2017 International Workshop on Productivity, Innovation and Intangible Investments in Assisi and two external referees to this journal for helpful comments on previous versions of this article. Responsibility for the content of the article and for remaining errors is ours alone.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Francesco Venturini http://orcid.org/0000-0002-7295-268X

Notes

1 See Lane, Koka, and Pathak (2006) for a detailed discussion of AC measurement difficulties.

2 Bolli, Renold, and Wörter (2018) focus specifically on the impact of ‘vertical education diversity’ on innovation performance rather than on the contributions made by different skill groups (the focus of the present paper). However, as Bolli et al. note (127–128), in Switzerland vertical education diversity is strongly related to the relatively large proportions of workers at firm level who have been trained in the country’s high-quality vocational education and training system.

3 Franco, Marzucchi, and Montresor (2014) define skills as the presence of innovation-related training programmes at firm level and/or no reported problems due to lack of qualified workers.

4 Specifically, Escribano, Fosfuri, and Tribo (2009) derive AC as the principal component of four variables, two related to R&D spending, one related to training provision and one related to the employment share of engineers and scientists.

5 Using a proxy measure of skills based on formal qualifications, a common definition of ‘intermediate’ refers to certificates or diplomas which lie below university graduate (Bachelor degree) level but are above proficiency levels regarded as ‘semi-skilled’.

6 Horn and Cattell (1962) define fluid intelligence as reflecting the impact on intellectual abilities of heredity and injury (such as impairment with age) while crystallised intelligence reflects the impact on abilities of learning acquired over time, for example, through work experience and continuing education and training, whether formal or informal in nature.

7 If $\phi$ is unitary, this points to constant returns to scale in knowledge production. If $\phi$ is less than unity, this implies decreasing returns, whilst the reverse holds when$\phi$is greater than one.

8 Logs are not taken for the openness measure since, as described in Section 4 below, it is derived from data on foreign trade and FDI as a factor score with mean zero and standard deviation of one.

9 Source: https://www.oecd.org/eco/growth/indicatorsofproductmarketregulationhomepage.htm#indicators Copies of the relevant files accessed in 2016 are also available from the authors on request.

10 FDI flows and total gross output are aggregated to three-year periods because of unevenness in annual FDI flows at country/industry level.

11 Factor test scores: Cronbach’s alpha measure of internal reliability: 0.696; Kaiser-Meyer-Olkin measure of sampling adequacy: 0.510; Bartlett’s test of sphericity: p < 0.001***.

12 Under this assumption a measure of quality-adjusted total labour input is obtained by weighting each different skill group (as signified by qualification levels) by the share that each skill group occupies in total labour compensation (see, for example, Jorgenson et al. 2005).

13 Supplementary estimates and appendices are available at: https://works.bepress.com/francesco_venturini/60/

14 Full descriptive statistics are available in an on-line supplement to this paper (see Note 13)

15 Source: OECD Research and Development Statistics.

See http://stats.oecd.org/Index.aspx?DataSetCode=ONRD_COST

16 See Note 13.

17 In these second-stage regressions, standard errors are bootstrapped with 200 replications.

18 See Note 13.

19 We are grateful to a referee for this journal for making this suggestion.

20 Considerable resources have recently been invested in cross-country comparisons of STEM graduate supplies by organisations such as the OECD and the US National Science Foundation.

See http://www.oecd.org/sti/oecd-science-technology-and-industry-scoreboard-20725345.htm; https://www.nsf.gov/statistics/2018/nsb20181/report/sections/overview/workers-with-s-e-skills.

However, upon investigation, these comparisons have so far focussed exclusively on annual output (flows) of different kinds of graduate for entire countries. By contrast, in the present paper our analysis relies on the availability of data on the stocks of workers with different qualifications at industry level in each country concerned.

Additional information

Funding

This work was supported by the UK Economic and Social Research Council under Grant ES/J019135/1 via the Centre for Research on Learning and Life Chances (LLAKES), UCL Institute of Education, London.