1. HUMAN CAPITAL: EXTENDING THE
MEASURES
MARY O’MAHONY
PRESENTATION AT THE SEM CONFERENCE 2015
PARALLEL SESSION D: MEASURING CAPITAL AND WEALTH
This research benefited from funding from BIS,
CEDEFOP, LLAKES - an ESRC-funded Research
Centre – grant reference ES/J019135/1, and
the INDICSER and SPINTAN projects financed by
the EU 7th Framework Programme – grant no.
244709 and grant no. 612774.
Department of Management
2. Overview
• Conventional measures of human capital
• Extensions:
– Employer provided training
– Age cohort effects
– Health and Human capital
3. Measures of human capital
• Many international exercises to measure human capital , e.g.
Barro and Lee database, Inclusive Wealth, World Bank plus
international testing exercises such as PISA and PIACC
• Fraumeni (2015) summarises methods and impacts on rankings
of countries
• Human capital stocks measured in satellite accounts by many
statistical offices
• Most popular measures are those based on the Jorgenson
Fraumeni discounted life time income approach
– Applied now in 20 (mostly OECD) countries
– Some countries prefer accumulated expenditure on education
inputs
4. Measures of human capital: JF model
• Applies neoclassical theory of investment to human capital
• Value of human capital depends on an individual’s discounted
lifetime income
• Calculations divide population into various life stages (in
education, in the labour force and retirement)
• Implementation requires data on:
– working population and school enrolments,
– educational attainments and earnings associated with those
attainments
– survival rates
5. Measures of human capital: JF model
• Implementation also depends on a number of assumptions
including:
– Returns from human capital accumulation accrue to
individuals (no role for additions to human capital provided
and paid for by employers)
– Relative wage rates by educational attainment levels are
determined by contemporaneous relative wage rates (the
lifetime income of a 25 year old with a university degree
measured by income earned by older age groups with the
same level of attainment)
– Assumptions on depreciation and survival rates (frequently
assume all persons retire at age 65 and no role for impact of
health)
6. Extending the model: Employer provided training
• Literature on measuring intangible assets (e.g.Corrado, Hulten
and Sichel, 2005) include measures of firm specific human
capital.
• Investments in training provided by firms which does not
feature in the earnings of employees, cumulated to stocks of
intangible training capital
• Estimates suggest that this training capital is sizeable and
important for productivity
– O’Mahony (2015) suggests that in the EU15 investments in
intangible training account for about 2% of GDP
– Comparable to expenditure on secondary education in these
countries
7. Extending the model: Employer provided training
• Econometric estimates in Mason et al. (2014) show significant
direct effects from training capital on productivity
• and significant interactions between training and certain types
of skills (degree and above and upper intermediate)
8. 1995-2007 1995-2007
Capital-labour ratio 0.3962*** 0.2991**
[0.137] [0.147]
Higher skills 0.0665 0.2219***
[0.059] [0.079]
Upper intermediate vocational skills -0.0016 0.0928
[0.051] [0.060]
Lower intermediate vocational skills -0.1244 0.0842
[0.078] [0.091]
Lower intermediate general skills 0.0125 0.0868
[0.061] [0.072]
Average high-skilled training capital per
hour worked 0.0963 0.2779***
[0.077] [0.087]
Average intermediate-skilled training
capital per hour worked -0.0898 0.2549
[0.095] [0.180]
Training capital (higher)*Higher skills 0.0776***
[0.028]
Training capital (intermediate)*Upper
intermediate vocational 0.0563
[0.041]
Training capital (intermediate)*Lower
intermediate vocational 0.058
[0.042]
Training capital (intermediate)*Lower
intermediate general 0.0313
[0.046]
Observations 1456 1456
Adj. R2 0.566 0.608
Fixed effects estimates of average levels of labour productivity,
1995-2007, All Countries
9.
High-
apprentice
countries
High-
apprentice
countries
Low-
apprentice
countries
Low-apprentice
countries
Capital-labour ratio 0.4938*** 0.3928*** 0.3593** 0.3338*
[0.159] [0.141] [0.164] [0.176]
Higher skills 0.1119 0.1246 0.2413** 0.4415***
[0.103] [0.148] [0.109] [0.107]
Upper intermediate vocational skills 0.0843* 0.1660** -0.0144 0.0255
[0.047] [0.073] [0.089] [0.082]
Lower intermediate vocational skills 0.6522 0.7687* -0.2089** 0.0194
[0.445] [0.430] [0.086] [0.110]
Lower intermediate general skills 0.0924* 0.1822** -0.1098 0.0424
[0.051] [0.076] [0.103] [0.097]
Average high-skilled training capital
per hour worked 0.3040*** 0.2851*** 0.0088 0.2986*
[0.072] [0.101] [0.088] [0.153]
Average intermediate-skilled training
capital per hour worked -0.0515 0.6769** -0.0504 0.1024
[0.107] [0.302] [0.099] [0.200]
Training capital (higher)*Higher skills 0.0200 0.1099***
[0.051] [0.038]
Training capital (intermediate)*Upper
intermediate vocational 0.0954** -0.0014
[0.046] [0.068]
Training capital (intermediate)*Lower
intermediate vocational 0.2592 0.0799**
[0.203] [0.037]
Training capital (intermediate)*Lower
intermediate general 0.1124** -0.0023
[0.044] [0.049]
Observations 624 624 832 832
Adj. R2 0.6169 0.6346 0.6085 0.6516
Fixed effects estimates of average levels of labour productivity, 1995-
2007, Comparing high- and low-apprenticeship countries
10. Extending the model: Employer provided training
• The impact of training varies by systems of educational
provision
– countries that target more general education (e.g. UK,
Scandinavian countries) see greater productivity impacts from
training the high skilled
– Countries with more vocational orientation education systems
(e.g. Germany and Austria) also derive significant productivity
benefits from training at upper intermediate level
• Econometric estimates in O’Mahony and Riley (2012) suggest
significant spillovers from high skilled labour that are facilitated
by employer provided training
11. (a) (b) (a) (b) (a) (b)
Share of tertiary education hours 1.364** 1.024*** 1.121** 0.836*** 0.383 0.293
(0.013) (0.001) (0.040) (0.008) (0.436) (0.319)
Log tertiary training capital per tertiary hours -0.018 -0.011 -0.087 -0.061 -0.139** -0.109***
(0.818) (0.813) (0.228) (0.158) (0.047) (0.010)
Interaction between education and training 0.843*** 0.351** 1.063*** 0.474*** 0.879*** 0.395**
(0.003) (0.042) (0.000) (0.005) (0.002) (0.022)
Log IT capital per hour -0.015 0.030 0.008 0.060** -0.021 0.018
(0.745) (0.298) (0.865) (0.034) (0.656) (0.529)
Log fixed capital per hour -0.451*** -0.414*** -0.333*** -0.358***
(0.000) (0.000) (0.002) (0.000)
Log gross value added per hour 0.204*** 0.180*** 0.128* 0.117***
(0.006) (0.000) (0.073) (0.007)
Employment growth (over last 5 years) -0.073 0.035 -0.005 0.106**
(0.296) (0.407) (0.946) (0.010)
Observations 8,095 8,095 8,095 8,095 8,095 8,095
Individuals 2239 2239 2239 2239 2239 2239
Country/Industry fixed effects 48 48 48
Individuals*Country/Industry fixed effects 2879 2879 2879
Country/Industry/Wave random effects 333 333 333 333 333 333
(1) (2) (3)
Notes: Dependent variable is the log hourly wage; Additional controls include year effects, indicator for managerial and professional
occupations, marriage, quadratic in age and quadratic in job tenure, workplace size, permanent contract, vocational training course.
Education spillovers and training
12. Finland France Germany Netherlands Sweden UK United
States+
Output per person
hour 2002-2007 3.32 1.60 1.53 1.84 . 2.70 2.09
2008-2013 -0.36 0.18 0.44 -0.26 0.71 -0.45 1.13
Percentage point contribution
Training 2002-2007 0.08 0.08 0.04 0.15 0.06 0.07 0.00
2008-2013 0.05 -0.02 0.00 0.05 -0.02 -0.04 0.00
Labour
composition(Skills) 2002-2007 0.29 0.32 0.15 0.51 0.19 0.47 0.27
2008-2013 0.29 0.36 0.29 0.05 0.19 0.54 0.33
Recent work that extends growth accounting to the period after the financial
suggests significant declines in the contributions of training capital to labour
productivity growth
Extending the model: Employer provided training
13. Extending the model: Relative wages
• Results from the recent PIACC survey questions the
contemporaneous relative wage assumption
• Fraumeni (2015) shows country rankings according to the JF
model and those from PIACC. Of the 20 countries for which JF
estimates are available:
– The US ranks first on JF and only 10th
on PIACC
– Similarly the UK ranks 2nd
on JF and 8th
on PIACC
– In contrast Australia ranks 9th
on JF and 4th
on PIACC
– And the Netherlands is 1oth on JF and 2nd
on PIACC
• Results suggest the need to revisit whether earnings reflect skills
especially for older age cohorts
• Probably needs a second wave of PIACC to fully understand the
implications
14. Extending the model: Including health
• Simultaneously with work on measuring human capital many
countries also experiment with estimating health capital stocks
• Need to integrate the two measures
• Health affects human capital through:
– Affecting the retirement age (survival in the JF model)
– Impacts on earnings of those who continue working in poor
health
– Incentives to invest in human capital – greater longevity
might increase the returns to investing in education
• Health effects depend not only on the individuals own health but
also on their dependents
– Retirement can also occur due to caring activities
15. Extending the model: Including health
• Current work Estimating the impact of health (both own and
relatives) on retirement decisions
• Estimating the impact of health on earnings
• Constructing human capital stocks dividing also by health status
– (Lea Samek - PhD at King’s)
• First estimates for the UK with possible extension to Germany
• Also modeling the impact of health on education incentives
– Martin Weale for SPINTAN project
16. Extending the model: Conclusions
• Highlighted three areas for further work
• Employer provided training is most complete and there is a
strong case for including this in national accounts as part of
intangible investments
• Combining health and conventional human capital in satellite
accounts is likely to be both feasible and important for policy
• More work required to fully understand the implications of
PIACC
17. References
Corrado, C., C Hulten and D Sichel (2005), “Measuring capital and
technology: an expanded framework”, In C Corrado, J Haltiwanger and D
Sichel (eds), Measuring Capital in the New Economy, The University of
Chicago Press, p. 11-46.
Fraumeni, B. (2015), Choosing a human capital measure:
Educational attainment gaps and rankings, NBER working paper
no. 21283
Mason G, M O’Mahony, A. Rincon, R Riley (2014) “Macroeconomic
benefits of vocational education and training”, Cedefop Research Paper
No. 40.
O’Mahony M (2012) ‘Human Capital Formation and Continuous
Training: Evidence for EU countries’, The Review of Income and Wealth,
Vol. 58, No. 3. p. 531-549
O’Mahony M and Riley R (2012) “Human capital spillovers: the
importance of training”, INDICSER Discussion Paper No. 23.