Developing people’s knowledge and skills – their ‘human capital’ – is thought to be critical for improving productivity and sustaining economic growth. Empirical research suggests this link is surprisingly dubious, but measurement challenges may help explain this.
Human capital is the knowledge, skills, competencies and health embodied in individual people (OECD, 2001). Like physical capital – such as buildings or machinery – human capital can be ‘accumulated’, usually through investment in education, training or better health. Similarly, it can depreciate due to the lack of use, ageing or even obsolence – when new and disruptive technologies emerge, making old skills redundant. For example, the development of artificial intelligence (AI) and automation are likely to replace many jobs in manufacturing and services that currently rely on human labour.
What do the data say about the relationship between human capital and productivity?
Education – as a major form of human capital investment – has become one of the focal points of economic policy. Yet, evidence suggests that, in OECD countries, the correlation between average educational attainment and ‘multi-factor productivity’ – the overall efficiency with which labour and capital inputs are used together in the production process – is negative (and statistically significant). Counterintuitively, this would suggest that the more a person is educated or trained, the less productive they become. But when the data include non-OECD countries, the relationship is positive – a more predictable finding.
These results are from the widely used Barro-Lee data set, which measures average educational attainment from 146 countries between 1950 and 2010. Surprisingly, the trend showing the negative correlation between education and productivity in OECD countries has been increasing over the past few decades.
Another recent study casts doubt on the expected positive correlation between education and earnings. Despite the percentage of the US population with a college or university education having increased dramatically between 1970 (11%) and 2010 (27%), real wages have stagnated for the majority of US college and secondary education graduates. The only improvement in earnings is observed for the very highest earners, suggesting an increase in income inequality.
What does economic theory tell us?
These findings from the data are puzzling as economic theory tells us that human capital is a fundamental determinant of a country’s growth rate. This also helps to explain the policy focus on education in recent decades. In the modern ‘knowledge economy’, the supposed relationship between education and economic growth suggests a need for an ever-increasing share of the population to be highly educated, either to adopt existing knowledge or to generate innovative ideas for future growth.
To explore the relationship between human capital, labour productivity and income, Figure 1 shows how GDP per capita is determined primarily by two factors: labour productivity (GDP in volume terms, divided by the total labour hours worked) and the employment rate.
Labour productivity is itself influenced by ‘multi-factor productivity’ (also known as total factor productivity) and ‘capital deepening’ (more capital per worker). What Figure 1 illustrates is that, at least according to economic theory, higher levels of human capital should improve labour productivity and consequently lead to higher per capita income in the economy.
Figure 1: How human capital affects labour productivity and income
Source: Author’s diagram, also see Égert et al, 2019
Multi-factor productivity reflects how efficiently various production inputs are used within the economy and is generally estimated as a ‘residual’ – the part of GDP left over when all inputs are accounted for.
Complexity would arise when deciding whether or not human capital is explicitly counted as one of the inputs, in addition to capital and labour, in the production function (a calculation that gives the relationship between physical inputs and the quantity of goods produced). If human capital is not properly accounted for, it will be treated as part of the multi-factor productivity estimates. As a result, estimates for multi-factor productivity can differ significantly based on different specifications used.
What role does measurement play in the difference between theory and evidence?
Some researchers believe that challenges in measuring human capital may explain the results found in empirical studies that seem to contradict the presumption that increases in human capital should boost productivity.
A recent OECD study identifies several shortcomings of the previous measures of human capital used in past research. The output/income-based approach (also known as the ‘lifetime income-based approach’) is a common method for estimating the existing ‘stock’ of human capital. This method assumes that the lifetime incomes received by individuals are the returns to their investment in education. As a result, the value of the stock of human capital depends on the rate of return to education, educational attainment and the total educated population within the economy.
Based on this approach, the Office for National Statistics (ONS) currently measures the UK human capital stock in monetary terms – as the working age population’s discounted lifetime earnings. The UK’s real full human capital stock is estimated to have increased from £18.75 trillion in 2004 to £21.4 trillion in 2018, which is about ten times national GDP. The returns to the accumulation of human capital could take various forms – for example, in increased labour productivity, more economic output or higher income in the UK.
For the rates of return to education, measures used in previous empirical studies relied on two simplifying assumptions. First, that the returns to education remain constant across countries and over time; and second, that there are constant or diminishing returns to the number of schooling years.
Empirical findings do not support either assumption. Average returns to education for the OECD countries have been rising consistently since the 1970s. And the marginal return to different levels of education appears to have a ‘U-shape’, with average returns for both primary and post-school (college, university or vocational) education exceeding the returns for secondary education.
Figure 2: Returns to additional years of education
Source: Botev et al, 2019
Although the new quality adjusted measure for human capital has been proposed to reflect these empirical realities, and has obtained encouraging results, there are other methodological weaknesses when applying the approach described above.
The estimates only capture the returns from formal education and training. Other aspects of human capital such as informal learning, an individual’s innate abilities and their health are ignored in this system. What’s more, simply adding up an individual’s time in education may fail to take ‘spillover effects’ into account – these can include social benefits, such as access to networks or increased social status.
In addition, some knowledge and skills are specific to particular contexts, networks and organisations. For example, employees who possess strategic positions in firms are better compensated than others, which implies that their social network can also be a component of their human capital.
To reflect the multidimensional nature of human capital, statistical agencies across the world have been trying to improve existing measurements. The ONS has developed an indicator-based framework to complement its stock measure. Similarly, the World Bank has developed a composite human capital index (HCI) as part of the World Development Report.
The indicator-based framework is more effective at capturing other important dimensions of human capital, such as health. For example, within the four sub-indicators of the World Bank’s index, three are health-related – mortality rates, stunting prevalence for children under five, and adult survival rates.
What can policy-makers do to boost human capital?
Although the current measures of human capital are far from perfect, they can still shed some light on education policies. Six such policies (listed in Figure 3) have been tested and found to be associated with human capital measures across all the OECD countries.
All of these have room for improvement on at least one of the six fronts. The UK, for example, could benefit from improving attendance in pre-primary education and decreasing the student-teacher ratio in primary and secondary education (see Figure 3).
Early childhood education has been identified as one of the most cost-effective strategies for promoting economic growth. Pre-school education not only provides a solid foundation for better school performance, but is also essential for the development of a child’s emotional, social and overall wellbeing.
Investment in early childhood education is particularly beneficial for disadvantaged families. Currently, pre-primary education attendance in the UK is far below the OECD average. By closing this gap with the top three performing countries, it is estimated that GDP per capita would increase by 1.7% in the long run.
Figure 3: Long-run impacts of education reforms by closing the gap with the average of top three OECD performers, % change in GDP per capita
Similarly, the UK has a relatively high pupil-teacher ratio compared with its European counterparts (see Figure 4). This ratio is regarded as a measure of the teaching resources available in schools.
A lower ratio is generally correlated with better learning outcomes. The potential increase in GDP per capita is estimated to be 1.7% in the long run, should the UK be able to reduce the ratio to the average level of the top three OECD performers. For developing countries, the benefits of making progress on this policy could be even more significant, although the quality of teaching matters, not just its quantity.
Figure 4: Ratio of pupils and students to teachers and academic staff in upper secondary and primary education, 2017
Panel A: Secondary
Panel B: Primary
Accumulating more human capital also has positive spillover effects for society, as each person benefits from others being better educated. But when the decision is left to individuals, there is a tendency to ‘under-invest‘, suggesting policy-makers need to encourage the accumulation of skills and education.
In addition, regional and intergenerational inequality can affect people’s wish to invest in their human capital, perhaps due to low perceived returns from education. This lack of investment can result in a self-fulfilling prophecy, causing inequalities to persist.
Despite the empirical puzzle of not being able to demonstrate a clear positive link between human capital and economic growth, it appears that measurement challenges could be a contributing factor in this result. These challenges include accounting for the quality of human capital and incorporating other dimensions like health or social networks into the human capital ‘stock’.
For lower-income (non-OECD) countries, the benefits of investing more in education are clearer in the data, as well as in the theory. Individuals do not have the incentive to invest as much in their own education as is desirable for the economy as a whole, and when people perceive a low rate of personal return to acquiring skills, this can turn into a vicious circle.
Education policies such as improving the ratio of teachers to pupils and investing where returns are high – particularly at primary stage – are therefore essential for improving productivity and fostering long-term prosperity.
Where can I find out more?
- Education and economic growth: Summary of the research evidence from the LSE’s Programme on Innovation and Diffusion (POID).
- How intangible assets can help alleviate the regional disparity: Discussion of the role of human and social capital in addressing regional inequalities in the UK.
- Human Capital Project: World Bank project to close the human capital gap in the world during and after the Covid-19 crisis.
- People matter – improving our estimates of human capital: Report from the Office of National Statistics on the development of indicators to measure human capital.
- Quality-adjusted labour input (QALI) QMI: ONS blog on methodology for QALI and multi-factor productivity estimates.
- Productivity, human capital and educational policies: OECD articles and resources on human capital.
Who are experts on this question?
- Angus Deaton, Princeton University
- Claudia Goldin, Harvard University
- Phillip Brown, Cardiff University
- Hugh Lauder, University of Bath
- Sin Yi Cheung, Cardiff University
- James Heckman, University of Chicago
- Steven Durlauf, University of Chicago
- Anna Valero, London School of Economics