Measuring Human Capital: Learning Matters More than Schooling

Lambert here: Credentialed agnotology FTW!

By Noam Angrist, Fellow, Oxford; Consultant, World Bank; Cofounder, Young 1ove, Simeon Djankov, Policy Director, Financial Markets Group, London School of Economics, and Pinelopi Goldberg, Elihu Professor of Economics, Yale University. Originally published at VoxEU.

Human capital is a critical component of economic development. But the links between growth and human capital – when measured by years of schooling – are weak. This column introduces a better measurement, using a database that directly measures learning and represents 98% of the global population. The authors find that the link between economic development and human capital is strong when measured in this way. They also show that global progress in learning has been limited over the past two decades, even as enrolment in primary and secondary education has increased.

For decades, human capital research has relied on measures of schooling (Mincer 1984, Mankiw et al. 1992). But proxying human capital with schooling assumes that being in school translates into learning. Evidence suggests that this is often not the case (Pritchett 2013). Recent analysis reveals that six out of ten adolescents worldwide cannot meet basic proficiency levels in maths and reading (UNESCO 2017). This gap between schooling and learning is particularly acute in developing countries. In Kenya, Tanzania, and Uganda three quarters of grade 3 students cannot read a basic sentence such as “The name of the dog is Puppy”. In rural India, half of grade 3 students cannot solve a two-digit subtraction problem such as 46 minus 17 (World Bank 2018).

Measuring Human Capital Worldwide: The Harmonised Learning Outcomes (HLO) Database

Several studies have presented the argument that when human capital is measured by schooling, it fails to deliver the returns predicted by growth models. But when measured by learning, human capital is more strongly associated with growth (Pritchett 2006, Kreuger and Lindahl 2011, Hanushek and Woessmann 2012). Understanding growth is critical (Akcigit et al. 2020), and while good country studies exist (Fazzio et al. 2020), we still need comparable, global, timeseries data.

To date, much of the effort to measure learning has focused on advanced economies. This has come at the expense of a significant portion of the global distribution, in particular those countries with the most potential to gain from human capital accumulation. This limitation is due to the absence of comparable measures of learning in developing countries.

In a recent paper, we introduce a database of globally comparable learning outcomes for 164 countries, covering 98% of the global population from 2000 to 2017 (Angrist et al 2021). This is the largest and most current global learning database, and one of the first to disaggregate learning results by gender or introduce methodological improvements such as the inclusion of standard errors to quantify uncertainty. The database is available for public use and updates are expected periodically as new learning data become available. A large-scale effort to track human capital formation using this database is the World Bank’s Human Capital Index. We contribute to a body of research in this area (e.g. Rindermann 2018).

Figure 1 Average learning from 2000-2017 across 164 countries

Source: Harmonized Learning Outcomes (HLO) database.

The Human Capital Gap: Schooling Is Not Learning

We explore the contrast between changes in schooling and changes in learning over time, from 2000-2015. We observe a clear pattern of high and increasing school enrolment, while learning progress appears to be limited in many cases. For example, in the Middle East and North Africa enrolment rates achieved a high of 99% by 2010, up from 95% in 2000. In contrast, learning levels stayed low and remained the same (around an initial score of 380) from 2000 to 2015 in these regions. This pattern has been referred to in previous studies and by the international education community as ‘the learning crisis’. Using the HLO database, we demonstrate that this crisis holds on a global scale.

Figure 2 Enrolment versus learning by region, conditional on country-fixed effects.

Source: The data in this figure include primary enrolment rates from Lee and Lee (2016) and learning estimates from the Harmonized Learning Outcomes (HLO) database.

Learning Accounts For Substantial Economic Development

Several studies in development accounting have explored the relative contribution of human capital to cross-country income differences. The results have been inconclusive, in part owing to difficulties in measuring human capital. Our results suggest that human capital accounts for between a fifth to around half of cross-country differences in income, sitting somewhere in the middle compared to other work (Hall and Jones 1999, Jones 2014, Caselli and Ciccone 2019), which ranges from zero to nearly all (Table 1). This result is not meant to provide definitive estimates, but rather to motivate the use of a direct measure of learning in future analysis in the development accounting literature.

Table 1 Baseline development accounting results and comparison to the literature

We also find stronger links between our measure of human capital and growth than human capital measures which rely largely on schooling, such as the Penn World Tables (Feenstra et al. 2015) and the United Nations Human Development Index (UNDP 2013).


This database comes at a moment when a series of global efforts have been launched to measure and track learning on a global scale. Although recent analysis suggests that the world is on track to achieve the goal of universal primary enrolment by 2030, if learning continues to stagnate, this achievement will mean little.

Our findings have substantial implications for the future of work. As developing economies become increasingly service-oriented and automation and digitisation continue to expand, skilled work will be critical to enable workers to secure the jobs of the future (Baldwin and Forslid 2020). Human capital is a critical enabler of this transition. Our findings reveal that schools – which have for decades been assumed to impart knowledge and skills – have often fallen short. But if they deliver in terms of learning, the development gains could be substantial.

Print Friendly, PDF & Email
This entry was posted in Income disparity, Social policy on by .

About Lambert Strether

Readers, I have had a correspondent characterize my views as realistic cynical. Let me briefly explain them. I believe in universal programs that provide concrete material benefits, especially to the working class. Medicare for All is the prime example, but tuition-free college and a Post Office Bank also fall under this heading. So do a Jobs Guarantee and a Debt Jubilee. Clearly, neither liberal Democrats nor conservative Republicans can deliver on such programs, because the two are different flavors of neoliberalism (“Because markets”). I don’t much care about the “ism” that delivers the benefits, although whichever one does have to put common humanity first, as opposed to markets. Could be a second FDR saving capitalism, democratic socialism leashing and collaring it, or communism razing it. I don’t much care, as long as the benefits are delivered. To me, the key issue — and this is why Medicare for All is always first with me — is the tens of thousands of excess “deaths from despair,” as described by the Case-Deaton study, and other recent studies. That enormous body count makes Medicare for All, at the very least, a moral and strategic imperative. And that level of suffering and organic damage makes the concerns of identity politics — even the worthy fight to help the refugees Bush, Obama, and Clinton’s wars created — bright shiny objects by comparison. Hence my frustration with the news flow — currently in my view the swirling intersection of two, separate Shock Doctrine campaigns, one by the Administration, and the other by out-of-power liberals and their allies in the State and in the press — a news flow that constantly forces me to focus on matters that I regard as of secondary importance to the excess deaths. What kind of political economy is it that halts or even reverses the increases in life expectancy that civilized societies have achieved? I am also very hopeful that the continuing destruction of both party establishments will open the space for voices supporting programs similar to those I have listed; let’s call such voices “the left.” Volatility creates opportunity, especially if the Democrat establishment, which puts markets first and opposes all such programs, isn’t allowed to get back into the saddle. Eyes on the prize! I love the tactical level, and secretly love even the horse race, since I’ve been blogging about it daily for fourteen years, but everything I write has this perspective at the back of it.


  1. divadab

    Is our children learning? Yes but mostly not at school? Credentialism is an indicator of decadence, imho. Credentialed careerists, short-termers, are too successful in the current system. Whatever, smart families will raise and train their youth despite systematic credentialism clogging up the works.

    Good article. Points to much of what ails us as a society.

        1. drumlin woodchuckles

          Even more rarely asked would be the follow-up question: if our children is learning, what is our children learning?

  2. efschumacher

    So if learning isn’t happening in schools, where is it happening? Indeed skilled work requires relevant skills, and as Matt Crawford tells us, Shop Work has been struck from curricula all across the land.

    If employers are unwilling to train apprentices from scratch, where on earth is this requisite learning actually going to happen then?

    1. Zamfir

      If you’re talking about the US, than that’s a very different issue than the article. By the standards of the article, the US scores at the top for both schooling and learning.

    2. jefemt

      Imagine a net-growth global population.
      Imagine a net loss of jobs over time with automation and AI.

      Imagine being an employer:
      -Post job vacancies, pluck the very best and brightest at your leisure.
      Flog the existing workforce until/ if when you finally fill a position.
      Every day that goes by without having that position filled accrues directly to the bottom line.

      The capitalist employer/ floggers in our town are lamenting no workforce— directly laid at the feet of skyrocketing rents and real estate prices. No workforce housing, converted to air bnb’s/ vrbo’s.

      The air bnb’s / vrbo’s are pretty full… white flight/ covid flight to the land of the wingnut
      no-distance, no-mask don’t-tread-on me anti-science all-about-me flock

      Pretty broken system. Well, not that pretty….

    3. drumlin woodchuckles

      If learning isn’t happening in schools, then what is happening in schools?

      ( Let us hope that some learning is still happening in schools).

  3. xformbykr

    This is intriguing to be sure. But I don’t yet comprehend a definition of what the harmonized learning numbers actually mean. There didn’t seem to be a definition statement in the post, nor in a brief check of the provided link.

    1. larrymotuz

      I agree, xformbykr. I too cannot fathom what this article is trying to say. Seeing (and understanding) what measures of ‘learning’ are being used would be more than helpful. I’m simply perplexed.

    2. Jason

      Yes. Took me a while to find out that it is basically making test scores comparable across many countries by creating conversion factors based on countries that have multiple test scores (e.g. PISA and the country’s own test). Excerpt:

      “The central intuition behind the methodology we use is the production of a conversion factor between international standardized achievement tests (ISATs) such as PISA, TIMSS,and PIRLS and their regional counterparts(RSATs) such as SACMEQ, LLECE, and PASEC. This conversion factor is derived by comparing scores for countries that participate in both an RSAT and an ISAT in a given time period, schooling level(primary and secondary), and subject. Following Altinok, Angrist and Patrinos (2018), we call these countries “doubloon countries.” A detailed description of each assessment is provided in Altinok, Angrist and Patrinos (2018). This approach puts all tests on a common scale enabling credible comparison across countries.These tests have been designed and scaled to be comparable over time since the late 1990s. Thus, there is no need for an intertemporal adjustment over the 2000-2017interval.”

  4. km

    Keep in mind that corruption is endemic in lots of countries’ educational systems, both on the side of the bureaucracy (jobs and contracts for political supporters) and students (cheating and buying grades).

    In Ukraine, it was inadvisable to go to a young doctor, because a medical degree in post-Soviet Ukraine was often purchased, but this was impossible in the Soviet days.

    1. km

      Also, upon meeting a recent graduate from a top Ukrainian law department, the question one asked oneself is “OK, so who is your father?”

    2. Felix_47

      I investigated India and Sri Lanka some years back for medical school for my kid as a safety. She made it to a US school. Many of the schools, esp the private ones, train and give degrees to Indians for only the export market largely the US. They are often in storefronts with perhaps reading lists but limited med school as we understand it here and they hand out a degree. If the students passed the ECFMG which is easy they were on their way to the USA. The English language test was a bigger hurdle. The large number of East Asian doctors make it clear that the US med school system is overkill. It is mostly OJT.

  5. John Emerson

    My bet is that in many poor countries, various sorts of outside aid key on the percentage of school age children enrolled in school, without attention to the quality of the education. Once the kids are signed up, the goal has been met.

    1. noonespecial

      The study above indicates that, “…the world is on track to achieve the goal of universal primary enrollment”.

      At the local, very rural level, in some countries to achieve this goal outside help may be required.

      As an example (which is not novel to rural peoples of the Global South), in a tiny area known as Coromoro in Colombia school children walk a few hours a day due in part to the collapse of a pedestrian bridge in 2016. The article linked has a pic. It is estimated by the school’s director that school dis-enrollment hovers at 70%. The local mayor and governor’s office have yet to develop and execute a solution.

  6. Anonymous 2

    Reminds me of a saying I came across many years ago:

    ‘Nothing has been taught until something has been learned’.

  7. liz

    So how is” learning” defined in this article.. and how is it differentiated from “schooling”?We need to know!

    1. jrkrideau

      Based an a quick look at the paper in Nature and an even shorter glance at the data it looks like they are using scores on standardized tests (PISA for example) . Not all countries/provinces will use the same tests so they have IIUC standardized the various scores into an HLO scores. The number crunching probably was a brute.

  8. JEHR

    When I came across the words “human capital,” it put me off. Please define “human capital.” Exactly.

    You cannot stop human beings from learning (unless they are comatose) because it takes place whether or not you can define it numerical of non-numerically. FTW indeed!!!

    Looks like April Fool’s joke.

    1. kees_popinga

      My former employer had a Personnel Dept. After a while they renamed it Human Resources. Then they changed it to Global Human Capital. I quit shortly thereafter.

  9. Mme Generalist

    I’m skeptical about the ability to extract any supportable claims from the data, as they appear to be largely based on extrapolations that are not clearly defined. They’re telling us what we’re looking at rather than showing it, seems to me at first glance.

    This reminds me of the data presented in support of No Child Left Behind that compared the US schooling outcomes with those of European countries and Japan. Turns out that those countries test kids (in some countries as young as 11) and track them into disparate courses, academic or vocational, whereas in the US we largely offer only an academic track and we offer it to all students for free. Of course our academic outcomes would be lower or, I would say, more noisy.

    Still, I can say that the innate problem-solving skills of my new employees went off a cliff around 2010, forcing me to have to completely rethink and retool training that had been effective for decades. And it was true that education level didn’t correlate to skill level at all. So something is definitely going on.

    Also, based on my son’s (now 22) experience, the schools were engaged in warehousing and indoctrination of all but the most “gifted and talented” kids back to at least 2005, although Jonathan Kozol’s work says it goes back to school desegregation and crept from disadvantaged schools out into the greater public schools over about 30 years.

    It’s all very worrying.

  10. Ghost in the Machine

    “Several studies have presented the argument that when human capital is measured by schooling, it fails to deliver the returns predicted by growth models. But when measured by learning, human capital is more strongly associated with growth (Pritchett 2006, Kreuger and Lindahl 2011, Hanushek and Woessmann 2012). Understanding growth is critical (Akcigit et al. 2020), and while good country studies exist (Fazzio et al. 2020), we still need comparable, global, timeseries data.“

    Of course, the fundamental concern is about growth, “the ideology of the cancer cell.”

  11. Ben Oldfield

    In the UK where the exam results are how the schools are measured, concentrate on teaching how to pass exams and so learning is secondary.

    1. Eustachedesaintpierre

      Yes & much of it came from Blairite targets which were initiated into to the public sector. My stepson was once married to a deputy head who talked about how her school was effected negatively by the measures, to which she added the totally unrealistic expectations of parents even if their little precious had the cognitive capacity of a house brick. Many of the parents who could moved house in order to get their kids into schools which by the above rating had a better score, while this in turn resulted in the left behind becoming ever more so. This was not helped by good teachers taking early retirement due to the increasing all round pressure.

      She worked very long hours as I remember which I believe led to a divorce, but later I heard that she gave up & left teaching altogether for 35hrs & a higher salary as an administrator for Halfords, which she apparently could accomplish almost in her sleep. Nice bright lass of Italian parents who was a great loss to learning.

  12. Arizona Slim

    I am reminded of what Mark Twain said about never allowing schooling to interfere with his education.

  13. Peter Dorman

    Years of schooling themselves should have little direct effect on anything. The main “outputs” are learning, socialization and credentialing; years are at best a proxy measurement. In 2008 I did a study for the International Labor Organization that looked at the evidence for consequences of child labor on “education” (and also health), and it found that effects on learning, measured by test scores and grade progression rates, were far greater than on just years of schooling themselves. Of course, education is not some isolated treatment that people get or don’t; it is part of a glop of interacting factors — social institutions, cultural norms, and personal characteristics — where each affects the others. When I wrote the review, I was still somewhat naive about the ability of econometric techniques to disentangle them. I no longer believe this.

  14. Fazal Majid

    Much schooling, specially in higher education, is a form of signaling. It builds social capital, but not necessarily human capital. That the former is often more valuable than the latter is proven by the competition for Ivy League and other prestige schools.

    1. Yves Smith

      I have to disagree. I learned to write at Harvard. I effectively majored in long papers. But what made the difference was my sophomore tutorial in an elite major (as in you had to apply and they only accepted 1/3 of applicants). We had 2 tutors for 5 students. Every paper I wrote got more comments than the original text. It was painful but I came out with the writing equivalent of hand-eye coordination.

      I also learned how to do what the French call l’explication de texte, which is close and exact reading of written passages.

  15. David

    “We observe a clear pattern of high and increasing school enrolment, while learning progress appears to be limited in many cases. For example, in the Middle East and North Africa enrolment rates achieved a high of 99% by 2010, up from 95% in 2000. In contrast, learning levels stayed low and remained the same (around an initial score of 380) from 2000 to 2015 in these regions.”

    Since we have no idea what “learning” is in this context, how it is defined or how it is measured, the statement is essentially meaningless. “Literacy”, for example, means something very different in Japan from what it means in Jamaica. It’s always been accepted that simply putting children is school doesn’t magically educate them: it simply creates the conditions for doing so. I wonder, in addition, to what extent this reflects the neoliberal crapification of education that we see in so many countries. In France, for example, a 16-year old school leaver has “learnt’, by any measurable standard, perhaps only what a 13 or 14 year old would have been expected to know fifty years ago.

    1. eg

      I take issue with any hazy comparison of what is or is not learned across generations. Often the assertion glosses over a former exclusivity now replaced by relative universality or ignores entirely new categories of “what is learned” which didn’t exist during the supposed golden age gone by.

      I assure you, children have not changed.

  16. Grebo

    The results have been inconclusive, in part owing to difficulties in measuring human capital.

    Not surprising; economists don’t even know how to measure capital when it is made of machines.

    Blair Fix compares Human Capital Theory to eugenics.

    This article is actually about the slightly less nebulous topic of education. Its conclusions will surprise no-one who ever went to school. I guess sticking in buzzwords like “human capital” helps get into journals.

Comments are closed.