An educated labour force is key to ‘modern’ economic growth. While economists have firmly anchored human capital in the ‘new’ growth theory since the 1990s, they have struggled to demonstrate a positive relationship between educational expansion and aggregate productivity growth in the developing world. At the micro level the link is clear: on average, the higher the education level of an individual, the higher the income. At the macro level, however, the association between higher educated societies and higher productivity levels remains very weak. Since economic theory predicts positive social returns to schooling, the lack of such returns in developing economies is puzzling: why does the growth theory of the ‘West’ not seem to hold for the ‘rest’ of the world? Or, as most memorably phrased by Lant Pritchett (2001) in light of the rapidly rising educational attainment rates in the developing world: where has all this education gone?
So far, two main explanations have been offered for this ‘micro-macro paradox’: bad institutions and poor educational quality. The bad institutions view suggests that higher educated cohorts engage in rent-seeking activities, allowing them to earn high incomes through activities that do not add much to the overall growth of a developing economy. The poor educational quality view, in contrast, points fingers to the schools themselves: attainment rates are indicators of schooling quantity, but it is the quality of schooling that determines its economic gains. In other words, people may spend more years in school and still not learn much (Pritchett 2013). While the signalling effects of a higher level of education may raise individual incomes, it does little for aggregate productivity growth. This latter view was strengthened in recent years by cumulating empirical evidence documenting a much stronger correlation of measures of educational quality with productivity indicators (Hanushek and Woessman 2012). However, both explanations miss an important part of the picture.
In our paper (Frankema and van Waijenburg 2019), we take a long-term historical perspective to shed new light on this puzzle. Our approach differs from the existing literature in two ways. First, we consider a longer timeframe for assessing the economic gains of the ‘global schooling revolution’, which started in the mid- to late-19th century in most developing economies (Barro and Lee 2015). Second, we focus on a different indicator of human capital accumulation, namely the long-term trends in skill premia for various blue- and white-collar occupations. The skill premium is defined as the wage premium of a skilled worker over the wage rate of an unskilled labourer working in the same sector. Using insights from the economic historical literature, we interpret the relative magnitude of the skill premium (in the long run) as an indicator of the relative scarcity of a particular skillset and the effectiveness of educational institutions in tailoring supply to demand. Indeed, the historical record for Western economies suggests that over the course of many centuries, skill premia tend to move towards a lower bound ‘steady state’ when educational institutions work well (Goldin and Katz 2008, van Zanden 2009). We know remarkably little, though, about the comparative levels and trends of skill premia in developing economies.
Our paper is the first to explore the long-run development of skill premia in Africa and Asia, and we hope it inspires further historical research into the price structure of labour skills in the developing world. We focus on the skill premia of adult male urban workers of indigenous descent. Based on a large body of colonial and post-colonial sources, we create a new skill premium dataset for 50 African and Asian countries for the period 1870-2010 (database available here). Our dataset focuses on five occupations: carpenter, electrician, car mechanic, (entry-level) clerk, and bank teller. While the skill premium trends follow more or less similar patterns for all five occupations, we will illustrate the main ‘stylised facts’ that result from our study by presenting evidence on the carpenter’s premium.
Figure 1 shows our carpenter premia for African and Asian countries in the early stages of the schooling revolution, and reveals the first two new stylised facts. First, around 1900, artisanal labour was far more expensive in Africa and Asia than in Western economies. Only in Japan, which was by then well into its industrial ascent, were carpenter premia at comparable levels to those in the West. Hence, Western economies had a major cost advantage when using skilled workers in productive activities. Second, skill premia for carpenters were, on average, much higher in Africa than in Asia. While an average carpenter in Asia commanded a premium of roughly 100-200%, this premium was around 200-500% in Africa. We find similar gaps in skill premia between Asia and Africa for most other occupations. This difference in magnitude between the two regions in the early 20th century raises an important new historical question: how and why did pre-existing systems for the transfer of skills differ across Africa and Asia?
Figure 1 Skill premium for carpenters, c. 1900 (%)
Source: Figure 2 in Frankema and van Waijenburg (2019)
Figure 2 places our new African and Asian series in a long-term comparative perspective, and reveals the third stylised fact. In both Africa and Asia, skill premia have fallen dramatically over the course of the 20th century, ultimately converging to Western ‘steady state’ levels. Although the pace of this free-fall varied from place to place, the ‘great convergence’ in skill premia was a universal phenomenon. The economic and historical significance of this convergence is profound: it means that the relative costs of hiring skilled labour in a diverse set of economic activities have fallen dramatically and that a structural transition has taken place in Asian and African labour markets over the last century. While a lower skill premium in itself is not a sufficient condition for ‘modern’ economic growth, the Western experiences suggests it may well be a necessary one.
Figure 2 African and Asian carpenter premia in global historical perspective, c. 1250-2000 (5-year average, %)
Sources: Figure 3 in Frankema and van Waijenburg (2019)
What drove this massive decline in artisanal and white-collar skill premiums in the global South? Barro and Lee’s (2013) recently expanded education dataset allows us to do a rough test of the relationship between the fall in skill premiums and the expansion of school attainment rates. Figure 3, where each observation is a half-decade value for a single country, plots this relationship. The figure points to a negative relationship between the level of the skill premium and the share of the (male) population that has received at least some basic education.
Formal testing of this relationship yields mixed results. While our baseline regressions and the ones in which we control for country-fixed effects show a strong negative relationship between skill-premium levels and educational attainment rates, the coefficient of the education variable is reduced to zero when we add both time- and country-fixed effects. It is likely, however, that this result is driven by the weaknesses of the education variable as a proxy for supply of skilled labour in a particular occupation. To be useful for identification under time- and country-fixed effects, the half-decade to half-decade variation within countries in the supply of carpenters needs to correspond closely with the concomitant variation in the general levels of education. It is hard to imagine that such a close short-run correlation ever existed. Additionally, alternative explanations for the decline in skill-premiums – the forces of globalisation, de-skilling, and changes in (colonial) labour market policies – had rather diverse and ambiguous effects on the observed trends in skill-premiums and, on their own, cannot account for the universal pattern we documented. In other words, while several forces co-determine the pace and magnitude of the decline, we argue that the spread of mass education was the only real conditio sine qua non.
Figure 3 Carpenter skill premium and the relative supply of male educated workers, 1870-2010
Sources: Figure 8a in Frankema and van Waijenburg (2019)
What does this negative relationship imply for the puzzle on the weak relationship between education and growth? And what policy lessons for future educational investment may be drawn from a historical comparative perspective? First of all, as the historical path towards Western industrialisation illustrates, there is a long run-up to the transition to ‘modern’ economic growth. In this run-up phase, a range of pre-conditions need to be met. It is highly likely that the schooling revolution has brought all of the Asian and African economies in our sample a step closer to sustained growth. In this sense, the economic gains from educational investments over the last century, even if still of insufficient quality, deserve a more optimistic evaluation. Of course, this finding does not mean that improvements in the quality of education have become redundant – on the contrary.
The fact that individual returns to the acquisition of skills have fallen so dramatically in relative terms has consequences for government policies targeting education systems and the labour market. If a skilled construction worker or a white-collar worker can barely make more than an unskilled worker who has spent fewer years in school, the incentives for paying high school fees decline and the incentives for migrating to societies with higher skill premia rise. In other words, even though the ‘Great Convergence’ in premia has raised possibilities of economic diversification and Smithian growth locally, it has not put low-income countries on an equal footing in the global market for skills. The wider implication of a long-term perspective, then, also points to the importance of collecting and comparing data on the remuneration of labour skills worldwide. For instance, what we do not know yet, is whether the trends that we uncovered in our study also apply to the higher segments of the labour market. What are the relative costs of hiring workers who can operate high-tech equipment, who understand computer programming, or who can integrate artificial intelligence into business and production systems? Is the development of a knowledge economy feasible in Africa if the brightest and best educated workers can find jobs elsewhere? For the design of policies governing the local and global labour markets of the 21st century, such questions warrant serious attention.
Barro, R and J-W Lee (2015), Education Matters: Global Schooling Gains from the 19th to the 21st Century, New York, NY: Oxford University Press.
Goldin, C and L Katz (2008), The Race between Education and Technology, Cambridge, MA: Harvard University Press.
Frankema, E and M van Waijenburg (2019), “The Great Convergence. Skill Accumulation and Mass Education in Africa and Asia, 1870-2010”, CEPR Discussion Paper 14150.
Hanushek, E A and L Woessman (2012), “Do Better Schools Lead to More Growth? Cognitive Skills, Economic Outcomes, and Causation”, Journal of Economic Growth 17(4): 267–321.
Pritchett, L (2001) “Where Has All the Education Gone?”, The World Bank Economic Review 15 (3): 367–91.
Pritchett, L (2013), Rebirth of Education. Schooling Ain't Learning, Washington: Center for Global Development.
Van Zanden, J L (2009), “The Skill Premium and the ‘Great Divergence”, European Review of Economic History 13(1): 121–53.