Human Capital and the International Division of Labor

The international division of labor is the great polarizer of the world. Individuals, regions, countries, and macroregions that specialize in high-skill activities are systematically richer than their counterparts who specialize in low-skill activities. So, everyone wants to climb the ladder. The big question was how? How could poor nations overcome the entrenched advantages of the rich nations and grab a share of the high-skill work?

The reigning thesis at midcentury, around the time most nations gained independence from the imperial metropoles, was that the key was to industrialize at whatever cost. The Soviets went the furthest in using state power to force the pace of industrialization. But the Soviet Union was special—it was one of mere two nations on the globe (the other being the US) that could hope to survive and develop in near-autarchy. Modernizing elite coalitions in the rest of the poor world had neither the sort of power centered at the Kremlin at their disposal (although many attempts were made in different forms), nor could they hope to survive in autarchy. In the 1950s and 1960s, for instance, India was dependent on food imports from the US to survive in the face of intermittent famine and poor agricultural productivity. And the Nehruvian elites may have inherited the formidable power structure of the British Raj. But this massive edifice of power were simply not up to the task of completely passifying, much less enslaving, the entire country, as Stalin had done in the 1930s. So, what was the formula to be?

The consensus formula for industrialization at the midcentury moment was tariff protection and import substitution, together with mixed-economy policies that reserved the strategically-important “commanding heights” of the heavy industry, capital goods, and infrastructure to the state, and left what were seen as the less important consumer goods and services to private enterprise. Ultimately, this formula just did not work. The tariff walls, import controls and onerous regulations conspired with the small market size of third world countries to condemn their firms to incompetence and their cleverest citizens to rent-seeking. Whatever the theoretical merits of nascent industry arguments, in practice, it simply did not work. Between the 1970s and the 1990s, beginning with Chile and ending with Vietnam, it was progressively abandoned everywhere in the southern half of the globe.

The new model that emerged was in line with the neoliberal consensus that had emerged in the core industrial nations from the crises and counterrevolutions of the 1970s and 1980s. The elite coalitions atop third world states had themselves transformed. The anti-western postcolonial generation had given way everywhere to a less moralizing generation that had soured on midcentury socialist ideals, which turned out to be more moral in theory than practice. Instead, the new generation of third world elites were attracted to the glimmer of Western affluence reinforced by the spread of televisions and cable. To hell with socialism and social justice, they just wanted the American way of life, even if just for themselves.

This new formula was based on the theory that what had gone wrong was incentives. The midcentury statist regimes distorted economic incentives of everyone from bureaucrats to politicians to firms to individuals. The way out of the mess was to kick out the state from the economy and let the market do its magic. This was the heart and soul of the Washington Consensus. Sooner or later, it was embraced by all third world elites. By the late-1990s, even former socialist strongholds were privatizing and liberalizing.

There were some successes. At first look, the successes were those who focused on export-oriented industries. In reality, exposure to the rigors of global market competition was necessary but not sufficient for success. For every success in export-oriented industrialization in Asia, there were dozens of failures, in Asia as elsewhere. For a while, much of the trouble was blamed on poor macroeconomic management and corruption, ultimately due to entrenched structures of political economy. The technocratic solution proposed of the Washington institutions, sometimes imposed with great coercion when the beggar had run out of options, was ‘structural adjustment’: market-friendly reforms, bureaucratic transparency and rules-based governance, fiscal and monetary discipline.

But just like the midcentury formula, neoliberal market reforms and structural adjustment also failed to help most nations industrialize and modernize. Clearly, there was some other blockage besides those related to market and state institutions. What was this blockage?

The question can be approached in different ways. One can look at the deep structures of family and kinship institutions are the ultimate source of the blockage, with the status of women playing the role of the mediator. But perhaps before reaching for anthropological particularism to explain large-scale order, we should pay attention to middle-range theories that provide a compelling proximate account of the performance of the developmentalist states.

Economists had argued for decades that human capital is a crucial conditioner of economic performance. More recent work has made considerable progress in operationalizing this middle-range theory. The literature centers on the work of Hanushek and Woessmann, among others. In a series of papers and monographs, these scholars have shown that educational attainment (average years of schooling) is a poor proxy of the average level of cognitive skills of a population because the vast majority of kids in third world countries are not learning much at all. Instead of years of schooling, the average level of the human capital of a population is better proxied by learning outcomes, preferably as measured by internationally-standardized tests (PISA, TIMMS, etc).

One of the most compelling pieces of evidence is that the great divergence between Latin America and East Asia—which were at the same level of per capita income in 1960, whereas now the latter is ten times richer—which remains unexplained if we use educational attainment, is fully explained once we switch to learning outcomes as the measure of human capital. Here’s the money shot (from Hanushek and Woessmann, Science, 2016). As you can see, Asian countries fall well above the regression line and LatAm nations fall well below it if one used years of schooling (left graph), but the residuals are totally random, and the fit much better, if one uses learning outcomes instead (right graph).

So, we have a very powerful proximate explanation of why some nations in Asia made it, while the rest of the global south did not. It’s not exactly rocket science: countries that invested heavily in lifting up the bulk of their population in health and education (China, Korea, Taiwan) developed the cognitive skills to succeed; countries that failed to so, and that instead poured their resources into projects that favored elites, like India’s IITs, failed to develop the cognitive skills at the scale required to take-off.

National economic performance is very strongly conditioned by cognitive skill-sets. Hanushek and Woessmann show that a standard deviation unit higher aggregate score predicts 2% higher growth in per capita income per year. That’s enough to account for the entire divergence between, say, East Asia on the one hand, and Africa, South Asia, and Latin America on the other. Put simply, global polarization in standards of living is explained to first order by the global distribution of cognitive skills:

It follows immediately that countries with great cognitive skills should win the struggle over the global division of labor. The only paper that has investigated this question is Ciccone and Papaioannou (2009). They showed that countries with higher educational attainment do indeed specialize in industries that are skill-intensive. I asked Woessmann over email if someone had updated their work using learning outcomes. Unfortunately, no one seems to have done so yet. This led me to investigate the question myself.

We’re interested in pinning down the degree to which skill-differences between countries can account for differences in employment shares between less and more skill-intensive industries.

We obtain the global distribution of cognitive skills from the World Bank’s Harmonized Learning Outcomes (HLO) dataset. We select the HLO for math at the secondary level because it seems to contain the strongest signal of cognitive skills. This will be our feature or predictor.

We obtain employment and wage levels by country and industry (we average over the decade for which data is intermittently available for most countries) from UNIDO. We use wage levels in the US to sort the industry into wage terciles. For the cross-country regression, our response is the difference between the proportion of a country’s labor force employed in (1) the highest third or tercile of industries and (2) the lowest tercile of industries, when industries are sorted on the average US wage level (our proxy of the skill-intensity of the industry). The larger and more positive this difference, the more the country specializes in high-skill industries; the lower and more negative, the more the country specializes in low-skill industries.

If the theory is correct, we should find a strong positive relationship between our measure of the average level of cognitive skills and skill-biased specialization. We throw out countries that are large oil and gas producers, as this is a clear confounder. The next graph present our main result.

We find a very strong positive relationship (r=0.67, P < 0.001) between our proxies for the average level of cognitive skills and the skill-bias of countries’ employment profile. Our results are congruent with those in Ciccone and Papaioannou (2009). The interpretation is very straightforward because both cognitive skills and employment shares are slowly-moving structural parameters: the international division of labor, the great polarizer of the world, is determined to a great extent by the accumulated cognitive skills of the bulk of nations’ populations.

Some very serious implications follow for both developed and developing nations, and for high geopolitics. First, nations like India need to invest a whole lot more in the general health and educational outcomes of their kids and their mothers—flashy airports and software isn’t gonna cut it.

Second, developmentalism does not end at the technological frontier. Even for countries at the cutting edge, the race is neverending because others are always catching up. Developed nations too must therefore pay more attention and invest more in developing both the broad skill sets of their population, and reform their immigration systems to make themselves more competitive at the upper levels of skill. Tariff walls ain’t gonna cut it—if we hunker down behind tariff walls, we may survive but we will surrender the rest of the world to Chinese domination.

Finally, the foreign policy community needs to come to grips with the fact that Chinese supremacy in cognitive skills is such that it completely overwhelms China’s demographic decline. And the scale of oncoming cohorts is such that we’ve only seen the tip of the iceberg yet. At the minimum, we should not launch a cold war can we cannot hope to win. China’s literally No 2, while the US lags behind Russia in average cognitive skills. And China’s population is 1.4bn while the US’s is 0.4bn. Maybe it’s time for the blob to wake up from their unipolar dreamworld.