Convergence, China and Colonisation

Some opening thoughts about divergence:

If differences in output between nations can largely be attributed to human capital then, given knowledge is non-rivalrous, we should expect knowledge to flow between countries. In the long run, all countries ought to converge onto the same steady state determined by the stock of knowledge in the world. However, this is not what we see in the data. Only certain countries seem to adopt modern technologies whilst others take decades to do so. Hence, convergence must be conditional.

Solow model of growth tells us that convergence should occur when countries have the same saving rate, population growth rates, and efficiency of labour (human capital).

What explains convergence?

Baumol examines data for 16 advanced nations between 1870 and 1993 and finds a strong inverse correlation between initial productivity levels and subsequent productivity growth. This indicates remarkable convergence of output per labour hour among industrialized nations, with nearly all leading free-enterprise economies closing the gap with the productivity leader.

A key explanation for this convergence is that productivity-enhancing measures, such as innovation and investment, exhibit public good characteristics meaning that the benefits spillover to other industrialised economies. Innovation is largely non-rivalrous and partially non-excludable so any effective new growth policy or innovation will soon be adopted by other industrialized countries. Furthermore, investment is akin to a public good because reallocating capital toward high-productivity sectors increases imports of other goods, which in turn raises wages and productivity in trade partner countries, spreading the benefits internationally.

Do all countries converge to similar levels?

However, this convergence pattern disappears when applied to a wider set of countries. Using Summers and Heston’s dataset of 72 countries, Baumol examines real growth rates from 1950 to 1980 and finds no clear negative relationship between initial income levels and growth. In fact, the regression line has a slight positive slope, indicating that some of the poorest countries grew most slowly.

Summers et al. (1984) categorise the countries into four groups: industrialised economies, centrally planned economies, middle-income market economies, and low-income countries. Examining the period between 1950 to 1980, Baumol observes stark differences in growth rates - 3.1% for industrialised countries versus just 1.5% for low-income countries - and finds no evidence of convergence within any subset. This implies the existence of multiple “convergence clubs” rather than a single global pattern.

What explanation does Baumol give for lack of convergence?

What is needed is an explanation of why poorer, less developed countries have benefited to a relatively small degree from the public good properties of the innovations and investments of other nations. Baumol suggests that it is lack of education and skills that prevent the effective adoption of advanced capital-intensive technology. However, institutional weakness, poor infrastructure and underdeveloped financial systems may also play a role.  Without these complementary capabilities i.e. strong property rights, adequate transport and electricity, skilled labour, and developed capital markets, the benefits of global innovations will not be effectively exploited, despite the public good nature of innovation. For example, rent seeking elites may simply block the adoptions of new technologies if they feel it is going to undermine their position.

How did globalization contribute to convergence?

Williamson (1996) argues that globalization was the primary driver of convergence in the late 19th century through the open economy forces of trade and mass migration.

Firstly, he shows that mass migration drove convergence, not through the human-capital channel, but through the transfer of workers from labour-abundant Europe to the United States. Emigration from European nations like Ireland and Sweden made labor scarcer at home, thereby raising real wages and living standards in the sending countries. At the same time, immigration depressed wages in the receiving countries (for example, US wages fell by 9%). Williamson’s counterfactual analysis suggests that mass migration accounted for 70% of real wage convergence, 75% of GDP per worker convergence, and 50% of GDP per capita convergence from 1870–1910.

What was the other channel?

Alongside this channel, it was also commodity market integration that drove convergence among OECD countries. Falling transport costs reduced price gaps between countries (commodity price convergence), which encouraged trade in goods between nations. This trade shifted production toward each country’s comparative advantage and effectively equalised returns to factors of production across borders (factor price convergence), raising wages in poorer economies and reducing them in richer ones.

Complete Williamson (1996):

Compared to the forces of international trade and labor mobility, Williamson finds that education and human capital played only a very minor role in late nineteenth-century convergence. In fact, the contribution of schooling to GDP per worker growth was never statistically significant. This is unsurprising given that the technologies of the 1890s were far less skill-intensive than those of the late twentieth century.

Williamson shows that convergence halted between 1914 and 1950, marked by divergence in real wages and slower GDP per worker-hour convergence. He attributes this to global deglobalization during and after the First World War, such as autarchic policies, trade barriers and migration restrictions, which disrupted the open-economy trade and migration that was driving convergence.

The gradual reconstruction of world commodity markets since 1950 then contributed to the resumption of convergence.

Williamson succeeds in showing that globalization drove convergence among OECD countries (Europe, its settler offshoots, plus Argentina and Brazil) however such convergence is limited to this subsection countries. When Pritchett (1997) includes the Third World and Eastern Europe in his analysis, he actually demonstrates global divergence in GDP growth. Thus, globalization-driven convergence is very real but also limited in scope to advanced economies.

Who provides compelling evidence for divergence?

Pritchett (1997) provides compelling evidence of economic divergence in modern economic history, particularly between developed and developing nations, which disproves the theory of universal convergence in incomes.

Firstly, he shows how divergence rather than convergence is the dominant feature of modern economic history. From 1870 to 1990, the ratio of per capita incomes between the richest and poorest countries increased by approximately a factor of five (from 8.7 to 45), and the standard deviation of natural log GDP per capita across all countries has increased by 60% since 1870. From 1980–1994, the average growth rate for advanced countries was 1.5%, compared to 0.34% for less developed countries, indicating continued divergence.

What is Pritchett’s greatest contribution?

The article’s greatest contribution to proving divergence beyond doubt is the establishment of a minimal plausible GDP per capita, estimated at $250 in 1985 dollars. Anything below such a benchmark, alongside never having been observed historically, would be inconsistent with plausible nutritional intake or sustainable population expansion, falling far below extreme poverty lines.

The world’s richest country in 1960, the US, increased their per capita income fourfold between 1870 and 1960, implying a growth rate of around 1.7% per annum, and convergence theory predicts that poorer countries should grow even faster than this as they catch up. However, 42 countries in 1960 had per capita incomes below $1,000, less than four times the $250 minimal plausible GDP. If these countries had grown as fast as the US, their 1870 incomes would have been below the benchmark and thus empirically impossible. Hence, these countries must have grown slower, contradicting the predictions of convergence theory.

What does Pritchett say about convergence clubs?

Looking only a sample of today’s advanced capitalist countries, Pritchett finds that they all displayed remarkably similar long-run growth rates since 1870, leading to considerable convergence in absolute income levels. The poorest six in 1870 generally had the fastest growth rates, and the richest five had the slowest. However, this observed convergence is almost tautological due to the sample selection, as it includes only countries that are rich now. It also excludes examples of divergence, such as countries that declined from relative riches to poverty, such as Argentina.

Looking at all other countries, Pritchett finds much slower growth on average than richer countries between 1960 to 1990. However, unlike the developed sample, there is no single pattern. Some countries experienced rapid growth (such as Korea at 6.9%) whereas others stagnated (28 countries had growth below 0.5% per annum) and many declined (such as Mozambique at -2.2% per annum). This lack of consistent catch-up growth again contradicts convergence theory.

Who explores the catch-up thesis?

Abramovitz (1986) explores the catch-up hypothesis, which posits that countries with lower productivity levels tend to experience faster productivity growth rates, resulting in convergence over time. This is because they can just adopt all the existing technologies used by productivity leaders paving the way for rapid advancement.

Using productivity data for 16 industrialized countries between 1870-1979, Abramowitz shows that productivity variance among countries steadily declined and that rank correlations between initial productivity and subsequent growth rates became strongly negative. This indicates that countries starting further behind tended to grow faster and supports convergence.

What is Abramowitz’s crucial qualification?

However, Abramowitz makes a crucial qualification to this hypothesis: while technological backwardness creates the potential for rapid productivity growth and economic growth, whether or not a country realizes that potential depends heavily on its “social capability.” This encompasses technical competence, such as education and literacy, alongside the relevant political, financial and industrial institutions, like experience with large-scale enterprise. This social competence also includes attitudes, particularly an openness to change: to competition, the establishment of new firms, and the sale of new goods. A country’s potential for rapid growth is strongest not simply when it is technologically backward but only when it is technologically backward and socially advanced.

What was the importance of the slave trade on Africa?

Nunn’s (2008) paper examines the importance of Africa’s slave trades in shaping subsequent economic development. He uses a variety of historical data sources to construct measures of the number of slaves exported from each African country between 1400 and 1900. Running IV regressions, Nunn then shows a robust, statistically significant negative relationship between slave exports per area and current economic performance, measured by real GDP per capita in 2000. Thus, the African countries that are poorest today are the ones from which the most slaves were taken. This correlation remains statistically significant even after controlling for various factors such as the identity of the colonizer, geographic and climatic variables, and natural resource endowments.

The paper also addresses the concern that initially underdeveloped countries might have selected into the slave trades, which would make the observed relationship spurious. However, Nunn shows that it was actually the most developed and densely populated areas of Africa that tended to be most affected. These more prosperous areas had the institutional capacity to facilitate trade and efficiently supply large numbers of slaves.

Finally, Nunn also explores two possible channels of causality.

What is the first channel?

Ethnic fractionalization is frequently associated in the literature with reduced social cohesion, weaker domestic institutions, and poorer provision of public goods, all of which hinder economic development. Thus Nunn hypothesizes that the slave trade fostered internal conflict which in turn weakened ties between communities and impeded the formation of broader ethnic identities. He then reveals a strong positive relationship between historical slave exports and current ethnic fractionalization in African countries to support this channel.

What is the second channel?

Secondly, due to the political instability it brought, Nunn suggests that the slave trade led to the weakening and underdevelopment of political structures, which he supports by demonstrating a clear negative relationship between slave exports and 19th-century state centralization. Due to the lack of significant political development during colonial rule, Africa’s postindependence leaders in slave-trade-heavy nations inherited states that did not have the infrastructure necessary to extend authority and control over the whole country. Many were, and are still, unable to collect taxes from their citizens and so unable to provide a minimum level of public goods and services. Indeed, the income gap between low-slave-export and high-slave-export countries significantly widened after the late 1960s and early 1970s, coinciding with widespread independence.

What held China back? First factor: institutions.

The one civilization that was in a position to match and even anticipate the European divergence was China (Landes, 2006).

First, China lacked a free market and institutionalized property rights. The Chinese state frequently interfered with private enterprise: restricting trade, setting prices, reserving key sectors for state ownership, and discouraging private wealth accumulation, which suppressed innovation and capital accumulation. The Ming Dynasty, for example, banned overseas trade, punished any unauthorized travel and enforced rigid social hierarchies and professional immobility. This undermined economic dynamism and prevented a merchant class from growing powerful.

Next two Chinese factors:

Furthermore, Chinese society was highly totalitarian. The all-controlling state bureaucracy instilled a culture that valued stability and conformity, and which viewed innovation as a threat to the established order. Technological breakthroughs like water-powered spinning machines and blast furnaces were achieved some 500 years before Europe. However, China failed to exploit the economic potential of these inventions due to lack of institutional support.

Furthermore, cultural isolationism prevented China from capitalizing on European advancements. As European contact increased in the 16th-18th centuries, the Chinese elite frequently rejected foreign knowledge as they saw it as inferior or threatening Chinese cultural superiority. This meant that potentially useful European developments in science and industry were either downplayed or dismissed.

Final differences between China and Europe:

Finally, Europe’s political fragmentation and competition among states created an environment where innovation was rewarded. Rulers had to tolerate innovation and avoid excessive predation to survive because an inventor could always escape to another region with their new technology.  China’s imperial unity, on the other hand, meant that escape from repression or rejection was impossible so a single emperor’s disapproval could halt technological progress across the entire empire. There was no incentive for local Emperors to admit innovation and risk disrupting the status quo.

In imperial China, merchants were seen as the lowest class in society and viewed as selfish, immoral and exploitative. As a result, they enjoyed little social respect. Furthermore, this wasn’t just cultural but backed up by law: during the Ming dynasty, for example, merchants were banned from taking civil service exams and weren’t allowed to wear luxury materials like silk or fur. Thus, even if they grew rich, merchants were blocked from gaining political influence or rising socially. This institutional discrimination discouraged commercial ambition and prevented a powerful merchant class from emerging.

Thus, while China was technologically advanced for centuries, its cultural triumphalism, fear of innovation, and failure to protect economic freedoms, explains why it fell behind the West and missed the Industrial Revolution. (Lanes, 2006)

Why did Japan develop before China?

In 1873, China exported three times as much raw silk as China but by 1930 Japan exported three times as much as China. Using this case-study of the silk industry, Ma (2004) seeks to explore why Japan developed before China and argues it was down to Japan’s successful implementation of induced institutional and technological innovation.

Following the Meiji Restoration, Japan’s leaders embraced Western science, invested in public infrastructure – such as research centres, transport and education – and attempted to foster private enterprise as part of a modernization strategy. This “induced innovation” enabled the development and rapid diffusion of high-yield technologies.

In contrast, China resisted modernization, blocked institutional and technological improvements, such as railroads and steamships, and refused to invest in education and public goods. They also continued to focus on military and state-owned factories, ignoring rural areas and not caring about private enterprise. As a result, Chinese silk farmers kept using older, less effective methods, and their cocoons started to fall behind in both quantity and quality.

How did the length of colonial rule affect development?

Feyrer and Sacerdote (2009) used a novel IV estimation strategy to provide the first causal estimates of the impact of the length of colonial rule on subsequent development. They ague that how early an island was discovered and colonized was, in part, determined by its location relative to prevailing wind patterns. Crucially, these wind patterns likely do not affect long-term development other than through the island’s date of discovery. The authors therefore use these wind vectors to estimate the causal effect of the length of colonial rule on subsequent development.

Their first-stage estimates show that stronger westerly winds are associated with earlier discovery and more years under colonial rule. Then, according to their second-stage estimates, they find that the length of colonial rule has a positive effect on per capita income in 2000.

This result may seem surprising because it suggests that colonial rule was actually good for economic development. However, it should be noted that the estimated effects are for the length of colonial rule conditional on being colonized, not the average effect of being colonized relative to not being colonized.

Why would colonial tenure be positively related to modern economic outcomes?

Any explanation must allow for events which happened centuries ago to matter today, pointing to the importance of institutions given their longevity. The authors emphasize  the promotion of trade as, for many of the islands, the very concept of households generating a surplus and exchanging this in exchange for money and goods arrived with the colonizers. Colonial island regimes needed to make their colony profitable or at least self-sustaining so that they would make more from tax revenue than they were spending through subsidies. As a result, colonial governments introduced and expanded the production of export commodities like sugar, coffee and tobacco to generate cash flow. They also played a large role in building schools (which raised human capital), roads and drinking-water infrastructure, and helped establish property rights and judicial courts.

Interesting finding on colonisation:

Interestingly, later years of colonialism are associated with a much larger increase in modern GDP than years before 1700. This suggests that colonialism in the post-enlightenment era led to more efficient and beneficial institution transfer than colonialism previously. In fact, an additional 100 years of post-enlightenment colonialism is associated with  64% higher per capita income in the colony. This is not to say that colonialism pre-1700 was detrimental to modern GDP but, given the human toll on the early natives, it is not unreasonable to consider pre-1700 colonialism as a net negative.

Is there a link between European population during colonisation and development today?

Using OLS, Easterly and Levine (2012) find a positive and statistically significant relationship between income today and the proportion of Europeans in a country’s population during the colonial era. They also confirm AJR’s prediction that indigenous population density was inversely associated with the share of European settlers during colonization. However, they point out, that this does not show that political institutions are the principal channel through which Europeans shaped the reversal of fortune.

Crucially, Easterly and Levine’s results show a clear positive effect on development for even very small European settlements, which is inconsistent with AJR’s binary extractive vs inclusive narrative. To be precise, their OLS estimates indicate that once European settlement is above 4.8%, the settlers have a positive effect on development today compared to no European settlement. This suggests that any adverse effects arising from the extractive institutions created by small colonial settlements were more than offset by other things that Europeans brought, such as human capital, technology, institutions and integration into global markets, which had lasting, positive effects on economic development.

Only colonies with less than 4.8% European settlement would have suffered (developmentally) from colonization. This is still the vast majority (89 vs 35 in their sample).

What was the scramble for Africa?

Michalopoulos and Papaioannou (2016) argue that the arbitrary political boundaries drawn by European powers during the “scramble for Africa” is a significant factor contributing to underdevelopment in the continent. These artificial borders, which persist today and were often created without knowledge of local divisions, partitioned ethnic groups across different states. This, the authors argue, has contributed to higher levels of civil conflict across Africa because partitioned groups can draw support from co-ethnics on the other side of the border, reducing the opportunity cost of engaging in armed rebellions. Governments, recognising this potential, tend to view partitioned groups as threats, leading to political exclusion, discrimination and violence against these groups. The effects of partitioning can also spill over to neighbouring regions, either through the spread  of wars and conflict or through large refugee flows which can be destabilizing.

How do they support this thesis on borders? Why is it convincing?

The authors show that partitioned and non-partitioned groups enjoyed similar outcomes before the borders were drawn. Then, using the Scramble for Africa as a quasi-natural experiment, they demonstrate that partitioned ethnic groups experienced significantly higher levels of civil conflict and lower individual well-being compared to non-split groups, even after controlling for other geographic and historical factors. This indicates causality. Thery also show that members of partitioned ethnicities have fewer household assets, poorer access to utilities, and worse educational outcomes – even when compared to non-split groups within the same country.

The authors argue that the drawing of these arbitrary borders, rather than colonisation itself, was a major channel through which Europeans harmed African development.

This is an intuitively convincing narrative given that many places around the world with ethnic partitioning, such as India & Pakistan or the Dombas between Ukraine & Russia, tend to exhibit higher levels of conflict and violence.

How does Austin criticise the reversal of fortunes thesis?

Austin (2008) argues that Acemoglu, Johnson and Robinson’s (2002) reversal of fortunes thesis oversimplifies the causation of economic outcomes by “compressing” different historical periods and development paths.

For example, AJR compare just two moments, 1500 and 1995 - nearly half a millennium apart, which ignores all the significant historical processes that occurred between these dates. The nature and content of key analytical categories like “colonial rule” and “property rights” were not stable across this long period but fluctuated significantly.

This compression of history also occurs horizontally due to their research design’s binary assignment of colonies which ignores the true diversity of historical paths. They divide all European colonies across five continents into just two groups: ‘settler’ and ‘non-settler’ colonies, with all African colonies falling into the latter. However, this is a gross oversimplification as there were both colonies with and without European settlement in Africa.

Two more issues with AJR’s thesis for Africa:

Furthermore, the depiction of all colonial rule in Africa as purely extractive is historically inaccurate. For example, in many non-settler colonies, it was in the interest of the colonial administrations to support and encourage African economic enterprise rather than solely extracting from it. Thus, there is far more variation in colonial experiences within Africa than AJR accommodates.

Finally, Austin points out that scholars such as AJR and Nunn neglect African agency, particularly when discussing the slave trade. European colonizers were rarely in a position to dictate the terms or choose the rate of extraction, as is often believed, but most slaves were purchased from African rulers and traders. The need for labour coercion was a factor of pre-colonial land-abundant conditions, not solely a result of the slave trade. Furthermore, it is incorrect to view institutions in Africa as simply imposed and maintained by colonial governments. Instead, African producers often resisted colonial attempts to coerce or direct their economic activities, especially in non-settler colonies where they retained land control. Colonial administrations often had to adapt their policies in response to African economic success and political action.

He also criticizes the data used by AJR as incomplete or circumstantial, and challenges the use of urbanization metrics as a proxy for development.

What is an outlier for colonisation?

Botswana is the standard outlier cited against the AJR-style "colonisation produced extractive institutions" story: despite being colonised, it achieved sustained growth and democratic stability after independence, with one of the fastest GDP-per-capita growth rates in the world from the 1960s onwards.

AJR (2003) attribute this to the survival of pre-colonial Tswana institutions - particularly consultative assemblies - combined with light-touch British indirect rule that left them largely intact. But this reading is strained. Botswana was one of the poorest countries in the world at independence in 1966, and serious growth only began after the discovery of large diamond reserves shortly afterwards; the institutions didn't produce growth before the diamonds, growth started when the diamonds did.

The simple reading is that Botswana is a resource-windfall case where the rents happened not to be captured. The more accurate reading is that Botswana required both — diamonds produced kleptocracies in Angola, the DRC and Sierra Leone, while good institutions without resource windfalls produced stable but poor democracies elsewhere in Africa. Botswana is therefore best read not as evidence for either pure institutions or pure resources, but as a case where the interaction mattered

What is a question we should ask about articles?

• Is the explanatory variable being used as a proxy for something else?

• Is the relationship causal or just correlational, and how do they distinguish?

• What's the direction of causality?

• Is the instrument really exogenous?

• Does the proposed mechanism actually do the work, or is it asserted on top of a correlation?

• Is this a single-case study or a cross-country claim?

• Does the claim generalise outside the specific historical or geographic context tested?

• Does the paper compress history?

• Does this paper place too much weight on broad structural forces and not enough on agency?

• Could an alternative mechanism fit the same data?

• Could this be a downstream consequence of something else?