Great article! Do you expect that, conditioned on rapid growth in India, a similar period of growth might reoccur? I would imagine that depends on the nature of Indian growth.
Yes! India is arguably the key variable here. If India were to enter the sort of capital-hungry growth phase that China saw in the 2000s, that could certainly relaunch commodity booms across Latin America and Africa. India is certainly growing strongly, but a lot of that growth is concentrated in the services sector, so we haven't seen massive capital accumulation on the scale of what China saw (though capital accumulation in India has certainly been nothing to sneeze at over the last ~5 years).
India's big population would seem to make it a promising market for food. I think they have built a lot of ports and roads in the last few years. Lots of power plants, too. They should continue to be a good market for LNG, coal, oil, and copper, too, to name some other examples
What are some consistently useful investments that poor states can make during boom periods that can support growth through ups and downs? Nuclear plants? Lots of solar and batteries cut energy imports or allow more exports? Basic reading and arithmetic tutoring in school by older students? Bukeleish round-up of repeat offenders?
Also does this imply that countries that specialise in low skilled manufacturing are likely to do better now that China has industrialised and getting older?
Yes, probably! Vietnam/Malaysia made a relatively good bet. Remains to be seen how it goes with Bangladesh but probably better than the bet India made (on service exports)
For Malaysia it was a choice because they had a sizeable oil and gas industry. For Vietnam and Bangladesh we didn't have a choice in the matter. Shove everyone into a factory or have more peasent rebellions.
Aren’t there reasons to be skeptical, though, about manufacturing continuing to be a route to development? I think you and Henry Williams touched on this in your essay published in American Affairs. That is, consumption patterns in rich countries are probably changing as populations shrink and grow older (my understanding is older people spend much more of their income on services), plus labour-saving automation increasingly reduces the key benefit of manufacturing, namely that it can suck in large numbers of workers with varying levels of skills. (I take it automation hits two ways: it reduces the total number of workers plus the jobs that remain or are created, eg maintenance or supervision of robots, tend to require more skills.) Dani Rodrik and Raghuram Rajan have certainly taken this view re manufacturing as a route to development and have both said a services-based development is the way of the future now, although from what I can tell Rodrik is a cloud of pessimism and mostly recommends this because he thinks the other paths to development have even worse prospects now.
Communism meant that a big chunk of the world (mostly china) was punching below its human capital weight. Taking off the fetters of communism was a one time boost though. Eventually everyone reaches the limits of their genetic capital.
My whole life what you might call middle income countries always get hyped up during resource booms only to fall back to earth every time there is a bust.
So that leaves the poor governance explanation intact. It was covered up for a while by the cmmodities boom but then reappeared.
And even with feed forwards (higher income my promote better policies, make bad policies less damaging) the difference in governance between Singapore or South Korea of China today compared to 1950 and compred to Zimbabwe in 2026 is enought for me to think that developmet economists ought to be focused on reforms of polciy in poor countries.
Domains of latent comparative advantage, state capacity to provide the commercial environment for capital to flourish in them, interested capital and an environment that allows large economy of scale to be tapped are factors that can allow lower productivity countries to ascend. Not sure there’s any benefit to convergence, higher education countries will always generate more new high productivity domains that diffuse to others, at least until high education diffuses. The observation of china’s growth driving resource boom probably is the case. Perhaps this shows that large amalgamated scale of state can facilitate growth and productivity ascent. And Akamatsu’s observation of stages of productivity ascent through diffusion of demand and technique from higher to lower countries may describe the underlying accumulation. Extrapolating to current economy and simplifying groupings, countries face one of 3 productivity levels their latent comparative advantage can transition to: material processing before export, industrialization or automation. Pursuing latent comparative advantage, via state capacity equipping capital, is the strongest basis for productivity ascent, the sum of conditions may or may not help achievement at any given time. Convergence is less important than productivity ascent.
I didn't think Solow intended his model of growth to be the basis of a claim there would be convergence. The economics profession did expect convergence, but the Solow model was just focussed on the growth effects of capital accumulation. Whether low capital economies eventually catch up also depends on other considerations. Also, China's own experience shows that capital accumulation does matter.
That said, it was brave of SS&P to proclaim (in their 2021 paper) that "unconditional convergence" had arrived. And it is good of them to retract that claim today.
You do make a good point about China. It was only a few months ago that Branko Milanovic (who has done important work on inequality within and across economies) assert something to the effect that neoliberal prescription basically led to economic development in the Global South. (I can't recall the exact claim, but that was the thrust.) It wasn't clear to me whether his claim took China's growth into account, and attributed China's growth to neoliberal prescription (!) or had excluded China's growth. But I can't help wondering if a lot of the celebration about the success of development among World Bank and IMF staff control for China's growth.
I do like the Solow growth model. It provides a framework to think about growth based on capital accumulation, and that seems a genuinely valuable contribution to economics and the broader field of political economy.
I didn’t finish reading this because the first half is so completely wrong. It grossly misrepresents the existing literature, history of empirical work and even the model itself. There was a HUUUUGGGGGEEEEEE literature on growth and convergence in the 90s. Mankiw Romer and Weil, the paper that set off literally “millions of growth regressions” (to paraphrase Sala i Martin), are 1992. And really it goes back another decade, to Barro. And all these papers found *support* for Solow model. It never came “under attack”, although starting in mid 70s, it got expanded upon with New Growth Theory which basically tried to explain the part that Solow left unexplained - the residual.
But that whole literature found ton of support for *conditional*, not absolute, convergence, which is exactly what Solow-Swan predicts. You want Penn Tables. Sure, let’s take Penn Tables. Regress growth on initial income. But control for investment rate. Boom, you got yourself a significant negative coefficient. Convergence. That’s all it takes, controlling for just one confounded rather than doing a naive bivariate correlation. Which is why no one, at least not since 1980s, actually tried to publish anything with just growth and initial income as everyone and their mother, who studied sociology in college, knew that that was not what Solow model said. The model tells you to control for, at least, the investment rate (s) to account for the steady state that the economy is converging to.
Solow model and convergence never went away. Only thing that happened is that for a minute there it looked like we might be having *absolute* convergence, not just conditional convergence. Great! But it doesn’t matter since the Solow model says nothing about absolute convergence.
Crack open Barro and Sala i Martin’s textbook.
It’s also incorrect to say that in Solow model, because of diminishing returns, in steady state an extra unit of capital doesn’t get you anything. That production function is strictly increasing, it’s not Leontief. More capital still gets you more output, it’s just that’s just enough to offset the decrease in your capital stock due to depreciation and population growth. If you want to get really technical then it might very well be the case that you have diminishing returns to capital, constant returns to scale and the economy still grows forever! Again Barro and SalaiMartin have a nice example. Y=AK+(K^a)(L^(1-a)). That’s why you need the Inada conditions to guarantee existence of steady state.
Eventually it will be permissible to discuss certain unfortunate aspects of our species such as the marked differences in average IQ across racial groups. This 800 lb gorilla cannot be ignored by truth-seeking economists indefinitely.
Please correct me if I have this wrong, but this definition of convergence seems flawed to me n that the units in the regressions are countries, as represented by their per-capita incomes, unweighted by differences in population, so that India, Bangladesh, and Indonesia enter the calculations with the same weight as, say, Togo, Benin, and Ecuador. Measures of the global distribution of income among individuals or households, no matter where they live, like the global Gini coefficient, are much more meaningful, not being dependent on the arbitrariness of political dividing lines. Those measures give appropriately higher importance to populous counties like China and India, which are, on average, still converging. I’m not saying that it’s not worth noting that Africa and Latin America have been lagging, but explanations for that might have to be much more regionally specific. It might be too early to conclude that globalization has failed, especially in light of India’s recent high-growth experience. Moreover, if, for example, Nigeria and S. Africa were to grow faster, which may be starting to happen, the convergence coefficient would be very little affected but the global distribution of income would react more positively in the direction of greater equality.
The “Great Convergence” was not development succeeding. It was a temporary alignment with a single external growth engine. When that engine slowed, the illusion disappeared.
I really loved this article. It's well researched and I followed the argument, though it got a bit dry in spots. But I'm confused about the timeline and I'm hoping you can clarify.
The SS&P graph shows convergence starting around 1995, but China only joined the WTO in 2001. What was driving convergence in those first five or six years? In the 1990s China was mostly making cheap plastic stuff and textiles, not the heavy industrial goods that came later. That kind of manufacturing doesn't use nearly as much copper or iron or other commodities. The commodity boom really happened in the 2000s and 2010s when China was building massive infrastructure, right? So what explains the growth before that?
Also, what about India? Its economy has been growing fast recently and it has a huge population. Even if it's more service-based, people working in services still buy stuff and India is trying to build up manufacturing because of nationalism. Why aren't poor countries just trading with India now instead of China? If your theory is right, shouldn't we be seeing another convergence wave from Indian demand?
And here's what really bothers me: Japan, South Korea, Taiwan, Singapore all industrialized before China. Japan was huge in the 60s-80s, Korea and Taiwan in the 70s-90s. They also needed commodities for manufacturing. Maybe not China-level scale, but combined they should have created some bump in the data, right? But the graph shows nothing until 1995. Why?
I think China is probably part of the story, especially for the peak years in the 2000s-2010s. But the timing doesn't quite add up for me and I'm wondering if there were other big factors in 1995 that you didn't get into as much. Do we have good trade data for the 90s showing commodity exports to China before 2001?
Respectfully, here are some reservations about this piece:
1. The article doesn't quite answer the question posed in the title. Rather than explaining why poor countries stopped getting better, it reframes the premise: "they never were getting better in the first place." That's a provocative reframing, but the underlying causal mechanism still feels underexplored.
2. I'm not fully convinced by the claim that China's commodity boom is the main driver behind the "great convergence." Temporal coincidence doesn't establish causation, and these phenomena are structurally interconnected in ways that resist simple one-directional explanation. Other factors such as financial deregulation also played significant roles but weren't addressed here.
These critiques aside, I appreciate the article raising important questions about development economics narratives. Would be curious to hear the your thoughts!
Great article! Do you expect that, conditioned on rapid growth in India, a similar period of growth might reoccur? I would imagine that depends on the nature of Indian growth.
Yes! India is arguably the key variable here. If India were to enter the sort of capital-hungry growth phase that China saw in the 2000s, that could certainly relaunch commodity booms across Latin America and Africa. India is certainly growing strongly, but a lot of that growth is concentrated in the services sector, so we haven't seen massive capital accumulation on the scale of what China saw (though capital accumulation in India has certainly been nothing to sneeze at over the last ~5 years).
India's big population would seem to make it a promising market for food. I think they have built a lot of ports and roads in the last few years. Lots of power plants, too. They should continue to be a good market for LNG, coal, oil, and copper, too, to name some other examples
What are some consistently useful investments that poor states can make during boom periods that can support growth through ups and downs? Nuclear plants? Lots of solar and batteries cut energy imports or allow more exports? Basic reading and arithmetic tutoring in school by older students? Bukeleish round-up of repeat offenders?
On existing evidence it would be reasonable to assume that not all populations have identical
capacities but the model presumably assumes they do because to do otherwise is not permitted
Turns out merit matters more than resources.
Also does this imply that countries that specialise in low skilled manufacturing are likely to do better now that China has industrialised and getting older?
Yes, probably! Vietnam/Malaysia made a relatively good bet. Remains to be seen how it goes with Bangladesh but probably better than the bet India made (on service exports)
For Malaysia it was a choice because they had a sizeable oil and gas industry. For Vietnam and Bangladesh we didn't have a choice in the matter. Shove everyone into a factory or have more peasent rebellions.
Aren’t there reasons to be skeptical, though, about manufacturing continuing to be a route to development? I think you and Henry Williams touched on this in your essay published in American Affairs. That is, consumption patterns in rich countries are probably changing as populations shrink and grow older (my understanding is older people spend much more of their income on services), plus labour-saving automation increasingly reduces the key benefit of manufacturing, namely that it can suck in large numbers of workers with varying levels of skills. (I take it automation hits two ways: it reduces the total number of workers plus the jobs that remain or are created, eg maintenance or supervision of robots, tend to require more skills.) Dani Rodrik and Raghuram Rajan have certainly taken this view re manufacturing as a route to development and have both said a services-based development is the way of the future now, although from what I can tell Rodrik is a cloud of pessimism and mostly recommends this because he thinks the other paths to development have even worse prospects now.
Yea but old people need non-tradables services so that's not going to develop any poor country.
Communism meant that a big chunk of the world (mostly china) was punching below its human capital weight. Taking off the fetters of communism was a one time boost though. Eventually everyone reaches the limits of their genetic capital.
My whole life what you might call middle income countries always get hyped up during resource booms only to fall back to earth every time there is a bust.
So that leaves the poor governance explanation intact. It was covered up for a while by the cmmodities boom but then reappeared.
And even with feed forwards (higher income my promote better policies, make bad policies less damaging) the difference in governance between Singapore or South Korea of China today compared to 1950 and compred to Zimbabwe in 2026 is enought for me to think that developmet economists ought to be focused on reforms of polciy in poor countries.
So the Krugman-Venables model still stands.
Domains of latent comparative advantage, state capacity to provide the commercial environment for capital to flourish in them, interested capital and an environment that allows large economy of scale to be tapped are factors that can allow lower productivity countries to ascend. Not sure there’s any benefit to convergence, higher education countries will always generate more new high productivity domains that diffuse to others, at least until high education diffuses. The observation of china’s growth driving resource boom probably is the case. Perhaps this shows that large amalgamated scale of state can facilitate growth and productivity ascent. And Akamatsu’s observation of stages of productivity ascent through diffusion of demand and technique from higher to lower countries may describe the underlying accumulation. Extrapolating to current economy and simplifying groupings, countries face one of 3 productivity levels their latent comparative advantage can transition to: material processing before export, industrialization or automation. Pursuing latent comparative advantage, via state capacity equipping capital, is the strongest basis for productivity ascent, the sum of conditions may or may not help achievement at any given time. Convergence is less important than productivity ascent.
Besides Subramanian, Sandefur, and Patel, Michael Kremer wrote about convergence around the same time:
https://www.journals.uchicago.edu/doi/10.1086/718672
I didn't think Solow intended his model of growth to be the basis of a claim there would be convergence. The economics profession did expect convergence, but the Solow model was just focussed on the growth effects of capital accumulation. Whether low capital economies eventually catch up also depends on other considerations. Also, China's own experience shows that capital accumulation does matter.
That said, it was brave of SS&P to proclaim (in their 2021 paper) that "unconditional convergence" had arrived. And it is good of them to retract that claim today.
You do make a good point about China. It was only a few months ago that Branko Milanovic (who has done important work on inequality within and across economies) assert something to the effect that neoliberal prescription basically led to economic development in the Global South. (I can't recall the exact claim, but that was the thrust.) It wasn't clear to me whether his claim took China's growth into account, and attributed China's growth to neoliberal prescription (!) or had excluded China's growth. But I can't help wondering if a lot of the celebration about the success of development among World Bank and IMF staff control for China's growth.
I do like the Solow growth model. It provides a framework to think about growth based on capital accumulation, and that seems a genuinely valuable contribution to economics and the broader field of political economy.
I didn’t finish reading this because the first half is so completely wrong. It grossly misrepresents the existing literature, history of empirical work and even the model itself. There was a HUUUUGGGGGEEEEEE literature on growth and convergence in the 90s. Mankiw Romer and Weil, the paper that set off literally “millions of growth regressions” (to paraphrase Sala i Martin), are 1992. And really it goes back another decade, to Barro. And all these papers found *support* for Solow model. It never came “under attack”, although starting in mid 70s, it got expanded upon with New Growth Theory which basically tried to explain the part that Solow left unexplained - the residual.
But that whole literature found ton of support for *conditional*, not absolute, convergence, which is exactly what Solow-Swan predicts. You want Penn Tables. Sure, let’s take Penn Tables. Regress growth on initial income. But control for investment rate. Boom, you got yourself a significant negative coefficient. Convergence. That’s all it takes, controlling for just one confounded rather than doing a naive bivariate correlation. Which is why no one, at least not since 1980s, actually tried to publish anything with just growth and initial income as everyone and their mother, who studied sociology in college, knew that that was not what Solow model said. The model tells you to control for, at least, the investment rate (s) to account for the steady state that the economy is converging to.
Solow model and convergence never went away. Only thing that happened is that for a minute there it looked like we might be having *absolute* convergence, not just conditional convergence. Great! But it doesn’t matter since the Solow model says nothing about absolute convergence.
Crack open Barro and Sala i Martin’s textbook.
It’s also incorrect to say that in Solow model, because of diminishing returns, in steady state an extra unit of capital doesn’t get you anything. That production function is strictly increasing, it’s not Leontief. More capital still gets you more output, it’s just that’s just enough to offset the decrease in your capital stock due to depreciation and population growth. If you want to get really technical then it might very well be the case that you have diminishing returns to capital, constant returns to scale and the economy still grows forever! Again Barro and SalaiMartin have a nice example. Y=AK+(K^a)(L^(1-a)). That’s why you need the Inada conditions to guarantee existence of steady state.
Eventually it will be permissible to discuss certain unfortunate aspects of our species such as the marked differences in average IQ across racial groups. This 800 lb gorilla cannot be ignored by truth-seeking economists indefinitely.
Please correct me if I have this wrong, but this definition of convergence seems flawed to me n that the units in the regressions are countries, as represented by their per-capita incomes, unweighted by differences in population, so that India, Bangladesh, and Indonesia enter the calculations with the same weight as, say, Togo, Benin, and Ecuador. Measures of the global distribution of income among individuals or households, no matter where they live, like the global Gini coefficient, are much more meaningful, not being dependent on the arbitrariness of political dividing lines. Those measures give appropriately higher importance to populous counties like China and India, which are, on average, still converging. I’m not saying that it’s not worth noting that Africa and Latin America have been lagging, but explanations for that might have to be much more regionally specific. It might be too early to conclude that globalization has failed, especially in light of India’s recent high-growth experience. Moreover, if, for example, Nigeria and S. Africa were to grow faster, which may be starting to happen, the convergence coefficient would be very little affected but the global distribution of income would react more positively in the direction of greater equality.
The “Great Convergence” was not development succeeding. It was a temporary alignment with a single external growth engine. When that engine slowed, the illusion disappeared.
Poor countries failed because their elites became intertwined with the elites of rich countries
They enriched themselves and their progeny while beggaring their own countries
India included
I really loved this article. It's well researched and I followed the argument, though it got a bit dry in spots. But I'm confused about the timeline and I'm hoping you can clarify.
The SS&P graph shows convergence starting around 1995, but China only joined the WTO in 2001. What was driving convergence in those first five or six years? In the 1990s China was mostly making cheap plastic stuff and textiles, not the heavy industrial goods that came later. That kind of manufacturing doesn't use nearly as much copper or iron or other commodities. The commodity boom really happened in the 2000s and 2010s when China was building massive infrastructure, right? So what explains the growth before that?
Also, what about India? Its economy has been growing fast recently and it has a huge population. Even if it's more service-based, people working in services still buy stuff and India is trying to build up manufacturing because of nationalism. Why aren't poor countries just trading with India now instead of China? If your theory is right, shouldn't we be seeing another convergence wave from Indian demand?
And here's what really bothers me: Japan, South Korea, Taiwan, Singapore all industrialized before China. Japan was huge in the 60s-80s, Korea and Taiwan in the 70s-90s. They also needed commodities for manufacturing. Maybe not China-level scale, but combined they should have created some bump in the data, right? But the graph shows nothing until 1995. Why?
I think China is probably part of the story, especially for the peak years in the 2000s-2010s. But the timing doesn't quite add up for me and I'm wondering if there were other big factors in 1995 that you didn't get into as much. Do we have good trade data for the 90s showing commodity exports to China before 2001?
Respectfully, here are some reservations about this piece:
1. The article doesn't quite answer the question posed in the title. Rather than explaining why poor countries stopped getting better, it reframes the premise: "they never were getting better in the first place." That's a provocative reframing, but the underlying causal mechanism still feels underexplored.
2. I'm not fully convinced by the claim that China's commodity boom is the main driver behind the "great convergence." Temporal coincidence doesn't establish causation, and these phenomena are structurally interconnected in ways that resist simple one-directional explanation. Other factors such as financial deregulation also played significant roles but weren't addressed here.
These critiques aside, I appreciate the article raising important questions about development economics narratives. Would be curious to hear the your thoughts!
The best economies are the ones that build and maintain roads. When it is difficult to travel within a country that country stagnates economically.
And power grid, communications, etc.