Dani Rodrik on premature deindustrialization

Consider Brazil and India, two emerging economies that have done comparatively well in the last decade or so. In Brazil, manufacturing’s share of employment barely budged from 1950 to 1980, rising from 12% to 15%. Since the late 1980’s, Brazil has begun to deindustrialize, a process which recent growth has done little to stop or reverse. India presents an even more striking case: Manufacturing employment there peaked at a meager 13% in 2002, and has since trended down.

It is not clear why developing countries are deindustrializing so early in their growth trajectories. One obvious culprit may be globalization and economic openness, which have made it difficult for countries like Brazil and India to compete with East Asia’s manufacturing superstars. But global competition cannot be the main story. Indeed, what is striking is that even East Asian countries are subject to early-onset deindustrialization.

China is a very large country, of course, with much of its workforce still in rural areas. But most migrant workers now find jobs in services rather than in factories. Similarly, it is extremely unlikely that the new crop of manufacturing exporters, such as Vietnam and Cambodia, will ever reach the levels of industrialization attained by the early industrializers, such as Britain and Germany.

An immediate consequence is that developing countries are turning into service economies at substantially lower levels of income. When the US, Britain, Germany, and Sweden began to deindustrialize, their per capita incomes had reached $9,000-11,000 (at 1990 prices). In developing countries, by contrast, manufacturing has begun to shrink while per capita incomes have been a fraction of that level: Brazil’s deindustrialization began at $5,000, China’s at $3,000, and India’s at $2,000.

There is more here.


I've seen this thesis before--something about how globalization allows you to use preexisting "horizontal" supply chains vs having to build "vertical" ones de novo, as alluded to in TC's post. But perhaps the remainder of the puzzle can be explained by the fact poor people don't need infrastructure like rich Westerners do. Instead of a steel suspension or cable stayed bridge, the poor folk just use a rickety rope bridge. Instead of modern sanitation infrastructure, they poop behind a bush. But they must have their iPhones of course--that's more essential than utilities.

I'm confused about what point you are making

I write for the 120+ IQ crowd, but I'll try to dumb down my post for the rest.

1) Globalization is one reason for premature industrialization (PI).

2) Poor folk don't need fancy bridges, so they don't build them. That's another reason for PI.

Got it now? Flame on!

No, I don't get it. I assume you mean DEindustrialization, but even so it doesn't make sense.

Ray, I maybe wrong, but I would have thought the 120+ IQ types would have no need of your admittedly informative comments, as they would able to evaluate Tyler's posts and attain the correct answers fairly quickly - so, perhaps you have a little confirmation bias going on there? But please don't let that stop you - I enjoy your comments and I have a fairly low IQ. Also, what's your opinion of the effectiveness of horizontal supply chains across rope bridges? Wouldn't they have a tendency to rapidly go vertical??

He is saying that since 1980, "capitalism" is interpreted as pillage and plunder of capital, not creating capital.

"Capitalism" in Brazil is cutting the prime woods and sending them to Asia to be turned into cheap short lived consumer goods without and construction of capital assets in Brazil, ie no steel span bridges. It is drill baby drill to burn the capital assets of Brazil to fund consumption without building a sustainable multigeneration capital asset stock in Brazil.

Labor plus capital equals labor consumption plus capital consumption. In a capitalist society, capital consumption exceed capital production. If Brazil extracts a billion in capital to be burned, then Brazil would add 1.1 billion in multigeneration capital assets with the addition of labor, even after labor has been compensated with labor's consumption.

Consider the US in the 19th century and China in the past few decades: mining of iron and coal and cutting timber to build ribbons of steel and iron horses to run on them spanning the entire nation. That capitalist enterprise was contained mostly in the 19th century, yet today that capital asset is still generating huge returns on the iron ore and coal and timber and labor consumed 100-200 years ago.

Since Reagan, rather than consume, capital simply chases its tail driving up the price of old and decay capital assets. US rail operates run slower today than a hundred years ago, but the price of the railroad assets that have decayed is higher. Rather than taking returns on capital and consuming more to increase the capital stock, the argument is to sell off capital assets in creative destruction, because less capital stocks will yield higher rates of return. Higher rates of return means an asset worth a billion can be priced at two billion. In a decade or two, the capital asset will probably be worthless but then it can be creatively destroyed.

Railroads can be treated as assets to pillage and plunder because they are privately owned. For roads, they are simply free assets if you can block efforts to charge you for their decade, block efforts to charge you for improving them. But that's another example of capitalism as pillage and plunder.

The strongest growth is in China where individual sacrifice is highest in order to rapidly build China's capital asset stock. China is like the US in 19th century and from 1930-1970 - industrial policy that promotes heavy consumption by capital to increase rapidly the capital assets. And high capital consumption means high labor demand, but labor must consume less than they produce, but in exchange, they share in greatly increased production.

If they don't use toilets, no one has to make them. If the industrial workers don't drink, don't drive, don't build nice large fancy homes, don't have expensive hobbies like shooting, fishing, photography and the like, no one has to make the stuff. No internal consumption, simply low cost production for export.

Are there places where the average person has an iPhone but poops behind a bush?

Central park?

In India, over 40 percent of people do not use toilets. I don't know what percent have cell phones

The typically cited figure is 2:1 mobiles to toilets.

How would one know which bush to poop behind without the proper iPhone app? I think its safe to assume that iPhones facilitate increased use of bushes.

Costanza will be on that after his next divorce.

Robots, efficiency and automation?

How much of this is Brazil & India versus just a global decrease in workers needed for manufacturing.

That was my first thought to this. Isn't it just in part that manufacturing output per capita is so much higher than it was 60 years ago? And the consumption of goods hasn't grown as fast as total income, due to the consumption basket shifting more towards services. So between the two, you just need less global manufactoring workers percentage.

Indeed, that was my first thought as well. How does the situation change if you look at manufacturing production as a percentage of GDP in various countries?

I really don't understand why this isn't the first thought anyone should have- it isn't like Brazil, India, or China use 1950s technology to produce industrial output. Their industrial technology is light years ahead of what the US had 50 years ago.

This would explain the lower peak in labor force participation rates, but it doesn't explain why deindustrialization is starting at lower levels of per capita income, which is the bigger concern.

Because robots make higher wage employees in First World nations more profitable, and are replacing workers in both low and high wage countries?

"Deindustrialization" in this sense seems to be just the labor force share of manufacturing starting to decrease. In an earlier age, it decreased in high wage countries because of moves to lower wage countries. But if the jobs are being lost to robots (and just more productivity per worker), then it will decline in all countries, not just high wage ones.

John Thacker has already hit it, but I would point out that US isn't even deindustrialized by your metric- until recently, its manufacturing output was greater than China's entire economy.

You must lack a 120+ IQ.
It's because of pooping horizontally behind bushes.

There's a fair amount of "rooster causing the dawn" in Rodrik's thesis: where in the newly-industrializing world has anything like "Fordism" developed? I don't see anything like the AFL or CIO developing in China, and even there, productivity is too high for more than a few percent of the workforce to be engaged in manufacturing. Also, Chinese industrial productivity is increasing rapidly just like everywhere else involved in manufacturing as ongoing mechanization/robotization and Moore's Law tech improvements do their thing, so the future is less people doing manufacturing, not more.

Employment in manufacturing may be low in Brazil but manufacturing still contributes ~30% of their GDP.

Ergo manufacturing hasn't declined; manufacturing just needs fewer people these days. That's a global trend.

Yes, agreed. Manufacturing and agriculture don't require many bodies anymore. So people flock to "services" and BS jobs.

That sounds like a synonym for a wealthier society. The same or greater amount of outputs with fewer inputs. That was the definition of efficiency they taught us in my sixth grade intro to economics unit.

Lucky you. In India only economics students are taught or even exposed to economics, at the 11th grade level in school or later in college. And mostly to socialism and communism and Gandhian self sufficiency. Not a single judge, lawyers, politician, bureaucrat, journalist, scientist, engineer or doctor knows the difference between "value in use" vs "value in exchange" or has ever heard of "comparative advantage." Oddly we have an economist as Prime Minister.

.....never let your schooling interfere with your education. :)

A wealthier society where all the gains go to the top 5 percent of the population.

Is the article claiming that peak manufacturing in the developing world is (a) too low according to some well-ground absolute standard, or (b) too low relative to countries that industrialised earlier. Or is it both? Perhaps I will find out when I do more than skim, but my quick thought is why should there be a "correct" peak?

The world has changed. For example, we can achieve some things with plastic that used to require metal - and thus need thus a little less heavy industry relative to GDP. Similar things will apply in thousands of other sectors. Also, these countries can use the manufacturing base of others. Chinese middle-managers can buy German cars, while their German predecessors had to buy German cars.

I suspect that Rahul is correct. In the good (bad?) old days, the labor component of manufacturing output was much higher. As countries industrialized, they would "steal" manufacturing from developed countries on the basis of labor costs. Since labor was a big share of the cost, there was a big opportunity for savings: first, with commodity products, but increasingly with more advanced products as labor productivity improved in the developing country.

Now, manufacturing is much more automated; particularly, in developed economies, but also in developing nations. I have an investment in a small US manufacturing company where sale tripled in ten years, while the number of employees declined by 40-50%. Manufacturers in developed economies have invested heavily in automation to blunt the threat from low labor cost zones. Manufacturers in developing countries face similar pressures to automate more quickly than in previous decades, and so manufacturing employment never reaches the levels achieved in countries that industrialized first.

Another reason is that highly automated factories that prove to be uneconomic to their original builder can easily be put to profitable use by another due to their low operating costs, whereas less automated factories that are uneconomic are typically shuttered and never reopened. Over time, this tends to leave behind more automated factories even if these factories are not, once all costs are considered, more profitable than less automated factories.

'Brazil has begun to deindustrialize'

Nope - employment is not the definitive measure, capability is. Just ask anyone interested in true deep water oil drilling. The odds of whoever ends up exploiting the South China Sea oil resources will use cutting edge Brasilian technology are quite good.

Look, its manufacturing following the same trajectory as agriculture and mining: ever increasing productivity within the sector combined with competition for labour with other, newer sectors spells a long term decline in share of the workforce. Brazil, India and China just reflect what is surely a global trend. Of course they will have relative manufacturing employment peak lower than the first wave of industrializing countries! The tertiary sector(s) practically didn't exist the first time the secondary sector got huge.

So in a sense, yes, globalization matters, in as much as the BRICs vs historic OECD path will show the contrast most clearly - because as Ray said horizontal supply chains are part of what helps manufacturing productivity, and because globalized, trade exposed "service" industries like software development can capture a share of the employment market even early on in a nation's development trajectory. But its not the underlying cause of manufacturing's secular decline. In some number of decades people will be remarking on how quickly Somalia and Haiti's "tertiary" sectors peaked in total employment share when compared to India.

(scare quotes on tertiary and services because I don't think either captures the "everything that's not extraction or manufacturing" bucket very well.). Is a license for a copy of Windows, or a Beatles song, a "service", or just an intangible good? Its certainly a very different beast from say a haircut, or conveyance on a house sale. Is energy a service? If not, does it differ that much from bandwidth, or hiring computer storage off Amazon?)

That's a really good point, especially as even services become globalized. We think of service jobs as things that can't be outsourced... but why not? I can't remember the last time I called a local repairman for anything. To the extent retail is a service, it's being globalized. I would bet that a lot of what we think of as custom installed goods will become more like custom pre-fabbed goods, built and outfitted to order and just drop-shipped to the customer.

It has been my understanding that the decline in Chinese manufacturing was entirely in the state owned manufacturing sector and that private manufacturing employment is growing rapidly.

Can anyone disprove that line of analysis?

By the way where do most of you think US manufacturing output is relative to its 2007 peak just before the recession.?

US manufacturing output in August was 99.4 of an index where 2007 =100.and the peak month was 100.8 in January 2007. So US manufacturing output is still below its pre-recession peak.

December 2007 not January 2007.

India never industrialized, thanks to Economic Giants ike Rodan, Myrdal, Hirschman....anything government want to take over and eventually destroy, their disciples provided intellectual support. Even today, India is progressive economists dream state - any cutting edge economics you can think of, the "elites" in India are implementing them. Example - "nudge economics" - place anti-smoking/drinking messages right on the TV/movie scenes when they show smoking/drinking. Forget about individual liberty, the 'elites' know what is right for the masses. Life is always easier on the hallowed halls of Oxford/Harvard/Princeton, so you can "ponder" about the "striking cases" and advice the governments to get more involved to save the masses.

Comments for this post are closed