Economic development in an “Average is Over” world

Here is my new paper on that topic (pdf, new link here), commissioned by the Asian Development Bank (but not yet approved or refereed by them).  The key question is what kind of development path will follow, given the realities of premature deindustrialization in emerging economies today.  Here is one bit from the paper:

…trickle-down growth from price discrimination and the erosion of intellectual property rents becomes more important as a source of economic improvement. I call this mechanism “cell phones instead of automobile factories.” Many economic ideas are subject to non-rivalrous use, as they can be deployed by many people once they exist. That phenomenon may sound separate from the substitution of capital for labor outlined above, but that is part of the same broader process. If the wealthier nations use smart software to displace imports from the developing world, poorer nations will benefit from the software in other ways, including a trickle-down of goods and services.

The cell phone (and by extension the smart phone) is a paradigmatic example of trickle-down consumption. The technologies behind the cell phone were invented across a variety of nations, none of them poor (although China contributed to the finishing process), and yet cell phones are extremely prominent in poor and lesser developed nations. Internationally, cell phones and smart phones have brought significant benefits and often at relatively low cost. In the poorer parts of Asia, cell and smart phones are available for much lower prices than in the West. Part of that is the result of price discrimination, such as when Samsung sets deliberately lower prices for most of Africa and the poorer parts of Asia. In other cases the poorer countries buy a somewhat lower quality product, but one still effective for many of their needs. The Blackberry was not long ago state of the art in the United States, but now it sells primarily in poorer countries, including Indonesia, Vietnam, and South Asia, in addition to parts of Africa, and of course it sells to these regions at lower prices.

And this:

Or in other words, rather than Indonesia or Cambodia exporting manufactures to buy imported goods, an alternative development path is that some of those imports trickle down and enter poorer countries at especially low prices. Poorer economies can’t get constant cost goods and services for any cheaper than they are available in wealthier countries and in fact they may have to pay more because of shipping costs, poor institutions, and less efficient retail systems. If the wealthy nations produce more cement, the trickle down benefits from that activity may be slight. But for declining cost commodities, it is a different story entirely.

The more the economies of the wealthy countries are focused on increasing returns to scale sectors, the more important this version of trickle-down growth will become. And for the last few decades, many of the most important innovations in the wealthy countries have been shifting into increasing returns to scale sectors, most notably in the tech world. The tech world is geographically clustered, and centered in Silicon Valley, which are both classic signs of an increasing returns to scale sector. Some of the outputs are given away for free (Google, Facebook), and others show high degrees of market concentration, with a single dominant supplier providing a network good (eBay, Facebook, Instagram, Twitter). When it comes to the hardware behind the tech sector, there is an emphasis on new models, upgrades, and differential pricing plans, again all signs of increasing returns to scale.

In the limiting case, if everything in the economy looks and acts like the tech sector, this source of growth could be quite significant indeed. In other words, a world where “software eats the world,” to borrow Marc Andreessen’s phrase, is a world where the developing nations end up doing pretty well, even if the traditional export-oriented path to convergence has gone away.

Most forms of economic growth are fundamentally imbalanced (Hirschman 1958), but in this “cell phones scenario” we see a new form of imbalance. The new imbalance would be based on increasing returns to scale goods, which would trickle down to poorer countries, vs. constant and increasing cost goods, which would not trickle down. Developing nations thus would be very well supplied with (cheaper versions of) increasing returns to scale goods, but have relatively stagnant supplies of constant and decreasing returns to scale goods.

Comments of course are welcome.  The paper also includes some brief discussions of how the main arguments might apply to China, India, the Philippines, and Central Asia, in line with its ADB origins.


I suspect that Moore's Law and the Flynn Effect are related, which, if true, ought to be good news for poor countries:

Moore's Law may be petering out, but used smartphones will eventually become abundant the way used Ford Model Ts became abundant in the 1920s.

>... but used smartphones will eventually become abundant the way
> used Ford Model Ts became abundant in the 1920s.

Smartphones are much more abundant. 16.5 million Model-T's were made in total. Each of the top four phone manufacturers shipped more than that in just Q1 2016.

All those figures are world-wide totals, but that means more now than it used to.

I was thinking quantum stuff would help carry on Moore's law for longer than I might have assumed. But then I got thinking about how remote influences from electromagnetic radiation would make quantum computing a very vulnerable means of computing for the average user, even if they become affordable in principle.

Perhaps only some very few companies or government departments, those with the ability to develop very high-end and effective Faraday cages, will be able to use such technologies and enjoy computing security at the same time. In time, perhaps they will see fit to share these technologies which are paid for by taxpayers with the taxpayers themselves.

You may already know about this , but there's an informative (and long) article here:

Interesting. It seems that magnetic barriers are an important part of the quantum computing process, one that they've already been working on a lot.

However, this suggests to me a huge asymmetry between the dark arms of the state that can afford such things and the rest of us. Since quantum computing is well-attuned to cryptography, the NSA's wet dream of being able to crack all encryption that the common man can access will be achieved.

Of course, real threats, of the sort that are more dangerous than slippery bathtubs, will be able to sidestep this. So, as with many advances in such fields in recent years and decades, the potential for turning these technologies against the populace abounds but the utility for legitimate threats is virtually zero. Surely it will help with some cool science research too - whether the general public will in fact be the beneficiaries of such technologies is not, I believe, a foregone conclusion.

Quibble on page 5:

Trade between nations falls off rapidly, indeed exponentially, as the geographic distance between those nations
increases (the so-called “gravity equation,” which now is one of the most firmly established regularities in empirical economics).

I object to the use of "exponentially" in this context, to describe a power-law model. Since the coefficient in question is close to 1, why not just say trade is inversely proportional to distance?

It's not exponential at all, it's a power law. You would call it exponential if the distance were in the exponent.

I call this mechanism “cell phones instead of automobile factories.”

The Koreans prefer both, but then, what do they know about development economics?

'If the wealthier nations use smart software to displace imports from the developing world, poorer nations will benefit from the software in other ways, including a trickle-down of goods and services.'

Almost as if the sorts of imports don't matter - why, with the right software, Thailand and Vietnam won't be exporting rice (the world's 2nd and 3rd largest exporters, respectively), but they will be able to benefit from using software that displaces rice imports to allow a trickle down of goods and services.

'and yet cell phones are extremely prominent in poor and lesser developed nations'

Being able to skip the entire creation of a copper wire network spanning those nations might just have a little to do with that.

'In other words, a world where “software eats the world,” to borrow Marc Andreessen’s phrase, is a world where the developing nations end up doing pretty well'

The GPL, birthing as it did the major cell phone OS on the planet, was truly a glorious capitalist achievement, right?

GPL = General Public License? Pretty good analysis.

My quibble with TC is the writing is a bit opaque and oblique, but maybe that's a feature not a bug.

"The GPL, birthing as it did the major cell phone OS on the planet, was truly a glorious capitalist achievement, right?"

Most GPL software is created by for-profit companies intending to earn a return on their investment. As was certainly the case with Andy Rubin and Android, Inc. In general, GPL has nothing to do with lack of a profit motive:

Thus, “free software” is a matter of liberty, not price. To understand the concept, you should think of “free” as in “free speech,” not as in “free beer”. We sometimes call it “libre software” to show we do not mean it is gratis.

'A matter of liberty not price'. Hmm. Sounds rather libertarian, doesn't it?

Except most open source software companies make is NOT licensed under the GPL, specifically because they don't care about the extra features it has vs, say, the MIT or the Apache license. GPL embodies copyleft, and there's no copyleft in most of the open source out there. You could argue that most of the one we already have is only there for historical reasons, as, by design, switching out of the GPL is pretty much impossible.

True. But the Apache licensing model is even more compatible with commercial, for-profit uses (since it does not require all derivative works to remain open-source) , and that is seen as a major reason it has become more popular than GPL:

While Moore's Law will end in 3 to 5 years, what almost everyone thinks is Moore's Law will go into the 2020s without a hitch. By 2019, a smartphone should have the computer power that Watson did in 2011.

What about the advice for the government to stay out of everything? Despite massive government interventions in promoting domestic accumulation of technical and production abilities in basically any economy that ever got rich?

I understand that agricultural productivity improvements have huge potential to reduce the extent and severity of poverty. But is "stick to farming" really good advice at the macro level? The infant industries argument should not be completely thrown out the door, although standard neoliberal critiques should always be near the forefront in any such debate.

Increasing returns industries have the potential to improve lives even in countries that refuse to lessen intervention. The cell phone is a classic example. The major breakthrough in most poor countries was just allowing cell phones in the first place (quite a battle in some areas). But once in place, widespread acceptance of phones destroyed land line monopolies that kept household use of phones to the low single digits under the old regime. No monopolies had to be formally abolished. Indeed, even giving out a cellphone monopoly was still vastly more efficient than the land line monopoly since most households have found ways to afford a cheap phone. This is how technology "invades" nations that are resolutely interventionist without regard to efficiency.

pg 25. small thing:
"For several decades after World War II, Indian practiced protectionism"

I think India was intended

Tyler, I enjoyed reading the paper. As a comment I was left curious about the opportunity in agriculture and would've liked a bit more detail on comparative productivity levels of agriculture to gauge differences. Similarly I wondered even with best practices if much farmland simply wouldn't be competitive due to non-optimal growing conditions?

China: 1] "But elevating 1.3 billion people, many of whom live in distant rural areas, through export-driven growth is a much more difficult endeavor." China hasn't been export-driven nearly to the extent commonly misunderstood as adjusted figures (in 2008) showed exports accounted for just 10 percent of GDP. (

2] "Furthermore China, like most of the developing world today, is likely to undergo premature deindustrialization. I’ve already mentioned that American started deindustrializing in the 1970s, when it was already a fully developed, middle class country. Chinese deindustrialization is likely to come well before a comparable income level, shifting to services and also to industrial robots before it attains the American standard of living from the 1970s; that also will keep Chinese inequality high."

First, post-transfer inequality would not have to be high in China. The U.S. was a "fully developed" (not sure of the criteria for this) middle class country" by the 1960s as well. OECD stats start at 1970 for the U.S. but China currently has the same GDP per capita (ppp) as the U.K. did in 1967.That seemed middle class. Didn't the U.K. deindustrialize in the 1970s as well? Britain's per capita GDP was $18,000 in 1975 (middle class) and $19,000 by 1977 (even more middle class!) If China grows at 6 to 6 percent, it will reach that level by 2020. It seems to face "premature" deindustrialization only if the U.S.per capita GDP is used as the arbitrary standard.


Alok Rai spoke of Indian development under colonialism (though the stricture would equally apply to 'Nehruvian' License Raj pre-Liberalization India) as featuring ' A sort of damaged modernity ... at once embryonic and addled".
Alfred Marshall, writing to a promising Indian ex-student of his, said that India needed entrepreneurs like the Tatas who fully embraced modernity and thus were tech savvy and willing to think long term, thus creating new globally benchmarked industries like Iron & Steel.
The Tatas did indeed show the way to other 'speculators' who sent their scions to MIT or Wharton and were similarly willing to create new knowledge based industries and new 'Mill Towns' which were to feature an upwardly mobile class of manufacturing workers. However, Marshall (this was circa 1910) proved prescient in warning that the Indian entrepreneurial class as a whole would find it easier to chase speculative gains and invest in political patronage so as to capture rents. What he did not envisage was that this class of entrepreneurs, who never embraced modernity except of an 'embryonic and addled' sort, would get jealous of modern industrialists like the Tatas and strike back at them through Organized Labor, local political agitation, and Govt. intervention. Thus, India, in the Fifties, turned out to be not the promised land of the Nationalist industrialists- Tatas, Birlas, Bajajs and so on- who actually invested in their manufacturing work force, providing good accomodation, schools etc- but rather their Waterloo.
Some Indians think this had something to do with 'Socialism'. It didn't. The thing was financed by the sort of entrepreneurs Marshall had already identified in 1910 as likely to impede or destroy the positive initiatives taken by the Tatas.
One result was that the Tatas and Aditya Birla and other tech savvy, 'modern' entrepreneurs looked abroad to expand because they were hamstrung at home. This was Hirschman 'Exit' which the Indian State was disabling for enterprises within its borders- of the thousands of 'sick' enterprises being artificially kept alive, not a few are the 'embryonic and addled' ventures of the bad type of entrepreneur who cleverly transfers his losses to the Nationalized Bank while corruptly hoping to profit from an eventual real estate windfall.
Kalecki gave an analysis of what he observed in countries like India which, unlike Marshall's highlighting of the role of pre-modern or 'damaged modernity' speculators, focused on the 'intermediate class' which had some education and social clout but no defining class characteristic as yet. Furious factionalist politics within this class meant that rent-extraction mechanisms were robust to what might otherwise appear to be an active 'circulation of elites'.
Leaving Marshall, Hirshcman and Kalecki aside- Indians have to admit that there is no moral constituency for industrialisation. From the 1890s onwards there was opposition to female employment in manufacturing because of the 'sacred' character of the female gender. Mahatma Gandhi wrote of his horror at the sight of women going to work in factories in England. Greed and Avarice, in his view, had reduced these Angels to the level of street-walkers. Gandhi warned that if women start working in factories, they will start clamouring for the vote. Indeed, this was already happening in England at the time. Gandhi pointed out that Parliament is actually a brothel because its members voluntarily choose to give themselves to a new man every few years. (This may sound bizarre, but a lot of Indian State Legislatures do in fact feature Legislators sequestering themselves in a hotel while their votes are put up on the block)
Indian legal and cultural barriers to the employment of women hindered industrialization as did the capture of better paying jobs by high caste male part time agriculturists who challenged the authority of overseers and imposed a 'rural work rhythm'- i.e. picking up the pace once in a while but otherwise sociably lounging around.
Since members of the industrial working class could always, individually, improve their position by not doing their job but still getting paid- allowing them leisure to pursue other avenues of enrichment, India never possessed a political constituency for industrialization. Formal sector jobs have long been prized by those who lack them not because they provide remunerative work but because they are a gateway to welfare- housing, schools, health care- without the requirement to actually do any productive work at all. Middle Managers have an incentive to farm the actual productive work out to contractors (who may be related to them) who bring in poorly paid temporary laborers from outside the region. Meanwhile Head Office has discovered that it pays them better to ignore productivity and focus on round tripping money from the Nationalised Banks, Life Insurance Corp etc. by all sorts of dubious means.

In a democracy, like India, if no one wants a growing Manufacturing sector because no one envisages himself or herself as actually belonging to an industrial proletariat, how is it going to magically appear? Instead, if everybody agrees that labor should retain a stake in its agricultural roots and not be permanently mobile, how can we prevent manufacturing being taxed so as to subsidise yet more agricultural involution?
One theory is that even the modest development India has seen has been the product of exogenous shocks and the spectacular stupidity of its own Mandarin class. 'India grows by night'- i.e. when the State is asleep- is Gurcharan Das's lapidary slogan. Unfortunately 'trickle-down growth from price discrimination and the erosion of intellectual property rents' increases the capacity of the State and the Kaleckian intermediate class to exercise a vigilance corrosive of manufacturing success.

Great info and analysis, thanks!

(Small piece of advice: breaking up your comment into smaller paragraphs will make it a lot more readable.)

Thanks Kris! The next post on this blog ties up serendipitously with this one. I've left a comment there- with paragraphs!- which ties up with my observations here.

Here's an interesting (and less cryptic) article by Jessie Romero out of the Federal Reserve Bank of Richmond, titled "Goodbye, Globalization?": Among other things, he distinguishes "global value chains" (GVCs, or vertical specialization), which marked the spectacular growth in trade up until the great recession and the Great Trade Collapse, from value-added trade, the latter (it seems to me) similar to what Cowen is describing. But here's the flip side (from the article): "Weak demand [caused by the recession in Europe and America], along with a strong yuan, also has depressed exports from China, and there are signs of longer-term changes in the Chinese economy. "Two dimensions of the Chinese economy have changed," says the University of Houston's Kei-Mu Yi. "First, as they've become more technologically proficient, they can make a lot of the intermediate inputs themselves, and to the extent they do, their demand for imports would fall. Second, as their economy has gotten bigger, they are selling more domestically rather than exporting." Just as China's entry into the global market boosted trade for the world as a whole, a persistent decrease in China's trade could depress global trade growth."

In technical terms, it's the decline in trade elasticity: "Just how much trade elasticity has declined, and when that decline started, is the subject of considerable debate among economists. But some research suggests the process actually started well before the global financial crisis. With Cristina Constantinescu and Michele Ruta, also of the World Bank, Mattoo found that the trade elasticity started falling around 2001, to about half of what it was between 1986 and 2000. According to their analysis, this decrease in elasticity explains about half of the trade slowdown in 2012 and 2013. The authors pointed to a slowdown in the adoption of GVCs as one major reason the trade elasticity has decreased. Comparing the elasticity of gross trade to the elasticity of value-added trade, which has been relatively stable over time, they find the measures have converged since the early 2000s, suggesting a slower pace of vertical specialization."

On the optimistic side is this (again similar to Cowen): "Still, there are factors that could lead to faster trade growth in the future. For example, technology has made it increasingly possible for small and medium-sized enterprises (SMEs) to reach customers around the world. (International organizations generally define a medium-sized enterprise as one with fewer than 250 employees and a small enterprise as one with fewer than 50.) SMEs continue to account for only a small portion of trade relative to their share of businesses in the economy; in the United States, for example, SMEs are more than 99 percent of all businesses, while accounting for only about 15 percent of exports and 10 percent of imports. Policy changes that make it easier for SMEs to participate in international trade, such as raising the threshold above which an importer must pay customs duties, reducing trade compliance costs, or harmonizing postal systems, could help boost trade growth. Another potential source of trade growth is trade in services, such as computer programming or accounting."

Finally, does trade matter (from the article, quoting the University of Houston's Kei-Mu Yi): "The period when the global economy did really well happened to be the period when globalization increased a lot. Was it just an accident that the global economy did really well when global trade did really well, or are they somehow connected?"

Here's my interpretation: The spectacular growth in trade benefited developing countries (in particular China), but also benefited western firms, which enjoyed spectacular profits as the result of much lower costs - the interests of developing countries and western firms (but not western labor) were aligned. The future of trade, however, will be much different, with less benefit to western firms and greater benefit to smaller firms in the developing countries. In that respect, the interests of western firms and developing countries may not be aligned and the impetus for trade from western firms may decline. Is that concerning: yes, if trade wars and actual wars are concerning.

"Trade wars" are engaged by governments at the behest of mercantile forces. In a real free market the maximum amount of trade would take place without governments attempting to distort the process of individuals doing business.

Governments are people too (to paraphrase Mr. Romney), the people being those with the resources to influence "government". Trade wars (and resulting real wars) are at the behest of the people with the resources to influence "government". During the explosion in the growth of trade, the interests of western manufacturing firms (but not western manufacturing labor) were aligned with developing countries (mainly China). That alignment has about run its course. Tabarrok's answer (in the next blog entry) is an alignment of the interests of western service firms (but not western service labor) and developing countries. The fear of western manufacturing and service firms is that developing countries don't need them anymore. It's then that all Hell breaks loose in the global economy.

I advise some shareholders in classic Mittelstand Capital good exporters. As you can imagine, Chinese growth was a great boon- which is why I suddenly got involved.
As an impartial observer, or trustee, I feel the real culprit for a diminishing future for the enterprises I am concerned with has to do with the German govts. failure to update infrastructure and to continue living in a fool's paradise.

I also feel, China is becoming very corrupt because those who made money through the market have such a big incentive to join the hegemonic class rather than continue to diversify and, at the margin, exercise 'Exit'.

Nice analysis. I'd argue though, that the same forces apply to autos as to smart phones (though not quite to the same degree). So, just as $59 smart phone provides 95% of the functionality of one that costs $599, so too does a $20K Honda provide nearly all the functionality of an $80K BMW (And, in fact, the Honda will almost surely be the more reliable and durable vehicle and still be on the road years after the BMW has become too uneconomical to continue repairing). And, as with smart phones, auto-manufacturers engage in price discrimination to defray global R&D, offering lower-cost developing-market specific models that, again, offer the majority of the functionality at a much lower cost.

Smartphones reduce the costs of transmitting, storing and using information.

But, what matters is that there has to be someone well educated enough on one end of the line to use the information to improve their production.

One of my neighbors is an ag economist with a background in animal nutrition. He tries to get farmers to keep records, improve animal nutrition with ration plans, etc. using smartphones.

Real gauchos do not use data he tells me, nor do they want to collect it, or follow a nutrition plan sent to them on a smartphone.

Perhaps there is an intrinsic value in being a 90% productivity maximizer which leaves a little humanity to the animals instead of a 100% profit maximizer?

I think it's different with leafy stuff. Even in poorly educated parts of Africa, those I've met who started to learn more advanced methods of keeping records and optimizing output for their specific plot conditions, were rather excited about the whole prospect of MO MONEY! It takes a certain type of character to want to fatten up the calf 0.5 days faster or 1.2kg more. It is obvious who will make more money, but not who will be more satisfied with the money they make.

The only farm I've worked on that had cattle, it was mostly just to munch up the plant matter after harvests (timed to occur throughout the year), stomp and churn up the dirt a little, and leave behind a little fertilizer. He was an absolute maximizer with the plants, but I think he also just liked having some mammals wandering around plots on the farm. It was also my first (and only) run-in with an electric fence. Good guy to work for, but MAN that is not a detail you forget to mention!

My friend was recounting his experience in Latin America, where many of the landowners are actually well off city dwellers who have a manager run the farm. The city guy derives value from coming to the farm and riding around on his prize horse. Macho.

He has also done work in Africa. He tells me the women are the ones who are quick to adopt new technology...using simple machines to weed and plant. Women influence each other to adapt quickly, watch the money carefully, and even communicate across villages to find prevailing prices.

I like it Tyler.

I can think of another example of price discrimination that benefits developing countries. Football. I mean European Football (soccer), and specifically the English Premier League. (I know Tyler doesn't "get" soccer, but hopefully he understands this).

The English Premier League sells its broadcast rights domestically and in Europe for very high prices on a per viewer basis, but much lower prices in Africa and Asia. Unbelievably, there are now 469 million fans of the Premier league in Asia alone. Africa has 260 million fans. These figures dwarf the west. European audiences are 67 million and North America a mere 28 million. Source ( Yet broadcast revenue from the West is still heavily supporting the game.

In large part because of revenue from rich parts of the world, the quality of soccer has improved. English (as well as some Spanish and German) teams are buying the best players in the world and creating the best, most game that's ever been seen. Poor nations are getting the exact same, world-class entertainment that Europeans and Americans are getting... in part because of price discrimination among TV rights packages. It's not unlike smart phone technology.

Those Africans and Asians are watching the game on television. It's imaginary, a display of pixels. It's not real. A television broadcast of a sporting event is simply a more technologically sophisticated version of a black and white photo of an occurrence during that event or even a seventeenth century oil painting of one. Watching an actual event is much different than observing its television facsimile.

Not sure I get this. So Development doesn't mean roads, urbanization, industry? Development doesn't require that the population have jobs which pay salaries which they use to buy clothing, housing and food? Sure, smartphones are cheap, but roads aren't cheap. Having a mobile Facebook account is enough to be counted as developed now?

So how does this information affect development policy recommendations, if at all?

It seems that developing countries gaining access to cheap, but powerful, platforms should induce entrepreneurs to solve local problems in new ways.

And its not only platforms like phones. The best-in-class SW and HW development tools are increasingly accessible to the willing without financing (e.g., it's amazing what can be done even today with free auto-CAD and a 3D printer). Will this accelerate "the average is over" within developing nations?

Nassim Talib preaches the wisdom of optionality and trial-and-error. Your mechanism fantastically increases the options available for potential entrepreneurs to exploit and cheapens the cost of the trial-and-error process.

How can policies assist these tailwinds? Publically funded entrepreneurial incubators seems obvious, but likely have non-obvious downsides (maybe fairness).

Suppose software eats the world. We can see a lot of software development in the Software-as-a-Service sector. It may be that SaaS providers are not as willing to price discriminate as hardware manufacturers (at least that's what I would expect: I suppose it depends on the provider's marginal costs). If that's the case, the benefits of future technology may not be as widely available at low cost to developing nations as otherwise expected.

Markets produce prices. Not the other way around. Communication of prices affects all industries. Whether or not you participate in those markets is another story.

Comments for this post are closed