If technology has arrived everywhere, why has income diverged?

That is the topic of a new paper by Diego Comin and Martí Mestieri, published in AEJ: Macroeconomics, here is the abstract:

We study the cross-country evolution of technology diffusion over the last two centuries. We document that adoption lags between poor and rich countries have converged, while the intensity of use of adopted technologies of poor countries relative to rich countries has diverged. The evolution of aggregate productivity implied by these trends in technology diffusion resembles the actual evolution of the world income distribution in the last two centuries. Cross-country differences in adoption lags account for a significant part of the cross-country income divergence in the nineteenth century. The divergence in intensity of use accounts for the divergence during the twentieth century.

I am struck by the strength of the two major stylized facts in this paper.  The mean adoption lag for spindles, classified as a 1779 technology, was 130 years, or in other words that is how long it took for the technology to move to poorer countries.  For ships, listed as a 1788 technology, the mean lag is 110 years.  Synthetic fiber is a 1931 technology, with a mean adoption lag of 29 years.  For the internet, a 1983 technology (is that right?), the mean adoption lag is only 6 years.

But the overall story is not so simple.  The more advanced countries use more of these technologies, and use them more effectively (“intensity”), and that gap has been growing over time.  Yes, Ghana has the internet, but it is Silicon Valley that is working wonders with it.  Some technology use begs more technology use.

If you calibrate those parameters properly, it turns out you can explain about 3/4 of the evolution of income divergence across rich and poor countries.


So the Art Deco style where Barton Fink works where "You can imagine it peopled by failed commercial travelers, with pathetic sex lives, who cry alone in their rooms" is a hotel where he lives indefinitely telling the horror of technology transfers in the age of Hercules.

"For the internet, a 1983 technology (is that right?), the mean adoption lag is only 6 years."

It is vague, but the internet is usually considered to have started in 1989.

Some of us were using the internet before 1989.

The internet was developed earlier but was not part of the commercial landscape before the 1990s. I added a modem to my pc in grad school in 1989, I believe, and we used both bitnet and internet email in my (large Ph.D. granting) economics department through the campus mainframes. Bitnet was considered to be quicker in delivery then (which was not yet instantaneous). There were also usenet discussion groups, but again they were primitive and noncommercial. My sense is that the internet only began to put a noticeable footprint on higher education and business with the development of Mosaic and Netscape in the mid 1990s.

"Commercial Internet service providers (ISPs) began to emerge in the very late 1980s. The ARPANET was decommissioned in 1990."

The internet began with the invention of the MPLS router at Stanford on the late seventies. The first commercial internet service was The Well in the early eighties, so the date used is basically correct.

No, it is not basically correct.

"The WELL was started by Stewart Brand and Larry Brilliant in 1985, .... ... became one of the original dial-up ISPs in the early 1990s when commercial traffic was first allowed, and changed into its current form as the Internet and web technology evolved.

Depending on what you like to call the Internet you can choose several dates. The Internet went live in October 1969. You will see the 1983 date a lot because that is when the used second major version of the basic protocol, TCP/IP, uses was required replacing the earlier version called NCP. The modern structure of the Internet was implemented in the first part of the nineties with the first Internet Exchanges (where networks exchange traffic), at the same time the commerical Internet was created with the early exchange at Sprint. So pick you date. I go with the 1969 date since the first killer application, email, was invent in the 1971 and people were using the Internet.

Is usage the same as adoption? Is wearing clothes made from a spindle the same as adopting a spindle?

Where I'm going with this is: is using the internet and a phone the same as designing chips, assembling RAM? Or developing apps? How do we accurately measure adoption of technology?

Ever hear of socialism?

You can bring the socialist to technology, but you can't make him adopt the market mechanisms necessary to use the technology effectively.

On the shrinking tech lag shown in Table 1: My impression is that this trend continues if you go back further, as well. The medieval advances in agriculture took a century or two to spread throughout Europe alone, i.e, about as long as it took for the spindle to spread worldwide. And it famously took a thousand years for paper-making to spread from China to Europe, which still boggles my mind no matter how much I try to wrap my head around it.

“The more advanced countries use more of these technologies, and use them more effectively (“intensity”), and that gap has been growing over time.”

Interesting. The more advanced countries have the advantage of having more infrastructure in place (possibly including social infrastructure), while the less advanced countries have the advantage that it's easier to copy standardized, well-tested methods from higher-tech countries. This suggests that the first advantage may outweigh the second as far as rates of return are concerned. Of course, it's also possible for countries to sprint ahead if they can innovate successfully at scale, like England in the 1700s or America in the 1800s/1900s, but this is rare enough that I don't expect it to show up in aggregate data of this type.

Ben LT: And it famously took a thousand years for paper-making to spread from China to Europe, which still boggles my mind no matter how much I try to wrap my head around it.

Intensity of use is probably something to do with this; if you consider the history of papermaking in China, book production is still pretty low to the degree that China can be surpassed in paper and books by the less populous Islamic world (https://en.wikipedia.org/wiki/History_of_paper - "However despite the initial advantage afforded to China by the paper medium, by the 9th century its spread and development in the middle east had closed the gap between the two regions. Between the 9th to early 12th centuries, libraries in Cairo, Baghdad, and Cordoba held collections larger than even the ones in China").

Printing, likewise, in China, remained a relatively laborious woodblock process, without widespread use of movable type and the press (as much as experimentation with metal movable type did happen in East Asia prior to Europe.)

If the internet were fairly slow, relatively expensive, gave moderate sized marginal improvements on print, you would not see the most intensive use, and that would lowers the probability that it would have spread.

But something about the drive to repeatedly iterate and improve technologies (even those developed first elsewhere) arose in Europe in the latter half of the second millennium AD and has continued at an increasing pace. Technologies iterate quickly and either develop advantages or sink, rather than persisting at a low margin of efficiency over previous technologies for a long time. This is probably the effect of competition on a national and capitalist level, with the increased cooperation brought about by more widespread political unity and science and education.

I'd say that talking about "explaining the evolution of income divergence" is a misspecification. Rather, the lower use of an adopted technology is one of the manifestations of lesser success by imitators. After all, even before the new tech, the laggards were often not even fully exploiting technology in agriculture, etc. that had existed for centuries. In the same way that countries as different as Argentina, Russia, or Egypt do not fully exploit industrial era technologies to produce even more growth in conjunction with better overall organization.

'but it is Silicon Valley that is working wonders with it'

If one feels that creating the first for profit surveillance society is a wonderful thing, where people buy the devices and provide all the information necessary to ensure pervasive monitoring.

Isn't there an inverse correlation between the availability of abundant cheap labor and the rate of adoption of new technology? Doesn't capital usually favor the former more than the latter?

The the questions become: why it’s happening and why have trends changed over time?

"If you calibrate those parameters properly, it turns out you can explain about 3/4 of the evolution of income divergence across rich and poor countries."

This might be illuminating something about practices and culture in each country, but this seems like you are measuring symptoms rather than disease. Why is the adoption rate different? Could this be an observable quantity that is the result of different levels of freedom, or is caused by cultural differences, or education levels, or any of hundreds of underlying causes?

Regressions give you a model that, for better or (usually) worse can generate predictions. These, of course, are of varying accuracy and quality, but in the end don't actually explain, but just illuminate correlation.

This is what I thought too... It's easy to imagine a scenario where tech adoption follows from wealth or where the two reinforce each other.

Fracking has just about doubled US oil production in about 10 years. The technology is available. Mostly it's government policy that prevents its use elsewhere.

Cowen: "Yes, Ghana has the internet, but it is Silicon Valley that is working wonders with it." Wonders? What Silicon Valley has mostly succeeded in doing is exploiting the internet as an advertising medium in a wealthy country with lots of consumers. Hows does one exploit the internet in that way in Ghana?

Trying to argue that nothing about daily life has changed since 1980 except for more advertising just makes one sound delusional.

Trying to pretend that Silicon Valley does not generate most of its enormous profits and wealth from digital advertising just makes one sound delusional. Equally delusional would be one in Ghana who would pay billions for digital advertising for products for which there are no buyers.

Silicon Valley has given up on the idea of making a reliable driverless car and is limiting its effort to producing the software. This has freed up some of the tech talent so they can develop the technology for converting the windows in the driverless cars to tv screens to sell digital advertising to those riding in the car (and, perhaps, to those in other driverless cars passing by). This is the actual tech rather than the fictional tech.

>If you calibrate those parameters properly, it turns out you can...

... make a stack of data say literally anything you want.

It's hilarious that Tyler is happy to begin a sentence with these words.

It's also hilarious that he has inadvertently described climate "science."

"For ships, listed as a 1788 technology": but why?

WKPD: "The Charlotte Dundas was the first practical steamboat, in that it demonstrated the practicality of steam power for ships, and was the first to be followed by continuous development of steamboats." She proved herself in 1803. So clearly it's not steam power that's decisive. What was it that made 1788 the year for "ships"?

Yes, a 1788 square rigged sailing ship such as those of the British Navy or merchantmen would not have been significantly different from an 1818 or 1758 ship. And if it took 130 years to be adopted in poor countries, which countries were adopting square rigged wooden sailing ships for the first time in 1918?

There is no mystery unless you accept the Standard Social Science Model. If you accept the SSSM then the whole world is enormously more mysterious. Just don't believe a wrong model. Reject the wrong model and your cognitive load for figuring out the world gets a lot lighter.

Why do you call it a model as if its some kind of theory? Genetic equality was proven by Darwin in his great work "On the Origin of Species."

Do you even science, bro?

Someone did a study on the differential adoption of the Spinning Jenny in England and France, and found the two main factors to be wage rates and interest rates. Which makes sense, for technologies in general. If you are paying people $X per hour and can borrow at 9%, buying a labor saving technology is going to be a lot less attractive than if you are paying $2X/hour and can borrow at 4%.

Two different effects, the diffusion of existing tech and the creation of new tech, so despite diffusion increasing, creation increases more and is concentrated where most used.

It's not just Silicon Valley: China is doing just as well, if not more so, but behind their big firewall.

There are many factors to this, but a big one is a matter of scale and capital. It's not that there aren't silicon valley style ideas happening somewhere else: The ideas happen everywhere. The shiniest tech startups had good people, but the difference isn't that stark: Execution can happen anywhere. The main issue is funding, including funding for said exponential growth: It's hard enough to have a startup succeed in a small US market, imagine how hard it is to do it in Ghana.

In other parts of the grand market that is the world, we have no problems figuring out that small advantages in performance, or in ability to be seen by gatekeepers, lead to outcome differences widening. Tyler has talked about this in plenty of other areas, so why not tech? We have a wider pool of consumers available, but without protectionism, tiny advantages compound, and most of those advantages happen to firms with an early presence in the Bay. China only competes because they closed themselves and invested an immense amount of money letting domestic companies grow while being immune to foreign competition.

Unfortunately the genie is out of the bottle, and the capital advantage that the big giants in China and California have is not one that can be matched without even larger capital investments, which no country in the world has the will and ability to make. Given how software is going to keep eating the world, I suspect that countries are going to have to find ways to bring more of the wealth creation where they want. Maybe we'll see barriers, large changes in how taxes for international companies work. Another alternative is to copy the military industrial complex in the US, creating a bunch of high paying jobs in different places for political reasons. It'd sure be the most Cyberpunk way around the problem.

"Given how software is going to keep eating the world"

Is it going to keep eating the world? While software kept growing in areas it already made inroads, there seem to be very few industries where software made a big impact that didn't rely on software five years ago. Similarly, I can't think of a major consumer software product that has been developed in the last five years comparable to Uber/Google/Facebook/etc. except Venmo and Zelle. Why wouldn't the software industry mature just like every other industry before it?

It is not surprising that higher-IQ countries are better able to exploit the intellectual advances made by high-IQ people.

"Yes, Ghana has the internet, but it is Silicon Valley that is working wonders with it. "

I spend all my waking hours struggling to figure out why Ghana doesn't have the technical output of Silicon Valley. It's a total mystery. Everyone has the same question, and yet nobody has any answers. Is it time to accept that this conundrum will never be solved? Some things are simply unknowable.

It is the people and their individual knowledge. Today's technology is far to complex for any individual to know more than a piece of it. To work wonders you need to organize large numbers of individuals and not have some third party say NO.

In Silicon Valley, you have the individuals to make the teams and it is permissionless.

In other areas of technology such as aquaculture, California says NO and that field is not growing nearly as fast as in Ghana.

Good point. I remember how there were all those Ghanian chip fabs and aeronautical startups the Ghanian government said 'no.' If only the Ghanian government would permit tech companies to incorporate there we wouldn't have this problem.


1) Intensity of technology use has always mattered for productivity; see higher late 19th century-early 20th greater British and US productivity on textiles (workers working more machines) relative to German, Greek, Japanese, etc.

2) At the same time as intensity of technology use may be diverging income between countries, note, the Preston Curve also converges (https://twitter.com/pseudoerasmus/status/1008749943551614976), and greater income produces seemingly diminishing returns to human life, happiness and education. Income predicts less about human welfare.

Raw presence of technology seems like it should be limited by state capability and average individual capability. But intensity of technology use may reflect more people's interest in technology and their willingness to make it part of their life, and their incentives to do so.

Ghanaians have far lower human capability than Silicon Valley, yes, but Silicon Valley is also strongly selected for people who are peculiarly willing to make technology central to their lives. Intensity of technology use by other human beings, outside the narrow technophile circle, will be driven more by its ability to make useful, positive changes to their lifestyle. No one is particularly motivated towards turning themselves into a highly productive robot (unless they're German).

If intensive technology use fails to do this, then intensive and adoption may becomes random and predicted by cultural and personality factors (taste for extra conspicuous consumption on the margin of welfare with high effort:low welfare return, taste for technology, etc.).

It's really great that Ghana has the Internet.


It is a snowball effect on technology as it's always been with powerhouse countries and the use and implementation of technology.

Comments for this post are closed