32 Comments
Mar 6, 2021Liked by Noah Smith

I could be convinced that solar and biotech (and maybe ML) will be GP technologies that will boost TFP in a not so distant future. But in an economy where services dominate it seems to me the aggregate boost as to be somewhat contained or of short duration as the IT boost in the mid 90s. What do you think?

Expand full comment
author

I do think the boost will be contained by the shift to services. Which is why I'm more optimistic about technology's contribution to TFP than I am about TFP in the aggregate!

Expand full comment

One thing:

- I'm not sure about that college enrollment graphic. My understanding is college enrollment is countercyclical, and 2010 was different than 2017 as far as the state of the US economy: for example, 2010 and 2017 had unemployment numbers of ~9% and ~4% respectively.

Expand full comment
author

I think that's part of it. But I also think we're hitting saturation on college.

Expand full comment

Interesting point. Do you think it's across the board saturation, particular fields, or a combination of both?

Expand full comment
Mar 6, 2021Liked by Noah Smith

From your "All Futurism is Afrofuturism" post, I learned that the UN medium variant forecast says that 50% of the population growth in the next 20 years will be in Africa (and 95% in the next 80 years). I'd be interested in your thoughts on how this trend might interact with your overall techno-optimism.

Expand full comment
author

Great question. I'll write about this in the future. Short version is that I think solar and batteries/storage will give a big boost to African industrialization, and mRNA vaccines against malaria (if that pans out) will remove the last big disease burden from the continent. So in other words, I think new technology is especially suited to the task of industrializing the world's last great un-industrialized densely populated regions!

Expand full comment

Malaria is not the only disease burden left - sleeping sickness still renders vast swathes of Africa uninhabitable.

Expand full comment

In 2018, 977 cases of sleeping sickness were recorded. (WHO)

There were an estimated 219 million cases and 435 000 related deaths from malaria in 2017. (WHO)

I mean, one's an estimate and the other is reported, so that will distort things. And the CFR for sleeping sickness is much higher, if you get it.

There are many other diseases to confront too. Dengue, HIV, TB, infectious diarrhea all horrible widespread impacts. On the high CFR end, sleeping sickness, rabies, Marburg, and prion-based diseases are all awful.

But all things considered, malaria is a pretty big deal for humanity, it should be high on our todo list, and it's hard to imagine it not having a major impact.

Expand full comment

I am cautiously optimistic. Three big reasons. One, generational change. Just as the old hard-money generation has given way to a new generation of monetary policymakers, I have some hope that the YIMBY movement will gradually infect the conventional wisdom over the next decade. If we have full employment and a big loosening of land use restrictions, that would be great for productivity. Second, the green energy revolution. The oil-powered growth of the 50s through to the 80s was terrible for the environment, and justifiably triggered a tsunami of political and regulatory pushback. The technologies of the 2020s (e.g. e-bikes, solar+storage, hydrogen) are great for the growth and the environment, meaning there's more likely to be broad support. NextEra is now more valuable than Exxon - that changes the lobbying dynamic. Three, I see Silicon Valley starting to disrupt some of the most unproductive and stagnant areas of the economy, e.g. construction, healthcare. My gut feel is that growth in the 2020s is going to rip.

Expand full comment

I'm unclear what is included and what not in "technological progress", nor for that matter what a focus on the US tells us about how technology is or isn't advancing.

Take China. Per capita income has increased almost twenty times over the past 30 years. This is the greatest achievement of curtailing poverty and creating better lives for the most people in all of human history. Pretty good! Now, of course, there's much more than "technology" behind that. But among the factors are: the development of container-based shipping; the development of extremely large and economical container ships; the transformation of ports to handle those ships and quickly move vast numbers of containers; improved transportation infrastructure to move containers speedily across large markets (e.g., the building of the Interstate Highway System in the US).

Are those part of "technological progress" and are they included in TFP measures? Fully or partially? And to the extent our measures don't capture this kind of stuff, how useful are they when they miss, fully or partially, the greatest increase in the materiel lives of people in all of history?

Expand full comment

I've been pondering the GDP slowdown for decades, and can suggest more factors apart from technology.

Cleanup: until the 1950s or '60s, industry polluted almost without restraint (including tobacco smoke into people's lungs) and booked hefty profits. We now pollute with more thought for the consequences, but most importantly, we spend labour and materials trying to make the old messes less toxic.

Safety: we attach higher value to human life and health, and so we spend on protecting the one and restoring the other. Whether we get good value for money is a separate argument, but probably yes outside the USA, and the results do not show up directly in GDP (mumble mumble broken window fallacy). Yes, this does overlap with point one.

Courts: certainly in the USA, and probably in other places, far more disputes are litigated, at huge cost, and with ... various levels of benefit, mumble mumble patent trolls mumble ambulance chasers.

Education: I for one am convinced that additional years at educational establishments have led, not to more being learned, but to the same things being learned at a more relaxed pace. And yes, my primary and secondary schools emphasised sports too. Even nerds like me had to play, no exceptions, and some of us got extra exercises to fix our posture.

With a big yes to the other points that Noah and most economists make.

Expand full comment

^^^As you can see, each period of productivity growth now looks a bit slower than the last, meaning that the exponential growth rate is slowing down.^^^

I have to say for this non-economist, the above seems highly intuitive. I mean, video calls are nice, but are they *that* much of an improvement over the non-video calls enjoyed by Ward and June Cleaver? And phone calls, are nice, too, but are they that much of an improvement compared to the improvement enabled by the telegraph? (Before the telegraph it took many days for important communications to travel!). Indeed, the telegraph itself arguably pales compared to what railroads enabled. (Six weeks for a letter to get from the frontier to New York compressed down to three days!).

And so on.

Technological revolutions are, erm, revolutionary. Except for the ones that directly preceded them. I mean, it doesn't get much more revolutionary than agriculture and the domestication of animals 10k years ago. That is until you arrive at the mastery of fire-making technology 400,000 years ago! I feel diminishing returns on the productivity accompanying new technologies set in from the very beginning.

That said, I don't buy the "pessimistic" case on technology and productivity, mainly because these improvements, while arguably in many cases decreasing in terms of *marginal* improvement, nonetheless can seem pretty spectacular as they aggregate. We humans are all truly standing on the shoulders of the giants (and their discoveries) who came before us. And, while I've argued that diminishing returns long ago set it, I don't contend they've done so smoothly or remorselessly. So we occasionally get wonderful *bursts* of improvement (the printing press, steam power, combustible fuels, electricity, etc).

I expect AI + fusion will be central to the next burst.

Expand full comment

Life expectancy took off after the industrial revolution and generally (that is, outside of world wars, American gilded ages, Communist revolutions, and other catastropies) has increased even faster closer to the present. While life expectancy took no such leap going from hunter/gatherer to farming/herding. In fact, it seems to have gone down (though the concentrated state-directed violence capacity of the farming and herding polities eventually squeezed out the hunter-gatherers). So no, I disagree that recent advancements are somehow lesser than earlier ones.

Also, focus on Zoom meetings or whatever, but we are in the midst of an information revolution that is as big as any that came before. Massive data processing will change our lives. Our descendants will look at us the same way we see medieval scribes who thought that anyone rich enough to possess 100 books had in his possession a mind-boggling amount of knowledge.

Expand full comment

On diminishing returns to technology over history, I don't think you're making an apples-to-apples comparison here. I'm no anthropologist, but my guess is fire-making technology was adopted across a long period of time, and I'm certain agriculture was. Of course, the reason for that is people didn't have the ability to spread technology quickly across the human race – but that's relevant here! That's a function of technology! Measuring each technology by "if everyone adopted this, how much would the world change" makes everything look like small potatoes (if you will) compared to agriculture, but doesn't factor in *how long it takes for everyone to adopt the technology*. I'm also no economist, and I don't know how economists account for this, but doesn't using new technology to reduce the uptake time for further new technologies leads to *increasing* productivity returns to technology? And couldn't that dynamic outweigh the diminishing returns dynamic in certain scenarios?

Expand full comment

It's undoubtedly true that in ancient times new technologies took longer to spread than is the case recently. Fire-making technology might well have been perfected in fits and starts, independently, by diverse, widely dispersed human populations over many thousands of years. And this was almost certainly the case with agriculture, too (indigenous Americans obviously didn't hear about agriculture from the Sumerians).

By I believe my point still stands: the telegraph (to cite but one example) represented a quantum leap in humankind's ability to transmit information. But it still wasn't nearly as big a deal as, say, the wheel. And the wheel wasn't as important as agriculture (and agriculture wasn't as groundbreaking as fire!). Although there are occasional examples to the contrary (as well as the "bursts" I mention above), it seems inarguable that the *broad trend* of technological innovation is one of declining disruptiveness at the margin; we just don't notice this effect because the impact of all this innovation (down through the ages) is cumulative. How could it be otherwise? What could humans possibly come up with that will be as revolutionary to us as fire or farming or the sail or gunpowder were to our ancestors?

The diminishing returns trend that has characterized the trajectory of human technological advancement seems undeniable to me despite the fact that, indeed, as you point out, the earliest of these required more time (in some cases much more time) to be perfected and disseminated. In any event, the overall point I'm trying to make is: *of course* innovation is slowing; it's been happening for millennia.

Expand full comment

What about dysgenics and fall in cognitive ability over time because of the negative iq-fertility correlation?

Expand full comment

I'm not going to get into the techno-optimist versus techno-pessimist debate per se but simply focus on the problem of measurement of TFP. The trick in ferreting out measurement of TFP, the so-called "residual factor" is how to specify the growth equation you are working with. The procedure I have come up (basing my reasoning on my reading of the works of Simon Kuznets and Edward Dennison) with my own work on growth is to use a growth formulation with the following inputs: (1) an augmented labor input that takes into account hours worked and efficiency of each hour's worked, namely L* = h e(h) W. In effect you are taking into account the flow of labor services that reflects among other things the age/gender composition of the labor force and - potentially - the deskilling of workers due to the substitution of capital for labor: (2) capital flow K* = K q(K) where q is the quality of capital reflecting embodiment of technology and institutional change in capital; (3) augmented land input, La = La q(la) that reflects the way land is used. If you do this you can factor out what I believe is a true residual, one that takes into account infrastructure embedded in land and capital. The point is much of the TFP measured by other methods disappears using this approach. In particular investment in capital and land loom larger. The possibility that some economies are reducing the quality of land and or undermining the efficiency of labor by investing in automation comes to the fore. If AI for instance is deskilling labor one might conclude that automation has a negative edge to it; something I believe is true. What of the conclusions flowing from this is the worsening income distribution seen in the declining share of labor in national income accounts - observed for a variety of countries - may reflect a bias in the way capital quality is being improved, away from labor augmenting technical change toward labor quality reducing technical change.

Expand full comment

In some ways our current situation resembles that of 3rd century Western Europe. Rather than apply Roman know-how to maintaining and improving the Western Empire, the emphasis as on better control of the peasantry.

Expand full comment

TFP is also a product of managerial innovation which is, again, not the same as technological progress. One big reason for high TFP growth in the 1930s as better industrial design (e.g. Dreyfuss), improved processes (e.g. the Gilbreths) and changing management decision structures and incentives (e.g. switching from sector to route management in rail operations).

The modern process trend is to use technology to deskill jobs and provide tighter central management control. Centralized intelligence actually removes smarts from the periphery. In the past, employees with local expertise could compensate for home office idiocy. Now, that is impossible. Shorter job tenures also eliminate local intelligence. Sometimes good central procedures can compensate, but in the 1960s, a lot of factories failed in their efforts to modernize when they fired the obsolete old timers who understood how the process actually worked as opposed to how it was supposed to work.

Combine tighter central control with a trend towards mergers and oligopolies, and it is no surprise that productivity is stagnant. People used to joke about the Soviet Union and People's Bread Bakery #345, but now, that's where we get our baked goods, at one unit in a large centralized chain.

I propose a lot more anti-trust enforcement, breaking up large firms as was often done during periods of rising productivity. Stronger labor and higher pay would also encourage managers to develop and use local intelligence rather than running everything from the top down. Higher marginal tax rates would encourage for 90% rate executives to increase 10% rate work salaries. By building value in the company as an asset, rather than simply riding the inflationary wave, productivity could become a real corporate goal again. It isn't now.

Expand full comment

This piece answers compellingly the thesis if Cal Newport Stop. Breathe. We Can’t Keep Working Like This. https://nyti.ms/3kU4HdN

Where Newport has a good point that intersects with the educational attainment aspect of this piece is that senior managers impose many technological instruments and methods absent a willingness to invest in the learning necessary in order to realize the advantages of those instruments and methods. Updates appear automatically on computers with changes in features and functionality but without any concern for how they disrupt existing routines. The problem is not the technology, but the people.

Expand full comment

Is "subtle and circumspect" a polite way of saying Tyler Cowan's "The Great Stagnation" doesn't make clear claims? Or that he is successful in hiding his political allegiances? There was an awful lot of good criticism of it back then: how has it done over the past 9 years?

<a href="http://critiques.us/index.php?title=Reviews_of_%22The_Great_Stagnation%22">I have an index to criticisms of that book (libertarianism being my hobby horse.)</a>

'Amusingly, "The Great Stagnation" is a revival of Alvin Hansen's old idea of "secular stagnation", which said that "... the American economy would never grow rapidly again, because all the growth ingredients had played out, including technological innovation and population growth...." This was stated in the 30's, just before the record-breaking postwar boom.'

Expand full comment
author

Cowen's book is retrospective; he doesn't try to predict grand laws of technolgical evolution, as Gordon does and as a few others do. That is what I mean by "subtle and circumspect".

As for secular stagnation, it does involve a technolgical stagnationist component (and there have been many many technological stagnationist over the centuries), but it's chiefly a story about aggregate demand.

Expand full comment

If past performance is no guarantee of future results, and we can always drag Little Orphan Annie back on stage to belt out another rendition of Tomorrow, then why even bother talking about stagnation? And what happens when your evergreen argument for optimism bumps up against the "bend at the end"?

I'd like to learn more about the nature of technological revolutions. How do they play out? At their dawn, are we ever able to predict their course? Where are we in the the various revolutions of, say, the past one hundred years? Perhaps your next post will touch on some of these concerns.

Expand full comment
author

My argument for optimism is not evergreen! In fact it's quite specific to our times. I will definitely write about these topics more. To be brief, I think technological progress comes in waves, when we unlock new general-purpose technologies. And I think we are due for another wave, thanks to solar and biotech (and maybe ML).

Expand full comment
Mar 6, 2021Liked by Noah Smith

If progress comes in waves, then it seems one can always find reasons for optimism. In any case, I'm enjoying this series and look forward to the next instalment.

Expand full comment
author

Thanks! The next one will be the last and toughest -- I'll take on the argument that scientific ideas are getting harder to find.

Expand full comment

That red trend seems like pure noise mining. If you were to fit from just 2 years earlier to 3 years later, you’d be smack back in line with the green trend. Further, it doesn’t even step the green trend up - extrapolating forwards/backwards from before/after the red period predicts not just the trend but the data itself for the period after/before it. It just looks like noise to me.

Expand full comment
author

If it's noise, that just reinforces the central point that TFP is hard to predict! 😊

Expand full comment