The Effect of Aging on Economic Growth in the Age of Automation

Many of you requested that topic, here is a new NBER paper by Daron Acemoglu and Pacual Restrepo:

Several recent theories emphasize the negative effects of an aging population on economic growth, either because of the lower labor force participation and productivity of older workers or because aging will create an excess of savings over desired investment, leading to secular stagnation. We show that there is no such negative relationship in the data. If anything, countries experiencing more rapid aging have grown more in recent decades. We suggest that this counterintuitive finding might reflect the more rapid adoption of automation technologies in countries undergoing more pronounced demographic changes, and provide evidence and theoretical underpinnings for this argument.

This is not the final word, but it is interesting to see that the expected effect does not materialize.

I am wondering, by the way, if I might not cover all the requests you made in response to that post, except of course the ones I already have done in the past.


This is interesting, but I think a bit premature. We haven't seen a complete cycle yet.

There may be hope for Italy yet.

From a strictly theoretical point of view, yes.
Meanwhile, whenever I go to visit the friends that I knew in college, I go to Plymouth, Bristol, London, Southampton, Berlin, or Weimar.

They look at GDP2015/GDP1995 v the change in the ratio of population over 50 to that 20-49 as if there should be a linear relationship between growth and change in the rato over a 20 year period. Why not plot annual GDP growth v the population ratio for each year?

I can only read the abstract: they say " If anything, countries experiencing more rapid aging have grown more in recent decades "

Is this really true of Japan ? (GDP Growth Rate in Japan averaged 0.51 percent from 1980 until 2016) from

From the article:

"The relationship between aging and adoption of robotics technology is established in Acemoglu and Restrepo (2017) using data from the International Federation of Robotics (IFR), which provides information on industrial robots across a range of industries for 49 countries.........This figure excludes Japan, since the IRF notes that Japanese data are not comparable over time because of a change in classification"

So, it's data from 169 countries but no Japan. I have no quantitative way of telling if taking Japan out of the sample has a significant impact on the article's conclusions or not.

That is the nominal growth rate.

The real GDP/capita growth rate for Japan from 1980 to 2015 was about 2.4%.
The real GDP/capita growth rate for the U.S. from 1980 to 2015 was about 2.3%.

The real GDP/capita growth rate for Japan from 2000 to 2015 was about 0.7%.
The real GDP/capita growth rate for the U.S. from 2000 to 2015 was about 0.9%.

Note the similarity.

Duh! Of course, the GDP deflator is an "inflator" in the case of Japan. Even zero nominal growth, means real growth with ongoing deflation.

In the US, at least, the enormous amount of money generated by horribly expensive health care for the elderly, inflated house prices due to close to record low mortgage rates, higher auto sales (set to reverse) due to subprime auto loans, along with the trillions of dollars generated by the world's largest defense costs have kept GDP numbers from going negative..

The intuition relies on a series of misunderstandings (the whole "greying of Europe" genre):

(1) Median age of the whole population changes a lot with higher/lower fertility, but median age for the population of working age only slightly. The latter is relevant. The former means you have more pensioners, but also fewer children who contribute little or nothing. Both changes are of about equal size but opposite sign so that the share of working age remains pretty stable.
(2) With lower fertility you have slightly (!) fewer people of working age below 40 years, and slightly more above 40 than with replacement or above fertility. However, great scientific contributions, new patents, etc. peak around age 40, and the distribution is rather symmetric. So shifting some people from below to above 40 mostly cancels out.
(3) Actually, older people at least contribute something to innovation, while children don't. So an older population could even be more innovative.

Run a regression for innovation indices (e.g. Bloomberg Innovation Index) versus median age, and you find a positive slope (older = more innovative). Take only developed countries, and the relationship is flat. That should be expected from the above considerations: no big shifts for working age population, most effects cancel out, maybe a slight edge for older societies.

The naive intuition is that older societies collectively sit in a wheelchair and play with their dentures. But aging of a society is not like aging of a person. Conflating the two leads you to "intuitive", but false conclusions.

"Several recent theories emphasize the negative effects of an aging population on economic growth, either because of the lower labor force participation and productivity of older workers or because aging will create an excess of savings over desired investment, leading to secular stagnation."

Oh, no, secular stagnation! Excess savings! Lower labor force participation! Oh, no! Gotta get that nose to the grindstone. Put that money to work so the financial sector can thrive. Increase productivity! This BS sounds like something the Politburo would put on billboards. Or the Krauts. Arbeit macht Frei!

Countries with aging populations are the most economically developed countries, while countries with young and rapidly growing populations are the most economically undeveloped countries. Hence, one would expect the former countries to continue to outperform the latter countries, if for no other reason than their starting lines differ so radically. And, of course, there are cultural differences. Because of the many differences, the better comparison is the historical trend line for economic growth of a country and the more recent trend line for the same country after its population began aging.

Acemoglu talks about the importance of central institutions that provide justice, enforcement of contracts and education. Those goals are easier to achieve with an older population structure. Acemoglu´s new findings are consistent with his good institutions = growth story. Anyway, he enriches the discussion by pointing the counterintuitiveness of research results.

Acemoglu may have found that economic worries are overrated. But what about social worries? Population decline followed Roman Empire's decline, population decline was a consequence that time. I'm no expert in the topic but the question remains interesting: could population decline trigger something bad or the fears are unfounded?

Of course, in the ancient world just maintaining the population required every woman to have at least six or seven pregnancies, since many if not most mothers and their babies died in childbirth and life expectancy was much shorter for both men and women. The life cycle of a woman in the ancient world has been described as birth, sex, death, and decay, inducing many "liberated" young women to choose asceticism, threatening the tribe's survival. Indeed, fear of population decline is so ingrained in us that birth control is prohibited or discouraged in many if not most cultures, even many economists are subject to the fear by exaggerating the consequences. Humans just can't get over their past.

Deaths of women in childbirth actually peaked in the 19th century due to the medicalization of childbirth before the need for hygiene was understood. Fewer women died from childbirth attended by midwives than doctors- the doctors were more likely to spread sepsis from other ailing patients. Yes, women died from childbirth in antiquity and the Middle Ages, but at rates lower than is often assumed

As the American system collapses like a castle of dominoes, the new oligrches will seize increasingly bigger fortunes and the workers' share of the economy will experience a free fall. Robot workers can't rebel.

Do people really think aging demographics will have a large effect on GDP growth per capita? A little, sure, I suppose, but its overall GDP growth that aging demographics will surely hit hard.

Unfortunately, I still not convinced as:

1) Japan & Germany have had decent growth but depend upon high exports and government debt. At some point, I believe Japan is the canary for the demographic spiral on government which might last another decade or so.
2) I still think the low birth totals are accenting the debt problems of Italy, Ukraine and Greece. Not primary cause but it is increasing the problems of debt crisis.
3) It does slow the total economic growth. Population size is the one variable that heavily effects both the AS and AD curves. Total growth has fallen and with less consumers buying stuff, that limits productivity growth.
4) The Great Recession really did show over several years nomial/real wages can go down but the working classes are also making the decision to have smaller families. (The decrease of wages in the US is masked by lower medical benefits. Ask anybody in an office)
5) At least in the US, the labor market is becoming increasingly tight and most businesses need workers to expand but they do not want to hire them at increasing wages.

5) If the market is setting wages higher employers need to accept that fact, just as the rest of us have to deal with prices that increase.

China is the prime example of a country with very rapid growth and an aging population because of the old one child policy. I wonder how much of his results was due to China and if you remove China from the data do you still get the same results?

Demographers predict (the mid-range prediction) that China's population will continue to grow through the first half of this century, and then decline by over 300 million in the second half. Think about that: China's population will decline by an amount equal to the entire population of the United States. You think that might have social, political, and economic consequences?

Actually, I believe for 20 - 30 years low birth rates accelerate growth in a nation and I believe one of the reasons for Japan Inc. success in 1970 - 1990. Why:
1) Small families keeps the cost of living down and keeps lower wages (esp in the cities) 2) Less education investment needed and family focuses more attention on children

+1. Isn’t this just math, or more precisely measurement error? If you magically eliminated all child-raring and kept everything else the same, you instantly get a GDP bonus and a GDP-per-capita bonus, and you’d grow more for about a generation. Then you’d see GDP fall as the working population falls, followed with lag by GDP-per-capita falling as you start to lose scale effects and certain things become uneconomic. The problem is that you weren’t correctly counting the productivity that went into child-raring, so when you eliminated it, you didn’t notice anything was missing for a long time.

Now if you have the Lite version of this, where only some women choose to be childless and some families choose to have fewer children, you get a Lite version of the effects.

One of the strangest things in modern economic measurement is that we count the GDP contribution of stay-at-home parents at zero. We should count them at some premium to the wages they are forgoing, because revealed preference shows that’s what it’s worth.

This seems like a fair point to me.

Even if a stay-at-home parent would be more productive working at a firm (which, as you suggest, is questionable given their revealed preference), it should only by this marginal productivity effect that is reflected in GDP if they suddenly decide that they were wrong and would rather work. Instead, when this parent goes to work, not only is their productivity in that out-of-home job now counted, but the productivity of having someone outside the home take care of the child is now counted (unless family watches them for free). While the latter suggests that we might also be missing some of the benefits of an older population in the GDP numbers (if grandparents are watching children for free while parents work), it seems like the effect you suggested is much more substantial.

I was wrong about the spelling of child-rearing...

There's some complexity related to taxes in my revealed preference point. When a parent stays at home, you get their full productivity without any tax loss. When they work, you see perhaps half that benefit, the after-tax portion. To you the taxes are pure loss, since you see no benefit from them. But "society" sees some benefit.

Short term I believe low birth accelerates growth but long term it will evidently slows down growth because:
1) Population size is the one big variable that effects AD and AS curves. In the short run, it effects the AD (Say Japan 1980s which exported excess goods) but long run effects the AS. Look at the US labor market today. It is tight on the skilled side.
2) Look at Japan's economy and government. Government debt is dangerously high and several big corporate names, especially Sony, look like GM did 10 - 15 years ago. History could change.

Yeah, that's what I'd guess too.

Presumably there are optimal age distributions leading to maximum expected real GDP and maximum expected real GDP growth (and not necessarily the same distribution for each measure), with those numbers decreasing in either direction as you move away from the optimal distribution. Currently-older economies may be closer to the optimum than currently-young economies. That wouldn't mean that they can keep getting older and always be better off in terms of economic growth. The important question to try to understand is where the optimum is.

Sorry about the changed topic, but I happened to see the Peter Thiel/Cowen conversation referenced elsewhere:

The female program host is a high talker, just like the NPR ladies. This is another nail in the coffin in the hypothesis that all college professors are liberals, often voiced by Art Deco.

I probably should have said "Nail in the coffin for any denial of the hypothesis that all college professors are liberals"!

In my local community one of the chain fast food outlets has hired mostly retired people. I'm not sure how this happened, by design or by chance but the difference in service is amazing. Someone is always cleaning the tables and soda machines. The person behind the counter is polite and actually aware they are there to serve you not shout back and forth to their friends. And when the order is served up they don't just throw it on the counter and turn their back on you. Amazing!

And get off my @#%^& lawn! Boy, when I was your age....

I'm generally suspicious of economics but this paper lends support to my priors so I will praise it uncritically....

But to be serious, ask yourself about the "lower productivity of older workers?" Unless you're talking about string theory or increasingly rare manual labor, it doesn't exist. If it did, it would be reflected in the price for older people's labor. Right, libertardians? Because otherwise, there is massive discrimination against young people, discrimination that in an efficient market should not be allowed to exist.

Africa's population is growing at 500 million every 15 years. These people are going to have to live somewhere and I don't think Africa itself can possibly cope with that growth rate. The idea of a graying population would then probably only be a very short term problem as more intensified levels of migration are required.

If only half of those new people migrated that would be around 15,000,000 people a year looking for new opportunities.

What about lower populations leading to fewer brains working on research, leading to fewer discoveries?

It’s obvious that there is a lot of growth to come, but we then you can’t really stop such things instead it’s better to plan it out better. I always focus on working out things in proper strategy and being a Forex trader, I do it all so nicely through OctaFX, as they are very special because of wide range of features and facilities from having low spreads at 0.1 pips for all major pairs while there is also rebate scheme where I get 50% back on all trades, so all this is superb!

Comments for this post are closed