We often hear about the growth rate of real GDP in various countries. But it’s important to know, especially when the growth rate is low, what the growth rate of real GDP per capita is. Take Japan, which has had a feeble growth of real GDP. But it’s also had a feeble growth of “capitas,” that is, population.
Here’s what Alex Pollock wrote in August 2016:
If we measure by growth in real gross domestic product (GDP), without considering changes in population, Japan’s economic growth is far behind that of the United States. From 2000 to 2015, its real GDP grew an average of 0.72 percent per year, while U.S. real GDP grew an average of 1.77 percent.
In average growth rates, more than 1 percent per year is a big difference, indeed, as it compounds over time. Over 15 years, this annual growth rate difference would add up to U.S. GDP being 30 percent larger, compared to 11 percent larger for Japan, a difference of 19 percentage points.
However, economic well-being is not measured by aggregate GDP, but by GDP per capita. The question is how much production there is per person. In this case, measuring per-capita growth gives us a very different outcome.
In 2015, Japan’s population was essentially the same as it was in 2000, with an average annual growth rate of 0.01 percent. The corresponding annual growth rate of the U.S. population was 0.87 percent. So the U.S. added 39 million more people over the period to provide for.
Thus real GDP growth per capita in Japan was 0.71 percent per year. In the United States, it was 0.89 percent – a much more similar number. The growth rate advantage over Japan, measured per capita, is reduced to a modest 0.18 percent.
Of course, real GDP matters too. One reason Japan will likely have a much worse (than our) problem of funding seniors’ benefits over the next few decades is that Japan’s population growth is essentially zero. So there are not as many young people, relative to the population of seniors, to pay those payroll taxes and other taxes that fund the benefits.
Maybe the Japanese government should consider allowing people age 18 to 35 to immigrate. A million of them a year would help.
HT2 Jeff Hummel.
READER COMMENTS
Alec Fahrin
Nov 4 2017 at 1:02pm
A bigger issue here is that Japan’s GDP per capita is $35,000 while our’s is $55,000, yet their GDP per capita is growing slower.
I’m a big believer that feelings and perceptions matter as well. If Japanese see their GDP in permanent stagnation, they’ll feel weaker than the US even though both have the same per capita growth rate.
That’ll weigh on Japan in the long-term (and has if you compare 1990-2015 instead of 2000-2015).
Also, an interesting fact. Japan has added 350,000 immigrants since 2013.
Todd Kreider
Nov 4 2017 at 3:19pm
GDP per capita growth since 2000:
Japan 0.8% per year, currently $38,000 (2010 dollars)
US 1.0% per year, currently $55,000 (2010 dollars)
So real per capita growth has been almost identical for the past 16 years.
David Henderson’s projection for Japan is exactly what academics and journalists have been doing for 15 years: run simple population growth projections out to 2050 and 2060 for Japan and assume no other changes in health or automation despite strong evidence that both will uproot social security systems and health care in all countries by 2030.
Japan doesn’t need to accept more immigrants, although probably will at around its current rate.
David R Henderson
Nov 4 2017 at 4:33pm
@Todd Kreider,
David Henderson’s projection for Japan is exactly what academics and journalists have been doing for 15 years: run simple population growth projections out to 2050 and 2060 for Japan and assume no other changes in health or automation despite strong evidence that both will uproot social security systems and health care in all countries by 2030.
It’s true that I projected Japan’s population growth rate. Are you willing to say what you think the changes will be that will uproot social security systems and health care, and also what you mean by “uproot?”
Mark Bahner
Nov 4 2017 at 7:38pm
Probably he’s referring to estimates such as computers will add 10 billion human brain equivalents (HBEs) by 2030, and over 1 *quadrillion* HBEs by 2050. (That’s right…1,000,000,000,000,000…1 followed by 15 zeros.)
Why Ray Kurzweil says the Singularity will occur before 2050
So it won’t matter how many people a country has.
Todd Kreider
Nov 5 2017 at 12:10am
Many medical breaktroughs are coming in the 2020s including far more effective cancer treatments, the end of Alzheimer’s and Parkinson’s disease, health pills that reduce the risk of getting heart disease , stroke and cancer as well as increase strength and endurance. Then there is CRISPR…
The probllem with those writing on Japan almost never have any science background so set advances out to 2050 and 2060 to zero. (Tyler Cowen said in 2011 that he doesn’t expect any medical advances out to 2030 despite so much occuring now in the labs and as the computer exponential curve is holding up.)
Automation will displace workers, not just assist them. There will be new jobs as well but by 2027 the entire career/retire paradigm will likely look very different than today. And what about changes by 2037? Even then, the technological changes are being set to zero 20 years into the future as well as 40 years out to 2060.
Remember Zeke Emanual’s “Why I hope to die at 75”? He is saying when 2032 medical technology and treatments come around, he will have no part of it. His op-ed was just a publicity stunt but it does provide an example why you can’t assume an 80 year old in 2050 or even 2030 will look, feel or think like a typical 80 year old today.
So social security will also change significantly but at this point can’t tell how in any detail.
Todd Kreider
Nov 5 2017 at 5:19am
I didn’t notice that Mark posted before me. One difference I have with Ray Kurzweil is that since the late 1980s, I haven’t been convinced that the exponential computer power curve will continue to accelerate into the 2040s and 2050s. To me, the numbers I extrapolated got very, very large between 2040 and 2060 where society would be essentially unrecognizeable to sommeone visiting those decades from the year 1989.
I’d usually no matter what the year that Moore’s Law should continue another 10 to 15 years whereas Kurzweil has been almost 100% certain that the curve would continue after Moore’s Law had ended in the early 2020s to beyond the 2040s and 2060s and into the 22nd century. His prediction is that my “black hole” for society, what he calls “The Singularity”, (which I also used but not to non math type friends as much), will arrive at 2045. I also said I thought the probability of this happening was maybe 50% where I just meant that I didn’t know.
Yet there can be tremendous changes in medicine and with jobs by 2030 even if nothing as life altering as a Singularity occurs. It has been interesting watching essentialy everyone ignore future technological changes when writing on Japan.
[We apologize, sincerely, for the occasional delay in posting comments. Sometimes legitimate comments are temporarily trapped by our spam filter. Both Mark’s comment and yours were accidentally and temporarily trapped. I try to do a pass through the spam at least once daily, even on weekends. We appreciate your patience, and we understand that when a conversation is underway, waiting 12-24 hours can be a frustrating eternity. Thanks for remarking to readers that you hadn’t seen Mark’s comment. Indeed, you couldn’t have seen it because it hadn’t been released from spam. But, to be fair to each commenter, when comments are released from spam, they receive a date/time stamp of the original date and time they were submitted.–Econlib Ed.]
David R Henderson
Nov 5 2017 at 8:15am
@Todd Kreider,
Thanks.
I would think, though, that advances in medical technology that extend life, which I also expect, will make the social security problem even tougher.
Abe
Nov 5 2017 at 9:18am
Economic well-being is not measured in GDP per capita, particularly when the population being measured changes over time. The growth in the U.S. population over that period was largely due to low-skill immigration; i.e. people who are poorer than average. Those people lowered the U.S.’s GDP per capita merely by being included in the calculation, but it is much less clear (probably false) that they made other Americans poorer. If you were to measure the “GDP per capita” of just the people who were in the U.S. in 2000 through today, you would find that it grew faster than the 0.89 figure reported.
If I were to join the New York Knicks, the average height of the team would be less. But that does not mean Kristaps Porzingis would shrink.
David R Henderson
Nov 5 2017 at 9:35am
@Abe,
Well done.
Todd Kreider
Nov 5 2017 at 10:23am
That might be the case if social security wasn’t going to have enormous pressure to change but my point is that such a different society will also have a different type of safety net as the current system won’t make sense.
@Abe: The U.S. has a population of around 315 milion and the number of immigrants added each year is a tiny fraction of that so the distortion is trivial. Also, where is the evidence that “The growth in the U.S. population over that period was largely due to low-skill immigration” as you claim? (my emphasis)
There is a distortion between the US and Japan in terms of inequality as the post tax/transfer gini coefficient for the US is around .45 whereas Japan is at .38 so for the middle 70% of Japanese, the standard of living is somewhat closer to that of the middle 70% of the US.
Mark Bahner
Nov 5 2017 at 1:18pm
No biggie! (I did think briefly of re-doing the comment without what I assume was the offending link to my blog post about the future of AI. :-))
EconLog represents the very best of the web, and an occasional glitch can’t take away from that.
Best wishes,
Mark
David R Henderson
Nov 5 2017 at 1:54pm
@Todd Greider,
That might be the case if social security wasn’t going to have enormous pressure to change but my point is that such a different society will also have a different type of safety net as the current system won’t make sense.
I agree that it won’t make sense. Not that it ever did make sense, but I agree that it will make even less sense. That doesn’t mean that it will change much. Remember that these are government programs.
Jon Murphy
Nov 5 2017 at 2:27pm
@Todd Kreider
Also, where is the evidence that “The growth in the U.S. population over that period was largely due to low-skill immigration” as you claim?
If I am understanding your objection (and please correct me if I am wrong), you are questioning where the evidence is that immigrants have been largely unskilled. Brookings had a report out about 8 years ago now and found approximately half of the US immigration population has a high school degree or less (See here).
The U.S. has a population of around 315 milion and the number of immigrants added each year is a tiny fraction of that so the distortion is trivial.
Approximately 11% of the US population are immigrants.
The U.S. has a population of around 315 milion and the number of immigrants added each year is a tiny fraction of that so the distortion is trivial.
The US population grows by approximately 22 million people per year, one million (~5%) of which are immigrants, so it’s not that trivial. It’s small, smaller than it should be, but I’d disagree it’s trivial.
Mark Bahner
Nov 5 2017 at 9:04pm
Hi,
Here’s how I look at it: In 2030, the world will be adding about 10 billion human brain equivalents per year. Let’s say that there’s nothing extraordinarily difficult about creating the other parts of the human body…if needed. (For example, an autonomous vehicle doesn’t need a robot with hands and fingers or feet.)
So let’s say that we’re adding the equivalent of 10 billion human beings every year, just as capable as humans for the overwhelming majority of jobs. Since there are less than 10 billion people right now, all humans could just stop working, and the machines would take over, and the world GDP would actually increase slightly. (Or the *nominal* GDP would contract, but the *real* GDP would increase, because the nominal contraction would be the result of lower prices for just about everything.)
Then let’s say that progress in computers essentially stops, and we just keep adding 10 billion human equivalents in 2031, 2032, 2033, etc, rather than adding exponentially. That means that in 2050, the effective world population is something like 200 billion, rather than 1 quadrillion. That’s still a completely different world from even 2017.
Even circa the 2030s, there should be no problems with Social Security, because no one will need to work.
Todd Kreider
Nov 6 2017 at 8:28am
The above has been my guess with respect to social security for the past 15 years or so: social security will just fade away around the late 2020s or early 2030s. I usually just say I don’t know because I may be missing something, but I think Mark will be correct.
There is another scenerio that I almost never see discussed about the exponential computer power curve. Instead of it stopping at 2030 or even earlier in 2025, it could slow significantly but still rise each year and if continued, the world from a technological sense in the 2070s or 2080s would look like what was expected for the 2040s by those who agree with Kurzweil’s timelime. so, rather than shoot straight into a Singularity, the world would instead ease into it.
But the point that can’t be emphasized enough is that computer acceleration continuing to 2030 is still a ‘completely different world’ from today, as Mark wrote. You don’t need a Singularity for that.
As I said, it has been fascinating that so many social scientists set technological change to zero for the next several decades to make some alarmist point about Japan’s or China’s future. Since almost unanimous, it makes me think they aren’t really serious about this topic.
Comments are closed.