Some enlightening charts here. More people means more ideas, more innovation, and more labor, key components of economic growth. This is especially true at the city level where, as I will discuss in an upcoming Risk & Progress essay, innovation and GDP grow superlinearly to population growth.
It also presents a stark warning for those who seek to “close” the borders to immigrants on the right, and those on the left that celebrate shrinking populations. Neither outcome will bring the utopian future they promise.
Dump a load of people down and they don't innovate on day 1.
Pragmatically, the ability to innovate depends a lot on training and culture. If you want innovation, a small number of well trained and well funded top talent may be better than a large number of people with basic and mediocre training.
What's more, the American economy can grow from innovations that happen elsewhere.
This is a weird article. Population has a large impact on total GDP, but much less of an impact on GDP per capita, yet it's the latter that this article is focusing on. GDP per capita is mostly a measure of productivity, which itself is mostly a function of a society's technological progress. Tech progress has been slowing quite a lot since the 1950s in nearly all areas except for computers. Forecasting GDP per capita into the future would need to almost entirely focus on which effect would dominate, either science continues slowing to a crawl such that productivity remains nearly constant, or computers (perhaps through AI) become so great that it leads to another era of rapid growth similar to the 19th century.
Total population levels would have almost nothing to do with it, outside of black swan events (e.g. total population crash) or something that impacts productivity/science like older people working less.
But remember, tech progress comes from people. It's a staple of models of economic growth that more people leads to more ideas and more innovation. I think we would both agree that the last ~century has seen:
1. increasing GDP/capita
2. Rising number of researchers
3. Falling productivity per researcher
In this situation, the only way to increase GDP/capita when research productivity is falling is to increase the number of researchers! And the only way to do that long term is to grow the population (i.e. if population were fixed, there's only a fixed number of researchers you can have).
What I haven't done here is try to establish whether the relationship is causal. If it isn't, pro-growth policy might not have the intended effect.
> It's a staple of models of economic growth that more people leads to more ideas and more innovation.
That innovation doesn't need to happen in any particular geographic region. Someone inventing more efficient machines will increase GDP/capita wherever the inventor lives.
Given how bad a lot of America's schools apparently are, it might be better for America's GDP/Capita if more people were living in places that had a functional education system.
Your analysis in "Extrapolating GDP per capita growth" makes no sense. You assume an increase in GDP/capita of 2%/year. Independent of population. So we could 10x the population or reduce the population to one American and the GDP/capita in wouldn't change based on your methodology. You're assuming the link between population and GDP growth rate, then using that to predict increases in GDP/capita as population increases, and then using that result to support the connection. Circular.
The reason I assume that GDP/capita will grow by 2%/year is because that's been happening for almost 150 years, do you see a reason why that will stop?
It's really just a second attempt to extrapolate wealth at larger populations assuming a certain population growth rate and GDP/capita growth rate. If you have a better idea for how to do this let me know!
> I would like to see someone more qualified answer this question. If these numbers are even remotely true, then simply staying the course and growing the American economy over the next century would create radical abundance.
This is assuming a fairly steady rate of economic growth. Personally I suspect this won't happen.
Personally I suspect that when AI gets smarter than us, it self improves very fast. The AI then get's nanobots which can double every hour until they cover earth, and then build a dyson sphere in a month.
This is Much faster than the 2% economic growth. This is economic growth that can easily double every day (for a few weeks).
This may be very good or very bad for the humans, depending on how the AI is programmed.
I think you are assuming a strong causal link between population and GDP per Capita.
I don't think that link is very strong in reality.
Pick any tiny place, like Liechtenstein, they have a GDP of $98k /Capita and a population of 40k.
That's Way off your trend lines for the USA.
Alternative model.
Population started growing when people had reliable access to enough food and basic hygiene. Population can stop growing if people use lots of contraception.
GDP/capita is all about economic efficiency. Think robots. If your ploughing fields by hand, it's impossible to be rich. If you have robots that can manufacture anything, it's really hard to be poor.
For some averagely paid who works at a car factory to be able to afford a car with a weeks wages, it is necessary for the factory to produce more than one car for every week of labor. If cars are manufactured by a long and arduous process and every part is filed into shape by hand, this is not the case. If cars are made by robots, this can be the case.
There are a few subtleties. But basically, average wealth per hour depends on what people can produce with an hours work. Which grows as we automate.
As a forecast for the future (conditional on continuing population growth) I don't think we need to assume anything about the causal link between the variables. As long as they remain correlated and population grows, the forecast should hold. Though I agree that the implications for policy depend heavily on the direction of causality!
I also agree that this only works within a single country at a time. Liechtenstein would need its own regression.
Thinking about average wealth per hour and automation, I think most of the progress in automating stuff comes from people. Like in your other comment, you can boost innovation by training researchers better, or by having a higher proportion of them, but both of these "dials" have a finite limit. A common point in economic growth theory is that only population can increase without bound. This is especially important in the context of empirically falling productivity per researcher. If we want to automate more, we need to train both more and better researchers, growing the population plays a role in that.
> As a forecast for the future (conditional on continuing population growth) I don't think we need to assume anything about the causal link between the variables. As long as they remain correlated and population grows, the forecast should hold.
There is a sense in which this is trivially true. If A continues to go up at this rate, and the relation between A and B stays about the same, then B goes up at about the same rate.
If (average global temperature/ the age of the president) goes up, and that continues to be correlated with GDP/capita, then the forecast should hold.
At some point you are just doing curve extrapolation, but with a random extra variable that makes things more complicated.
I think the relation between population and GDP/capita is something that may well break in the next 100 years. And the population curve could radically change too.
> I think most of the progress in automating stuff comes from people.
True. So far. AI isn't doing that much automating yet. In the future, that is likely to change.
But not all the automation experts need to live in the USA to put up USA gdp/capita. This is part of why Lichtenstein can be rich. They use inventions that were made elsewhere.
Most of the people are not doing automating most of the time. Automating stuff takes rare and specific skills that most people don't have.
The AI doing all the R&D dial technically has a finite limit. But if it's anywhere near it's limit, humans are not needed for R&D by rather a large margin.
> A common point in economic growth theory is that only population can increase without bound.
If the theory says that, it's a rather stupid theory.
Given our current understanding of physics, population probably can't increase without bound. And of course, if there is some exotic physics that can be exploited in the distant future, it doesn't matter now.
> If we want to automate more, we need to train both more and better researchers, growing the population plays a role in that.
Are we talking about moving people from one part of the globe to another, or increasing global population? The first won't do much by itself. I mean it might make people be better researchers if moving means they can attend better schools.
Growing the population via birth would help a bit. Science reform would help a bit. Genetic engineering for higher IQ would lead to better researchers. If I wanted to Maximize research, I would choose +30 IQ over 2x population.
Some enlightening charts here. More people means more ideas, more innovation, and more labor, key components of economic growth. This is especially true at the city level where, as I will discuss in an upcoming Risk & Progress essay, innovation and GDP grow superlinearly to population growth.
It also presents a stark warning for those who seek to “close” the borders to immigrants on the right, and those on the left that celebrate shrinking populations. Neither outcome will bring the utopian future they promise.
Innovation is in the derivative.
Dump a load of people down and they don't innovate on day 1.
Pragmatically, the ability to innovate depends a lot on training and culture. If you want innovation, a small number of well trained and well funded top talent may be better than a large number of people with basic and mediocre training.
What's more, the American economy can grow from innovations that happen elsewhere.
This is a weird article. Population has a large impact on total GDP, but much less of an impact on GDP per capita, yet it's the latter that this article is focusing on. GDP per capita is mostly a measure of productivity, which itself is mostly a function of a society's technological progress. Tech progress has been slowing quite a lot since the 1950s in nearly all areas except for computers. Forecasting GDP per capita into the future would need to almost entirely focus on which effect would dominate, either science continues slowing to a crawl such that productivity remains nearly constant, or computers (perhaps through AI) become so great that it leads to another era of rapid growth similar to the 19th century.
Total population levels would have almost nothing to do with it, outside of black swan events (e.g. total population crash) or something that impacts productivity/science like older people working less.
But remember, tech progress comes from people. It's a staple of models of economic growth that more people leads to more ideas and more innovation. I think we would both agree that the last ~century has seen:
1. increasing GDP/capita
2. Rising number of researchers
3. Falling productivity per researcher
In this situation, the only way to increase GDP/capita when research productivity is falling is to increase the number of researchers! And the only way to do that long term is to grow the population (i.e. if population were fixed, there's only a fixed number of researchers you can have).
What I haven't done here is try to establish whether the relationship is causal. If it isn't, pro-growth policy might not have the intended effect.
> It's a staple of models of economic growth that more people leads to more ideas and more innovation.
That innovation doesn't need to happen in any particular geographic region. Someone inventing more efficient machines will increase GDP/capita wherever the inventor lives.
Given how bad a lot of America's schools apparently are, it might be better for America's GDP/Capita if more people were living in places that had a functional education system.
Your analysis in "Extrapolating GDP per capita growth" makes no sense. You assume an increase in GDP/capita of 2%/year. Independent of population. So we could 10x the population or reduce the population to one American and the GDP/capita in wouldn't change based on your methodology. You're assuming the link between population and GDP growth rate, then using that to predict increases in GDP/capita as population increases, and then using that result to support the connection. Circular.
The reason I assume that GDP/capita will grow by 2%/year is because that's been happening for almost 150 years, do you see a reason why that will stop?
It's really just a second attempt to extrapolate wealth at larger populations assuming a certain population growth rate and GDP/capita growth rate. If you have a better idea for how to do this let me know!
>It's really just a second attempt to extrapolate wealth at larger populations
That's my issue with this, it's not. You're using a per capita number so it's completely independent of population.
> I would like to see someone more qualified answer this question. If these numbers are even remotely true, then simply staying the course and growing the American economy over the next century would create radical abundance.
This is assuming a fairly steady rate of economic growth. Personally I suspect this won't happen.
Personally I suspect that when AI gets smarter than us, it self improves very fast. The AI then get's nanobots which can double every hour until they cover earth, and then build a dyson sphere in a month.
This is Much faster than the 2% economic growth. This is economic growth that can easily double every day (for a few weeks).
This may be very good or very bad for the humans, depending on how the AI is programmed.
I think you are assuming a strong causal link between population and GDP per Capita.
I don't think that link is very strong in reality.
Pick any tiny place, like Liechtenstein, they have a GDP of $98k /Capita and a population of 40k.
That's Way off your trend lines for the USA.
Alternative model.
Population started growing when people had reliable access to enough food and basic hygiene. Population can stop growing if people use lots of contraception.
GDP/capita is all about economic efficiency. Think robots. If your ploughing fields by hand, it's impossible to be rich. If you have robots that can manufacture anything, it's really hard to be poor.
For some averagely paid who works at a car factory to be able to afford a car with a weeks wages, it is necessary for the factory to produce more than one car for every week of labor. If cars are manufactured by a long and arduous process and every part is filed into shape by hand, this is not the case. If cars are made by robots, this can be the case.
There are a few subtleties. But basically, average wealth per hour depends on what people can produce with an hours work. Which grows as we automate.
As a forecast for the future (conditional on continuing population growth) I don't think we need to assume anything about the causal link between the variables. As long as they remain correlated and population grows, the forecast should hold. Though I agree that the implications for policy depend heavily on the direction of causality!
I also agree that this only works within a single country at a time. Liechtenstein would need its own regression.
Thinking about average wealth per hour and automation, I think most of the progress in automating stuff comes from people. Like in your other comment, you can boost innovation by training researchers better, or by having a higher proportion of them, but both of these "dials" have a finite limit. A common point in economic growth theory is that only population can increase without bound. This is especially important in the context of empirically falling productivity per researcher. If we want to automate more, we need to train both more and better researchers, growing the population plays a role in that.
> As a forecast for the future (conditional on continuing population growth) I don't think we need to assume anything about the causal link between the variables. As long as they remain correlated and population grows, the forecast should hold.
There is a sense in which this is trivially true. If A continues to go up at this rate, and the relation between A and B stays about the same, then B goes up at about the same rate.
If (average global temperature/ the age of the president) goes up, and that continues to be correlated with GDP/capita, then the forecast should hold.
At some point you are just doing curve extrapolation, but with a random extra variable that makes things more complicated.
I think the relation between population and GDP/capita is something that may well break in the next 100 years. And the population curve could radically change too.
> I think most of the progress in automating stuff comes from people.
True. So far. AI isn't doing that much automating yet. In the future, that is likely to change.
But not all the automation experts need to live in the USA to put up USA gdp/capita. This is part of why Lichtenstein can be rich. They use inventions that were made elsewhere.
Most of the people are not doing automating most of the time. Automating stuff takes rare and specific skills that most people don't have.
The AI doing all the R&D dial technically has a finite limit. But if it's anywhere near it's limit, humans are not needed for R&D by rather a large margin.
> A common point in economic growth theory is that only population can increase without bound.
If the theory says that, it's a rather stupid theory.
Given our current understanding of physics, population probably can't increase without bound. And of course, if there is some exotic physics that can be exploited in the distant future, it doesn't matter now.
> If we want to automate more, we need to train both more and better researchers, growing the population plays a role in that.
Are we talking about moving people from one part of the globe to another, or increasing global population? The first won't do much by itself. I mean it might make people be better researchers if moving means they can attend better schools.
Growing the population via birth would help a bit. Science reform would help a bit. Genetic engineering for higher IQ would lead to better researchers. If I wanted to Maximize research, I would choose +30 IQ over 2x population.