Population and real GDP per capita

I noticed a few comments to another of my posts about possible links between population size and economic performance.  My working assumption is that, on average, across all countries, there isn’t any such relationship.   Apart from anything else, if there were a positive relationship –  that was more than chance –  it would suggest that two countries merging would increase their respective real incomes.  And yet for at least the last 70 years, we’ve had steadily more countries emerging.  No doubt economics isn’t the only thing at work in those choices –  people might be willing to pay a price to be “free” and self-governing –  but it isn’t likely to be an irrelevant consideration either.

But what do the data show?   Here I’ve just used the IMF World Economic Outlook database data for 2016.

The first chart shows the relationship –  for the 193 countries/territories the IMF reports data for –  between real GDP per capita (in purchasing power parity terms) and population (each dot is one country).   The population term is expressed in logs.

popn and real GDP pc

As (I would have) expected, there is basically no relationship at all.   The simple linear regression line is actually slightly downward sloping, but that won’t pass any test of statistical significance.  Perhaps one could craft a story in which the top 10 countries (in terms of per capita income) all have quite small populations –  the biggest is around 5 million people – but since oil plays a big part in most of those individual cases even then one shouldn’t make too much of the point.

And here is the chart if we look only at the countries with populations from 0.5 million (a tenth of New Zealand’s) to 50 million (ten times New Zealand’s).  Since that is a much more compressed scale –  not everything from Tuvalu to the People’s Republic of China –  this time the population variable isn’t expressed in logs.

popn and real GDP pc 0.5 to 50m For those with sharp eyesight, New Zealand is a dot coloured orange.

Again, there really isn’t any sort of relationship.  Again, the simple regression line is downward sloping, but there are lots of countries with very small populations and very low per capita incomes.   But even within this more-compressed range of populations, there is no sign at all of any sort of upward sloping relationship –  the idea that, on average, a higher population will be associated with higher per capita incomes.

Of course, within each of these dots there are complex historical relationships, as to how population in any particular country came to be what it was (some about conquest, making big countries out of small one; sometimes the historical carrying capacity of the land; in some the role of slavery (eg forced depopulation from Africa), in others the role of immigration policy.   Some locations offer better prospects than others and will, typically, have attracted or retained poeple accordingly.

But this post isn’t attempting to get into any of that. it is simply observing that at the most elementary level of numerical analysis there is no sign that countries with larger populations tend to be richer (whether as a matter of cause, or of effect).

23 thoughts on “Population and real GDP per capita

  1. I think trade is a big part of the explainer here. In the old days when trade was low empires tended to be the dominate form. Now trade is much greater the effort of maintaining empires is not worth it for many.

    North Korea has low trade and is doing poorly, while a country in the Eurozone with near perfect trade opportunity is much less impacted by being small.


  2. quote :- “there is no sign that countries with larger populations tend to be richer”

    can the opposite be said

    countries with larger populations tend not to be richer


    • China in 2000 or China in 2018 or China in 2050? Perhaps it is more to do with how you use that population. Any resource can be squandered but any resource if used properly, directed properly does yield immense returns.


  3. “”As (I would have) expected, there is basically no relationship at all. The simple linear regression line is actually slightly downward sloping, but that won’t pass any test of statistical significance. “”

    To my untrained eye it seems that large populations are slightly less productive than small ones. That matches our experience with Auckland council since the merger; bloated bureaucracy with our councilors having less control. It is also my experience of most IT departments – the bigger they are the less effective. And I have vague memories of hearing that mid-sized companies are the most profitable and the majority of mergers result in a reduction in profits.

    Of course countries are different to those three organisations. But the distance between me and my concerns and Jacinda Ardern is smaller than say a retired Chinese computer programmer and President Xi or American equivalent and President Trump. The famous 6 degrees of separation is probably 1 or 2 between me and our NZ government but in China it might well be the full 6 degrees. A drastic example would be the estimated 45m who died in Mao’s ‘Great leap forward’ 1958-62; a book I read recently said it was impossible to tell whether it was caused by Mao actively trying to kill his own citizens or just incompetence resulting from nobody wanting to pass the bad news up through their hierarchy.

    Liked by 1 person

    • The problem with metrics is that it can be manipulated. My mother was in Auckland hospital for 4 months but they kept moving her around to different wards and different beds on the same level. I can understand when she had move from different levels for surgery to recovery. It started to dawn on me that Auckland Hospital had a performance metrics measure of some kind based on how quickly you are able to rotate the use of the beds in the recovery wards.

      Auckland Council also manipulates metrics ie time to answer queries. It gives itself 15 days to respond but officers get around these 15 days by asking silly questions. They throw it back at you with some nut case query in order to restart the clock. Therefore if you check the metrics they would have done an excellent job because every query is answered by another query.


    • NZ does not have a large population at 4.5 million but it does squander it’s resources by having 10 million cows, 1 million deer and 30 million sheep for a very low $15 billion export GDP. That is why NZ is full with 4.5 million people. Feeding 10 million cows is equivalent to feeding 200 million people with the equivalent waste handling, water usage and land resources.

      A highly government subsidised company like Samsung trades $350 billion with only 350k people.


    • It is also my experience of most IT departments – the bigger they are the less effective.
      and the more the top dogs rely on them the worse the performance.

      Companies run by the IT dept are hopeless. Fletchers being a good example.
      Went to get some pipes the other day. Waiting for an invoice and the office staff were waiting. One commented that I would have the pipes in the ground before their “new” system got the invoice out.
      No wonder our productivity is appalling.
      Had to use the IRD’s site the other day. My god what an abortion of a thing. Used to be easy whan I last used it. Now it just awful.

      That’s cost us how much.
      Like the Insus effort by the police.
      I think there is a connection between IT use and lack of increasing productivity. Perhaps Michael might look at that relationship.


      • There is no magic solution to computing disasters but here are a few tips – never computerise a good manual system – Japanese manufacturing has been very successful with its reluctance to computerise. Secondly never computerise a mess – once you sort out the mess it is surprising how little needs to be done – consider the Australian computer system that left our school offices taking on extra staff to sort out the payments.

        Try “CRASH: Learning From The World’s Worst Computer Disasters” by David Collins & Tony Bicknell.

        Liked by 1 person

  4. Michael

    The argument about population and productivity is Simon Kuznets thesis, which is explained in a number of papers and in his Nobel Lecture. (1) It is not the population of “developing” countries but “developed” countries only that matter. The emphasis is on developed countries. Your sample include heavily populated – less productive developing countries. These developing countries have explosive population growth with zero research efforts. (2) It is the growth rate of population in the productive developed countries not the level or log level that matter. (3) it is real GDP per hour-worked not real GDP per capita. Jones (2002) AER model and consequent literature, which I can cite many link LONG-RUN STEADY-STATE productivity growth to population growth. TFP growth is a function of GLOBAL RESEARCH EFFORTS GROWTH, which is tied to population growth in the DEVELOPED countries only. Population growth in developed countries ==> Research efforts ==> new ideas ==> TFP productivity ==> GDP per hour. So increasing population in NZ, for example, may contribute noting to productivity in NZ and elsewhere. If population growth continues to decline in the advanced world, research efforts will decline proportionally and will productivity everywhere. That said, Kuznets explained these issues in a number of classic paper.

    Liked by 1 person

    • When population increases in the developed world it’s quality of life decreases (therefore it is less productive – in a sense ?) There isn’t much green space for the citizens of Tokyo.

      Liked by 1 person

    • Weshah

      You jumped a long way ahead of me in this comment. My very quick post was simply a response to a commenter here the other day who provided a list of the 10 largest population coountries in the world and suggested this was a considerable economic advantage. My point simply was that there is no obvious correlation, in the cross section, between population size and GDP per capita (or productivity). I’m not aware of any serious observer arguing otherwise – since otherwise (say) chopping the US or China in two would make a material difference to their living standards and I don’t know of anyone who believes that (having said that, obviously the extent of trade and currency barriers matters, and one of the arguments why there are many more states now than there were 80 years ago is that trade barriers are a lot lower – the price of self-governing is less than it was).

      I had intended to pick up the issues around population growth rates today (and have done so previously). Kuznets’ is, of course, one hypothesis, and perhaps the seed of probably the only sort of story on which more people is likely to boost per capita living standards. I remain pretty sceptical about that story – as distinct from reallocations within countries, such that many more now live in productive cities – and was interested to find Friedman similarly sceptical in one set of comments he made on Kuznets.


      • NZ has a huge reserve of metals and mining based industry potential. But this mining wealth has been squandered due to environmental concerns.

        Similarly population is a resource. You can either direct it into high skilled high productivity manufacturing or you can direct them into low skilled and low productivity service based industries.


      • That story suggests onward and upwards. Thinking of Japan some electronics factories have closed (Panasonic in Tsunashima) and instead of Japanese buses New Zealand bus companies are buying Chinese Zhong Tong busses. What’s more there is near zero green space in big cities and nowhere to sit. I pointed out a few days back how child care workers have to wait their turn to get pregnant and are payed below average wages.


      • If Chinese buses sell better than Japanese buses then profitability outweighs productivity in global marketing. Cheaper unproductive laborers in China gives a higher margin return to the buyer than high quality robot manufactured buses I guess.


      • Michael

        You are not alone. Many people and economists agree with you that the positive relationship between population growth and productivity growth are not observed in the data. Friedman was skeptical of Kuznets. Solow’s model has the same sign like yours (negative). But Richard Quandt proved mathematically that Kuznets argument is correct if the production function is increasing-return-to-scale. Simple and straightforward. You know that increasing return to scale is the main assumption behind endogenous growth. Essentially, under such assumption one can double output without doubling inputs. This can happen when population growth (in advanced nations only) is high. This scale effect was treated in the second generation endogenous growth models. I am not sure that immigration helps either. If we keep admitting waiters and taxi drivers as you showed before we will get no growth for sure. Sorry to you and your readers for my long comments.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s