Scott Sumner  

The problem of population

PRINT
Henderson on Piketty, Part 6... Two Tullock Stories...

Tyler Cowen has a column in the New York Times that discusses the issue of population. I mostly agree with his policy recommendations, but what interests me is the underlying assumptions. What is the optimal global population? What is the optimal population for each country? How should global population be distributed?

Tyler notes that global population will increase sharply over the next century, with almost all of the growth occurring in relatively underdeveloped Africa and South Asia. In contrast, population will actually decline in some countries, and indeed is already doing so in Japan. So why should we care?

It's an area that will prove central to understanding whether nations will grow richer -- or will stagnate and lose global importance.
This begs the question of what we mean by "richer" and "stagnate" and "global importance." Later Tyler notes that many economists have steered clear of the difficult problem of population:
Many economists are uncomfortable with population issues, perhaps because they aren't covered in depth in the standard graduate curriculum, or because they touch on topics that may be culturally controversial or even politically incorrect. That's unfortunate. In the future, population economics -- and associated social issues -- are likely to be at front and center of our most important policy concerns.
This is probably correct, but leaves out one additional problem---we don't have a good model. In my area (monetary economics) I take population as a given, and look for policies that will maximize aggregate utility, or utility per capita. If we take population as a given then those two goals are identical. Not all economists are utilitarian, but most use utilitarian assumptions in their analysis.

Even if population is assumed fixed, utilitarianism raises all sorts of thorny problems. For instance, can we really make interpersonal comparisons? But if we allow population to be endogenous then the problems multiply exponentially. Perhaps the biggest problem is determining our objective function; what are we trying to maximize? (And of course, who is "we?") Is it total aggregate utility? Is it utility per capita? Those two objectives might lead to radically different policy conclusions.

For the sake of argument, let's assume that utility is positive, on average. Even this seems like a leap of faith to me; I can't even imagine how we could reach that conclusion scientifically. You'd expect the forces of evolution to program us with strong survival instincts even if most people "lead lives of quiet desperation." Nonetheless, it seems completely unproductive to make any assumption other than that most people are net positive in utility.

The much tougher problem is whether to focus on average utility, total utility, or some third category (which seems implied in Tyler's essay.) If average utility is the right criterion, the optimal global population might be quite small. Or it might not, we simply don't know. I've lived in both Australia (1991) and England (1986), which are near the extremes of population density for the developed world (England is far more densely populated than metro Atlanta, and Australia has 1/10th the US population in an area almost as large as the continental US.) It seemed to me that living standards were considerably higher in Australia, mostly because it was much less crowded. But that's obviously highly subjective; Australia lacks a city as sophisticated as London.

On the other hand if total utility is the right criterion, and if people in even poor countries are often surprisingly happy (as many surveys suggest), then the optimal population might be extremely large. Bryan Caplan has made a similar argument from a non-utilitarian perspective, as do religions like Christianity.

There are probably intermediate criteria that put some weight on both average utility and maximizing the number of geniuses (and hence culture and science,). Here I have something in mind that might view Germany as in some sense more successful than both Luxembourg and India, despite having a smaller total GDP than India and a smaller GDP per person than Luxembourg.

In any case, it seems clear to me that one reason that economists steer clear of the population question is that they don't have any confidence in any particular "model."

I also have a few observations about Japan's falling population, which is something that Tyler views as being worrisome. I'm also a bit pessimistic about Japan, but it's worth noting that it's really hard to make an objective argument that falling population is a problem, in and of itself. Consider a few possible scenarios:

1. Suppose Japan's population kept falling until it reached about half its current level of 125 million. It would still have almost as many people as Britain and France do today. Would that sort of population reduction significantly impact its ability to influence world events? A little bit, but It's hard for me to imagine that Japan's ability in the long run to hold onto the Senkaku/Diaoyu islands, for instance, will hinge on that sort of change in population. China's already 11 times more populous, and has nukes. Either China will get he islands or it won't--I doubt Japanese population will play much of a role. And are those tiny uninhabited islets actually very important--for reasons other than national pride?

2. Now let's suppose that after Japan's population falls in half, real estate becomes so inexpensive that the Japanese start living in Dallas-style McMansions and having 2.1 kids per family. So the population levels off and the Japanese islands are less crowded. Is that population "wrong"? It's hard to say. Also suppose New Zealand's population grows from 4 million to 40 million, at which point they call a halt to immigration and level off. New Zealand would still have fewer people than Japan. What should we focus on, levels or changes? Which country would have the "better" population policy? It's a historical accident that these two highly fertile Pacific islands have such vastly different populations. Why should we regard either country's current population as being even close to "optimal?"

3. Is aging really a problem? Aging is generally associated with better health. Suppose we made the retirement age for public pensions equal to life expectancy minus 25% of the gap between life expectancy and age 20. In other words, if life expectancy was 80, people would retire at 80 - 0.25(80-20), which equals 65. If life expectancy rose to 100, the retirement age would rise to 80. An aging population by itself does not create any special challenges for fiscal policy, unless we allow it to. I.e. unless we arbitrarily keep reducing the share of adult years that people are required to work before getting a public pension. On the other hand aging combined with a low birth rate, as in Japan, does put a temporary burden on the public sector, until Japan's population levels off. But it's a transitional problem, not a long run problem.

To summarize, I remain an extreme agnostic on all population questions. I have no idea what the optimal population is for planet Earth. If there is a "true" answer to that question, it might well be 20 billion, or 2 billion, or zero. And how much weight should we put on animal welfare? Given all that uncertainty, I'll keep working to improve living standards for the people who are actually here, by advocating non-destructive monetary policies such as NGDPLT. I'll let much smarter people like Bryan and Tyler wrestle with the big questions.

PS. You might think my real estate price argument is implausible, as Japan would still be much more densely populated than places like Australia. But Australia has strict zoning laws, and hence I'd guess that in 50 years houses in Sydney will cost much more than in Osaka.


Comments and Sharing






COMMENTS (19 to date)
Henry writes:

Not super important, but I was wondering about your comment that Australia has 1/10 the US population density. While it's be no means especially dense by any measure, the vast percentage of its population lives along the coast, while the interior is virtually empty.

The typical Australian might observe density at a similar or greater degree than a typical American. Certainly this would be the case for a few states (Vermont, maybe) with greater overall density, but few large urban areas.

I wonder, is there any measurement or estimate of population density that attempts to account for how the typical person experiences it?

Rob writes:

There is one strategy that all utilitarian versions have in common: Use the scientific method to enhance humans so that the most intense suffering is reduced to much lower intensity levels and the most intense pleasures are increased beyond current natural limits.

Then maximize those enhanced humans to carrying capacity.

Humanism demands this because it is the most pro-human thing to do. Animal welfare demands this because humans are animals, too, and a large population of these enhanced humans would replace otherwise suffering wild and domesticated animals. Average utilitarianism demands it because it increases average utility considerably beyond what it is now. Total utilitarianism demands it because it maxes out total utility. Negative utilitarianism demands it because it replaces current suffering.

None of these moral philosophies would look only at ordinary economics of unenhanced humans. They would all look at technologies which vastly reduce suffering intensities and increase pleasure intensities, and then max out population.

How to sell these ideas to an unwilling public, I do not know.

Airman Spry Shark writes:

I would hypothesize that maximizing the product of global utility (or, equivalently, the sum of log[U_i]) would largely match people's intuition about optimality.

There would be higher total utility than when averaging and higher per-capita utility than when summing.

Also, logs are relatively easy to work with, mathematically.

Seth writes:

Just one kibitz - I don't think Cowen's argument "beg the question" in the sense of assuming its own conclusion.

R. Jones writes:

Premature optimization is the root of all evil. The hope for long run utility maximization lies in ensuring that societies are stable and well regulated enough to allow for technological progress that fundamentally changes human nature.

David Friedman writes:

For my attempt to construct a criterion to use to compare futures with different populations, see:

"What Does Optimum Population Mean?" Research in Population Economics, Vol. III (1981), Eds. Simon and Lindert.

JKB writes:

I found this passage from Alan Macfarlane's "The Invention of the Modern World" interesting in thinking of population growth and the "worrisome" decline in fertility in the developed countries:

IN MOST CIVILIZATIONS in history, the major check on population is mortality. Here there are two main variants. Either perennial diseases and high infant mortality keeps population more or less in balance – or, in another variant, the population is moderately healthy, numbers build up quite rapidly, and then periodic crises caused by war, famine and epidemic disease occur. In these two cases, humans have experienced the threat of high mortality in one form or another and tend to be anxious to have many children in order to combat the dangers. This is the pattern observed through much of Chinese, Indian or European history.21

The ‘modern’ pattern is one where it is lowered fertility which keeps population in check, rather than high mortality. Different mechanisms are used, late marriage and high rates of non-marriage, various forms of controls on the numbers born alive through infanticide and abortion, and nowadays high levels of contraception. These are what Wrigley calls ‘low pressure’ regimes. Until quite recently it was widely believed that this ‘low pressure’ regime is the product of some ‘demographic revolution’, perhaps caused by improvements in contraceptive technology in the nineteenth century.

It is now quite clear from the work of Tony Wrigley and other that in England a combination of late age at first marriage (often over twenty-five for women), plus selective marriage (with up to a quarter of women never marrying) was enough to keep population more or less static for some centuries.22

What is interesting is it seems that a decline in population in richer countries as well as aging populations would be the natural consequence the decrease influence of mortality and the increase in the ability to control fertility.

I've come to think that the 20th century, mostly post WWII, we've been in a transition as more and more of the world's population have escaped the war, disease, famine, childhood threats, etc., thus population exploded. But after the 1960s, fertility control has come to dominate population growth leading to declining birth rates, aging populations, etc. Where we still see the high birth rates and growing population is in the still non-modern world but where modern medicine, sanitary practices, etc. has penetrated via Western aid.

Back to Macfarlane

As Malthus argued, the only force strong enough to stand against the biological desire to mate and have children, was the even stronger social desire to live comfortably and avoid poverty.

Brett writes:

I don't think there really is an optimal human population level, since what counts as "acceptable" living conditions is rather subjective. There is an upper bound to human population, in the sense that human beings have minimum needs for food, water, clothing, and air per person in order to survive. But aside from that, it just depends on the society - 30 billion humans might be able to live "comfortably" in dense urban areas where most of them eat creatively flavored vegetarian cuisine and spend a ton of time with cheap electronic entertainment, for example.

As for Japan, I think the decline in population levels is being offset by greater automation and the shift of the population into the Tokyo Metro Area. Japan could end up as a handful of city areas with countrysides that are almost completely empty of human beings, especially if farms end up being mostly automated.

ChrisA writes:

Another illustration of the difficulty in making moral calculations. Which is basically what you would expect if our moral calculating machinery is basically an evolutionary kludge to allow us to cooperate in small groups of closely related people.

Of course you can override your own internal moral system but here is the rule; application of any consistent moral system (like Utilitarianism) to any bigger questions other than small in-group ones comes up with results that either are obnoxious or provide no guide whatsoever.

Brian writes:

"The much tougher problem is whether to focus on average utility, total utility, or some third category (which seems implied in Tyler's essay.) "

Scott,

To the degree that GDP/capita serves as a reasonable measure of utility, we already know the answer: there's no difference between average versus total utility. In a well-connected free-market economy, GDP/capita increases linearly with population size, so very large global populations give the best outcomes. More is simply better. In this case, actual maximization of utility is not well defined.

It's also the case that in a well-connected global economy, optimum population size for a given country is not well defined, any more than it is for a state or town. Porous economic borders make such distinctions artificial.

As for the distribution of fertility around the world, fertility is likely to drop for those who have other means of being highly productive. Folks who have a comparative advantage in producing wealth through innovation are more likely to avoid the opportunity costs of childrearing and let that task be done by those who have little to offer other than procreative production. That's what leads to higher fertility rates for the poor in a given country and for poor countries as a whole. This is also why education tends to reduce fertility--educated people are more likely to have other means of productivity.

M.C. writes:
But Australia has strict zoning laws, and hence I'd guess that in 50 years houses in Sydney will cost much more than in Osaka.
No need to wait 50 years, Sydney already costs much more than Osaka today: http://www.numbeo.com/property-investment/compare_cities.jsp?country1=Japan&city1=Osaka&country2=Australia&city2=Sydney&displayCurrency=USD

[shortened url changed to full url--Econlib Ed.]

Nathan W writes:

You could look at the thesis of B Housseini called "Essays on Demographic Changes, Health and Economic Development"", which includes a few essays on modelling structures for assessing/comparing well-being in ways which may provide some useful structures for thinking about these questions.

I part company with libertarians over environmental issues (including population). Laissez faire policies here fail, as Garrett Hardin observed in "The Tragedy of the Commons" (Science, 1968).

1. The Earth's human population cannot grow without limit.
2. The Earth's maximum possible instantaneous human population exceeds its maximum possible sustainable human population.
3. The Earth's maximum possible sustainable human population leaves little room for wilderness or large non-human terrestrial animals.
4. Value is determined by supply and demand; a world in which human life is precious is a world in which human life is scarce.
5. The Earth's human population will stop growing when (a) the birth rate falls to meet the death rate or (b) the death rate rises to meet the birth rate.
6. The Earth's human population will stop growing as a result of (a) deliberate human agency or(b) other.
7. Deliberate human agency is (a) democratically controlled or (b) other.
8. All human behavioral traits are heritable.
9. Voluntary programs for population control selectively breed non-compliant individuals.
10. Human misery is like heat: in the absence of barriers it will flow until it is evenly distributed.

Prayer is not a viable policy. A Chinese-style population policy is a best-case scenario.

Where do disagree?

Scott Sumner writes:

Henry, Good point, and I don't know. But it's worth noting that the coast of Australia is far less crowded than the coast of the US, which makes visiting the coast much more enjoyable.

Rob, Interesting, but I'm not sure I completely follow where you are going with that.

Airman, But what good does that do if you can't measure utility?

Seth, Yes, I need to drop that bad habit.

David, I should probably read that before claiming expertise.

JKB, Interesting.

Brett, That sounds plausible.

ChrisA, Good point, but I wouldn't go that far. Surely there are SOME macro problems for which utilitarianism works, say monetary policy?

Brian, As suggested in my Australia-Britain comparison, I don't agree.

M.C. Interesting, then I'd guess the gap will be even wider in 50 years.

Thanks Nathan.

Malcolm, On point 9, I'm not convinced that the selection effect is strong enough to overcome the decline that occurs as incomes rise. Time will tell.

mico writes:

"3. Is aging really a problem? Aging is generally associated with better health. Suppose we made the retirement age for public pensions equal to life expectancy minus 25% of the gap between life expectancy and age 20. In other words, if life expectancy was 80, people would retire at 80 - 0.25(80-20), which equals 65. If life expectancy rose to 100, the retirement age would rise to 80. An aging population by itself does not create any special challenges for fiscal policy, unless we allow it to. I.e. unless we arbitrarily keep reducing the share of adult years that people are required to work before getting a public pension. On the other hand aging combined with a low birth rate, as in Japan, does put a temporary burden on the public sector, until Japan's population levels off. But it's a transitional problem, not a long run problem."

What if we are able to extend total lifespan faster than we can extend healthy lifespan? In that case longer life expectancy means decreasing productivity in old age and a low quality retirement, albeit with more years of total life. I don't know if that is a good trade.

Brian Donohue writes:

Good post Scott.

But zero population optimal? Even Thoreau wasn't such a misanthrope.

David writes:

Brent and Malcolm:

"There is an upper bound to human population, in the sense that human beings have minimum needs for food, water, clothing, and air per person in order to survive."

"The Earth's human population cannot grow without limit."

I disagree, in the extreme:

First, expand "Earth" to include "All reachable matter". Right now that is pretty much Earth only, but there is reason to believe that the limits will change dramatically in a very short time.

OK, so now you have a solar system to work with, which has a much higher, yet still finite, carrying capacity (limited by phosphorus requirements, as I remember). So you start going outward, exploring the galaxy. But (as we understand physics today) you can't expand faster than the speed of light.

So you have the total volume of the universe available to mankind growing (at most) at the cube of time, while population grows exponentially. Crazy Eddie, sounds bad for infinite growth, right?

But that is unlikely to happen - we don't see a bunch of aliens filling the universe, after all. The most likely scenario (IMHO) is that physical humans are replaced with virtual humans.

You start with a real human, that spends all of his time in a virtual world. (In a modern teenager, for example, you can see the beginnings of this ;-} ) After a hundred years or so, his body starts wearing out - most importantly, his brain does. So various parts are slowly replaced, until he is entirely a program running on hardware, living in a world running on other hardware. At some point, you skip the middleman and just include simulating the person's hardware as part of the universal simulation.

OK, so thus far all we have done is increase the human universe's carrying capacity by an extremely large, yet finite, amount. Good, but not infinite.

But these are creative humans we are perfectly simulating. So they spend billions of years optimizing the simulation. Specifically, compression of data and compression of processing time. As you add more and more people to the simulation, the possible differences between each individual's experiences and personality asymptotically approaches zero. As you add more years to the simulation, the possible differences between this year and all previous years also asymptotically approach zero. Given that, compression can achieve exponential effectiveness, counterbalancing exponential population growth. In addition, the energy required to run the simulation can be decreased with time as well.

So not only can we eliminate any physical limits to humans in this way, we can even outrun the heat death of the universe itself!

(Malcolm): "Prayer is not a viable policy."
(David): "First, expand 'Earth' to include 'All reachable matter'. Right now that is pretty much Earth only, but there is reason to believe that the limits will change dramatically in a very short time. OK, so now you have a solar system to work with ... "
Neither is science fiction a viable policy.

Daniel writes:

We absolutely can not grow forever, we live on a finite planet and this means that there are limits: link

Comments for this entry have been closed
Return to top