=economics =sociology =cities


Incomes are higher in big cities like NYC and San Francisco, so in the future everyone will live in megacities. Or will they? This isn't just an academic question. In China, the government is actively trying to make huge cities.



People come together in cities to reduce transportation costs, which makes trade cheaper, which is important because of specialization. The advantage of a big city is the potential for more specialization, but there are limits to how specialized jobs should be.


Many jobs have limited specialization. Janitors and cashiers do mostly the same thing everywhere. In NYC, the high cost of living makes them more expensive, but they're not more productive than anywhere else. Their high salaries are being carried by something else.


There are advantages to having a group of specialists in a particular area in the same place. They might teach other people about that area. Companies want enough people in that area that they can hire a new employee, and workers want enough companies that they can find a new job.


The implication is, large cities will specialize in certain areas, areas which have many subfields that companies might need a specialist in. For example, NYC does finance, and Silicon Valley does programming. You can find programmers many places, but if you need specialists in a particular combination of tools / languages / problems / etc, maybe then you want to go to Silicon Valley despite the premium that you'll be paying with the high cost of living there.



What are the limits to this?



There's a limit to how narrow a specialization is really meaningful or good. No field is an island, and learning about related areas can be good for a specialist. The best programmers in any particular programming language all know other programming languages too. Companies often require lots of "keywords" in resumes that include things that don't really matter, things that are effectively the same, and things that some people could learn quickly.

Why is it that people drive (a long way) to get to an (expensive) office to work on a computer, when they could just use a computer at home? Most people are incapable of collaborating over the internet effectively, but that's not true of everyone. There are even companies (like DeviantArt) that use all remote workers. Usually software companies; programmers tend to like email/IRC/etc more than other employees, but programmers aren't the only people who can work over the internet. I think specialists in a narrow area are more likely to be able to, because those computer/internet skills tend to be important for learning about specialized things. But companies are generally bad at recognizing when remote work is possible and enabling it.



So, those are possible limits to the advantages of bigger cities, but what are the disadvantages? Why not combine smaller cities into one big city?


Pollution, infectious diseases, and proximity to natural resources are all issues with megacities, but the biggest problem is, I think, traffic.


For there to be a benefit from the cities being adjacent, people need to drive between them. That's a greater distance than traffic within a city goes, and that traffic has externalities like pollution and congestion.

Putting remote work aside, I think there are significant dropoffs in returns to scale around 1 million for most cities, and 10 million for cities specializing in particular areas. If my view is correct, then making only megacities is a flawed plan, and it would be better to make cities of ~1.2m and ~11m people at around their current relative proportion.

At those sizes, I think you want typical buildings to have at least 3 and 11 floors respectively. (That's for current construction, not construction in 1950.) What things are much more expensive with multiple levels? Roads, parking, big box / warehouse stores, and factories. What are the implications? Maybe 10m person cities will have more multilevel retail selling fancy expensive stuff. Industry/manufacturing based towns would probably tend to have ~350k people, but at that size it would make more sense to have 1+ clusters of ~200k people doing industrial stuff next to a 1m person city, unless there's some sort of natural resource nearby. Smaller towns than that would mostly make sense as tourist destinations or areas with retirement homes or places selling stuff to farmers. So this is where I think things are headed.



back to index