Change my mind: Density increases local but decreases global prices

But price is not a function of density!

You think building more apartments in Cumberland, MD will increase prices, or decrease prices? I should point out: Cumberland is already a very densely built city. Its inner urban geography is not unlike Oakland. It just happens to be in a long, painful period of decline. So many of those buildings are empty.

The answer is pretty obvious, here. Prices would likely drop. The "induced demand" is caused by economic success, not density itself. Plenty of dense cities have declined. (Cumberland isn't even that cold! And it has legal marijuana! And it's close to ski resorts! Eat your hearts out, armchair prognosticators!)

>we find that indeed, the denser an area, the higher its house prices

The data in this graph skew California cities upward because their suburbs are not part of their core urban areas — Concord/Walnut Creek is considered separate from the Bay Area, as is Dublin/Pleasanton — while cities outside of California are more likely to be connected. This is due to mountains intersecting dense urban geography in California, which doesn't happen in most other major cities (but Miami is constrained by swamps and New York does hit mountains to its north). But there are other unusual forces on housing prices in California (Proposition 13, construction difficulties), and without the major California cities the correlation becomes much weaker, before we even consider reverse causation.

>No; I think the missing insight is that there’s some pool of geographically mobile Americans1 who are looking for new housing (or who might start looking if the right situation presented itself). These people have various combinations of preferences and requirements. One common pattern is to prefer any big city - they would be happy to live in Seattle, or NYC, or the Bay, if the opportunity came up.

High decadal growth rates in Nashville, Charleston SC, Fayetteville AK, Boise, Jacksonville and Bend among other places put the lie to this. There are certainly "trends", but they aren't limited to big cities, and they don't really affect all big cities: the Los Angeles metro (of all places!) grew a measly 3% from 2010-20. Surely LA's high prices aren't primarily driven by a growth rate just a little higher than that of Pittsburgh.

There might be some induced demand from certain kinds of housing construction in certain contexts, but this blog post is a long way from demonstrating any such effects. The most well-justified negative effect I've heard of from new housing construction is that businesses serving poorer populations are closed due to increasing commercial rents and competition from businesses serving the new clientele. And this can increase the cost of living for local residents, as well as causing job losses (even if employment statistics improve). What we do know is that sometimes people are worse off after construction happens (hence the uproar in the first place), and unfortunately I don't think this kind of 10000-feet analysis brings us closer to understanding why.

Wanted to mention though that your sentence "the total cost of living in Tokyo blows NYC and the Bay out of the water" risks being interpreted with two very different meanings.

The idea for roads is that congestion moderates the utility of roads, so over a wide range of capacities roads will remain congested or even become more congested as you add roads, since if you build more you make the roads more useful (reaching more places faster, through better connections or increased capacity) and so people will use them more, potentially resulting in more congestion but almost certainly not less.

The parallel idea for housing would be that cost moderates the utility of the city, people move into the city until the cost matches the benefits of density for them (driving up costs and pushing out people for whom the equation no longer works out). If you add units, you've just allowed more people to move from Idaho into the city, and made the entire city potentially more valuable from increased network effect and caused costs to go up further.

This process shouldn't be expected make it affordable to live in the city, unless you've managed to either damage the utility of living in the city through the secondary effects of increased population/density or you've managed to build enough supply the you can sink the entire worldwide demand for living in one of the best and most valuable place, such that there are no more wealthier people anywhere in the world to sell to and you must go down market to get more customers.

It would however, increase the number of people who can have the benefits of living there, but at the cost of displacing more people who can no longer afford it... the benefits go to people outside of the city: The wealthy ones that can move into the city, and the rest who enjoy lower competition for housing outside of the city.

> Denser housing is very efficient

Housing itself increases in efficiency to a moderate level of density-- a duplex is somewhat more efficient to construct than a single family home, owing mostly to reduced land costs (materials are somewhat higher due to the increased isolation and fire protection required). But the effects fall off with increasing density, and you run into issues that building higher involves structural and safety related costs that hurt their efficiency.

E.g. https://ti.org/pdfs/APB36.pdf Page 2--

  high-density construction is significantly greater,
  per square foot, than low-density construction. Due to the
  increased need for steel, concrete, and similar materials,
  the cost of building three or more stories is progressively
  greater the more stories are built
  [...]  building three stories costs 30 to 50
  percent more, per square foot, than two-stories; four to
  seven stories costs 200 to 300 percent more; and eight or
  more stories costs 450 to 650 percent more" )