[mathattack]

Very interesting. A couple comments:

- The "Not in my backyard" is economically rational, but still wrong. I'm not sure how to fight it.

- My general view is to remove the roadblocks to let economics help out. Large public works doesn't seem to be the right solution either. (High crime, bad schools, etc)

- The better case is not "luxury vs low-income" but rather to put the lower (or even medium) income apartments somewhere that there's less competition from luxury. There's no natural law that says they need to be side by side. (But I would argue we should ensure the quality of schools for both is similar)

[aetherson]

One could argue -- I'm not sure that I care to -- that the natural evolution of things would be for all new housing stock to be built well above the median of the market, and that as structures are a durable, but not appreciating asset, the deal would be that the rich live in new stock, the middle class live in older stock, and the poor live in yet older stock. As new stock comes into the market, it depresses the value of the older stock, and moves it down the income line.

In which case there would be no concept of "affordable housing" as a kind of development -- affordable housing would just be older housing.

Since structures are a significantly durable asset, this would operate over the course of decades, or even a century, not years.

In which case we could imagine that changes in the supply stock over 35 years (ie, rent-controlled stock in SF) would have a magnified effect on the well-being of the poor.

That said, I suspect this model is too simple to have that much predictive power. You are right to focus on schools (though note that in SF, public schools are built on a lottery system, and being local to the school gives you only limited access to it).

Many people are very hostile to the option to say, "Well, fuck it, SF is expensive -- people who want cheaper housing should live in more geographically remote places." It's hard for me to figure out where any concept of a right to live near place X begins or ends.

[mathattack]

Interesting idea. The main issue is that location matters. Most of the value of high value real estate is in the location. (This is why a 4BR house in Rochester NY costs 1/20th what it would cost in Manhattan or SF)

People would rather live in an old house with good schools and close to work (say Palo Alto or Los Altos) than new houses a significant drive away.

You can't tell someone "Take a risk on the schools, get a bigger commute, and pay up" but you can say "Get a nice new apartment, roll the dice on the schools, but save a lot of money and buy a nice car for the commute"