No points for guessing!
This essay (hat tip: Jon Rabinowitz) is a great read. Yes, the econometrics are pretty basic. Sure, the data is a little less-than-robust. (The guy collected rent data for S.F. back to 1949, but did not adjust for quality.) But the basic conclusion seems pretty good: letting development rip will at best hold back the tide.
I would add two points, however.
First, I worry about the model, simply because San Francisco is part of a metropolitan housing market. It really needs to include suburban development. Of course, the Bay Area has slammed suburban building shut even further than the County of San Francisco, but it is easier to imagine doubling the number of housing units in San Mateo and Alameda than it is to imagine doubling them in S.F. proper. Surely that would have a major effect.
Second, S.F. may be setting itself up for a major recession some day. I suspect not, to be honest: a gilded metro area looks depressingly plausible, particularly for the center of the computer industry. But a combination of a tech boom bust combined with an exhaustion of the supply of skilled people willing to live in the world’s first gilded metro area will kill the place dead.
At that point, New York will get another filip and the people who didn’t buy in Upper Manhattan and the Bronx when it was still relatively cheap will regret it. You know who you are. It isn’t too late.