I knew this was going to happen: people have heard a trite sound-bite about "counter-intuitive" statistical logic so many times that they ignore that correlation is usually a pretty damned good signal for a causal relationship.

To wit: of course there's a correlation between price and new housing units -- because it is causally related. When prices go up, developers have greater incentives to build. More importantly, in a city with as high a density as San Francisco (and yes, folks, it is dense -- the second-densest city in the US, in fact), with as many architectural challenges (seismic, geographic, etc.) the limiting factor for new construction is land and materials, not red tape.

Nobody wants to hear this, but it's true. The fixed costs of building here are so high that developers won't do it unless rents go up. That's why new construction costs upwards of $4k for a 1-bedroom unit, and why new construction in SF doesn't place any real downward pressure on rents -- except (perhaps) in the very long term. At best, you're treading water. Developers don't build into a falling market.

But really, the best response for the people who keep asserting that "building up" is the magical solution is to point to Manhattan: it's the densest city in America, yet it's just as expensive as San Francisco, if not more so.

There are no magic bullets. San Francisco is expensive because there's a lot of money chasing a tiny little bit of land. You don't need a dissertation to explain it.