1) When a lot of money comes into a city quickly,
I'd expect rents to rise, until new construction
can bring them down. And since much of the new
construction might be far from the center of the
city, rents near there might go up and stay at
an elevated level.
2) I would expect this effect to be more severe in,
(taking an example *totally* at random) a city
squeezed between a bay and some mountains, because
geography would constrain building and transportation.
3) A quantitative analysis would be necessary to figure
out how much of the housing price increase was due
to the swift inflow of dot-com cash, the geographic
constraints of San Fransisco, and rent control. Prof.
Krugman did not (IIRC) mention this even in passing
(a single sentence mentioning this would have sufficed
to those who know, and not been an obstacle to those who
don't know such things).
4) I live in a non-rent-controlled city. My landlord wanted
references. This is to give a good chance that I would
not trash the place, and that I wouldn't leave in the middle
of the lease period without paying (I live in a university
town - it's a landlord's market in general, but finding
a new tenant in the middle of the year can be hard). I can see
similar reasons for doing so in SF, with the added reason
that the dot-com employment situation is more volatile. A landlord
might want to make sure that the tenant will be able to pay
the steep rents for the reasonable future.
This leads me to the conclusion that Prof. Krugman was leaping from real-world event to the economic principle that he wanted to illustrate, bounding over the actual causality which might or might not be involved.
Barry