Menzie Chinn discusses the evidence on sectoral shifts hypotheses. See also the piece by Valletta and Cleary.
These are intriguing and useful studies but I don't think they get at the core of the matter, mostly because sectoral shifts and aggregate demand shocks are so closely intertwined in this recession.
Here's a very simple story. The prices of homes and stocks fall, plus there is some panic, so people spend less. On the surface, that's an AD story, following from an economy-wide negative wealth effect. But it's also a sectoral shifts story, because people are not cutting spending proportionately on all items. For instance luxury consumption and debt-financed consumption have been hit especially hard, not to mention real estate and financial services (for other reasons). And since I do not expect a quick rebound of real estate or stock prices, this is more or less a permanent change in sectoral priorities. Still, in the data the AD shock might well absorb most of the "credit" for what happened.
We're also seeing job losses in virtually every sector. It's not for instance a "sectoral shift away from services and into matchstick production and tungsten." It's a shift out of jobs which are revealed as unprofitable and a lot of people not knowing where the new jobs will be created.
If someone wants to insist that "this is really an AD shock, not a sectoral shift," I'm not so keen on fighting to keep one term over the other. I would insist, however, on an issue of substance, namely that not all AD shocks are alike. If we are going to switch terminology, it could be said that this is a real AD shock and not just a nominal AD shock. (Though there have been nominal AD shocks too.) A nominal AD shock can be offset more easily by goosing up some mix of M and V and restoring the previous level of nominal demand. If you want an example of a nominal AD shock, imagine a more neutral change in monetary variables and indeed those have happened in the postwar era. Or read David Hume's parable of the money under the pillow. In those cases you don't need to make people feel wealthier in real terms, you just need to get the flow of spending up again. Today, part of the problem is that people feel less wealthy in real terms and that influences the content of their spending and investment decisions.
When a real AD shock comes, policy still should be expansionary in response, but there is an important difference. In absolute terms, nominal expansion won't much help the labor market, which still has to reallocate workers from some sectors to others, given the collapse in asset prices and expectations.
You'll see indirect recognition of this from many current Keynesian writers, when they talk of the jobless recovery or fear that the economy will fall back next year after the stimulus money runs out. In general I agree with those points. Yet these writers are less willing to consider the implied conclusion that a bigger stimulus won't much help — and may hurt — the longer-run adjustments which are required. Boosting MV will restore employment only to a very limited extent. It's still the case that recovery will require a great deal of sectoral readjustment and that will take a good bit of time.