I'd like to get a few opinions on record or simply recap some previous points.
The current downturn is a mix of AD and real shocks, in uncertain proportions, and in a manner which is hard to separate empirically. It is now obvious there is a lot of structural unemployment and there is a quick and probably unjustified rush to define it all as AD-influenced unemployment turned sour. The structural theories have their problems, but they can better explain why corporate profits are high and can better explain the distribution of unemployment across income and educational classes. The regional distribution of unemployment is persisting because of labor immobility, which involves both AD and structural issues. The sectoral shift view is more about shifting out of optimism-linked activities, within any particular sector, rather than about shifting out of construction and finance per se.
The mix of structural and AD factors does not, in any case, support liquidationist policies. It supports AD-stabilizing policies, though it suggests that in absolute terms those policies will do less well than expected. The failure of the fiscal stimulus is consistent with a number of different views, not just the claim that it should have been bigger. We've yet to see a good theory of how stimulus scales up to produce a bigger and better multiplier at higher levels.
I don't trust stimulus analyses which fail to assign a central role to confidence and confidence is hard to model.
Current experience is also consistent with (but not unambiguously favoring) an Austrian-like view that the stimulus boosts some activity and then shortly thereafter pulls away the rug, leaving us more or less back where we started, albeit with some smoothing gains in the short run and some adjustment costs in the longer run. The persistence and scale-up from the initial fiscal boost is hardly guaranteed. The empirical papers on multipliers are not to be trusted and the results are in any case hard to generalize from one period to another.
Harald Uhlig's paper is one statement of the case against stimulus. There is nothing measured by the Alan Blinder study which rules out the central result of this paper, namely transitory gains in the short run and high costs in the longer run.
Macroeconomics is rarely simple.
Elizabeth Warren is unlikely to prove an effective agency head, and the two sides to this debate ought to switch positions. Yet…politics very often isn't about policy.
On the AD front, Scott Sumner has been vindicated more than any other writer. His best critic is Arnold Kling, especially with regard to whether there are only two kinds of inflation regimes, low or high and variable. A related question is what a looser monetary policy would have done to financing the long-run debt burden and the use of the interest rate spread to recapitalize banks.
We still don't know what we are doing.