My comment on this post is so intemperate that I will put it only here -- I use this blog as a trash can for excessive rudeness.
I must admit that I didn't finish reading this post.
Soon after reading
"Interestingly, the biggest challenge to large multipliers does not seem to come from the empirical front, but rather from the theoretical front." I realized that no useful conversation could include both of us. At that point I should have left without typing. But I won't.
You argue that empirical estimates are refuted by theoretical considerations. The possibility that a DSGE model might not correspond to the true true economy, you know the world out there, was not considered when I gave up reading.
Do you agree that you claim that empirical research is challenged by a purely theoretical argument ? My sense is that to make such a claim someone has to be incapable of even grasping the concept of science.
On the empirics, yes it is true that consumption is lower than say Krugman would guess when there is rationing (I mean explicit rationing with ration coupons). Also private investment is lower when it is illegal without explicit permission which was often not granted. The ambiguity in the empirical literature is explained to anyone who looks at history with other than mind boggling ignorance.
But i have slipped and discussed evidence not DSGE models. I ask is there any reason to pay any attention at all to what would happen if a DSGE model were a good approximation to reality. I believe that DSGE modelling is a complete total 100% failure -- that the models require one arbitrary parameter for every summary statistic they fit. I think they have absolutely no merit whatsoever. For about a year I have been challenging New Keynesian economists to present any counter argument, any evidence that the models work better than one would expect a totally false worthless theory to work.
It might not be surprising if people ignored me (but they didn't) but Paul Krugman who you note, gets attention. He has made the same challenge and agrees that NK DSGE model defenders have failed to meet it.
(I don't agree entirely with Krugman -- the stagflation effect was predicted by Samuelson and Solow in 1960 -- the claim that those old Keynesians got that wrong is based entirely on taking the legend of a figure out of context http://bit.ly/TDqz9R so I count 0 not 1).
Why should I or anyone give a damn about the properties of DSGE models. I stress I am not just saying that I think they are not useful to policy makers. I think that any pure social scientist who seeks the truth should recognize that they have no merits as contribution to social science.