John Quiggin provides an excellent discussion of macroeconomics. It is much too good to summarize. Just click the link and read.
Paul Krugman is also (as usual) brilliant. In particular, I fear he understands the sociology of the profession.
I have some comments on Quiggin.
This is, as usual, brilliant. I can think of a few things to add.
1) the claim that medium and long run outcomes are determined by tastes and technology does not imply that there is a unique long run equilibrium growth path. This would follow if technology were exogenous, but, of course, it isn’t. It is standard in business cycle theory to assume that technological progress is exogenous, but really believing that it is exogenous is much crazier than believing in rational expectations and such. In growth theory it is conventional to attempt to model technological progress. This is enough for equilibrium to be indeterminate and for demand side policies to have permanent effects.
2) There was a rather large literature on coordination failures which cause fluctuations (you know Benhabib and Farmer and such like). There was nothing wrong with this literature as math of fun theory. It seems to have vanished. It just wasn’t true in 2007 that if you want to insist on rational expectations then the available choices are new classical vs sticky price models in which long run averages were as in new classical models.
3) Actual general equilibrium theory did not stagnate from 1950 on. Actual general equilibrium theorists studied models with incomplete markets in which equilibria can be indeterminate, sunspots can affect outcomes and equilibria are generically not constrained Pareto efficient.
4) Persistent fluctuations due to aggregate demand were renamed “hysteresis” by Blanchard and Summers in 1986. European data already massively rejected the not yet developed old new Keynesian models. This paper was considered to be relevant to a relatively minor field (the study of strange countries which aren’t the USA) and ignored in mainstream macroeconomics eg by Blanchard and Quah in 1987. The study of the strange unusual case of developed countries other than the USA didn’t even remain central to the modelling of European macroeconomies by European central banks.
All four points imply that the very widespread conviction among macroeconomists that long run outcomes are unique and determined by exogenous variables had no basis in theory. All over the place highly mathematical theory showed how that needn’t be true (typically theory based on extreme over use of the assumption of rational expectations do a degree which would make Lucas blush).
The assumption of a unique exogenous long run growth path absolutely does not follow from the D, S, G or E parts of DSGE. It is a separate assumption –a methodological a priori not an implication of other standard assumptions. I think you have explained why. If equilibrium is indeterminate, then economists can’t design optimal plans and economists are reluctant to admit this.
In contrast, it is possible (by extreme abuse of the assumption that the world is in Nash equilibrium) to write down models in which economic agents magically know which equilibrium they are in and in which there are Nash equilibrium with possible persistent depressions. The assumption of rational expectations makes no sense at all when there are multiple Nash equilibria. However, this does not reliably embarrass game theorists. The calculations required by agents in DSGE models with a unique equilibrium are obviously completely alien to actual people. I don’t see an intimate connection between multiple long run growth paths and people having to rely on rules of thumb. Real people rely on rules of thumb even in situations with a unique equilibrium (say a zero sum game with a 5 by 5 game matrix). Imaginary economic agents can know which of a continuum of Nash equilibrium they are in (solving the models feels about the same).