Brad DeLong has a (way too) long post. But the end gives the gist of his argument:
The problem is that the macroeconomics that Paul Krugman learned at Jim Tobin’s knee wasn’t just 1930s-style Hicks-Hansen Keynesianism. It was the 1970s adaptive-expectations Phillips Curve neoclassical synthesis–nearly the same stuff that I first learned at Marty Feldstein and Olivier Blanchard’s knees in the spring of 1980.
That is the framework that Marty is using now, and that generates his puzzlement. That framework had a short run of 1-2 years, a medium-run transition-dynamics phase of 2-5 years, and a long run of 5 years or more baked into it. You cannot–or at least I cannot–just throw away the medium run transition dynamics* and the declaration that the long run Omega Point is five years out, and say that mainstream economics does well. You need to explain why the back-propagation induction-unraveling worked at its proper time scale in the 1970s and the 1980s, but is nowhere to be found now.
And so I am much less confident that I have solid theoretical ground under my feet than Paul Krugman does.
The “solution” (or “explanation” for the absence of the “back-propagation induction-unraveling”), towards which even Brad´s pal Larry Summers is warming up to is….NGDP-Level Targeting. In fact, the Fed has (implicitly) established a much lower level target for NGDP, a level that is consistent with Summers´ “Great Stagnation” thesis.
So I hope that when Summers says “I didn’t quite endorse NGdp targeting. I said that I would prefer a shift to NGdp targeting to a shift up in inflation targets”, I also hope he has a level target at the back of his mind!
HT Scott Sumner