Must-Read: Olivier Blanchard says that he and Paul Krugman differ not at all on the analytics but, rather, substantially on "tone". When I read Olivier, I find his tone so measured and reasonable that casual and even more-attentive-than-casual readers are likely to completely miss the point.
When Olivier believes the Federal Reserve did right to raise interest rates in December. But when he says "each of the last three conclusions presents challenges for the conduct of monetary policy", what he means what I conclude is: The Federal Reserve has made and is making three mistakes in its assessment of the relationship between inflation and unemployment:
It believes that the relationship is tight, so that you can make policy by simply looking at the forecast without looking at asymmetric consequences in the tails of the distribution of future outcomes. But the relationship is not tight, but loose. It has always been loose.
It believes that the gearing between unemployment and inflation is strong, so that minor falls of unemployment below the natural rate produce substantial increases in inflation even in the short run. But that gearing has not been strong since the early 1980s. It is weak.
It believes that increases in inflation substantially and rapidly affect expectations of future inflation, so that we are never far from a wage-price spiral. But that gearing has not been strong since the late 1980s, if then. Inflation expectations are anchored.
Why the Federal Reserve is working today as if the Phillips-Curve relationship is still what it was in the years around 1980 is a great mystery. But it is, I think
--and I think Olivier thinks, though with his reasonable tone it is hard to tell--leading the Federal Reserve to place some bad monetary-policy bets right now:
The US Phillips Curve: Back to the 60s?: "The US Phillips curve is alive...:
...(I wish I could say “alive and well,” but it would be an overstatement: the relation has never been very tight.) Inflation expectations, however, have become steadily more anchored, leading to a relation between the unemployment rate and the level... rather than the change in in inflation... [that] resembles more the Phillips curve of the 1960s than the accelerationist Phillips curve of the later period. The slope of the Phillips curve... has substantially declined.... The standard error of the residual... is large.... Each of the last three conclusions presents challenges for the conduct of monetary policy...