Paul Krugman returns to a familiar theme:
Economics: What Went Right: ...I’m at EconEd; here are my slides for later today. The theme of my talk is something I’ve emphasized a lot over the past few years: basic macroeconomics has actually worked remarkably well in the post-crisis world, with those of us who took our Hicks seriously calling the big stuff — the effects of monetary and fiscal policy — right, and those who went with their gut getting it all wrong. ...
One thing I do try is to concede that one piece of the conventional story hasn’t worked that well, namely the Phillips curve, where the “clockwise spirals” of previous protracted large output gaps haven’t materialized. Maybe it’s about what happens at very low inflation rates.
What’s notable about the Fed’s urge to raise rates, however, is that Fed officials, including Janet Yellen, are acting as if they have high confidence in their models of inflation dynamics –which is the one thing we really haven’t done well at recently. I really fear that we’re looking at incestuous amplification here.
Agree about the uncertainty about inflation dynamics, but fear Fed officials will interpret it as risks on the upside that must be nullified through interest rate hikes. As for the Phillips curve, here's a graph from his talk:
As Krugman says, "Maybe it’s about what happens at very low inflation rates." I would add that the combination of the zero bound, low inflation, and downward wage rigidity may be able to explain the change in the Phillips curve -- I'm not quite ready to give up yet.
More generally, estimating inflation dynamics has been far from successful. For example, in many VAR models (a widely used empirical specification for establishing relationships among macroeconomic series), a shock to the federal funds rate often causes prices to go up (theory says they should go down). This can be overcome somewhat by including commodity prices in the model. The idea is that when the Fed expects inflation to go up it raises the federal funds rate, and since the policy does not complete eliminate the inflation, the data will show a positive correlation between the federal funds rate and inflation. Commodity prices are thought to embody and be sensitive to future expected inflation, so including this variable helps to solve the "price puzzle" as it is known. Even so, the results are highly sensitive to specification, and when you work with these models regularly you come away believing that the estimated price dynamics are not very good at all.
But the Fed must forecast in order to do policy. There are lags (though I've argued they are likely shorter than common wisdom suggests), and the Fed must act before a clear picture emerges. The question is how the Fed should react to such uncertainty about its inflation forecasts, and to me -- given the corresponding uncertainties about the state of the labor market and the asymmetric nature of the costs of mistakes about inflation and unemployment (plus the distributional issues -- who gets hurt by each mistake?), it counsels patience rather than urgency on the inflation front.