But What About Inflation?
In my last post, I observed that the macro models of 2006 would have assigned a probability of essentially zero to either of these two events:
In this post, I turn to the behavior of inflation. Here, the situation is quite different. The Federal Reserve staff’s forecasts of inflation (during a rather challenging time frame) were quite accurate. They certainly don't suggest a need for any kind of fundamental re-think to their baseline modeling approach.
To show why I say so, let me start with the run-up to the crisis – that is, 2006 and 2007. In those pre-crisis years, the Fed’s Greenbook did not include forecasts beyond a two-year horizon.
In December 2006, the Greenbook predicted that PCE core inflation would be 2.3% in 2007 and 2.1% in 2008. Actual PCE core inflation was 2.3% in 2007 and 1.6% in 2008. Both realizations are within the relevant 70% confidence intervals.
In December 2007, the Greenbook predicted that PCE core inflation would be 2% in 2008 and 2% in 2009. Actual PCE core inflation was 1.6% in 2008 and 1.4% from in 2009. Again, both realizations are within the relevant 70% confidence intervals.
Inflation was distinctly lower in 2008 and 2009 than the Fed staff anticipated. But the size of the miss is, from a model-based probabilistic sense, not at all surprising.
Now, let me turn to 2008, 2009, 2010, and 2011. (The Fed has not yet released its staff forecasts for any post-2011 meetings.) In these meetings, the staff provided forecasts for core inflation over the next five years. Stars will indicate that the actual core inflation rate is outside the 70% confidence interval. (Apologies for formatting.)
In all of the entries in these various tables, we see only three stars. (Actually, this suggests to me that the confidence intervals for these forecasts may be systematically too large.) Perhaps more importantly, aside possibly from 2008, the magnitude of the misses don’t strike me as all that large from an economic point of view. The forecasting performance is consistent with the good news conclusion that, even in these challenging conditions, the Fed’s staff retained their ability to predict the twists and turns to US inflation.
In contrast, and as John Cochrane has emphasized in his recent NBER Macro Annual, the past few years have revealed distinct gaps in academic understanding of inflation dynamics. Some academics predicted that the high unemployment rate would lead inflation into negative territory. Others predicted that the Fed’s easy monetary policy would trigger rising inflation. But, all told, most were surprised by how stable inflation remained in the face of huge slack, large expansions of the money supply, and constant nominal interest rates.
Why did Fed staffers do a better job of forecasting inflation than most academics? In academic models, medium-term inflation expectations are relatively sensitive to economic conditions and monetary policy choices. In the Fed's model, that's much less so. That's probably a lesson that academics should take on board as they seek a better understanding of inflation. I've tried to do so myself in this paper.
N. Kocherlakota
Rochester, NY
May 21, 2017