What if I didn't favor NGDP targeting?
By Scott Sumner
For the past nine years I’ve been promoting market monetarist ideas in the blogosphere. How important is NGDP targeting to the MM agenda? Much less important than many people assume.
Kurt Schuler left the following comment in response to my previous post:
Nominal GDP targeting has not yet been implemented anywhere. Accordingly, you have the luxury of comparing an untested policy whose defects (if any) have not yet been revealed in practice with well-tested policies whose defects are a matter of record. Advocates of inflation targeting were in the same position when it was first widely discussed. Then it was implemented, and after some years of apparent success came the Great Recession. If you are plan to advocate nominal GDP targeting in your book, you should specify what results (if any) would lead you to revise your favorable opinion of it.
Let’s suppose I switched my view away from NGDP targeting, and moved toward the Fed’s current “dual mandate” approach, which aims at 2% inflation and high employment. What then? How much would change?
The first thing I’d do is create a single variable that incorporates both of the policy goals in the Fed’s dual mandate. After all, the Fed can only hit one target at a time. That variable could be set up in a wide variety of ways, but here’s one very simple example:
AD = PCE inflation plus employment gap.
Where the employment gap is defined as the percentage difference between actual employment and the Fed’s best estimate of the natural rate of employment.
Thus if inflation were 2.7% and employment were 1% above the natural rate, then the AD variable would come it at plus 3.7%.
Next I would have the Fed try to target AD at 2%, that is, I’d have them set policy at a level where expected future inflation plus the employment gap equaled 2%.
Here I’d like to emphasize that there are many other ways of doing this. For instance, you could put a coefficient of 0.5 on the employment gap, not 1.0 as in the example above. Indeed there is a whole class of dual mandate targets, which share certain common characteristics. I don’t currently have strong views as to which one is best.
So let’s suppose the Fed sets up the formula, and then I blindly adopt it. What then? How much does that change my blogging over the past 9 years?
Hardly at all; these formulas are different from NGDP growth, but not radically different. In either case, money was far too tight during late 2008, and in subsequent years. In either case the Fed was failing to target the forecast. We didn’t have a Great Recession because the Fed was targeting inflation and employment instead of NGDP; we had a Great Recession because the Fed was setting policy far too tight to hit its own inflation and employment composite goal.
If you compare NGDP targeting to the ECB’s single inflation mandate, then the differences are a bit larger. But even in that case, ECB policy has often been too tight to hit their 1.9% inflation target. (But not in 2011, when inflation targeting really was a big problem.)
Don’t get me wrong, I definitely believe that NGDP targeting is superior to the Fed’s flexible inflation targeting. But that’s not the core problem here, the core problems are:
1. Failure to target the forecast
2. Failure to rely on market forecasts
3. Failure to do level targeting (at least at the zero bound, as recently recommended by Bernanke.)
What would make me change my mind about NGDP targeting? I suppose if it were adopted and employment became more unstable (than under recent policy) then this would tend to refute the notion that NGDP targeting is superior to the Fed’s current policy. How much data would we need? That’s a judgment call, which would actually involve two variables—the number of years operating under the new system, and the extent to which employment became more unstable. The greater the increase in employment instability, the more quickly NGDP targeting would be discredited. I can’t give you an exact number, like most things in economics it’s a matter of degree. (Or you might use Bayesian terminology and talk about changing probabilities of NGDP targeting being superior.)
As far as the use of futures targeting, a refutation would occur if NGDP futures targeting resulted in actual NGDP become more unstable.
PS. Eliezer Yudkowsky left a very interesting comment after my previous post. (A rare comment that I had to read multiple times to fully absorb.)