I have recently become aware that I have lost the heavy-weight strawman-attacking championship title to a quite unlikely opponent. That’s right. There is someone far better at creating and attacking strawmen than the queen of all strawman attackers, yours truly. And the latest champion is…
Kling starts off a post entitled, AD-AS Debunked in One Graph, with this:
In the WSJ blog. It shows the behavior of the Fed’s preferred inflation measure, the year-over-year percent change in core consumer prices in the GDP deflator, over the past several years. Early in 2007, it was 2.6 percent. It is now at its low point, of 1.1 percent. In between, there was a local trough of 1.2 percent late in 2009 and a local peak of 2.0 percent, about a year and a half ago.
If in 2006 you had given this data to practicing macroeconomic forecasters and asked for a prediction about the behavior of unemployment, what would have been the response? My guess is that the majority would have predicted no rise in unemployment at all. The remaining forecasters would have predicted (or inferred, based on an estimated Phillips Curve) a small increase in unemployment in 2009, followed by another small increase in 2012 and 2013. No one would have predicted or inferred an enormous increase in unemployment late 2008 and early 2009, followed by a gradual decrease and then a more pronounced decrease this year.
And then he jumps to this:
Let me re-state my concern with measuring aggregate demand as nominal GDP. If inflation remains absolutely constant, then any fluctuation in nominal GDP is arithmetically a fluctuation in real GDP. At that point, “explaining” changes in real GDP by changes in nominal GDP becomes completely circular.
First of all, the Fed’s preferred measure of inflation is not the GDP deflator. If it were, as I understand the market monetarist argument, the economic crisis that ensued from 2007 onward would not have been as severe. And of course, if you hand faulty data to a forecaster, you get a faulty forecast. If the inflation measure that the Fed does use, the PCE headline measure, is provided, then it is far more likely to produce a more accurate prediction of the behavior of unemployment from 2008 to present.
Instead of proving any real defect in the argument for NGDPLT, Mr. Kling actually provides reinforcement for the argument against inflation targeting by bringing to light the point that there are too many ways to measure inflation, with some being marginally useful for the purpose of monetary policy, while others should not be used for that purpose at all – unless of course we could all agree that the Fed should be tightening policy during a negative supply shock. Would Mr. Kling agree that would have a real cost associated with it in terms of employment?
Herein lies the entire argument for NGDPLT, at least from my perspective. The Fed can control one of two nominal variables at any given point in time, either inflation or NGDP. Inflation is nebulous, difficult to measure, has sources other than monetary, and the various ways to measure it provide many opportunities for fluid policy choices from one measure to the next. Measuring NGDP is more concrete and thus much less discretionary, reducing the amount of adverse surprises (like wow, we couldn’t have expected unemployment to soar) with no one able to understand exactly what happened.
Given this circumstance, it is hardly relevant to assume that inflation would remain constant, much less that the intent of market monetarists is to explain anything in terms of real GDP. In fact the expectation of market monetarists is that inflation would fluctuate with supply shocks.
That NGDPLT strawman has plenty of bruises, and there is lots of straw flying around. But my guess is that NGDPLT is still in very fine shape.
HT: Becky Hargrove