NGDP Targeting: Altering Theory to Fit History

In an article from last Fall, titled Re-Targeting the Fed, Scott Sumner provides his reasoning for instituting a nominal GDP (NGDP) target to improve monetary policy. According to Sumner (emphasis mine):

NGDP growth, which is made up of the inflation rate plus "real" (or inflation-adjusted) GDP growth, would have been a better indicator of the severity of the crisis. Between 2008 and 2009, NGDP declined at the fastest rate since 1938, while inflation (at least the "core" inflation rate, which excludes food and gas prices and which the Fed uses as its key inflation measure) raised no red flags. Because inflation cannot quickly adjust to such sudden drops in spending, this decline in nominal GDP brought about a sharp decline in real GDP — thus, a severe recession.

In an important sense, the sharp drop in NGDP precipitated the crisis: It was the proximate cause, even if the housing crisis was the ultimate cause. This suggests that NGDP is useful not only as a predictor and indicator of trouble, but as a target for monetary policy. Setting a goal in terms of nominal GDP could provide a superior alternative to the Fed's current inflation targets.

Unless I’m mistaken, Sumner is suggesting here that a change in NGDP has a causal effect on RGDP, which can help explain the recent recession and continuing economic malaise.

This past weekend, in a Reply to Karl Smith, Sumner:

pointed out that during the 1970s we had normal (3.2%) growth and that 100% of the high inflation was due to NGDP growth being much higher than 5%, indeed about 10.4% on average between 1970 and 1980.  So even during the 1970s, inflation would have been only around 1.8% under NGDP targeting.

Did you notice the change in assumptions? If not, read that quote again.

Referring to the 1970’s period of stagflation, Sumner is now claiming that had NGDP been reduced from an average of 10.4% to 5% during the decade, RGDP would have remained exactly the same (5% NGDP target - 3.2% RGDP = 1.8% inflation). What happened to the causal effect whereby a “decline in nominal GDP brought about a sharp decline in real GDP”? If we are to believe Sumner’s theories, the causal effect is therefore a relatively new phenomenon.

Even accepting this fact, how can we be sure that the relationship won’t change again if policy is altered (See NGDP Targeting: Changing Policy Changes Relationships and NGDP Targeting: Changing Policy Changes Relationships (Part 2))? Separately, if Sumner’s theory about the 1970’s is correct (that RGDP is unaffected), then why bother targeting NGDP at all? One might claim that people subjectively feel better with higher prices, but that seems like awfully weak support for monetary policy.

In my opinion, the more likely case is that NGDP does have a causal relationship with RGDP (corresponding to changes in private credit outstanding), however that theory and the 1970’s stagflation fail to align. Under NGDP targeting in the 1970’s, both inflation and RGDP would probably have been lower, with the relative decreases remaining unknown. Although I remain open to the idea of NGDP targeting, attempts such as these to fit the theory to the story only heighten my skepticism.

(Note: The discussion above implicitly assumes that the Federal Reserve could successfully implement NGDP targeting, but I remain doubtful that monetary policy can achieve those goals.)