Monetary Policy: AIT seems on target
By Scott Sumner
Last year, the Federal Reserve announced a policy termed “flexible average inflation targeting.” The basic idea is that the Fed commits to insuring that PCE inflation will average roughly 2% over an extended period of time, and that any short run discrepancies will be offset by future overshoots in the opposite direction. Although they did not specify a starting point, it’s generally assumed to be roughly the beginning of the decade (say January 2020.) Thus inflation should average 2% during the 2020s.
As of today, we see the following TIPS spreads, which are a crude proxy of bond market inflation forecasts:
5 years: 2.20%
10 years: 2.14%
30 years: 2.14%
Because TIPS holders are compensated according to the CPI, and because the CPI inflation rate tends to run about 25 basis points above the PCE inflation (which the Fed is actually targeting, these TIPS spreads imply roughly this sort of expected PCE inflation:
5 years: 1.95%
10 years: 1.89%
30 years: 1.89%
PCE inflation, however, has been only about 1.1% over the past 12 months. So in principle, you want to see PCE inflation expectations of slightly over 2%/year for the next 5 years–say 2.2%, and very close to 2% over the next 30 years.
Nonetheless, these recent TIPS spreads are really good news. In my view, they are not statistically different from what you’d expect if the Fed’s new AIT policy were completely credible.
There are two reasons why the current 2.14% 30-year TIPS spread is consistent with Fed credibility. First, my estimate of the CPI/PCE discrepancy (0.25%) is backward looking. It’s quite possible that markets expect the gap to be slightly lower going forward. Second, it’s possible that the TIPS spread slightly underestimates actual market inflation expectations, due to the fact that conventional bonds are slightly more liquid than TIPS, and hence can be sold at a slightly low expected yield.
These two factors together can explain the small discrepancy between actual TIPS spreads and the sort of spread you’d see with perfect AIT credibility. Fed policy may not be perfect, but any imperfections are not statistically significant. This is a major achievement.
Of course I’d prefer level targeting, and especially NGDP level targeting. If they must target inflation, I’d prefer they target core PCE inflation. But even a successful implementation of AIT would be pretty good, relative to what we saw in many earlier decades such as the 1960s, 1970s, 1980s, and 2010s.
Now they need to carry through with what they’ve promised. PCE inflation needs to actually average 2% during the 2020s. But don’t underestimate the importance of moving market expectations close to the policy target. That’s an important first step. As recently as March 19, 5-year TIPS spread had fallen to 0.14%. So today’s 2.20% looks pretty good.