The Fed is currently contemplating a set of monetary policy changes that might be viewed as “second best”. These include yield curve control and average inflation targeting. With yield curve control the central bank would peg the yield on longer-term Treasury bonds. This was Fed policy during the 1940s. Under average inflation targeting the central bank tries to make up for an inflation undershoot or overshoot, so that inflation will average 2% over the business cycle.
In my view, these are second best policies. Holding down long-term interest rates might be expansionary, but it also might end up being contractionary. After all, contractionary monetary policy can easily decrease long-term interest rates by reducing expectations of inflation and economic growth.
Average inflation targeting might be expansionary during a recession, but it also might lack credibility for exactly the same reason that the current inflation targeting regime lacks credibility; there is no hard and fast commitment to make up for inflation undershoots, just an intention to try to do so.
In my view, first best policies would involve level targeting (of prices or better yet NGDP) and a “whatever it takes” approach to monetary expansion to hit those level targets. That’s the only policy regime that would give me confidence that the Fed would actually hit its target over the very long run.
So here’s the question. Should we view the adoption of second best policies as a step forward, even if not optimal? Or would they be a step that makes the optimal policy less likely? That is, would average inflation targeting involve the opportunity cost of rejecting level targeting?
What do you think?
READER COMMENTS
Garrett
Jul 29 2020 at 4:52pm
I’d say the analysis should go something like this:
What is the distribution of future outcomes in the current monetary policy regime?
What is the distribution of future outcomes in the “second best policies” and “first best policies regimes? By how much is second-best better than current and how much is first-best better than second-best?
What is the probability of future adoption of first-best conditional on adoption of second-best, and what is the probability conditional on rejection of second-best?
Is the conditional distribution of adoption better or worse than the conditional distribution of rejection?
For a simplified example with the distributions collapsed into expected values, say the current regime is worth 0, second-best is worth +1, and first-best is worth +2. If second-best is adopted there’s a 20% chance that first-best will be adopted later, but if second-best is rejected then there’s a 50% chance that first-best will be adopted in the future. So adoption of second-best is worth 0 + 1 + (2 * 20%) = 1.4, while rejection is worth 0 + (2 * 50%) = 1. In this example, adoption of second-best policies should be supported.
The debate is regarding not only the values I assumed, but also the shapes of the distributions and how to rank the different distributions.
Scott Sumner
Jul 29 2020 at 8:34pm
I agree with that framing. Of course the hard part is estimating those probabilities.
tpeach
Jul 29 2020 at 7:04pm
The RBA has long had an average inflation target and has performed pretty well, except the last few years when they were too concerned with high house prices and kept monetary policy too tight even though inflation was consistently below target.
They have also recently pegged the 3 year interest rate at 0.25%. That’s been so credible that they haven’t needed to purchase any more govt bonds since a few weeks after the target was announced in March.
I think that if Michael Woodford were to run a central bank it would look a lot like the RBA. Both policies are very New Keynesian, which I see as second-best to Market Monetarism. I think for NGDPLT to be introduced any time soon then the Fed would really have to screw up again big time, which would certainly be sub-optimal.
Scott Sumner
Jul 29 2020 at 8:36pm
I agree that central banks can often peg the bond yield without many purchases. I seem to recall, however, that the BOJ actually had to cut back on purchases to prevent the rate from falling below the peg. In that case the policy might end up being (unintentionally) contractionary.
Thomas Hutcheson
Jul 29 2020 at 7:42pm
Why cheer or jeer? Just keep advocating for what one thinks is optimal policy.
For me, that is price level targeting starting no later than 2016 (implying some catching up to do at an uncertain rate) as the “price stability” half of the Fed’s Congressional mandate plus the other half of the mandate, full employment, preferable with the expanded sense of full employment of all resources, not just labor. This policy would be slightly less expansionary during a supply shock than NGDPL targeting, but I think the current situation shows just how small the supply shock really is, so the difference would not be great.
I’m a total agnostic about which (at least two) instruments should be used to achieve these targets, perhaps intervention in the TIPS market for price level trajectory control and the foreign exchange markets for employment.
Michael Sandifer
Jul 30 2020 at 1:44am
Average inflation targeting is a welcome step forward, though insufficient. I don’t think there’s much chance that NGDP targeting in the near future is the opportunity cost, as it seems unlikely it would have been adopted soon.
I don’t like trying to hold down long rates for the reasons you mention. It would be better to actually target higher long nominal rates, with sufficient stimulus. I fear this could be an example of one step forward, and one or two steps back.
Todd Ramsey
Jul 30 2020 at 9:35am
We should welcome and cheer second-best. Millions of lives were not ruined because the Fed took unprecedented massive, rapid action in mid-March.
We should also continue to push for first-best, NGDP level targeting. Now, where can we find a burgeoning public intellectual, who has already influenced Fed policy (including the March intervention) to eloquently articulate the case for NGDP level targeting?
We should leave no stone unturned in scouring the streets of Mission Viejo searching for such a person.
Scott Sumner
Jul 30 2020 at 1:40pm
Todd, You said:
“We should leave no stone unturned in scouring the streets of Mission Viejo searching for such a person.”
I’ll put my mask on and help you look for him or her.
marcus nunes
Jul 30 2020 at 7:22pm
Any policy that does not REQUIRE make-up policy is sub-optimal. The opportunity cost of that sub-optimal policy is immense (the sum of lost output, or the area underneath the trend line).
An example is given here: https://thefaintofheart.wordpress.com/2020/07/20/after-covid19-inflation/
Comments are closed.