On August 2020, Fed officials announced a new approach to monetary policy, which they called “Flexible Average Inflation Targeting”. The idea was to allow some variation in inflation in the short run, but aim for an average inflation rate of 2% in the mid- to longer run.
What they actually did was something radically different. In 2021, the Fed adopted a 1960s-style highly stimulative monetary policy in an attempt to “create jobs” by printing money. Just as in the 1960s, that policy led to high inflation.
Later, the Fed claimed that they had never intended to target the average inflation rate. Rather the policy aimed to make up for periods where inflation ran below target, but not for periods when it ran above target. I felt like a dummy, as I had naively believed that average inflation targeting meant average inflation targeting.
London School of Economics Professor Ricardo Reis is certainly no dummy, and he had the same view as I had:
So where could Reis and I have gotten this crazy idea that average inflation targeting meant average inflation targeting? Perhaps from the Fed itself. In an April 6, 2021 paper, Dallas Fed economists Enrique Martínez-García, Jarod Coulter and Valerie Grossman also claimed that the policy was symmetric:
Notably, the Fed changed its language on inflation, replacing its 2 percent inflation target commitment, and instead said it will “[seek] to achieve inflation that averages 2 percent over time.”
This change is a substantial departure from the previous flexible inflation-targeting regime. Monetary policy under inflation targeting was symmetric—the Fed would equally respond to overshooting and undershooting of the target. The Fed lets “bygones be bygones,” since it does not attempt to make up for past inflation deviations from target.
By comparison, average inflation targeting means that policymakers would consider those deviations and can allow inflation to modestly and temporarily run above the target to make up for past shortfalls, or vice versa.
Note that the phrase “vice versa” is italicized in the original. They thought this point was worth emphasizing.
In a recent tweet, David Beckworth suggests that Jerome Powell is leaning toward an abandonment of FAIT, and a return to a flexible inflation targeting (FIT) regime:
David’s entire twitter thread threat is worth reading. He points out that the FAIT policy was based on a long series of important papers that I have dubbed the “Princeton School” of monetary policy. These papers emphasize the need for some sort of level targeting regime, focusing either on the price level or nominal GDP. These proposals aimed to correct very specific flaws in the previous inflation targeting regime, which led to the big policy failure of 2008-15.
So let’s review what happened here:
1. In 2020, the Fed adopted FAIT, based on highly respected research into the question of what went wrong in 2008.
2. The plain meaning of the term “average” suggests the policy was symmetric. I thought it was symmetric. A Dallas Fed publication said the policy was symmetric.
3. The policy did generate a robust recovery, but it ended up creating too much inflation.
4. To the extent that the policy failed, it failed because it was not symmetric. The Fed aimed to correct inflation undershoots, but not overshoots. It is not a question of the Fed failing to achieve flexible average inflation targeting after trying really hard; they never even attempted FAIT. They attempted something entirely different, 1960s-style monetary stimulus.
Unfortunately, in our culture words have an almost magical power, a talismanic power. If an institution announces that it will undertake policy X, and then undertakes policy Y, any success or failure will be based not on the policy that was actually undertaken, rather it will be attributed to the policy that was announced. The Fed announced that it would do FAIT, did something entirely different, and now (if Beckworth’s tweet is correct) seems about to abandon FAIT and replace it with something far worse.
On the bright side, a cynic might argue that perhaps next time they’ll announce policy Y (FIT), but actually do policy X (FAIT). Unfortunately, in order for these sorts of policies to work they need to be well understood by the financial markets, and at least somewhat credible.
I understand that the Fed feels a need to do something different after the fiasco of 2021-22. So why not announce a policy of NGDP level targeting at 4%/year? Given the long run US growth rate of roughly 2%, that sort of policy will produce an average inflation rate of close to 2%, and it will be more “flexible” when there are supply shocks like Covid and the Ukraine War.
PS. In the 30 years before average inflation targeting, PCE inflation averaged 1.9%. Since August 2020, it has averaged 4.2%, or 3.6% if you take a five year average to avoid Covid distortions.
READER COMMENTS
Kevin Dick
Nov 23 2024 at 2:13pm
Flexible Average Inflation Targeting is neither average or targeting?
It certainly appears to be flexible!
David Henderson
Nov 23 2024 at 2:28pm
Great post.
One possible typo: “David’s entire twitter threat is worth reading.” I think you mean “thread.” But I haven’t read it yet. It could be a threat. “-)
Scott Sumner
Nov 23 2024 at 8:22pm
Thanks, I fixed it.
David S
Nov 23 2024 at 2:34pm
How will the pattern of rate cuts, or potential lack thereof, in the first half 2025 give us a clear picture of Fed policy stance?
If FAIT is dead, then should we expect modest rate cuts even if PCE inflation is slightly elevated? Or if NGDP growth stagnates should we expect even more aggressive cuts?
I’m curious to see if persistent or accelerating inflation during 2025 as a consequence of GOP moves tariffs and taxes actually inspires the Fed to tighten. Which I think they could do simply by signaling that there will be no cuts until inflation moderates.
Scott Sumner
Nov 23 2024 at 8:21pm
I don’t like to think of monetary policy in terms of interest rates. A slowdown in rate cuts could mean tighter money, but it could also indicate that money was easier, causing faster NGDP growth, which then explains why rates fell less than expected.
But money should probably be a bit tighter even under a simple FIT approach, as inflation looks set to at least slightly exceed 2% for a while.
Thomas L Hutcheson
Nov 25 2024 at 5:38pm
A (big enough) tariff increase woud be a shock requiring changes in relative prices just as much as an increase in international prices of an important traded good like petroleum, or a sudden fall in the labor force like COVID. To such a shock the Fed would exercise “flexibility” to temporarily raise inflation above the targeted average to facilitate the needed adjustments in relative prices. Tightening would not be appropriate until it sought to return inflation to target.
Garrett
Nov 23 2024 at 11:15pm
The idea that the Princeton school could be discredited by the association with FAIT reminds me of the discrediting of the Chicago school by the association with Chile in the 1970s-80s
Scott Sumner
Nov 23 2024 at 11:45pm
Yes, sometimes life is unfair.
Craig
Nov 24 2024 at 12:46am
Higher for ‘longer’?
One surely muat take the Fed seriously, but never quite literally.
Chris Gardner
Nov 24 2024 at 7:58am
Since the preferred core PCE has been above CPI, I anticipate a new preferred rate “ex- energy, food, housing, services, clothes, cars, travel, etc” and call it “super duper ultra mega core CPI” and claim it smooths out volatility. It will only include the price of apples since those have recently dropped in price.
Lizard Man
Nov 24 2024 at 8:18am
How high was inflation during the late 80’s early 90’s? It must have been really high to offset a decade of undershooting 2% inflation during the Great Recession. I am thinking that if the Fed were to do real average inflation targeting, they would need a goal of 3% inflation. Because average inflation targeting 2% gave us the Great Recession. And if that’s true, 3.6% isn’t far off from target.
Scott Sumner
Nov 24 2024 at 1:26pm
The late 1980s don’t matter for the data I cited. Inflation was a bit elevated during the first year or two of the 1990s, but since then it’s mostly stayed close to 2%. It was a bit high in 2007-08.
Jose Pablo
Nov 25 2024 at 12:28pm
I had naively believed that average inflation targeting meant average inflation targeting
Well, this was, in fact, naive, since it is easy to see that “correcting” for periods of low inflation and “correcting” for periods of high inflation don’t have, by any means, the same political cost. Fed officials are, after all, humans (I think). So, if the political costs are asymetrical the FED policies will also be so.
It is very difficult to imagine the FED pushing interest rates higher in today’s economy and causing more economic pain, just to bring inflation under the 2% totally arbitray level just because “average inflation” (defined as measured from a totally arbitray starting point) should be kept at 2%.
After all, being unemployed or not being able to afford a house, sound like more pressing problems than being seen as “heretics” in the cult of an arbitrary lane in the quick sand of inflation rates.
In any case, monetary policy the way the humans at the FED helm are able to practice it, does result in a 2.4-2.5% average inflation rate over long periods of time. Just read 2.4-2.5% whenever they say 2%. Read what they meant not what they say (I do the same when my kids answer the question “what time did you came back home yesterday?” and it works pretty accurately)
Thomas L Hutcheson
Nov 25 2024 at 5:51pm
Of course the target is not supposed to be arbitrary. In principle it is chosen as the optimal amount of inflation to allow relative prices to adjust to average level of “Brownian movement” mini shocks given average stickiness of some nominal prices.
But some above average shocks do come along requiring temporary over-average inflation, so backward-looking average will always be above the targeted average as you say.
Jose Pablo
Nov 28 2024 at 1:03pm
Thomas L Hutcheson
Nov 25 2024 at 5:29pm
I can see how one might interpret the “average” as backward looking, but the idea a targeting is forward looking, so should the average being targeted. But leaving aside the semantics, the whole idea of flexibility is that extraordinary shock require temporarily over target inflation until relative pries have adjusted to the shock and then inflation is brought back down to target. But what kind o event would make under-target inflation optimal. Sure, a central bank might re-estimate their model and decide that a different, lower target was optimal, but within a regime in which the optimal target is unchanged, what would qualify as an “anti-shock” to justify an anti-over target 🙂 inflation?
Thomas L Hutcheson
Nov 25 2024 at 5:41pm
“It is very difficult to imagine the FED pushing interest rates higher in today’s economy and causing more economic pain …”
or any advantage to doing so.
Thomas L Hutcheson
Nov 25 2024 at 5:59pm
This presents the Fed with an interesting problem. What if the public does not like the amounts of over-target inflation that the Fed thinks is necessary to keep markets clearing after shocks (that is, to avoid unemployment) and is prone to punish the party holding the Presidency at the time. Should the Fed act “politically” to preserve the Administration or just power through with income maximization?
MarkLouis
Nov 26 2024 at 3:20pm
My personal theory (which I cannot prove) is that the Fed intended it to be symmetric but never expected to have an overshoot. When confronted with an overshoot they looked at what needed to be done and simply lacked the courage.