Excellent piece. If I were them, I'd say two things. The pandemic was an extraordinary situation, so we're starting the new FAIT clock today. And going forward, it's symmetrical. Well, actually, I'd just say one thing. That we are now switching to 4% NGDPLT. But easy for me to say though.
I suspect that if they hit the soft landing, and don't trigger a recession that brings inflation back to near-zero, what they will do, in effect, is go back to Bernanke-ism for 5-7 years, very-slightly under-shooting 2% for a long enough time that on a ten-year window, just before the inflation spike of '21-'22 falls out, you'll have your 2% average.
This is a solid explainer -- well done. My one quibble is the over-emphasis on clarity as a strong goal.
In my field of software engineering, I've worked for managers and organizations that fantasized about systems that would impose analogous kinds of clarity on complex systems or processes. The results were typically bleak: a lot of meetings, debate, and sometimes engineering effort to achieve very little net benefit.
Clarity, like most virtues, is good ... up to a point. For example, if one really wanted clarity, the Fed could implement a computer program that would take a variety of economic indicators and spit out the next interest rate adjustment. Most of us have read enough dystopian fiction to have a vague sense that rigid rule-based systems always, eventually fail to anticipate every possible situation.
With complex systems, you ultimately need to balance the rigidity of total clarity against the flexibility and wisdom that a decision making body of subject matter experts can bring to bear. I think we're better off with roughly as much clarity as we currently have from the Fed: general guidelines, yes, but enough ambiguity to allow experts to make decisions within those parameters.
Datum :) point. The 10 year TIPS is currently at about 3% pa c.f. _my_ estimate of 2.3% as the equivalent of the Fed's 2% PCE target. Not bad but not on (barely within 2 sigma) target, either.
Mathematically speaking, overshooting one year only means the Fed should undershoot the next if the window for the average is one year, which would be silly. Too long a window would also be silly because then annual changes would have little effect on that moving average. The Fed should specify a time window they’re using for their average (if they haven’t already).
Superb post, Milan. I haven’t thought about a lot of this stuff since MacroEcon in college, but it is probably some of the most important policy shaping America today.
Is the following too obvious to point out? Or does it show insufficient economic training? Or might it actually be interesting: weighing low inflation against low unemployment is a mismatch, because employment costs are only one of many industrial inputs. We should weigh low inflation against low unemployment plus strong antitrust enforcement. In other words, the Fed should scold the Executive to always be denying individual industrial actors pricing power, in order to help the Fed do its job. In other other words, antitrust is not merely nice to have to help insure consumer satisfaction. It is essential to have to enable the Fed to do its job.
How would lax antitrust lead to economy-wide inflation? Antitrust only focuses consolidation in a single market sector. For consolidation to lead to economy-wide inflation every single market sector would have to controlled by a monopoly, which is definitely not the case.
One quibble: is it possible you originally wrote it back in March? The quoted line below was true then but it’s out of date now:
“the Fed raised rates by a quarter of a percentage point last week, with six more rate hikes expected in 2022.”
After raising policy rate by 0.25% in March, the Fed raised their policy rate an additional 0.5% last week, with five more meetings scheduled for the rest of 2022.
Your best work yet, Milan!! Great explainer, thanks!
Maybe I am over-simplifying, but if the previous ceiling of 2% resulted in an average inflation of 1.5, why wouldn’t they just raise the ceiling to 2.5%?
Do you have an informed view of whether the Fed's decade long F---up was a fear of inflation going over 2% or a reluctance to do more QR? And either way so how did that feed into an objective function? What bad would have happened if.
Interesting. So, I’m an engineer. When we make drawings for manufacturing, we usually put a dimension plus or minus a certain amount (“tolerance”) knowing that it’s physically impossible to precisely hit nominal. That said, the manufacturing guys will often target slightly on the “material on” side of nominal (like, a hole’s diameter is slightly smaller than nominal on the drawing) so that they have the ability to re-work the item to fit the drawing if they undershoot. If they overshoot, the part is scrap- very difficult to put material back on. Kind of like targeting slightly higher inflation due to the zero lower bound. Anyway I guess what I’m suggesting is a target of 2% +/- 0.5%. Maybe that would have been politically toxic in an era of 1.5% inflation but I think it would sound pretty good to most folks now!
Another option is level targeting. Price level (or even better, nominal GDP level targeting). So if the price level is 100 today, in 5 years it should be 1.02^5 = 110.41. So each miss (over or under) is compensated for going forward. In 10 years, price level should be 121.90. Then you have to get to the 2% on average without making 2.5% a target.
Also an engineer here, and I was thinking something similar.
If the problem is that a simple "targeting an average of 2%" causes too much uncertainty for investors/markets, why not just add in a tolerance/bounding so that everyone knows that they won't oscillate severely from year to year?
"Targeting an average of 2%, with a yearly boundary of +/-0.5 or 1%"
Unless something like that would significantly reduce their flexibility in addressing weird situations? Like the one we find ourselves in now.
Sometimes I forget that we engineers are odd :) But, like, there is a whole manufacturing philosophy around this, Statistical Process Control, where you monitor a metric and can use an array of triggers to indicate that the process is drifting "out of control." Now, inflation is a bit of what we'd call a "lagging indicator." It takes time to start to see the effects of your policy changes. So it would be important to be really specific about what calculation of inflation you're tracking. The monthly year-over-year inflation number seems like a good one to track. I had to google to jog my memory, but the traditional triggers are:
-A single point outside the control limits
-Two out of three successive points are on the same side of the centerline and farther than 2 σ from it
-Four out of five successive points are on the same side of the centerline and farther than 1 σ from it
-A run of eight in a row are on the same side of the centerline. Or 10 out of 11, 12 out of 14, or 16 out of 20
-Obvious consistent or persistent patterns that suggest something unusual about your data and your process.
It’s essentially unarguable that the Fed / Govt (remember the Fed is “independent”) has for decades underestimated real inflation - see John Williams on “shadow” inflation - to justify loose monetary policy to drive down unemployment & keep the economy rolling along. At the same time, with every central bank around the globe essentially doing the same (even if we presumably led the way, others would have adopted this methodology irrespective) we were forced to play along; if not, our $ would be so strong (due to higher rates) our exports would not be competitive, along with further impairing our own economy. And we know this would not have been politically acceptable. Central banks over stuffed the world with liquidity - as, let’s face it, the global populace (and electorates) - preferred the illusion & ez “juicing” of the global economies. However, this served to destroy purchasing power…having a particularly negative regressionary impact (lower income earners were forever crushed). It also destroyed seniors that count on fixed income vehicles & were forced to take higher equity risk to create needed wealth. Decades of excessive central bank economic juicing is now coming home to roost as we move to quell even the sharply understated inflation of today.
Thanks Jason. Again, I’m happy to acknowledge that John Williams methodology may not be on point. Frankly, I’ve never analyzed it. However, the way the CPI is calculated, which I’ll take for granted that you’ve looked at, is a farce. Improved productivity of something going into the basket drops it’s price. Many other elements serve to understate inflation IMHO. Regardless, I’ll simply ask, what sort of inflation do you believe you & your family are subjected historically (you pick the timeline you want to address) and say in the past 12 months?
> Improved productivity of something going into the basket drops it’s price.
Yes, I generally agree with this methodology of calculating inflation. If I buy a TV today for $500, and next year I can buy a better (larger) TV for the same price, then clearly I'm getting more for my money, and some amount of disinflation has occurred.
I agree that it can be hard to calculate that tradeoff, and I'm not trying to say that the CPI is perfect, but in general I agree with what they are trying to do.
None of this is to say that we are not in an inflationary period today, but that has no bearing on the "correctness" of the CPI.
In fact the CPI, as you are aware, is 8% the last couple of months, so clearly it is representative of a sharply higher inflationary environment than we’ve had over the past 30+ years. But, no, that does not speak to the “correctness” or absence thereof…agreed. At the same time, I’m (genuinely) happy to hear that you and/or others feel that improved productivity is disinflationary. Thank you & wishing you a great May weekend.
My apologies for word choice & reference to John Williams. IMHO (shared by many, if not most), CPI severely understates true inflation. I am not conspiracy-minded (socially liberal, fiscally conservative, left leaner, Mike Bloomberg enthusiast, Trump hater). Oh, and I’m a career security analyst (now 40 years), as well as a portfolio manager and real estate investor. ..oh, and a CFA. I’ll assume you understand how CPI is calculated.
Equally so, the CPI is not an accurate gauge simply cause you or myriad / most economists suggest it is. The 2020 election was not stolen from Trump. But I’m not going to spend hours proving it or linking to news stories. One could always suggest that the references of someone else are “vague” or the “proof” inadequate. Believe what you like.
Excellent piece. If I were them, I'd say two things. The pandemic was an extraordinary situation, so we're starting the new FAIT clock today. And going forward, it's symmetrical. Well, actually, I'd just say one thing. That we are now switching to 4% NGDPLT. But easy for me to say though.
1. The Fed had never brought inflation like this without triggering a recession.
2. A 2% inflation target is stupid. At 2% inflation you lose half the value of your money in a generation (35 years)
Very informative - thank you Milan. :)
You're quite welcome
I suspect that if they hit the soft landing, and don't trigger a recession that brings inflation back to near-zero, what they will do, in effect, is go back to Bernanke-ism for 5-7 years, very-slightly under-shooting 2% for a long enough time that on a ten-year window, just before the inflation spike of '21-'22 falls out, you'll have your 2% average.
This is a solid explainer -- well done. My one quibble is the over-emphasis on clarity as a strong goal.
In my field of software engineering, I've worked for managers and organizations that fantasized about systems that would impose analogous kinds of clarity on complex systems or processes. The results were typically bleak: a lot of meetings, debate, and sometimes engineering effort to achieve very little net benefit.
Clarity, like most virtues, is good ... up to a point. For example, if one really wanted clarity, the Fed could implement a computer program that would take a variety of economic indicators and spit out the next interest rate adjustment. Most of us have read enough dystopian fiction to have a vague sense that rigid rule-based systems always, eventually fail to anticipate every possible situation.
With complex systems, you ultimately need to balance the rigidity of total clarity against the flexibility and wisdom that a decision making body of subject matter experts can bring to bear. I think we're better off with roughly as much clarity as we currently have from the Fed: general guidelines, yes, but enough ambiguity to allow experts to make decisions within those parameters.
Datum :) point. The 10 year TIPS is currently at about 3% pa c.f. _my_ estimate of 2.3% as the equivalent of the Fed's 2% PCE target. Not bad but not on (barely within 2 sigma) target, either.
Datums.
I would argue that "data point" is now also correct by virtue of being common use. Words mean the things they get used to mean
I agree. I as just having fun. :)
And maybe more importantly, "data point" is almost surely better writing because everyone will understand it.
Mathematically speaking, overshooting one year only means the Fed should undershoot the next if the window for the average is one year, which would be silly. Too long a window would also be silly because then annual changes would have little effect on that moving average. The Fed should specify a time window they’re using for their average (if they haven’t already).
Excellent, balanced post about a thorny and technically complex issue. Thanks Milan! Great work.
The fed does not target core inflation, and that suggestion changed the analysis considerably
Excellent post.
Superb post, Milan. I haven’t thought about a lot of this stuff since MacroEcon in college, but it is probably some of the most important policy shaping America today.
Is the following too obvious to point out? Or does it show insufficient economic training? Or might it actually be interesting: weighing low inflation against low unemployment is a mismatch, because employment costs are only one of many industrial inputs. We should weigh low inflation against low unemployment plus strong antitrust enforcement. In other words, the Fed should scold the Executive to always be denying individual industrial actors pricing power, in order to help the Fed do its job. In other other words, antitrust is not merely nice to have to help insure consumer satisfaction. It is essential to have to enable the Fed to do its job.
How would lax antitrust lead to economy-wide inflation? Antitrust only focuses consolidation in a single market sector. For consolidation to lead to economy-wide inflation every single market sector would have to controlled by a monopoly, which is definitely not the case.
This is a great piece.
One quibble: is it possible you originally wrote it back in March? The quoted line below was true then but it’s out of date now:
“the Fed raised rates by a quarter of a percentage point last week, with six more rate hikes expected in 2022.”
After raising policy rate by 0.25% in March, the Fed raised their policy rate an additional 0.5% last week, with five more meetings scheduled for the rest of 2022.
Yes I did write this circa March 16, it’s just been in the pipeline for a while
Being able to explain complicated things in a clear way indicates a thorough grasp of the subject (and makes the reading pleasurable besides). Thanks!
Your best work yet, Milan!! Great explainer, thanks!
Maybe I am over-simplifying, but if the previous ceiling of 2% resulted in an average inflation of 1.5, why wouldn’t they just raise the ceiling to 2.5%?
Do you have an informed view of whether the Fed's decade long F---up was a fear of inflation going over 2% or a reluctance to do more QR? And either way so how did that feed into an objective function? What bad would have happened if.
What Sahm told me is that politically doing that would be dead in the water
But did Sahm like it?
Interesting. So, I’m an engineer. When we make drawings for manufacturing, we usually put a dimension plus or minus a certain amount (“tolerance”) knowing that it’s physically impossible to precisely hit nominal. That said, the manufacturing guys will often target slightly on the “material on” side of nominal (like, a hole’s diameter is slightly smaller than nominal on the drawing) so that they have the ability to re-work the item to fit the drawing if they undershoot. If they overshoot, the part is scrap- very difficult to put material back on. Kind of like targeting slightly higher inflation due to the zero lower bound. Anyway I guess what I’m suggesting is a target of 2% +/- 0.5%. Maybe that would have been politically toxic in an era of 1.5% inflation but I think it would sound pretty good to most folks now!
Another option is level targeting. Price level (or even better, nominal GDP level targeting). So if the price level is 100 today, in 5 years it should be 1.02^5 = 110.41. So each miss (over or under) is compensated for going forward. In 10 years, price level should be 121.90. Then you have to get to the 2% on average without making 2.5% a target.
Also an engineer here, and I was thinking something similar.
If the problem is that a simple "targeting an average of 2%" causes too much uncertainty for investors/markets, why not just add in a tolerance/bounding so that everyone knows that they won't oscillate severely from year to year?
"Targeting an average of 2%, with a yearly boundary of +/-0.5 or 1%"
Unless something like that would significantly reduce their flexibility in addressing weird situations? Like the one we find ourselves in now.
Sometimes I forget that we engineers are odd :) But, like, there is a whole manufacturing philosophy around this, Statistical Process Control, where you monitor a metric and can use an array of triggers to indicate that the process is drifting "out of control." Now, inflation is a bit of what we'd call a "lagging indicator." It takes time to start to see the effects of your policy changes. So it would be important to be really specific about what calculation of inflation you're tracking. The monthly year-over-year inflation number seems like a good one to track. I had to google to jog my memory, but the traditional triggers are:
-A single point outside the control limits
-Two out of three successive points are on the same side of the centerline and farther than 2 σ from it
-Four out of five successive points are on the same side of the centerline and farther than 1 σ from it
-A run of eight in a row are on the same side of the centerline. Or 10 out of 11, 12 out of 14, or 16 out of 20
-Obvious consistent or persistent patterns that suggest something unusual about your data and your process.
Source: https://asq.org/quality-resources/control-chart
It’s essentially unarguable that the Fed / Govt (remember the Fed is “independent”) has for decades underestimated real inflation - see John Williams on “shadow” inflation - to justify loose monetary policy to drive down unemployment & keep the economy rolling along. At the same time, with every central bank around the globe essentially doing the same (even if we presumably led the way, others would have adopted this methodology irrespective) we were forced to play along; if not, our $ would be so strong (due to higher rates) our exports would not be competitive, along with further impairing our own economy. And we know this would not have been politically acceptable. Central banks over stuffed the world with liquidity - as, let’s face it, the global populace (and electorates) - preferred the illusion & ez “juicing” of the global economies. However, this served to destroy purchasing power…having a particularly negative regressionary impact (lower income earners were forever crushed). It also destroyed seniors that count on fixed income vehicles & were forced to take higher equity risk to create needed wealth. Decades of excessive central bank economic juicing is now coming home to roost as we move to quell even the sharply understated inflation of today.
https://fullstackeconomics.com/no-the-real-inflation-rate-isnt-14-percent/
Thanks Jason. Again, I’m happy to acknowledge that John Williams methodology may not be on point. Frankly, I’ve never analyzed it. However, the way the CPI is calculated, which I’ll take for granted that you’ve looked at, is a farce. Improved productivity of something going into the basket drops it’s price. Many other elements serve to understate inflation IMHO. Regardless, I’ll simply ask, what sort of inflation do you believe you & your family are subjected historically (you pick the timeline you want to address) and say in the past 12 months?
> Improved productivity of something going into the basket drops it’s price.
Yes, I generally agree with this methodology of calculating inflation. If I buy a TV today for $500, and next year I can buy a better (larger) TV for the same price, then clearly I'm getting more for my money, and some amount of disinflation has occurred.
I agree that it can be hard to calculate that tradeoff, and I'm not trying to say that the CPI is perfect, but in general I agree with what they are trying to do.
None of this is to say that we are not in an inflationary period today, but that has no bearing on the "correctness" of the CPI.
In fact the CPI, as you are aware, is 8% the last couple of months, so clearly it is representative of a sharply higher inflationary environment than we’ve had over the past 30+ years. But, no, that does not speak to the “correctness” or absence thereof…agreed. At the same time, I’m (genuinely) happy to hear that you and/or others feel that improved productivity is disinflationary. Thank you & wishing you a great May weekend.
“Its” and not “it’s” …typo. Excuse me.
As I noted to David & John, thank you for pointing out the fallacies of J Williams’ work.
My apologies for word choice & reference to John Williams. IMHO (shared by many, if not most), CPI severely understates true inflation. I am not conspiracy-minded (socially liberal, fiscally conservative, left leaner, Mike Bloomberg enthusiast, Trump hater). Oh, and I’m a career security analyst (now 40 years), as well as a portfolio manager and real estate investor. ..oh, and a CFA. I’ll assume you understand how CPI is calculated.
Equally so, the CPI is not an accurate gauge simply cause you or myriad / most economists suggest it is. The 2020 election was not stolen from Trump. But I’m not going to spend hours proving it or linking to news stories. One could always suggest that the references of someone else are “vague” or the “proof” inadequate. Believe what you like.
Thank you David & Thank you John for highlighting the inadequacy of Williams’ work…I do appreciate that.