I have enormous respect, trust, and admiration for anyone willing to admit they were wrong. I have to see these kinds of admissions before I can fully listen to someone with an open mind and heart.
Attributing current inflation to the ARP is dubious. Very little of that money has been spent yet. Furthermore, the ARP is about 40% as big as the Trump eta stimulus measures. Most of the inflation that occurred last month resulted from pent up demand from 2020 and had little to do with the Georgia runoffs. That being said, turning supplemental unemployment into a reemployment bonus is an excellent idea that moderates should be able to find the votes to pass. Why aren’t Manchin and Sinema and Murkowski ramming that one through? It makes no sense.
I took a weekend trip to Boston for a family event and hotel prices were astoundingly low for that market. But when I went to rent a car the rates were outrageous. So I got a room very close to a relative and took Uber to and from the airport. So my anecdotal evidence is that the pandemic impacts are very disparate but rather extreme. This sort of volatility can't be good. Oh, and the plane flight both ways was completely booked so airlines have some capacity adjustments to make. Or not.
I have been working on a large capital project for my company, and since about January we have just been getting hammered by commodity prices, especially steel. Starting in February, we began getting hammered on the services side, with various contracting services (electrical, mechanical, etc.) suddenly coming in with much higher bids and longer lead times.
Last month both of those trends stopped getting worse for the first time. I am seeing futures prices in things like lumber and steel generally starting to cool off, so I think much of the durables squeeze will have resolved itself by the fall, with things like cars probably cooling off by next year.
My mid-brain take is it's actually good that maybe the boom didn't start immediately. Voters have really short memories, so maybe delaying the boom, even just a little, until a little further into this Congressional terms will keep it just a bit more front of mind when mid-terms come. I think voters are still willing to give economic issues some benefit of the doubt as we emerge from the pandemic, so I don't think Democrats are going to pay much of a price for this stuff if it indeed does cool off by Q4.
Nice to see Matt grapple with this. I found this a useful article in a service journalism way: I have a hard time wrapping my head around the mechanisms of inflation for the most part, so nice to see perspectives on the why of it all.
Some very good stuff in this post. Here are some things I would add:
(1) Obviously, the biggest problem with the forecast was to frame it in terms of y-o-y inflation, which missed the huge impact of the "base effect," that is, the fact that last year's price level was abnormally low. To see how misleading it is to quote y-o-y inflation now, imagine that you had almost died of hypothermia after a skiing accident. The next morning you wake up in a hospital room. The doctor looks at your chart and says "Wow! This guy's temperature has gone up 7 degrees in 24 hours! Better throw him in an ice bath quick!"
(2) People should pay more attention to measures of long term inflation. The best, which no one but the wonkiest wonks even know about, is the NY Fed's Underlying Inflation Gauge.
(3) Matt is definitely on the right track when he says we need to focus on how much prices are driven by psychology and how much by temporary scarcities. Back in the 1960s and 70s, when we had real inflation in this country, psychology played a big role. A key psychological component of economic models is inflation expectations. Back then, the thing driving inflation was the fact that people projected current levels of inflation forward, expecting next year's inflation to equal this year's, or this year's plus an acceleration factor. That produced the "Shifting Phillips Curve" or "Accelerationist" phenomenon made famous by Milton Friedman. Eventually, tight monetary policy, at great cost, brought inflation expectations under control, leading to the long era of price stability. We don't want to get back into that. For the complete story with charts and data, see this: http://dolanecon.blogspot.com/2014/05/what-ever-happened-to-phillips-curve.html (Memo to self: Update this post.)
(4) Mentioning the Fed: Part of the high cost of getting rid of inflation last time was that the Fed had to do it alone. It would be much better if we could coordinate monetary and fiscal policy to fight inflation. That should be done not so much through spending austerity as through flexible taxes that increase when inflation threatens and ease off when it does not. That is one thing the MMT crowd are right about. (It is too bad the cultishness and weirdness of many in the MMT crowd discredit a few points on which they think soundly.)
Love your analysis generally, but the main reason you missed on inflation is because you like the stimulus. Same reason a conservative would miss on any negative effects of a tax cut. There are no free lunches.
When talking about weird behaviors with deficits, inflation, etc., the weirdness is always attached to some big event, like the Depression or WWII or the oil embargo. Yet here we are just barely starting to see the light at the end of a global pandemic and everyone is freaking out over some weird inflation numbers; weird because inflation was low for a long, long time and suddenly it's rising.
Political opportunism notwithstanding, wouldn't it be even weirder if there wasn't any weirdness in the economic data right now? Matt specifically predicted lower inflation than is happening right now, but that was, in effect, a bet that this huge, unforeseen random event wouldn't have a huge impact, which was a risky bet. Also, if weirdness continues and inflation suddenly drops in the fall, will Matt's prediction count as correct? (I cannot remember if it was time-bound.)
It seems obvious in retrospect, but how many people correctly guessed the post-war economic bonanza of the 1950's?
Another compounding impact on the car makers problems with semi conductors is that the node sizes they typically used in thwir vehicles were quite ancient technology-wise for the fabs, and that once they pulled themselves out of line and the fabs easily replaced their orders (which utilized smaller more modern - and cheaper to produce - node sizes that most consumer electronics rely on) it actually started to cost the fabs money to bump up production of the outdated nodes for the car companies.
Additionally, some car companies (namely Tesla) mitigated some of these impacts by reprogramming their vehicles' firmware to utilize different, more readily available, ptofessor types. Tesla can do this because, more than Ford or Toyota, they are basically a modern software company. But this didn't remove 100% of the supply chain impacts.
So the cart companies inability to upgrade their products to more modern / efficient components (like the rest of the consumer electronics industry had been) also hurt them, because they put additionally costs on fabs in the form of leaving money on the table to work with them once supply was maxing out fab production.
Your effort to predict specific things with actual percentage forecasts is really cool, and I really respect owning up to your miss. I really hope it catches on. Since getting into forecasting as a hobby last year it's become even more frustrating that pundits talk about what they think will happen in loose and vague terms. There's a better way!
For what it's worth those of us who forecasted on a similar question on Metaculus really missed the high inflation too. (The question was intended to be the same as what you were predicting on but mistakenly was about annualized month over month instead of month over 12 months ago, so it's not a 1 to 1 comparison but we clearly were off on how high it would be). https://www.metaculus.com/questions/6645/highest-us-core-cpi-growth-in-2021/
The rest of the "Matthew Yglesias Predicts 2021" series (https://www.metaculus.com/questions/6634/matthew-yglesias-predicts-2021/) running on there has a lot of interesting questions so it'll be cool to see how that plays out. Since one of the tenets of forecasting is updating frequently I'd be interested to see you revisit your forecasts and see which ones have changed.
"I’m also not hearing people say things like “better go buy that table now, because the price will only get higher in six months.”"
Ok, I'll say it then. I looked at the inflation numbers and decided to run off and buy some durable goods --- and am planning some more purchases. I do think that the most likely outcome is that prices drop some in 6 months as supply catches up, but I also see a long tail where inflation gets a little wild. I've got more cash sitting in my bank account than I need, and my dishwasher and air conditioners already suck, so I bought replacements. That said, they've sucked for a long time, and I think that inflation fears pushed me over the edge into replacing them.
Like most journalists, when Kevin Drum uses jargon he does not not understand (MRP) and does so in an an offhand way intended to suggest he does he is wrong. He very clearly does not understand what that means and what it is used for. Furthermore the idea that one simply does an economic forecast with their abysmal record of getting things right and scores a hit every time is extremely foolish. Parenthetically ask a journalist what Planned Obsolescence means and they have about a 100 percent record of getting it wrong.
Automobile manufacturers do have some tried and true statistics for estimating future demand for their product. The most important is probably the average age of the fleet currently on the road. And it just set a new record. The idea that this would instantly translate into sales is not a given. If you've been watching this industry over the last few years you might have noticed that quite a few plants dedicated to small sedan production have been shuttered for the very obvious reason that they couldn't give the damn things away and profit margins on them were low. CAFE standards be damned. If people will not buy them you can't force them to. Which brings us to now.
What exactly were auto manufacturers supposed to have foreseen? Overall demand or which particular models were going to be in demand? I ask because if you were going to use MRP to schedule production in a JIT environment then you have to guess right. If you issue a PO and order parts you are going to get them barring unforeseen supply chain problems. And you had better need them otherwise you are going to pay a lot of money to warehouse stuff you don't need. Manufacturers have been destroyed by guessing wrong.
While I think the idea of slowing fleet purchases is particularly good, and any leverage they have over the semiconductor sector should be used, I wish everyone would stop playing amateur mass psychologists. Is the Biden administration running around and "doing stuff" about inflation like zoning reform or licensing reform that have no hope of having any effect on inflation in any reasonable time frame, if at all since these are state issues, going to look reassuring or panicked? IDK.
Or will dropping steel tariffs, which since it's not the binding constraint on car production probably will not decrease prices, and Biden is clearly resistant to doing, look like they're capitulating because of inflation so you should worry about it too or a savvy move bowing to reality? IDK, but neither does Matt. My conditional quantified prediction would 60% chance anything not clearly effective would be read as panic by the political press.
You give a lot of good reasons that *in December*, you didn't expect ARP to be as large as it was, which was why you underrated the risk of inflation. But in February, you said it was sized appropriately: https://www.slowboring.com/p/full-employment
"It is unlikely that we will see high inflation (above 3%, say) this year" (I'm not sure if you meant YOY here)
These things aren't necessarily in contradiction with one another, but I just want to understand what the through line is. Is it that even *if* you foresaw inflation would be higher, you would still have supported the same $1.8T? Do you still think concerns about the mismatch with the output gap were wrong? Is the answer to both of these related to *when* the $1400 got paid out?
GOP 2016: If you don't do something about it, you're going to have taco trucks on every corner.
GOP 2021: The cost of burritos is too damn high!
I have enormous respect, trust, and admiration for anyone willing to admit they were wrong. I have to see these kinds of admissions before I can fully listen to someone with an open mind and heart.
Let’s make this the norm and not the exception.
The honesty and accountability are very much appreciated. It's why I want to pay for the takes. We can learn and calibrate by proxy.
Attributing current inflation to the ARP is dubious. Very little of that money has been spent yet. Furthermore, the ARP is about 40% as big as the Trump eta stimulus measures. Most of the inflation that occurred last month resulted from pent up demand from 2020 and had little to do with the Georgia runoffs. That being said, turning supplemental unemployment into a reemployment bonus is an excellent idea that moderates should be able to find the votes to pass. Why aren’t Manchin and Sinema and Murkowski ramming that one through? It makes no sense.
I took a weekend trip to Boston for a family event and hotel prices were astoundingly low for that market. But when I went to rent a car the rates were outrageous. So I got a room very close to a relative and took Uber to and from the airport. So my anecdotal evidence is that the pandemic impacts are very disparate but rather extreme. This sort of volatility can't be good. Oh, and the plane flight both ways was completely booked so airlines have some capacity adjustments to make. Or not.
I have been working on a large capital project for my company, and since about January we have just been getting hammered by commodity prices, especially steel. Starting in February, we began getting hammered on the services side, with various contracting services (electrical, mechanical, etc.) suddenly coming in with much higher bids and longer lead times.
Last month both of those trends stopped getting worse for the first time. I am seeing futures prices in things like lumber and steel generally starting to cool off, so I think much of the durables squeeze will have resolved itself by the fall, with things like cars probably cooling off by next year.
My mid-brain take is it's actually good that maybe the boom didn't start immediately. Voters have really short memories, so maybe delaying the boom, even just a little, until a little further into this Congressional terms will keep it just a bit more front of mind when mid-terms come. I think voters are still willing to give economic issues some benefit of the doubt as we emerge from the pandemic, so I don't think Democrats are going to pay much of a price for this stuff if it indeed does cool off by Q4.
Nice to see Matt grapple with this. I found this a useful article in a service journalism way: I have a hard time wrapping my head around the mechanisms of inflation for the most part, so nice to see perspectives on the why of it all.
Some very good stuff in this post. Here are some things I would add:
(1) Obviously, the biggest problem with the forecast was to frame it in terms of y-o-y inflation, which missed the huge impact of the "base effect," that is, the fact that last year's price level was abnormally low. To see how misleading it is to quote y-o-y inflation now, imagine that you had almost died of hypothermia after a skiing accident. The next morning you wake up in a hospital room. The doctor looks at your chart and says "Wow! This guy's temperature has gone up 7 degrees in 24 hours! Better throw him in an ice bath quick!"
(2) People should pay more attention to measures of long term inflation. The best, which no one but the wonkiest wonks even know about, is the NY Fed's Underlying Inflation Gauge.
( https://www.newyorkfed.org/research/policy/underlying-inflation-gauge ) It hasn't gone over 3 percent yet, although it looks like it might before the end of the year.
(3) Matt is definitely on the right track when he says we need to focus on how much prices are driven by psychology and how much by temporary scarcities. Back in the 1960s and 70s, when we had real inflation in this country, psychology played a big role. A key psychological component of economic models is inflation expectations. Back then, the thing driving inflation was the fact that people projected current levels of inflation forward, expecting next year's inflation to equal this year's, or this year's plus an acceleration factor. That produced the "Shifting Phillips Curve" or "Accelerationist" phenomenon made famous by Milton Friedman. Eventually, tight monetary policy, at great cost, brought inflation expectations under control, leading to the long era of price stability. We don't want to get back into that. For the complete story with charts and data, see this: http://dolanecon.blogspot.com/2014/05/what-ever-happened-to-phillips-curve.html (Memo to self: Update this post.)
(4) Mentioning the Fed: Part of the high cost of getting rid of inflation last time was that the Fed had to do it alone. It would be much better if we could coordinate monetary and fiscal policy to fight inflation. That should be done not so much through spending austerity as through flexible taxes that increase when inflation threatens and ease off when it does not. That is one thing the MMT crowd are right about. (It is too bad the cultishness and weirdness of many in the MMT crowd discredit a few points on which they think soundly.)
Love your analysis generally, but the main reason you missed on inflation is because you like the stimulus. Same reason a conservative would miss on any negative effects of a tax cut. There are no free lunches.
When talking about weird behaviors with deficits, inflation, etc., the weirdness is always attached to some big event, like the Depression or WWII or the oil embargo. Yet here we are just barely starting to see the light at the end of a global pandemic and everyone is freaking out over some weird inflation numbers; weird because inflation was low for a long, long time and suddenly it's rising.
Political opportunism notwithstanding, wouldn't it be even weirder if there wasn't any weirdness in the economic data right now? Matt specifically predicted lower inflation than is happening right now, but that was, in effect, a bet that this huge, unforeseen random event wouldn't have a huge impact, which was a risky bet. Also, if weirdness continues and inflation suddenly drops in the fall, will Matt's prediction count as correct? (I cannot remember if it was time-bound.)
It seems obvious in retrospect, but how many people correctly guessed the post-war economic bonanza of the 1950's?
Another compounding impact on the car makers problems with semi conductors is that the node sizes they typically used in thwir vehicles were quite ancient technology-wise for the fabs, and that once they pulled themselves out of line and the fabs easily replaced their orders (which utilized smaller more modern - and cheaper to produce - node sizes that most consumer electronics rely on) it actually started to cost the fabs money to bump up production of the outdated nodes for the car companies.
Additionally, some car companies (namely Tesla) mitigated some of these impacts by reprogramming their vehicles' firmware to utilize different, more readily available, ptofessor types. Tesla can do this because, more than Ford or Toyota, they are basically a modern software company. But this didn't remove 100% of the supply chain impacts.
So the cart companies inability to upgrade their products to more modern / efficient components (like the rest of the consumer electronics industry had been) also hurt them, because they put additionally costs on fabs in the form of leaving money on the table to work with them once supply was maxing out fab production.
Your effort to predict specific things with actual percentage forecasts is really cool, and I really respect owning up to your miss. I really hope it catches on. Since getting into forecasting as a hobby last year it's become even more frustrating that pundits talk about what they think will happen in loose and vague terms. There's a better way!
For what it's worth those of us who forecasted on a similar question on Metaculus really missed the high inflation too. (The question was intended to be the same as what you were predicting on but mistakenly was about annualized month over month instead of month over 12 months ago, so it's not a 1 to 1 comparison but we clearly were off on how high it would be). https://www.metaculus.com/questions/6645/highest-us-core-cpi-growth-in-2021/
The rest of the "Matthew Yglesias Predicts 2021" series (https://www.metaculus.com/questions/6634/matthew-yglesias-predicts-2021/) running on there has a lot of interesting questions so it'll be cool to see how that plays out. Since one of the tenets of forecasting is updating frequently I'd be interested to see you revisit your forecasts and see which ones have changed.
"I’m also not hearing people say things like “better go buy that table now, because the price will only get higher in six months.”"
Ok, I'll say it then. I looked at the inflation numbers and decided to run off and buy some durable goods --- and am planning some more purchases. I do think that the most likely outcome is that prices drop some in 6 months as supply catches up, but I also see a long tail where inflation gets a little wild. I've got more cash sitting in my bank account than I need, and my dishwasher and air conditioners already suck, so I bought replacements. That said, they've sucked for a long time, and I think that inflation fears pushed me over the edge into replacing them.
Like most journalists, when Kevin Drum uses jargon he does not not understand (MRP) and does so in an an offhand way intended to suggest he does he is wrong. He very clearly does not understand what that means and what it is used for. Furthermore the idea that one simply does an economic forecast with their abysmal record of getting things right and scores a hit every time is extremely foolish. Parenthetically ask a journalist what Planned Obsolescence means and they have about a 100 percent record of getting it wrong.
Automobile manufacturers do have some tried and true statistics for estimating future demand for their product. The most important is probably the average age of the fleet currently on the road. And it just set a new record. The idea that this would instantly translate into sales is not a given. If you've been watching this industry over the last few years you might have noticed that quite a few plants dedicated to small sedan production have been shuttered for the very obvious reason that they couldn't give the damn things away and profit margins on them were low. CAFE standards be damned. If people will not buy them you can't force them to. Which brings us to now.
What exactly were auto manufacturers supposed to have foreseen? Overall demand or which particular models were going to be in demand? I ask because if you were going to use MRP to schedule production in a JIT environment then you have to guess right. If you issue a PO and order parts you are going to get them barring unforeseen supply chain problems. And you had better need them otherwise you are going to pay a lot of money to warehouse stuff you don't need. Manufacturers have been destroyed by guessing wrong.
While I think the idea of slowing fleet purchases is particularly good, and any leverage they have over the semiconductor sector should be used, I wish everyone would stop playing amateur mass psychologists. Is the Biden administration running around and "doing stuff" about inflation like zoning reform or licensing reform that have no hope of having any effect on inflation in any reasonable time frame, if at all since these are state issues, going to look reassuring or panicked? IDK.
Or will dropping steel tariffs, which since it's not the binding constraint on car production probably will not decrease prices, and Biden is clearly resistant to doing, look like they're capitulating because of inflation so you should worry about it too or a savvy move bowing to reality? IDK, but neither does Matt. My conditional quantified prediction would 60% chance anything not clearly effective would be read as panic by the political press.
You give a lot of good reasons that *in December*, you didn't expect ARP to be as large as it was, which was why you underrated the risk of inflation. But in February, you said it was sized appropriately: https://www.slowboring.com/p/full-employment
And even after it passed in March, you said inflation would be mild: https://www.slowboring.com/p/inflation-coming
"It is unlikely that we will see high inflation (above 3%, say) this year" (I'm not sure if you meant YOY here)
These things aren't necessarily in contradiction with one another, but I just want to understand what the through line is. Is it that even *if* you foresaw inflation would be higher, you would still have supported the same $1.8T? Do you still think concerns about the mismatch with the output gap were wrong? Is the answer to both of these related to *when* the $1400 got paid out?