In several recent posts, Tyler Cowen has stressed the need for better models of inflation. In one case, he expressed exasperation at my claim that (price) inflation is an almost meaningless concept:
4b. More seriously, Scott seems to dismiss the price level concept altogether. For instance he once wrote: “In the past, I’ve frequently argued that inflation is an almost meaningless and useless concept. I’m not even aware of any coherent definitions of the concept.” I don’t think this is a defensible point of view, and you have to compare Scott’s criticisms of the o1 model to his own approach, which is fairly nihilistic. And I think wrong. If inflation were higher and someone offered Scott an inflation-indexed contract to sign, would he be unable to evaluate such a transaction? Obviously not.
Yes, there’s some hyperbole in the phrase “almost meaningless”. But I suspect there’s much less exaggeration than most economists would assume. I’ll present my case with an example and then discuss Keynes’s view on the subject, which I believe is more accurate than either my previously expressed view or Tyler’s view. Then I’ll discuss China’s economy, an area where I seem to view the price level as important, but most other economists “dismiss the price level concept altogether”. No one will come out looking very good (except Keynes.)
What led me to such an overheated claim about inflation being almost meaningless? It would help to look inside the “sausage factory” and see what’s going on when the government estimates inflation. The more I look at official government estimates of TV inflation, for instance, the more skeptical I become about the entire process:
According to the U.S. Bureau of Labor Statistics, prices for televisions are 99.15% lower in 2024 versus 1960 (a $495.77 difference in value).
Between 1960 and 2024: Televisions experienced an average inflation rate of -7.18% per year. This rate of change indicates significant deflation. In other words, televisions costing $500 in the year 1960 would cost $4.23 in 2024 for an equivalent purchase. Compared to the overall inflation rate of 3.76% during this same period, inflation for televisions was significantly lower.
To me, that estimate doesn’t just seem wrong, it seems borderline insane. And that’s despite the fact that I’m probably in the top 1% of snobs who really care about picture quality. A few years back, I paid thousands of dollars extra to get a 77-inch OLED TV. Yes, in a technical sense modern sets are much better. But more that 100 times better? Please define the term ‘better’.
If you pressed an economist, they’d probably say “better” means more utility. Fine, but what utility measuring device determined that viewers derive 100 times more utility from a modern TV? In 1960, I was five years old. I don’t recall picture quality having much effect on how hard I laughed while watching I Love Lucy. In what meaningful sense is a modern TV 100 times better?
Economists obsess over whether the CPI or the PCE is closer to the “true rate of inflation”. But how can there be a true rate of inflation if economists cannot even precisely define what they mean by “better”?
If TVs were the only good, I’d stand by my claim that government inflation estimates are “almost meaningless”. But they are not the only good. And I would have to concede that inflation estimates for a gallon of gasoline or a dozen eggs are far from meaningless. The overall CPI is a hodgepodge composite of meaningless and meaningful data points, all mixed together.
Here’s Keynes in the General Theory, discussing the question of whether inflation data is meaningful:
But the proper place for such things as net real output and the general level of prices lies within the field of historical and statistical description, and their purpose should be to satisfy historical or social curiosity, a purpose for which perfect precision — such as our causal analysis requires, whether or not our knowledge of the actual values of the relevant quantities is complete or exact — is neither usual nor necessary. To say that net output to-day is greater, but the price-level lower, than ten years ago or one year ago, is a proposition of a similar character to the statement that Queen Victoria was a better queen but not a happier woman than Queen Elizabeth — a proposition not without meaning and not without interest, but unsuitable as material for the differential calculus. Our precision will be a mock precision if we try to use such partly vague and non-quantitative concepts as the basis of a quantitative analysis. . . .
In dealing with the theory of employment I propose, therefore, to make use of only two fundamental units of quantity, namely, quantities of money-value and quantities of employment. . . . We shall call the unit in which the quantity of employment is measured the labour-unit; and the money-wage of the labour-unit we shall call the wage-unit. . . .
It is my belief that much unnecessary perplexity can be avoided if we limit ourselves strictly to the two units, money and labour, when we are dealing with the behaviour of the economic system as a whole; reserving the use of units of particular outputs and equipments to the occasions when we are analysing the output of individual firms or industries in isolation; and the use of vague concepts, such as the quantity of output as a whole, the quantity of capital equipment as a whole and the general level of prices, to the occasions when we are attempting some historical comparison which is within certain (perhaps fairly wide) limits avowedly unprecise and approximate.
In general, I find the General Theory to be wildly overrated. Of course it’s got some good stuff, as Keynes was brilliant. But overall it is a far less useful guide to macroeconomics than is the earlier Tract on Monetary Reform.
The preceding quotation, however, is a very insightful observation. Keynes was right; fuzzy concepts like the price level can be useful for some purposes, but are inadequate for more rigorous scientific investigations. And whereas price inflation is not very useful, wage inflation should be a central concept in any macroeconomic model.
On the other hand, while inflation is a fuzzy concept, it is obviously not a meaningless observation to say that Venezuela’s nominal GDP growth overstates its real GDP growth due to a fast rising price level. We do have some rough but reasonable estimates of price inflation that can help to illuminate comparisons between time periods, or between countries.
Consider my frequent claims that China has the world’s largest economy. That statement only makes sense if you compare the US and Chinese economies in real terms. In nominal terms, the US has the largest economy. So in that sense, I’m a bit of a hypocrite.
When Tyler says that I “dismiss the price level concept altogether”, a reader might be forgiven for assuming that I hold some fringe views outside the mainstream. So I decided to google “world’s second largest economy”, to see what I got. At the top of the list was AI overview:
There followed a long list of links that mentioned China, not the US (which is the actual second largest economy.) And yet the claim that China is second only makes sense if one “dismisses the price level concept altogether.” There is simply no plausible estimate of US and Chinese price levels that would have China in any position other than world’s largest economy.
So let’s compare the views of Keynes with the views of mainstream economists:
1. Both Keynes and I believe that wage inflation and employment are the two key macroeconomic variables. While price inflation is not completely useless, its marginal value is almost zero, once you have accounted for wage inflation.
2. Recessions occur when aggregate demand falls relative to nominal wage rates.
3. The price level may be of interest to people making very general comparisons about the relative size of economies, or when estimating the change in living standards over very long periods of time, but should not be treated as if they were precise scientific concepts.
4. The original Phillips Curve utilized wage inflation. I am almost certain that Keynes would have shared my view that the later shift to price inflation was a mistake.
To summarize, economists tend to use price inflation in places where it is not appropriate–where wage inflation would be far more useful. Even worse, they often “dismiss the price level concept altogether” when considering exactly the sort of broad generalizations where price level adjustments would be highly appropriate, such as the question of whether the US or China has the world’s largest economy.
And don’t try to argue that when discussing “the economy”, the AI Overview assumed we meant “nominal economy”. I am quite confident that if you asked any AI a question about recent US economic growth, they would cite data for real GDP, not nominal GDP. That’s also true of the media. “The economy” seems to mean real GDP when discussing the business cycle, but it suddenly means nominal GDP when people wish to show the supremacy of the US economy.
PS. Off topic: Happy birthday to my stepfather Maxwell Freeman, who turned 100 today. Max earned two Purple Hearts fighting in places like Leyte and Okinawa during WWII. He is still going strong.
READER COMMENTS
Jim Glass
Dec 20 2024 at 4:19am
I think there are more dramatic examples than TVs. I’ve seen computer people say the technology in today’s smart phones would have cost $200,000,000 million in the 1990s — equivalent computing power was being used to design nuclear weapons — *if you could get it*, which you of course you couldn’t as all the tech and services packed into your little smart phone didn’t exist. And what parts did exist, if you could get them, would fill rooms.
(Some of us remember the big part of the military budget spent on the Cold War’s top secret spy satellites of the 1980s. Now you can see the dwindling number of tanks the Russian have in their depots looking at your phone in almost real time. A while ago I saw a Cold War documentary in which a retired Soviet general talked about their spy satellite program, looked at his phone, and laughed. But I digress…)
I agree with you and JMK. It’s not just absurd price adjustments, but even more that over not many decades the goods in one year’s “pricing basket” simply don’t exist in another’s. How does one price adjust for that? OTOH, for practical purposes short-term calculations are significant to prevent future 1930s, 1974s, and 2008s.
Of relevant interest: Some economists have tacked these issues. MeasuringWorth.com gives several different ways to measure relative values over long periods with calculators functioning back to 1790. Some of their articles are interesting. E.g.: “The value of a Model T in today’s dollars” by various measures. I’ve cited “Measuring Slavery in 2020 Dollars” here previously.
Jim Glass
Dec 20 2024 at 4:25am
“$200,000,000 million” seems a bit much. More like $200 million.
Anonymous
Dec 27 2024 at 12:09pm
Keep in mind that today you can easily get a TV (second-hand) for free which is way better than you could get in 1950 or whatever, so in some sense TVs are infinitely less expensive.
Arqiduka
Dec 20 2024 at 5:01am
But measuring inflation should have nothing to do with estimating how much “better” newer models are. You track the price of the same model in time, as far as you can, and when that one stops selling, you track the price of the next. As long as there’s any overlap between the two, inflation can be meaningfully measured with no recourse to how much better or worse the new model is.
Scott Sumner
Dec 20 2024 at 12:01pm
No, that technique does not work, and it’s easy to see why. When a new TV comes out at a very high price, it’s bought by the rich, for whom price in no object. Then the price falls dramatically, say from $10,000 to $1,000. Then a very slightly better TV comes out and is sold to the rich for $10,000, and its price quickly falls to $1000. Rinse and repeat.
Using that technique you’d have even more extreme TV price deflation, say from $500 to just a few cents. It won’t work.
Arqiduka
Dec 20 2024 at 2:34pm
Nah, you forget to account for quantity.
The rich toy sells a few units and thus barely moves even the TVs entry, let alone the CPI. The mass sales are what move the needle, and here is where you see the 8 or so percent loss year on year.
Scott Sumner
Dec 20 2024 at 11:54pm
I strongly disagree. That would lead to absurd estimated rates of TV price deflation. The fact is, people spend similar amounts on TVs as they did many decades ago. The claim that TV prices have fallen more than 99% must be based on the assumption that the new sets are “better”, or the claim means nothing at all.
Arqiduka
Dec 21 2024 at 1:08am
This algorithm obviously wouldn’t produce 99% deflation rates, I don’t see how that could even be entertained. I don’t understand this comment.
Arqiduka
Dec 21 2024 at 1:25am
Oh, you meant compounded over time. What can I say, such is the magic of chain linkage.
No, you don’t need to make any quality assumptions at all. Indeed, you can construe an example where quality falls (say, due to loss of knowledge as people retire) but the prices of what is sold still fall YoY, and you would still get secular deflation.
Scott Sumner
Dec 21 2024 at 9:39am
I still don’t get it. Someone buys a TV for $500 in 1960. Another person buys one for $500 today. The BLS says the new set costs only 1% as much as the old one. If that’s not a claim that the new set is 100 times better, what is it? What’s the point of measuring the price level?
Arqiduka
Dec 21 2024 at 2:37pm
I try to picture it like so.
At some point in the past, the price of a then-modern TV set was falling year on year, and may have continued to if left to its devices. Yet, when a new model became available, the market judged it better enough to pay more for it. So much, that the older, far cheaper model got discontinued. Rinse and repeat.
So, this is an indirect claim that the quality has been going up, but not by the BLS, but by the market. If, somehow, we had a 60s TV on sale at 5 bucks, people still wouldn’t buy it in bulk.
Dylan
Dec 22 2024 at 12:15pm
I don’t get it. The first television I bought on my own was $800 in 2002 on a Black Friday sale. It was 24″. A couple of months ago I bought a 40″ TV not on any particular sale, and it cost me $79.*
Let’s translate that into beer. In 2002, a 6-pack of good beer costs around $5. Today, it’s more like $18. So, in 2002 that TV costs me 160 6-packs of beer. The TV I just bought was about 4 of them, a decrease of ~98% in just 22 years.
*I think you’d be right to point out that today’s TV’s are monetized outside the purchase price since TVs spy on you and sell the data to advertisers.
John Hall
Dec 20 2024 at 9:50am
Even something like wages has complications in that you aren’t tracking the same cohort of workers or composition of industries over time. (hence the ECI: https://www.bls.gov/eci/)
Something like labor compensation per worker has very similar properties as wages (and likely susceptible to the same problems), but maybe helps remember that the focus of the analysis is on the aggregate.
Alan Goldhammer
Dec 20 2024 at 10:44am
Is not the price of groceries the best measure of inflation? One needs to eat, one does not need to watch TV.
Scott Sumner
Dec 20 2024 at 12:03pm
No, inflation is the average change in all goods prices, not just those you regard as necessities. There’s no reason to exclude luxury goods (and most people don’t even regard TVs as a luxury.)
Vaidas Urba
Dec 20 2024 at 1:15pm
Tyler Cowen used an unbiased prompt for ChatGPT – and guess what? Market monetarism is the favorite of AI: https://chatgpt.com/share/675f5f89-50dc-8010-9f8a-a303eb9ee7f8
Scott Sumner
Dec 20 2024 at 11:58pm
I must be doing something wrong, as that link doesn’t work for me. It is cut off at the bottom. Any chance you can copy it to an email and send it to me? Thanks.
Vaidas Urba
Dec 21 2024 at 4:24am
I forwarded it to your Bentley address – it is the only I have
Scott Sumner
Dec 21 2024 at 9:42am
Thanks, Very helpful. I didn’t see him literally say MM is best, but it was a very good response. Tyler’s right that the o1 Pro is giving more sophisticated analysis than the earlier AIs. I probably should do another post.
Would I have to be a paid subscriber to have gotten that complete report that you sent me? Just wondering what I did wrong.
Henri Hein
Dec 20 2024 at 3:59pm
I was glad to read this, as I’ve long considered inflation problematic, if not as a concept, at least as a measure. I’m skeptical when I see econometrics where they adjust for inflation across long time horizons, like multiple decades. I get that it can be potentially illuminating to estimate the relative purchasing power across time, but if you use the wrong measure of inflation, the potential to confuse seems equally potent, and I don’t know how you can be sure of the inflation rates over that long a time period.
What do you think of the Measuring Worth project? I don’t have an opinion myself, just genuinely curious what you think about it.
Scott Sumner
Dec 21 2024 at 12:02am
It seems like they are relying on government estimates.
jose pablo
Dec 20 2024 at 5:52pm
such as the question of whether the US or China has the world’s largest economy.
I wonder what any possible interest in that question could be. Except, maybe, as an example of the “useless questions” category.
Scott Sumner
Dec 21 2024 at 12:03am
I’d say it’s similar to comparing Queen Victoria and Queen Elizabeth—fun to debate, but otherwise not important.
Jim Glass
Dec 20 2024 at 11:51pm
If “inflation” has only dubious meaning, if any, over the long run, then remember this has an effect on GDP data too, as real GDP is calculated with an inflation adjustment. So how real has been real growth of GDP from 1960 until today?
The Measuringworth people say one should take care to select the particular value measure that meets one’s analytical needs. One could say this about GDP itself. Real, nominal, dollar terms, PPP? Each is best for different purposes, it’s not one rules over all.
And more: Does China still have a developing economy, or now an advanced economy? It’s GDP per capita suggests developing. Past Premier Li Keqiang’s report of 600 million Chinese living on less than $141 a month certainly suggests developing. OTOH, its world’s largest economy, world’s largest export sector — growing now at 12% annually compared to world export growth of 3% — and world-leading position in so many tech sectors, its space program and all the rest, certainly suggest an advanced economy. The CCP’s ability to draw on the resources of the world’s largest economy to organize state action (surging military, growing nuclear arsenal, space programs, etc.) exceeds that of most advanced economies. So what’s the definition of “advanced”? Why?
If we say China still has a developing economy, what’s the purpose? Is it to preserve China’s current eligibility for special economic benefits such as “longer timeframes to meet trade obligations within international agreements, lower annual budget assessments at the UN, access to preferential loans from institutions like the World Bank [and more]” (as explained to me by my calendar-challenged Google AI), as per the same metrics used to determine the status of other countries as “developing”? OK. But do other such countries invest the value of such favored-case benefits at home to develop, or instead spend them abroad as China as has spent $1 trillion (so far) on the Belt & Road, plus on military growth such as China does to enable claiming the South China Sea, bully the Philippines, and be ready to fight the USA in a war over Taiwan maybe by 2027? Does that spending matter? What’s the purpose of using the word?
When I was a kid there was an endless argument over who was the greatest center fielder, Mickey Mantle or Willie Mays. That dispute was settled when Bill James, the former econ student who pretty much invented modern sports analytics, showed that Mantle had the highest peak value and Mays the greatest career value. So each one could rightly be declared the best, depending. What was the greatest NFL team of all time? Many say the 17-0 1972 Dolphins with the only perfect record ever, against an easy-ish schedule, while many analytics types say the 1962 Packers who were a mere 14-1 against a tougher schedule. Though, of course, any good modern college squad would crush both those teams by running its 300+ pound linemen over their 240 pound guys. Which means both those teams would have been annihilated by *any* NFL team of the last couple generations. (Except the Jets). So which is the greatest NFL team of all time? I don’t know. It depends. And picking which sports team is better than an other is easy.
Scott Sumner
Dec 21 2024 at 12:06am
China is clearly a developing country, unless you think India is more advanced than Switzerland because it has a space program and a bigger GDP.
Jim Glass
Dec 22 2024 at 12:28am
Why does it matter? With 600 million people living on $140 a month, I don’t disagree. But what’s the point of saying so? I’m not arguing, I’m asking.
E.g., The advanced countries of the EU are screeching and organizing a defensive trade war because China is hammering their advanced tech industries with its advanced tech exports while also supporting a real shooting war on their eastern flank that they see as a genuine threat to their security. When they see the world’s largest economy doing all this to them, they don’t perceive it as “not developed”. Perhaps very reasonably. The major parts of its economy that are hitting them are *plenty* advanced.
So when you tell them it is a “developing economy”, what’s the gain? The benefit of doing so?
China has 5x the GDP per capita of India. And India isn’t hammering the advanced economies of the world with anything but scam phone calls.
Todd Ramsey
Dec 21 2024 at 9:26am
Please tell Mr. Freeman “Thank You” on behalf of my family and me.
Scott Sumner
Dec 21 2024 at 10:01am
Thanks, I will do so.
Andrew_FL
Dec 21 2024 at 11:20am
Google has defaulted for years to giving GDP comparisons in exchange rate adjusted dollars, so I wouldn’t be so sure Google’s AI would give you the answer you expect it to if you’re not very specific about what you’re asking.
Jose Pablo
Dec 22 2024 at 7:03pm
my claim that (price) inflation is an almost meaningless concept:
And yet, having an answer to the question: “How much purchasing power is my capital in US$ going to lose next year?” is a very relevant one.
However, from this perspective changes in relative prices* are much more interesting than “inflation”. The question “How much purchasing power is the retribution of my job market skills going to lose (win) next year?” is an even more relevant one.
* Including (in particular) the comparison between interest rates and inflation.
Jim Glass
Dec 23 2024 at 12:26am
[Henri Hein: My 4th attempt to reply. As hitting ‘reply’ hasn’t worked I try here. The penalty is each attempt gets longer. (Jim, please see our Econlib note below. The actual penalty would be that your comments don’t get published at all if you don’t abide by our simple comment policies, like not using obscenities.–Econlib Ed.) ]
For one thing, it right away provides a much broader perspective on history than one gets from the automatic default ‘everything in CPI adjusted dollars’. For instance…
George Washington’s 1789 salary of $25,000 as our first President CPI-adjusts to $826,000 in 2022 dollars. (Biden’s salary is $400,000.). But, as we all ask, what does that really mean? Well, relative to GDP per capita George’s $25,000 was equivalent to $39,300,000 in 2022. And relative to share of GDP it was $3,370,000,000. (Plus he had a famous expense account too.) That doesn’t mean he was paid $3 billion in today’s money. It does mean that his relative position in the teensy, poor economy of the time was that of an oligarch. It also suggests that for all the current woe about ever-rising “inequality”, the oligarchs of today are a lot poorer relative to average people than they were then. Measuring Slavery in 2020 dollars gives a lot of context about what was going on with the Civil War that answered questions we still hear all the time, such as “Could the war have been avoided by buying out the slaveowners like the British did?” (No way!) As to the Civil War itself, historians say it had…
And as to more current inflation issues …
[] Babe Ruth’s $80,000 salary in 1930 CPI-adjusts to a $1.4 million today — a utility infielder’s compared to the current average MLB salary of $4.5+ million. But considering the growth of the economy (the market) by GDP it adjusts to $24+ million, comparable to today’s largest.
[] The Apollo Moon Landings. NASA put the cost at $23 billion 1972 dollars. By CPI that’s $190 2022 dollars “but this would not be a very good measure since the CPI does not reflect the cost of rockets and launch pads … By using the GDP per capita, we are measuring the cost in terms of average product and would get a number of $380 billion. Finally, a way to consider the “opportunity cost” to society, the best measure might be the cost as a percent of GDP, and that number would be $642 billion.”
[] Gasoline! Everybody’s hot topic when it goes up. The Price of Gasoline in 2021 Dollars annually since 1929, by seven different measures. (The people paying the lowest price complain the loudest. Ain’t it always so?)
And for the record, $500 spent on a TV is 1960 was in 2022 dollars worth: $4,940 by CPI, $6,510 relative to average wage, $12,900 relative to average income, and $24,000 relative to GDP. So that real lower price measured by CPI was the least of the declines.
[Jim: Try removing the curse words from your comments. I’ve edited your comment to remove the foul language word from this comment for you, and it sailed through. You’ve been a commenter on EconLog for a long time. We shouldn’t have to remind you that obscenities–in any form–are not allowed here. –Econlib Ed.]
Knut P. Heen
Dec 23 2024 at 8:59am
Isn’t long-run inflation meaningless? I heard Hayek (on some Youtube-video) say that Keynes had been worried about the power of British labor unions during the 1930s (they were beating up strike breakers), and that Keynes thought inflation was a smart way to reduce real wages (when labor unions made it impossible to reduce nominal wages). This strategy clearly requires short-term unanticipated inflation hikes to work. Hence, the test is to see if there is a relationship between short-term unanticipated inflation and unemployment. The quality of TV-sets do not vary that much in the short-term.
Jim Glass
Dec 23 2024 at 6:48pm
What curse words? I’ve NEVER used a curse word on this site.
Email me your copy of what you are talking about.
I HAVE NOTED BEFORE that there have been comments here under my name THAT I DID NOT WRITE. I never received a response from you about that.
And all those years – more than a decade — how many obscenities did I use?
Email, me please. Send me the offending text.
[I’ve sent you an email detailing the offending word precluded by our spam filter. I do not think you want to claim that the four comments you’ve submitted in quick succession were not sent by you. But they did each contain an obscenity–a commonly used stand-in for an obscenity with no possible other interpretation.–Econlib Ed.]
Comments are closed.