It is always surprising how even the financial press is confused about inflation. I am not speaking of complex conflicting theories on the causes of inflation but of the very basic distinction between inflation and changes in relative prices.
For example, The Economist just wrote:
America’s annual inflation rate stood at 5.3% in August, down slightly from the 13-year highs of the previous two months. There was some evidence that inflationary pressures may be levelling off, such as an easing of prices for used cars, which have driven some of this year’s inflation. [my emphasis]
An increase in the price of used cars cannot generate or “drive” inflation for a simple reason: it works the other way around. Inflation, which is defined as a continuous and sustained increase in the general price level, causes all individual prices (including wages, interest rates, etc.) to rise more or less proportionately, over and above the change in relative prices caused by changes in demand or supply on specific markets. To the extent that used car prices were partially caused by both inflation and a relative price change, they could not be a cause of inflation.
Suppose that a decrease in the supply of new cars causes an increase in their prices. This in turn boosts the demand for used cars, a substitute, and causes an increase in their prices of, say, 27%. The consumers who continue to buy cars at the higher prices must spend less on other goods and services, whose relative prices would decrease. Assuming no independently generated inflation (no increase in the general price level), that would be the end of the story: fewer cars and more of other goods or services are produced. Prices of cars have increased relative to the prices of other goods; only a change in relative prices has occurred. If there is inflation, though, the increase of used car prices will be augmented by the general rate of inflation in the economy, say, 5%.
To illustrate, take the Bureau of Labor Statistics (BLS) data for August. The BLS report on the Consumer Price Index (CPI) reveals a 5.3% increase in the index compared to August of last year. In other words, the BLS estimates that the general level of consumer prices in the economy grew by 5.3%. It also calculates that the prices of used cars and trucks (the available statistical category) increased by 31.9%. If the 5.3% represents a good estimate of inflation (in the price level), this means that the relative price of used cars and trucks increased by (roughly) 26.6% (= 31.9% – 5.3%), that is, relative to other goods in the economy. (I say “roughly” because an exact calculation would require excluding the increased price of used cars and trucks from the total increase of the CPI, not a large correction as used cars and trucks account for only 3.5% of the goods and services in the CPI.) The remaining 5.3% in the increase of used car and truck prices is what is due to inflation. It is nonsensical to say that the 26.6% “has driven some of this year’s inflation” of 5.3%, because the latter gets added to the former.
Look at it in another way. If T=R+L, it is nonsensical to say that R contributed to, or “exerts pressure” on, L. If the Total price change of used cars and trucks (32%, rounded) is equal to the Relative price change (27%) plus the change in the general price Level (5%), it is nonsensical to say that the relative price change (R) contributed to the change in the general level of prices (L).
The Economist is not the only media to be confused. For another example, see the Wall Street Journal’s related story, “Inflation Eased in August, Though Prices Stayed High,” September 14, 2021. God knows what the New York Times, the Washington Post, or Fox News must have said!
Steve Ambler, an economics professor at the University of Québec in Montréal, adds a point that may partly explain the confusion of non-economists (and perhaps some economists):
There’s a significant empirical correlation between the rate of inflation and relative price variability. … In an inflationary environment there’s likely going to be a few price increases that commentators will be able to point to that are in the right hand tail of the distribution. Then they conclude that those extreme price increases are “driving” inflation.
READER COMMENTS
David Seltzer
Sep 20 2021 at 10:23am
Pierre, Brilliant, clear explanation. It seems some, such as myself, don’t grasp the nuances of economics until they are explained this well. Thank you.
Pierre Lemieux
Sep 20 2021 at 10:27am
Thanks, David.
Craig
Sep 20 2021 at 11:27am
Aren’t they really just speaking for how the CPI is mathematically calculated? I mean, this past month, inflation cooled to .3 and in part the reason it cooled is because of a relative decrease in used car prices, ie any category where the prices are below the monthly reading are dampening the CPI, any category above tend to be pulling it up.
I understand the point that money spent on items above the CPI is money that can’t be spent on other things, no question, but those other things might not be measured by the CPI, ie money spent on bitcoin, stonks and other stuff apparently doesn’t count. Indeed, neither does money spent on homes, and indeed one might say that, mathematically, this method of measuring inflation, OER tends to dampen inflation since the OER is less than the increasing in housing prices.
Pierre Lemieux
Sep 20 2021 at 3:46pm
Craig: Good question in your first paragraph. I did not raise it in my post because I wanted to keep my argument simple (perhaps too simple?). You are right to note that the indirect measurement of an unobservable phenomenon (inflation) is not the phenomenon itself. The arithmetics of the CPI can lead somebody into thinking that inflation or deflation is higher or lower than it is because the index incorporates a mix of relative price changes and inflation. Consider the following. Case #1: If there is no inflation (or deflation), the CPI should change by 0%, because price decreases compensate for price increases (all relative prices). In reality, the CPI increase won’t be exactly 0% because some prices might be sticky (and not all prices are measured, etc.). Case #2: If there is inflation and the included relative price adjustments do net to 0%, the CPI will give a good measure of inflation (barring other problems in the index). But you still cannot conclude that if a relative price had been higher, inflation would have also been higher because the higher price increase would normally have been compensated by a higher price decrease elsewhere in the economy, and this should have been caught in the index.
On your second paragraph, the CPI by design measures the change in the prices of produced goods and services, not of financial claims on future goods and services. And “money spent on homes” is indeed included in rents and imputed rents on owner-occupied houses and apartments: “shelter” accounts for 32,6% of the August CPI.
John Hall
Sep 20 2021 at 1:11pm
People who read stories like these largely understand all this. CPI is a measurement of inflation, not inflation itself. What people debate is whether the increase in price X means prices Y,Z,etc. will go up in the future. It is rather difficult to say, so it is useful to know whether the increase in CPI driven by one category that may correct in the future. Similarly, more and more attention is paid to trimmed mean estimates of CPI (and PCE) or sticky/flexible measures of inflation, rather than just core CPI.
Pierre Lemieux
Sep 20 2021 at 4:04pm
John: My point is that a change in a relative price tells us nothing about future inflation. A relative price change does not even tell us how the inflation rate would have been different in the past had the relative price change not happened. My answer to Craig above gives a slightly more detailed explanation.
About your last sentence, even The Economist comes close to arguing that you can find an index, or trim something in an existing one, that will produce the inflation rate you want.
As you say so well,
Jose Pablo
Sep 20 2021 at 5:21pm
Very interesting Pierre and yet …
If used cars represent 3.5% of the price index and their prices go up by 32%, used cars are “causing” a 1.12% increase in the “general level of prices” (just by the way the “general level of prices” is “calculated”). Being this 21% of the total increase of the general level of prices, it could be acceptable to say that “used cars” are “driving” (no pun intended) the general level of prices.
If it is true, as you point out, that the “relative price increase” of used cars do not pressure inflation (by definition of “relative price increase”). But you can imagine a two goods economy: “used cars”, representing 3.5% of the total consumption and “other goods and services”, representing 96.5%. If used cars increase their price by 32% and “other goods and services” by 0%. Then the “inflation” would be 1.12% and you can understand somebody saying that the 1.12% increase was “driven” by the price increase of used cars and having a hard time admitting that, in reality, it was “the increase in the general level of prices” the cause behind part of the increase in used cars.
Craig
Sep 20 2021 at 10:25pm
I would tend to say that at minimum the recent news stories about, first, lumber, and then used cars might be in the public eye sufficient to stoke consumers’ expectations of inflation. For sure if, in anticipation of inflation, you buy a car today in fear that it will cost even more tomorrow, the money spent on the car can’t be spent on other things, but we live in a ‘bad credit, no problem’ economy where not having the actual money doesn’t seem to be a problem, you just need to have sufficient credit. Sometimes there are credit standards and other times all you need is a pulse.
Jose Pablo
Sep 21 2021 at 11:02am
Reading your comment (and keeping Pierre distinction in mind), I think that people don’t change their behavior when they anticipate “inflation” (maybe bond traders do, but not the layman). What changes people behavior is the anticipation of “changes in relative prices”: they buy a car if they think car prices are going up (regardless of what the general level of prices is doing) or when houses go up …
Not sure about the implications of that, if any. But it does seem that “inflation” and “change in relative prices” are the same in consumer minds. If they are not the same in the economist minds (when economist try to develop a model than explains consumer behavior) then, maybe, the economist are the ones getting it wrong.
[I have the same feeling with the Ricardian Equivalence, by the way. None of my friends, not even the economists among them, change their behavior the same when the government increase taxes that when the government increase their debt. Not at all. Tax increases hurt them for real. Debt increases just make them complain about government expending at the dinner table]
Thomas Lee Hutcheson
Sep 22 2021 at 6:25am
The question ought to be should a change in the supply or demand of some important item cause a change in monetary or fiscal policy? I think what is usually meant by saying that X “drives” inflation is that it should not.
Comments are closed.