![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
grok_mctanys) wrote2023-06-16 07:46 am
The way inflation is measured and reported makes no sense.
Measurement and reporting
Inflation is measured using a basket of goods:
Consumer price inflation is the rate at which the prices of goods and services bought by households rise or fall. Imagine a large "shopping basket" containing those goods and services. As the prices of the various items change over time so does the total cost of the basket.
Inflation (or deflation) is then the amount that the total price of that basket of goods goes up (or down) from month to month.
Fairly typical examples of reporting are:
Inflation set to ease...: The Office for National Statistics (ONS)
is set to reveal Consumer Prices Index (CPI) inflation dropping to 8.2% in April
from 10.1% in March
Inflation falls to lowest level...: The Office for National
Statistics said on Tuesday inflation fell to 8.7 per cent in April - down from
10.1 per cent in the previous month.
UK Inflation down...: The UK's inflation rate fell to 8.7% in
April, down from 10.1% in March, [...] Core inflation - which strips out food
and energy prices - rose to a 31-year high of 6.8% in April
What I thought inflation measured
Given the typical reporting of "The rate of inflation in April was 8.7%", I had always thought this meant that the amount prices had risen in April, i.e. from the start of April to the end of April⁰, was at a rate of 8.7% per year.
Although the rate of inflation is often described as being "over a year" or "for a year" or "in a year", that sounded to me that while the rate was being measured for a given month, it was just being expressed in a per-year unit. This made sense to me for a number of reasons.
- You don't have to have been travelling for an hour to describe your speed in miles per hour. If you want to calculate your (average) speed along a given stretch of road, you just need to measure how long that stretch of road is, and how long you took to travel it. You don't have to go further back than that. And there's nothing odd about expressing a rate in units of time that differ from how long the measurement took.
- Inflation expressed as percentage points per month would normally only have fractional values, which isn't as easy to communicate. We could measure speeds in miles per second rather than miles per hour, but then a speed limit of 30mph would come out as 0.0083mps, which is awkward to convey or put on speedometers and signposts. Using a unit where the values typically fall from 1 to 100 is easier to get across. (Inflation could be measured using per mille points or basis points per month, but then those units would probably also need to be explained, whereas most people already know how percentages work.)
- The interest rates for loans, mortgages, bonds, and other lines of credit/debt are also normally expressed as percentage points per year. Being able to compare inflation rates to interest rates, and to other annual monetary changes (e.g. salary changes) is incredibly useful.
- Months are different lengths. If a smooth constant rate of inflation caused a basket of goods to go up by 31p in January, the same smooth constant rate of inflation would cause that same basket of goods to only go up by 28p in February. Expressing just the amount of inflation in a given month would give inconsistent rates.
After all, Economics is, like, a science¹. It uses maths, and tables, and graphs, and equations. I'd expect it to use the word "rate" in the correct mathematical sense - the slope of the graph of measurement vs. time over the interval described.
Therefore, the "rate of inflation in April" surely should be most straightforwardly interpreted as "the amount that prices rose during the month of April", expressed in "percentage points per year".
What "the rate of inflation in April" actually measures
I recently came across a few bits of reporting where the way inflation was being being described didn't align with how I understood it. There might have been a number of reasons for this, one of which was that a reporter had either misunderstood inflation, or understood it but was communicating it poorly. If you've ever read an article about which you have first-hand knowledge or expertise in, you'll recognise how feasible this is. Another, of course, is that I might have misunderstood inflation.
However, given that these separate reports seemed to have got inflation wrong in exactly the same way, I thought it was worth double-checking my understanding.
From the ONS Consumer price inflation, UK: April 2023 glossary:
Annual inflation rate
The most common approach to measuring inflation is the 12-month or annual inflation rate, which compares prices for the latest month with the same month a year ago. In any given month, the annual rate is determined by the balance between upward and downward price movements of the range of goods and services included in the index.
So "the rate of inflation in April" (or, as it is more accurately but only occasionally reported, "the rate of inflation in the year to April") actually measures "the amount that prices rose between one April and the previous April", rather than "the amount that prices rose in April".
Huh.
I learned something. Great!
Now I have a better understanding of what the reported "rate of inflation" is, that's going to help me navigate the world more effectively, right? This is a good thing.
Except...
...the more I thought about what the reported rate of inflation actually measured, the less sense it makes to me.
A thought experiment
I'm going to give an example that plays out over three years.
The first year is calm. The price of the basket of goods rises at a steady "inflation" rate of around 2.4%/year, which is roughly what modern western governments target.
In the second year, a shock at the end of April causes prices to rise suddenly between May and July, before (due to some sound government policy intervention and strong consumer protection laws against greedflation profiteering(!)²) they stabilise and fall back down to near previous levels by the end of the year.
In the third year, prices go back to rising steadily at around the original target rate.
This is a work of fiction - any similarities to actual rates of inflation (living or dead) are purely coincidental.
| Yr/Mon | Price | Infl | Yr/Mon | Price | Infl | Yr/Mon | Price | Infl | ||
|---|---|---|---|---|---|---|---|---|---|---|
| (202)1/Jan | £100.00 | ? | (202)2/Jan | £102.40 | 2.40% | (202)3/Jan | £105.00 | 2.54% | ||
| (202)1/Feb | £100.20 | ? | (202)2/Feb | £102.60 | 2.40% | (202)3/Feb | £105.00 | 2.34% | ||
| (202)1/Mar | £100.40 | ? | (202)2/Mar | £102.80 | 2.39% | (202)3/Mar | £105.25 | 2.38% | ||
| (202)1/Apr | £100.60 | ? | (202)2/Apr | £103.00 | 2.39% | (202)3/Apr | £105.50 | 2.43% | ||
| (202)1/May | £100.80 | ? | (202)2/May | £106.00 | 5.16% | (202)3/May | £105.75 | -0.24% | ||
| (202)1/Jun | £101.00 | ? | (202)2/Jun | £111.00 | 9.90% | (202)3/Jun | £106.00 | -4.50% | ||
| (202)1/Jul | £101.20 | ? | (202)2/Jul | £112.00 | 10.67% | (202)3/Jul | £106.25 | -5.13% | ||
| (202)1/Aug | £101.40 | ? | (202)2/Aug | £111.00 | 9.47% | (202)3/Aug | £106.50 | -4.05% | ||
| (202)1/Sep | £101.60 | ? | (202)2/Sep | £112.00 | 10.24% | (202)3/Sep | £106.75 | -4.69% | ||
| (202)1/Oct | £101.80 | ? | (202)2/Oct | £110.00 | 8.06% | (202)3/Oct | £107.00 | -2.73% | ||
| (202)1/Nov | £102.00 | ? | (202)2/Nov | £108.00 | 5.88% | (202)3/Nov | £107.25 | -0.69% | ||
| (202)1/Dec | £102.20 | ? | (202)2/Dec | £106.00 | 3.72% | (202)3/Dec | £107.50 | 1.42% |
What is the relationship between price changes and "inflation"?
If prices fall, inflation can stay positive.
The UK Inflation down... article makes the claim: A slower rate of
inflation does not mean prices are coming down - it means they are not going up
so quickly.
It is not the only piece of reporting to do so.
I have heard similar statements by a number of other genuinely respectable
commentators in the past.
Those comments stood out to me previously, as they seemed so self-evidently,
blindingly obvious that I was surprised that the point needed to be expressed at
all.
The thing is, based on what "inflation" is actually measuring, it isn't necessarily true. I mean, it might be true, but it might not. Looking at the example from September of year 2 - prices fall every month for 4 straight months, yet "inflation" remains positive the whole time.
The reverse is true for the 7 months in year 3 where "inflation" is negative. Prices rise every month during that time.
There can be sudden shocks in inflation, even if everything is stable.
Looking again at the period in year 3 where "inflation" turns negative, it goes from a "healthy" +2.43% in April to suddenly diving to a shocking -5.13% in July! What caused this sudden, unexpected shift in "inflation" in year 3?
Nothing.
There was no change in the rate prices were rising between April and July in year 3. Prices were under control, and rising slowly and steadily the whole time. The apparent change in inflation in year 3 is actually entirely due to the sudden rise in prices in year 2. But because that sudden change in prices was not repeated in year 3, "inflation" goes completely wild.
There's no way to tell what inflation has done over, e.g. a 6-month period.
Even if you've got 10 years of worth of inflation numbers, there's no way to tell what inflation has done over any timescale other than a whole number of years.
Every April measurement only tells you about what inflation did since the previous April, and the April before that, and the April before that. Ditto for every October. There is no way of taking those numbers and figuring out how much prices have changed from any given April to the following October.
Why is inflation reported the way it is?
Seriously, why?
These inflation numbers gathered for the government and reported by the media only make sense in the right context. But not only is the right context not given, the wrong context is given instead.
Comparing the rate of inflation from month to month is pointless. If inflation in March is 10.1%, and inflation in April is 8.7%, that says nothing about what the price of goods did between March and April³. It only says something about what prices did between March and the previous March, and completely independently between April and the previous April.
It doesn't even tell anyone whether prices are currently rising or falling.
Furthermore, none of these issues would occur if inflation actually measured what I originally thought it did.
The only possible justification I can think of for reporting inflation this way is to account for periodic seasonal pricing variations. For example, if prices fall every January due to retailers always having sales then, or due to wide scale post New Year belt-tightening leading to lower consumer demand and consequently a drop in prices, then inflation could look like it's doing something weird in January when everything is actually fairly normal, historically speaking. Measuring inflation from January to January and February to February makes that kind of variation irrelevant.
But, the ONS already knows how to calculate seasonal variations in prices and account for it - because it already does that for other statistics it gathers, including other types of inflation. They could easily publish a "seasonally adjusted" measure of what inflation was really doing in a given month.
So, why is inflation measured and reported the way that it is? What is anyone supposed to do with that information? What use is it, like, at all?
Answers on a postcard...
⁰ Actually, more like from the start of April to the start of May, in order to avoid a fencepost error.
¹ Sorry, that's meant to say that Economics is like a science (but not actually one).
That's unfair.
The trouble is, Economics appears to desperately want to be a "hard" science
with reductionist theories and precise mathematical laws, evidenced by its
models of spherical utility-maximising rational economic actors in a vacuum.
But because humans are highly chaotic, being both stubbornly contrarian and
also vulnerable to groupthink, this approach doesn't really appear to work.
See the well-known Economic aphorisms that markets can remain irrational a lot
longer than you and I can remain solvent
⁴ and Wall Street
indexes [sic] predicted nine out of the last five recessions!
⁵.
Nevertheless, Economists seem to keep using the language of precise mathematical
models anyway.
² You know, like competent governments who work on behalf of voters against the companies who would exploit or endanger them.
³ Although, unless prices fell by more than 1.4% in the previous March to April, an inflation reduction from 10.1% to 8.7% means that prices will have fallen between the current March to April. Also note, that's unless they fell by more than 1.4% in a single month, not that unless they fell at a rate of more than 1.4%/year. So, yeah, prices probably did actually fall.
⁴ A. Gary Schilling, 1986:
markets can remain irrational a lot longer than you and I can remain solvent.
⁵ Paul Samuelson, NewsWeek, September 19, 1966, p. 92:
To prove that Wall Street is an early omen of movements still to come in GNP, commentators quote economic studies alleging that market downturns predicted four out of the last five recessions. That is an understatement. Wall Street indexes predicted nine out of the last five recessions!
Note that Wall Street indices don't actually predict recessions. Rather, commentators and the authors of economic studies (mis)interpret the movement of those indices to predict recessions.
no subject
But a possibly even bigger issue is *what items are in that basket and who decides they are representative!* Certainly not someone living on the breadline, and relying on supermarket own brand and other bottom of range choices, for whom those prices matter a hecking lot more.