INFLATION FOLLOW-UP

Usually I let my writing sit a day before I post it, so I have time to check it over, and to allow time for OH NO, THAT'S TOTALLY WRONG to pop into my head -- some little thing that I forgot, maybe, that makes all the difference

Didn't do that with yesterday's "Inflation show-and-tell" because I was so happy with it.

But when I went to bed I couldn't sleep. I was going over the numbers in my head and nothing was working out as I described it in the post. After a while I got up and took the post off-line so I could sleep.

Reading the thing over this morning I come to this:

And now that inflation has gone high, I think we can expect it to stay high for about a year, even if price increases drop back to what they were in 2020. Why? Because as one new monthly low is added in to the change-from-year-ago, the oldest monthly low drops out of the calculation. Once a monthly value comes into the calculation, it stays in the calculation for a year.

I don't like saying stuff like that without checking it. Things that are obvious in the daytime keep me awake at night. "Obvious" sometimes goes down with the sun.

So I looked over yesterday's graphs and decided to make a new graph showing just one inflation dataset instead of two, and to show "monthly" and "from year ago" on one graph instead of two. So I tossed aside the inflation measure I usually use, the "All Items" measure, shown in red on yesterday's graphs.

That leaves me with the one I don't prefer, the "All Items Less Food and Energy" measure. If that keeps me awake tonight, maybe I'll write about it tomorrow.

Graph #1: Monthly (red) and From Year Ago (blue) Inflation Rates

The red runs low because it shows the change in prices from a month ago. The blue runs higher because it shows the change in prices from a year ago.

After January 2021, the blue line goes up a lot while the red goes up much less. Doesn't that bother you just a little? Bothers me.

The red line peaks in April at about 1%. That would be 12% for the year, or more. That's unacceptable. But one month before that peak, the number was in the normal range, or close to it. And for six or seven months before that, the numbers were low; before that, only one high one; and before that a month of low inflation and three months below zero -- three months of falling prices. This takes us back to March 2020. Almost all the numbers were low, and the one high (July 2020) was more or less counterbalanced by the low of April 2020.

So how did blue inflation skyrocket in April 2021? Yes, May and June also show high inflation rates. But May and June didn't happen yet when they figured the April numbers.

I want to see how the changes are related: how a change in the monthly number influences the course of the "from year ago" inflation rate. To make it easier to see the effects of a change, I made some dummy data that always has the same monthly inflation rate.

It turns out that a unchanging inflation rate of 0.165% per month results in 2% annual inflation. I went with that because the 2% annual rate was the "target" rate for a long time. Still is, maybe.

The colors don't match the FRED graph, but here's the graph of unchanging inflation:

Graph #2: Dummy Data showing a Constant Monthly Inflation Rate

The numbers across the bottom represent months. The monthly inflation numbers are shown in blue, and the from-year-ago numbers in Excel's Uninspiring Orange. The orange line starts late because I waited until there was 12 months of blue data. 13 months actually, so the first month can be replaced by the 13th month. January-to-January, or whatever. After that late start, orange runs at 2% annual.

Now let me modify the dummy data. After month 20, I will replace three numbers with the three below-zero values from graph #1: March, April and May 2020. The covid recession appears as a vertical bar on that graph; for reference, the recession started on February 1 and ended on April 1.

The modified Excel graph, with the blue line showing the three-month low:

Graph #3: Constant Inflation except for the  Low of March-April-May 2020

According to the FRED data, the three low values are -0.021, -0.369, and -0.075. Add the three together, you get -0.465. That's a decline of less than half a percentage point. But the orange line shows a full percentage point decline. How did that happen?

I dunno. But that's a question for another time. The question that must be answered now is this: What's going on with the graph? Why does the blue line return immediately to the 0.165 level while the orange line runs low for so long? And then why does the orange line come back up when it does?

Just like the red line on graph #1, the blue on #3 has just three points below zero. I even used the same numbers. And after those three points, the blue goes back to the constant 0.165% monthly inflation rate. So that's the blue line.

The orange line measures a 12-month change. After the 3 low values visible on the blue line have come into the orange calculation, the other 9 values in that calculation can only be some of the 0.165 numbers that are everywhere else on the blue line. As the months go by, the low values gradually move from "newly entered" to "ready come out of the 12 values". Then, over the next three months, those lows come out of the calculation, and the line works its way back up to the 2.0% level.

But if, by some chance, something pushes the monthly inflation number up just as the three monthly lows are coming out of the from-year-ago inflation number, that number will not just go back to normal. It will go high, as the low number comes out and the high number goes in at the same time. 

In the real world, and on graph #1, the unusual low of April 2020 was coming out of the from-year-ago calculation just as the unusual high of April 2021 was being counted in. If the timing was different, we might not have had a high from-year-ago number for April 2021. But the timing wasn't different.

Again in May, the low of 2020 came out of the from-year-ago inflation just as the high of May 2021 was being counted in. As a result, we got two fluke months in a row where from-year-ago inflation (blue on graph #1) shot up a lot, while the monthly numbers (red) did not increase enough to explain the large increase in the from-year-ago number.

This explanation doesn't work so well for June because the 2020 number wasn't so low. Maybe people saw the high inflation of April and May and decided to get on the bandwagon and raise their prices. Dunno.

 

Now I'm adding the highs from graph #1 to the Excel graph: April, May and June of 2021:

Graph #4: Constant Inflation except for the 2020 Low and the 2021 High

I cut a few years off the start of the graph and added a few at the end to make room for the return to normal a year after the 2021 high.

Month 34 on this graph is April 2021, the first of three highs. That puts us, in October, at month 40. So we're only about half done with this high level of "from year ago" inflation. Of course it depends on how reality turns out. But if you go by graph #1, we are already three months past the monthly (red) highs, and well into the Lets Make Things Look Bad For No Reason stage.

We'll see how it goes. 

Oh by the way: Yesterday's post is online again, unchanged. Maybe I'll sleep better tonight.