> The Fed massages the inflation rate by throwing out "extreme" deviations.

The rationale is valid though. Price-conscious consumers would simply switch brands, stores, or even go for substitute goods if the price of a particular product they chose happened to skyrocket for no reason while the price of all other alternatives barely changed.

It's not valid though. Inflation driven substitution means lower quality. You used to barbecue steaks every weekend. Now you can only afford bologna. You used to wear leather shoes. Now you can only afford plastic crocs. Etc, etc, etc.

Depends on the degree and the duration. If the local slaughter house has to shut down because of Covid and steaks triple in price, does it really make sense to indicate that inflation is ~50% (number made up) over my basket of goods? Or does it make sense to acknowledge that I’ll shift over to chicken or no meat for a few weeks until prices return?

Similarly, if the price of beef permanently puts it out of reach for my family, then I absolutely would call foul on substituting it from my basket of goods. That would be inflation, and excluding beef would be gaming the system.

The problem is that determining between these cases is very hard to do in the moment, and only becomes obvious in hindsight.

> It's not valid though. Inflation driven substitution means lower quality.

It really is valid. Not only is it valid, it's precisely the whole point of tracking a consumer price index. The goal of CPI is to track the prices of goods and services consumed by the population, not brand- or product-loyalty. When prices go significantly up, it's a known fact that consumers react by seeking affordable alternatives, or in the case of luxury items simply going without.

Toy model: There are 10 goods you buy. They're reasonably substitutable for one another (maybe they're different foods, or types of computer game, or something).

Every year one of these increases in price by 20%. A different one every year, cycling between them all. So every 10 years, they all increase by 20%, an effective inflation rate of about 1.8%.

We have a price index based on a basket of these goods. At the start of year 1, our basket contains equal quantities of all the goods. In year 1, the price of good #1 abruptly goes up 20%, having been stable for the last ten years. OK, we say, this is a thing that's experiencing some sort of anomalous increase in price, so we'll take it out of the basket. At the end of the year, we look at our basket of goods 2..10. The prices haven't increased at all! Zero percent inflation! Splendid.

At the start of year 2, our basket contains equal quantities of goods 2..10. In year 2, the price of good #2 abruptly goes up 20%, having been stable for the last ten years. Another anomalous increase. Better take this one out of the basket too. At the end of the year, we look at our basket of goods 3..10. The prices haven't increased at all! Zero inflation again.

At the start of year 3, our basket contains equal quantities of goods 3..10. In year 3, the price of good #3 abruptly goes up 20%, having been stable for the last ten years. Dang, better take this one out of the basket. Our basket is getting a little short of this kind of good. Hey, good 1 has been stable for a couple of years after its blip in year 1, let's put it back in. At the end of the year, we look at our basket of goods 1,4,..10. The prices haven't increased at all! Zero inflation.

And so we continue. Each year we remove an item that obviously doesn't belong in the basket because it's suffered an anomalous increase in price, so of course consumers will be switching to something else. Sometimes we put back in things that have been stable for a while. Every year we get an inflation figure of zero.

And yet, somehow, ten years after we started all the prices are 20% higher. How on earth did that happen?

Great, so when prices increase so much people can't afford to buy something there's no inflation with regard to that item. That's handy.

You're right, I suppose the homeless don't experience inflation in housing costs, therefore it doesn't exist!

This is as beautiful as the NSA's redefinition of surveillance whereby they only collect _all_ the data they can. But it's not surveillance until they look at it, which they assuredly don't because there's so much of it.

Further to the point above, when this happens the GDP drops. These metrics need to be looked at in combination with each other, not in isolation. There is no "one metric", its a series of metrics that tell a story.

If the economists track the price change in 5 brands of spaghetti and report on the median, that would represent people seeking alternatives, for sure.

But if they track the price change in 1 brand of spaghetti, treatment of 1 broken arm, and 1 million-transistor CPU? Those aren't goods that can substitute for one another.

right, so CPI is not a good measure for actual inflation

CPI is a good method to measure inflation, but it's managed by people who have an interest in getting a low number, so it is a poor measure in practice.

> right, so CPI is not a good measure for actual inflation

No, quite the opposite in fact. CPI is a accurate measure of inflation as it tracks how much the real-world consumer prices vary where it really matters: real-world consumer expenses.

Perhaps the mistake you're making is assuming wrongly that demand for specific brands/makes/models is perfectly inelastic and consumers do not change their consumption patterns at all even if their price increases so much that they can no longer afford them. Meanwhile, if a low/middle class family sees the price of stake skyrocket, you can bet they'll start to switch a few meals to hamburguers or hotdogs.

The argument is that there is no inflation if people can't afford stuff, because they'll just be priced out and forced to not buy stuff anymore, thus their expenses will be stable. Duh. Inflation is not how much money people spend, inflation is ever increasing prices leading to people affording less and less stuff with their rapidly depreciating money.

If inflation happens and food becomes more expensive and you keep buying the same food you have less money left over for non food products. The CPI will adjust and weight your food expenses more which will raise the CPI much faster than if it also looked at fitness equipment that went down in price because you didn't buy it because food was too expensive.

If food gets more expensive and people don't spend more on food then inflation didn't happen by definition. They still spend the exact same amount after all, they just got different food instead.

By that definition inflation basically never happens until people get higher wages so they have more to spend.

That’s clearly absurd, and so defining inflation in a way that means it never happens when people substitute goods is also clearly absurd.

That is exactly how the ruling class sees the situation. Gallon of milk $10? Chump change, maybe once it hits $100 it will make for a bit of faux concern table chatter at our $10k/seat fundraising dinner next week. The plebes agitating for $15/hour? Catastrophe.

If inflation means measuring value of money over time, substitution is not correct. You could substitute before - you just didnt, because it was not necessary. With a measure you describe we can slide into poverty and it doesnt show.

Yet it is so far from rigorous which makes it ripe for abuse. The "extreme" deviations are defined arbitrarily, evidenced by the fact that different offices of the Fed come up with different inflation numbers. Hence the chairman can choose one which is conveniently on target.