[eru]

Thanks for checking!

Btw, many economist argue that CPI overstates inflation. Mostly because CPI has a relatively fixed basked, but people adjust what they are buying if relative prices shift. (Eg if bread becomes relatively more expensive than pasta, people buy less bread and more pasta. The CPI basket does not reflect that.)

A modern alternative is to use the GDP deflator. https://en.wikipedia.org/wiki/GDP_deflator Basically, it means to use everything that got produced / consumed as the basket for your inflation measurement.

[bendbro]

> if bread becomes relatively more expensive than pasta, people buy less bread and more pasta. The CPI basket does not reflect that

That makes sense. But say bread at $1(2023) is objectively worse than pasta at $1(2023); is there any way to account for that? It seems like a generic form of shrinkflation: without holding the basket constant, how do we know we are getting the same value?

To be fair, it could work the other way and the pasta at $1(2023) is better than the bread at $1(2023).

[eru]

> But say bread at $1(2023) is objectively worse than pasta at $1(2023); is there any way to account for that?

See https://www.bls.gov/cpi/quality-adjustment/ and https://www.bls.gov/cpi/quality-adjustment/questions-and-ans... for how the CPI attempts to control for quality. How well they succeed is another question.

Quality adjustments in inflation data are always something that requires judgement. (Even though they try to minimise that.)

A recent example that was not reflected in the CPI:

During the pandemic choice and quality dwindled. Instead of buying your favourite toilet paper, you just got whatever wasn't sold out. Eating out got a lot worse, when it was possible at all. But if a burger still sold for 2 dollars before or during the pandemic, the CPI did not budge. So official inflation was understated.

Conversely, as we came out of the pandemic quality recovered without that being fully reflected in the CPI. So official inflation was overstated.

---

Because inflation requires judgement, I try to avoid having to rely on it in any analysis.

The discussion we had could mostly be re-formulated in terms of

> How has nominal median pay developed over time relative to nominal per capita GDP?

That framing leaves out inflation completely.

Of course, it ignores absolute advances. And deliberately so!

But those are better left to a separate discussion.