Hacker News Reader

Comment by aeternum | Hacker News Reader

The non-uniformity of inflation is interesting and has been brought up by many economists recently.

I wonder if eventually the Fed will try to track inflation as a vector rather than a single number.

Indeed, it always annoyed me when the CPI a) excludes actual costs of living and b) alongside things like "food" included "flat-screen TVs".

Staple foods might cost 25% more, but hey, those TVs are down 20%, so yay!

An old company of mine was very proud that everyone's payrise began equivalent to CPI increase, before any other performance related increases.

That was nice, but it's a CPI that excludes rent/mortgage payments in a massive property bubble. When your rent that was already 40% of your net income goes up by 25%, CPI is meaningless.

When we suggested their base pay raise also consider that aspect, we got a blank stare and "no, no, CPI... <vague hand gestures>"

Mind you, it's like the "unemployment rate", for statistical purposes, you're not unemployed if you worked somewhere for one hour plus, paid or unpaid.

So the percentage of our population on the unemployment benefit always tracks higher than the official unemployment rate.

I mean, I guess that's their measure, and it's useful for statisticians, but it's not meaningful for citizens.

> Staple foods might cost 25% more, but hey, those TVs are down 20%, so yay!

It's even worse than that because most of the time the TV price don't really goes down by 20%, but by 3% and hedonic adjustment makes it appear as if it went down by 20% in CPI because some people estimated that going from HD to 4K increased the value by 17% …

That's how we get figures saying that computers cost 20 times less than what they used to be when in reality it costs a little less than 2 times less.

> That's how we get figures saying that computers cost 20 times less than what they used to be when in reality it costs a little less than 2 times less.

Can't you buy old PCs for 20 times less than a (second-hand) modern PC?

I bet this Pentium 4 desktop cost around NZ$2000 new: https://www.trademe.co.nz/a/marketplace/computers/desktops/n... but now it's going for NZ$100.

> Can't you buy old PCs for 20 times less than a (second-hand) modern PC?

You can, but it's basically a paperweight at this point (I doubt you can even find a recent OS that boots it). Whereas back in the day, such a computer allowed me to play AAA games, browse the web, watch movies, and use the MS office suite for work or studies. Just as the average today's computer.

There have been gains (engineering or scientific calculation benefit a lot from the additional available computing power), but for most user, the improvement is at best marginal (the games' graphics are better, but I doubt it makes kids this days enjoy their games more than what I did when I was 12) or even nonexistent (what's the productivity gains from Word 2021 compared to Word 2000 for a student writing a thesis ?).

You can still do all those things on the Pentium 4 (boot up Windows XP, install Morrowind, use MS Office 2000 or whichever version was current, watch The Matrix), but the cost to buy the PC is significantly less (even taking into account the second-hand nature of the PC).

It isn't comparing like to like to compare the cost of a modern PC to one 20 years old, with modern games, modern websites, modern Office products etc.

A decade ago you could get a new computer for $2000, now that same computer is $100 isn't apples to apples. Used items always cost less than new and decade old used items doubly so.

> I wonder if eventually the Fed will try to track inflation as a vector rather than a single number.

They already do, they just publish it as a single number for simplicity and ease of comparison. Multiply your vector by weights corresponding to "relevance" and you get PCE.

https://www.thebalance.com/personal-consumption-expenditures...

What does tracking inflation as a vector mean? My memory of vectors was that they are a direction and a magnitude. But I don’t see how this relates to inflation.

If we measure a basket of 1,000 products that is direction and magnitude in a 1,000 dimension space. Those are not the important things of that vector. What's important is that we have 1,000 measurements of inflation for 1,000 different products. One of the points of the article is that it could be possible to collapse that vector into a single scalar number as governments do, but it usually isn't because the variance between each component is too big and makes that number meaningless.

And yet it's useful to be able to tell my customers that I have to raise my fee because of a 6% inflation instead of "gas went up 20%", "but clothing went down 5%."

a multi-dimensional vector i presume the OP meant - one value per item.

you can still measure the distance between such a vector (even though you can't really visualize it). This distance is then the change in inflation, and can be compared across years.

Yes, I should have said multi-dimensional vector

The BLS already publishes per category CPIs, alongside the aggregates[1]. Presumably you'd want the Fed to use different weights than the BLS (overweight "necessities"?) for monetary policy. But they already do, they vastly underweight food and energy prices, using "core inflation"[2]. (Besides, they typically use the BEA PCE deflator[3], rather than BLS CPI for monetary policy.)

Occasionally you'll hear that this is part of some government conspiracy to suppress true cost of living measures or some such. I think that's largely nonsense as CoL adjustments in government departments are always done with plain CPI, only the Fed uses Core. The logic of using Core (excl. food and energy) is mostly pragmatic, it's just a simple smoother, they could also use trimmed or rolling averages. The raw inflation is pretty volatile, based on that they'd have to adjust policy rates up and down all the time.[4]

[1] https://www.bls.gov/news.release/cpi.t01.htm

[2] https://en.wikipedia.org/wiki/Core_inflation

[3] https://www.bea.gov/data/personal-consumption-expenditures-p...

[4] https://krugman.blogs.nytimes.com/2010/02/26/core-logic/ (Sorry, can't find a non-paywalled link)

Reply on news.ycombinator.com