Probably so it can tie in with some smart contract that's based off the CPI (eg. Treasury Inflation-Protected Securities). As for why that's useful, I'm not sure. At the end you'll still need a trusted oracle for the scraped prices.

It would be very useful for on-chain financial derivatives based off a more accurate inflation gauge. I don't trust the official CPI numbers and don't see them as accurately reflecting my cost of living increases over the years. Couple in the fact that there are literally trillions in pension benefits tied to CPI, so the incentive to manipulate these numbers is huge. It's already very opaque. If you look at the constituents [0], housing, education and health care make up around 60%, so it makes no sense that inflation has been flat for the last ten years.

The prices would ideally come from a large number of sources and some averaging would take place.

[0] https://www.advisorperspectives.com/dshort/updates/2021/07/1...

>I don't trust the official CPI numbers and don't see them as accurately reflecting my cost of living increases over the years.

What makes you think a NGO would do a better job? On one hand they might not have the same incentive to manipulate the numbers in the same way that the government would, but that would be replaced with an incentive to come up with whatever numbers that their donor base would like to see (see also: media companies adjusting their coverage to retain subscribers).

>If you look at the constituents [0], housing, education and health care make up around 60%, so it makes no sense that inflation has been flat for the last ten years.

Are you missing a part of your argument here? You just say that a few categories make up 60% of the CPI, then immediately conclude that "it makes no sense that inflation has been flat".

>The prices would ideally come from a large number of sources and some averaging would take place.

The problem is that you need to do various subjective adjustments (eg. the controversial "hedonistic adjustments") for the numbers to make sense.

> What makes you think a NGO would do a better job?

Maybe they wouldn't do better, but they wouldn't have a trillion dollar incentive and ideally there'd be competing metrics.

> Are you missing a part of your argument here? You just say that a few categories make up 60% of the CPI, then immediately conclude that "it makes no sense that inflation has been flat".

In the US, over the last 10 years housing, education and health care have grown by a lot while official inflation has been historically very flat (0 - 2%).

Health care: On a per capita basis, health spending has increased over 31-fold in the last four decades, from $353 per person in 1970 to $11,582 in 2019. [0]

Housing: Median home price was around ~210k in today's dollars ten years ago and is now ~314k [1]

Education: The cost of college increased by more than 25% in the last 10 years [2]

> The problem is that you need to do various subjective adjustments

Sure, but again, I don't trust the officials to do this given the trillion dollar incentive to fudge the numbers.

[0] https://www.healthsystemtracker.org/chart-collection/u-s-spe....

[1] https://dqydj.com/historical-home-prices/

[2] https://www.cnbc.com/2019/12/13/cost-of-college-increased-by...

>Maybe they wouldn't do better, but they wouldn't have a trillion dollar incentive and ideally there'd be competing metrics.

The fact that no private entities tried to do anything similar (eg. bloomberg, s&p, msci) despite the massive incentive to do so (it's hard to price bonds/mortgages/loans if you can't trust the inflation numbers), suggests that the current number is already pretty good and/or that coming up with comparable numbers would be massively expensive. I think the problem is of expectations. You want a competitor to the BBC. You hope that it'll be something like propublica, but given the crowd associated with crypto, you'll probably end up with something like zerohedge.

>Health care: On a per capita basis, health spending has increased over 31-fold in the last four decades, from $353 per person in 1970 to $11,582 in 2019. [0]

That seems pretty consistent with the figures from CPI: https://fred.stlouisfed.org/series/CUSR0000SAM2

From 1999 to 2019, the source you gave (Total national health expenditures, US $ per capita) 154% increase. The "Medical Care Services" component of the CPI grew by 112% in the same time period. The figure you gave also doesn't account for an aging population (older people spend more on healthcare, irrespective of cost), so the "true" difference is probably smaller.

>Housing: Median home price was around ~210k in today's dollars ten years ago and is now ~314k [1]

Because looking at home prices is a poor way of assessing housing cost increases. If you factor in falling interest rates, the cost of a house has actually gone down. Houses have also gotten bigger and better equipped in the last few decades, which might increase the overall sticker price, but arguably shouldn't contribute to CPI increase.

see: https://awealthofcommonsense.com/2021/03/what-if-housing-pri...

>Education: The cost of college increased by more than 25% in the last 10 years [2]

The "Tuition, Other School Fees, and Childcare" component of the CPI has increased 37.4% in the same time period.

https://fred.stlouisfed.org/series/CUSR0000SEEB

The reason there's no concerted effort for Bloomberg, S&P, etc to create an alternative inflation index is because there's no point. It won't be able to be integrated into financial products. Finance is extremely regulated and apart from some bespoke CDS, getting acceptance of a main stream financial product with this weird index would be very difficult. With smart contracts there is no approval process and a very low cost to create such products. But who knows, maybe no one is interested, but it's still an important project

> That seems pretty consistent with the figures from CPI:

CPI 1970: 31.1 CPI 2021: 570.6 Increase: 570/31 = 18-fold compared to 31-fold. Don't just say "looks similar" when you could do simple division and realize it's not similar. I don't think there is a single economist that would argue that health care spending has not grown considerably faster than the rate of inflation.

Here's the continuously compounded rate of growth of health expenditures as per the st louis fed

DATE HLTHSCEXPHCSA_CCH

1/1/01 8.61255

1/1/02 8.53011

1/1/03 6.8683

1/1/04 7.1269

1/1/05 6.22677

1/1/06 6.17762

1/1/07 6.12021

1/1/08 4.61058

1/1/09 4.60561

1/1/10 3.86769

1/1/11 3.5515

1/1/12 3.53598

1/1/13 2.54628

1/1/14 5.1474

1/1/15 6.26148

1/1/16 5.01743

1/1/17 3.98121

1/1/18 4.16224

You can do the same for housing [1]

[0] https://fred.stlouisfed.org/series/HLTHSCEXPHCSA#0

[1] https://fred.stlouisfed.org/series/MSPUS

>The reason there's no concerted effort for Bloomberg, S&P, etc to create an alternative inflation index is because there's no point. It won't be able to be integrated into financial products. Finance is extremely regulated and apart from some bespoke CDS, getting acceptance of a main stream financial product with this weird index would be very difficult.

1. you're thinking that the index would have to be integrated into the product itself (eg. this bond pays s&p inflation index + 2%) for it to be useful. This isn't the case. Even for forecasting or internal rate-setting purposes it would be massively useful.

2. If you check the website for s&p or msci, they got an endless list of indexes. Some of them are even integrated into financial products (eg. tracked by ETF or derivatives). Exotic indexes is definitely not a problem for them.

>I don't think there is a single economist that would argue that health care spending has not grown considerably faster than the rate of inflation.

I think the problem here is that you're conflating per-capita spending with price increase. As mentioned earlier an aging population would increase per-capita spending without necessarily increasing price. People opting to buy better, but more expensive care would also increase per-capita spending without necessarily increasing price.

>You can do the same for housing [1]

What's your contradictory source for the actual price of housing?

The CPI basically measures the median American's budget. So if wealth inequality is unchanging, then it's a decent measure of economy-wide inflation. Otherwise, it measures inflation x delta wealth inequality.

What if those budgets have significantly less buying power? Especially in critical areas like housing, healthcare, and education?

Then people will tend to buy less of them to fit their budget. Here is the most telling chart: inflation-adjusted wages in the US [0]. They never change, because the CPI uses a survey to determine what the average American is buying, and how much it costs, then they adjust weights based on how much they spend on it (which cancels out the cost because people change their buying preferences based on current prices) and leaves you with the median American's total expenditure. Yeah, that's really how they do it. And that's why CPI is a joke.

[0] https://www.pewresearch.org/fact-tank/2018/08/07/for-most-us...

Is it possible people are buying less by volume, even if amounts spent level off? For example smaller homes, less healthcare, getting less educational value per (adjusted) dollar spent?

Yes, that's exactly it. Such changes are not well reflected in the CPI model.

Including the computation and data on-chain enables other parties to use it trustlessly and verifiably. If you, for example, wanted to make an algorithmic central bank that reacted to inflation numbers you would need on-chain CPI in order to do that. Similarly you may want to index your interest rate to the CPI in some way (like LIBOR). It becomes a lot more useful when on-chain than in a central database

It’s not trustless, because you’re trusting the scraping NGO to put the correct price data on the chain. You can verify that it came from the NGO with a simple signature; you don’t need the chain for that.

nah fam that won't likely be the final topology of the network. you can have oracles all the way down and a slashing risk for those that deviate significantly from the consensus prices and CPI numbers. so you're trusting the economic design of the system to appropriately incentivize the participants to give honest data

I think you are right, this can be done without a blockchain.

Once done, it could be useful to make an oracle with it and put it on ethereum so it can be used by other smart contracts. But not the calculation itself nor the original data points

If all you have is a hammer…

These are people who think the government wants to censor their prices database... so not exactly geniuses.

Take a look at Argentina 2007-2016 [1,2,3]

[1] https://en.m.wikipedia.org/wiki/MIT_Billion_Prices_project

[2] https://www.hbs.edu/faculty/Pages/item.aspx?num=54491

[3] https://en.m.wikipedia.org/wiki/Cristina_Fernández_de_Kirchn...

What happened in Argentina 2007-2016 that would suggest censorship-resistance in a prices database is a feature that makes sense?

The government misreported inflation figures. (According to the people who made the price database)

Mining fees?

For about the same reason that you should use blockchain for inventory management.