What I'd really like to see is a multi-dimensional inflation vector that would also be future-proof. People really need diverse items: they need food, but also real estate, the prices of which are going through the roof currently, and that's a form of inflation. Also, there are many profiles of consumption. People have different needs. I'd say that it could be a 100 floating point vector; with the first 40 values arr…
Take these examples:
- The housing market is extremely diverse in certain countries, where some cities have experience price increases the past decades, some have been stagnating, and some have seen price decreases.
- Same goes for salaries. In extremely centralized cities, there's likely been above inflation rise, while in extremely decentralized cities, salaries have been stagnating, sometimes just standing still.
- Fuel prices can depend on local policies and taxation laws.
- Foods etc. can depend on how far away the places are in the logistical chain, how many competitors there are in the area, how the stores and chains are organized, etc.
- Travel (airfare, trains, etc.) can depend on location. Some larger hubs have experienced airfare that's become cheaper and cheaper with years, while rural places have seen the same going up.
Just in my country (Norway), there are hundreds of towns and villages with prices that are all over the place, and you'd need to scrape thousands of websites. Some of the data would be more or less impossible to get, due to the data never being digitized to begin with.
Of course, I would love to have such a database ready. I'd love to be able to run a query on my home town, and see what the prices for top 10 items were, say, 45 years ago. And compare that to the national inflation, as well as local prices of the more expensive cities.
So I think that whomever is going to work on this, probably just need to take the very rough and general (historical) inflation rates, which most government websites offer. To make a DB which accounts for all the micro- and macro changes would just be too consuming for any small team. You'd basically need a Wikipedia-esque team of individual contributors to cover all the prices.