- They have changed what goods are thrown in the basket and in what proportion repeatedly.
- They have made adjustments, or when they haven't they arguably should, for variation in quality of goods since for example any TV you buy now is higher quality than anything available in 1990. You don't have the option of buying the previous quality anymore though.
- It's a bit murky how to account for substitutions over time such as people buying Impossible burgers instead of beef.
Many of the adjustments they consider are completely reasonable but involve subjective and game-able choices which makes the measure hard to trust at face value.
One more problem with CPI is that the way it works is they survey a bunch of median households about what they bought and how much of it they bought, and they don't adjust for how much money they saved. So in economic times when the median household was able to save some money, that resulted in a different CPI than when households spend their entire paychecks (where they have to adjust their spending preferences/how they allocate according to the maximum they could spend -- all their money).
For example, if grocery prices go up 30% on average, but only 10% on some staple like beans, I might buy more beans than I used to, putting more weight on the item whose price increased less, but not out of preference for that item. That increases the BLS weight for beans and decreases it for prices that are rising more quickly. Back in 1980 I might have bought different items and also saved some money each month. Now that saving portion of my "expenditures" is gone, but that also isn't accounted for by the CPI.
There's a reason that charts of real wages over time are largely flat or even increasing [0], despite that being inconsistent with the economic reality, and it's this BLS metric to blame.
You've described preferring to buy beans over other food items.
First of all, this is incorrect. They do track per-item [0].
The point that I think you are missing is that, to me, beans are of a lower quality than what I used to purchase. However, hedonic adjustments don't go the other way to account for this.
You see, because I do not save any money, and I allocate the same proportion of my median salary to spend on food as I did before (because I have no additional money to put in), my contribution to inflation is based on the inflation of items that I can afford, not of what I used to be able to afford. That's just how they calculate it, resulting in a lagging indicator that misses some inflation altogether when combined with increasing wealth inequality. In an environment where the median person has no disposable income, a portion of inflation is always hidden by their changing consuming habits, where those changes are driven by the inflation itself and not by any change in quality or qualitative preference.
I fully understand you feel like beans are lower quality, but that doesn't matter because it is not possible to track what you might have bought, so we track what you did buy. You may buy beans because bean companies go viral with their marketing, or because your local store is out of potatoes, or because you found a really good looking chili recipe you want to try. CPI doesn't try to explain why you bought beans, or didn't buy potatoes.
Strawman. No one ever claimed that.
I'm not sure why you're continuing to argue here.
> but that doesn't matter because it is not possible to track what you might have bought
I can think of several ways of improving on the current system, two of which can be easily derived if you understand my previous post.
Spoiler: track the inflation of goods weighted by quantities purchased 1-2 years prior. And incorporate savings as a significant category such that the category CPI_savings = -(net_savings_this_month - net_savings_X_months_ago) / ((income_this_month + income_X_months_ago)/2). Still imperfect if most of the people answering the survey have no savings, but at least it would have captured the undocumented inflation from the 1980s-2010s that was hidden by decreasing savings.
Fundamentally the BLS in the US has no incentive to accurately report inflation. They are a department of the executive with no real safeguards against political influence. Reducing the official CPI reduces the government's expenditure and understating inflation is a great way to do so without officially cutting any budget item.
Further, see my sister comment to yours, inflation is personal, a country-wide single-number inflation measure is broken by design.