CPI is also a broken measure due to its inability to measure wealth inequality (sampling median purchasers only) and discretionary hedonic adjustments. Real inflation has likely been ~33% higher than reported for decades (the Big Mac index is probably the true gold standard).
>CPI is also a broken measure due to its inability to measure wealth inequality (sampling median purchasers only)
And that's fine, because the CPI's job is to measure how prices have changed (ie. inflation), not wealth inequality. Just because a given metric doesn't support your pet cause doesn't mean it's "broken". It's like complaining GDP is a broken metric because it doesn't measure how oppressed minorities are.
> Real inflation has likely been ~33% higher than reported for decades (the Big Mac index is probably the true gold standard).
I tried to confirm this and the results were far from conclusive. The big mac index data can be obtained from github[1] and gives a 129.9% increase between april 2000 (earliest data available) and july 2022 (latest data available). To compare this against the CPI data, I checked on FRED and the "Consumer Price Index for All Urban Consumers: Food Away from Home in U.S. City Average"[2] component of the CPI gives a 87.0% for the same time period. If you take those numbers and convert them to annualized rates, you will indeed find a 33.3% difference between them. Looks like your claim is confirmed, right? But not so fast. The CPI category listed above actually breaks down into more detailed categories, including "Limited service meals and snacks", which presumably the big mac falls into. That data isn't available on FRED but can be found on BLS's data viewer[3] and gives you a 105.8% increase for the same time period. If you convert that to an annualized rate the difference between that and the big mac index drops to 15%. I suspect if we try harder we can eliminate more of the discrepancy. For instance, the "Limited service meals and snacks" index was probably computed from a basket of items, not just big macs. Items that have fatter margins (eg. fries or drinks) might have inflated slower than big macs, thereby dragging the growth of the basket down and explaining the difference.
At some point if food, housing, medicine, education etc are consistently rising faster than inflation for long periods perhaps there’s a constant issue with how inflation is being measured.
In the end if your spending doesn’t map to the basket of goods used to calculate inflation then the official figures aren’t meaningful.
> At some point if food, housing, medicine, education etc are consistently rising faster than inflation for long periods perhaps there’s a constant issue with how inflation is being measured.
The CPI is computed from a basket of goods, so you're always going to be able to pick some components that are growing faster than average. The fact that such components exist doesn't mean that "there’s a constant issue with how inflation is being measured".
>In the end if your spending doesn’t map to the basket of goods used to calculate inflation then the official figures aren’t meaningful.
Sure, they might not match your consumption patterns 100%, but that's because everyone's consumption patterns are different and it's impossible to come up with a metric that maps 100% for everyone. What matters is whether the basket accurately captures the spending patterns of the public as a whole. The BLS makes the basket weights publicly available[1]. I skimmed the list and they look fairly sensible. Are there any specific items from that you think don't accurately capture how the average american spends?
> The fact that such components exist doesn't mean that "there’s a constant issue with how inflation is being measured".
The problem is consistency. People need to eat. So if food increases significantly faster than inflation then eventually people’s basket of goods will consist of 99% food and 1% everything else.
If you look at a family of 4’s spending in 1990 vs today some stuff has gotten cheaper but to have an equal lifestyle you also need to buy the stuff that’s dramatically more expensive. A cheap Disneyland vacation isn’t cheap any more. Granted people’s actual purchasing reflects the new reality so Netflix substitutes for that movie tickets they can’t afford, but that isn’t an equivalent good.
If you want a specific example Apparel. Clothes the average person buys has gotten much worse over time. Fast fashion isn’t just about keeping up with trends, it just doesn’t last.
>If you look at a family of 4’s spending in 1990 vs today some stuff has gotten cheaper but to have an equal lifestyle you also need to buy the stuff that’s dramatically more expensive. A cheap Disneyland vacation isn’t cheap any more. Granted people’s actual purchasing reflects the new reality so Netflix substitutes for that movie tickets they can’t afford, but that isn’t an equivalent good.
Ironically the hedonic adjustments that you decry so much about is supposed to adjust for this. Also, while you can bring up examples of stuff getting lower in quality, you fail to bring up instances where quality improved (eg. google maps on your phone vs paper maps). Without actual hard numbers quantifying either side arguing over this just turns into a handwaving war.
Hedonic adjustments are argued about quite often because they let the BLS change inflation numbers almost arbitrarily. Your Google Maps example is replacing something that was very cheap with a slightly better version of it that was also very cheap it’s got negligible impact on inflation compared to big ticket items.
Here’s one issue that stuck with me: “While there are arguments that the new goods do exhibit atypical price behavior—entering at a high price and following a “U-shaped cost curve”2—this aspect of the new goods problem is an issue more for sampling than for quality adjustment procedures. Nevertheless, we will argue below that the existence of differential price trends within item categories does have implications for the way in which hedonic regression techniques are implemented.” https://www.bls.gov/pir/journal/gj17.pdf
The BLS uses improving outcomes to adjust for healthcare costs. In terms of new treatments that’s perfectly reasonable, but it undermines increased costs for basic treatments like broken bones. People may enjoy higher quality but not if they are priced out of all options.
>Hedonic adjustments are argued about quite often because they let the BLS change inflation numbers almost arbitrarily.
except that BLS lists which CPI categories have hedonic adjustments applied[1] and even how it's calculated for some categories[2], so they can't decide to arbitrarily sink energy inflation by saying that gasoline has gotten 50% better. Likewise, because they break out CPI by components, you can calculate a hedonic-less or food-only CPI if you so desire.
> Your Google Maps example is replacing something that was very cheap with a slightly better version of it that was also very cheap it’s got negligible impact on inflation compared to big ticket items.
It also replaced nearly all the electronics you bought from radio shack back in the day (eg. computer, alarm clock, radio, calculator, cd player, camera/camcorder, cordless phone, tape recorder). Quality improvement also applies to other categories as well, like appliances or cars. Finally, like I said before, the BLS breaks out all the inflation figures by category, so if you really wanted to you could make a "bare minimum to survive" index or whatever, although your initial comment seems to take issue with how the individual categories are calculated rather than how the categories are weighted.
>Here’s one issue that stuck with me:
I'm not sure what the actual issue is from your quote. Can you reexplain using your own words?
>The BLS uses improving outcomes to adjust for healthcare costs. In terms of new treatments that’s perfectly reasonable, but it undermines increased costs for basic treatments like broken bones. People may enjoy higher quality but not if they are priced out of all options.
But the CPI seems to be capturing healthcare just fine? Look at how high "hospital services" rose compared to overall inflation.
Moreover, the core issue seems to be "what do you do if things in a category are getting better but people are spending the same or even more on it?". The whole idea behind CPI is that you're supposed to measure price changes for the same goods. In that sense it's obvious that we should lower the CPI even though people are spending more on it. However, I can see where you're coming from given how people conflate "CPI" with "inflation" and "inflation" with "cost of living", the latter of which ignores like-for-like comparison for goods and only looks at cashflow. If you want to measure how has that changed, you'll need to come up with another metric, but I suspect that opens a whole can of worms. eg. how do you compare costs for phones between 2005 and today, if people in 2005 were buying $200 flip phones and today are buying $1000 iPhones, even though a $100 android would totally be serviceable and blow the $200 2005 phone away?
> It also replaced nearly all the electronics you bought from radio shack back in the day (eg. computer, alarm clock, radio, calculator, cd player, camera/camcorder, cordless phone, tape recorder). Quality improvement also applies to other categories as well, like appliances or cars. Finally, like I said before, the BLS breaks out all the inflation figures by category, so if you really wanted to you could make a "bare minimum to survive" index or whatever, although your initial comment seems to take issue with how the individual categories are calculated rather than how the categories are weighted.
Alternatively, multiple cellphones are required to replace a single calculator because they don’t have nearly the lifespan. They need constant recharging as they aren’t solar powered. And that’s not even getting into the UI issues due to a lack of keys etc. Objectively cellphones are a different compromise not a strictly better replacement. (let alone vs a computer.)
So again these comparisons are very subjective especially when something has upsides at the cost of dramatically reduced lifespan. The simplest argument for the BLS’s bias is comparing how often they accused of over vs under estimating inflation.
PS: Did you consider how fragile that 100$ Android phone is in your comparison? If so how much should you adjust the equivalent price for that fault? Which is the core problem as it’s easy to compare numbers while ignoring say Windows 11’s built in advertising when comparing equipment over time.
>So again these comparisons are very subjective especially when something has upsides at the cost of dramatically reduced lifespan.
Yeah a phone might be more fragile than a pocket calculator, but if you need a phone anyways and it comes with a calculator for free, the "replacement" calculator is effectively free and has infinite durability. You can knock off some points for not having tactile buttons, but then again it being effectively weightless and volumeless seems like a benefit.
> The simplest argument for the BLS’s bias is comparing how often they accused of over vs under estimating inflation.
That's a poor heuristic because even assuming the BLS was 100% accurate, you'd expect more people to complain about CPI being higher than lower. Squeaky wheel gets the grease as the saying goes, and there simply isn't much to complain about for low inflation. There are certainly downsides, but those are more academic/abstract than having to pay for bread with wheelbarrows full of bills. Same goes for conspiracies, eg. "everything is getting more expensive and the government is covering it up!". From there, motivated reasoning takes over and it's just a numbers game.
Fair enough, though I was more speaking in the academic context.
Anyway, a much better example of what I am referring to would be the very long slow decline of Red Delicious apples. What started as as a very tasty fruit was optimized for retail sales until it became unpalatable. Year to year it’s not obvious at any specific point that the product is worse, but over longer stretches of time say 1970 to 2010 it became so.
This related to shrinkflation, just harder to quantify because so many different things can be adjusted. However, if you’re measuring inflation over longer stretches to validate BLS calculations you find this kind of systematic bias more obvious.
> And that's fine, because the CPI's job is to measure how prices have changed (ie. inflation), not wealth inequality.
Allow me to elaborate. Does it seem odd to you that "real wages" (i.e. inflation adjusted) have barely budged in decades, and even slowly increased? The wages aren't decreasing, despite people apparently being able to buy a nice house and live the American dream on a median wage up until around the early 90's.
The way CPI is calculated is that it surveys the median purchaser about their purchases (how much they bought and how much it cost). And as prices increase faster than wages, the median purchaser is spending a larger percentage of their paycheck on goods. More expensive purchasers / nicer goods are largely excluded from these surveys. For the last decade or so, it's probably pretty much all their money. That is NOT captured in CPI. They can only spend up to their entire paycheck, so inflation is limited to median wage growth. And then the hedonic adjustments can be used to help the government meet its targets ("when a measure becomes a target it no longer serves as an accurate measure").
