What's the relationship between real dollar adjustment and inflation?

October 21, 2021 1:30 PM

Is the adjustment used for computing "real dollars" in econometric datasets the same as compensating for inflation, or is it something different? For example, if I look up real median US household income on FRED, I see that the value for 1984 is $53,337 in 2020 USD. If I then use an inflation calculator, I see this has an equivalent buying power as $21,068 in 1984 USD. Am I correct in understanding that the median household income in 1984, represented in 1984 dollars, is $21,068?
posted by biogeo to Science & Nature (7 answers total)

Roughly, but the methods for calculating inflation/CPI/etc have changed over the years, and different datasets/calculators use different methods. So, unless your calculator is using the same math as the dataset, you won't necessarily get the exact same 1984 number when you work backwards.
posted by Jairus at 1:41 PM on October 21, 2021 [2 favorites]

No, because ‘real dollars’ is measured in changes in the prices of various commodities, not wages (because one of the most important things it measures in these datasets is comparative living standards, i.e. what people can and can’t afford).

There’s no fixed relationship between what people earn and what they can actually use that money for, since wages rise and fall. In the US, they’ve fallen very firmly in real terms since 1984.
posted by Fiasco da Gama at 1:43 PM on October 21, 2021 [2 favorites]

Best answer: Why not just use the median household income from 1984 directly? This shows $22,415. The difference is because the chart is using CPI-U-RS and the calculator you are using is based on CPI-U, which is slightly different based on changing methodology. If you look at the CPI-U-RS data, you'll see that you can calculate exactly $22,415 from $53,337.
posted by ssg at 1:44 PM on October 21, 2021 [8 favorites]

Best answer: As Jarius said, you have to use a consistent measure of inflation. On the real median income page you linked to, it says: "Units: 2020 CPI-U-RS Adjusted Dollars, Not Seasonally Adjusted". The inflation calculator you linked to uses CPI-U, so the results would be inconsistent. (As explained here, here, the CPI-U-RS is generally considered more accurate than the CPI-U.)
posted by Mr.Know-it-some at 2:00 PM on October 21, 2021 [1 favorite]

Response by poster: This is great, thanks all! So if I'm understanding correctly, the adjustment based on CPI-U and CPI-U-RS are indeed both measures of inflation, but I was making an error in using two different measures?

Thanks for the link to the unadjusted median household income dataset, ssg. I'm not that experienced searching on FRED and failed to find that one despite looking for it!
posted by biogeo at 9:24 PM on October 21, 2021

That's right. The CPI-U is just the annual estimate of inflation made every year — it does not change later. The CPI-U-RS is adjusted to re-estimate past years' inflation with current methods. So if you're looking for the most accurate estimate of inflation, the CPI-U-RS is your best bet (but as you see, the difference is not huge even back to 1984) on the assumption that methodology now is better than it was in the previous decades. The differences are more important if you are comparing inflation across decades, where if you used CPI-U, you'd wouldn't be comparing apples to apples because the methods were different in past decades.

As I understand it, the business of estimating inflation is kind of tricky and requires adjustments to keep the estimate from appearing to diverge from reality (and even then, economists will argue about details). It's an estimate, not an inarguable measurement of something factual.
posted by ssg at 10:33 PM on October 21, 2021 [2 favorites]

Great answer, but the sound you hear is hordes of economists laughing at your understatement: "the business of estimating inflation is kind of tricky."

Inflation would be easy to calculate if:
1) Everyone bought the same goods and services (or at least in the same proportion).
2) No one ever changed the types or amounts of stuff they bought from month to month.
3) No new goods or services were ever sold and all goods and services remained the same (no improvements in quality).
4) The relative prices of all goods and services changed at the same rate, rather than some prices increasing at different rates or decreasing.
5) Everyone shopped at the same stores and paid the same prices, and they didn't change where they made their purchases.

Since none of those are remotely true, the overall inflation measure has to adjust for all of those factors.
posted by Mr.Know-it-some at 11:08 AM on October 25, 2021