Discussion about this post

User's avatar
Econoboi's avatar

How did I get these numbers?

I decided to put this in the comments for all you sick freaks who actually care about methodological details. The data here are from the BLS’s ‘Consumer Expenditure Survey,’ with data from 1996 to 2020.

I take the nominal decile figures for pre-tax income and divide that by the square root of each decile’s household size to get the ‘equivalized’ household income for each decile. I do this because it’s a good way to control for household size when thinking about ‘true’ relative income levels. The reason I don’t just divide by total household size (choosing its square root instead) is that households share resources.

Imagine a single person in household A vs. a family in household B. If household B dedicates $700 in income to buy a stove, the same $700 stove enriches all three household members. If the person in household A buys a $700 stove, he only enriches himself. Taking the square root of household size adjusts household B for its shared resources so as to not overstate how much they’re splitting their income between non-shareable resources (like bottles for only the child or a gym membership only for Dad).

Astute readers will have noticed that decile data is only available as of 2014, so in order to get decile-level data pre-2014, I take the nominal quintile data from 1996-2013. Then I calculate the growth rate going backwards in quintiles, so if the bottom 20th percentile was $10,000 in 2014 and $9,800 in 2013, this registers as around a 2% decrease (from 2014 to 2013). Then, I apply this growth rate to 2014’s bottom 10th and 10th-20th percentile data to get 2013’s numbers. I repeat this step for all growth rates and years to get each decile’s nominal income from 1996-2013.

In terms of my own household income, I combined tax return information (as far back as I had available) with personal adjustments I knew existed at the time, like child support payments in or out of the house, personal transfers from family, etc. In years where I don’t have tax data, I use recollection and some rough inflation adjustments to get both my household income and size for each year.

Admittedly, the further back I go, the fuzzier the data becomes, but I tried my best to avoid any gross misrepresentations.

Nonfon's avatar

Really enjoyed this!

9 more comments...

No posts

Ready for more?