
We are doing some market analysis in Texas and surrounding states. One of the issues is to identify populations that might be potential purchasers of a particular offering. That is at least partially a function of income.
The graph below shows estimates of real per capita income trends within Texas household income quintiles.

To get this, we started with state income distributions from the U.S. Bureau of Economic Analysis (BEA) at https://www.bea.gov/data/special-topics/distribution-of-personal-income. This provided nominal incomes by household income quintiles for
- Total personal income
- Net earnings by place of residence
- Proprietors’ income
- Net compensation
- Dividends and interest income
- Rental income
- Personal current transfer payments
For this graph, we didn’t work with any of the detailed categories. We stuck with total personal income.
Data came in a zip file with data for every state from 2012 to 2022. There were separate workbooks for every state. For every state there were separate worksheets for every year. Job one was to extract the data and combine all the years for Texas.
The downloaded data was not adjusted for inflation. We could easily see that some quintiles had seen income growth. With others, however, we could not immediately see if that was growth or if that was inflation. Step two was to download Consumer Price Index (CPI) data and adjust all of the years and quintile values to 2022-equivalent dollars. CPI data is available for download at https://www.bls.gov/cpi/data.htm. We used data for all urban consumers in the Southern region of the U.S. We used annual measurements that were not seasonally adjusted.
With inflation-adjusted data for quintiles of Texas households, we still could not see if individuals were gaining or losing ground. This is because every year the quintiles each give data for one-fifth of the households, but we have no idea of household or population growth.
We made a simple assumption that households averaged the same size across all five quintiles. That allowed us to take annual Texas population estimates divided by five as the number of people in each quintile. Dividing inflation-adjusted quintile incomes by population gave us the per capita income estimates shown in the graph. We utilized Texas population estimates from the BEA at https://www.bea.gov/data/by-place-states-territories, because data from the BEA is remarkably easy to locate, download, and use.
There are a few things about the data and the data manipulation that deserve note.
First, for every year the total income received by the top quintile was greater than the income received by the bottom four quintiles combined. This was not changed by any of the manipulations described above.
Second, the assumption that household sizes are the same across all quintiles was convenient and gave us the ability to normalize the data for population size but is probably not completely accurate. For any quintile where household sizes are larger than the overall average, the quintile’s per capita estimate would shift down. Conversely, for any quintile where households are smaller than the overall average, the quintile’s per capita estimate would shift up.
Our best guess is that the lower quintiles have larger households and that the higher quintiles have smaller households. This is consistent with the demographic arguments in the recent post, “The Coming Depopulation.” If so, the lines for the bottom quintiles would drop and the lines for the top quintiles would rise.
Third, the data estimates current realized income. That is pretty close to total income for the bottom quintiles. Households in the upper quintiles, however, are likely to have significant levels of unrealized unearned incomes in the form of appreciation or capital gains on investments. These streams are reported and show up in the data as they are realized. If they are realized in a constant steady stream over time, the data is probably an accurate reflection of reality. To the extent that unrealized income streams are growing over time, the data will underestimate them during any period.
This was an interesting exercise undertaken as part of a larger analysis of market potential in the Southern U.S. It is possible to replicate this for any state and to engage the data at a more specific level. While multistate regions can also be analyzed, they require additional manipulation because income ranges on household quintiles will be unique to every state. In all cases, a careful disclosure of assumptions made and the potential implications of those assumptions is required.