ACS Data Users Group

Hello Fellow Users,

I’m working through ACS 5-Year Estimates from 2019, looking for a way to get an estimate of demographic counts (or percentages) under a certain income limit (80% Area Median Income to be specific) to analyze racial outcomes for a HUD CDBG program in Indianapolis.

I can calculate the last bit pretty easily but I’m having a problem locating the intersection of a racial identifier (ex. Black) and any (continuous or categorical) measure of income.

If I had source data I would be able to create my own categorical variables for income under 80% AMI and each of the racial groups I’m interested in looking at. Am I able to get anything like this from publicly available ACS sources? I've looked around the MDAT but there don't seem to be any good measures of income that I can utilize.

Anyone have an experience with a similar problem? These sort of "racial program audits" are becoming more popular after 2020. I could settle for the overall population percentages but I think that leaves out the important caveat that low income and non-white racial groups are positively correlated. Many hud programs serve predominantly low-mod income persons; if I use overall population percentages won't it seem like we are helping "more minorities" than would be expected from a random sample.

Any comments or thoughts on this subject would be appreciated.

Rez

Hi Rez—

You won't find what you seek thru data.census.gov.

What you're looking for is the Comprehensive Housing Affordability stats database (known fondly as "CHAS"). It's put together by the Special Tabulations team at Census, and then published on HUDuser.gov https://www.huduser.gov/portal/datasets/cp.html

The most recent database is always 1-2 years older than the most recent ACS base tables. But it does provide what you're seeking: incomes and rents recategorized into percent-of-AMI bands.

Get familiar with the CHAS data by opening the "Data Dictionary" https://www.huduser.gov/portal/datasets/cp/CHAS/data_doc_chas.html

Then obtain the tables you need with the Data Download utility https://www.huduser.gov/portal/datasets/cp.html#2006-2017_data

By the way: One of my colleagues has R programming that merges together some of the separate worksheets into a single table. If that's of interest, inquire by email.

--Todd Graham

Metropolitan Council Research

There are a couple aspects of the CHAS data about which users should be aware and that are not well documented.

If you are familiar with how federal housing funds works, you probably know that what we call "80%" of Area Median Income (AMI) is often not actually 80% of the AMI figures released by HUD. This is because under certain circumstances the 80% AMI figure is subject to a national cap. Similarly, I believe that the 50% figure is capped at times. However, in other cases the "caps" can be higher than the 50% or 80% figures you would get using just multiplying the the AMI. Here is an example from the city where I work (Note that the 100% of HUD median figures in the table are the 100% values published by HUD):

https://www.cambridgema.gov/CDD/housing/resourcesandadditionalinformation

The income levels used to produce the CHAS data are based off the 50% and 80% figures for the local metro or state taking into account the cap. If you try to reproduce the numbers using PUMS data that uses "actual" 80% you may end up with widely differing percentages than you will find in the CHAS data.

Another aspect of CHAS that is poorly documented, and which we learned about several years ago after tracking down the person at HUD who was then responsible for producing the CHAS at the time, is that the income cut offs used in the CHAS are derived from proportions of the 80% figure. What this means is that if the 80% figure for your metro is "actually" let us say 75% of AMI, then the income cut off used to produce the 100% cut off is 75% * 1.25 = 93.75%.