How many could lose coverage in your CONGRESSIONAL DISTRICT?


For weeks now, I've been painstakingly compiling the estimated number of people whose healthcare coverage would be at risk if the ACA is indeed repealed, by state and by county.

While my state-wide estimates cover all 50 states (+DC), I've only been able to compile this data at the county level for 35 states (DC is basically one big county anyway).

Now, by popular demand, I'm presenting my estimates of how many are at risk of losing coverage by U.S. Congressional District.

Unfortunately, like the county-level data, estimates for some states are harder to compile than others. The Kaiser Family Foundation has done the tedious work of piecing together ACA exchange Qualified Health Plan enrollment data from every zip code in each Congressional District for every single state, which is awesome, but it still takes a lot of tedious work to put together (the KFF data is embedded within an interactive map but not available for CSV/text file export, so I have to dig through every CD in every state individually).

In addition, the KFF analysis doesn't include Medicaid expansion enrollees in those zip codes/Congressional Districts, so it's much easier to run estimates for the states which didn't expand Medicaid under the ACA (plus states with only a single district, duh!)

UPDATE 2/04/17: In addition to the image versions below (suitable for sharing on social media), I've also ported all Congressional District-level data into a GOOGLE DOCS SPREADSHEET version to make it easier to make corrections/updates as well as so people can more easily copy/paste data for their state/CD themselves.

Here's the basic methodology:

For ACA Exchange enrollments:

  • Start with the raw number of QHP selections via the ACA exchanges as of 1/31/17
  • Subtract the percentage of unsubsidized enrollees (that is, people paying full price for their policies)
  • Subtract another 10% on top of that, since some people are only receiving nominal tax credits (these are people who would be inconvenienced by losing the tax credits but would most likely be able to scrounge up enough money to hold onto their policies)
  • Finally, I knock another 10% off of that number to account for the ~10% of enrollees who simply never pay their first monthly premium anyway, and are therefore never actually enrolled in an effectuated policy in the first place.

Nationally, this adds up to roughly 8.3 million people out of the expected 12.3 million people who selected QHPs during the 2017 Open Enrollment Period; an additional 4 million will either be paying full price or only receiving modest tax credits.

For Medicaid Expansion enrollments:

  • The statewide numbers for Medicaid expansion are far more precise, as they come directly from the various state health departments or ACA exchange reports.
  • Breaking these out by Congressional District, on the other hand, is much trickier, since none of the state/exchange reports break them out by CD. Instead, I have to figure out which CD every county is located in, then add up the the county-level data.
  • This is made even more difficult by the fact that most counties are split into multiple CDs thanks to gerrymandering/etc. In those cases, I'm dividing the county totals by the number of CDs it's split into.
  • For instance, Oakland County, Michigan has 53,868 people enrolled in ACA Medicaid Expansion. Oakland County is sliced into 4 separate CDs (MI-08, 09, 11 and 14), so I add 13,467 people into each of those Districts.
  • Yes, I realize this is a simplistic system, since there's no guarantee that all 4 "chunks" of the county have equal populations, but its' the best I can do. If you have a better method, feel free to contact me, but remember I do this on my own time.

All that being said, here's what we're looking at:

  • UPDATE2/17/17: CALIFORNIA now added.
  • UPDATE 2/19/17: HAWAII AND IDAHO now added. (After weeks of having to track down the individual congressional districts by rolling over the appropriate part of the map--which wasn't easy in many districts due to gerrymandering--it looks like the KFF has added the actual raw source data, making it much easier to finish up the remaining states!)
  • UPDATE 2/19/17: INDIANA now added.(Important caveat: The Medicaid Expansion total may be ~74,000 too high; will correct if I receive confirmation of this)
  • UPDATE 2/19/17: IOWA now added.
  • UPDATE 2/19/17: LOUISIANA now added.
  • UPDATE 2/19/17: NEVADA now added. 
  • UPDATE 2/19/17: MASSACHUSETTS now added. Two important things to note re. Massachusetts: First, until recently I was under the impression that MA had already expanded their Medicaid eligibility beyond the ACA's 138% FPL threshold anyway, and that therefore there weren't any MA residents enrolled specifically due to the ACA; it turns out that I was very, very wrong about this. According to this January 2015 MassLive article, at least 190,000 Baystaters would indeed be kicked off of Medicaid ("MassHealth") if the ACA were to be repealed.
  • The second Massachusetts oddity is that the Kaiser Family Foundation analysis has all 9 Congressional Districts having the exact same number of exchange enrollees as of last March. Obvoiusly this isn't the case. KFF rounded each county off to the nearest 100, but it still seems unlikely that all 9 within the state would be that close (within +/- 50 people) regardless of the demographic/income similarities. However, in the absence of any other data sources, I have to use that as my as a result, my own estimates of both high- subsidy exchange enrollment and Medicaid expansion enrollment are also identical for each of the 9 CDs.

With the addition of Massachusetts, I've now analyzed all 50 states + DC. I'll probably be making some additional minor revisions as corrections/updated data comes in, but this should pretty much complete the project.