Cross-posted at ACA SignupsAfter being cited by both the Washington Post as well as PolitiFact as one of only two sources for a reasonably accurate tally of ACA-enabled Medicaid/CHIP expansion (Avalere Health is the other one, although their analysis only included data through the end of December), I decided to go back and review my methodology. I've pored over both the HHS and CMS reports again, as well as reviewing the "Targeted Enrollment Strategy" and other special "bulk transfer" cases.
I also looked more closely at the reporting criteria of Washington and Colorado, both of which have (thankfully) gone above and beyond the other states in providing details on their own Medicaid/CHIP numbers, defining which enrollments are specifically for people who are legally eligible for Medicaid only due to the Expansion provisions. Finally, for non-expansion states, I took another look at this report which estimates South Carolina's (a non-expansion state) "woodworker" enrollments, as well as some additional information about the woodworker effect compiled by contributor Ruth37.
As a result of all of this, if you look at the current Medicaid/CHIP spreadsheet (Week 21), you'll see some significant changes down the right-hand column (you'll have to scroll horizontally if your display is less than 1,440 px wide, sorry!!). It now has Legal Expansion ONLY and "Woodworkers" clearly separated out from the total number of new enrollees (which also includes "baseline churn").
In some cases the data is precisely defined by either the CMS reports or the state Exchange (or state Health Services department) itself, such as Colorado, Washington and West Virginia. In cases where I don't have an exact number, I'm using the following rule for normal Medicaid/CHIP enrollments (ie, this doesn't apply to special "bulk transfers", which are all ACA-enabled by definition):
—Expansion States: 30% Baseline Churn, 20% Woodworkers, 50% Expansion
—NON-Expansion States: 85% Baseline Churn, 15% Woodworkers, 0% Expansion (obviously)
I want to emphasize this as strongly as possible: the above percentages are educated guesses only. I'm not saying that exactly 15% of enrollments in non-expansion states (or 20% in expansion states) are "woodworkers", and I'm not saying that the number of expansion-only enrollments in those states which are doing so is exactly 50% either. These ratios will be higher in some, lower in others. I may tweak these numbers more as additional data comes in. At this time, however, these represent my best estimates based on everything I've been researching and compiling over the past 4+ months. I think I've earned the right to be given the benefit of the doubt, but I could end up being dead wrong.
Having said that, the end result of all of the above gives the following estimates, which will either please or disappoint you depending on your perspective:
—Legal ACA Expansion ONLY: 3.27 Million (was 2.6 Million)
—Expansion + Woodworkers: 4.29 Million (was 4.8 Million)
In other words, the lowest estimate has gone up about 670,000 (yay!), but the highest estimate has come down about 510,000 (boo!)The good news either way is that I'm pretty confident that I've finally come up with a fairly reliable formula for removing the "baseline churn" which has been causing so much noise to drown out the signal, to use Nate Silver's lingo :)
I'll have a better idea of how reliable this methodology change is after the January CMS report is released, which I'm hoping will be any day now.
Please note that I will likely be removing the "top line" portion of The Graph (currently displayed as 7.16 million) as a result of this new methodology, which would result in the appearance of a dramatic 2.88 million drop on the Medicaid side. It's important to understand that this does not change the number of total, woodworker or even expansion-only enrollments, it simply removes the "baseline churn" portion which shouldn't be included anyway, but which I wasn't able to come close to defining previously.UPDATE: OK, here's The Graph, updated to remove the baseline churn and with the new methodology baked in: