Huh. State-based exchanges appear to be better at retaining enrollees year-round than HC.gov.

OK, this was a bit unexpected.

Last week I analyzed the monthly enrollment effectuation numbers through all 4 years of ACA Open Enrollment (2013 - 2017), to help visualize how the enrollment numbers flow throughout the year (and to also show how misleading the Trump CMS "attrition" report from last week was).

Then, on Thursday, I wrote about a CMS powerpoint presentation from last year which provided a Holy Grail of sorts for me: Not only did it provide the monthly effectuation numbers for 2015, it broke them out by starting and ending month. That is, it divided the numbers into how many enrollees for that month had originally enrolled starting in January, in February and so forth. This is really, really important because having "an average" of 100,000 enrollees each month doesn't mean much if it's a different 100,000 people every month--that is, if a lot of people are signing up, having expensive treatments and then dropping out and then replaced by someone else, that hurts the risk pool, since the original batch aren't paying for the other months of the year when they don't require treatment.

Here's what the table looked like for 2015:

The only catch here is that this table only includes HealthCare.Gov states, not the more than a dozen other states which operate their own exchange sites.

Now, 8,838,291 people actually selected policies via HC.gov during the 2015 Open Enrollment Period out of 11,688,074 nationally, or roughly 75.6% of the total. I naturally assumed, therefore, that the monthly retention/attrition rate would probably be around the same...that the numbers in the table above should represent around 76% of the national total each month.

It turns out this is wrong.

While HC.gov absolutely accounted for ~76% of enrollments during OE2, the RETENTION RATE for HC.gov states vs. the SBMs deviates greatly. Take a look at the monthly NATIONAL effectuation numbers for 2015 from my 2013-2017 post. I have estimates for 8 months, but the March, June, September and December numbers are official from CMS itself:

• March: 10,187,197 nationally vs. 7,320,968 on HC.gov: HC.gov made up 72% of the national total.
• June: 9,949,079 nationally vs. 6,871,008 on HC.gov: HC.gov only made up 69% of the national total.
• Sept: 9,313,323 nationally vs. 6,393,564 on HC.gov: HC.gov only made up 69% of the national total.
• Dec: 8,780,545 nationally vs. 6,052,168 on HC.gov: HC.gov only made up 69% of the national total.

In other words, the SBMs are (or at least were, in 2015) doing a MUCH better job of retaining enrollees throughout the year than HC.gov does/did. How much better?

• In March, HC.gov states had 7,320,968 effectuated enrollees out of 8,838,291 OE2 signups (82.8%)
• In March, SBM states had 2,866,229 effectuated enrollees out of 2,849,783 OE2 signups (100.6%)...yes, thanks to SEP additions (and Massachusetts' special year-round "ConnectorCare" policies), the SBMs managed to have more people enrolled, paid and effectuated in March than had officially signed up as of the end of Open Enrollment.

• In June, HC.gov states had 6,871,008 effectuated enrollees (77.7%)
• In June, SBM states had 3,078,071 effectuated enrollees (108%)

• In September, HC.gov states had 6,393,564 effectuated enrollees (72.3%)
• In September, SBM states had 2,919,759 effectuated enrollees (102.5%)

• In December, HC.gov states had 6,052,168 effectuated enrollees (68.5%)
• In December, SBM states had 2,728,377 effectuated enrollees (95.7%)

As noted above, Massachusetts accounts for part of this, since they allow people to enroll in some categories of exchange policies year-round rather than limiting them to the official enrollment period, but that only accounts for a small part of this deviation, since MA only makes up around 1.2% of national enrollees and only 5% of the SBM enrollments. The only other factor I can think of which makes sense is that the SBMs might have been stricter about enforcing/verifying SEP enrollments than HC.gov in 2015?

Anyway, it's an interesting little quirk.