Is subsidized enrollment in the ACA marketplace really up since 2016?

Subscribe to xpostfactoid via box at top right (requires only an email address; you'll get 2-3 emails per week on average)

CMS's report on long-term enrollment trends in the ACA marketplace released this week emphasized the 40% drop in unsubsidized enrollment from 2016 to 2018. But the counterpoint came as something of a surprise: subsidized enrollment, according to CMS, is up since 2016.

If you look at the most often cited enrollment numbers for each year -- total plan selections reported by CMS annually at the end of the open enrollment season  -- total subsidized enrollment is down substantially -- 7.1% from 2016 to 2018, and 7.7% from 2016 to 2019. But average monthly enrollment was higher in 2018 than in 2016 -- and probably will be slightly higher this year.

It might appear that retention has improved -- more people stay in their plans for more of the year. But that's not clear, as average monthly enrollment in 2018 is not quite the same measure as in years prior.  Let's look at the numbers:

Changes in subsidized enrollment, ACA marketplace 2016-2019

Year	Subsidized* as of end of OE	Subsidized - avg monthly	Avg. monthly as % of plan selections
2016	10,525,955	8,248,839	78.3%
2017	10,100,808	8,025,959	79.4%
2018	9,770,291	8,356,247	85.5%
2018 v 2016	-7.2%	+1.3%	+7.2%

Source: CMS state-level Public Use Files

The problem with drawing conclusions from the contrast between signups as of the end of OE and average monthly enrollment is that in 2016 and 2017, open enrollment was not completed until Jan. 31, whereas it ended on Dec. 15 in 2018. Thus, in 2016-2017, some enrollees' coverage was not effectuated in January or February, and totals in those months don't reflect full enrollment. Since about 10-12% of those who select plans never make a first payment, you could say this "shorting" of average monthly enrollment is something of an illusion. But then, it's not clear either that first-month payment rates were better in 2018.  So where are we exactly?

Charles Gaba has charted (natch) effectuated enrollment month by month since 2014 - information pulled from the latest CMS enrollment snapshot. Unfortunately, there's no breakout for subsidized enrollment that I can find. The overall totals do show, however, that enrollment was higher in December 2018 than in December 2017 and December 2016. Those differences would probably be sharper for subsidized enrollees. 

One factor that very well may have improved retention in 2018 is silver loading -- a practice that began that year, after Trump cut off direct reimbursement of insurers for Cost Sharing Reduction (CSR) subsidies in October 2017. Since CSR is available only with silver plans, insurers were allowed to price the benefit into silver plan premiums only, which created steep discounts in bronze and gold plans in many states and regions. An awful lot of people got free or near-free bronze plans, and a free plan is easy to retain. A lot of people also obtained gold plans that cost less than silver.  

Then too, those who enroll in a shorter enrollment season are possibly more highly motivated and likely to retain their plans.  If that's the case, the higher retention may not show up in states that run their own marketplaces and maintain longer enrollment seasons. Maybe that's an examination for another post.

* Two notes re the enrollment totals reported above:

1. For 2016, subsidized enrollment was reported only as a percentage (83%) of total enrollment.

2. For subsidized enrollment as of the end of OE, I used the totals who received premium tax credits APTC) . A very small number of enrollees each year obtain Cost Sharing Reduction subsidies without APTC (22,000 in 2018). But that slightly larger total is not available for all states in 2016 and 2017.