A CMS misinformation byte is getting into the woodwork

Axios today reports on the Kaiser Family Foundation's latest data note on individual market enrollment. The main takeaway is that unsubsidized enrollment is down by two million in 2018: brutal premium hikes are driving the unsubsidized out of the individual market.  I was brought up short, however, by this offsetting claim:

Subsidized enrollment grew by about 500,000 people.

Kaiser here is retailing CMS's comparison of effectuated enrollment as of March 2017 and March 2018, tallying those who enrolled on-exchange paid their premiums in February.  These snapshots show total on-exchange enrollment at 10.6 million in March 2018 and 10.3 million in March 2017, and subsidized enrollment at 9.2 million in 2018 vs. 8.7 million in 2017.

As I noted three weeks ago, that comparison is erroneous:

As Charles Gaba pointed out last year (and revisits here), however, the 2017 "snapshot" exaggerated early attrition by failing to take into account the fact that those who enrolled between 1/15 and 1/31 (the final day of OE in 2017) did not have payments due until March 1.  There were 539,352 enrollees in that time frame.  None of them could effectuated their coverage for February, which is the population counted in the "snapshot." If those enrollees effectuated coverage at the same rate as enrollees before 1/15 (88.5%), there were 10.8 million who had effectuated or would soon effectuate as of the time of CMS's tally. That total outstrips this year's by 2% .

Kaiser notes the error, rather vaguely, in a footnote (#4):

Open enrollment for 2017 continued through the end of January, but for the 2018 plan year it ended on December 15, 2017. It is likely, therefore, that effectuated enrollment for 2017 was somewhat understated relative to 2018.

The error carries over to the totals for subsidized enrollment, which Kaiser, relaying CMS's reported totals, reports increased by 522,000 in 2018 (rounded by Axios to 500k). If 84% of the late enrollees who paid their first premium were subsidized (the percentage reported for subsidized enrollment overall), about 400,000 should be added to the subsidized enrollment total for 2017, leaving subsidized enrollment up by about 122,000 this year.

The error may not seem that consequential -- but as I noted in the prior post, CMS leadership has exploited it for political purposes for two years running. In 2017, the text prefacing the effectuated enrollment snapshot highlighted first-month attrition, which the report exaggerated. This year, CMS flipped the script to claim an enrollment gain, which it used to justify its Draconian cuts to the ACA navigator program.

It is disturbing to see this error, which Gaba flagged more than a year ago, get into the health news bloodstream.