In which Megan McArdle & Bob Laszewski have an interesting theory (Updated)
2018 MIDTERM ELECTION
Time: D H M S
While I've been pretty much vindicated regarding the (eventual) 1st month premium payment rates, off-season enrollment rates and monthly attrition rates for exchange QHPs, there have still been two items which have bothered me.
The first is what I've been complaining about since May 21st, when the HHS Dept. announced that they would be discontinuing their official monthly QHP enrollment reports. Advocate of the ACA I may be, but stopping issuing these reports was simply a bad policy decision regardless of how good or bad the off-season enrollments have been. The fact that the numbers since mid-April have actually been very good (around 9,000 per day enrolling on average) makes it even more perplexing that they'd stop reporting them.
Frankly, I’m kind of surprised that this is the first time we’ve heard those numbers. They’re pretty good. Oh, sure, they’re not as good as the optimists hoped, but it’s on the low end of estimates from industry expert Bob Laszewski, and certainly a lot better than the numbers we’ve been hearing from insurers. I expected them to be a lot worse, based somewhat on the insurer statements, but also because I figured the administration would have released them if they weren’t really awful. Obviously, the administration would prefer to tout the 8 million enrollment figure, but it has still exceeded the enrollment projected by the Congressional Budget Office for 2014. Why not stand proud?
OK, she actually mixes some flat-out wrong statements in the above quote as well: The numbers are actually better than optimists like myself hoped (I was assuming the currently-enrolled number as of August would be around 7.0 million, not 7.3 million), and it's actually pretty much on-target with "the numbers we've been hearing from insurers." However, her larger point here is one I agree with 100%: The lack of enrollment data since the 4/19 report has made no sense whatsoever.
The second thing which has bothered me is this: A 2% monthly average attrition rate actually sounds a bit low, even to me. There, I said it. The attrition numbers I've compiled so far have ranged from as low as 0.33% per month in Maryland to as high as 7.7% in Nevada...but my general sense has been that it's likely to be around 3% overall.
And here's where both McArdle and Bob Laszewski (also an ACA critic) may have put together a possible answer as to why HHS/CMS went from dead silence on QHP enrollments since mid-April to unexpectedly releasing the currently enrolled/paying number last week. Both make similar points, but since I quoted McArdle above, I'll quote Laszewski this time:
As I thought of the administration's release of 7.3 million it occurred to me that they said the 7.3 million included all enrollments through mid-August.
Under Obamacare, after a person has paid their first premium, a health plan can't cancel anyone until they have gone three months without making a payment.
So, they are effectively double counting by including the "adds" while also keeping the "deletes." The 7.3 million figure includes all of the enrollments that have occurred through mid-August. But the number also still includes every person who has failed to make a premium payment in June, July, and August––since the carriers can't yet knock them off the rolls. The health plans tell me there is a 2% to 4% monthly attrition rate.
Now, before ACA supporters start thinking that I've Turned To The Dark Side or whatever, take another look at what I said above: Until the 7.3M figure was released on Thursday, I had been operating on the assumption that the average attrition rate of around 3% per month; that's where I got my 7.0M current-as-of-August estimate from. When CMS announced the actual August number as 7.3 million, I dropped the attrition rate down to 2% and voila, it matches up.
What McArdle and Laszewski are arguing here is that the CMS Dept. deliberately picked August 15th as the "currently enrolled" number to report (as opposed to an earlier or later date) because they expect this to be the "high point" for the year: The point when off-season additions are at their highest level RELATIVE to the number of people dropped due to more than 3 months of non-payment of premiums.
If I'm understanding them correctly, they're essentially arguing that a whole bunch of the people who enrolled in late March and early April--whose policies didn't actually take effect until May 1st--are about to be dropped off the rolls due to never paying their May premiums (but that the insurance companies had to wait until after mid-August to actually drop their coverage due to the 3-month rule).
If this is true--if, say, only 80% of the March/April enrollees have been paying up instead of 90% of them--then that would mean that as many as an additional 290,000 people have been dropped since August 15th (the date of the 7.3 million figure announced last week).
If that's true, it would bring the September "currently enrolled & paying" figure down to...wait for it...7.0 million people....exactly what I was assuming in the first place using a 3% average monthly attrition rate instead of 2%.
In addition, of course, there's also the matter of the 115,000 enrollees who have some sort of immigration/citizenship data issue yet to be resolved. Some of these enrollees may end up being kicked off of their plan on September 30th, either due to them not qualifying legally or never bothering to send in the proper paperwork at all. Assuming that, say, half end up with that outcome, that's another 58,000 to lop off the total.
Here's the thing, though: Even if that's the case, the November 14th currently enrolled/paying tally should still end up being around 7.1 million people using a 3% average attrition rate...still 100K above the "magic" 7 million figure:Furthermore, even using the high end of Laszewski's "2-4%" range at 4% per month, the November tally will still be almost 6.6 million...a bit below the 7 million figure, but still well above the CBO's revised 6 million projection from February, which took the early technical problems into account:
In other words, even assuming that McArdle/Laszewski's theory is correct, the HHS Dept. has been playing dumb for the past 4 months for no good reason. They've refused to release any numbers since the report which came out in early May (which cut off the data as of April 19th), raising all sorts of suspicion, criticism and doubt from detractors and supporters alike when there was no reason to do so.
And if McArdle/Laszewski's theory is incorrect, then the radio silence since May makes even less sense.
UPDATE: Hmmmm...as Ken Kelly points out in the comments:
The no-payment crowd, however, were never covered - there's nothing special about Aug 15 for them. That's your 10% never-paid group. However, M&L seem to believe that many people paid in May, and then missed June, July and August, but were included in the 7.3M "paid up" figure. This smells a bit desperate to me, particularly given that:
a) This applies only to people getting advance premium tax credits and
b) HHS is informed as soon as one payment is missed. Tavenner was being rather dishonest if she called these people "paid up".
I'm not sure that HHS actually is informed whenever a payment is missed (at least not "as soon as" it happens...although I suppose this would have to be the case for those who receive their tax credits monthly instead of all at once at tax time), and the tax credit factor would still apply towards about 85% of those who enrolled, but this is still worth considering.
In addition, Ken's point re. this only applying to those who made their first payment but then stopped paying for the next 3 months is a valid one as well. I'm sure some folks fit into this category, but it seems unlikely that there would be that many. Finally, as I noted in response to someone else,
IF the "3 month delay" theory is correct, the numbers UP UNTIL NOW would all look very good indeed. It's not until AFTER August that they would seem to drop off significantly.
In other words, politics would explain not releasing any numbers AFTER August (until after the election), I suppose, but not beforehand.
So...the mystery continues.