New Data on ACA Enrollment Shows Problems in Many States

The Department of Health and Human Services released updated data yesterday on enrollment on the Exchanges including, for the first time, greater breakdowns on enrollment by several key categories: age, gender, and the metal level of purchase.

The result of this long awaited and much requested data is, at first glance, very much a mixed picture. Some of the overall statistics do not look as problematic as some — including me — had feared they might be. But it looks as if there is a very serious potential for large adverse selection problems brewing in a number of states,  most notably West Virginia, Mississippi, Maryland and Washington State.

The good news for the ACA from the data

There are three major pieces of good news for those who support the goals of the ACA.

1. The overall gender distribution of enrollees, 54% female, 46% male does not appear on preliminary inspection to be sounding “red alert.” To be sure, the problem may be a little greater than would otherwise be suggested by the aggregated numbers if the middle age group is more heavily female and the oldest group of enrollees more heavily male that the aggregated numbers suggest.  And Mississippi is troubling with 61% female enrollment (and for other reasons, see below).

But, overall, and if they hold up, these do not appear to be the the kind of numbers that would be way beyond what insurers likely expected or that, standing by themselves, would be devastating to an insurer on an Exchange.

2. Several states have total enrollments and the age distributions that should reduce the possibility of a serious death spiral getting started. New York and California are the two big states doing better than most.  Connecticut is doing very well also.

3. The metal tier distribution is 80% for Bronze and Silver policies and only 20% in Gold and Platinum. That’s comforting for adverse selection. A higher proportion of enrollment in the more generous plans would have been a warning sign that enrollment was coming disproportionately from the sick.

There’s a footnote on this point later on — we are not out of the woods — but this is definitely better news for the ACA than a distribution of, say, only 50% Bronze and Silver purchases.

The bad news

Just because the ACA is doing better than some had forecast on an overall basis does not mean there will not be very serious problems in some states.  Given that the statute is presently unamendable as a practical matter, problems in just a few states can hurt a lot of people.

The data released by HHS yesterday shows that there are a number of states in serious trouble.

The table below shows current enrollment in states that appear at first glance to be in significant trouble. (There may be states that are as bad as those shown in the table below; I kind of eyeballed the data, which was not provided in a nice electronic format, to select what appeared to be a problem).

I also show the projected number of purchasers, the percentage of current enrollees as a fraction of purchasers, the percentage of “young invincibles” as a fraction of the enrolled population and the percentage of the expensive 45-64 year old age group.


I also construct a “back-of-the-envelope”/quick and dirty “Maelstrom Index” that roughly measures the problems confronting the individual Exchanges in the state.  The maelstrom index is calibrated to run from 0 to 1 with a 0 score meaning the state is in excellent shape and a 1 meaning there is an extremely high risk of an adverse selection death spiral materializing.  It is not a probability measure.

The formula for the maelstrom index is shown below.

Max[0, (5*(1.4 – enrollmentPercentage – youngInvinciblePercentage))/7]


As one can see, there are a number of states that have serious potential problems. West Virginia is problematic with 66% of its enrollees ages 45 to 64 and only 17% ages 18-34. Almost all experts would concede that this poses serious risk of adverse selection. And the overall fraction of enrollment relative to projections, just 20%, suggests that West Virginia’s insurance pool may be disproportionately populated by the sick. Mississippi is likewise a problem with enrollment only 14% of that projected, 58% in the age 45-64 bracket and, as mentioned above, 61% female enrollment.

I have to believe that insurers writing in Mississippi are deeply concerned about these numbers.

Also  troubling are the numbers in Texas, Washington State, and Ohio simply because, as an absolute number, there is such a large absolute difference between the number of people projected to purchase and actual enrollment. I’ll speak more about all of these numbers and try to update the chart in the days ahead.

The unknown

There’s one key piece of data not found in the HHS release yesterday. The number of “enrollees” who have actually purchased policies.  If this conversion rate is, as I fear, hardly 100%, the numbers presented today overstate the extent of purchases in the Exchanges and understate the dangers of adverse selection breaking out in many states.

Seth J. Chandler is a Professor of Law at the University of Houston at the University of Houston and author of acadeathspiral.org, where this post originally appeared.

7 replies »

  1. AgenteNews piece you linked is not germane to the major elements of the discussion re: ACA enrollments. Actually comes off if indirectly as a “puff piece” endorsing the Law (don’t “rock the boat” no matter how bad/dismal the impacts), emphasize the positives regardless how many are made worse off.

  2. John: Nice “deflection” attempt. But just dissing those who pursue intelligent (and very useful) inquiry into certain important facets of ACA w.r.t its financial integrity perhaps amounts to little more than doing the “whistling-in-the-dark” thing? And as for Klein’s piece you kindly provided link for — pay special attention to his Item #7 therein. That is the true canary in the mineshaft for ACA exchanges — unfortunately the relevant data on the important enrollment dimension discussed there (i.e., distribution of enrollees re: health status) are not yet available for analysis, not to Klein, Levitt (KFF), Laszewski, anyone. Hence little substantive assessment to date beyond some very general mentions. Lacking the relevant data of course does not render that important factor as something to dismiss so casually.

  3. Another key piece of information left out is the number of enrollees that had previous coverage (not just the recently canceled). I will bet a significant portion of these Connecticut enrollees had policies but “re-enrolled” through the public exchange for lower premiums because they had relatively higher premiums that they were managing to pay (higher due to existing conditions and compounded year over year increases) either with or more likely without taxpayer subsidies, because they are relatively wealthier.

  4. Good suggestion. I’m tired of beating dead horses.
    Let’s pay more attention to science (or a large and growing anti-science population), best practices, the importance of advance directives, the madness of marketing designer drugs/devices during the evening news…

  5. I think THCB needs to rename itself to “THIB.” The Health Insurance Blog. Pretty much all that’s being posted and discussed any more.