Credibility & Accountability

This morning, The New York Times ran an Op-Ed by Steven Rattner, including an impressive infographic (the type that represents the U.S. population as lines of silhouette icons representing 1 million people apiece), breaking out the entire country by what type of healthcare coverage they currently have. The point, of course, was to demonstrate just how many people either have healthcare coverage at all thanks to the ACA or how many have improved coverage thanks to it. Every data point presented is credited directly to this website.

I'm not saying this to brag (although it is pretty mom is pretty geeked about her kid's work showing up in the Sunday New York Times). I'm mentioning it because I was expecting to receive a flood of attacks on the data and my credibility. There's a bunch of numbers listed there, and the "these are rough estimates" disclaimer at the bottom is pretty innoucuous. Instead...crickets.

Don't get me wrong--I've received some traffic from the Times piece, and there's over 160 comments, many of them heated, on the story. However, the traffic has been a small fraction of that usually generated by, say, Paul Krugman giving me a shout-out...and as for the comments, out of curiousity I decided to run a search through all 165 of the ones there as of this writing to see how many of them either attack or ask about the source for the data.

I figured most people wouldn't even see the "Source:" link at the bottom ("Where is this Rattner guy getting these numbers?"), and that those who did see the link would post stuff along the lines of "What the hell is" or "Is that run some blogger in his mom's basement?" or "that guy's just a partisan Democratic hack", etc. etc.

Instead, while there’s plenty of debate, discussion & argument about other aspects of the law—narrow networks, tax implications for the middle class, whether deductibles/premiums are increasing, whether it’s “just another giveaway to moochers”, the fact that other numbers haven’t been included, the presumption of calling discontinued plans “inferior”, etc etc...unless I’ve missed them, I can’t find a single comment which has actually questioned the data itself, as presented.

On the one hand, this is flattering. On the other hand, it's also a little disturbing.

A year and a half ago, when a few media outlets first started noticing this site, my wife jokingly suggested that perhaps they were starting to use me as their source. I, in response, tongue-in-cheekly snarked that perhaps I should just pull some completely made-up number out of my ass, just to see if anyone notices. I was kidding at the time.

These days, it's not so funny. For good or for bad, this site has become a Reliable, Trusted Source by many serious news media outlets. When I post a new data point, there's a lot of people watching and taking it seriously. I'm proud of this, but it's also a responsibility I treat with respect.

Therefore, please allow me to clear up a couple of points:

There has been one person today who has challenged an actual data point in the New York Times piece: Chris Conover, the Forbes contributor and American Enterprise Institute scholar, with whom I've butted heads on this site in the past.

Over the past year, Conover and I have developed a sort of grudging respect for each other. Earlier this evening he inquired as to where the "No longer uninsured: 13.4 MILLION" data point came from, since, as he noted, most of the respected national surveys from Gallup, Urban Institute, Commonwealth Fund and RAND Corporation have put the net reduction in the uninsured at roughly 11-12 million. He was wondering where the other 1.4 - 2.4 million in the Times came from, so here it is:

  • First of all, none of these surveys include children (presumably because they can legally only survey adults over 18). I assumed that adding in newly-covered children could add anywhere from several hundred thousand to perhaps 1 million to the tally. Mr. Conover stated that:

However, Urban Institute figures show that the uninsured rate for children was essentially flat between the first six months of 2014 (7.3%) relative to the same period in 2013 (7.4%). Anyone with the most rudimentary knowledge of statistics will recognize that this is not a statistically significant change, but even if we accept it as such, that trivial decline would add only 74,000 to the ranks of the previously uninsured (according to Census (Table 2), there were 74.055 million children under 18 in March 2014)

A sidenote: The 74,000 figure above was originally incorrectly listed as 5,400. As noted in Mr. Conover's update this evening:

Update #3: February 22, 2015

Charles Gaba who runs ACASignups. net pointed out an error I’d made in Footnote 3. I’d originally stated that even if we accounted for the statistically insignificant 0.1% decrease in the uninsured rate for children, it would add only 5,400 to the ranks of the previously uninsured. But he pointed out that I’d incorrectly applied the percentage change to the number of uninsured children (5.4 million) rather than total number of children (74.1 million). I appreciate his detecting this goof and have corrected the footnote.

Of course, Conover's larger point, about a 0.1% reduction being statistically insignificant, would still hold true. However, assuming that the "first six months" of each year was measured as an average across the time period, this is an apples to oranges comparison, for the following reason: More than half ot the 2014 QHP enrollments didn't actually become effectuated until either March or April. Out of the 8.02 million total QHP selections through April 19th, 2014, 500,000 were under 18 (2.7M were 0-34; 2.2M were 18-34).

84% (6.7 million) of the 8.02 million total were still effectuated by the end of 2014. Assuming this was even across all age groups, that means that about 420,000 children were enrolled in exchange QHPs. Further assuming that the 57% previously uninsured rate for adults found by the Kaiser Family Foundation also held true across age groups, that means a good 240K of these kids were newly insured. Of course, children are far more likely to already be insured than adults, but even if we cut that number in half it's still 120,000.

In other words, there's strong evidence to suggest that somewhere between 100 - 200K children were newly insured via the exchanges last year who aren't accounted for by these surveys. On the one hand, that's a lot less than "up to 1 million". On the other hand, that doesn't include Medicaid/CHIP (OK, let's just use CHIP for now since we're talking about kids), and it's still a lot more than nothing.

  • Next up: Medicaid/CHIP. The various national surveys that Mr. Conover refers to (Urban Institute, Gallup, etc) generally only run through the 3rd quarter of 2014, and in any event, the most recent monthly Medicaid/CHIP report only runs through the end of November. What's happened since then? Well, every month last year, net Medicaid/CHIP enrollment went up by at least a few hundred thousand people. All the data so far for 2015 suggests that this trend is continuing for now...and why wouldn't it? States like Pennsylvania have only recently added Medicaid expansion (admittedly, the 134K who signed up by early January are running into confusion due to former GOP Governor Corbett's messy "alternative" system, which is, thankfully, being converted back to "standard" Medicaid expansion by new Democratic Governor Wolf). Michigan, which was supposed to top out at no more than 500K Medicaid expansion enrollees, currently sits at 560K, up some 90,000 people since the end of November, and so on. Add them all up and there should be 2 million+ more people added to the Medicaid/CHIP roles as of mid-February than there were as of November. How many of these folks are newly insured? I don't know, but if it's even 25% that's another 500K or so.
  • Third: Actual QHPs (either on or off the exchanges). Of the 11.4 million confirmed exchange QHP selections for 2015, around 6.3 million were renewals from 2014 (either auto or active). That leaves around 5.1 million new entries. Of those, about 88% (OK, at least 87%) either have already paid their January/February/March premiums or will do so very soon. That's about 4.4 million people. How many of those folks are newly insured? Well, last year the KFF survey placed it at 57%. This year, a similar rate would be about 2.5 million. Even if you cut that in half, however, that'd still be at least 1.25 million newly insured Americans as of March 1st, when the latest batch of enrollments kick in.
  • Finally, note that the Times folks decided (I agree with this) to leave out some other smaller numbers--2015 SHOP enrollments, for instance (I doubt this totals more than perhaps 200K nationally right now), a million or so on the Indian Health Service and so on. (As an aside, I also urged them to completely scrap the whole "under 26 on their parents' plan" category, since just about all of that group are already represented in the other categories anyway).

So, add all of this up and you're talking about a good 2 million additional newly insured people on top of the 11-12 million noted as of last fall. That would bring the net grand total to somewhere between 13 - 14 million people. Split the difference and you get...13.5 million.

Now, there are a couple of legitimate quibbles here:

  • The disclaimer at the bottom states: "Obamacare figures are rough estimates for the period 2014 through Feb. 20, 2015". Since a couple million of the current QHPs + Medicaid enrollments don't actually go into effect until March 1st, they probably should have noted that clarification.
  • Speaking of the disclaimer, I would have been more comfortable with placing it more prominently, along the lines of: "Some numbers are rough estimates; there’s a lot of constant churn between each of the categories."
  • I suppose it would have been better of Mr. Rattner to go with a range of 13-14 million (or even 12-14 million if he wanted to be extra-cautious).
  • Finally, I admit to not being entirely comfortable with the labelling of non-ACA compliant policies as "inferior plans" (instead of simply "non-compliant"), although I didn't push that point. Obviously some people disagree as to whether their discontinued policies were "inferior" or not.

Other than this, however, I stand by everything in the NY Times piece. As I stated to the folks who approached me about using my data:

The way you have it now is as close to the as-of-today accurate state of affairs as I can estimate.

We're talking about 320 million people. The population has gone up several million since the ACA was enacted. Teenagers have turned 18. 25 year-olds on their parents' plans have turned 26. 64 year olds have turned 65 and moved to Medicare. People have moved, gotten married, gotten divorced, gotten jobs, lost jobs. Some have moved to Medicaid; some have moved to a new state; some have died.

The bottom line is that it's impossible to keep track of the status of that many people every moment. I think I've done a pretty good job of accurately representing the situation as it stands as of this point in time.