The Wall Street Journal looks to have used an interview by Ezra Klein on Obamacare tech woes as the basis for a reported piece on how insurers were having so much trouble with the Federal marketplaces that they were having to check and in some cases process enrollments manually. That’s clearly unworkable at anything resembling the scale of expected participation. But not only does the Journal article corroborate and add more color to an ugly story, but it also mentions a different type of problem, that of eligibility, in passing. That’s actually a hugely important and presumably separate subroutine in the system, and if that is also broken or buggy, it has serious implications of its own. The Journal reporters appear not to have understood the significance of what they were told about eligibility, so they didn’t follow up on in their account, but we’ll point out what the consequences of glitches on this front might be.
Klein’s interview with Robert Laszewski of Health Policy and Strategy Associates described the problems he was hearing from insurance companies, and what that implied about the IT systems. The critical section:
The insurance industry is literally receiving a handful of new enrollments from the 36 Obama administration-run exchanges. It’s really 20 or 30 or 40 each day through last week. And a good share of those enrollments are problematic. One insurance company told me, “we got an enrollment from John Doe. Then five minutes later we got a message from CMS disenrolling him. Then we got another message re-enrolling him.” On and on, up to 10 times. So insurers aren’t really sure if the enrollments they’ve got are enrollments they should have.
And remember, the insurers have automated all this. They don’t have a clerk sending out a welcome letter and an enrollment card. So if you just let the computer run, it could theoretically issue a welcome letter, a cancellation letter, a welcome letter, a cancellation letter, etc. Now, they’re not doing this right now because it’s all screwed up. They can manage a few dozen per day by hand. But when you’re talking about thousands or tens of thousands or hundreds of thousands, it becomes completely unmanageable.
The Journal confirms this picture:
Insurers say the federal health-care marketplace is generating flawed data that is straining their ability to handle even the trickle of enrollees who have gotten through so far, in a sign that technological problems extend further than the website traffic and software issues already identified.
Emerging errors include duplicate enrollments, spouses reported as children, missing data fields and suspect eligibility determinations, say executives at more than a dozen health plans. Blue Cross & Blue Shield of Nebraska said it had to hire temporary workers to contact new customers directly to resolve inaccuracies in submissions. Medical Mutual of Ohio said one customer had successfully signed up for three of its plans.
The flaws could do lasting damage to the law if customers are deterred from signing up or mistakenly believe they have obtained coverage.
The worst of these problems is that the insurer gets bad information. Some may be resolved when the IT glitches are cleaned up (one executive reports having enrollees show up with multiple spouses). But others may be due to problems with data integrity and database integration, and they are more difficult to remedy. Again from the Journal:
As more of those users attempted to sign up for plans this week, insurers began noticing problems with enrollment data. For now, they say they are largely able to manually correct the errors. But as enrollment increases—up to 7 million consumers are expected to sign up in the next 5½ months—that may not be possible, they worry.
Scott & White Health Plan in Temple, Texas, has received 25 enrollees from the federally run exchange so far. “There are some missing data elements that are requiring a lot of research on our part,” said Allan Einboden, the health plan’s chief executive. “If we’d received 5,000 and they all had to be worked, that’s a lot of extra administrative costs,” said Mr. Einboden, who said he expects the problems to be fixed…
At Priority Health in Michigan, health-plan staff are calling new customers to confirm each of their “couple of dozen” enrollees accurately picked the plan, said Joan Budden, chief marketing officer, after realizing some had enrolled in multiple health plans, likely owing to user error linked to slow healthcare.gov response times. “Sometimes they pushed the [submit] button three times,” Ms. Budden said.
Lambert jumped on the last issue:
You’ve seen this happen, every so often, on blogs like Corrente. I’m tellin’ ya, if somebody gave me a billion, or a million, or a hundred thousand, or a thousand, or a couple hundred bucks to stop double posting at Corrente, it woudn’t be happening. No matter how slow the site is (and Corrente is no speed demon).
So did all the programmers for the Federal Exchange take the brown acid? Was there a spec? Was there testing? This is just so bad!
Notice that for every insurer that the Journal contacted, the fix was a manual intervention or even an manual enrollment. Now some glitches may recur reliably so that the insurer can override or automate how it deals with them once it gains experience. But if this continues, and I can’t see how it can get much better unless the folks in the Administration admit what a mess this is, take the project off line, test and patch it up a ton and relaunch, we’re going to get an amusing side effect: the insurers will be hoist on their own petard.
Under the ACA, a stipulated percentage, 80%, of the insurance premiums are required to go to health care. That was presented as a boon to consumers (“health insurer profit margins are capped”) when it was intended to be a gimmie. Health insurers have been price gouging for decades. In the early 1990s, they spent 95% of health care premiums on medical expenditures. As of the passage of the ACA, it was down to under 85%. But if they get stuck with a lot of unexpected enrollment expenses, bye bye profit margins! Remember, the ACA, despite its billing, was designed to enrich the health insurers and Big Pharma. The insurers would gain more customers, both through the subsidies and fines inducing more customers to sign up for policies. So their top line would increase and they expected to have comparable, maybe even better profits on the new business (they are pushing more volume through largely existing infrastructure). But a boatload of extra up front costs will complicate this pretty picture.
Now of course, you might say, the solution for the insurers is simply not to help out the enrollees all that much. Just bounce all but the cleanest enrollments that get through and make it the Administration’s responsibility to deliver them decent data. But that intersects with another problem: the people who will absolutely insist on getting enrolled will be the sickest. That is the biggest nightmare in insurance, being stuck with the worst risks, or what is known as adverse selection. Thus the insurers need to get as many people through the process as possible and do what they can to minimize the pain for customers so that some of those young, healthy people they really wanted to get on their books do indeed sign up.
Let’s turn finally to the additional potential huge can of worms that the Journal tripped over but apparently did not probe. And credit to Lambert for catching it. Here’s the key sentence from the top of the second paragraph, emphasis ours:
Emerging errors include duplicate enrollments, spouses reported as children, missing data fields and suspect eligibility determinations, say executives at more than a dozen health plans.
Remember how Obamacare works. In an normal online shopping experience, you poke around a site or several sites, compare products and pricing, pick one and check out, which is when you provide payment information. The Obamacare system puts the payment part, which is eligiblity, up front, because it is eligibility for subsidies. They did not want people in income categories that would be eligible for subsidies first looking at plans and going, “OMG, that plan for our family will cost $1200 a month” when that was the pre-subisdy cost and they need to put their income parameters in to see what the net cost to them will be.
Remember, people below a certain income level are not eligible for Obamacare but instead shunted to Medicaid. If you are eligible for Obamacare, you get a subsidy if your income is below 400% of the poverty line, or $45,960 for an individual and $94,200 for a family of four. See here to see the various buckets of subsidies (scroll down to the table under “The Federal Poverty Guidelines Help to Determine Subsidies”).
Now it is possible that the “suspect eligibility determination” is due merely to determined shoppers managing to submit an enrollment prematurely while inputting approximate information that was too approximate. As Lambert wrote:
To calculate subsidies, the Exchange just needs some parameters or, in the vulgate, some boxes to fill in on the online form: Income, for example. Those parameters do not need to be validated — and it’s validation, not eligibility calculation, that is the cause of the bottleneck.* Just feed some fake parameters to the eligibility engine, and get back a number.
So some customers may think just like Lambert and are causing problems, which might point to other deficiencies (as in people may wind up submitting enrollments when they think they are just shopping, which could be the result of bad interface design).
But it could be that there is something more seriously wrong with the enrollment engine. It has to be fairly pronounced for the insurers to pick it up. How would insurers detect a “suspect eligibility determination”? Are they seeing enrollee’ estimated incomes? Why should that even be something the insurer is even aware of? You would assume the path is: Customer enters income and other eligibility information. HHS system slots individual/family into appropriate subsidy bucket. Consumer then pokes around among various plans and gets prices adjusted for his subsidy level. Customer ALSO gets to see whether he is eligible for “Advance Tax Credits” which means all he pays his insurer is that net price. When he files his tax return for the year, if his income on his tax return differs enough from what went in the enrollment form to put him in another bucket, he will have gotten too much or too little subsidy. That difference is applied to his taxes.
That’s a bit long-winded, but you can see where this is going. First, for insurers to detect eligibility issues, it would seem likely that they are getting the entire enrollment file, including the income information. I don’t like the idea of that from a privacy and security standpoint. Bad enough the NSA knows everything, now your insurer is in that loop too? Second, the errors have to be both pretty gross and reasonably frequent for the insurer to notice and be concerned about them (after all, in theory if the customer does not pay the rack rate, the Administration picks up the rest of the tab, so you would not expect them to be looking at this information all that closely).
Now the error could be a simple formula error that is dumping people into the wrong subsidy bucket. Gross errors like that would produce ugly results but would not be hard to fix. But if it’s not something like an easily-remedied computation mistake, the ramifications would be large. A big raison d’etre for Obamacare was the subsidies, both from the public’s and the insurers’ perspective. If that part of the software doesn’t work well, you have the prospect of consumers getting stuck with IRS bills that are wrong or unexpected (as in they provided the right information but the subsidy was calculated incorrectly). And where does the poor insured go then? HHS apparently will have an Obamacare ombudsman, but how much power will it have, and will that office have remotely the needed staffing to deal with this sort of mess?
While the discussion above may sound unduly pessimistic, Lambert’s big mistake on Obamacare so far has been in failing to anticipate that the start-up would be as horrifically screwed up as it has turned out to be. Based on results to date, the worst case scenario is likely to be the actual state of play. On the eligibility matter, we’ll see soon enough whether Obamacare’s Federal exchanges launch is continuing to run true to form.