Improving the Affordable Care Act Markets (Part 2)


In a previous post, I described how some features of the Affordable Care Act, despite the best intentions, have made it harder or even impossible for many plans to compete against dominant players in the individual and small employer markets. This has undermined aspects of the ACA designed to improve competition, like the insurance exchanges, and exacerbated a long term trend toward consolidation and reduced choice, and there is evidence it is resulting in higher costs. I focused on the ACA’s risk adjustment program and its impact on the small group market where the damage has been greatest.

The goal of risk adjustment is commendable: to create stability and fairness by removing the ability of plans to profit by “cherry picking” healthier enrollees, so that plans instead compete on innovative services, disease management, administrative efficiency, and customer support. But in the attempt to find stability, the playing field was tilted in favor of plans with long-tenured enrollment and sophisticated operations to identify all scorable health risks. The next generation of risk adjustment should truly even out the playing field by retaining the current program’s elimination of an incentive to avoid the sick, while also eliminating its bias towards incumbency and other unintended effects.

One important distinction concerns when to use risk adjustment to balance out differences that arise from consumer preferences. For example, high deductible plans tend to attract healthier enrollees, and without risk adjustment these plans would become even cheaper than they already are, while more comprehensive plans that attract sicker members would get disproportionately more expensive, setting off a race to the bottom that pushes more and more people into the plans that have the least benefits, while the sickest stay behind in more generous plans whose premium cost spirals upward. Using risk adjustment to counteract this effect has been widely beneficial in the individual market, along with other features like community rating and guaranteed issue.

However, in other cases where risk levels between plans differ due to consumer preferences it may not be helpful. For example, it has been documented that older and sicker members have a greater aversion to change (changing plans to something less familiar) and to constraints intended to lower cost even if they do not undermine benefit levels or quality of care, like narrow networks. These aversions tend to make newer plans and small network plans score as healthier. Risk adjustment would then force those plans to pay a penalty that in turn forces enrollees in the plans to pay for the preferences of others.

If new plans, or innovations to control costs and avoid over-utilization such as narrow networks and tiered pharmacies were not in the public interest, then those penalties could be justified. But since the ACA relies on contracting strategies available to private plans (not rate setting) to control costs, penalizing such innovations through risk adjustment will undermine competition and efforts at cost control. These constraints may also actually improve the quality of care and quality of life by avoiding unnecessary care and the complications it can create, and better coordinating care among more integrated providers. There is evidence in Massachusetts and elsewhere that risk adjustment penalties due to these preferences harm the public interest.

A small insurer that contracts with the same providers as a large one is going to pay more for each service provided, especially at hospital systems and medical groups that have a strong position when negotiating rates. A careful analysis of the New York market found that this was likely the single largest driver of losses among start-up health plans, but the risk adjustment model makes no accommodation for it. The only way for a smaller plan to partially counteract this large plan cost advantage is to have a more narrow network and form a close alliance with a limited number of provider partners. In doing so, that insurer will exclude some important institutions in the local area, and people who are least amendable to this tend to be older and sicker. The risk adjustment model makes no allowance for this selection preference.

The increased willingness of younger and heathier people (and small businesses) to try something new could help mitigate the benefits of size and incumbency, but the current risk adjustment model throws up roadblocks to this opening for innovation.

The measurement methodology itself also creates issues. As mentioned in my previous post, inevitably a dominant insurer will pull the average statewide risk score to be closer to its level than competitors, and what can seem like a small payment to it is huge to its competitors as a share of their premium. The largest incumbent plans also have a lot of historical data needed to effectively confirm each year what conditions (scorable risks) its members have, and sophisticated operations to find and confirm new diagnoses every year, as the ACA requires. This results in plans receiving higher risk scores than others that do not reflect true underlying differences in risk.

So, what can be done? Below are three suggestions for further discussion:

  • Add Value Control Factor – A Value Control factor could be established for select innovations that address cost and quality but may be selected against by enrollees with higher costs. Such factors could include small integrated networks, restriction to in-network benefits, reliance on a PCP to coordinate care, and tiered formularies. The amount of any risk transfer payments from a plan with eligible cost/quality controls could be reduced by this factor, in order to reduce penalties for innovative solutions that align incentives more closely between payers and providers, avoid waste and inflation, and improve collaboration in care management. This factor would be empirically established in each state, and if there is no measured impact then no factor would be applied. Plan designs that simply increase member cost-sharing would not be eligible for a Value Control factor.
  • Provide Relief for Small Plans – If a plan in a highly concentrated insurance market has less than a certain market share (e.g., 5%), it should be allowed relief from some of the burden of risk adjustment transfers. Plans could be made to opt-in to this program in advance to avoid gaming, and market share should be set at the level of the insurer in a particular market (namely the individual or small group market) rather than a specific insurance product. This relief is necessary to counter some of the advantages larger plans typically have from their long-tenured membership, favorable contracted rates with providers, and close relationships with brokers who steer coverage decisions.

The ACA has done well given the limitations it has faced. Millions more Americans have insurance than before, and those with existing health conditions are freed from the fear of being rejected for coverage simply because they are sick. Even under an administration hostile to it, enrollment has been stable for those receiving subsidies on the exchanges. However, the unsubsidized individual market and the small group market have been shrinking, while the dominant insurers keep getting more dominant. We face the prospect of stagnant marketplaces with shrinking competition, higher premiums and fewer enrollees. HHS has repeatedly invited states to submit their own proposals for alterations to the federal risk adjustment program to suit the conditions of each state, though so far only Alabama (in 2018) and New York (in 2017, as an emergency measure) have done so. It is better late than never for states to take HHS up on its offer.

Jonathan Halvorson is a Senior Healthcare Consultant at Sachs Policy Group and has a long-term interest in the transformative potential of technology on the health care system

8 replies »

  1. Barry, there are also at least 80 categories used in CMS’s HCC risk scoring framework. Last count I saw was 83. The goal for each system is to predict future costs based on past data (diagnoses, demographics, costs per service, etc.), and any such system creates opportunities for participants to maximize their score within the rules. The trick is to try to align these practices with actions that benefit members in terms of better health or quality of care, rather than just rewarding those who chase charts the best.

  2. Spike, seeing this a little late, but totally agree on all points.

  3. I tend to agree with both those points: reinsurance would be more efficient at the federal level, and we should remove the hard stop to subsidies at 400% FPL.

  4. My understanding is that Germany’s Central Fund uses 80 different factors to determine risk scores. If it’s working OK for them, why can’t we just copy that and build on it from there?

  5. I think you’re spot on to identify poor risk management models as a key challenge to the success of the ACA’s exchanges. Part of it is that it’s a concurrent risk model that only takes into account diagnoses identified that year. Part of it is the “zero-sum” nature.

    Medicare Advantage does not use a “winner-take-all” risk management calculation and uses both prospective and concurrent risk models to make it more fair. You’ll notice that most health plan startups are starting with Medicare Advantage first. I believe the better risk scoring model in the Medicare Advantage market is a key reason we’re seeing so much innovation there. That said, United Healthcare was the subject of a $1B qui tam lawsuit regarding accusations of fraudulently “up-coding” the risk of their populations to get larger risk adjustment payments from Medicare. That case was dismissed under seemingly suspicious circumstances… but if United is deploying the same tactics they used to get a little extra money from the federal government to bankrupt competitors, that’s a major issue.

    The future of healthcare is some version of the ACA where a government-sponsored system is carried out by private insurers with a risk scoring algorithm sitting in the middle as an umpire. That much is clear. Working out the risk scoring model to be harder to game and fairer to all competitors is critical to any plan that hopes to care for the sick and involve private carriers.

  6. I think reinsurance should be done at the federal level with one uniform set of rules regarding the amount of the attachment point and the percentage of costs above the attachment point that reinsurance will pay for. It’s very hard for some of the more financially strapped states to justify spending state money for reinsurance so the federal government can save money on subsidies. It’s perfectly logical that money spent on reinsurance should reduce aggregate premiums by approximately the same amount.

    At the very least, we should remove the 400% of FPL income cap to qualify for an ACA subsidy. I don’t think anyone should have to pay more than 10% of what Medicare calls modified adjusted gross income to pay for a silver level health insurance policy. This would especially most of the older couples who have not yet aged into Medicare but can be charged up to three times as much in premiums as younger people are charged.

  7. Barry, I mostly disagree. If just two competitors have something like 80% of the market, that is not healthy and usually does not work well for consumers. Many of your examples are not really apt (Target and Walmart together have far less than half the market share for the consumers segments they compete in: clothing, electronics, groceries, etc. They compete with Costco and Sams Club and Amazon, not just with each other.). When Walmart does dominate a small town by bankrupting small competitors, famously its prices are no longer as low. More generally, sometimes specific regions do have greatly restricted choice and the lack of competition, just like economics 101 says, are generally higher than in comparable areas that have more competition.

    You may be thinking of data which shows that when a health insurance market is competitive but the provider market is not (say, 6 insurers with between 5-25% market share each, but one hospital system has 60% market share and the other has 40%) then the larger number of plans isn’t better but actually results in higher prices. There is evidence this is true, so my comments on health insurance competition should be qualified to refer to environments in which there is not a strong imbalance in relative competitiveness.

    Reinsurance is a good idea and it is already being done in around 12-14 states where the risk pool was really deteriorating, so I see these as complimentary strategies. When it is done to stop runaway growth, it actually mostly pays for itself from a federal perspective (because tax credits are tied to premium cost, so lowering premium saves federal expenditures. The ratio of reinsurance spend and premium savings can be nearly 1:1). Look at Alaska, for example.

  8. I don’t think we need more competition in the health insurance market from smaller competitors. Between 35%-40% of the market is already controlled by non-profit insurers, mainly the Blues not owned by Anthem, Kaiser, Harvard-Pilgrim, Medica, Puget Sound and others. Even the for profit insurers including United Healthcare, Anthem, Aetna, Cigna and Humana all have pretty low pretax profit margins.

    We could lower premiums by using federal reinsurance to cover most of the cost of the highest cost claims like we already do for Medicare Part D but that would cost a sizable amount of money. Ditto for high risk pools which would cost even more money and, at least at the state level, politicians have never been willing to spend the money to finance effective high risk pools for all who needed them because they felt it was too much money to spend on too few people.

    In the retail world, just two competitors can provide more than adequate competition. Think Walmart and Target, Costco and Sam’s Club, Home Depot and Lowe’s, Walgreen’s and CVS, etc. Even supermarkets usually have only two or three competitors in most regional markets.

    I don’t think lack of competition among health insurers is anywhere near as big a problem as you suggest. Instead, high hospital prices extracted from private payers by powerful health systems, high drug prices, especially for specialty drugs, and obnoxiously high bills from out of network doctors and ambulance companies are bigger problems that need to be addressed.

    It’s interesting to note that in both healthcare and higher education, costs have gone up significantly faster than general inflation for decades. Both of those industries have the highest government involvement as either a payer or a lender or both.