Risk Adjustment Gone Wrong

The Affordable Care Act was intended to usher in a new era of competition and choice in health insurance, and at first it succeeded. But increasingly, provisions in the law are undermining competition and wiping out start-up after start-up. If something isn’t done soon, the vast majority of new insurers formed in the wake of the ACA will fail, and many old-line insurers that took the opportunity to expand and compete in the new markets will leave. It’s a classic story of unintended consequences and the difficulties of regulation.

Flush with optimism after the ACA passed, dozens of new insurers formed to take advantage of the environment created by the law. Twenty three of these were co-ops given start-up funding by the ACA. In most states the new plans only grabbed a small share of the market, but enough to put pricing pressure on larger incumbent plans. In a few states, like New York, the start-ups and other new entrants grabbed over half of the business on the exchanges.

To the surprise of many, price increases in health insurance remained low by US historical standards even as the recovery continued and people who had been without insurance were finally able to get it. How much of that modest cost trend is due to an improved competitive marketplace on the exchanges is speculation, but what is clear is that the doomsayers about the ACA were wrong. Costs did not explode, and even with higher 2016 rate increases we are not back to the bad old days (yet).

One of the cornerstones of the ACA, however, has had the opposite effect intended, and it is now threatening the destruction of much of the good that has happened to health insurance competition and choice. The ACA had three measures to stabilize the exchanges and improve competition—reinsurance, risk corridors and risk adjustment. Reinsurance has worked largely as intended to mitigate the effect of unexpectedly high claims. Risk corridor payments were similarly intended to soften the highs and lows of bad guesses about medical costs, but were dramatically reduced by Congressional Republicans. This played a role in forcing about half of the co-ops to close by the end of 2015. It was a deliberate political outcome rather than a regulatory misjudgment. In any case, both of those programs were intended to be temporary until insurers could gain experience in the new post-ACA environment. They end in 2016, so their benefits and harms have essentially been delivered, unless Congress is willing to reverse course and fully fund the risk corridors before time runs out.

On the other hand, the risk adjustment payments are permanent, and their scope expands beyond the exchanges to all individual and small group health insurance. These risk adjustment payments are having a large and surprising impact on insurers across the country for reasons that sometimes have no real competitive justification, making some companies that are already dominant and profitable even more dominant and more profitable.  The point of giving a bonus to plans with higher than average risk and penalizing plans that enroll members with lower than average risk is to prevent the business strategy of skimming the market to find the lowest risk enrollees. The idea is to prevent health insurers from profitably pricing below their competition because their products attract members who use fewer health care services, and force insurers instead to compete on the quality and efficiency of services delivered. Is that happening?

Take New York, where I live and work. UnitedHealth, the largest insurer in the US, is also one of the largest players in the NY individual market and is by far the largest player in the small group market (under the name ‘Oxford’). It is profitable locally and nationally, and has been for some time. In the small group market, it did not participate in the new exchange at all, yet it was due a net $281 million in risk adjustment payments. In the individual market, it had limited participation and was due a net $86 million in payments in 2015. The next highest beneficiaries of payments received $13.5 (small group) and $48 million (individual).

In the small group market in New York, about 9 of every 10 dollars paid in risk adjustment went to United/Oxford in 2015, largely for clients that United had before the ACA took effect and that it continued to serve off the exchange. There is no change in these payments based on profitability, so a company breaking even or losing money because it priced aggressively to grow is still forced to pay, while a profitable company will still receive money so long as it has members in higher risk demographics or with higher risk diagnoses.

A key feature of the way the risk adjustment program works is that the payments come from other plans in the same state. It is a zero sum game. So, other plans in New York gave their largest competitor a total of about $367 million dollars in 2015. This was also the year that the fastest growing ACA co-op in the nation, HealthRepublic of NY, went bankrupt owing its competitors $191 million in risk adjustment payments, because the formula determined that its members had health risk that was too low. It owed about 37% of its total 2015 premium revenue. Even if HealthRepublic did not have other problems, it could not have long survived these kinds of payments in a business (health insurance) that, despite its reputation, typically has a profit margin of around 5%. It could easily have taken a decade for Health Republic to make back in profit what it lost in one year’s worth of risk adjustment. Of that $191 million owed in risk adjustment payments, the vast majority was slated for United and its Oxford subsidiary. Similarly, other smaller plans trying to expand the small group market have been paying out millions.

In New York’s individual market, the second and third largest recipients of risk adjustment funds have been venerable Blue Cross/Blue Shield plans (Excellus and Empire, a subsidiary of Anthem). Blues plans in many states are the ones getting the best shake out of the risk adjustment program. In contrast, those paying out the large sums have mostly been smaller competitors that are either start-ups, or non-profit Medicaid plans that expanded into the individual commercial market. The companies that are paying are often losing money even before the payment. How is this helping competition or stability?

The problem is by no means isolated to New York. On August 1st, New Mexico Health Connections (NHMC) and Minuteman Health of Massachusetts sued the federal government, charging that the risk adjustment program unfairly penalizes new entrants and advantages old line insurers that have long dominated the individual and small group markets. For 2015, NMHC was required to pay out 15% of its total premium in risk adjustment payments. In the case of New Mexico, and many other states, the primary beneficiary is the local Blue Cross/Blue Shield plan.

Similar scenarios (and in some cases, lawsuits) are playing out in Maryland, Washington, Oregon, California and other states.

What Went Wrong?

Risk adjustment works relatively well in other nations and in the Medicare program, so why does it seem to be failing here? I don’t think anyone has the full answer at this point, but a few trends seem to be emerging.

Risk adjustment requires an insurer to report on the health risk of its members, and to do that it needs good data. Plans that played the game better from the start set a high priority on collecting and reporting on that information. However, it is much harder to get good data if a member just joined than if you have had that member enrolled prior to the ACA exchanges and can mine your data warehouse for all those ICD codes that boost the risk score. The dominant pre-ACA players had more years of member data, and mature analytics capabilities, to help them maximize their risk scoring. This has created a serious penalty for new entrants in the first few years which CMS has not addressed.

The Maryland co-op suing the federal government has taken this concern a step further and alleges that incumbent insurers actively encouraged their members to seek services in order to be diagnosed as higher risk so that the plan would receive increased risk adjustment payments. This may be straying into tinfoil-hat territory, but the potential for some gaming is there.

More important than the reporting issues, some of the difference in member risk between insurers is also real, but it is not clear that the difference is well-captured in the risk adjustment model, or that it should always generate a payment.

For example, another advantage of incumbency is that older, higher-risk people are more likely to want to stick to names they have known for years in health insurance (e.g., the local Blue Cross/Blue Shield plan) and not want to take a chance on upstarts. Brokers in the small group market may steer clients towards the companies they have long been most comfortable doing business with. These forces of inertia create a barrier to newcomers, which is normal, but it gets compounded as an issue because for each sicker person who doesn’t switch to a new or less-known brand, it contributes to the financial transfer payment the upstart plan owes under the risk adjustment program.

It’s also not clear if some plans are being penalized for being too good at preventive health. The CMS model does not seem to be able to tell the difference between a plan that has low risk people because they enrolled that way, and one that has low risk people because it works more effectively to prevent chronic diseases and cancers. It’s unlikely this is a major factor, just because so few health plans actually seem to be able to move the needle on preventive health care. It may, however, be an issue in some cases and certainly could become a bigger issue down the road.

Finally, and this is the real wild card, products that attract sicker patients have historically tended to be those with richer benefits, larger networks, lower penalties for going out of network, and fewer approvals to get care. That’s understandable, but having people gravitate to such plans will make controlling costs harder. And by rewarding plans with less cost-control that attract higher risk patients, the ACA may be inadvertently incentivizing health insurers to promote such products because they can predict that the plans with worse cost control will be subsidized by the stricter cost-control plans that attract healthier people. That’s a scary thought for the future of premiums on the exchanges, and doesn’t seem to have been accounted for in the policy papers outlining the risk adjustment goals and methodology.

With the ACA’s risk adjustment, it is possible that there will be a perpetual skew of risk towards products that are less well-designed to control it, which is reinforced by the risk adjustment mechanism in a manner that doesn’t allow more tightly controlled products to compete well, with the result that certain products are pushed out of the market, competition is lessened, and premiums have less pricing pressure.

All of the issues described above are fixable, but time is rapidly running out to do so after the damage of the risk corridor reductions, and now these large and ongoing risk adjustment payments. The seven remaining co-ops are hanging by a thread, with one or two exceptions. Firms that expanded into new markets with the ACA have had to retrench or reconsider further expansion. Even well-capitalized start-ups can’t continue forever to hemorrhage dollars through risk adjustment payments to their competitors. It’s hard for David to win in a fight against Goliath when you risk-adjust the rock out of his slingshot.

All that said, it is true that the Goliaths are not all standing tall. While United has done well in New York, it made headlines a few months ago when it announced it was losing about $500 million annually on the exchanges and was pulling out of most states. Aetna and Cigna, other insurance giants with well known brands, have also experienced losses and are pulling back on expansion plans. The Blues, including Anthem, have generally been winners, but even adding profits from the Blues and the few others that have stayed in the black, the average plan had a net loss of 4% on the individual market in 2014. The small group market has remained profitable at about a 3% margin overall, though as the example of New York shows there are big winners and losers hidden in the average.

The data is clear that insurers have been playing an aggressive game on pricing, especially on the individual exchange. This can’t continue much longer, and plans will need to price to maintain a margin of 3-5%. With the current risk adjustment model giving such large and hard to predict results, achieving pricing discipline to yield a stable margin will be a challenge, to say the least.

A fix to the risk adjustment model can’t wait. It will have to be thoughtful to minimize gaming and to achieve the objectives of bringing stability and encouraging competition based on quality and efficiency rather than actuarial risk selection. Perhaps a solution in the short term is to lower transfer amounts so their impact is lessened until the reporting and methodology can be ironed out. For a longer term solution, one of the more simple refinements may be to add a restriction that plans that are already profitable or have medical loss ratios below a certain percent (say 85%) should not receive these payments, or payments should be made on a sliding scale related to MLR. Similarly, unprofitable plans could be removed from some or all obligations of making risk payments. Likely, more radical changes are needed.

CMS is well-aware that there are problems. Let’s hope it makes enough changes before the innovations and competition among insurers that sprang up with the ACA whither and blow away.

Jonathan Halvorson edits the New Economy section for THCB. FD: As a consultant Jonathan works with startups, providers and health plans, advising clients on policy issues, strategic direction and related topics.

Categories: Uncategorized

Tagged as: , ,

17 replies »

  1. Dramzi, I’m not sure what I wrote that contradicts your first several points. Yes, RA applies after adverse selection and is intended to mitigate its effects. My point was that not all adverse selection is the same, and there are various circumstances where it appears to harm rather than help competition to apply the same overly-simple formula on transfers from one plan to another.

    I also agree that RA doesn’t predict all of the variation in expenditures. The problem, though, is that the transfers are too large given the inadequacies in the models. CMS recognizes this, for different reasons than I discussed here. In proposed rules published on 8/29, CMS seeks to blunt the force of the model for transfers at the high and low end of the risk scale and provided several options for doing so.

    Sadly, I also think that gaming the system is not necessarily a positive thing that achieves the goals of the ACA. I think I was a little too easy on the Medicare risk adjustment process. There are serious allegations that plans in the Medicare Advantage program go on fishing expeditions for additional health conditions, and in doing so end up identifying diagnosable conditions that patients are not inclined to treat (or that are minor enough that treatment isn’t really important), so that claims cost does not go up as much as the risk-adjusted capitation rate received by the plan. Also, regardless of this, you and I agree that smaller, newer and/or less-sophisticated plans are at a disadvantage because they lack the data and analytics needed to fully identify and report on the risk of their population. To some extent (as for unsophisticated plans) that is tough luck and it is good that they have pressure to change. But to some extent that is anti-competitive (as when a new entrant that could put pricing pressure on a dominant plan needs to make payments to its competitor because it doesn’t have a long history of claims to mine for conditions).

    So, it may certainly be true that I have misunderstood something about RA, but I don’t know what it is from your comments.

  2. I believe you have misunderstood some features of RA. First RA applies after adverse selection and is a way of mitigating its effects. After all the sickest patients are always a greater liability because insurers are funded on the basis of other actuarial factors. Second, RA only predicts about 60% of the variation in expenditures, so cannot completely adjust for the risk and can still only have value in concert with other features of plans, like risk corridors, and reinsurance. Frankly volume too tends to water down the risk; once you get to 400-500 people, risk tends to recent to the mean. Unless, of course the insurer is exposed to an inordinate number of people from a disastrous zip, since health risk tends to cluster geographically according to socioeconomic factors.

    Many of the coops came in with bright-eyed optimism; more optimism than competence. They mis-priced their policies disastrously, but would have survived if they had the capital. However no business survives both mistakes and a lack of capital.

    Gaming the system in this case was designed, it seems to me, to achieve he goals of ACA. By reaching out to patients, getting them in, evaluating for prevention and identifying risks, you incidentally produce more codes that could push up your risk score. These people now have heir chronic conditions addressed rather than ignored due to system structure. The system now rewards the care of the sicker people.

    I suspect the reason incumbents do better in an RA is simple: they have more data and the competence to utilize that data to competitive advantage. Will anyone argue that an insurer exists with more data than United? And you remember the old adage: “In healthcare, the one with the most data wins.!”

  3. There are two parts to the formula. The first part assesses risk without reference to the value of claims paid by the plan (though it does reference average cost by condition, and the ICD codes come in through claims typically). The second part, after relative risk levels are assessed and transfer amounts are decided, does reference total claims paid. So the answer is partly yes, partly no. I have heard complaints that the methodology does not “true up” risk levels to actual claims paid by each plan. When reinsurance and risk corridors were in place (and fully funded) that criticism had less force, but now it seems to be a bigger issue.

  4. Is The Risk Adjustment Formula based more on the expected risk of the pool rather than on actual claims paid?
    It seems the risk can be fudged so much that the biggest fudger, wins

  5. All very well intended, i.e., to build a better mouse trap. None of this will solve the “social dilemma” as augmented by Parkinson’s Law that represents the pressing problem for the high cost and low effectiveness of our nation’s healthcare.

    Resep Jus Sayuran yang Disukai oleh Anak

  6. All very well intended, i.e., to build a better mouse trap. None of this will solve the “social dilemma” as augmented by Parkinson’s Law that represents the pressing problem for the high cost and low effectiveness of our nation’s healthcare.

    Paul Nelson, M.D.

  7. Thanks. I was informed of the unintended consequences of ACA from your article. Yes, this conversation is somewhat in a vacuum, but good discussion.

  8. Andre, that’s certainly an option to consider. To be clear, my proposal was really just intended as a conversation-starter that seemed to directly address one of the unintended consequences of the current risk adjustment model. I am sure whatever final solution is created will be more complex, and might look very different.

  9. Thank you. There seems to be a building consensus, so I am optimistic something will be done, but will it be enough and will it come quickly enough? That’s the multi-billion dollar question.

  10. Powerrp, your point has a narrower application than you seem to suppose. Yes, better capitalization would have helped the start-ups, but they were not in a good position to be better capitalized, and they were relying on funding that was promised in the ACA but never materialized due to hostile acts of Congress. I did not blame the co-op bankruptcies primarily on risk adjustment, and indicated that the risk corridors played a role. I should have been clearer, so let me clarify now: under-capitalization, risk corridor shortfalls, too-aggressive pricing and risk adjustment all played a role in the failed start-ups to date.

    But as for United Healthcare and others taking a step back, that has nothing to do with capitalization. And going forward, none of the concerns about competition relate to risk corridors or reinsurance. It is the risk adjustment methodology that is creating problems for competition and appears untenable in its current form.

    I think you may have misunderstood my argument. I am not saying we should do away with risk adjustment. I agree that it works in other places. It’s just not working here, in the post-ACA individual and small group markets on and off the exchanges.The solution is not to eliminate it, but to figure out what’s wrong and improve it.

  11. It might be as simple as building a better mousetrap. The Federal government, with the implementation of ACA, started toying with risk in the private market in ways that are one-dimensional and unconventional. Risk is dynamic, and cannot sufficiently be captured with medical billing codes or actuarial projections – alone. The Gov’t re-organized the incentive for an insurer taking on risk, and gave more support to (the mechanic with the messiest shop) – not a good market-based idea. The risk assessment process and reporting under the ACA should probably be handled by private market participants (specializing consultants), and, notwithstanding the article’s suggestion on payment-related MLR benchmarks as a fix, an insurer could pay a third-party company (and pay less money into the risk-adjustment model and diminishing the imbalance inherent in the current risk assessment process) to produce a customized risk-report based upon a uniform reporting standard, get better data on their risks, and level the playing field for risk assessment and risk adjustment in the process.

  12. The author has “gone wrong” when he blames the straw for the camel’s broken back. Risk adjustment is an absolutely-essential piece of permanent infrastructure. In no way does RA deserve to be cited as the cause of these failures.

    Typically, the failed issuers were grossly under-capitalized (even net of unpaid risk corridors) and/or they engaged in magical thinking. The data submission process for RA is rigorous, but it ain’t rocket science. CMS went to extraordinary lengths to help every issuer. Yes, issuers could be forgiven for big misses on RA reserving in the ’14 cycle, but in no way could they have been surprised for the ’15 cycle! Sure, certain parts of RA can be improved. Did the author even broach those topics? No, he did not.

  13. Excellent post by Mr. Halvorson. Unfortunately we have spent decades asking for transactional, fee for service care based in treating disease, rather than paying for prevention. Almost certainly it will take years to unwind these business rules. I am no fan of regulation but currently don’t see more effective free market tools. Human nature will continue to look for every advantage- either cherry picking for healthy populations or sick populations based on prevailing subsidies. One has to wonder if all of this complicated rule setting will ultimately collapse into a single payer system. Some have suggested this was the original intent of the legislation.

  14. From a Primary Healthcare stand point, the high deductible plans in Obamacare virtually eliminate any meaningful health care for Basic Healthcare Needs, with or without a Medical Savings Account. As the Medicare eligible number of citizens mushrooms in the next 15 years, we will be looking at another round of aggressive hospital expansion projects. Is it possible to conceive that tinkering with the alternatives for universal health insurance will solve the exorbitantly high cost of our nation’s healthcare? Presumably, the best minds around got us into Obamacare. But, the law of unintended results continues to plague these efforts. Meanwhile, the excess cost of our nation’s healthcare in 2015 represented 60% of the Federal deficit, AND our nation’s maternal mortality ratio continues to worsen, as it has for the last 20 years.
    Overall, the health insurance industry has slowly, but surely, decreased the payment processes for Basic Healthcare Needs as a means to pay for the Complex Healthcare Needs of each citizen and the employment benefits of its executive teams. Is there any health insurance payor that doesn’t have a contract with more than one medical school hospital? Probably, not many. As we begin to reconstruct the distant horizon of healthcare reform, let us not ignore the fact that Primary Healthcare will probably never improve without first acknowledging that it is basically not insurable. It realistically should be considered a pre-paid expense with a re-insurance process managed by the payor with a stop-loss arrangement spread-out among the Primary Healthcare clinics contracted with the plan. The Design Principles applicable to managing a common-pool resource would apply.

    Paul Nelson, M.D.

  15. It seems the two key tools that insurers need to survive and thrive have been taken away from them by the government: 1. Insurers have been experts in assessing risks and in accepting, pricing and shedding risks. This has been taken away from them by taking away underwriting and forcing community rating. And the government thinks it can do the same job as expertly as the insurers and making them whole by risk adjustment and corridors. This failure you cover well in your fine article above.

    And 2. The government has taken the other critical tool from insurers–the sine qua non of insurance theory–the ability to use interest rates to manage the float (the premium monies sitting there) so that they look around and find zero or de facto zero rates all over the place. Thus they have to invest in crazy equties or risky foreign bonds all over the world to make a profit….or simply price their policies at actual current costs + profit. So beneficiaries are not getting any help from the firm’s investments at all.

    No wonder premiums are zooming.

    I don’t see why any insurer sees anything but gloom ahead in this milieu.