ACA Exchange Initiatives Program | Actuarial Specialties | Health | Web Exclusives

Financial Fairness

Looking for ways to level the playing field among health insurance carriers under the ACA

Kristi M. Bohn Web Exclusive

Now that we have access to risk adjustment results from 2014 and 2015, we can see the extent of risk adjustment’s importance as an element of the Affordable Care Act (ACA). This is evidenced by the relatively large transfers of dollars among carriers in many states, which has the effect of bringing carriers’ loss ratios closer together.¹ The carriers whose enrollees chose bronze and silver plans generally owe a lot to the carriers with a higher proportion of gold and platinum enrollees, and they should. The adverse selection against the platinum plans has made them unavailable in many areas, which is a good indicator that the system was not adequately compensating high-risk cases (that is, people who were more likely to elect platinum plans).

The Centers for Medicare & Medicaid Services (CMS) will be making two very important improvements to risk adjustment in the near future: increasing the risk scores for those who spend a short time in the market (2017) and including certain drug information in risk scoring (2018).

In Minnesota, we have been studying whether we should adopt our own risk adjustment system. With the help of a team of consultants, we found that carriers with a disproportionate share of partial-year enrollees are saddled with uncompensated risk.

Including Drug Data in Risk Scoring

Pharmacy data inclusion is very important for a couple of reasons. First, it is important in light of high-cost drugs, such as very expensive hepatitis cures. Hepatitis cures are particularly concerning, given the disease’s disparate coverage. Hepatitis is common enough that such disparate coverage among carriers—and moreover markets—means that uncompensated cures in the individual market easily can bankrupt a well-behaving carrier. In talking with one state’s regulator, the impact of Hepatitis C has already been a major contributor to the insolvency of a health insurance carrier (though for 2014, this consequence also can be attributable to the risk corridor program’s failure to deliver on its promise).

There are many articles relating to the major financial dilemma that the expensive hepatitis cure places on countries, states and insurers (see the Related Links at the end of this article.) Hepatitis is relatively common, so the expensive cure is an enormous burden. Moreover, some people still would be at risk for reinfection due to their continued exposure after they are cured. The individual market is far too small for a carrier to be burdened with other markets’ (and other individual market carriers’) non-coverage of the cure.

A refinement in the model to reflect actual delivery of the cure (versus statistically possible delivery) is critical, both in supporting financial fairness as well as supporting the public health goal of promoting the inclusion of the hepatitis cures on carriers’ formularies. There are other circumstances that warrant similar special risk adjustment handling, such as rheumatoid arthritis drugs, multiple sclerosis drugs, HIV treatments and hemophilia drugs (synthetic versus biologic hemophilia drugs are remarkably different in cost, so much so that a risk score based on average payment is simply unfair and a low-hanging fruit in terms of statistical accuracy in such high-cost scenarios).

The second reason why the inclusion of prescription drugs is an important improvement to the risk adjustment system is that its inclusion will help mitigate the unfair advantage that larger, more experienced carriers have in terms of having much greater access to historical claims and diagnosis data. Such a bias in information access can support better claims mining for more appropriate claims submissions from doctors and hospitals, as well as care management outreach to patients directly in order to encourage visits that trigger higher-risk adjustment scores. Drug data inclusion into the model will help level the playing field between small and large carriers.

There also are some important findings from 2014 and 2015. Many carriers’ risk adjustment payment changed dramatically from 2014 to 2015, both in total dollar magnitude and on a per life basis. One carrier in Minnesota’s individual market went from a payment that was just under $4 per member per month (PMPM) to a payment that was just under $100 PMPM. And this carrier was not small. The data of small carriers across the country is just as varied. CMS should reconsider the statistical credibility of small carrier data, as the current model seems to lend too much credibility too soon. To that point, some state regulators are very concerned about the financial stability of small carriers participating in the individual and small group markets due to risk adjustment, and a change in credibility would help mitigate the variability and improve predictability in a fair way—particularly compared to some of the other proposed alternatives recently offered to CMS and state regulators. For a basis for evaluating the statistic credibility of risk adjustment based on enrollment size, see the Society of Actuaries’ (SOA’s) study on this topic.²

Refining the Risk Adjustment Model

This variability in carrier payments between 2014 and 2015 shows something we all can appreciate—the demographic profile of any given carrier is not yet stable. The ACA’s goal of competition is working: People are shopping. This fact draws attention to something actuaries need to strongly and repeatedly counsel against: A prospective risk adjustment model is unthinkable any time soon and should not be on anyone’s list of considerations. Even if stable distributions are reached, adoption of a prospective model should be taken as a sign of resignation on the ACA’s goal of competition and consumeristic shopping. While prospective risk adjustment models may perform well in group, Medicaid and Medicare markets, they are wholly inappropriate in the individual market. It was a relief to see that CMS is not considering changing the use of a retrospective risk adjustment model.

Minnesota’s exploration of potentially adopting its own risk adjustment system also provided some unexpected insights. First, CMS told us to expect that high-cost case conditions would be overpaid. This was CMS’s openly stated and purposeful strategy to correct for the typical finding that risk adjustment systems tend to underpay for high-cost cases. A strategy to overpay for high-cost cases would discourage a potential marketing bias away from older and less healthy patients. Further to this point, because the transfers are based on premiums (instead of the typical basis on claims), we were expecting high-cost cases to be further overcompensated, because premiums theoretically also would include retention items, such as administrative costs, profit loads, taxes and assessments. However, based on Minnesota’s experience to date, both expectations were not realized. While some of the high-cost cases are indeed being overpaid (particularly in 2014 and 2015, when the reinsurance program was very material), some simply are not. Unfortunately, the types of high-cost cases that are not fully compensated are occurring at a much higher frequency than anyone had expected.

End Stage Renal Disease, multiple sclerosis, certain types of diabetes, personality disorders, autistic disorders pervasive developmental disorders, and HIV/AIDS are examples of this phenomena in Minnesota. For more on this topic, Minnesota’s report on considering its own risk adjustment system should be available to the public soon, though note that the End Stage Renal Disease has been a recent finding outside of the study.

To add to this aggregate shortfall to certain carriers with a disproportionate share of high-cost cases, the individual market enrollment demographics are simply not stable yet—premiums are rising dramatically in Minnesota in order to keep up with claims. This case is so material in Minnesota that the overall level of the risk adjustment transfers, which are based on premiums instead of claims, are much lower than the carriers would transfer had the transfers been based on claims. Claims are now known to be materially higher than the corresponding premiums in Minnesota, even after reinsurance is reflected, for both 2014 and 2015. This is a problem that is exacerbated by the logical tendency for consumers to choose lower-cost plans, as well as the failure of the risk corridor program to have delivered on the promise to largely mitigate this major underwriting risk.

State regulators can help somewhat on the variation of price points risk by ensuring that carriers do not materially underprice in relation to the statewide risk pool. Given statewide risk pooling, there theoretically should be minimal premium variation among carriers. The primary sources for variations in any given geographic rating area are:

Administrative and taxation/assessment differences
Network and provider discount differences
Care management differences
Morbidity differences that the risk adjustment system fails to capture

Another suggested improvement is for the CMS risk adjustment model to be recalibrated on the individual and small group populations, rather than initial and separate commercial data that currently is being used. Based on Minnesota’s study, the team found that altering the model’s data source would not significantly change the model’s predictive ability, because the relativities in the current model are approximately accurate regardless of which market the scores are based on for all but a few conditions. However, using such data greatly would improve other adjustments, such as metal level induced demand adjustments. Further, such adjustments likely need to be different in the individual market versus the small group market, given that these two markets have very different adverse selection opportunities and effects.

In my view, the metal level adjustment also should consider that it is likely there are other selection biases, regardless of condition. In other words, while the model focuses on chronic conditions, the accuracy and fairness of the model can be improved by capitalizing on existing covariances to all enrollees’ acute conditions in order to help improve the overall fairness for a carrier’s morbidity burden. Without such considerations, we should expect the availability of platinum plans to continue on their path to extinction in the individual market.³

Market Sustainability

While these highly technical topics are important to consider in future improvements to the risk adjustment system, these concerns pale in comparison to the sustainability problems that the individual markets face in many states. Because the risk adjustment system is a zero sum game for each state, the bigger problems we must grapple with are overall individual market affordability (though more important than affordability is value) and sustainability in light of adverse selection.

To put it simply, there has been too much adverse selection against the individual market as a whole, and the question of transfers among carriers is rather minor in relation to the overall policy problems that have emerged in many states. This includes adverse selection from providers, such as encouraging dialysis patients to use the individual market when they should be on Medicare (see 1882(d)(3) of the Social Security Act). Add to that a handful of employer brokers sending sick employees to the individual market, and then individuals choosing to enroll and then leaving the individual market when they need care. The individual market is simply too small to accept and subsidize even a small portion of this behavior. CMS has reacted to these concerns by issuing guidance against such employer practices⁴ and allowing carriers to prohibit many kinds of third-party payments. But more policy changes are needed to reduce overall adverse selection, because it would take very few high-cost cases to send the very small individual market into unsustainable zones.

On the state regulatory side, we may need to pursue “suitability” breaches against brokers and agents (in Minnesota, up to $10,000 per incident and/or license revocation/suspension) in order to deal with these concerns if additional federal policy and operational changes are not enacted soon. But state conduct processes are slow and burdensome, and state staffing is light; state pursuits may be too little, too late.

Concluding Thoughts on Minnesota

As a final note, Minnesota is an interesting state to observe. Prices across most rating areas of Minnesota in 2014 were the lowest in the nation, even though the rate increases that carriers priced were significantly higher than in 2013 and were in line with Minnesota-specific and national public health studies. For example, the 2014 rate increases were typically near the 2017 increase that the SOA’s ACA study had predicted. That said, premiums have increased substantially due to continually emerging new entrants’ unexpectedly high morbidity.

Many critics credit the Minnesota increases to the wrong factors because they simply do not know about Minnesota’s structure (including one of this article’s citations!). The most material policy issue affecting Minnesota’s steeper rate trajectory is its unique structure. Our experience—though unique—is informative. Minnesota is one of only two states that offers a Basic Health Plan. Thus, in Minnesota, there is no one in the 87 percent or 94 percent cost-sharing reduction plans, and we have a very small proportion of overall enrollment in the 73 percent plans. This is because nearly all who qualify for cost sharing reduction (CSR) plus advanced premium tax credit (APTC) subsidies are triaged to the Basic Health Plan. Without this more stable, healthy demographic of price-insensitive enrollees, Minnesota’s individual market experiences far fewer risk pool supports than other states. We expect our state will need to make some very important policy decisions in the coming year in order to address the trajectory the individual market risk pool has taken. Risk adjustment will play a role in that process, as fairness and sustainability will always be a concern to regulators and policymakers.

Financial Fairness

Including Drug Data in Risk Scoring

Refining the Risk Adjustment Model

Market Sustainability

Concluding Thoughts on Minnesota

Related Links

References: