Sunday, February 28, 2021

Does mandatory maximum silver loading require a risk adjustment fix?

Subscribe to xpostfactoid

I want to address a somewhat technical issue concerning regulatory action to require that gold plans in the ACA marketplace be priced below silver plans -- that is, regulatory action on the state or federal level mandating maximal silver loading.

Gold plans should be priced below silver plans because silver plans --on average -- carry a higher actuarial value than gold plans. "Actuarial value" refers to the percentage of the average enrollee's costs covered by the plan, according to a formula created by CMS (that formula could use adjustment, but that's another story). Gold plans have an AV of 80%, plus or minus a few percentage points. Silver plans have a baseline AV of 70%, but that's only for enrollees with incomes above 250% of the Federal Poverty Level (FPL). Below that threshold a secondary subsidy, Cost Sharing Reduction (CSR), available only with silver plans, boosts AV to 73%, 87%, or 94%, rising as income falls.  Most silver plan enrollees obtain AV of 94% or 87%.

Until 2018, the federal government reimbursed insurers directly for CSR, and silver plans were priced as if their AV was always 70%. When Trump cut off that reimbursement in October 2017, insurers started pricing CSR directly into silver plan premiums.  That created discounts in bronze and gold plans. But the price adjustments stopped halfway; silver plans are still priced proportionately lower in most markets than their real AV (the average AV of all an insurer's silver plan enrollees) would warrant.

Arguably, "silver loading" -- pricing silver plans according to their real AV -- should be maximized in accord with a self-fulfilling prophecy: no one with an income above 200% FPL will choose silver, because silver is priced as if its AV is higher than it is for enrollees above that threshold. At an income above 200% FPL, you're paying for someone else's CSR. It hasn't worked out that way, because silver is relatively underpriced. 

Why? In Health Affairs,  David Anderson and I recently posited two causes:

Market watchers have identified two likely causes for the only-partial effect. First, competitive pressures drive insurers to underprice silver plans, when allowed. A majority of on-exchange enrollees are eligible for CSR, and the cheapest silver plans gain an outsized share of enrollment. 

The second factor is the risk adjustment program that CMS administers to deter insurers from trying to attract the healthiest enrollees. The current formula, based on usage in employer-sponsored insurance, favors silver plans, at the expense of gold and bronze plans, by assuming that CSR will stimulate more usage of medical services than has proved to be the case. That's probably because CSR enrollees have lower incomes than those in employer plans, and even the reduced out-of-pocket costs inhibit care more than anticipated. Insurers that attract too much enrollment at other metal levels may suffer in the zero-sum risk adjustment game. 

Here is the debate. If regulators require insurers to price plans at each metal level in proportion to their real actuarial value without fixing the risk adjustment formula -- as at least three states have done, directly or indirectly -- will insurers react by making their bronze and/or gold plans as unattractive as possible, since those plans carry an implicit risk adjustment penalty?  

David Anderson outlines the danger:

Some states have mandated that insurers price at actuarial value relativity.  These states say that if blended silver has an average actuarial value across all products of 87% and blended gold actuarial value is 79% (a more than plausible situation), then silver should be priced roughly 10% higher than gold.  This produces short term gold discounts below the silver benchmark while producing opportunities for aggressive rule-lawyering insurance companies to find ways to optimize their portfolio of plans offered to only sell silver and not sell not-silver.  There are plenty of ways to do that.  I don’t think that we want insurers to play games on how not to cover people.

From this point of view, only CMS should mandate maximal silver loading, since only CMS can fix risk adjustment (actually, states can create their own risk adjustment formulas, but none have -- it's a major enterprise).  Actuary Greg Fann, on the other hand, argues that if states have to play on a level field, under rational pricing rules, any skew in the risk adjustment formula is unlikely to create major distortions.

I am not qualified to assess the potential of the risk adjustment problem to distort markets if maximal silver loading is mandated. I do, however, have questions (and an observation) about the premise:

1) We’ll get a natural experiment in whether mandating maximal silver loading distorts markets, as at least three states (Pennsylvania, Maryland, Virginia) have done it by one means or another.

2) Enrollees with income over 200% FPL have already in large part abandoned silver in most markets. Even with a risk adjustment skew, will most insurers opt to disincentivize gold/bronze?

3) Given required AV constraints, how bad can you really make bronze/gold plans?

4) I gather risk adjustment isn’t exactly exact science. If gold becomes cheaper than silver, might that change relative induced demand between the metal levels, as more healthy people choose gold?

5. How many insurers understand the market well enough to sabotage their bronze/gold offerings as a deliberate strategy? And, per all the points above, is that likely to be an effective strategy?

Thoughts welcome!

UPDATE: Some "welcome thoughts" below from actuary Daniel Cruz, a frequent coauthor with Greg Fann and proponent of state action to intensify silver loading (I say "intensify" rather than "maximize," because, as Daniel clarifies below, states that have taken action have not required insurers to price silver as if all enrollees have incomes under 200% FPL). 


First and foremost, it needs to be repeated, that health plans are not allowed to develop metal level pricing factors based on the impact of risk adjustment payments. Risk adjustment amounts paid or received can only impact rates on a global level. What this basically means is if the health plan receives risk adjustment payments that would lower rates by 2%, then that 2% has to be applied to ALL metal levels and all plans. If you want to understand this mechanically, this is called the market adjusted index rate and you can read about it in the following ( page 20). I understand the need to talk about why health plans are incentivized to target certain enrollees and how risk adjustment creates certain incentives but we shouldn’t lose sight of the regulations that currently exist. It would be like banks violating some banking law and everybody is trying to figure how to disincentivize the banks from violating the law…or you could just enforce the law that already exists. So I repeat, health plans are not allowed to consider risk adjustment payment outcomes between metal levels when pricing the rate differentials between their Bronze, Silver and Gold plans.

The second thing to note is that risk adjustment is always improving and will never be perfect. Starting this past year, CMS has started to use EDGE server data or individual/small group data to calibrate the risk adjustment factors and it will be entirely phased in over the next two years ( The basic impact is HCCs staying somewhat flat and demographic factors decreasing. The decreasing may be due to moving accident items to HCCs (thoughts relayed to me by Gabriel McGlamery) but it may also have to do with income as presumably healthy enrollees utilize health care less in the individual market since preventative care or elective care is more cost prohibitive for them than say an average employer market consumer. Just a guess. But the takeaway is that CMS is constantly improving risk adjustment, starting to base it on ACA data and we should acknowledge that no matter what they do it will be imperfect.

The third thing to note is specifically how risk adjustment treats the CSR enrollees. The first item is the 1.12 induced utilization bump health plans receive. This bump has little justification at the moment other than to reduce federal spending. It was originally there because health plans could not charge premiums for CSR benefits. Instead, they received payments from the federal government which were calculated based on an estimate of how much the health plan would have cost if they had silver benefits vs what they cost with CSR benefits and the feds paid the difference. One item was missing from that which was that CSR benefits would induce more care which wasn’t technically covered from the feds payment and so they put that in the risk adjustment formula to account for the lack of payment. In today’s world, health plans can and do charge premiums for CSR benefits. So there is no strong reason to have the 1.12 bump and as it sits there it creates an unneeded incentive to lower silver premiums (again incentives but from a regulatory perspective still shouldn’t happen). Getting rid of the 1.12 factor was discussed early on but later dropped. (

The only other aspect to deal with for CSR folks is should risk adjustment have an income based factor? I think there is good reason to advocate for this but this would be an added complexity which is always something to consider when weighing whether to implement changes to a large scale nationwide model. I don’t have a strong opinion on this and I’m not an expert on the specifics to give one.

The fourth thing to note is the maximization fix. It should be noted that none of the states mentioned, PA, VA and MD, have implemented maximization. The only state that has is WY, which was implemented by the monopoly health plan. The three states mentioned regulated the metal level pricing differentials and focused on the induced demand factor being used which aligned their metal levels more appropriately but have allowed for the silver load to remain partially implemented and not at full equilibrium (aka 100% platinum). Induced demand is a complicated little term to describe in the ACA world but what’s key to remember is per regulations ( -  page 40 and 41) it can only represent changes in behavior of an AVERAGE ACA consumer based entirely on the differences in benefits. It CANNOT reflect health status of enrollees selecting the plan nor the income and it cannot reflect metal specific enrollment or net risk adjustment outcomes of metal levels. Again this is a regulatory issue, already defined fairly explicitly.

I have advocated for standardizing the induced demand factors used because this has been the source of wild disparities between health plans and across states. In places like California where you have standard plan designs, you see huge differences in health plan pricing differentials of their metal tiers which is an obvious sign of a problem. I’m generally in favor of health plan flexibility but this issue hasn’t worked out in reality for the induced demand factor and its hurting consumers/markets. I would compare this to getting married to a Kardashian. It would be nice to trust the relationship and hope it will work out but all signs and past experience suggests you should sign a pre-nup. So as a rule, if you marry a Kardashian, sign a pre-nup and if you allow for induced demand pricing in the ACA market, standardize those pricing factors and move on to the next issue. That’s just my personal opinion on the matter, there is an argument for flexibility of course. The 1.44 is standardizing another assumption that creates disparities between health plans. 1.20 can be justifies but so can 1.44 based on the rational consumer assumption of 100% CSR enrollees in Silver. 1.44 > 1.20 for the consumer and the markets so both are fine but for those interested in helping consumers and the market why not go with 1.44? I’ve tried to lay out the case for the 1.44 as best I could ( )

The last comment is a pet peeve of mine. There is nothing intrinsically bad about Gold plans (look at the small group market enrollment) that would make health plans run away from them. In fact, if the carrier can be profitable on a higher priced plan, it will likely be able to cover their admin better and make more profit on a per member basis. What health plans have been trying to avoid is outlier risk not Gold plan designs. As long as Gold is overpriced and enrollees represent a small portion of total enrollment, those members will likely reflect the highest risk members and the profitability of that group is dependent upon risk adjustment which is improving. At 5-10% enrollment, you take a risk. But as Gold becomes valuable to a broader set of consumers with more reasonable pricing, a more average level of risk will gravitate to Gold and you will have the outlier risk blended with average risk. Carriers will be happy to get more average risk in higher priced plans which is evident when a monopoly in WY decided to go that route. PA, MD and VA all have good amount of Gold enrollment and will likely have stability on this front. Assuming carriers run away from Gold as opposed to running away from being stuck with all the outlier risk is a static view of the market that doesn’t consider what happens when enrollment shifts due to a value shift for consumers. Forcing induced demand factors evens things out by not allowing individual health plans to try to avoid outlier risk, pushing the market further towards equilibrium. I have always liked the idea of watching the “good states” though, see how it plays out. To the contrary, in California, Sharp and Kaiser initially tried to lower Gold at or below Silver but competitive pressure drove them back the other direction which happens when you don’t put health plans on an even footing. I live in California so pardon the Cali references but that market provides a lot of examples of issues. 

Subscribe to xpostfactoid

No comments:

Post a Comment