December 10, 2023

Health Insurance

Follow Your Health Insurance

ACA Metal-Tier Mispricing: Improving Affordability By Solving An Actuarial Mystery

It wasn’t supposed to be this way.

In 2015, HHS’s Assistant Secretary for Planning and Evaluation (ASPE) described what would happen if the cost-sharing reductions (CSR) that silver-tier Marketplace plans give their low-income members were financed by premiums rather than separate federal payments.

ASPE compared the number of silver enrollees with incomes above 250 percent of the federal poverty level (FPL), who received coverage with 70 percent actuarial value (AV) because they were ineligible for CSRs, with the number of silver enrollees with less income, who qualified for CSRs that raised AV to between 73 percent and 94 percent. ASPE found that average silver AV was “about 85 percent.” (The AV indicates the percentage of an enrollee’s medical expenses that a plan will cover, on average.)  Silver premiums, which determine premium tax credit amounts, would therefore rise to “cover about 85 percent …. of total health care costs…” Gold plans would remain priced based on 80 percent AV. As a result, consumers formerly in 70 percent-AV or 73 percent-AV silver would “be better off buying gold than silver plans,” since consumers would receive “lower premiums and greater financial protection.”  

Soon,  the only consumers left in silver would have 94 percent-AV or 87 percent-AV coverage. This would require a further ”increase in silver plan premiums … to cover an AV of about 90 percent,” significantly above premiums for 80 percent-AV gold.

After federal CSR payments ended in late 2017, premiums in few states met ASPE’s expectations. By 2022, the fifth full year without CSR payments, silver premiums remained stubbornly below gold premiums in 37 states, by a median margin of 11 percent—even though silver’s average AV, based on national enrollment, was 87 percent.

Instead of obtaining affordable access to gold plans, which had median deductibles in below $1,400 in 2022, millions of consumers were left in 73 percent-AV silver, 70 percent-AV silver, or 60 percent-AV bronze, with median deductibles that exceeded $3,300, $5,000, and $6,900, respectively. In 2022, 51 percent of all marketplace enrollees received high-deductible bronze or silver coverage in the median state.

Such metal-tier mispricing costs consumers roughly $6 billion a year, according to one analysis. Health equity concerns are implicated as well, since gold enrollment among consumers without CSRs appears to be substantially lower in Latino and African American families than among non-Hispanic Whites.

How did this happen? And how can policymakers fix this problem?

Searching for answers, we investigate three topics familiar to murder mystery fans: motive, means, and opportunity.


The motive is classic: money.

When the ACA took effect, insurers quickly discovered that risk adjustment overpaid carriers for CSR members. Risk adjustment estimates the risk level of each carrier’s enrollees, charging insurers with low-risk membership and paying insurers with high-risk membership. Insurers expecting high-cost enrollment can thus afford to follow the Affordable Care Act (ACA) and “ charge the same average premium … as a plan enrolling a higher proportion of low risks.

In 2016, a leading actuarial firm reported that risk adjustment payments significantly exceeded CSR claims, so CSR members generated revenue far above their actual costs. The actuaries concluded, “CSRs are Key to Success.” The risk adjustment formula assumed that lower cost-sharing would increase utilization for low-income CSR consumers above levels in lower-AV silver coverage, ignoring income’s impact cutting health care use.

Carriers soon began aggressively underpricing silver, the obvious strategy for attracting profitable CSR consumers, whose low incomes make them sensitive to small premium differences. Those incentives persisted even after federal CSR payments ended, as we explain below.


Insurers’ means of underpricing silver were explained in a Health Affairs Forefront article and a letter to the Centers for Consumer Information and Insurance Oversight (CCIIO) from nationally respected health actuaries. To justify common pricing practices, Joyce Bohl and colleagues argued that, in paying estimated claims resulting from CSRs, silver premiums should reflect “specific experience data” showing costs actually incurred by CSR enrollees.

Their rationale focuses on the termination of federal CSR payments more than five years ago: “The purpose of [the portion of silver premiums covering CSR costs] is to ensure that an insurer receives sufficient revenue in aggregate to cover the expected costs of the unfunded CSR subsidy.”  That supposed purpose, which has no discernable roots in ACA pricing law, incorporates experience rating into silver premium setting, since federal CSR subsidies reflected actual costs for low-income CSR beneficiaries.

All else equal, people with less income use fewer services. As explained by the American Academy of Actuaries, “In general, lower-income individuals have … a significantly higher threshold for determining when a service is sufficiently valuable to be utilized. As such, we would generally anticipate much lower utilization among low-income enrollees…”

This income effect’s magnitude is extraordinary. Consumers with incomes below 200 percent of FPL qualify for 94 percent-AV or 87 percent-AV plans, which had median deductibles of $47 and $635, respectively, in 2022 in But because those plans are limited to low-income people, actual paid claims are lower than in non-CSR silver plans offering 70 percent AV, which has a $5,166 median deductible. Incorporating the utilization experience of low-income CSR members, as urged by Bohl and her co-authors, thus depresses silver-tier premiums, causing corresponding increases to premiums at other metal levels.  

Incorporating the reduced utilization associated with low-income CSR members into pricing violates the ACA. Subject only to specific, statutorily defined exceptions, the characteristics of people expected to choose a particular plan should not affect premiums. This “modified community rating” principle is codified in subparagraph (A) of 42 U.S.C. 300gg(a)(1), which permits premiums to “vary with respect to the particular plan … only by (i) whether such plan …  covers an individual or family; (ii) rating area…; (iii) age … ; and (iv) tobacco use…” Subparagraph (B) adds, “such rate shall not vary with respect to the particular plan … by any other factor not described in subparagraph (A).” 

Traditional medical underwriting, which bases premiums on each consumer’s likely claims, violates that law. But the law is also violated when insurers base premiums on the distinctive characteristics of individuals expected to enroll in a particular plan. The reduced utilization associated with low-income CSR enrollees is a “factor not described in subparagraph (A)” and thus may not lawfully affect plan premiums.

To effectuate modified community rating, ACA §1312(c)(1) requires each insurer to consider all its individual-market enrollees to be “members of a single risk pool.” Accordingly,  42 CFR §156.80(d) has rate setting begin with the insurer establishing a market-wide index rate for all individual market plans, “based on the total combined claims costs for providing essential health benefits within the single risk pool.” The insurer adjusts that rate to fit individual plans to incorporate only five factors: Actuarial value and cost-sharing; provider network, delivery system, and utilization management; coverage that exceeds essential health benefits; administrative costs; and for catastrophic plans only, the characteristics of eligible people.  CCIIO’s pricing instructions thus require estimating utilization for every plan based on “the average demographic characteristics of the single risk pool.” This leaves no room to base silver premiums on the anticipated characteristics of silver enrollees.

If the regulation allowed insurers to vary plan premiums based on possible shortfalls in federal CSR payments, or the characteristics of people eligible for silver coverage, the experience rating Bohl and colleagues advocate would be lawful. But the regulation explicitly provides that “a health insurance issuer may vary premium rates for a particular plan from its market-wide index rate … based only” on the five enumerated factors (emphasis supplied).  Their recommended approach may describe how actuaries often price plans, but it does not follow federal law.

In its regulatory preamble, CCIIO explained the single risk pool’s purpose:

“Without a single risk pool rule, [the ACA’s] prohibitions against traditional underwriting could incentivize issuers to find ways to segment the market into separate risk pools and charge differential premiums based on segmented risk, a de facto mechanism for underwriting.”

The approach defended by Bohl and her co-authors creates multiple risk pools. One, limited to CSR members whose low income suppresses utilization, estimates the CSR costs that are incorporated into silver premiums. It is not clear whether Bohl and colleagues would use different risk pools for bronze and gold tiers or combine all non-CSR-eligible enrollees into a single pool that is used to calculate all non-CSR claims. Either way, their approach lets insurers “find ways to segment the market into separate risk pools and charge differential premiums based on segmented risk, a de facto mechanism for underwriting.” It yields the precise result CCIIO sought to prevent with the single-risk-pool regulation.


Inadequate enforcement of rate regulation offers insurers ample opportunity to aggressively underprice silver premiums while raising premiums at other metal levels. Policymakers can fix this problem by enforcing rating requirements. They can also end carriers’ incentive to cheat by fixing risk adjustment.

Rate Regulation

In recent years, New Mexico and Texas have  required carriers to strictly follow the single-risk-pool requirement. The results were stunning:

  • In 2022, the first year New Mexico’s approach took effect, the proportion of Marketplace enrollees receiving high-deductible coverage in bronze or low-AV silver fell from 49 percent to 23 percent.
  • In 2023, the first year Texas’s approach took effect, the proportion of Marketplace consumers offered zero-net-premium gold coverage rose from 43 percent to 73 percent. Final results are not yet available, but by the end of January’s first week enrollment was already 32 percent higher in 2023 than in 2022.

Recognizing that no insurer can afford to follow the law if competitors are allowed to break it, New Mexico and Texas strictly apply two rules to assure a lawful and level playing field.

First, carriers must use the same assumptions, for all metal tiers, about the relationship between cost-sharing and utilization.

All carriers must use CCIIO-approved factors estimating the relationship between AV and utilization.

Bohl and colleagues object that federal assumptions may not match local and plan-specific conditions. But ACA-compliant pricing reflects utilization by a carrier’s entire statewide pool. To prevent gaming, regulators should require all carriers to estimate market-wide demand using the same population-wide relationships between AV and demand, such as the CCIIO-approved ratios used in New Mexico and Texas.

Second, carriers must use the same assumptions about the distribution of silver enrollees among plans with varying AV.

Without such a requirement, insurers could skirt the law and underprice silver by assuming unrealistically low enrollment into 94 percent-AV and 87 percent-AV plans.

Texas’s required distribution reflects total statewide enrollment the previous year. New Mexico’s distribution reflects the equilibrium anticipated by ASPE, with silver enrollment limited to 87 percent-AV and 94 percent-AV plans.

Each distribution assumption yields a “CSR loading factor.” It quantifies the claims that would increase if, instead of receiving 70 percent-AV coverage, a carrier’s single risk pool was distributed among silver variants in state-specified proportions.  

Bohl and colleagues object to New Mexico’s standard as unrealistically assuming rational consumer behavior. However, a state making Texas’s choice will migrate to the equilibrium ASPE anticipated and New Mexico assumed, as silver premiums rise relative to gold. New Mexico’s movement to equilibrium immediately, rather than over time, hardly seems a fatal flaw.

Both states’ policies cause plan-level claims and revenue to diverge, which Bohl and co-authors label, “actuarially unsound.”  But such divergence is inevitable with modified community rating, which forbids a particular plan’s premiums from reflecting the distinctive characteristics of expected members. To achieve the goal of “actuarial soundness” — assuring sufficient but not excessive revenue — each carrier’s anticipated market-wide claims and revenue must balance. New Mexico and Texas achieve that goal, since they require carriers to use actual CSR costs in setting market-wide index rates.

Risk Adjustment

Risk adjustment should

  • estimate the cost of CSR consumers based on actual claims and
  • estimate revenue based on ACA-compliant premiums that reflect the distribution of silver members among CSR variants.

By filling the gap between foreseeable costs and lawful premiums, such a risk-adjustment formula would make CSR consumers and others equally profitable for law-abiding insurers, curbing carriers’ incentives to aggressively underprice silver. Below we delve into the ACA’s risk-adjustment framework and the needed revisions in risk adjustment in more detail.

Before the ACA’s coverage provisions took full effect in 2014, staff and independent researchers contracting with the Centers for Medicare & Medicaid Services (CMS) explained risk-adjustment as follows:

“[A] risk adjustment program … will assess charges to plans that experience lower than average actuarial risk and use them to make payments to plans that have higher than average actuarial risk… Without risk adjustment, plans that enroll a higher proportion of high risk enrollees would have to charge a higher average premium (across all of their enrollees) to be financially viable… Risk adjustment—if it functions as intended—allows a plan enrolling a higher proportion of high risks to charge the same average premium, other things being equal, as a plan enrolling a higher proportion of low risks.”

The American Academy of Actuaries noted that risk adjustment is most effective when it “appropriately offsets the costs of risk selection so that an insurer is ambivalent to any characteristics (such as age, income level, health status, or preferred metal tier) of any individual who chooses to enroll.” To achieve that result, the risk transfer formula, which determines what each carrier pays or receives through risk adjustment, should “align the methodology determining the risk scores with the actual factors that influence risk assumed by health plans.”

That formula has two main components. The “risk term” estimates a carrier’s likely paid claims, given the characteristics of the carriers’ individual-market enrollees. The “rating term” estimates the carrier’s ACA-allowed premium revenue. The difference between the two terms determines the carrier’s liability for or entitlement to risk-transfer payments.

After analyzing risk adjustment for CSR consumers in 2021, CCIIO reached several conclusions:

  • Originally, the risk term overestimated CSR enrollees’ costs by assuming that people in higher-AV silver plans would use many more services, without any adjustment based on income. In fact, low-income consumers enrolled in high-AV coverage used fewer services than middle-income consumers who received low-AV silver coverage.
  • Once federal CSR payments stopped, the resulting revenue loss offset the excessive risk-adjustment payments previously generated by the risk term. Nationally, the two mistakes’ effects balanced out in 2019, so the risk term turned out to be a satisfactory overall predictor of CSR enrollees’ costs.
  • However, the end of CSR payments created a problem with the rating term, which assumes that silver premiums pay for 70 percent-AV coverage. Such premiums now fund coverage with AV above 70 percent. Risk adjustment accordingly continues to overpay plans for silver enrollees, but the overpayment now results from the rating term rather than the risk term.

CCIIO should fix CSR risk adjustment by changing both terms. The risk term needs revision to ensure accurate prediction of CSR enrollee costs. There is no guarantee that two mistakes in the current formula—overestimating utilization and ignoring the absence of federal CSR payments—will continue to balance out as they did in 2019. Moreover, their national balance does not mean that they precisely offset each other with all groups of CSR enrollees. A much sounder approach would have CCIIO separately calibrate risk factors to actual claims incurred by individual-market enrollees in high-AV silver plans.

The rating term needs revision to fit average silver AV under current conditions. States vary in their distribution of silver enrollment, as illustrated by Exhibit 2 in the article by Bohl and colleagues. To accommodate major differences between states, the rating term should incorporate three alternative factors for average statewide silver AV: one would apply in states that remove all high-AV CSR members from the Marketplace by implementing the Basic Health Program (BHP); another would be used in states that remove many of the lowest-income CSR enrollees from the Marketplace by expanding Medicaid to 138 percent of FPL; and the third would be for states that neither implement BHP nor expand Medicaid.     

The Law And Good Policy Align   

As Bohl and colleagues suggest, good policy outcomes cannot justify federal legal violations. But in this case, the law and the good align: strictly enforcing ACA pricing requirements can yield significant savings for consumers, especially in communities of color. Moreover, fixing risk adjustment can incentivize insurers to align premiums with coverage generosity, as required by the ACA.

The largest remaining mystery is how quickly other states will follow New Mexico’s and Texas’s lead.

Authors’ Note

Mr. Dorn previously worked for Families USA, which advocated for the premium alignment policies adopted in New Mexico and Texas and which contracted with New Mexico to provide consulting services to inform the state’s development of premium alignment policies.