June 28, 2017
Stabilizing the Individual Market: Risk Adjustment and Risk Mitigation
Instability and premium increases in the individual health insurance market have figured prominently in recent debates over national health policy. One important but sometimes overlooked question in these debates is the role of risk adjustment and other risk-mitigation measures in overcoming these challenges. To analyze this issue under the current rules for the individual market and under the proposed rules now under discussion in Congress and the Administration, the Urban Institute and the American Action Forum (AAF) convened a daylong, roundtable summit with the following experts from academia, industry, and the actuarial community:
- Al Bingham, Jr., FSA, MAAA, MS, Wakely Consulting Group
- Robert Book, PhD, MBA, MA, Health Systems Innovation Network, LLC
- M. Kate Bundorf, PhD, MBA, MPH, Stanford University
- James C. Capretta, MA, American Enterprise Institute
- Stan L. Dorn, JD, Urban Institute, Health Policy Center
- Robert Donnelly, MPP, Johnson & Johnson
- Paul Edattel, Energy and Commerce Health Subcommittee—Majority
- Melanie Egorin, Ways and Means Committee—Minority
- Janis Frazer, FSA, MAAA, MBA, Anthem, Inc.
- Bowen Garrett, PhD, Urban Institute, Health Policy
- Julie Goon, Anthem, Inc.
- Bradley Herring, PhD, Johns Hopkins University
- Christopher Holt, American Action Forum
- Douglas Holtz-Eakin, PhD, American Action Forum
- John Kaelin, Centene Corporation
- Timothy Layton, PhD, Harvard Medical School, Dept. of Health Care Policy (HCP)
- Kevin Lucia, JD, MHP, Georgetown University, Center on Health Insurance Reforms
- Shawn Martin, The American Academy of Family Physicians (AAFP)
- Derica W. Rice, MBA, Eli Lilly and Company
- Erin Trish, PhD, The University of Southern California
- Cori E. Uccello, MAAA, FSA, FCA, MPP, American Academy of Actuaries
This issue brief synthesizes key findings from the summit. It expresses the views of multiple experts who attended the meeting, rather than the authors’ views. After setting out common nomenclature, we discuss general themes that apply almost regardless of the rules governing the individual market, and we analyze the operation of risk adjustment (RA) in the current individual market. Finally, we explore how policymakers could structure RA and other risk-mitigation measures if the basic ground rules governing the individual market are significantly changed.
Terminology and objectives of risk-mitigation measures
- Risk selection can take several different forms.
- “Risk selection against the market” means that eligible individuals consider their expected health care costs, among other factors, in deciding whether to enroll. When such selection occurs, the market disproportionately consists of people who expect to incur higher health care costs, and this increases market-wide premiums.
- “Risk selection against a plan” means that market participants decide which plan to buy based on their anticipated health care needs. Typically, such choices are influenced by the plan’s covered benefits (including prescription drug formularies), provider networks, cost-sharing rules, and marketing practices. When high-cost consumers disproportionately select a plan, the sponsoring insurer can experience a cycle of increased premiums causing relatively healthy members to drop the plan, leading to further premium increases, and so on.
- “Risk avoidance by a plan” means that a plan implicitly discourages membership among consumers whose expected health care costs exceed anticipated revenue. Such “competition by risk avoidance” undermines the fundamental policy goal of encouraging insurers to compete by offering value to consumers. It may also spur the development of markets where consumers with significant health care needs cannot obtain affordable coverage that is comprehensive enough to meet those needs.
- Risk adjustment consists of payments to or from insurers that compensate for the difference between the expected claims costs of an insurer’s members—the members’ “risk level”—and the insurer’s premium revenue. RA can take several different forms, each seeking to minimize the impact of risk selection against plans and insurer’s consequent risk avoidance:
- RA can be “zero-sum,” shifting dollars from insurers with low-risk members to insurers with high-risk members. Alternatively, it can be “guaranteed,” where an insurer’s payments are not affected by the risk of other insurers’ members. Unlike zero-sum RA, guaranteed RA may require an external source of funding beyond premium payments in the individual market.
- RA can be “prospective,” with each member’s risk assessed based on information available at the time of enrollment, or “concurrent,” incorporating information available after the coverage year ends.
- Reinsurance reimburses insurers for a portion of covered claims when an individual member’s or a plan’s total costs exceed a specified amount. By limiting insurers’ risks for relatively unpredictable “outlier” cases, reinsurance promotes premium stability, encourages broader insurer participation, and lessens the necessary volume of RA transfer payments. Reinsurance can be
- internally funded by resources in the individual market (e.g., by lowering RA payments, surcharging individual-market premiums, or insurers voluntarily buying reinsurance); or
- externally funded by revenue sources other than individual-market premiums.
- With risk corridors, plans make or receive payments when claims, as a percentage of premiums, diverge from targeted amounts by more than a specified percentage. Like RA, risk corridors can be zero-sum, shifting funds between carriers, or guaranteed, potentially requiring external financing. In an unknown or changing market, externally financed risk corridors limit insurers’ risks of mispricing premiums because of their limited knowledge of costs in the market.
- High-risk pools (HRPs) remove certain consumers with preexisting conditions from the individual-market risk pool, thereby lowering premiums in the remaining market. Public dollars reduce HRP premiums below the amounts needed to cover the enrollees’ claims. “Invisible” high-risk pools can operate without segregating enrollees by using a form of reinsurance that bases eligibility on health conditions rather than claims costs.
Externally funded risk mitigation—guaranteed RA or risk corridors, externally funded reinsurance, or HRPs—can lower consumer premium payments market-wide by reducing the costs that such payments must cover. With lower premiums, more healthy consumers may enroll, reducing selection against the market.
At the summit, several general themes emerged that apply to RA and risk-mitigation measures, almost regardless of how the individual market operates.
First, RA is a vital tool in preventing community rating from causing harmful risk selection against plans and insurers’ consequent risk avoidance. RA lets insurers compete based on efficiency, networks, medical management, and components of consumer value, rather than by avoiding undercompensated risks. Other risk-mitigation measures serve different purposes. For example, a robust reinsurance program probably would not eliminate the need for strong RA. Reinsurance promotes market stability by helping plans cover the expenses of members who incur unexpectedly high costs, but it does not seek to fully adjust plans’ payments based on relative differences in enrolled members’ foreseeable risks.
Second, policymakers face a trade-off under almost any configuration of the individual market. Effective RA ensures that carriers can offer different levels of coverage, even though plans with the most comprehensive coverage are likely to disproportionately attract unhealthy members. However, unless RA receives substantial external funding, premiums for plans that attract comparatively healthy members must increase to finance RA for less healthy members. In a zero-sum RA system, policymakers must therefore weigh the relative advantages of
- an individual market that
- includes comprehensive offerings that meet the needs of consumers with known health problems but
- partially funds such offerings through increased premiums charged by plans that disproportionately serve young and healthy consumers; and
- a market that
- charges lower premiums because it mainly or only offers less comprehensive coverage that attracts younger and healthier consumers who pay full premiums out-of-pocket but
- either does not include options that meet the needs of people with known health problems or makes those options available only at a very high cost.
Third, RA is important but difficult to do well. If RA amounts do not correlate well with the foreseeable costs of consumers who join the individual market, plans have incentives to seek out members with excessive RA payments and to avoid those whose RA payments do not cover anticipated costs. That can destabilize markets, yield needlessly high premiums, and prevent some consumers from being offered affordable coverage that meets their health care needs. If insurers do not feel confident in their ability to predict cost exposure under an RA regimen, they may entirely avoid the affected market, limit their market offerings, or raise premiums to create a margin of financial protection. If the market’s rules change greatly, revising RA and other risk-mitigation measures could become more challenging, and federal officials will need to maintain or strengthen their existing staffing and analytic capacity. Even if the rules do not change, federal officials must continue their work of refining and strengthening RA.
Fourth, risk selection takes health status into account, independent of age. The correlation between youth and low cost is far from perfect. Within each age band, costs vary greatly based on health status. Addressing only age-based risk selection is not sufficient. Policymakers interested in limiting the harmful effects of risk selection must also address selection based on health status.
Fifth, operational details matter. Federal officials have developed an administrative structure that, after several years in operation, now works relatively efficiently. Each plan uses its own server to maintain information about members’ demographic characteristics and health care claims. Federal officials provide software that plans run on these servers to determine members’ risk. The results largely determine the amount of carriers’ payments to or from the RA system. Now that government agencies and carriers have invested the time and resources needed for this system to operate smoothly, the cost of shifting to a new system could be considerable.
Other operational details include how plans and providers gather member data and how they forecast costs and revenues. The details of RA and other risk-mitigation mechanisms can greatly affect how these functions are carried out, the resulting costs and uncertainties facing carriers and purchasers, and eventual market conditions.
The current individual market
Offsetting risk selection against plans and limiting risk avoidance by carriers
In the current individual market, RA is important in offsetting the effects of risk selection against plans, thereby limiting risk avoidance by carriers. When carriers enroll sicker populations, RA gives those insurers additional resources that generally correspond to those enrollees’ expected costs. For carriers, this increases the economic feasibility of offering plans designed to meet the needs of people with health problems in a setting where premiums cannot rise to pay for higher expected costs.
Though generally successful, RA does not perfectly account for all conditions. Limits on formularies and provider networks can contribute to keeping premiums down. However, such limits can also become vehicles for risk avoidance. Gaps in individual-market drug formularies, compared with those in group plans, may signal health conditions that plans are seeking to avoid. The same is true of provider networks that have limited representation of certain specialty groups. In these and similar cases, the details of plan design may indicate that carriers are trying to avoid members with particular health conditions, suggesting a need to increase or add RA for those conditions.
The Center for Consumer Information and Insurance Oversight (CCIIO) is making changes to improve RA’s correlation with predictable costs and thus its ability to reduce incentives for carrier risk avoidance. Starting this year, the risk adjustment model incorporates duration factors to better reflect the costs of partial-year enrollees, and beginning in 2018, data about prescription drug claims will be used to impute otherwise undiagnosed conditions and to refine RA payments based on condition severity. But some questions have emerged around CCIIO’s plan to use claims data from the individual market to calibrate RA starting in 2019. Currently, CCIIO does such calibration using commercially available data that primarily come from the large-group market. The contemplated shift has important advantages because the individual market differs from large-group coverage in the extent of risk selection against the market and in other ways. However, the shift may lock in current market inefficiencies. If certain conditions are now undercompensated, the claims data used to calibrate RA will reflect insurer responses to that underpayment.
For example, if cancer was undercompensated in current RA, insurers might discourage enrollment of cancer patients by restricting access to cancer care. People with cancer may thus appear inexpensive in EDGE server data that calibrate RA, compared with costs shown in data from other markets. Thus, using EDGE server data could lower payments to insurers for enrolling people with cancer, further increasing incentives for risk avoidance. If this problem becomes significant, CCIIO could consider a blended approach that calibrates RA based on a combination of individual-market data and data from other markets.
Another change involves CCIIO’s planned implementation, starting in 2018, of a limited reinsurance program. Internally funded by reducing RA payments nationally, the program helps carriers that enroll members who incur very high costs. As planned, it covers a very small number of people with annual claims exceeding $1 million per member. If reinsurance is later expanded so that the claims threshold drops well below $1 million, markets could become more stable. However, carriers serving low-cost geographic areas could subsidize those serving high-cost geographic areas. More comprehensive reinsurance also risks diminishing carriers’ incentive to manage very-high-cost cases efficiently. On the other hand, a substantially more expansive reinsurance program could perhaps be funded within each state, thus reducing the magnitude of cross-subsidies from low-cost to high-cost areas, and some argue that, with or without reinsurance, insurers (1) have incentives to manage the care of costly consumers, in general; but (2) have limited control over the cost of the small number of extraordinarily expensive individual cases,.
Some observers propose changes beyond those being adopted by CCIIO. Several of these suggestions may not improve RA or could prove problematic:
- Changing RA from a concurrent to a prospective model could improve plan incentives to manage care efficiently and make RA more predictable. But that modification would be challenging in the individual market because of high individual churn in and out of the market and between plans. Prospective RA requires data on members’ prior claims. Such data are unavailable for much of the individual market, where many members have recently transitioned from employer-sponsored insurance, Medicaid, or uninsurance.
- Adding socioeconomic factors to RA could potentially improve the model’s “fit” with claims. But such a change could paradoxically reduce insurers’ incentive to serve members with low socioeconomic status (SES). Some observers suggest that when SES is analyzed separately from health status indicators, low SES may reduce rather than increase RA payments.
One modification to the basic structure of RA warrants serious consideration: changing RA from a zero-sum to a guaranteed-payment program. Currently, RA creates uncertainty and instability because plans may not be able to predict how RA will affect them. A carrier’s status as RA contributor or recipient and the general magnitude of carrier payments or receipts can vary from year to year, in part because of the characteristics of other carriers’ members. If RA were guaranteed to each insurer based on the characteristics of that insurer’s members, plan liabilities and receipts would become less uncertain and more predictable. This would provide some market stabilization, but it would likely require funding external to individual market premiums. Because this step would likely lower premiums, the cost of external funding would be partially offset by savings on premium tax credits and cost-sharing reductions.
Limiting risk selection against the market
Preventing adverse selection against the market and preventing it against individual plans are two distinct goals that may require different solutions. RA that cushions individual plans against selection may worsen selection against the market because it increases the premiums charged by plans that disproportionately attract low-risk members. These plans may be the most affordable options for young and healthy consumers with incomes too high to qualify for premium tax credits, but higher premiums resulting from RA limit the plans’ ability to garner such new members.
External funding for RA, reinsurance, HRPs, or other risk-mitigation measures would lower overall premiums by substituting for premium dollars in paying a portion of claims. These lower consumer premiums would not come with any reduction in the comprehensiveness of coverage and would reduce selection against the overall individual market. If enough external funding were provided, premiums could fall to the average level required for those without chronic conditions, increasing enrollment of healthy consumers who pay full premiums.
But the desirability of external funding is affected by its source. Some policymakers may want funding to come from employer group plans. Charges could be levied as a percentage of premiums, or firms that have healthy employees could pay a higher proportionate share. Ultimately, people with group coverage would likely bear the cost of such funding, but they could benefit by receiving additional health security; the individual market would be more stable and have lower premiums, constituting a better safety net for people who lose group coverage because of layoffs or other events. At the same time, lower premiums would produce savings on tax credits and cost-sharing reductions, lessening the need for external funds.
More fundamentally, if policymakers aim to lower individual market premiums by increasing enrollment of young and healthy members, they must determine whether risk-mitigation payments that subsidize plans serving less healthy members provide optimal bang for the buck. Such payments would lower premiums market-wide, attracting some young and healthy consumers. But a finite pool of dollars might attract more of these enrollees and improve the risk pool more if those funds were spent to increase premium and cost-sharing subsidies or to finance outreach and application assistance, rather than to fund risk-mitigation measures.
A future individual market operating under different rules
Transitioning to a different market
If policymakers set very different rules for the individual market in ways that change the characteristics of participating consumers, risk-mitigation measures may also have to change. To help insurers manage the transition to a market with different enrollees, policymakers may need to create risk corridors like those that operate inside Medicare Part D, where funding is not limited to premium dollars. Such corridors would limit insurers’ losses and profits, protecting carriers against making major errors in predicting risk when markets undergo major changes. Unless safeguards like these are offered at least temporarily, carriers may not offer coverage at all or may raise premiums to provide a cushion against unknown risks.
The ACA provided for temporary risk corridors, but congressional action forestalled external funding, and plans received much less support than they were originally promised. Any future risk-corridor legislation must be crafted carefully for this risk-mitigation measure to be taken seriously as protection against unpredictable costs in a changed individual market.
Operational challenges of state flexibility
Some proposed policy changes would increase state flexibility to define the rules governing the individual market. For example, states could narrow the range of required benefits, increase insurers’ ability to vary premiums based on individuals’ age and known health risks, broaden the spectrum of permitted actuarial valuation or other parameters for out-of-pocket cost-sharing, and so on. Each state’s configuration may require the recalibration of existing risk-adjustment measures and perhaps the establishment of other state-specific risk-mitigation measures.
The details matter. Small mistakes can have grave consequences for market stability. States may find it difficult to develop sound models for RA or other risk-mitigation measures that fit the changed rules of a very different individual market; so far, every state has relied on the federal RA model rather than develop its own.
Fortunately, federal agencies can simplify this work for states. CCIIO could use national individual-market data to develop a set of general rules for adjusting CCIIO’s RA model to fit different state policy choices. For example, a rule could state, “If your state does X for EHBs, Y for age-based premium variation, and Z for permitted ranges of actuarial value, then your state’s RA model could change from the federal model by doing A, B, and C.” If such guidelines prove unsatisfactory, CCIIO might offer (for a fee that covers its costs) to recalibrate its RA model to fit state market rules and underlying demographics. In any case, federal legislation permitting increased state variation could provide for “default” options that promote administrative feasibility by retaining current insurance rules absent state action to the contrary.
Increasing carriers’ options for plan design
If a future individual market increases the range of variation among plans operating within a state, policymakers face a key choice between
- promoting plan diversity by helping insurers obtain the resources needed to offer comprehensive coverage that meets the needs of consumers with known health problems; and
- keeping premiums low for more limited health plans that may appeal to relatively young and healthy consumers.
RA could make it financially feasible for carriers to offer coverage significantly more generous than the minimum required under new rules. Without substantial external funding, premiums charged for less generous coverage would have to rise well above the amounts otherwise needed to finance such coverage, given the risk profile of likely enrollees. Those additional premium dollars would fund RA that goes to insurers providing more generous coverage, which would disproportionately attract members with high expected health costs. Without robust RA, more generous plans could cover the additional claims resulting from risk selection only by raising premiums substantially, potentially to unsustainable levels. Put simply, unless enrollees in less comprehensive plans pay increased premiums to defray the costs of enrollees in more comprehensive plans, the latter plans may no longer be offered in the market. This trade-off illustrates the cross-cutting theme noted earlier—that without significant external funding, policymakers must choose between an individual market that includes comprehensive coverage options to meet the needs of people with known health problems or a market with low premiums for the unsubsidized young and healthy that encourages more such consumers to enroll.
A related challenge faces a future individual market with substantial variation between competing plans. For RA to function effectively within a state, one set of rules must specify minimum benefits and actuarial value levels for all insurers that participate in RA. If those minimums are extremely low and other plans provide substantially more generous coverage, successful RA may be difficult to achieve. In the past, RA often succeeded in systems, like Medicaid and Medicare, that have less variation than current individual markets, not more variation.
Varying premiums based on risk
Some proposals for changing the individual market would increase plans’ ability to vary premiums based on consumers’ expected health care costs, instead of limiting premium variation to age, geography, family versus individual coverage, and (at state discretion) smoking status, as in today’s market. If risk rating were allowed market-wide, policymakers would again face a basic choice: Under one approach, RA would continue, limiting insurers’ need to vary premiums based on risk. If RA approximated known risks as estimated by actuaries, carriers would likely charge close to community-rated premiums, because higher premiums would not be needed to cover the increased claims costs of consumers with known health risks. Divergence from community-rated premiums would show a likely need to adjust the RA model.
Unless this approach was supported by significant external financing, it would have the disadvantage of raising premiums for coverage that attracts younger and healthier enrollees. Policymakers who want consumers’ premium payments to vary based on health status should instead dispense with RA entirely, possibly using HRPs to take comparatively unhealthy consumers out of the general individual market’s risk pool.
Some proposed legislation would let plans risk-rate premiums, but only for consumers who have experienced recent coverage lapses. Depending on its details, such an approach could encourage healthier consumers to drop coverage, because that would qualify them for premiums lower than the community rate. The resulting market bifurcation—healthier consumers in risk-rated coverage and less healthy consumers in community-rated plans—could destabilize community-rated coverage. The latter coverage could become extremely costly or even unavailable. RA and risk-mitigation measures may not be able to prevent such outcomes, unless policymakers forbid carriers from lowering risk-rated premiums below the community rate.
Combining strong RA with some flexibility to vary premiums and plan design based on consumers’ health status may offer one important advantage that could appeal to policymakers across the political spectrum. Some insurers could offer plans customized to provide unusually efficient care for people with specific health conditions. For example, plans may specialize in diabetes care with provider networks skilled in serving diabetics. If such a plan’s resulting costs for diabetic members fell below RA predicated on average costs, the plan would realize financial gains from enrolling such members. To further encourage diabetics’ enrollment, the plan could lower its premiums for people with the condition; this would go beyond current plan recruitment tools, such as reduced out-of-pocket cost-sharing for members with specific conditions. Attracting more people with chronic conditions to plans skilled in treating those conditions could prove beneficial. However, other steps may be necessary to prevent discrimination against people whose conditions are hard to control.
RA and other risk-mitigation measures involve technical issues that are rarely the subject of entertaining dinner-table conversation between ordinary humans. Nevertheless, resolving those issues intelligently and pragmatically is crucial to the effective functioning of individual markets. To date, federal agencies have shown considerable ingenuity in fashioning and improving RA to meet the needs of carriers and consumers alike. This work must continue, regardless of how the rules of the individual market change.
Support for this project is provided by Anthem, Inc., and Eli Lilly and Company. For more information on the Urban Institute’s funding principles, go to
 One can also speak of “adverse selection” against the market or against particular plans.
 Both forms of risk selection are influenced by premiums (including subsidies) and, in some cases, the individual mandate.
 Under the ACA, RA in the individual market is zero-sum. For Medicare Advantage and Medicare Part D, it is guaranteed. With zero-sum RA in the current individual market, premium payments that finance RA include costs borne by the consumer as well as premium tax credits and cost-sharing reductions that the federal government funds for enrollees with incomes below specified levels.
 Medicare Advantage uses prospective RA, and the ACA’s individual market uses concurrent RA.
 Community rating imposes a similar but, in some ways, more fundamental trade-off. By prohibiting premium variation based on individual risk, community rating raises premiums for healthier consumers and lowers them for consumers with foreseeable health problems. Effectively, healthier consumers cross-subsidize less healthy consumers.
 Regulators can act as a check on carriers’ ability to avoid risk, particularly in health insurance Marketplaces, where federal regulations prohibit benefit designs and marketing strategies that have the effect of discouraging the enrollment of high-cost consumers. Regulators also review benefit designs and formularies before approving plan operations.
 Most states’ individual markets—under current law and under the proposed changes—permit premium variation based on age. Such variations reduce but do not eliminate the need for age-based risk adjustment.
 For example, converting the risk-adjustment program from concurrent to prospective would greatly reduce plans’ uncertainty about the health status of both their own enrollees at the start of the year and enrollees in the market overall. However, a prospective RA program would require a centralized database with information about consumers who were formerly covered by other individual-market carriers, Medicaid, or group plans. This new kind of database would require major operational changes and likely raise privacy concerns.
 Michael Geruso, Timothy J. Layton, and Daniel Prinz, “Screening in Contract Design: Evidence from the ACA Health Insurance Exchanges,” Working Paper 22832 (Cambridge, MA: National Bureau of Economic Research, 2016), http://www.nber.org/papers/w22832.pdf.
 Mark Shepard, “Hospital Network Competition and Adverse Selection: Evidence from the Massachusetts Health Insurance Exchange,” Working Paper 22600 (Cambridge, MA: National Bureau of Economic Research, 2016), http://www.nber.org/papers/w22600.pdf.
 See: Centers for Medicare & Medicaid Services, U.S. Department of Health and Human Services. Patient Protection and Affordable Care Act; HHS Notice of Benefit and Payment Parameters for 2018; Amendments to Special Enrollment Periods and the Consumer Operated and Oriented Plan Program. Final rules. Fed Regist. 2016 Dec. 22; 81(246): 94058–94183.
 Data from other markets could be reweighted to fit the demographic characteristics of individual-market enrollees.
 60 percent coinsurance applies.
 For this to happen, reinsurance would likely need to become substantially more generous, going beyond a small number of costly outliers. Areas may have high health care costs because of limited competition between hospitals or insurers, prevailing patterns of physician practice, a high cost of living; or other factors.
 If the individual market becomes more predictable to carriers, the uncertainty introduced by zero-sum RA could diminish, and the need for guaranteed RA could become less acute.
 For most of Part D’s history, drug plans set premiums above costs, achieved favorable financial results, and reimbursed Medicare through the risk corridors program. More recently, new and costly drugs, such as those for Hepatitis C, led to plan losses that were partially reimbursed through risk-corridor payments.
 For a more detailed explanation of how risk-corridors work, see: Cynthia Cox et al., “Explaining Health Care Reform: Risk Adjustment, Reinsurance, and Risk Corridors” (Menlo Park, CA: Kaiser Family Foundation, 2016), http://files.kff.org/attachment/Issue-Brief-Explaining-Health-Care-Reform-Risk-Adjustment-Reinsurance-and-Risk-Corridors.
 See, for example: Robert Pear, “Marco Rubio Quietly Undermines Affordable Care Act,” New York Times, December 9, 2015, https://www.nytimes.com/2015/12/10/us/politics/marco-rubio-obamacare-affordable-care-act.html.
 The only state that began ACA implementation with its own RA model, Massachusetts, has now shifted to the federal model.
 This approach would depart greatly from the historical rationale for RA—that RA is needed to make the transition from risk-rated, medically underwritten insurance to community-rated coverage. RA does this by making up the difference between premiums and foreseeable health care claims.