Community Rating Failed

The prevailing view frames level funded health plans in the small group market as a form of regulatory arbitrage. Healthy small employers use ERISA preemption to escape community rating, cherry-pick low-risk populations into self-funded arrangements, and leave the sick and expensive behind in the community-rated pool. The growth of level funded is, on this account, a story of market gaming that undermines a public policy designed to make health coverage accessible to small employers with unhealthy employees.

This article takes the opposite position. The ACA’s adjusted community rating for the small group market did not solve the affordability problem it was designed to address. It priced healthy small groups out of a market segment and produced the adverse selection it was meant to prevent. Level funded is not regulatory arbitrage. It is the market’s correction of a pricing policy that failed on its own terms. The growth of level funded from a niche product in 2013 to the coverage structure for an estimated 36 percent of covered workers at small firms by 2024 is not evidence of bad actors gaming the system. It is evidence of what happens when a subsidy mechanism depends on the voluntary participation of the people being taxed.

What Community Rating Was Supposed to Do

#

Before the ACA, the small group market used medical underwriting, experience rating, and, in many states, preexisting condition exclusions. A small employer with a diabetic employee or a cancer survivor on staff faced premiums that reflected that population’s actual health risk. A competitor with younger, healthier employees paid substantially less. The policy problem was real: small employers with sick employees faced unaffordable premiums or were denied coverage altogether, and the employees themselves bore the consequence.

The ACA’s adjusted community rating, effective January 1, 2014 for small group plans, prohibited premium variation based on health status, gender, or claims history. Carriers could vary premiums by age (within a 3:1 ratio), geographic area, family composition, and tobacco use. Everything else had to be uniform across the risk pool. The design intent was to spread risk across all small groups so that no single group’s premium reflected its own health status. Healthy groups would subsidize sick groups implicitly, and every small employer would have access to coverage at a predictable price.

What Community Rating Actually Produced

#

The mechanism has a structural weakness that was identified by economists before the ACA was written. Rothschild and Stiglitz’s foundational 1976 analysis of competitive insurance markets demonstrated that pooling heterogeneous risks in a community-rated market is not a stable equilibrium when participation is voluntary. Groups whose risk is priced below their actuarial cost will want to stay. Groups whose risk is priced above their actuarial cost will want to leave. If a cheaper alternative exists for the lower-risk groups, they exit. The pool that remains has higher average risk than before. Premiums rise to reflect it. More low-risk groups exit. The cycle continues until the pool is composed entirely of the groups who cannot leave.

The strongest empirical counterargument comes from the 1990s. When New York implemented pure community rating for its small group and individual markets in 1993, Thomas Buchmueller and John DiNardo found no evidence of an adverse selection death spiral in data from 1987 through 1996. Coverage rates in New York did not fall relative to Pennsylvania (no reform) or Connecticut (moderate reform). The Rothschild-Stiglitz prediction did not materialize in that context. That finding needs to be addressed directly, not ignored.

Two structural differences distinguish the 1990s state-level community rating experiments from the ACA small group market. First, New York’s 1993 reforms had no broadly available exit vehicle for healthy small groups. ERISA self-funding existed, but level funded structures were not yet a scaled product marketed to small groups with health-status underwriting. The alternative market was thin. When Buchmueller and DiNardo measured the outcome, the groups that should have exited largely had nowhere cost-effective to go. Second, New York’s reforms did produce a structural shift: HMO penetration increased dramatically as healthy groups moved from fee-for-service indemnity products into managed care arrangements with lower premiums. The adverse selection manifested as managed care growth, not as market exit. The pool restructured rather than collapsed.

The ACA small group market presents a different topology. By 2014, level funded was a mature product offered by multiple national carriers and TPAs, actively marketed to groups with favorable health experience, with ERISA preemption providing clear federal protection from state community rating rules. The exit vehicle existed at scale from day one. CMS Office of the Actuary estimated in 2014 that 65 percent of small group employers offering insurance would face premium increases under ACA community rating regulations, creating an enormous addressable market for level funded as an alternative. The mass exit that theory predicted but Buchmueller and DiNardo did not observe in 1990s New York proceeded because the pre-conditions were different: a scaled product, a clear legal pathway, and active distribution infrastructure all existed simultaneously.

In the small group market post-ACA, the exit vehicle was ERISA preemption. Self-funded plans, and the level funded structures built on top of self-funding, are not subject to state insurance regulation. They do not participate in the community-rated risk pool. An employer who moves from a fully insured community-rated product to a level funded arrangement exits the subsidy mechanism entirely. The level funded product underwrites based on the group’s actual health experience, which is the pricing approach community rating was designed to prevent and which the market rationally prefers when the alternative is paying a community-rated premium that overprices the group’s actual risk.

A 2018 study published in the Journal of Risk and Insurance, using cross-state variation in pre-ACA rating regulations and KFF/HRET Employer Health Benefits survey data, found that lower-risk employers subject to laws limiting allowable premium rating variation had a predicted probability of self-insurance approximately 18 percentage points higher than otherwise similar higher-risk employers. The research confirmed what economic theory predicted: community rating creates a differential incentive to self-insure precisely for the groups the community-rated pool needs to retain in order to function. The exit is not a market failure. It is a rational response to mispricing.

The Enrollment Evidence

#

The enrollment data documents the outcome. Fully insured small group enrollment fell from approximately 17 million in 2013 to approximately 10 million in 2023, a decline of more than 40 percent over the decade following ACA implementation, according to Peterson-KFF Health System Tracker analysis of commercial market data. Oliver Wyman’s analysis of NAIC filings found a 26 percent decline in fully insured small-group enrollment from 2016 to 2023 alone, a compound annual rate of approximately 6 percent per year. The Urban Institute estimated in a 2023 analysis that self-insured and level funded plans had grown to account for roughly 45 percent of total small group enrollment when including the self-funded segment.

The deterioration of the remaining pool is documented in carrier rate filings. KFF Health System Tracker analysis of 318 small group insurer rate filings for plan year 2026, submitted across all 50 states and the District of Columbia, found a median proposed premium increase of 11 percent. Multiple insurers attributed the increases explicitly to declining enrollment and worsening risk pool morbidity. Anthem Health Plans of Maine reported that its small group ACA market size had declined 11.9 percent in a single year and projected a further 10 percent decline in 2026. A carrier in Indiana cited anti-selection: groups with better-than-average experience were moving to self-funded products, leaving behind those who could not qualify for health-status underwriting. This is the adverse selection spiral operating in real time, and it was produced by community rating, not by level funded.

The Subsidy That Was Not Funded

#

Community rating is an implicit cross-subsidy from lower-risk groups to higher-risk groups. This distinguishes it from explicit subsidies. The ACA Marketplace uses premium tax credits funded by federal tax revenue, meaning the subsidy comes from a broad tax base and does not depend on the continued participation of any particular market segment. The small group community rating subsidy is funded by the excess premiums of lower-risk groups who remain in the community-rated pool. When those groups exit, the subsidy disappears because the people paying it have left.

Explicit risk adjustment mechanisms exist within the fully insured small group and individual markets under the ACA. The ACA’s permanent risk adjustment program transfers funds annually from plans with lower-risk enrollees to plans with higher-risk enrollees within the same market and state, using HHS methodology to estimate plan actuarial risk relative to the market average. This mechanism operates entirely within the community-rated pool. An employer who moves to level funded exits the risk adjustment framework. The self-funded market has no equivalent mechanism, no contribution obligation, and no exposure to redistribution. This is the design gap that allows the adverse selection spiral to proceed. The exit vehicle exists and is legally protected by ERISA preemption. The mechanism to fund the subsidy from outside the pool was never built.

The practical consequence of this gap is measurable. As the 45 percent of small group covered workers estimated by Urban Institute to be in self-funded or level funded arrangements grow further, the base of lower-risk participants contributing to the community-rated pool’s risk adjustment shrinks. The exit comes disproportionately from healthier employers, precisely those whose participation would do the most to offset the higher-cost groups remaining. The risk adjustment program compensates for variation within the pool but cannot compensate for the secular exit of the pool’s healthiest members.

The policy response that follows from this analysis is not to attack the exit vehicle. Eliminating ERISA preemption for small group self-funded plans, or requiring level funded arrangements to contribute to state risk adjustment pools, would eliminate the mechanism that corrects community rating’s mispricing but would not fix the underlying problem. Healthy small groups priced above their actuarial risk would have no alternative market. They would drop coverage or absorb premiums that exceed the actuarial value they receive. The argument that constraining level funded would stabilize community-rated pools is empirically contestable: before level funded became widely available, unhealthy small employers still faced coverage access problems, and many healthy employers operated without coverage rather than cross-subsidize a risk pool with no mechanism to retain them.

Level Funded as Market Correction

#

The reframe this article argues for is direct: level funded did not break the small group market. Community rating broke the small group market for healthy employers. Level funded restored pricing that reflects actual risk for groups willing to accept the transparency and variance that risk-appropriate pricing entails.

The policy implication is not that community rating was wrong to attempt. The objective, making coverage accessible to small employers with unhealthy employees, was legitimate. The mechanism, an implicit cross-subsidy dependent on voluntary participation from the groups being taxed, was predictably unstable in the presence of an exit option. The ACA’s architects understood this risk: the individual mandate was designed to limit exit from the individual market by making non-participation financially costly. No equivalent constraint applied to the small group self-funded market, and the employer mandate’s applicability threshold of 50 or more employees left the small group market outside its reach.

The data on small group premium increases, risk pool deterioration, and insurer exits from the community-rated segment tells a story about a subsidy mechanism that has been losing its funding base for over a decade. A 2025 Oliver Wyman analysis cited by KFF Health System Tracker noted that individual market premiums had historically been 23 percent below small-group premiums through 2022, creating a competitive dynamic where even the ICHRA pathway to individual coverage was more economical than community-rated small group. The community-rated pool’s competitive position has deteriorated not because of predatory competition but because it overprices the majority of small groups relative to what they can access elsewhere.

That is not a market failure. It is a market signal. Level funded is the vehicle the signal has moved through.

The state-level pattern is instructive. Insurer exits from the community-rated small group market have been concentrated in states where alternative products are most accessible, where stop loss attachment point floors are low, and where ERISA preemption runs cleanest against state insurance regulation. States with aggressive small group community rating enforcement and high minimum attachment point floors for stop loss, such as New York, have retained larger community-rated small group pools than states where level funded can be deployed without those constraints. This is consistent with the hypothesis that the exit is price-driven and responds to the availability of alternatives, not to some intrinsic preference among employers for administrative complexity. Where the exit is harder, more employers stay in the community-rated pool. This does not mean the community-rated pool is working; it means the exit has been blocked. The underlying mispricing for healthy groups remains regardless of whether they can act on it.

The Honest Limit of the Counter-Thesis

#

This article’s argument has a limit that honesty requires acknowledging. The evidence that community rating produces adverse selection in the small group market is compelling. The evidence that a market without community rating is better for the employers and employees currently served by community-rated products is not. The employers who remain in community-rated small group plans are largely those whose health status prevents them from accessing level funded underwriting or who lack the administrative sophistication to move. Their situation does not improve if community rating collapses and no alternative subsidy mechanism exists to replace it.

The counter-thesis is not that community rating should be eliminated without replacement. It is that the current architecture, which nominally preserves community rating while allowing systematic exit by those who fund it, produces the worst possible outcome: the subsidy mechanism exists in name but is progressively defunded, premiums in the remaining pool rise, and the employers with the greatest need face the worst pricing. If the objective is to make coverage accessible to small employers with unhealthy employees, the current architecture is failing that objective while generating the appearance of policy stability.