CMS Lost the Broker Freeze in Court. 19,000 Agents Return to a Marketplace That Shrank by 2.6 Million People Last Year

CMS took effect on Sept. 22, 2026: an interim final rule pausing registration of any agent or broker without a Plan Year 2026 Exchange agreement on the federally facilitated marketplace, through Feb. 1, 2027. The National Association of Benefits and Insurance Professionals sued HHS and CMS on Oct. 15, and Health Agents for America filed separately later in October. On Nov. 5, the court struck down the pause and directed that training for new agents be available by Nov. 6, days after the 2027 enrollment period opened on Nov. 1 and more than two months before it closes in mid-January.

Two numbers in that table define the dispute. CMS's own rule says newly registered agents were about 11% of the registered base yet roughly 30% of the agents receiving termination notices, so the fraud problem was real and concentrated. But the pause would have frozen 19,000 new registrations, more than twice the 8,937 who joined in 2026, and 78% of those new registrants went on to enroll at least one consumer. The court did not decide whether the fraud was real. It decided that the only available mechanism for preventing it was disproportionate.

The pause was justified on program integrity grounds, and CMS put a price on it: preventing roughly $48 million to $877 million of improper premium tax credit payments a year, an average of $333.6 million, plus $280,000 to $1.1 million of avoided consumer administrative burden. Set against that, the enforcement action CMS announced the same day — cancelling 315,000 policies covering more than 760,000 people — returned about $2.2 billion of taxpayer funds in one stroke. The agency that modeled the pause at hundreds of millions of dollars a year had already clawed back more than two billion dollars from a single administrative action.

Restored distribution capacity is only as valuable as the products behind it, and the 2027 product map is being redrawn. Eleven carriers had announced marketplace exits for plan year 2027 as of Oct. 28, 2026, and the exits cluster among the national carriers that ran the largest multi-state individual books. The result is a market with fewer competitors, higher prices and a bigger share to go around — which is precisely the environment in which a restored agent network can be worth something.

The carriers that left did so on margin, not on demand. Molina's decision to shed seven states for 2027 comes with an expectation of a $1.0 billion decrease in marketplace premium as it trades volume for target margins, and its nine-month marketplace medical care ratio of 88.3% sits below the 92.6% it earns on Medicaid. Cigna's exit of roughly 369,000 members removes a book that was already down 17% year over year. What remains for the restored agents is a market where the marginal carrier has decided the individual book no longer clears its hurdle rate — except the ones with structurally low medical costs and the ones choosing to buy share instead.

The channel has a clear upstream and a clear downstream. Upstream sit the licensed agents and brokers whose registration CMS paused and the state-based exchange platforms that stayed open; downstream sit the carriers paying the commissions, the Medicaid managed care plans absorbing the churn, and the 23.1 million consumers who selected a plan in 2026. The ruling restores the upstream link without changing the size of the downstream pool, which is why the price of a restored agent channel shows up in carriers' commission expense rather than in their enrollment lines.

Molina is the cleanest illustration of the trade the ruling interrupts. Its marketplace book shrank from 655,000 members at the end of 2025 to 267,000 at Sept. 30, 2026 while Medicaid slipped from 4.57 million to 4.37 million, so the exchange absorbed the churn and Medicaid held the company together. Then comes the 2027 policy change: 43 states and the District of Columbia must implement Medicaid work requirements by the start of the year, and a person who loses Medicaid solely for failing a work requirement also loses eligibility for premium tax credits. The exchange stops being the fallback for the Medicaid loss it used to be.

The contrast in the table is the whole trade. Elevance grew individual exchange revenue to $2,514 million from $2,358 million while individual membership was flat, so it priced its book rather than chased volume, and it can afford agents because the cost sits inside a $43.4 billion segment. Oscar runs an 84.0% medical loss ratio on 2.69 million members and is buying counties in South Carolina and Ohio while rivals retreat. Centene's marketplace revenue fell to $8.7 billion from $11.0 billion on a 41.6% membership decline, and Molina's marketplace medical care ratio jumped to 92.5% in the third quarter, which is why Molina is cutting its 2027 marketplace premium by about $1.0 billion rather than paying more commissions into a risk pool it has already repriced upward.

The evidence gap is narrow and dated. CMS has disclosed no post-injunction registration count, no revised enrollment forecast for 2027, and no analysis of how much of the 760,000 people dropped on Aug. 31 came from the agent-assisted population the pause was designed to protect. Until the January plan-selection report and the spring effectuated enrollment tables arrive, the court has restored supply into a market whose demand is a matter of estimate, and the most defensible position is that the ruling changes competitive share rather than the size of the pie.