CMS finalized the 2027 Notice of Benefit and Payment Parameters on May 15, 2026 and folded a narrow duty into its overhaul of the Unified Rate Review Template: issuers that load silver premiums to cover unreimbursed cost-sharing reductions must now document that arithmetic in the actuarial memorandum for every year CSRs go unfunded.

Load factors already exceed 40% of the silver premium in some states. The requirement lands on a pool that shrank in the year the calculation is being rebuilt for.

Key Takeaways

  • Expected loading revenue "must not materially exceed" expected CSR costs, a substantive constraint CMS imposed alongside the disclosure, replacing the PY2026 flexibility to submit estimates rather than reconciled figures.
  • Marketplace enrollment fell 5% to 23.1 million in 2026 with healthier enrollees leaving disproportionately, so the CSR cost spreads across a smaller and more CSR-concentrated denominator.
  • Preliminary 2027 filings from 77 insurers across 16 states and DC show a median increase of 14%, with 20 requesting more than 20%, about four points of it attributed to the residual pool.
  • The 2027 risk adjustment models were recalibrated on 2021, 2022 and 2023 claims, years that predate the contraction the CSR load is now being reconciled against.
  • Eight provisions were enjoined July 16, 2026, four days before they applied, including the 130% bronze out-of-pocket band. The CSR loading duty was not among them.

What the Memorandum Now Has to Show

The requirement is specific rather than aspirational. Issuers filing 2027 rates must report actual 2025 CSR costs, an estimate of 2027 CSR costs, and an estimate of the additional revenue their rates generate through loading, then explain how the loading was developed and compare expected revenue against expected cost (Georgetown CHIR and SHVS, May 2026).

CMS adds a substantive test on top: expected loading revenue "must not materially exceed" expected CSR costs for the year. That is narrower than PY2026, when the rate filing instructions expressly permitted estimates rather than reconciled figures.

How CMS is enforcing it matters as much as the requirement. There is no new codified regulatory text. The reporting duty and the materiality test arrived through the PY2027 rate filing instructions carrying the URRT changes, which raises an open question about deference if an issuer ever challenged an enforcement action, since sub-regulatory guidance skips notice and comment. It also puts state regulators on the hook: CMS has told states that failing to ensure silver loading is not excessive could jeopardize Effective Rate Review State status.

That drafting choice is why the duty survived July. On July 16, 2026, Judge Brendan Hurson of the District of Maryland preliminarily enjoined eight provisions of the Payment Notice in City of Columbus v. Kennedy on Administrative Procedure Act grounds, effective July 20, the same day they would otherwise have applied (AHA, July 2026).

The blocked provisions include the wider bronze de minimis band that would have allowed a maximum out-of-pocket limit up to 130% of the standard annual limitation, or $15,600 for self-only coverage, with catastrophic plans following in 2028 (HFMA). Because the CSR loading duty lives in the filing instructions rather than the challenged regulatory text, it stands.

The Load Factor Rises Before Trend Is Applied

Cost-sharing reductions lower out-of-pocket costs for enrollees between 100% and 250% of the federal poverty level who buy silver plans, and federal law requires insurers to provide the benefit whether or not the government reimburses it. Since a 2017 Department of Justice opinion concluded the permanent appropriation could not fund reimbursement, most states have required or permitted issuers to concentrate the recovery on silver premiums (KFF). Only silver carries the obligation, so only silver needs to absorb it.

The effect travels because premium tax credits are set against the second-lowest-cost silver plan in a rating area. Loading raises the benchmark, the subsidy rises with it, and net premiums fall for subsidized enrollees who buy bronze or gold instead. Benchmark silver premiums rose roughly 17 percentage points more than bronze between 2017 and 2018, the first full year of widespread loading.

The pool underneath that arithmetic has moved. Marketplace enrollment fell 5% to 23.1 million in 2026 after enhanced premium tax credits expired, with healthier and higher-income enrollees leaving disproportionately. Preliminary 2027 filings from 77 insurers across 16 states and DC show a median proposed increase of 14%, most between 10% and 20%, with 20 above 20% (Peterson-KFF, July 2026). Issuers attribute roughly four points of the 2026 increase to the sicker residual pool and expect four more in 2027.

That runs directly through the loading calculation. Unreimbursed CSR cost scales with how many CSR-eligible enrollees a plan carries; the base it spreads over scales with total silver enrollment. As price-sensitive enrollees exit and the denominator shrinks and concentrates, the load factor rises before any medical trend is applied. An actuary rolling a 2026 load factor forward as a flat trend adjustment, rather than rebuilding it from the 2027 enrollment base, understates the figure the memorandum is supposed to reconcile.

The materiality test is where that lands. If a filing shows $40 million of actual 2025 CSR cost and projects $42 million for 2027 against $46 million of expected loading revenue, the roughly 10% gap is what a regulator weighs, not the dollar totals. CMS has published no numeric threshold, so the defensible submission documents the judgment calls behind the estimate, incurred-but-not-reported timing, enrollment mix and provider payment trend, rather than asserting a bottom-line percentage.

Two Loops the Disclosure Documents but Does Not Close

The first runs through risk adjustment. CMS recalibrated the HHS-HCC models for benefit year 2027 using enrollee-level claims from the 2021, 2022 and 2023 benefit years. Those years predate the 2026 contraction and the shift toward a smaller, sicker residual pool, so the risk scores determining an issuer's 2027 transfer are fitted to a healthier population mix than the one enrolling now.

An issuer squaring its CSR load against actual 2025 experience is therefore reconciling one part of its 2027 rate to current data while the transfer meant to offset that rate still reflects a distribution three to five years old. The memorandum documents around the mismatch. It does not resolve it.

The second runs through metal-tier shopping, and it feeds itself. A loaded silver benchmark routinely prices bronze and sometimes gold below or near zero net premium for subsidized shoppers, steering enrollment out of silver. Unsubsidized shoppers, above 400% of the federal poverty level or buying off-exchange, get none of that offset and face the full loaded premium, which pushes the healthiest and most price-sensitive of them toward bronze, off-exchange, or out of coverage.

That exit is the same movement producing the morbidity trend above. Fewer unsubsidized silver buyers means a smaller and sicker silver population carrying the load in the current year's arithmetic, which raises next year's load factor, which widens the subsidized bronze incentive again. A load documented in isolation from the metal-tier mix shift is one link of a loop, and the disclosure requirement does not ask anyone to model the rest of it.

Further Reading

Sources