Access is not enrollment: reading employer health-benefit benchmarks

The March 2025 snapshot

These are published BLS National Compensation Survey estimates for private industry, not all civilian workers and not state and local government workers. Size categories describe the sampled establishment, usually a particular location, rather than necessarily the total company.

Source: BLS Table 2, medical care benefits, March 2025, checked against the dated September 25, 2025 release. Rates are rounded as published. Take-up values are retained from BLS, not recalculated from the displayed access and participation percentages.

What the differences show—and what they do not

The published access estimates are 54% for workers at establishments with 1–49 workers and 90% for those at establishments with at least 500 workers: a descriptive difference of 36 percentage points. This is not a finding that 54% of small businesses offer health coverage, and it is not a claim that increasing a company's size causes coverage to improve.

Full-time and part-time access estimates differ by 62 percentage points. An employer's mix of work schedules therefore matters when selecting a relevant comparison group. It would be misleading to read an all-worker estimate as a benchmark for a workforce consisting entirely of full-time employees.

These differences have not been tested for statistical significance in this analysis. They describe the published, rounded point estimates. Industry, occupation, workforce composition and other differences may matter; this small extract does not adjust for them or establish a causal explanation.

The gap between access and participation also does not identify why someone did not participate. It cannot establish dissatisfaction, unaffordability, an enrollment error or lack of any health insurance. Coverage available through another source and the person's reasons for participating or not participating are not measured by the selected table. Do not turn a numerical gap into an employee story.

Keep the published take-up rate

BLS calculates take-up from unrounded estimates. Dividing the rounded percentages in a displayed table can produce a different answer. For example, the part-time row shows 10% participation and 25% access, while the published take-up rate is 42%. A simple division of 10 by 25 gives 40%, but it is not a correction to BLS's rate.

The table above preserves the official 42%. Any subtraction of published access and participation rates produces a descriptive percentage-point difference derived from rounded values, not a replacement take-up rate, an estimated count of employees or a measured explanation for nonparticipation.

Use a benchmark to improve the question

Before bringing an external statistic into a benefits review, record:

1. The reference period—not just the publication or retrieval date.
2. The covered population and whether the measure describes workers, establishments or companies.
3. The denominator: all workers, eligible workers, participants or another group.
4. Whether the size band refers to a location or the whole organization.
5. The measure being compared and the limitations of the underlying estimate.

Then compare like definitions where possible. Keep the employer's own information in the approved secure process; there is no reason to put employee health information into a public benchmark workbook. If the definitions cannot be aligned, record the mismatch instead of manufacturing a score.

This March 2025 snapshot can help structure a discussion. It does not prescribe a plan, contribution level, carrier or benefit package, and it should not be presented as a forecast for the next renewal.

Methods and limitations

This note selects seven rows and three private-industry columns from BLS's medical-care Table 2: 21 published values. The selected values were checked against both the dated archived release and the current Table 2 page on August 31, 2026. The table identifies the March 2025 reference period; these are not August 2026 observations.

This analysis keeps the published rates separate from descriptive percentage-point differences calculated by subtraction. No microdata, employee records, sample weights, modeled values, missing-value substitutions or significance tests were used. The selected groups overlap across the work-status and establishment-size classifications and must not be summed or averaged to reconstruct the total.

The BLS technical note explains the survey scope, establishment definition and availability of reliability measures. Among other exclusions, the private-industry population does not include the self-employed. The BLS glossary defines access, participation and take-up. For contributory plans, participation reflects required contributions and applicable service conditions; noncontributory-plan participation has its own treatment. This is not a count of medical visits or claims.

No New York estimate, employee-preference result, insurance-price forecast, retention effect or ranking of employers is produced. The BLS source data should be credited in any reuse. This material does not imply BLS endorsement, Jerome's personal authorship, or a new KFG study. AI assisted the extraction, calculations, drafting and checking; the underlying survey and estimates are BLS's. For later decisions or reuse, check the official release and any corrections rather than treating this dated snapshot as current data.