CMS · Medicare Advantage Console

CMS Console

Discovered · Catalogued · Analysed

source files
rows
96.2% PR MA penetration

Benchmark adequacy in the nation's highest-performing Medicare Advantage market

Puerto Rico operates the highest quality, lowest cost Medicare Advantage market in the United States.

By every performance measure CMS publishes, Puerto Rico's Medicare Advantage market leads the program. MA penetration stands at 96.2% — MA enrollees as a share of beneficiaries with both Parts A and B (CMS Medicare Monthly Enrollment, May 2026) — the highest in the nation, against 55.5% on the mainland. Three of the island's four largest contracts hold 5-Star ratings under the 2026 Star Ratings. MCS Classicare (H5577) is the largest 5-Star MA contract in the United States. 99.1% of Puerto Rico's MA enrollees are in plans rated 4.5 Stars or higher. No state combines this scale and this quality.

These outcomes are the stated objectives of Medicare Advantage policy: broad private-plan participation, competition on quality, minimal reliance on traditional fee-for-service, and lower per-member cost to the federal government. Puerto Rico has already delivered each of them, at scale, under CMS's own measurement system.

The payment side does not match the performance side. For CY2027, Puerto Rico's average benchmark is $778 PMPM59% of the mainland average ($1,313 PMPM) and 80% of the US Virgin Islands ($967 PMPM), an adjacent territory operating under the same Medicare statute. The program's strongest market is paid its lowest rates — an inversion no payment methodology should produce, and none should be able to defend.

The policy proposal is narrow: apply to Puerto Rico the same benchmark treatment CMS already applies to the US Virgin Islands. This is not a request for mainland rates or for any special adjustment. It is the application of an existing CMS payment level to a directly comparable jurisdiction — adjacent geography, identical statute, identical program rules — and it directs the correction to the only market in the program with a demonstrated 5-Star record at this scale. Of the corrections available to CMS, it is the most limited in scope and the most firmly grounded in the agency's own precedent.

The question this dashboard puts to CMS is analytical, not rhetorical: what payment-policy rationale supports a $189 PMPM differential between two adjacent US territories under the same statute, when the lower-paid market demonstrably outperforms the higher-paid one — and every state in the program? Absent such a rationale, benchmark parity with the US Virgin Islands is the defensible floor.

Every month · every member
$ less
What Medicare pays per member, each month Dollars per member, per month · 2018–2027
Puerto Rico
avg 0% bonus rate
US Virgin Islands
neighboring territory · same statute
Guam
Pacific territory · same statute
US mainland
avg 0% bonus rate
PR 5-star contracts
3
81.7% of PR MA enrollment
PR market HHI
3,514
highly concentrated · few carriers = fragile

Sources: rates — CMS CY2027 MA Ratebook (final Rate Announcement, April 2026); star ratings — CMS 2026 Part C & D Star Ratings; enrollment & penetration — CMS Monthly Enrollment by CPSC, August 2026.

Rate trajectory · 2018 to 2027

Puerto Rico vs the US mainland, in dollars and percent

Both panels read left-to-right by ratebook year. Top: dollar amounts at each point — with the US Virgin Islands as the territorial comparator (USVI receives $967 in 2027 vs Puerto Rico’s $778, despite both being territories). Bottom: PR as % of US and year-over-year growth.

Source: CMS Medicare Advantage Ratebooks, CY2018–CY2027 — unweighted average of county-level 0%-bonus benchmark rates per jurisdiction; full file list and methodology in "Sources & methodology" below.

Source: CMS Medicare Advantage Ratebooks, CY2018–CY2027 — unweighted average of county-level 0%-bonus benchmark rates per jurisdiction; full file list and methodology in "Sources & methodology" below.

Year by year

Year PR $US $ USVI $ PR as % US PR as % USVI PR Y/YUS Y/Y Gap vs US Gap vs USVI

What plans were actually paid — the CY2024 plan payment file

The charts above show the benchmark — the ceiling CMS offers each county before bidding. This section shows the payment: what CMS actually paid Medicare Advantage plans per member per month in CY2024, after bids and risk adjustment, straight from CMS’s own county-level plan payment file. Different instrument, same verdict — and the payment file adds a fact the benchmark can’t show: Puerto Rico’s scored population is sicker than the mainland average, yet its plans are paid the least in the nation.

Puerto Rico
$466
avg A/B payment per member-month · the only jurisdiction under $500
US Virgin Islands
$773
next-lowest · still $307 above PR
US (states + DC)
$903
mean of county-level rows, territories excluded
PR rank
55 of 55
lowest average payment of every jurisdiction in the file

Source: CMS 2024 Plan Payment zip file, 2024PartCCountyLevel.xlsx — unweighted mean of “Average A/B PM/PM Payment” across county × plan-type rows per jurisdiction (the file publishes no enrollment weights); verification register [18] on the Circularity tab.

The double edge — what a critic will find in the same file

Read the whole file before quoting it: the same spreadsheet that puts PR dead last on payment also shows PR plans receiving the highest rebates in the nation and carrying an above-average risk score. Both cut in Puerto Rico’s favor once the definitions are on the table.

Plan rebate · PR vs states+DC
$257 vs $177
highest of all 55 (2nd: Florida $234.87; states+DC mean $177.30). “Rebate” here = the share of the bid-vs-benchmark gap plans must return as extra benefits (not drug rebates) — PR plans bid far below even PR’s floor-of-the-nation benchmark, and beneficiaries get the difference as supplemental benefits.
Part C risk score · PR vs states+DC
1.53 vs 1.38
paid least despite a sicker-than-average scored population — 5th-highest of the 55 jurisdiction codes under this construction. High, not highest — DE (1.70), OH, PA and NC score higher, so the payment gap cannot be explained as a coding artifact unique to PR.
Part D low-income cost-sharing · PR vs national
$8 vs $69
per member-month, LICS subsidy. 70 of the 85 PR plan rows are $0.00 — five of six PR contracts at zero on every plan — because island duals are excluded from LIS. The NO-LIS exclusion, visible as a payment column.

Context, labeled as such: Puerto Rico’s CY2024 average MA payment ($466/member-month) is well below the national average FFS spending MedPAC reported nearly a decade earlier (≈$784/month in 2016) — a different program and measure, but a marker of how far the level itself lags. Console’s own comparison of two published figures.

Sources & methodology

Every dollar figure on this tab is the unweighted average of county-level capitation rates at the 0% quality-bonus tier (the base ratebook benchmark), computed per jurisdiction per year directly from CMS's published Medicare Advantage Ratebooks. Puerto Rico, the US mainland, and the US Virgin Islands are each averaged across their own counties/municipios. Rates are the "Part A + Part B" standardized county rates as released in each annual ratebook ZIP.

CMS Medicare Advantage Ratebooks (one per benchmark year)

Landing pages & methodology

All ten ZIPs are catalogued with their SHA-256 checksums on the Inventory tab; the per-year parsing logic (drift-tolerant column matching across CMS's changing schemas) is documented on the Architecture tab and reproducible from the SQL tab. Figures last verified against the 2027 final ratebook, April 2026.

★★★★★

The merit case

Five-star plans, two-star pay

Three Puerto Rico contracts hold five-star ratings in 2026 — covering 81.7% of all PR Medicare Advantage enrollment (552,146 of the 675,423 MA enrollees CMS reports for August 2026), with one of them (MCS Classicare, H5577) being the largest 5-star contract in the entire country at over 350,000 enrollees. Adding the 4.5⭐ Triple-S contract (H5774) brings coverage to 99.1% of PR MA in 4.5⭐+ rated plans. Yet these plans operate on benchmark rates that put them at the bottom of the national table. If CMS payments rewarded merit — as the Star Ratings bonus program was designed to do — Puerto Rico would be the showcase, not the exception.

Enrollment: CMS Monthly Enrollment by CPSC — August 2026  ·  Star Ratings: CMS Part C & D Performance Data — 2026 (2026 Star Ratings Fact Sheet)

Four-and-a-half stars

August 2026 snapshot

The Puerto Rico Medicare Advantage market

675,423 MA enrollees across 6 MA contracts and 78 municipios. Three corporate parents hold ~99% of the market. (Excludes 9,332 standalone Part D / PDP members, who are not in Medicare Advantage.)

Active MA contracts
6
H-contracts, 78 municipios
Total MA enrolled
675,423
August 2026 · excl. PDP
HHI
3,514
highly concentrated · few carriers = fragile
Top contract
52.9%
MCS Classicare alone

All PR contracts by enrollment

Data as of August 2026 · Source: CMS Monthly Enrollment by CPSC · Stars: CMS Part C & D Performance Data

ContractBrand · ParentStars EnrolledShare

Denominator for all star-share figures: CMS-reported Puerto Rico Medicare Advantage enrollment, 675,423 (Monthly Enrollment by CPSC, August 2026). Excludes 9,274 stand-alone Part D (PDP) enrollees — a separate program — and CMS-suppressed cells of ≤10 enrollees. The ~190 mainland MA contracts with Puerto Rico service-area rows report zero countable enrollment in PR.

Supporting charts

Interactive — hover for exact values, drag to zoom, click legend items to toggle series, double-click to reset.

PR rate trajectory 2018–2027

How to read it: the monthly amount CMS pays a Medicare Advantage plan per enrollee (the benchmark). Green is the US mainland average; red is Puerto Rico. Two adjacent US territories are shown for reference: US Virgin Islands (blue) and Guam (amber). In 2018: mainland $850 vs PR $483. In 2027: $1,313 vs $778. PR never converges with the mainland — the dollar gap widens from $367 to $535 per member, per month — and PR sits below both territories throughout.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — annual county rate books, CY2018–CY2027 (files 2018–2027-ma-rate-book), and the CY2027 Rate Announcement. Values computed from the published county rate book benchmarks.

PR vs other states & territories

How to read it: every jurisdiction’s average 2026 benchmark, highest to lowest. Alaska tops the list at $1,470; the US average is $1,227 (dashed line); the neighboring US Virgin Islands receives $939. Puerto Rico, at $730, is second-to-last of 56 — only American Samoa ($672) is lower.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — annual county rate books, CY2018–CY2027 (files 2018–2027-ma-rate-book), and the CY2027 Rate Announcement. Values computed from the published county rate book benchmarks.

PR as percent of US

How to read it: the same gap as a percentage: for every $1.00 CMS pays for a mainland enrollee, it pays 57–60 cents for an American citizen living in Puerto Rico (59.3% in 2027). The line is essentially flat — a decade of rate cycles has not closed the gap. For reference, the same ratio for US Virgin Islands (blue) holds near 74–77%, and Guam (amber) rises from 59% to 86% — both adjacent US territories under the same statute, both paid more relative to the mainland than Puerto Rico.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — annual county rate books, CY2018–CY2027 (files 2018–2027-ma-rate-book), and the CY2027 Rate Announcement. Values computed from the published county rate book benchmarks.

FFS vs MA gap — the key finding

How to read it: benchmarks are set as a multiple of local fee-for-service (FFS) spending. On paper, PR’s benchmark looks generous — 1.48× its measured FFS in 2024 vs 1.13× on the mainland. The next two charts show why that FFS yardstick is broken in Puerto Rico.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — FFS Data files, CY2010–CY2024 (files ffs-data-2010 … ffs-data-2024): per-capita fee-for-service costs and eligible counts by county, as published by CMS for MA rate-setting.

FFS dollar levels

How to read it: the yardstick itself: measured FFS spending per beneficiary is $992/month on the mainland but only $455 in Puerto Rico (2024). With ~95% of the island already in MA, almost no one remains in traditional Medicare — so the FFS base that anchors PR’s benchmark is built on a sliver of atypical enrollees.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — FFS Data files, CY2010–CY2024 (files ffs-data-2010 … ffs-data-2024): per-capita fee-for-service costs and eligible counts by county, as published by CMS for MA rate-setting.

FFS credibility — denominator

How to read it: how thin that sliver is: the average mainland county has ~9,400 FFS beneficiaries to measure costs from; the average Puerto Rico municipio has 376 (2024) — down from 687 in 2018. Note the logarithmic scale: the gap is roughly 25-fold.

Source: CMS Medicare Advantage Ratebooks & Supporting Data — FFS Data files, CY2010–CY2024 (files ffs-data-2010 … ffs-data-2024): per-capita fee-for-service costs and eligible counts by county, as published by CMS for MA rate-setting.

Market concentration over time

How to read it: the Herfindahl-Hirschman Index (HHI) of PR’s MA market. Above 2,500 (gold line) the US Department of Justice considers a market “highly concentrated”. Puerto Rico crossed that line in 2020 (2,505) and reached 3,514 by 2026 as carriers exited the island.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

Total PR enrollment

How to read it: total Medicare Advantage + stand-alone Part D enrollment in Puerto Rico, latest reported month of each year: 605,623 in 2020 → 684,697 in August 2026, of which 675,423 are Medicare Advantage (the remaining 9,274 are stand-alone PDP enrollees). Demand keeps growing even as payment lags.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

Top contracts' share

How to read it: market share by contract, latest enrollment file: MCS Classicare holds 51.8%; Triple-S 17.0%; MMM’s two contracts 15.0% and 14.0%. Three local organizations cover ~98.7% of the island — national carriers are nearly absent (UnitedHealthcare 0.7%, Aetna 0.1%).

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

PR vs US MA penetration

How to read it: the share of Medicare eligibles enrolled in Medicare Advantage — MA enrollees ÷ Medicare eligibles, the CMS penetration-file denominator: 85.2% in Puerto Rico vs 51.3% on the mainland (August 2026 file; that denominator counts every eligible, including Part A-only). Computed as a share of beneficiaries with both Part A and Part B (Medicare Monthly Enrollment, March of each year — the series charted here), the comparison reads 96.1% PR vs 55.4% mainland (March 2026; May 2026: 96.2% vs 55.5%) — and puts the USVI at 33.4%. Either construction, one conclusion: on the island, MA is not one option among several — it is the healthcare system.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

PR municipio enrollment

How to read it: MA enrollment in the 20 largest municipios: San Juan 63,954, Bayamón 38,926, Ponce 31,374, and so on — the geography of where American citizens living in Puerto Rico actually receive their Medicare.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

County variation within PR

How to read it: the 2026 benchmark rate for all 78 municipios. Even Puerto Rico’s highest-paid municipio (Humacao, $859) sits roughly $390 below the mainland county average ($1,247, green dashed line); the PR average is $730 (gold).

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

Part A vs Part B split

How to read it: measured FFS spending split by program part, 2022–2024: mainland Part A $376 + Part B $616, versus Puerto Rico Part A $206 + Part B $249. PR is far lower on both parts — and PR’s measured Part B actually declined ($292 → $249), another symptom of the shrinking, non-representative FFS base.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

Year-over-year growth

How to read it: annual benchmark growth: Puerto Rico and the mainland grow at similar single-digit rates (6.5% vs 5.3% in 2027). Equal percentage growth on an unequal base means the dollar gap compounds — percent parity is not payment parity.

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC) and Monthly Enrollment by Plan files, March of each year 2020–2026 (latest: March 2026). Computed from the published enrollment counts.

Chronic-condition acuity — what the FFS remnant can still see

Chronic-condition prevalence, Medicare FFS (2023 preliminary) — PR vs national vs USVI

How to read it: CMS's Mapping Medicare Disparities tool, pinned snapshot (fetched July 2026; the endpoint rejects non-browser clients, so values are baked, and any future endpoint change is itself a finding). Diabetes prevalence 46% in Puerto Rico against 26% nationally and 31% USVI; hypertension 67% vs 65%; CKD 24% vs 19% — a sicker population than either comparator, making USVI benchmark parity a conservative ask. Two disclosures printed deliberately: these rates are computed on the FFS remnant (the ~4% sliver the benchmark also relies on), and depression measures lower in PR (12% vs 19%; USVI 4%) — read that as under-detection in a thin-claims environment, because a critic will otherwise read it for you. Independently corroborated in CCW-based peer-reviewed literature.

[17] CMS Mapping Medicare Disparities tool, prevalence datasource 2023-preliminary (cms.gov/MMD) · pinned API snapshot, July 2026 · crosscheck: PMC11692657.

Excellence at the lowest price · CMS 2026 Star Ratings

The 5-star map of America runs through Puerto Rico

CMS rated 516 Medicare Advantage contracts for 2026. It awarded its highest mark — 5 stars overall — to 18. Three are in Puerto Rico: MCS Classicare (H5577) and both MMM contracts (H4003, H4004). And those three contracts do not merely appear on the honor roll — they dominate it: 547,600 of the nation's 820,833 five-star enrollees — 66.7% — are on this island. Puerto Rico holds twice the five-star membership of the other fifteen contracts combined (273,233) — more than every state in the Union put together. Not California. Not Florida. Puerto Rico.

MCS H5577, with 332,942 members, is the single largest 5-star contract in America — operated by a Puerto Rico–based company, serving Puerto Rican beneficiaries. The national standard of excellence in Medicare Advantage is not set in any state. It is set here.

And here is what the program pays for that performance: $730 per member per month — the second-lowest benchmark of all 56 US jurisdictions. The states hosting the other fifteen 5-star contracts collect $1,197 to $1,310 for theirs. The plans CMS itself certifies as the best in America operate on 55–60 cents of the dollar paid to their stateside peers. No payment methodology can certify a market as the best in the nation, pay it the least, and defend both results. Puerto Rico is not asking to be treated better than its performance — it is asking to stop being paid worse than every market it outperforms.

#1 in USA
MCS H5577 · the largest 5-star contract in the nation
332,942 enrollees — 40.6% of all 5-star MA-PD enrollment nationwide, in one Puerto Rico contract. The next largest in the USA (UnitedHealth H3256) has 69,229.
$730 vs $1,246
What 5-star excellence is paid — PR vs US avg
PR's 2026 avg benchmark is $730/mo; the US average is $1,246. The home states of the other fifteen 5-star contracts receive $1,197–$1,310.
3 of 18
Of the nation's five-star contracts are Puerto Rican
MCS H5577, MMM H4003 and MMM H4004 — a single island holds 1 in 6 of the top-rated MA-PD contracts in the entire USA.
66.7%
Of ALL 5-star enrollment NATIONWIDE — the entire USA — is in Puerto Rico
547,600 of the 820,833 enrollees across the nation's 18 five-star MA-PD contracts — every five-star contract in all 50 states and territories — live on the island (enrollment as of Oct 2025).

Stat sources — contract list and enrollment: CMS 2026 Star Ratings Fact Sheet (Nov 18, 2025), Table A1 — all 18 MA-PD contracts rated 5 stars overall, enrollment as of October 2025; shares are direct divisions of those published figures. Benchmarks: CMS CY2026 MA Rate Book, jurisdiction averages computed from published county benchmarks.

How to read the charts: the first chart ranks all 18 five-star contracts by enrollment — red bars are Puerto Rico, green bars are stateside contracts. The island's three contracts are #1, #3 and #4 in the nation. The second chart shows what each 5-star contract's home jurisdiction is paid per member per month in 2026: every stateside 5-star plan operates on a benchmark near or above the $1,246 US average (dashed line) — Puerto Rico's three operate at $730.

RATES -vs- PATIENTS — THE NATION'S FIVE-STAR MA-PD CONTRACTS, 2026 (ALPHABETICAL)

How to read it: green bars (left axis) = each contract's home-jurisdiction average 2026 benchmark in $/member/month — Puerto Rico's contracts in red; gold line (right axis) = enrollment as of October 2025. Contracts in alphabetical order, identified by contract number on the axis — the legend below the chart maps each contract to its organization. The line spikes exactly where the bars collapse: MCS H5577 — 332,942 patients, the largest 5-star contract in America — sits on the lowest bar in the chart ($730), with MMM H4004 and H4003 beside it. Every stateside 5-star contract is paid $1,197–$1,310. The three multi-state contracts (H3256 UnitedHealth, H4172 NHC Advantage, H6988 Anthem HP — 9.0% of national 5★ enrollment combined) operate across multiple states and carry no single home benchmark; their gray bars show the $1,246 US national average as a proxy (marked *), so all 18 contracts appear on the chart

Sources — enrollment: CMS 2026 Star Ratings Fact Sheet (Nov 18, 2025), Table A1, enrollment as of October 2025 — the complete list of all 18 MA-PD contracts rated 5 stars overall. Benchmarks: CMS CY2026 MA Rate Book (2026-ma-rate-book.zip), jurisdiction averages from published county benchmarks.

Who holds the nation's 5-star enrollment?

Source: CMS 2026 Star Ratings Fact Sheet, Table A1 (Oct 2025 enrollment). PR = H5577 + H4004 + H4003 = 547,600; other 15 contracts combined = 273,233.

Dollars per member behind each share

Source: enrollment per Fact Sheet Table A1; benchmarks per CMS CY2026 Rate Book jurisdiction averages. PR bar uses the PR average ($730); stateside bar uses the simple average of the 15 single-state contracts' home benchmarks.

THE COMPLETE 5-STAR ROSTER — ALL 18 MA-PD CONTRACTS, 2026

ContractOrganizationParentState Enrollment (Oct 2025) % of all 5★ enrollment Home 2026 benchmark ($/mo) 5★ last year

Source: CMS 2026 Star Ratings Fact Sheet (Nov 18, 2025), Table A1 — contract, organization, parent, October 2025 enrollment and prior-year 5★ status are the published CMS figures; percentage shares are direct divisions of those figures (total 820,833). Home benchmark = the jurisdiction average from the CMS CY2026 MA Rate Book; "multi-state" contracts carry no single benchmark and are shown as "—". Note: CMS also awarded the high-performing icon to 1 cost contract and 2 PDPs (21 contracts total); this table is the complete MA-PD set. Enrollment vintage differs from the Puerto Rico tab (which uses the August 2026 CPSC file — e.g. MCS H5577 at 356,977); each is cited to its own source.

Why this matters for the benchmark case: CMS's own quality system — the one that decides bonus payments nationwide — certifies that two-thirds of the best-served Medicare Advantage members in America live in Puerto Rico. The island is not asking to be paid a premium for excellence; it is asking not to be paid 59 cents on the dollar for it. If benchmark levels reflected delivered quality, Puerto Rico's would not be the second-lowest in the nation.

Beneficiaries' own verdict — #1 in America, five surveys running

Rating of Health Plan
94.26
#1 of 51 jurisdictions, 2025 · next best RI 89.08 · national 87.29
Rating of Health Care Quality
92.42
#1 · next best IA 89.57 · national 87.19
Rating of Drug Plan
93.88
#1 · national 87.93
Annual flu vaccine — the honesty tile
48.54
lowest in the nation (national 68.94) — disclosed before a critic finds it

Rating of Health Plan — Puerto Rico, five consecutive CAHPS surveys

How to read it: the member-reported Rating of Health Plan from CMS's official CAHPS contractor, all five surveys the contractor publishes (2021–2025). Puerto Rico ranked #1 of 51 reportable jurisdictions in every one, with margins of 2.9–5.2 points over the runner-up. The lowest-paid market in the program produces the best-rated member experience in it. (2021 values are published rounded to whole points.)

Source: CMS MA & PDP CAHPS national/state tables (HSAG, official CMS contractor), 2021–2025 surveys · ranks re-computed from the full 51-jurisdiction columns.

The rating CMS cannot compute — 88% of PR hospitals have no star rating

PR hospitals without a CMS star rating
53 of 60
88% · May 2026 file · vs a 41% all-type national norm
Acute-care hospitals rated (PR)
11%
6 of 54 · vs 86% nationally (2,670 of 3,115)
CMS's own stated reason (43 of the 53)
“Too few measures … to calculate a star rating”
footnote 16 — star measures are built on FFS claims that barely exist in PR

This is the same root cause as the benchmark circularity, visible in a second federal system: quality measurement starves where FFS claims vanish. To be printed with its caveats — unrated ≠ bad quality, and the comparison that matters is the national norm (both USVI hospitals, for the record, are rated — 1★ and 2★).

Source: CMS Hospital General Information (Provider Data Catalog), May 2026 vintage · counts tallied row-level via the datastore SQL API · footnote 16 text per CMS footnote reference.

The honest notes, before a critic writes them: of the 7 PR hospitals that are rated, Auxilio Mutuo holds 4 stars while most rate low; PR’s reported ED median time is 300 minutes vs 161 nationally (OP-18b) and sepsis-bundle compliance 24% vs 64% (SEP-1, July 2024–June 2025) — while left-without-being-seen is better than national (1% vs 2%). And the long-term-care layer barely exists on paper: 9 CMS-certified nursing homes in Puerto Rico, of 14,695 nationwide (June 2026) — consistent with the missing Medicaid nursing-facility benefit documented on the Duals tab. Provenance tier in register [23].

Puerto Rico MA · August 2026 · every MA brand · every municipio

Enrollment by Municipio

All 78 Puerto Rico municipios, broken down by plan. The table below carries every underlying number; beneath it, the statewide enrollment-by-plan summary for the most recent reporting date.

What a municipio is — and why it matters here. Puerto Rico is not divided into counties; it is divided into 78 municipios, and federal data treats each one as a county-equivalent jurisdiction. Every mechanism CMS runs at the county level on the mainland runs at the municipio level in Puerto Rico: each municipio carries its own FIPS code, its own row in the CMS rate book, its own Medicare Advantage benchmark, and its own enrollment reporting. When this console says “county,” in Puerto Rico that means a municipio.

The perspective this creates is central to the benchmark problem. Mainland counties average ~9,400 fee-for-service beneficiaries from which CMS measures local costs; the average Puerto Rico municipio has 376 (2024) — because ~95% of the island is already in Medicare Advantage. Yet the benchmark formula still sets 78 separate municipio-level rates from those vanishingly small samples. Seventy-eight jurisdictions, each priced on a sliver of data no mainland county would ever be priced on — that is the structural flaw documented in the Puerto Rico tab, seen here from the ground level, municipio by municipio.

Most recent reporting date · August 2026

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC), August 2026 file. Every row is the published CMS enrollment count; municipio and brand totals are direct sums of those rows.

Puerto Rico Medicare Advantage enrollment by plan

Statewide Medicare Advantage enrollment by carrier, totaled across all 78 municipios.

Total beneficiaries · every jurisdiction

Puerto Rico Medicare Advantage enrollment by municipio

All 78 municipios — each jurisdiction with its total Medicare Advantage beneficiaries and its share of the island. Click any column header to sort.

# Municipio Total MA beneficiaries % of Puerto Rico
TOTAL · 78 municipios 100.0%

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC), August 2026 file. Each municipio total is the direct sum of that jurisdiction's published CMS plan-level enrollment rows.

The underlying numbers · every row

The underlying numbers

Each carrier's plans are totaled within the municipio, with that brand's share of the municipio's Medicare Advantage enrollment. Filter by municipio or brand. Click any column header to sort.

Quick filter
Municipio Brand Enrolled % of municipio
TOTAL

Source: CMS Medicare Advantage/Part D Contract and Enrollment Data — Monthly Enrollment by Contract/Plan/State/County (CPSC), August 2026 file. Every row is the published CMS enrollment count; municipio and brand totals are direct sums of those rows.

The invisible duals · CY2023 → March 2026

The duals Medicare’s data cannot see

A “dual eligible” is a person enrolled in both Medicare and Medicaid — the poorest, sickest beneficiaries in the program, and the population around which CMS calibrates risk adjustment, D-SNP policy, the Low-Income Subsidy, and quality benchmarks. The mechanism that makes a dual visible to federal data is administrative, not clinical: in the 50 states, a beneficiary receives a federal “dual” flag when the state Medicaid agency reports them to CMS through the Medicare Savings Program categories (QMB, SLMB, QI, QDWI) in its monthly MMA state file — an exchange the statute mandates for “the fifty states and the District of Columbia” only. Puerto Rico is statutorily exempt from operating those programs — §1905(p)(4)(A) of the Social Security Act. No MSP categories, no MMA flags — and therefore no duals, as far as Medicare’s dual-status accounting is concerned. The blindness is precise, not general: CMS’s own T-MSIS-derived Medicaid files count 355,720 Puerto Rico full duals (2022 monthly average). CMS cannot claim the data doesn’t exist — its other ledger has it.

The result is a data artifact of extraordinary scale. CMS’s dual tables show Puerto Rico with 13,255 duals (1.7%) in CY2023 — dead last among the 50 states and DC — and CY2023 was the high-water mark: the Medicare-side count has since collapsed to 7,283 (2024), 6,187 (2025), and 5,581 (March 2026). Against the 307,571 D-SNP enrollees verified in CMS’s own SNP report, the undercount now runs ~55× — and the volatility is itself the tell: a real population does not halve in a year — the console’s own March-snapshot series swings from 5,662 (2017) to 12,283 (2018), peaks at 14,011 (2023) and collapses back to 5,581, a 2.5× peak-to-trough swing with no demographic explanation [12]. The island’s actual dual population, the Platinos, stands at 288,427 on ASES’s own enrollment rolls (Nov 2025; the certified CY2025 MCPAR annual report filed with CMS in May 2026 reports 285,876 — register [23]) and ~302,000 in CMS’s own D-SNP plan-enrollment files (MedPAC, 2023) — and CMS’s own publications now carry the count directly: 307,571 D-SNP enrollees in the agency’s SNP Comprehensive Report (March 2026) and 297,580 duals in its Medicaid managed-care enrollment report (July 2024). The estimate is no longer an estimate. At ≈39.5% of 765,273 beneficiaries, Puerto Rico would rank #1 in the nation — above DC’s 38.8%. The same agency that cannot see these duals in its status tables counts them in its plan files. MedPAC’s Data Book excludes Puerto Rico for precisely this reason, in its own words: “Data exclude Puerto Rico because enrollment data undercount dual-eligibility categories.”

The policy consequence is not cosmetic. Dual status is an input to nearly every payment and quality mechanism in Medicare Advantage: risk-adjustment coefficients, D-SNP frailty and benefit policy, Low-Income Subsidy administration, Star Ratings adjustments for socioeconomic mix, and the research base on which MedPAC, CBO, and CMS itself build territory policy. A jurisdiction whose poorest, sickest 39.5% are recorded as 1.7% (CY2023 — lower still today) enters every one of those calculations looking healthier, wealthier, and cheaper to serve than it is — a misclassification that compounds the benchmark gap documented throughout this dashboard, and that no actor in Puerto Rico can correct, because the missing flag is created by federal statute itself.

Medicare-side dual count
5,581
March 2026 · down from 13,255 (CY2023) → 7,283 (2024) → 6,187 (2025)
Actual dual population
307,571
verified D-SNP enrollees · SNP Comprehensive Report, March 2026
Actual dual share
39.5%
would rank #1 — above DC’s 38.8%
Medicare-side undercount
~55×
5,581 flagged vs 307,571 verified D-SNP enrollees · March 2026

One vintage, no story — the Medicare-side dual count, March of each year (2013–2026)

How to read it: one consistent vintage — the March snapshot of each year from the Medicare Monthly Enrollment file (DUAL_TOT_BENES). The count does not trend, it lurches: 6,049 (2013) → 14,011 (2023) → 5,581 (2026). Against the 307,571 verified D-SNP enrollees, every bar sits between 1.8% and 4.6% of the real population. A count this volatile, of a population this stable, is not measurement — it is noise. (The 13,255 CY2023 figure quoted above is the annual average; this chart holds the vintage fixed.)

Source — CMS Medicare Monthly Enrollment, DUAL_TOT_BENES, March rows 2013–2026 (March 2026 vintage) · refresh: tools/fetch-enrollment.mjs.

Puerto Rico's duals: who actually lives there vs. who the federal data counts

How to read it: ~302,000 dual eligibles live in Puerto Rico (CMS D-SNP plan files, MedPAC 2023). Federal dual tables count 13,255 (CY2023) — a figure that has since fallen to 7,283 (2024), 6,187 (2025), and 5,581 (March 2026). That's a 96% undercount on the CY2023 books alone — 288,745 people the data never sees.

Sources — total: MedPAC, July 2025 Data Book, Sec. 9 (~302,000 PR D-SNP enrollees, 2023). Flagged segment: CMS Program Statistics, MDCR ENROLL AB 42, CY2023 (13,255).

One population, three official counts

These three bars count the same people — Puerto Rico’s dual eligibles — in three different official systems. Bar 1 (red): CMS’s federal dual tables, CY2023 — 13,255, because no MSP flags exist for PR. Bar 2 (blue): 288,427 — ASES’s own Platino enrollment roll, Nov 2025: the island’s Medicaid administration counting its duals directly. Bar 3 (green): ~302,000 — CMS’s D-SNP plan-enrollment files as analyzed by MedPAC, 2023: the same duals, visible to CMS through the plans they join rather than through dual-status flags. The ~22× gap between bar 1 and bars 2–3 is the finding — and it has only widened since CY2023: the Medicare-side count slid to 7,283 (2024), 6,187 (2025), and 5,581 (March 2026), an undercount of ~55× against the 307,571 D-SNP enrollees verified in CMS’s own SNP report. The population is real and counted twice over — just not in the dataset that defines who counts as “dual.”

Sources — green bars: MedPAC, July 2021 Data Book, Sec. 9 (276,000 PR D-SNP enrollees, June 2020) and July 2025 Data Book, Sec. 9 (~302,000, 2023). Red bars: CMS Program Statistics, MDCR ENROLL AB 42, CY2020 and CY2023 (cross-checked against Medicare Monthly Enrollment — exact match).

Dual-eligible share of Medicare beneficiaries — 50 states + DC + Puerto Rico, CY2023

How to read it: each bar is the percent of that jurisdiction’s Medicare beneficiaries that CMS’s federal data flags as dual eligible in CY2023 (duals ÷ total beneficiaries, both from the same CMS dataset). Green bars: the 50 states and DC — ranging from Wyoming (10.4%) to DC (38.8%), national average ~19%. Solid red bar (bottom): Puerto Rico as federal data sees it — 13,255 of 765,273, i.e. 1.7%, last place. (And the CY2023 bar is the peak — the Medicare-side count has since fallen to 7,283 in 2024, 6,187 in 2025, and 5,581 by March 2026.) Striped red bar (top): Puerto Rico as it actually is — ~302,000 D-SNP enrollees (MedPAC, 2023) ÷ 765,273 total beneficiaries (CMS, CY2023) = 39.5%, which would be #1 in the nation. Hover any bar for the exact counts.

Source — state bars: CMS, Medicare Monthly Enrollment (data.cms.gov), state-level annual rows, CY2023: columns DUAL_TOT_BENES ÷ TOT_BENES. Identical figures in CMS Program Statistics — Medicare-Medicaid Dual Enrollment, table MDCR ENROLL AB 42. Striped bar: MedPAC, July 2025 Data Book, Sec. 9 (~302,000 PR D-SNP enrollees, 2023) ÷ CMS total above.

One agency, three ledgers — Puerto Rico's duals in CMS's own publications

How to read it: three CMS publications, one population. Green: 307,571 D-SNP enrollees in CMS's SNP Comprehensive Report (March 2026 — MCS 173,056, MMM 91,043, Triple-S 43,472; D-SNP enrollment requires verified dual status, making this a hard floor). Mid-green: 297,580 duals in CMS's Medicaid managed-care enrollment report (July 1, 2024 — the annual series has stayed within 297,580–384,146 every year since 2016). Red: 5,581 in the dual-status table Medicare payment policy actually consumes (March 2026). A 55× spread, inside one agency: CMS cannot claim the data doesn't exist — two of its own ledgers have it.

Sources — SNP Comprehensive Report, March 2026 · Medicaid Managed Care Enrollment (duals), Table 3 · Medicare Monthly Enrollment, DUAL_TOT_BENES.

Statutory basis & corroboration: MACPAC, Medicaid and CHIP in Puerto Rico (Medicare Savings Programs “are not available in Puerto Rico”; exemption under §1905(p)(4)(A) of the Social Security Act) · KFF (98% of PR duals enrolled in D-SNPs, 2021 — the highest share in the nation) · MedPAC, July 2025 Data Book, Sec. 9, verbatim: “Data exclude Puerto Rico because enrollment data undercount dual-eligibility categories. In 2023, Puerto Rico had about 654,000 Medicare beneficiaries enrolled in MA plans, and about 302,000 were enrolled in dual-eligible special needs plans.”

The Medicaid floor under the duals

For a mainland dual eligible, Medicaid stands under Medicare — it picks up cost-sharing and fills the coverage Medicare leaves. Puerto Rico’s Medicaid cannot play that role the way a state’s does, by statute. Federal funding is block-granted: SSA §1108(g) caps the federal contribution each fiscal year regardless of need (GAO B-333602). The federal match runs at 76% only through FY2027 — SSA §1905(ff), as amended by the Consolidated Appropriations Act, 2023 (CRS R47821) — after which it reverts toward the 55% territorial baseline of §1905(b), while a state at Puerto Rico’s income level would draw roughly the 83% statutory maximum. The benefit package is thinner by design: Puerto Rico does not provide all of Medicaid’s mandatory benefits — no nursing-facility coverage and no non-emergency medical transportation (MACPAC). The program itself — Plan Vital, roughly 1.3 million enrollees (medicaid.gov, January 2026), administered by ASES entirely through managed care — is the floor beneath the ~302,000 duals this tab counts. When the Medicare side underpays and the Medicaid side is capped, there is no third payer to absorb the difference.

[20] SSA §1108(g) / 42 U.S.C. §1308(g) (GAO B-333602) · SSA §1905(b), (ff), CAA 2023 P.L. 117-328 div. FF §5101(b) (CRS R47821) · MACPAC, “Medicaid and CHIP in Puerto Rico” · medicaid.gov Puerto Rico overview · CMS, “Managed Care in Puerto Rico” (ASES). Full verification register [20], Circularity tab.

The methodology trap

The Circularity Trap: how CMS methodology locks Puerto Rico into the lowest benchmarks

CMS uses Puerto Rico's own depressed hospital wages to set the rates that determine those wages. A frozen wage index feeds the MA benchmark formula, mechanically producing the lowest benchmarks in the nation — a self-reinforcing loop, not a market outcome.

Framework: Advantage Business Consulting, Medicare Circularity in Puerto Rico's Healthcare Industry (Feb 2024). Recognized by the U.S. OIG (2020) and the Congressional Research Service (2021).

PR Wage Index (computed)
0.35
FY2027 median of 56 hospitals (range 0.33–0.37) · lowest in the nation
All-Occupation PR/US Ratio
0.56
May 2025 economy-wide benchmark · 60% above the 0.35 computed index
PR Patient-Care Operating Margin
−7.9%
vs −2.9% mainland · FYE-2023 cost reports, Jan-2026 release
PR Physicians Age 60+
47.3%
vs 32.9% US · the retirement cliff

The five-step circularity chain

Step 01
Low wage index & other limitations
CMS computes PR's wage index at ≈0.35 — from the island's own suppressed wages.
Step 02
Medicare & Medicaid critical in PR
61.5% government payers vs 35.8% mainland.
Step 03
Insurers & hospitals treading water
Patient-care operating margin −7.9% vs −2.9% mainland.
Step 04
Pressure on healthcare salaries
$33,792 PR vs $89,002 US hospital staff.
Step 05 · loop
Worker shortage → back to Step 1
47.3% of physicians 60+; the shrinking workforce is re-counted into the next index.

Framework: Advantage Business Consulting (Feb 2024); recognition: OIG 2020, CRS 2021.

01 · The frozen wage index

Computed at ≈0.35 — the lowest in the nation

By CMS’s FY2027 final rule, all 56 Puerto Rico hospitals compute wage indexes between 0.3278 and 0.3655 (median 0.3500) — every one of them in the bottom national quartile. The post-Bridgeport transition pays them ≈0.5259 this year, stepping down 5% annually toward those computed values.

CMS Wage Index — Puerto Rico’s actual FY2027 values

How to read it. Every bar is a number in CMS’s FY2027 final rule tables. The wage index scales hospital pay against the national average of 1.00: Puerto Rico’s hospitals compute a median of 0.3500, are temporarily paid ≈0.5259 under the post-Bridgeport step-down, and the island’s rural floor — the “protection” of last resort — sits at 0.2511.

[1] FY2027 IPPS final rule (CMS-1849-F), Tables 2–3: 56 PR hospitals, computed wage indexes 0.3278–0.3655 (median 0.3500), FY2027 transition values 0.5155–0.5301; rural floor 0.2511, imputed floor 0.3278 · bottom-quartile concentration: OIG 2020.

02 · Healthcare below every sector

Practitioner pay sits near the bottom — the CMS index below everything

If the low PR/US ratio merely reflected a lower-wage economy, healthcare would track other sectors. It doesn't — practitioners (0.48) sit below the 0.56 economy-wide average and below all but two of the 21 published occupation groups, while hospital-staff pay (0.38, 2021 cost reports) and the CMS computed index (0.35) sit below every occupation in the economy — evidence that the CMS-determined rate is the anchor.

PR/US Salary Ratio by Occupation (May 2025)

How to read it. Most occupations cluster near 0.56; healthcare practitioners fall to 0.48 — third from the bottom — and the dashed CMS computed index line at 0.35 sits below every occupation in the economy.

[2] BLS OEWS, May 2025 (retrieved Aug. 30, 2026 via the registered BLS Public Data API; PR Legal group suppressed) · CMS IPPS · CMS Hospital Cost Report, 2021.

03 · Hospital & insurer stress

Losing money on patient care

Low benchmarks push hospitals into the red on patient care: patient-care operating margin −7.9% in Puerto Rico versus −2.9% mainland, and the median hospital tells the same story at −8.0% vs −2.6% (fiscal years ending 2023, CMS Hospital Provider Cost Report, January 2026 release — formula on the tile below; the earlier −10.1%/−3.4% figures came from a prior vintage of the same file and no longer reproduce, so they are retired here). Hospitals lose roughly 8 cents of every patient-revenue dollar and approach breakeven only through non-patient income — the distribution is wide, with Caribbean Medical Center at +11.0% and San Juan CP Hospital at +7.4% while most run negative and distressed systems report far lower (HIMA San Pablo Bayamón −127% in its bankruptcy-era filing). The early fiscal-2024 cohort looks worse still: −13.4% aggregate on the first 20 Puerto Rico reports. Medical-loss ratios run higher across Medicare and Medicaid lines, and individual comprehensive insurance is underwater.

PR Patient-Care Operating Margin (FYE 2023)
−7.9%
mainland: −2.9% · median hospital −8.0% vs −2.6% — Σ(patient revenue − operating expense) / Σ(patient revenue), rows reporting both fields (PR n=38) [3]
PR Individual Comp Loss Ratio
122.7%
claims exceed premiums — underwater
PR Medicare MLR (2022)
88.1%
US: 85.9% — thinner margin
PR Medicaid MLR (2022)
88.7%
US: 86.4% — higher cost pressure

The Medicare-specific view, honestly bounded. The margins above are all-payer. A MedPAC-style Medicare-specific margin requires worksheet-level cost-report data and, in Puerto Rico, would describe almost nobody — with ~96% MA penetration, the hospital ledger sees Medicare mostly as MA plans paying negotiated rates that track the low benchmarks. What the cost reports do show Medicare-specifically, from the same file and vintage:

Medicare FFS Payment per Discharge
$7,547
vs $12,184 mainland (−38%) — DRG + outlier payments ÷ Title XVIII discharges, FYE-2023 reports, IPPS-paid hospitals only (PR n=34, mainland n=1,905) [3]
Medicare Share of Inpatient Days
6.5%
vs 23.6% mainland — the FFS remnant as the hospital ledger sees it; nearly all PR Medicare inpatient care arrives through MA plans [3]
Wage-Index Echo
0.62×
PR's Medicare FFS payment per case is 62% of the mainland's — the ≈0.35 computed wage index and low standardized amounts, visible in actual paid claims per discharge

Hospital Staff Average Annual Salary (2021)

How to read it. PR hospital staff earn $33,792 — about 38% of the US average.

Insurance Market: Government Payers

How to read it. Government payers cover 61.5% of Puerto Rico versus 35.8% on the mainland.

[3] CMS Hospital Cost Report, 2021 · margins & Medicare tiles: FYE 2023, January 2026 release (HCRIS) · [4] KFF & OCI-PR, 2022 · [7] NAIC 2022, Tables 2 & 5.

04 · The widening gap

The salary gap is accelerating

From 2014 to 2025, PR healthcare pay grew 36.4% while the US mainland grew 45.1% (USVI: 30.6%). The PR/US ratio worsened from 0.537 to 0.505 — eleven more years of data, same direction: the gap is not closing.

Healthcare Industry Average Salary Index (2014 = 100)

How to read it. All three start at 100 in 2014; Puerto Rico's line stays below the US line throughout — 2025 average annual pay: $33,584 PR vs $66,520 US.

[5] BLS QCEW, NAICS 62 (private), annual average pay 2014–2025, retrieved via the QCEW open-data API Aug. 28, 2026 — PR · USVI · US.

05 · The physician cliff

Nearly half the workforce is retiring

Low salaries drive physician flight: 47.3% of Puerto Rico's physicians are over 60, with no pipeline to replace them at these wages. In critical specialties the shortage is already acute.

Specialties with >50% of Physicians Over 60 — PR (2020)

How to read it. Bars past the dashed 50% line mark specialties where most physicians are nearing retirement.

Critical Specialty Shortages — Population per Physician, PR (2020)

SpecialtyPop / PhysicianSeverity
Sports Medicine319,369Extreme
Pediatric Critical Care245,669Extreme
Neuroradiology228,121Extreme
Interventional Cardiology212,913Extreme
Thoracic Surgery212,913Extreme
Allergy & Immunology199,606Severe
Vascular & Interventional Radiology187,864Severe
Radiation Oncology168,089Severe
Pain Medicine145,168Severe
Critical Care Medicine138,856Severe
Neurological Surgery122,834Severe

[6] AMA Physician Masterfile, Dec 31 2020 · AAMC State Physician Workforce · population: Census 2020.

Medicare-billing NPIs (PR)
−30%
5,887 (2013) → 4,115 (2024) while the US grew +36% · FFS-visible supply, a lower bound
General Practice billers, 2013 → 2024
1,027 → 381
−63% — the legacy GP corps, gone from the FFS ledger (all-clinician decline: −30%) · verified live from the by-Provider API [15]
Graduated med school 35+ years ago
27.7%
vs 12.2% in Mississippi — similar population, 2.1× the clinicians (6,593 vs 13,759 unique NPIs)
Median career age
23 yrs
vs 16 in Mississippi · Doctors & Clinicians file, 2026 vintage

The label matters and is kept deliberately: the −30% series counts FFS-visible billers, a lower bound in a 96%-MA market — and that caveat itself feeds the invisibility thesis: the federal file that should measure the island's physician exodus can barely see its physicians at all.

[15] CMS Doctors & Clinicians national file (mj5m-pzi6, grd_yr) · Medicare Physician & Other Practitioners by-Provider files, 2013 & 2024 (distinct-NPI counts, verified exact) · construction: row-weighted histograms, distinct-NPI totals.

The pipeline, from Medicare’s own roster

A second, current-vintage angle on the same cliff — no press figures, no membership rolls. Every practitioner approved to bill or order under Medicare carries an enrollment record whose ID embeds its creation date. Take the January 2026 roster of primary-care practitioners (general practice, family practice, internal medicine) and ask: when did each one enter? Puerto Rico’s median currently-enrolled primary-care physician entered in 2012 — versus 2013–14 in Mississippi and 2016 in Connecticut. Only 27.4% of PR’s roster entered since 2020 (MS 32.9%, CT 38.5%). And the entry curve collapses exactly where the island’s crisis years sit: 351 entries in 2012 falling to 74 by 2016 and staying depressed through 2018 — the debt-crisis and María years — with no comparable trough in either comparator. One more composition fact: 70.7% of PR’s primary-care roster is legacy “general practice” (vs ~2% in MS and CT), an older-trained cohort with no board-certified replacement pipeline behind it.

Share of the current primary-care roster by year of Medicare enrollment — PR vs MS vs CT

How to read it. Each line is a state’s currently-enrolled primary-care practitioners, distributed by the year their Medicare enrollment record was created (share of that state’s roster, so shapes compare across sizes). Puerto Rico’s burgundy line towers on the left (pre-2013 entries) and craters in 2013–2018. This is a survivor-only view — physicians who left are absent from every state’s line equally — and in a 96%-MA market the PR level understates the true workforce; the shape is the signal [21].

[21] Medicare FFS Public Provider Enrollment (PECOS), January 2026 snapshot — provider types 14-01/14-08/14-11; counts via the dataset’s stats API, per-year sums reconciling exactly to cohort totals (PR 4,833 · MS 1,948 · CT 3,357).

06 · The measurement base, live

The denominator is vanishing

Every number in this loop is computed from Puerto Rico’s fee-for-service remnant — and CMS’s own enrollment file shows that base collapsing in real time: 77,372 FFS beneficiaries with Parts A&B in March 2013 (13.1% of the A+B population) down to 27,275 by March 2026 — 3.9%. The benchmark input for a 705,363-person A+B market is a 27,275-person sample, and it gets thinner every year — by May 2026, the newest CMS monthly file, it is 26,497 of 706,629 (3.75%).

The FFS measurement base — PR beneficiaries with Parts A&B in Original Medicare (March 2013–2026)

How to read it. This is the population whose claims feed the AGA and, through it, the MA benchmark. It has shrunk in nearly every year on record while the A+B population it must price for grew; the labels mark its share of all A+B beneficiaries. The series is a baked snapshot of the Medicare Monthly Enrollment API (March rows, state level) — the one source on this tab that updates monthly — and re-pulls with each data refresh via tools/fetch-enrollment.mjs.
FFS enrollees with Part B (PR)
26.8%
CY2023 · vs 84.6% nationally · cross-checked at 26.6–27.1% via Monthly Enrollment
Of the measurement base
~73%
cannot generate a Part B claim — yet the A+B benchmark is priced from this population
FFS spend per enrollee
$3,670
vs $12,023 US (CY2023) — meaningless without the composition stat to its left

Alone, the spending figure reads as “Puerto Rico is cheap.” Together with the composition stat, it proves the benchmark's input is broken: most of the population the A+B benchmark is computed from mechanically contributes $0 of Part B spending.

[12] CMS Medicare Monthly Enrollment, March 2013–2026 · A_B_ORGNL_MDCR_BENES / A_B_TOT_BENES, state level · March 2026 vintage.

[13] CMS Program Statistics, CY2023 — MDCR SUMMARY AB / ENROLL AB tables · Part B share cross-checked against Monthly Enrollment (A_B_ORGNL_MDCR_BENES vs ORGNL_MDCR_BENES, March and annual-average constructions).

07 · The unit-price rebuttal

Not cheap — unmeasured

CMS's strongest defense of the benchmark is that Puerto Rico is simply inexpensive. Its own claims file says otherwise: for the common office-visit codes, Medicare pays PR physicians within a few percent of the national rate — $57.25 vs $59.32 for a 99213, $83.50 vs $83.75 for a 99214. Low FFS spending is a vanishing-volume artifact, not a price discount: the entire PR Part B FFS market computes to roughly $71M in CY2024, and the state file carries only 1,792 service lines — Rhode Island, with a third the population, reports 2,910; the national state-file median is in the five figures (13,463 distinct lines) while USVI carries just 551. The longitudinal view is starker: Part B FFS payments in PR fell −76% from 2009 to 2024 ($277.4M → $65.5M) while USVI’s grew +93% over the same span (BESS carrier summaries; PR = carriers 09202 + legacy 00973) — provenance tier disclosed in register [23].

Medicare average payment per service — PR vs national (CY2024, office setting)

How to read it. Average Medicare payment per service (Avg_Mdcr_Pymt_Amt, office place of service). The common evaluation-and-management codes pay PR within 0.3–3.5% of national — and the annual wellness visit pays PR above national (+6.4%), shown deliberately: the honest pattern is “parity,” not “discount.” Unit prices are near-national; only the volume the system can see has collapsed.

[14] Medicare Physician & Other Practitioners — by Geography & Service, CY2024 · payment amounts verified to the cent via the live API · Part B totals: BESS carrier files (2009 baseline per the July 2026 research audit).

And thus, we come full circle.

The remaining low-paid healthcare workforce is counted into the next CMS wage index calculation → the computed index stays at ≈0.35 → low benchmarks persist → hospitals cannot raise wages → physicians leave → the cycle repeats.

The link to MA benchmarks. The wage index feeds the MA benchmark formula directly. A computed wage index of ≈0.35 — every one of the island’s 56 hospitals in the bottom national quartile — mechanically produces the lowest MA benchmarks in the nation — Puerto Rico's CY2027 benchmark of $778 PMPM versus the US mainland's $1,313 (59.3%).

This is not a market outcome — it is a methodological artifact. CMS uses hospital wages to set the rates that determine hospital wages. Both the Office of Inspector General (2020) and the Congressional Research Service (2021) have identified this circularity as a problem requiring correction.

Sources & verification register

[1] FY2027 IPPS final rule (CMS-1849-F), Tables 2–3 — PR computed wage indexes 0.3278–0.3655, transition values 0.5155–0.5301, rural floor 0.2511. cms.gov/…/fy-2027-ipps-final-rule-home-page
[2] BLS Occupational Employment & Wage Statistics (OEWS), May 2025 — retrieved Aug. 30, 2026 via the registered BLS Public Data API v2 (44 series resolved: PR statewide and national annual mean wage, all-occupations plus 22 SOC major groups; the PR Legal group is suppressed in the published data and omitted from the chart). Hospital-staff pay retains its separately-labeled 2021 HCRIS vintage. bls.gov/oes/tables.htm
[3] CMS Hospital Provider Cost Report (HCRIS), 2021 & fiscal years ending 2023 (January 2026 release, recomputed live from the dataset API 2026-07). Margins are dollar-weighted aggregates and medians over rows reporting both net patient revenue and net income from patient service (PR n=38, mainland n=3,389); cost-report vintages revise as late reports arrive, so the release date is part of the citation. Medicare tiles: DRG + outlier payments ÷ Title XVIII discharges over paired IPPS rows; Title XVIII share of total inpatient days. A MedPAC-style all-in Medicare margin needs worksheet-level HCRIS and in PR would describe only the FFS remnant — deliberately not printed. data.cms.gov/…/hospital-provider-cost-report
[4] KFF Health Insurance Coverage, 2022 · OCI-PR. kff.org · ocs.pr.gov
[5] BLS QCEW, NAICS 62 (private), annual average pay 2014–2025, QCEW open-data API (data.bls.gov/cew), retrieved Aug. 28, 2026. PR ENU7200050562 · USVI ENU7800050562 · US ENUUS00050562
[6] AMA Physician Masterfile, Dec 31 2020 · AAMC State Physician Workforce. ama-assn.org · aamc.org
[7] NAIC 2022 Health Insurance Industry Analysis Report — the latest edition with Puerto Rico state tables: the 2024 national commentary (checked Aug. 28, 2026) carries no PR-specific rows. content.naic.org
[8] U.S. OIG, Dec 2020 — "CMS Could Improve Its Wage Index Adjustment…" oig.hhs.gov
[9] U.S. OIG, Nov 2018 — "Significant Vulnerabilities in the Hospital Wage Index System." oig.hhs.gov
[10] Congressional Research Service, 2021 — "Medicare Hospital Payments: Adjusting for Variation in Geographic Area Wages." crsreports.congress.gov
[11] Advantage Business Consulting, Medicare Circularity in Puerto Rico's Healthcare Industry, Feb 2024.
[12] CMS Medicare Monthly Enrollment (data.cms.gov), dataset d7fabe1e-d19b-4333-9eff-e80e0643f2fd, March 2013–2026 state-level rows; March 2026 rows as retrieved Aug. 28, 2026 (UTC), re-verified unchanged Sept. 1, 2026; newest published month May 2026 — CMS revises this dataset monthly. data.cms.gov
[13] CMS Program Statistics, CY2023 (MDCR SUMMARY AB / ENROLL AB). data.cms.gov · Part B composition cross-checked via Monthly Enrollment API.
[14] Medicare Physician & Other Practitioners, by Geography & Service (data.cms.gov, 6fea9d79), CY2024. data.cms.gov · 99213/99214/93000/G0439 payments verified to the cent; service-line counts via /data/stats (PR 1,792 · RI 2,910).
[15] Doctors & Clinicians (mj5m-pzi6, 2026 vintage) & Medicare Physician by-Provider files 2013/2024. data.cms.gov · NPI series 5,887→4,115 verified exact; graduation-age figures re-derived (27.7%/12.2%, medians 23/16) — row-weighted construction, thresholds ±2pp.
[16] SSA §1860D-14(a)(3)(F) / 42 U.S.C. §1395w-114(a)(3)(F) (GPO, verbatim) · Part D Prescribers by Geography & Drug API · EAP substitute: SSA §1935(e). data.cms.gov
[17] CMS Mapping Medicare Disparities tool (data-api/v1/mmd-tool, datasource prev_final_long_fltr12_racecat_all_sexcat_all_23_p), pinned July 2026. cms.gov · FFS-remnant basis disclosed; depression under-detection framing per plan.
[18] CY2024 Plan Payment public use files (2024PartCCountyLevel.xlsx, 2024PartDPlans.xlsx), parsed from the published zip July 2026. cms.gov · Unweighted means across county × plan-type rows (no enrollment weights published); “US” = 50 states + DC, territories excluded — construction reproduces the file exactly (PR $466.13 / VI $772.85 / US $902.68; rebates $257.34 vs $177.30; Part D LICS $8.00 vs $69.48, 70 of 85 PR rows $0.00). Rebate = bid-vs-benchmark savings returned as benefits, not drug rebates. MedPAC’s ~$258 national rebate figure is a different, enrollment-weighted bid-data measure — not comparable.
[19] MA RADV final rule, 88 FR 6643 (CMS-4185-F2, Feb 1, 2023): extrapolation from PY2018, no FFS adjuster, PY2011–2017 enrollee-level only. federalregister.gov · CMS audit expansion, May 21, 2025 (all eligible contracts annually). cms.gov · Vacated on APA grounds Sept 25, 2025, Humana v. Becerra (N.D. Tex.); DOJ appeal to the Fifth Circuit pending — audits and enrollee-level recoveries continue, extrapolation paused. · HHS-OIG A-04-20-07090 (Aug 2024), MMM Healthcare H4003: 108/688 sampled HCCs unsupported, $165,312 net sampled refund. oig.hhs.gov · MedPAC coding intensity ~16% (March 2025, ch. 11); the March 2026 revision under v28 is cited directionally only — the exact revised figure awaits primary verification and is deliberately not printed.
[20] PR Medicaid financing: §1108(g) annual cap (GAO B-333602); FMAP 55% baseline (§1905(b)), 76% through FY2027 (§1905(ff), CAA 2023 §5101(b); CRS R47821); cap dollar levels are quoted only where attributed to KFF’s reading of 42 U.S.C. §1308(g)(6) — statutory text not independently fetched. MACPAC: not all 17 mandatory benefits (no nursing facility, no NEMT). macpac.gov · Plan Vital ~1.3M enrollees, medicaid.gov PR overview, Jan 2026 (figure surfaced via search index; page fetch blocked from this environment — vintage labeled). · ASES role per CMS, Managed Care in Puerto Rico.
[21] Medicare FFS Public Provider Enrollment (PECOS), January 2026 snapshot (dataset 2457ea29). data.cms.gov · Currently-enrolled practitioners, PECOS types 14-01 (general practice), 14-08 (family practice), 14-11 (internal medicine), bucketed by the enrollment-record date embedded in ENRLMT_ID; counted server-side via the stats API, per-year sums reconciling exactly to filtered totals (PR 4,833 · MS 1,948 · CT 3,357). Disclosed limits: survivor-only (departures invisible in every state equally); record-date, not license date; enrollment supports FFS billing and ordering/referring, so PR’s level understates a 96%-MA market — the cohort shape is the claim, not the head-count. Chart shows 2005–2025; 2003–04 entries and the partial 2026 month are in totals and medians but not charted.
[22] Territories tab — now fully live-verified (2026-08) against the published CMS files. Rate machinery: Risk2027.csv (USPCC $1,297.74 verbatim; PR 78 municipios avg $777.98 to the cent; AGA 49.9 PR / 68.0 USVI / 88.0 Guam; quartile 1.15 everywhere; caps binding 22/22 Guam villages, 1/2 USVI areas (St. Croix), 0/78 PR; San Juan $824.42 actual vs $1,178.82 pre-ACA cap). AGA series from risk2021/2023/2025 files: PR 52.2→49.9 flat-declining, USVI 72.4→68.0 falling, Guam 63.7→88.0 rising (CY2018 file no longer published at the archive URLs; series charted from CY2021). Risk scores: Non-PACE Risk Scores.csv 5-year county means 2020–2024 — PR 0.952, USVI 0.722, Guam 0.750; raw-cost indices computed as AGA × risk (PR ≈47.5, USVI 49.1, Guam 65.9). The 0.70 model re-run live: uncapped $1,044.68, capped $996.53, 37/78 municipios trimmed; construction note — the simplified formula (index × USPCC × 1.15, capped) yields $744 at current indices vs the published $777.98 (the published rate embeds GME/IME/KAC/VA-DOD factors), so the modeled $997 is a same-construction counterfactual, conservative vs the full formula. Enrollment/penetration/FFS-base series: Medicare Monthly Enrollment March rows (PR from this console’s live series [12]; USVI/Guam pulled and checksummed this session). Twelve Rate Announcement quotes: page-anchored links to the published PDFs; four precedent quotes pin-cited. cms.gov Risk2027 · ratebooks archive
[23] Content-parity port from the companion console, two tiers. Verified live by this console (2026-08): GP-biller counts 1,027 (2013) / 381 (2024) and all-clinician 5,887/4,115 via the by-Provider dataset stats API (filters Rndrng_Prvdr_State_Abrvtn=PR, Rndrng_Prvdr_Type=General Practice); national LIS receiving share 89.9% (5,617 of 6,250 plan rows) and $77.32 mean among receiving, from the CY2024 Part D plan file; duals volatility from this console’s own March-snapshot series [12]; rebate runner-up and risk rank from this console’s plan-payment state table [18]. Companion-verified, live re-check pending: BESS Part B carrier collapse (−76%, $277.4M → $65.5M, 2009→2024; USVI +93%; 13,463 national / 551 USVI service lines), hospital timely-care measures (OP-18b 300 vs 161 min; SEP-1 24% vs 64%; LWBS 1% vs 2%; Auxilio Mutuo 4★), 9 CMS-certified nursing homes (June 2026), and ASES MCPAR CY2025 certified count 285,876. Not ported pending recomputation on the current HCRIS vintage: aggregate total margin and negative-hospital counts.
All figures verified against primary sources. Wage index figures are taken directly from the FY2027 IPPS final rule (CMS-1849-F), Tables 2–3: 56 Puerto Rico hospitals with computed wage indexes 0.3278–0.3655 (median 0.3500), FY2027 post-Bridgeport transition values 0.5155–0.5301 (median 0.5259), statewide rural floor 0.2511, imputed floor 0.3278. Derived figures are arithmetic from the cited primary data; full verification register available on request.

A tale of two islands, one wage index

From Nantucket to Vieques

How one tiny island can reshape Medicare’s formula across the nation. Two American islands sit outside every metropolitan area. One of them turned that fact into hundreds of millions of dollars a year. The other one is Vieques. Every figure below carries its source; the irony is supplied free of charge.

Medicare’s hospital wage index has a rule, since 1997, that no urban hospital may be paid below its state’s rural hospitals — the “rural floor.” It was written to protect rural America. What follows is what two islands did with it, and what it did to them.

The island that played

In 2008, the only hospital on Nantucket, Massachusetts — newly merged into the Partners health system — converted from a Critical Access Hospital (whose wages don’t count) into an ordinary Medicare hospital (whose wages do). It thereby became the only rural hospital in Massachusetts, and its island wages became, by law, the wage floor for every urban hospital in the state. By 2022 that floor stood at 1.28 — built, in MedPAC’s words, “solely” on this one hospital’s data — raising nearly every hospital in Massachusetts, “in some instances over 35 percent.” By FY2027 the floor stands at 1.3071 — high enough to bind even Boston, whose final wage index is exactly 1.3071 — and the island payroll underneath it averages $78.99 an hour, 5.2 times the $15.17 behind rural Puerto Rico’s floor. The annual transfer to the state’s other hospitals was estimated at $256–367 million per year, paid by every other hospital in the country. Twenty-one states organized against it. The courts reviewed it. It stands. Sources: MedPAC, Report to Congress, June 2023, ch. 9; HHS-OIG audit A-01-15-00502; Boston Globe reporting, 2013; FY2027 IPPS final rule, Tables 2–3 (CCN 220177).

The formula that lifts every hospital in Massachusetts on the payroll of one small island is the same formula waiting for Puerto Rico. The day Vieques opens a certified hospital, the island’s real wages go to work for every hospital in Puerto Rico — automatically, legally, with no new law and nobody’s permission.

The proof of how much force that formula carries is on the federal record. When federal auditors examined the Nantucket hospital’s wage data (HHS-OIG, A-01-15-00502, March 2017), they found reporting errors of about 7.3% had moved an estimated $133.6 million to 55 other Massachusetts hospitals in a single year — nothing illegal, nothing recovered (“CMS has no mechanism to recover overpayments or remedy underpayments resulting from inaccurate wage data”), the bill quietly absorbed by hospitals in the rest of the country under ACA §3141. That is the scale of what one island’s payroll commands under this statute. Massachusetts has been collecting on it for fifteen years. Puerto Rico’s claim to the very same machinery is one certified hospital away — and this time the numbers would be real, working for the island that has been paying into everyone else’s floor.

Two floors, one statute

How to read it: the hospital wage index each island’s data produces, FY2027. Nantucket’s payroll sets the Massachusetts rural floor at 1.3071 — high enough that Boston’s own final wage index is exactly the floor. Puerto Rico’s best protection, the imputed statewide floor, is 0.3278; its rural index — computed from its own depressed wages — is 0.2511. Same sentence of the statute, a five-fold difference.

Source: FY2027 IPPS final rule (CMS-1849-F), Tables 2–3, FY2027 Final Rule home page (Table 2: CCN 220177 wage index 1.3071; Table 3: CBSA 22 rural MA 1.3071, rural PR 0.2511; Boston CBSA 14454); FY2027 Imputed State Floors PUF (same page); rural-PR methodology at 72 FR 47323, Federal Register, Aug. 22, 2007 (govinfo).

The payrolls underneath the floors

How to read it: average hourly hospital wage from CMS’s own FY2027 wage table. Nantucket Cottage Hospital pays $78.99 — 5.2× the $15.17 behind rural Puerto Rico’s floor. The island’s four highest payrolls — all public Centro Médico institutions — are proof of what the island’s real payrolls compute to: $24–$32/hour translates to a wage index of ≈0.42–0.50, roughly double the 0.2511 floor built on $15.17.

Source: FY2027 IPPS final rule, Table 2 (FY2027 Final Rule home page), Average Hourly Wage column, CCNs 220177 ($78.99), 400127 ($31.77), 400124 ($26.10), 400135 ($24.03), 400061 ($24.02); PR median computed across the 50 PR providers with FY2027 wage data; rural-PR basis per Table 3 and 72 FR 47323, govinfo.

The island that couldn’t

Vieques, Puerto Rico is also rural under Medicare’s definition — it sits in no metropolitan or micropolitan area at all. It has had no hospital since Hurricane María destroyed its health center in 2017. In January 2020, a 13-year-old girl, Jaideliz Moreno Ventura, died there of a flu-like illness amid a failed transfer off-island; FEMA obligated reconstruction money roughly two weeks later. The replacement — an $85 million emergency and ambulatory center — was delivered to the Department of Health in December 2025 and, as of mid-2026, awaits its formal opening. It is licensed as a diagnostic and treatment center, not a hospital; its wage data counts toward nothing. Sources: NBC News, Jan. 2021; San Juan Daily Star, Dec. 2025; El Vocero, Jul. 2026. Links verified Sept. 1, 2026.

Apples to apples: the two economies

How to read it: the same Census survey, run on both islands (ACS/PRCS 2019–2023). Note the axis is logarithmic — the gaps are too large for a straight scale: median household income 6.8× apart, per-capita income 7.8×, median home value 12.3×.

Source: U.S. Census Bureau, ACS & PRCS 2019–2023 5-year estimates for Nantucket County, MA and Vieques Municipio, PR: B19013 (median household income), B19301 (per-capita income), B25077 (median home value). Ratios computed from published values.

The safety net, inverted

How to read it: Vieques’ poverty rate is 19.8× Nantucket’s — yet SSI receipt runs the other way, because residents of Puerto Rico are excluded from SSI by statute (U.S. v. Vaello Madero, 596 U.S. 159 (2022)). The poorest island gets the smaller check because of where it sits on the map.

Source: ACS & PRCS 2019–2023 5-year, counties 25019 / 72147: C17002 (income-to-poverty ratio), B22001 (SNAP receipt; in PR the survey captures NAP), B19056 (SSI income), B28002 (internet access). SSI exclusion upheld in U.S. v. Vaello Madero, 596 U.S. 159 (2022) (slip op.). Rates computed from published counts.

A floor made of our own floor

Puerto Rico does have a rural floor — the statute requires one. In years with no rural hospital wage data, CMS computes it as the average of Puerto Rico’s own urban areas (FY2008 final rule, 72 FR 47323). Read that again: the island’s wage floor is the average of the island’s wage indexes. A floor made of the thing it is supposed to hold up. For FY2027, the rural Puerto Rico wage index stands at 0.2511 — below every urban area on the island, and roughly one-fifth of Nantucket’s handiwork. Source: FY2027 IPPS final rule (91 FR 49570), Table 3; Addendum, rural floor methodology.

We pay for the privilege

The rural floor is budget-neutral nationally: every hospital’s wage index is shaved by a uniform factor to fund the floors the rule produces — 0.973976 in FY2026, a 2.6% cut. Puerto Rico’s hospitals, at wage indexes of 0.33–0.37, absorb that cut in full. Which is to say: the lowest-paid hospitals in the United States help finance the Massachusetts rural floor every year. No legislature has organized about this. Source: FY2026 IPPS final rule (90 FR 36536), Addendum II.A.4.

Same Medicare card, different check

How to read it: what CMS offers a Medicare Advantage plan per enrollee per month, CY2026 ratebook, same entitlement, same year: $1,423.96 on Nantucket vs $840.23 on Vieques — 69% more for the rich island’s enrollee. The dialysis (ESRD) rate gap is 63%.

Source: CY2026 MA Rate Book page — direct file: 2026-ma-rate-book.zip, CountyRate2026.csv, rows for SSA county codes 22120 (MASSACHUSETTS, NANTUCKET) and 40740 (PUERTO RICO, VIEQUES), Parts A&B 5%-bonus and ESRD columns.

The all-urban island with twelve rural municipios

Since FY2025, twelve Puerto Rico municipios are rural for Medicare — including Vieques and Culebra. Puerto Rico nonetheless qualifies for the federal “all-urban state” floor (0.3278 in FY2027), because the island’s two hospitals located in rural municipios are both legally deemed urban through reclassification, leaving no hospital “classified as rural.” The island is all-urban because its rural hospitals have paperwork saying otherwise. This is currently Puerto Rico’s best wage-index protection. Sources: 42 CFR 412.64(h)(5)(ii); FY2027 Imputed State Floors PUF; FY2027 Tables 2–3.

The politest sentence in the Federal Register

The temporary policy that had roughly doubled Puerto Rico’s wage indexes was struck down by the D.C. Circuit in 2024. The replacement transition steps the indexes down 5% a year — and absent action the step-down does not end on October 1, 2027: each year’s wage-index decrease stays capped at 5% (42 CFR §412.64(h)(7)), so inpatient payments grind down roughly 2% every October for about eight more years until they settle ≈15.5% below the protected level. Asked to extend the transition, CMS answered: “we may consider this in future rulemaking.” On the rule where that consideration will happen, 982 public comments were filed. Puerto Rico institutions filed four. Sources: Bridgeport Hosp. v. Becerra, 108 F.4th 882 (D.C. Cir. 2024); 91 FR 49570; docket CMS-2026-1256; 42 CFR §412.64(h)(7).

The punchline is optional

Here is the part that isn’t cynical. The Nantucket rules are still the rules — the courts have forced CMS to honor rural wage data in the floor (Citrus HMA v. Becerra, D.D.C. 2022) — and Puerto Rico already holds both keys. First, Vieques needs no law and no petition. The municipio lies outside every metropolitan and micropolitan area, so the moment the Susana Centeno facility is licensed and Medicare-certified as a hospital it is a rural hospital automatically, by geography — the same sentence of the statute that made Nantucket’s hospital rural (SSA §1886(d)(2)(D), 42 U.S.C. §1395ww). Second, the strategy is Vieques — deliberately. Federal regulation does give urban hospitals doors into rural status (42 CFR §412.103(a)), and by CMS’s own FY2027 wage table the island’s four highest payrolls — ASEM ($31.77/hour), the Cardiovascular Center, the Comprehensive Cancer Center, and Hospital Universitario — are public institutions whose wages compute to ≈0.42–0.50. But this console does not lead with that route, for two reasons. The optics: a hospital in metropolitan San Juan arguing it is “rural” hands critics the very manipulation story this page documents. And the math: the protected wage level the island is stepping down from (≈0.5259) corresponds to $31.78/hour — almost exactly ASEM’s payroll — so a reclassified floor of 0.42–0.50 mostly duplicates protection the island still has while the 5%-a-year step-down plays out. Vieques needs zero petitions and one licensure decision: the cheaper strategy on every axis, and the honest one — real rural wages from a genuinely rural island. Source: FY2027 Table 2, Average Hourly Wage column.

What it takes — the checklist

Path A — the Vieques hospital (rural automatically; its own wage data counts ≈4 years after certification, the same lag Nantucket rode 2008→2011):

  1. License it as a hospital. The Susana Centeno facility is licensed today as a diagnostic and treatment center (CDT); the Departamento de Salud must license it as a hospital with inpatient beds. This is the only government act required — a licensure decision, not legislation.
  2. Enroll in Medicare as a hospital (Form CMS-855A through the Part A MAC, First Coast Service Options for Puerto Rico). CMS provider enrollment.
  3. Pass the certification survey against the hospital Conditions of Participation (42 CFR Part 482), via the State survey agency or a CMS-approved accreditor, and receive a CCN. Certify as an IPPS hospital, not a Critical Access Hospital — CAH wages are excluded from the wage index; converting out of CAH status is precisely how Nantucket started the clock in 2008.
  4. From the day the CCN issues, it is a rural IPPS hospital by geography — Vieques sits outside every CBSA — with no petition, law, or reclassification.
  5. File its first Medicare cost report (Form CMS-2552-10); its wages enter the wage index ≈four fiscal years later — the FY2031 option.

What to pay — the target is arithmetic, not aspiration. A wage index is the area’s occupational-mix-adjusted average hourly wage divided by the national average — $60.43 in FY2027, derived two independent ways from CMS’s own Table 3 ($78.99 ÷ 1.3071 = $15.17 ÷ 0.2511 = $60.43). That $60.43 is the effective divisor implied by the published indexes: the raw national average hourly wage in the FY2027 final rule is $58.83, and published indexes carry a budget-neutrality adjustment of ≈0.9736 ($58.83 ÷ 0.9736 = $60.43; Nantucket’s provider-specific ratio $78.99/$58.83 = 1.3426 becomes the published 1.3071 the same way). Both of Puerto Rico’s areas sit exactly at this pure ratio — across all 511 CBSAs in Table 3, the implied divisor tops out at 60.43–60.44, and PR is at the top — so dividing a target payroll by $60.43 converts it directly into a published index. So the Vieques hospital’s payroll converts directly: $25.40/hour computes a wage index of ≈0.42; $30.25 ≈ 0.50; $36.25 ≈ 0.60. The right target is backed into from the payment level Puerto Rico wants — the geographic adjustment that flows through hospital payments and, about a calendar year later via the AGA repricing, into the MA benchmarks (FY2031 wage index → CY2032 benchmarks) — not picked from a wage survey.

The targeting chain — from benchmark to AGA to payroll

How to read it: the AGA (average geographic adjustment) is where the wage index lands in the ratebook — CMS reprices five years of county claims at current payment rates, the county’s relative cost becomes its AGA, and the benchmark is USPCC × AGA × county factors × quartile. That formula, printed in CMS’s own rate file, reproduces Vieques’s published $840.23 and Nantucket’s $1,423.96 to the cent. Nantucket’s AGA is 1.23286; Vieques’s is 0.60441; the five lowest AGA factors in the entire nation are all Puerto Rico municipios. Backing in: every +0.01 of AGA is ≈$14.38 per member per month on Vieques’s bonus benchmark, and an AGA of ≈1.01 — essentially the national average, which is 1.0 by construction — would put Vieques at Nantucket’s benchmark. Pick the payment target first; the AGA, the wage index, and the payroll number fall out in that order. And Washington has already picked a number on this chain: two 2026 congressional letters ask CMS for a minimum AGA of 0.70 for Puerto Rico — the level the U.S. Virgin Islands already lives at (St. Croix 0.674, St. John 0.706 in the same CY2026 file).

Source: CY2026 Rate Calculation Data (cms.gov), risk2026.csv: 2026 FFS USPCC $1,230.52; formula “FFS Rate = FFS USPCC × AGA × (1 − GME factor) × (VA-DOD Adjustment Factor) × Credibility Factor − KAC − IME”; rows 40740 (Vieques: AGA 0.60441, credibility 0.87099, VA-DOD 1.1264, quartile+bonus 120%) and 22120 (Nantucket: AGA 1.23286). Computed rates match the published CY2026 ratebook exactly; sensitivity and parity AGA computed by inverting the same formula. Congressional 0.70-AGA asks: Rep. Hernández + 17 Members to HHS/CMS, Feb. 25, 2026 (minimum AGA 0.70 under SSA §1876(a)(4)); Gov. González-Colón / Rep. Murphy letter, Aug. 5, 2026 (minimum benchmark = 70% of national AGA beginning 2028).

Path B — urban reclassification under §412.103: considered, and set aside. The doors exist — rural referral center (§412.96), sole community hospital (§412.92), state designation — and the four Centro Médico payrolls ($24.02–$31.77/hour) qualify arithmetically. This campaign does not lead with them, for the two reasons above: the optics of a metropolitan San Juan hospital claiming rural status, and the math — a ≈0.42–0.50 floor mostly duplicates the protection the island still holds while the 5%-a-year step-down plays out, so the gain arrives in the out-years, which is exactly when Vieques’s real wage data arrives anyway. One certified hospital on Vieques does the work with no petitions at all.

What the campaign itself needs, then: one licensure decision by Salud for Vieques, an operator and a survey date; comments in the FY2028 rulemaking asking CMS to hold the transition while those wages ripen; and an island that files more than four comments in the rulemaking that decides this. Sources: 42 CFR Part 482; 42 CFR §§412.92, 412.96, 412.103; FY2027 Table 2; the lag from Nantucket’s own record (converted 2008, floor moved Oct. 1, 2011 — MedPAC 2011 letter, June 2023 ch. 9).

Mirror images: how each island holds its Medicare card

How to read it: CMS’s own 2025 enrollment file. Nantucket’s 2,325 beneficiaries are 93% Original Medicare — so the wage-index rules (where its 1.3071 floor lives) pay for nearly everyone. Vieques’ 1,662 beneficiaries are 78% Medicare Advantage — so the $840 county benchmark hits nearly every senior. Each island lives under the rule that binds hardest for it.

Source: CMS Medicare Monthly Enrollment (data.cms.gov), 2025 annual rows, county FIPS 25019 (Nantucket) / 72147 (Vieques): fields TOT_BENES, ORGNL_MDCR_BENES, MA_AND_OTH_BENES. Shares computed from the published counts. Values as retrieved Aug. 28, 2026 (UTC), re-verified unchanged Sept. 1, 2026; CMS revises this dataset monthly.

The exclusions, measured

How to read it: on Nantucket — poverty 3% — 11.7% of drug-plan enrollees receive the Part D Low-Income Subsidy. On Vieques — poverty 59.5% — at most 3.7% do, because the subsidy stops at the water’s edge (SSA §1860D-14). Meanwhile 56.4% of Vieques seniors tell the Census they hold Medicare and Medicaid together, versus 3.0% on Nantucket — 19× the need, none of the federal help that normally follows it.

Source: CMS Medicare Monthly Enrollment (data.cms.gov), 2025 annual, FIPS 25019 / 72147, fields PRSCRPTN_DRUG_TOT_BENES and the LIS columns (cells under 11 suppressed “*”, hence “at most”); LIS territorial exclusion at SSA §1860D-14, 42 U.S.C. §1395w-114; duals share from ACS/PRCS 2019–2023, B27010 (population 65+, Medicare & Medicaid together). Shares computed from published counts. Values as retrieved Aug. 28, 2026 (UTC); CMS revises this dataset monthly.

Nantucket needed one hospital and a merger — nothing else. Vieques needs exactly as much: a certified hospital. The day it opens its doors as one, the statute counts its wages automatically — no law, no waiver, no permission. The difference between the two islands was never the ocean — it was who had a hospital on the field.

The mechanics, the sequencing cautions, and the calendar live one tab over, in FY2028 Watch. This page exists so nobody mistakes the current arrangement for an accident of nature. It is an arrangement. Arrangements can be rearranged.

The territories' perspective · PR · USVI · Guam · one statute, three engines

Three territories, one statute — and only one left unprotected

Puerto Rico, the US Virgin Islands, and Guam all receive Medicare Advantage benchmarks under the same law — §1853 of the Social Security Act — and all three sit in the identical 115% quartile, the statute's maximum multiplier, in every county, every year (verified in CMS's rate-calculation files, 2018–2027). Thirteen years of Rate Announcements (CY2015–CY2027) contain zero USVI- or Guam-specific rate provisions. Yet by CY2027, Guam's average benchmark reached $1,135 (+127% since 2018) and USVI's $967 (+48%), while Puerto Rico's stands at $778 (+61%) — the lowest payment rate in the nation. Hold those three facts together and they force one conclusion — about the outcome, not the mechanics: with the multiplier already maxed for all three territories, the entire gap is manufactured upstream, in the FFS cost estimate the formula multiplies — the one step that belongs entirely to CMS. The machinery protected USVI (a standardization step that lifts its index to 68.0) and Guam (a pre-ACA cap trajectory that sets its rates) automatically, with no law commanding it — and a protection no statute had to command is a protection CMS can extend to Puerto Rico without waiting for one. That is the ask in the Outcome panel below: CY2028 Territory Stabilization at a 0.70 cost index — the same machinery pointed, at last, at Puerto Rico, landing its average benchmark at $997, within ~3% of USVI, with the statute's own caps guaranteeing it can go not one dollar further.

Getting the comparison honest: Guam diverged in growth (+127% vs. the mainland's +54%); USVI diverged in level (24% above PR, from a base that was already ~35% higher); Puerto Rico is the outlier at the bottom on both dimensions — and it is the only one of the three whose benchmark prices care for a large population: 678,245 MA members, versus 6,420 in USVI and 202 in Guam. The formula quietly protected the two markets where Medicare Advantage barely operates — and left unprotected the one place where it is not a program but the healthcare system itself, at 96.2% penetration, the nation's highest. Territory stabilization inverts exactly that: protection where the people are.

Five-star outcomes (2026 Star Ratings), last-place payment (CY2027 benchmarks) — the inversion, in one table

Puerto RicoUS Virgin IslandsGuam
Five-star MA contracts, 2026 Star Ratings3 — the nation's #1, #2 and #3 by enrollment00
Members in those five-star contracts (August 2026)552,146 · 81.7% of PR MA enrollment
Average CY2027 benchmark — what CMS pays, per member per month$778 — lowest in the nation$967$1,135
Paid vs Puerto Rico, per member per month (CY2027)+$189 (+24%)+$357 (+46%)

The quality system and the payment system point in opposite directions. The territory that holds the nation's #1, #2 and #3 five-star contracts — the densest Medicare Advantage market in the country, at 96.2% of A+B beneficiaries — is paid $189 a month less per member than a territory with zero five-star contracts, and $357 less than a territory with 202 enrollees. Five-star outcomes, last-place payment: the star system CMS built to reward quality is inverted by the benchmark system CMS runs beside it — for the territories, the more stars a market earns, the less it is paid. The 0.70 stabilization ends the inversion: it prices the nation's highest-rated market at $997 — still 24% below the mainland average, but no longer below territories where no five-star plan exists.

The bulletproof statement

Under one statute — §1853 — Puerto Rico's market out-performs every jurisdiction in the nation on CMS's own published quality measures (the #1, #2 and #3 five-star contracts in America by enrollment, serving 552,146 people; first in the nation on CAHPS Rating of Health Plan in every published table since 2018) and is paid the nation's lowest benchmark: $778. Under the same statute, the territories with zero five-star contracts — 6,622 MA members between them — are paid $967 and $1,135: 24% and 46% more per member than the best-performing market in the United States. Every number above is CMS's own. A formula that pays zero-five-star markets more than the market holding the nation's top three is not measuring quality or cost — it is broadcasting a broken input. The repair is the outcome: CY2028 Territory Stabilization at a 0.70 cost index — $997, administrative, capped by the statute itself at USVI parity, not one dollar more.

Sources: CMS 2026 Star Ratings Fact Sheet, Table A1 — the complete national five-star MA-PD roster (18 contracts; home states FL, GA, IL, NC, NV, NY, PR, TX plus three stateside multi-state contracts; none based in USVI or Guam) · CMS CY2027 Rate Calculation Data (benchmark averages) · CMS CPSC enrollment, August 2026 (H5577 356,977 + H4004 100,832 + H4003 94,337 = 552,146 of 675,423 MA members). Differences: $967−$778 = $189 (+24%); $1,135−$778 = $357 (+46%).

The picture that decides it: what CMS pays five-star excellence vs zero-star territories — CY2027

How to read it: five bars, one per market, all CY2027, all from CMS's own files. Red — Puerto Rico's three ★★★★★ contracts, the nation's #1, #2 and #3 five-star plans by enrollment (MMM H4003, MMM H4004, MCS H5577 — 552,146 members together): each paid $778 per member per month, the lowest benchmark in the nation. Gray — the US Virgin Islands and Guam, holding zero five-star contracts and 6,622 MA members combined: paid $967 and $1,135. Same statute. Same formula. The best-rated plans in America receive the lowest payment; the territories with no rated excellence receive up to 46% more. This single image is the inversion — and the CY2028 Territory Stabilization at 0.70 (the Outcome panel below) is what corrects it: $997, capped by the statute itself at USVI parity.

Chart sources: CMS 2026 Star Ratings Fact Sheet, Table A1 (star ratings; complete national five-star roster) · CMS CY2027 Rate Calculation Data (benchmarks) · CMS CPSC Monthly Enrollment, July 2026 (members).

Who lives there, who is on Medicare, and who the benchmark actually pays for

Puerto RicoUS Virgin IslandsGuam
Total population3,184,835
July 2025 Census estimate
87,146
2020 Census (latest official count)
153,836
2020 Census (latest official count)
Population 65+789,514 (24.6%)
ACS 2024
18,541 (21.3%)
2020 Census
17,190 (11.2%)
2020 Census
Medicare beneficiaries (Mar 2026)782,793 (~24.6% of pop.)21,133 (~24.2%)20,931 (~13.6%)
— enrolled in MA & other plans678,2456,420202
— in Original Medicare (FFS)104,54814,71320,729
MA penetration — all beneficiaries86.6%30.4%1.0%
MA penetration — A+B basis96.2%33.4%1.4%
FFS measurement base (A+B lives the benchmark is computed from)27,27512,71214,194
CY2027 average benchmark (0% bonus)$778$967$1,135
People that benchmark actually pays for (March 2026, MA & other plans)678,2456,420202

Population: US Census Bureau, Vintage 2025 Population Estimates (PR, July 1, 2025) · 2020 Census of the US Virgin Islands · 2020 Census of Guam — USVI and Guam have no post-2020 official estimates; their Medicare-share percentages use April 2020 denominators and carry vintage uncertainty. PR 65+: ACS 2024 1-year, table S0101. Enrollment: CMS Medicare Monthly Enrollment, March 2026, state rows ("MA & other" = CMS's MA_AND_OTH_BENES, which includes all non-Original-Medicare plan types). Benchmarks: CMS CY2027 Rate Calculation Data, unweighted average of the 0%-bonus rate column.

MA penetration: one territory is the program, two barely touch it

How to read it: the share of A+B beneficiaries enrolled in Medicare Advantage, March 2026. Puerto Rico: 96.2% — the program's highest. USVI: 33.4% (its MA market only appeared in 2021: 350 enrollees in 2018, 5,617 by 2021). Guam: 1.4% — 202 people. The benchmark CMS sets for Guam prices coverage for fewer people than fit in a movie theater; the benchmark it sets for Puerto Rico prices coverage for 678,245.

Source: CMS Medicare Monthly Enrollment, March 2026 (A_B_MA_AND_OTH_BENES ÷ A_B_TOT_BENES).

Benchmark growth since 2018: Guam +127%, PR +61% from the lowest base

How to read it: cumulative growth of each jurisdiction's average benchmark since 2018. Guam exploded (+127%); the mainland grew +54%; USVI +48% (but from a level ~35% above PR's); Puerto Rico +61% — yet PR both started and finished last in dollars: $483 → $778, vs Guam's $499 → $1,135. Two territories that began within $16 of each other in 2018 are now $357 apart.

Source: CMS MA Rate Calculation Data (CY2027 file), CY2018–CY2027 — each year's rate file is indexed on the linked archive page (0%-bonus rates, unweighted averages; same series as the payment chart on the Overview tab).

The twist, proven across all three territories: join MA and the formula turns on you (2018–2027)

How to read it — one graph per territory, everything in percent, one shared scale. The colored area is the share of that territory's beneficiaries enrolled in Medicare Advantage. The black line is the territory's benchmark expressed as a percent of the US mainland average — how the formula values it, relative to everyone else. Puerto Rico: 96.2% of people in MA, and its relative rate is stuck at 59% of the US average — last in the nation. USVI: it joined MA in 2021 (514 → 5,617 members in one year) — and its relative rate is already falling: 76.8% of US in 2018 → 73.7% by 2027, with its measured cost index down 73.1 → 68.0; only the pre-ACA cap softens the slide. Guam: barely 1% in MA, FFS intact — and its relative rate soared from 58.7% to 86.5% of the US average. That is the twist on one scale: where the area fills, the black line sinks; where the area stays empty, the black line climbs. The rate is computed from whoever stays behind in FFS — territory stabilization at 0.70 ends exactly this.

Sources: CMS Medicare Monthly Enrollment — annual state rows 2018–2025 plus March 2026; penetration = A_B_MA_AND_OTH_BENES ÷ A_B_TOT_BENES; MA year-over-year growth computed from the same columns · CMS MA ratebooks and CY2027 Rate Calculation Data — benchmark averages per year and the USVI cost index (73.1 → 68.0).

The engine room: why the same formula produces three different results

Engine A — Risk-score standardization manufactures the PR–USVI gap

The AGA (average geographic adjustment) divides each area's raw per-capita cost by its FFS population's average risk score, to "remove the effects of health status." On raw cost, Puerto Rico (48.4% of national) and the USVI (49.1%) are nearly identical — within a point under any weighting (unweighted across each territory's rate areas). But USVI's and Guam's tiny FFS populations carry risk scores of just 0.71–0.75, so the division inflates their cost indices 33–41%. Puerto Rico's FFS risk score is 0.963 — a ~4% lift. One division step converts two identical raw costs into a 24% payment gap. (CMS would answer that PR's higher score reflects a sicker residual FFS population; the counter — that 0.71–0.75 scores on sparse FFS claims likely reflect thin diagnosis coding rather than genuinely healthier islanders — is an argument, not a settled fact, and is presented as such.)

Proof: CMS CY2027 Rate Calculation Data (Risk2027.csv) — raw per-capita cost, FFS risk score (AVG5SCOR) and AGA columns reproduce 48.4/49.1, 0.963 vs 0.71–0.75, and 49.9 vs 68.0 · division step defined in the Medicare FFS Glossary 2027, formula #10.

Engine B — Guam's own measured costs genuinely tripled, on a growing FFS base

Guam's per-capita FFS cost went $285/month (2016) → $847/month (2024) — real growth in its own claims. Its FFS base grew from ~9,000 to ~13,700 A+B lives, because Guam has essentially no MA market to drain it. In most ratebook years since CY2022, Guam's own FFS rate is the binding value for its benchmark. Guam is also priced as one island-wide rate area — a single pooled calculation — while Puerto Rico is priced municipio-by-municipio from 78 thin claims cells, 69 of them below CMS's own 1,000-enrollee credibility threshold (median municipio: ~200 FFS enrollees; several suppressed entirely at fewer than 11 beneficiaries).

Proof: CMS FFS county file, CY2024 (Guam $847/month; per-year FFS files back to 2016, incl. the $285 value, are indexed on the ratebooks & supporting-data archive) · CY2027 Rate Calculation Data — Guam single rate area, PR 78 municipio cells, enrollee counts behind the 1,000-enrollee credibility line.

Engine C — The pre-ACA ratchet shapes their growth; it never reaches PR

Every county's benchmark is capped at its pre-ACA rate — a series that trends forward at the national growth factor and essentially never falls. That cap binds for all 22 Guam villages (every year since CY2021) and for one of USVI's two rate areas (every year 2018–2027): while capped, benchmark growth tracks the nationally-trended series instead of the area's own local data — that is the source of USVI's 2025→2027 acceleration ($862 → $939 → $967) even as its own cost index was falling. Precision matters: the cap is a ceiling, not a floor — it actually trims Guam and the capped USVI area below their uncapped values. Puerto Rico is subject to the identical rule; it just never binds, because PR's FFS-based benchmark collapsed ~26–27% below its own pre-ACA trajectory island-wide (San Juan: 30%). San Juan's protective cap sits at $1,178.82; its actual benchmark is $824.42.

Proof: CMS CY2027 Rate Calculation Data (Risk2027.csv) — 0%-bonus rate, Pre-ACA rate and "All Bonus Rates Capped" columns (sheet note, verbatim: "All 2027 rates are capped at 2027 Pre-ACA rate") reproduce the 22/22 Guam and 1-of-2 USVI cap binds, San Juan $1,178.82 vs $824.42 · prior years on the ratebooks archive.

The one variable that moved: the AGA cost index

How to read it: the AGA multiplies the national per-capita cost ($1,297.74 in CY2027) to set each area's FFS rate. Guam's index rose 71% (51.4 → 88.0). USVI's fell (73.1 → 68.0) — its benchmark rose anyway, courtesy of Engine C. Puerto Rico's has been flat around 50 for a decade: the collapsed FFS remnant keeps reproducing the same depressed measurement.

Source: CMS Rate Calculation Data (risk20XX.csv; CY2027 file), AGA factor column, CY2018/2021/2023/2025/2027; PR = unweighted municipio average.

The money chart: near-identical raw costs, 24% payment gap

How to read it: on raw measured per-capita cost (CY2027 files), Puerto Rico and the US Virgin Islands are nearly identical — 48.4% vs 49.1% of national (unweighted across each territory's rate areas). The AGA then divides each by its FFS risk score (PR ~0.96, USVI ~0.72). Result: 49.9 vs 68.0. The entire PR–USVI benchmark gap is created in that division.

Source: CMS FFS County data (CY2027 ratebook page incl. FFS Glossary) & Medicare FFS Glossary 2027, formula #10 ("Division by AVG5SCOR serves to remove the effects of the health status and demographics of the beneficiaries in the county"); CY2027 workbook values.

The unused safety net

How to read it: gold = each area's pre-ACA cap, the formula's protective ceiling; red = what the area actually receives (CY2027). For Guam the bars are equal — the cap binds island-wide, and its payment path follows the nationally-trended series (as it does for one of USVI's two rate areas). San Juan's cap sits $354 above its actual benchmark, unused. Every protective mechanism in the formula binds somewhere — except in Puerto Rico: 0 of 78 municipios capped, in any year examined.

Source: CMS Rate Calculation Data, CY2027 (Risk2027.csv): 0%-bonus rate, Pre-ACA rate, and "All Bonus Rates Capped" columns. Sheet note, verbatim: "All 2027 rates are capped at 2027 Pre-ACA rate."

Three measurement bases: one collapsing, one drained by MA entry, one growing

How to read it: the A+B FFS lives each territory's benchmark is computed from. Puerto Rico: 77,556 → 29,143 on 2025 annual data (−62% since 2013; by March 2026: 27,275; by May 2026: 26,497) as MA absorbed the island. USVI: flat ~17,500 until an MA plan entered in 2021, then ~12,600. Guam: growing to 13,723 — no MA market to drain it. By 2025, PR's benchmark — serving 678,245 MA members — rests on a measurement base only ~2.1× the size of Guam's, which serves 202.

Source: CMS Medicare Monthly Enrollment, A_B_ORGNL_MDCR_BENES, annual rows 2013–2025 (fetched July 15, 2026).

Territory stabilization: what the administration has done, and what it can still do

CY2027 rates are final (announced April 2026). The next lever is the CY2028 rate cycle: the Advance Notice arrives in early 2027 with a public comment window, and the final CY2028 Rate Announcement follows by the first Monday of April 2027. The record below separates what CMS has already proven it can do administratively from what it says requires Congress — in its own words.

The outcome · Territory stabilization at a 0.70 cost index · CY2028

The ask, stated precisely — and what it delivers

The outcome sought: in the CY2028 Rate Announcement, CMS adopts a uniform Territory Stabilization methodology: where a jurisdiction's FFS population fails representativeness — measurable triggers this dashboard documents, such as FFS falling below 5% of A+B beneficiaries (PR: 3.8%) or a majority of rate cells sitting under CMS's own 1,000-enrollee credibility line (PR: 69 of 78) — the effective cost index used in rate development shall be developed by similar-area estimation and shall be no less than 0.70 of national. One rule, all three territories: Guam (0.88) is untouched, USVI (0.68) moves marginally, Puerto Rico (0.51) is repaired.

Why it is administratively permissible — because it is drafted as estimation methodology, never as a benchmark floor: (1) §1876(a)(4) authorizes the Secretary to estimate per-capita cost "based upon an adequate sample… in a geographic area served by an eligible organization or in a similar area, with appropriate adjustments to assure actuarial equivalence" — where the local sample is inadequate, the statute itself names the remedy, and USVI (near-identical raw cost) is the natural similar area; (2) the zero-claims adjustment is ten years of precedent for Secretary-directed corrections to PR's FFS experience; (3) CMS's own characterization of the PR methodology (CY2020, reprised verbatim in CY2022 and CY2023) — "our exercise in discretion for the data used to develop the estimate for one geographic area" — is the agency's own statement of this authority. What is not permissible is a floor labeled a floor: CMS has refused that formulation on the record ("limited discretion to… applying floors for specific locales," CY2027), and the D.C. Circuit's vacatur of the IPPS low-wage-index policy (Bridgeport Hosp. v. Becerra, 2024) shows what happens to administratively-invented floors. The formulation is the difference between the outcome and a court loss.

What the outcome delivers — modeled against CMS's own CY2027 rate file (effective index floored at 0.70, every statutory pre-ACA cap honored, computation reproducible from the published Risk2027 file):

MeasureValue
PR average benchmark today (CY2027)$778
PR average benchmark under 0.70 stabilization, uncapped$1,045
Municipios where the statutory pre-ACA cap binds and trims the fix37 of 78
PR average benchmark under 0.70 stabilization, after statutory caps$997 (+28%)
USVI average benchmark (the parity target)$967 (within ~3%)
US mainland average (still far above the stabilized PR)$1,313 (PR remains 24% below)

The statute polices its own outcome. Because every benchmark remains capped at the pre-ACA trajectory — the same ceiling that already sets Guam's and half of USVI's rates — the stabilization cannot overshoot: it lands Puerto Rico at USVI parity and stops. The mechanism that protects the other territories becomes the guarantee that protecting Puerto Rico is fiscally bounded. The statute, verbatim — SSA §1853(n)(4), 42 U.S.C. §1395w-23(n)(4), "Cap on benchmark amount": “In no case shall the blended benchmark amount for an area for a year (determined taking into account subsection (o)) be greater than the applicable amount that would (but for the application of this subsection) be determined under subsection (k)(1) for the area for the year.” The “applicable amount” is the pre-ACA rate series — and because the cap applies after quality bonuses (“taking into account subsection (o)”), nothing can pierce it. CMS implements it in Risk2027.csv (Pre-ACA rate and “All Bonus Rates Capped” columns; sheet note, verbatim: “All 2027 rates are capped at 2027 Pre-ACA rate”), and this dashboard's model honors it: the caps bind in 37 of 78 municipios and trim the uncapped $1,045 average to $997.

Who does what, by when: comments making this ask, with this record, in the CY2028 Advance Notice window (~late January – early March 2027) · the §1853(d)(4) island-wide payment-area request filed by the Governor by February 1, 2027 as the structural companion · plans and the congressional delegation carrying the same single formulation. CY2027 is final; CY2028 is the actionable year — and every element above is within CMS's existing administrative authority.

Modeled figures are this dashboard's computation from CMS CY2027 Rate Calculation Data (FFS rate, pre-ACA cap, and 0%-bonus columns; USPCC $1,297.74; quartile 1.15): effective index = max(current, 0.70), benchmark = min(index × USPCC × 1.15, pre-ACA cap), averaged unweighted across 78 municipios — re-run live on the published file by this console: uncapped $1,044.68, capped $996.53, 37 of 78 municipios trimmed by the statutory cap. Construction note, disclosed: the simplified formula (index × USPCC × 1.15, capped) yields $744 at current indices versus the published $777.98 — the published rate embeds additional factors (GME, IME, KAC, VA-DOD) — so the modeled $997 is the counterfactual under the same simplified construction, conservative relative to the full formula. Statutory authorities quoted above: 42 U.S.C. §1395mm(a)(4); CY2020 Rate Announcement p. 24 (reprised CY2022 p. 41, CY2023 p. 33); CY2027 Rate Announcement pp. 46–47; Bridgeport Hosp. v. Becerra, 108 F.4th 882 (D.C. Cir. 2024).

The record · Twelve cycles of the ask · CY2016–CY2027

“The Ask,” year by year — in CMS's own words

Every quote below is from the final Rate Announcement for the year shown — each year is linked to the published PDF and opens on the first cited page (all twelve documents verified in full; all twelve links tested live). The request this dashboard documents has been on the administrative record continuously: a minimum benchmark since CY2016, an explicit “AGA floor” since CY2019, the specific 0.70 figure since CY2020 — restated by CMS every year, adopted in none, and never once answered on the terms of §1876(a)(4)'s similar-area clause.

CY (pages)The request, as CMS itself restated itCMS's response, verbatimGrounds
2016 (pp. 22–23)“establish a minimum benchmark level for the lowest cost counties in Puerto Rico… at 85% or 90% of the lowest average MA benchmarks among states”“we do not believe the approach suggested by these comments would be permissible under statute”Statutory only
2017 (pp. 22, 25)“establish a minimum benchmark level for counties in Puerto Rico”; “use another jurisdiction as a proxy”“the FFS data in Puerto Rico is sufficient for establishing accurate MA benchmarks as well as consistent with the statutory requirements”Statutory + data-sufficiency
2018 (pp. 29–31)no rebasing reductions; ESRD: “using the US Virgin Islands rate as a proxy”“We believe that the FFS data in Puerto Rico is sufficient for establishing accurate MA benchmarks”Data-sufficiency
2019 (pp. 25–26)“establishing an AGA floor, using a proxy benchmark, and/or applying a hold harmless minimum benchmark” — the first explicit “AGA floor”“We will continue to analyze these issues and consider whether any refinements to the methodology may be warranted in future years”Pure deferral
2020 (pp. 29, 33)the first explicit 0.70: “apply an AGA of 0.70 or apply a nationwide average AGA”; ESRD: “use the rates of US Virgin Islands”“the law requires that Medicare Advantage benchmarks be based on a county's average Medicare FFS per-capita cost, and there is no evidence that FFS costs in Puerto Rico are higher”The dual formula debuts
2021 (p. 35)“establishing an AGA floor/proxy (e.g., applying an AGA of 0.70 or a nationwide average AGA)” — citing CMS's own USVI GPCI proxy as FFS-side precedentsame dual formulaStatutory + evidentiary
2022 (pp. 43–44)“establishing an AGA minimum/floor (e.g., applying an AGA of 0.70)”; the benchmark is “39.6 percent below the national average”same dual formulaStatutory + evidentiary
2023 (pp. 34–35)“establishing an AGA minimum/floor (e.g., applying an AGA of 0.70) or applying a hold harmless minimum benchmark”same dual formulaStatutory + evidentiary
2024 (pp. 53–55)“establishing a minimum AGA of 0.70 for Puerto Rico” — with the full §1876(a)(4) theory: the statute “expressly permits the Secretary to use data from ‘a geographic area served by an eligible organization or . . . a similar area’”, USVI as the similar areaFFS floor precedents “are based on statutory provisions that are neither applicable to the MA program nor provide a direct analog to the provisions in section 1853” — the similar-area clause itself: unaddressedStatutory, sharpened; clause ignored
2025 (pp. 60–63)“A large number of commenters requested that we consider establishing a minimum AGA of 0.70 for Puerto Rico (similar to the AGA level in the US Virgin Islands)”“CMS has limited discretion to incorporate targeted adjustments or exceptions”; “we have not seen evidence to suggest that Medicare FFS costs in another jurisdiction are a reliable proxy”“Limited discretion” debuts
2026 (pp. 52–53)“establish a minimum AGA of 0.7 either as a national MA benchmark floor or specifically for Puerto Rico”same two formulationsStatutory + evidentiary
2027 (pp. 46–47)“use a proxy rate for Puerto Rico, establish a minimum floor benchmark (e.g., at USVI rates), and establish a minimum floor AGA of 0.7”“limited discretion to incorporate targeted adjustments or exceptions, such as applying floors for specific locales”Sharpest statutory form

What the record shows — three things. (1) The reasoning migrated. Bare statutory refusal (CY2016–17) → “the FFS data in Puerto Rico is sufficient” (CY2017–18) → pure deferral (CY2019) → the settled statute-plus-no-evidence dual formula (CY2020–24) → “limited discretion” (CY2025–27). Across those same years the agency described the identical authority as an affirmative “exercise in discretion” whenever it was granting Puerto Rico's data adjustments (CY2020 p. 24, reprised verbatim CY2022 p. 41 and CY2023 p. 33) — and as “limited discretion” precisely when floors were pressed hardest. (2) The similar-area clause has never been construed. The full §1876(a)(4) theory was put to CMS squarely once — CY2024, USVI as the similar area — and the response answered the fee-for-service floor precedents while never quoting, construing, or even mentioning the clause. Twelve years of responses; zero engagement with the statutory text at the center of the request. (3) The ask went congressional. In February 2026 — inside the CY2027 comment window — Resident Commissioner Pablo José Hernández and 17 other Members of Congress wrote to CMS and HHS pressing the USVI-level minimum, and the Medicare Advantage Integrity Act of 2025 (H.R. 6031, introduced November 13, 2025, with bipartisan sponsorship) would legislate the floor CMS says it lacks discretion to adopt. Every refusal in the table rejects a floor; none forecloses the estimation-methodology path — which is exactly the lane the CY2028 outcome above is drafted in.

All quotes: CMS final Rate Announcements CY2016–CY2027 (latest: CY2027, April 6, 2026), verified against the published PDFs on July 15, 2026. Congressional record: Office of the Resident Commissioner, H.R. 6031 press release (Nov 13, 2025) · El Nuevo Día (Feb 27, 2026) on the 18-Member letter to CMS and HHS.

Proven — CMS has already done these administratively for Puerto Rico

The precedents: four PR-specific actions, all taken without Congress

  • CY2012 — the A+B-only claims base (proposed February 2011; fully implemented CY2014). Because PR beneficiaries must opt in to Part B — unlike the automatic mainland enrollment — CMS restricted PR's FFS rate calculation to beneficiaries with both Part A and Part B, finding their standardized costs "on average 5% higher" (CY2012 Advance Notice, pp. 13–14). PR is the only jurisdiction measured this way; the convention is printed in the FFS data files themselves.
  • CY2017 — the zero-claims adjustment. "The Secretary has directed the Office of the Actuary to adjust the fee-for-service experience for beneficiaries enrolled in Puerto Rico to reflect the propensity of zero dollar claimants nationwide" (CY2017 Rate Announcement, pp. 24–25) — 13.9% of PR's A+B FFS beneficiaries generate zero claims vs 6.1% nationally (CY2027 Rate Announcement), so OACT reweights the cohort to the national proportion. Worth +4.1% to +4.7% every year since (CY2027: +4.4%), a Trust Fund impact CMS put at ~$320M/yr (CY2023). Re-adopted annually — never made permanent.
  • CY2018 — the double-bonus reinterpretation. CMS "reevaluated our interpretation of Section 1853(o)(3)(B)" (CY2018 Rate Announcement, p. 3), making 52 of 78 PR municipios double-bonus counties for star-rated plans — zero exist in USVI or Guam.
  • CY2020–CY2025 — a territory-tailored CMMI model. Puerto Rico participated in the MA VBID model for six years (Triple-S/GuideWell and MCS/MHH among participants), including a territory-specific design accommodation: socioeconomic targeting was defined, "in US territories," by dual-eligible status rather than LIS — because the LIS statute excludes the territories.

These are the proof of concept: when CMS concluded PR's data misrepresented PR's costs, it adjusted the data — administratively, under §1853's existing estimation authority. One warning flag: the CY2026 Rate Announcement added that CMS will "reassess the need for ongoing special adjustments" — the existing concessions are re-adopted one year at a time, and keeping them is itself a standing agenda item for every comment cycle.

All quotes: CMS Advance Notices & Rate Announcements (CY2012 Advance Notice pp. 13–14; CY2017 RA pp. 24–25; CY2018 RA p. 3; CY2023 RA; CY2026 RA), pin-cited above.

Available now — administrative actions for the CY2028 cycle and beyond

What the administration can do without Congress

  • 1 · Fix the standardization artifact (the largest lever). The AGA's division by the FFS risk score converts PR's and USVI's near-identical raw costs (48.4% vs 49.1% of national, unweighted) into a 24% payment gap. CMS can adopt a representativeness or coding-completeness adjustment to the standardization step for jurisdictions whose FFS remnant is a small, non-representative residue of the population being priced — the exact species of correction as the zero-claims adjustment it already made in CY2017. CMS's own words establish the authority — written about Puerto Rico itself, describing the A+B-only methodology: "Our exercise in discretion for the data used to develop the estimate for one geographic area, based on circumstances unique to that area, illustrates how there is more than one way to develop a reasonable and reliable adjusted average per capita cost estimate for purposes of the MA statute" (CY2020 Rate Announcement, p. 24; repeated verbatim in CY2022, p. 41, and CY2023, p. 33).
  • 2 · Consolidate Puerto Rico into a single island-wide payment area — a dormant statutory right, never used by anyone. Under SSA §1853(d)(4) (42 U.S.C. §1395w-23(d)(4); regulation at 42 CFR §422.304(e)), upon written request of a State's chief executive filed by February 1, the Secretary "shall make a geographic adjustment… to a single statewide Medicare+Choice payment area." "State" expressly includes Puerto Rico (SSA §1861(x) → §210(h); 42 CFR §400.200) — the Governor qualifies, and CMS's 1998 rule confirms approval is not discretionary ("Section 1853(d)(3) directs us to approve a Governor's request," 63 FR 35004). CMS itself noted in 2005 that "no State has availed itself of this option since its enactment in 1998" (70 FR 4588, 4656). Guam is already priced island-wide as a single area; Puerto Rico is priced from 78 municipio-level cells, 69 of them below CMS's own 1,000-enrollee credibility line. Honest scope note: the adjustment is budget-neutral within the state — it repairs fragmentation, small-cell volatility, and suppression, not the aggregate level. It is the structural cleanup; item 1 is the money.
  • 3 · Use the "similar area" estimation authority already in the statute. The FFS rate CMS computes flows through §1876(a)(4), which lets the Secretary estimate per-capita cost "in a geographic area served by an eligible organization or in a similar area, with appropriate adjustments to assure actuarial equivalence." Where a 27,000-person remnant cannot support a credible estimate, the statute itself names the alternative CMS says it lacks. And the clause is genuinely unaddressed ground: in twelve years of Rate Announcement responses, CMS has never quoted or construed the similar-area language — including CY2024, the one cycle in which commenters presented it squarely (pp. 53–55); the response answered the fee-for-service floor precedents and left the clause untouched.
  • 4 · Design a successor territory model at CMMI. §1115A directs the Innovation Center toward models that address "a defined population for which there are deficits in care" — Puerto Rico's 300,000+ federally invisible duals are such a population, and the precedent exists: PR participated in the VBID model for six years (CY2020–2025) with a territory-specific accommodation built into its design. VBID was terminated for excess Trust Fund costs, so any successor must be structured to survive §1115A's cost test — but the authority, and the PR precedent, are on the books.
  • 5 · The calendar. The CY2028 Advance Notice is due by early February 2027 (realistically late January, on recent practice: the CY2027 notice landed January 26) with a statutory 30-day comment window — roughly late January through early March 2027; the final CY2028 Rate Announcement lands by Monday, April 5, 2027. That comment window is where items 1 and 3 must be argued; a §1853(d)(4) payment-area request for CY2028 must be filed by February 1, 2027.
Requires Congress — only for permanence · the CY2028 outcome does not wait for it

Read the refusal record correctly: CMS closed the floor door — it never closed the estimation door

What CMS refused — and keeps refusing: a benchmark floor labeled a floor, USVI-parity by fiat, cross-jurisdiction proxy rates. Most recently, CY2027 Rate Announcement (pp. 46–47): “section 1853 of the Act prescribes the general approach that FFS per capita costs be used in developing MA rates and CMS has limited discretion to incorporate targeted adjustments or exceptions, such as applying floors for specific locales… we have not seen evidence to suggest that Medicare FFS costs in another jurisdiction are a reliable proxy.” Every one of those refusals — twelve consecutive cycles of them, tabled above — rejects the floor formulation. Not one of them touches the outcome this dashboard asks for.

What CMS never refused: similar-area estimation under §1876(a)(4). In twelve years of responses CMS has never quoted, construed, or answered that clause — including CY2024, when it was squarely presented (pp. 53–55). And CMS's affirmative description of its own authority — “Our exercise in discretion for the data used to develop the estimate for one geographic area… illustrates how there is more than one way to develop a reasonable and reliable adjusted average per capita cost estimate” (CY2020, p. 24; reprised CY2022 p. 41, CY2023 p. 33) — was written about Puerto Rico's own data adjustments. The tension between “limited discretion” (when floors are pressed) and “exercise in discretion” (when data fixes are granted) is not a contradiction to lament — it is a map. It marks where the authority lives: the estimation inputs. The Outcome panel's 0.70 stabilization is drafted exactly there, and its evidentiary record answers the one thing CMS says it has not seen.

What Congress adds — permanence, and the parts no agency can reach: a statutory territory benchmark floor (already pending: H.R. 6031, the Medicare Advantage Integrity Act of 2025, introduced November 13, 2025); extension of the Medicare Savings Programs and the Part D Low-Income Subsidy to Puerto Rico (the statutory root of the duals invisibility and the $0.00 LIS column documented on this dashboard); and Medicaid financing parity. The sequence is the strategy: CMS stabilizes in CY2028 under authority it already holds; Congress makes it permanent. Neither needs to wait for the other.

Quotes: CMS final Rate Announcements (CY2027 pp. 46–47; CY2024 pp. 53–55; CY2020 p. 24; CY2022 p. 41; CY2023 p. 33) · H.R. 6031: Office of the Resident Commissioner (Nov 13, 2025) · statutes: §1876(a)(4), §1853.

Methodology & primary sources for this tab: CMS Rate Calculation Data and Ratebooks CY2018–CY2027 · CMS FFS county claims files CY2016–CY2024 and the Medicare FFS Glossary 2027 · CMS Advance Notices and Rate Announcements CY2015–CY2027 · CMS Medicare Monthly Enrollment (data.cms.gov) · US Census Bureau: Vintage 2025 estimates, USVI 2020 Island Areas Census, Guam 2020 Island Areas Census, PR ACS 2024 · MedPAC March 2017 Report Ch. 13 · GAO-06-75 (gao.gov; loads in a browser, blocks automated checks). All figures verified against the primary files on July 15, 2026. Analysis and framing are this console's own; every number carries its source above.

Verification on this console: PR enrollment and penetration figures verified live [12]; the Risk2027 engine-room figures (USPCC, AGA, risk scores, caps, the 0.70 model) re-run live on the published files — provenance and verification tiers in register [22], Circularity tab.

More medicine, less money · CY2020–2023

Prescription drug utilization & cost

Puerto Rico's Medicare Advantage members fill ~19% more prescriptions per year than the national average — at ~15% lower gross drug cost per enrollee. Roughly $63 per prescription versus $89 nationally (CY2023). The pattern holds every year since 2020: the dollars convert into medication adherence, not overhead. The basis is unusually clean: 98.2% of Puerto Rico's 656,529 Part D enrollees are in MA-PD plans (vs 56.1% nationally), and 97.8% of them filled at least one prescription in CY2023 (vs 94.8% nationally) — on the island, the drug benefit is not a card in a drawer; it is used, by nearly everyone, every year.

+19.1%
More prescriptions filled
68.4 fills per MA-PD enrollee vs 57.4 nationally, CY2023 — a premium PR has held every year since 2020 (+17% to +21%).
−15.4%
Lower gross drug cost per enrollee
$4,331 per PR MA-PD enrollee vs $5,122 nationally, CY2023 — more medication delivered for fewer dollars.
$63 vs $89
Gross cost per prescription
$63.30 in PR vs $89.16 nationally per fill, CY2023 — PR dispenses at 71 cents on the national dollar.
97.8%
Of enrollees actually use the benefit
642,062 of 656,529 PR Part D enrollees filled ≥1 prescription in CY2023, vs 94.8% nationally.

Stat sources — all four figures computed directly from CMS Program Statistics, Medicare Part D, Table D6 (per-enrollee, MA-PD columns) and Table D7 (utilizers), CY2023: PR 644,605 MA-PD enrollees × 68.43 fills at $4,331.49; US 28,727,096 × 57.44 fills at $5,121.80. Cost per fill = published cost ÷ published fills.

How to read the charts: each group of bars is one year — green is the US national average, red is Puerto Rico; both use the MA-PD columns of the same CMS tables, so the comparison is plan-type to plan-type. Left: American citizens living in Puerto Rico fill more prescriptions every single year — 19% more by 2023 (the gold label above each pair). Center: and yet the island spends less on drugs per person — about 15% less. Right: divide one by the other — the cost of a single prescription — gross cost per prescription — where the gap is widest: PR fills a prescription for roughly 70–71 cents on the national dollar, every single year. In this third chart the US Virgin Islands — the neighboring territory whose MA benchmark runs ~24% above PR’s — is included for context: USVI pays $108 per fill versus PR’s $63.

How many prescriptions does each enrollee fill per year?

Source: CMS Program Statistics — Medicare Part D, Utilization Table D6 (by Area of Residence, MA-PD columns), CY2020–2023 files — prescription fills per enrollee column; Puerto Rico and US national rows.

How much is spent on drugs per enrollee per year?

Source: CMS Program Statistics — Medicare Part D, Utilization Table D6 (by Area of Residence, MA-PD columns), CY2020–2023 files — gross drug cost per enrollee column; Puerto Rico and US national rows.

Gross cost per prescription ($/fill)

Source: CMS Program Statistics — Medicare Part D, Utilization Table D6 (MA-PD columns), CY2020–2023. $/fill is the exact division of the two published columns (cost per enrollee ÷ fills per enrollee), computed identically for PR, USVI and the US.

44.1M fills: actual cost vs national price

One more angle, same files: Puerto Rico’s 44.1M CY2023 prescriptions would have cost $3.93B at the national $89.16/fill, versus $2.79B actual at $63.30 — an implied ~$1.14B/yr of pharmacy efficiency. Source: CMS Program Statistics — Medicare Part D, Table D6, CY2023 MA-PD cells — 644,605 enrollees × 68.43 fills × $63.30/fill actual, vs the same fills × the national $89.16/fill. Illustrative computation from published columns.

MORE MEDICINE, LESS MONEY — EVERY STATE AND TERRITORY, CY2023

Two national rankings. Across all 56 US states and territories, Puerto Rico ranks #1 on both measures: the lowest cost per prescription in the nation ($63.65, versus a US average of $93.48 and $144.93 in Washington DC) and the highest number of prescriptions filled per person (68 per year, versus a US average of 56). Puerto Rico delivers the greatest access to medication at the lowest unit cost — no other jurisdiction achieves both.

RANKING 1 · Cost per prescription (3 lowest vs 3 highest, of 56 US jurisdictions)

RANKING 2 · Prescriptions filled per person per year (3 highest vs 3 lowest, of 56 US jurisdictions)

Source: CMS Program Statistics — Medicare Part D, Table D6 (by Area of Residence, overall Part D columns), CY2023. Cost per fill = published gross cost per enrollee ÷ published fills per enrollee, per jurisdiction. 56 US jurisdictions plotted (50 states + DC + Puerto Rico + US Virgin Islands + Guam + American Samoa + Northern Mariana Islands); CMS’s "Foreign Countries" and "Unknown" rows excluded.

GROSS COST PER PRESCRIPTION — ALL 56 US JURISDICTIONS RANKED, CY2023

How to read it: the lowest-cost prescription in the Medicare program is filled in Puerto Rico: $63.65 — #1 of 56 US jurisdictions, 32% below the national average ($93.48, dashed line). At the other end: the US Virgin Islands at $127.84 (#55) — the territory whose MA benchmark runs ~24% above PR’s pays twice as much per fill — and the District of Columbia at $144.93 (#57), 2.3× Puerto Rico.

Source: CMS Program Statistics — Medicare Part D, Table D6, CY2023 — same computation as the rankings above, sorted from highest to lowest cost per prescription.

PRESCRIPTIONS FILLED PER PERSON PER YEAR — ALL 56 US JURISDICTIONS RANKED, CY2023

How to read it: the highest prescription use per person in the Medicare program is in Puerto Rico: 68.0 fills per year — #1 of 56 US jurisdictions, 22% above the US average of 55.6 (gold dashed line). West Virginia (67.3) and Kentucky (66.1) follow. The US Virgin Islands (33.5) ranks #54. Combined with the cost ranking above, Puerto Rico ranks #1 on both measures: the most medication per person at the lowest cost per prescription.

Source: CMS Program Statistics — Medicare Part D, Table D6 (by Area of Residence, overall Part D columns), CY2023 — prescription fills per enrollee column, all 56 US jurisdictions; sorted from highest to lowest. CMS’s "Foreign Countries" and "Unknown" rows excluded.

WHAT THE ISLAND ACTUALLY FILLS — PUERTO RICO’S TOP 12 MEDICATIONS, CY2024

How to read it: each bar is one molecule (all brand and generic forms summed), numbered by its Puerto Rico rank (#1 = most filled), measured in total Medicare Part D claims filled in Puerto Rico in CY2024 — 30.9 million claims across 1,156 distinct generics island-wide. The label on each bar shows where that molecule ranks nationally. Eleven of the twelve are low-cost chronic-disease generics — thyroid, statins, antihypertensives, diabetes, GI — the pharmacology of adherence. And the island’s profile is its own: famotidine ranks #5 in PR vs #25 nationally; clonazepam #8 vs #46. Levothyroxine — not atorvastatin, as in the nation — is Puerto Rico’s most-filled medication: 1,329,012 claims by 226,885 beneficiaries.

Source: CMS, Medicare Part D Prescribers — by Geography and Drug, CY2024 (file MUP_DPR_RY26_P04_V10_DY24_Geo.csv, May 2026 release). Rows aggregated by generic name across brand variants; claims include refills. Note: this dataset covers all Part D (MA-PD + PDP) and counts claims, not standardized 30-day fills — a different vintage and unit than the CY2020–2023 utilization tables above, cited separately.

YEAR BY YEAR — PR vs US (MA-PD basis)

Year PR fills/enrolleeUS fills/enrollee PR vs US PR $/enrolleeUS $/enrollee PR $/fillUS $/fill PR MA-PD enrollees

Source: CMS Program Statistics — Medicare Part D, Utilization Table D6 (by Area of Residence, MA-PD columns), CY2020–2023 files: CPS_MDCR_UTLZN_D_2020, CPS_MDCR_UTLZN_D_2021, MDCR_UTLZN_D_CPS12UPD_2022, MDCR_UTLZN_D_CPS12UPD_2023. All columns are the published CMS figures; the "PR vs US" ratios are direct divisions of those columns. Gross drug costs as published by CMS, before manufacturer rebates. PR figures are MA-PD basis — 98% of all PR Part D enrollment; PDP enrollment in PR is 11,923 (CY2023).

The $0.00 column — the low-income subsidy that stops at the water's edge

LIS cost-sharing subsidy, PR plan rows
$0.00
in 70 of the 85 rows under PR-market contracts — five of six contracts at zero on every plan (CY2024 plan payment file). Nationally, 89.9% of plan rows receive LIS cost-sharing: mean $69.48 across all plans, $77.32 among those receiving [18]
LIS paid per Part D claim
$0.019
vs $0.36 nationally — recomputed exactly from CMS's claims API
Enrollees with full LIS
1.8%
vs 24.4% nationally — in the program's poorest jurisdiction

The zero is statutory, and CMS's ledger prints it: “In the case of a part D eligible individual who is not a resident of the 50 States or the District of Columbia, the individual is not eligible to be a subsidy eligible individual under this section” — SSA §1860D-14(a)(3)(F). The substitute is the Enhanced Allotment Program (§1935(e)), capped near $42–49M a year; full stateside parity would imply on the order of $2B. The console's working figure for foregone assistance — ~$250M/yr — is our own deliberately conservative estimate between those anchors, labeled as such. This one is aimed at Congress as much as CMS. (Technical footnote: PR's per-claim figure isn't exactly zero because the dataset keys on prescriber geography — stateside LIS beneficiaries filling PR-prescribed scripts.)

[16] 42 U.S.C. §1395w-114(a)(3)(F), U.S. Code 2024 ed. (verbatim via GPO) · Part D Prescribers by Geography & Drug (per-claim LIS recomputed: $599,600 / 30.9M PR claims vs $622.0M / 1.71B national) · CY2024 Part D plan file (zero column; statute-guaranteed construction).

Ten conclusions

Findings — with provenance

The case rests on a single, evidenced premise: Puerto Rico already delivers the outcomes CMS wants Medicare Advantage to achieve nationally — at the lowest payment levels in the country. Stabilize its benchmark and it becomes the national model; let underpayment persist and the model erodes. The findings below are deliberately self-critical — they concede where CMS's position holds and isolate the argument that survives scrutiny. Every source links to the original document. The findings open with the proposed remedy — benchmark parity with the US Virgin Islands (Finding I) — followed by the evidence that supports it.

I.

The proposed fix: benchmark parity with the US Virgin Islands

The US Virgin Islands — an adjacent US territory operating under the same Medicare statute — receives a CY2027 benchmark of $967 PMPM. Puerto Rico receives $778 PMPM: $189 PMPM below the standard CMS itself has already determined appropriate for a Caribbean US territory. The proposal is narrow: apply to Puerto Rico the same benchmark treatment CMS already applies to the US Virgin Islands. Not mainland rates; not a special adjustment — an existing CMS payment standard, extended to a directly comparable jurisdiction.

The comparison removes every confounder typically raised in benchmark debates: same region, same statute, same program rules — yet the lower-paid market demonstrably outperforms the higher-paid one, and every state, on penetration, Star Ratings, and drug-cost efficiency. No payment-policy rationale has been articulated for a $189 PMPM differential between two adjacent territories in which the lower rate is assigned to the stronger performer. Absent such a rationale, USVI parity is the defensible floor — the operational remedy available now, while the structural basis reform in Finding VIII proceeds. The full argument is laid out in the Policy Position tab.

Sources
Benchmarks (PR & USVI) CMS CY2027 MA Ratebook · county/municipio rate books, territory averages
Finding I · Benchmark comparison · CY2027 · $/PMPM · Source: CMS MA Ratebook
II.

Anticipated rebuttal — CMS will point to its FFS data. That data measures the wrong population.

CMS's published FFS expenditure file shows Puerto Rico per-capita costs at $455.50 per month against a mainland $992.11 (45.9%, 2024). Expect CMS to cite this as proof that low benchmarks reflect low costs. The number is real — and it is the product of a measurement base that has collapsed. With 96.2% of PR's Parts A+B beneficiaries in MA, the FFS file is computed from the residual ~4% sliver (see Finding IV). A cost estimate built on the 4% who remain in FFS says nothing about the cost of caring for the 96% who left it.

The advocacy position is not "match FFS better." It is that the FFS basis itself is structurally unusable in Puerto Rico — the same conclusion the benchmark inversion in Finding I forces from the payment side.

Sources
Data file ffs-data-2024.zip · ma_ffs_expenditure sheet FFS24
CMS document CY 2027 Rate Announcement, p. 47
III.

Anticipated rebuttal — "PR already gets the program's highest multiplier." That multiplier is the proof, not the defense.

Puerto Rico's benchmarks run 1.48× its FFS baseline, versus 1.13× on the mainland. CMS may frame this as generosity. Read it the other way: the methodology is already stretching 48% above its own baseline in Puerto Rico — and the result is still the lowest payment rate in the nation ($730 vs $1,246 PMPM). When the program's largest upward correction still produces the program's smallest benchmark, the baseline is broken, not the correction excessive.

This is why the policy ask (Finding V / Policy tab) is framed on USVI parity — an existing CMS treatment of a comparable jurisdiction — and not on the FFS multiplier at all.

Sources
Joined data ma_ratebooks_county joined to ma_ffs_expenditure, 2018–2024
CMS landing page MA FFS Data, 2015–2024
Script & chart scripts/ffs_vs_ma_analysis.py · chart_ma_to_ffs_ratio.png
IV.

The credibility argument is the one that survives

The average Puerto Rico municipio has 376 FFS beneficiaries (2024) from which CMS measures local costs — 4% of the mainland county average of 9,441, a roughly 25-fold gap. And it is shrinking: 687 per municipio in 2018, down 45% in six years. Island-wide, the entire FFS base is roughly 29,000 people across 78 municipios (2024), down from ~54,000 in 2018. This is the strongest argument that the FFS basis is structurally inadequate.

MA penetration in PR is 96.2% of Parts A+B beneficiaries (mainland: 55.5%; on the penetration-file all-eligibles basis, 85.2% vs 51.3%). When CMS sets PR rates from FFS data, that data covers only ~4% of the PR Medicare-eligible population.

V.

PR plans are top-tier on quality, despite bottom-tier benchmarks

Three of the top four PR contracts hold five-star ratings in 2026 (MCS Classicare H5577, MMM H4003, MMM H4004), covering 81.7% of PR MA enrollment. Adding the 4.5-star Triple-S Advantage brings coverage to 99.1% of PR MA in 4.5+⭐ contracts.

Quality-payment paradox · CY2026 benchmark vs enrollment-weighted star rating · Source: CMS

This empirically confirms the 2022 Roberts & Song finding: after the ACA lowered PR's structural benchmarks, PR plans pursued quality bonuses and risk-score work to recapture revenue. They succeeded.

Sources
Peer review Roberts JM, Song Z. JAMA Health Forum 2022 · PMID 36218937
VI.

MedPAC has made the credibility argument for eight years

MedPAC's March 2017 report, Chapter 13, p. 20: "as MA penetration continues to grow, it leaves fewer, and perhaps less representative, beneficiaries on which to calculate FFS spending." MedPAC recommended expanding the Part A+B-only adjustment nationally; CMS declined.

MedPAC's March 2025 report goes further — it excludes Puerto Rico from its favorable-selection and coding-intensity analyses "due to the relatively small number of FFS beneficiaries in that territory." MedPAC will not use the same data CMS uses to set PR rates.

Sources
MedPAC March 2017 Report to Congress, Ch. 13
MedPAC March 2025 Report to Congress, Ch. 11
MedPAC July 2024 Data Book, Section 9
VII.

The PR MA market is a three-firm oligopoly

~98.8% of PR MA is concentrated in three corporate parents: MCS Classicare (MHH Healthcare, PR-local) at 52.9%, MMM (Elevance Health, mainland) at 28.9%, and Triple-S Advantage (Guidewell, mainland) at 18.2%.

HHI of 3,514 is deep in DOJ "highly concentrated" territory. Only one of the three dominant parents is genuinely PR-headquartered. This is the fragility argument: with so few carriers, if benchmark compression pushes one major plan to exit, beneficiaries have almost nowhere to go — so the highest-rated MA market in the nation is also one of the most exposed to a single carrier's withdrawal. Stabilizing the benchmark protects against that shock.

VIII.

The credible path is "fix the basis," not "fix the rates"

Use Innovation Center §1115A demonstration authority or amend §1853 to set PR rates on something other than a tiny, shrinking, selection-biased FFS sample. This is what plan commenters and the bipartisan Hernández/Fitzpatrick/Espaillat/Bacon letter (February 2026) asked for.

The Hernández letter proposes §1876(a)(4) authority for a 0.70 AGA minimum benchmark floor. CMS declined that remedy in the 2027 Rate Announcement.

IX.

More medicine, less money — the funds convert into care

Puerto Rico's Medicare Advantage members fill 19% more prescriptions per enrollee per year than the national average (68.4 vs. 57.4 fills, CY2023) — at 15% lower gross drug cost per enrollee ($4,331 vs. $5,122). The implied cost per prescription is roughly $63 in Puerto Rico versus $89 nationally — about 29% lower.

The pattern holds every year from 2020 through 2023 (fills +16% to +21%; cost −13% to −17%). This rebuts the inference that Puerto Rico is inexpensive because it under-delivers: utilization is higher, unit cost is lower. Medication adherence — precisely what CMS Star Ratings reward — is where the dollars go. Note: figures are gross drug costs as published by CMS, before manufacturer rebates.

Sources
CMS dataset CMS Program Statistics — Medicare Part D · Utilization Tables 6–7, by Area of Residence
X.

Puerto Rico’s duals are invisible in Medicare’s dual-status systems — even though CMS’s own Medicaid files count ~300–375K

CMS’s dual-enrollment tables show Puerto Rico with 13,255 dual eligibles (1.7%) in CY2023 — dead last among the 50 states and DC (national average: ~19%) — and the count has kept falling: 7,283 (2024), 6,187 (2025), 5,581 (March 2026). Against the 307,571 D-SNP enrollees verified in CMS’s own SNP report, the undercount now runs ~55× — volatility that is itself evidence of measurement failure, not measurement. The real population is ~302,000 D-SNP enrollees (MedPAC, 2023) and 288,427 Platino enrollees (ASES, Nov 2025) — ≈39.5% of PR’s 765,273 beneficiaries, which would rank Puerto Rico #1 in the nation, above DC (38.8%).

The cause is statutory, not demographic: Puerto Rico is exempt from the Medicare Savings Programs (QMB/SLMB/QI) that generate federal dual-status flags (§1905(p)(4)(A) of the Social Security Act), so Platino duals — 98% of whom enroll in D-SNPs (KFF, 2021) — carry no federal dual coding — the MMA state-file exchange that creates those flags is mandated for “the fifty states and the District of Columbia” only. The invisibility is specific to Medicare’s ledger: CMS’s own T-MSIS-derived Medicaid files count 355,720 PR full duals (2022 monthly average) — the population is fully documented inside the very agency whose dual tables cannot see it. MedPAC excludes PR from its dual analyses for exactly this reason: “Data exclude Puerto Rico because enrollment data undercount dual-eligibility categories.” The same federal blindness that hollows out the FFS benchmark (Finding IV) erases PR’s poorest, sickest beneficiaries from the data CMS uses to calibrate dual policy.

Sources
CMS dataset CMS Program Statistics — Medicare-Medicaid Dual Enrollment · MDCR ENROLL AB 42/45/48, CY2020–2023
Cross-check Medicare Monthly Enrollment · DUAL_TOT_BENES, state-level
Federal D-SNP count MedPAC July 2025 Data Book, Sec. 9 · ~302,000 of ~654,000 PR MA enrollees (2023)
Platino count ASES via Revista Seguros · 288,427 (Nov 2025)
XI.

The federal invisibility matrix — where Puerto Rico is absent, suppressed, or subtracted

The measurement failures documented above are not isolated — across the federal statistical apparatus, Puerto Rico is systematically missing, and often by design. CMS’s actuaries run an adjustment to exclude spending for U.S. territories from the $5.3T National Health Expenditure Accounts; the 2026 Medicare Trustees Report mentions Puerto Rico once in 271 pages. Each row below carries its scope caveat so none can be dismissed as overframed — and one row cuts the other way on purpose: T-MSIS proves the data exists inside the very agency whose other ledgers cannot see it.

Federal dataset / systemPuerto Rico’s status (verified)Caveat to print
MMCO dual-enrollment snapshots0 PR rows in all 54 snapshots (2007–2025); American Samoa (249) and Guam (139) do appearStatutory: MMA file mandate covers 50 states + DC
Geographic Variation PUF (FFS)Dual-share cell = "*" for PR & VI in all 11 years; PR's 78 municipios absent from county levelOnly 2 of 53 geographies suppressed
MA Geographic Variation PUFDual-share "*" all 7 years (2016–2022)File is stale (ends 2022)
National Health Expenditure AccountsTerritory spending actively subtracted from the $5.3T national ledger (documented method)Applies to all territories, by convention
MCBS (flagship beneficiary survey)PR sampled 1991–2016; all collection discontinued 2018"Counted for 25 years, then dropped"
MEDPAR Inpatient/Outpatient PUFsZero PR rows, every year 2013–2024 — an undocumented absenceSay "does not appear," not "excluded by policy"
Market Saturation & Utilization0 of 1,030,290 rows; Kauai County (10,504 FFS lives, 1 provider) gets 12 tracked metrics, San Juan gets noneFFS-only program-integrity tool
Hospital star ratings53 of 60 PR hospitals unrated (national norm: 41% unrated)Unrated ≠ bad quality; FFS-measure starvation
2026 Medicare Trustees ReportOne mention of Puerto Rico in 271 pages
Part D LIS payments$0.00 in 70 of 85 PR-contract plan rows vs $69.48 national mean (89.9% of plans nationally receive it)Statutory exclusion — aim at Congress too
HOS Public Use FilePR surveyed but blinded into Region 2 aggregate"Collected, then rendered invisible"
Medicaid monthly enrollment reporting (PI data)0 rows for all five territories, every month
T-MSIS TAF dual filesPR PRESENT: 355,720 full duals (2022) — proves the data existsDQ flag "not assessed"; ends 2022
State Drug Utilization DataPR absent through 2022, onboarded 2023 (43,173 rows) — pipes can be fixed within a year when required
Sources
Verification Each row checked against the named federal file — research sweep of July 15, 2026 across data.cms.gov, cms.gov/data-research, data.medicaid.gov, and the Provider Data Catalog.

The argument, in full

Policy position

The complete policy narrative for CMS, MedPAC, and Congress — Puerto Rico as the national benchmark for high-quality, high-efficiency Medicare Advantage, and one concrete proposal: benchmark parity with the US Virgin Islands. Every figure below matches the data shown elsewhere in this console.

Puerto Rico, with a stabilized benchmark structure, should not be viewed as a distressed Medicare Advantage market in need of rescue. The evidence supports the opposite conclusion: Puerto Rico already represents one of the strongest real-world demonstrations of what CMS is working to achieve nationally through Medicare Advantage — and, with a stabilized benchmark, it can stand as the national benchmark for the program.

The central paradox

Puerto Rico has achieved near-universal Medicare Advantage participation — approximately 96.2% MA penetration (MA enrollees as a share of Parts A+B beneficiaries, May 2026) versus roughly 55.5% on the mainland — while simultaneously delivering elite quality under CMS's own Star Ratings system. Under the 2026 CMS Star Ratings, three of Puerto Rico's four largest Medicare Advantage contracts earned the highest possible 5-Star designation — including MCS Classicare (H5577), the single largest 5-Star Medicare Advantage contract in the United States. In all, approximately 99.1% of Puerto Rico MA beneficiaries are enrolled in plans rated 4.5 Stars or higher — a concentration of quality unmatched in most mainland markets.

Puerto Rico produces these outcomes while operating at among the lowest benchmark payment levels in the nation. For CY2027, Puerto Rico's average Medicare Advantage benchmark is approximately $778 PMPM, against approximately $1,313 PMPM on the mainland — roughly 59 cents for every dollar paid in the continental United States — and against $967 PMPM in the US Virgin Islands, an adjacent territory operating under the same Medicare statute — roughly 80 cents for every dollar paid next door.

The market producing some of the strongest quality outcomes, the highest beneficiary participation, and the most consistent high-performance enrollment in the country is funded at among the lowest benchmark levels in the country.

The issue is not performance

The federal policy question is not whether Puerto Rico's Medicare Advantage system works. The data already shows that it does. Nor is the question whether Puerto Rico requires a bailout or special treatment. Puerto Rico is already producing many of the precise outcomes CMS seeks nationally:

  • exceptionally high Medicare Advantage participation;
  • consistent enrollment in elite-rated plans;
  • strong quality and compliance performance;
  • large-scale, population-level chronic disease management; and
  • efficient care delivery under materially constrained funding.

The actual question is whether CMS intends to stabilize and preserve this model before prolonged structural underpayment erodes the financial conditions necessary to sustain it.

The drug-utilization data makes the efficiency concrete: the funds are not absorbed — they are converted into care. Puerto Rico's Medicare Advantage members fill roughly 19% more prescriptions per year than the national average at roughly 15% lower gross drug cost per enrollee — about $63 per prescription versus $89 nationally (CY2023, CMS Program Statistics), a pattern that holds every year since 2020. Medication adherence — precisely what CMS Star Ratings reward — is where the dollars go, delivered at the lowest unit cost in the nation.

The risk-score counterargument — answered with the audit record

Puerto Rico’s average Part C risk score is high — 1.53 vs 1.38 for the states (CY2024 plan payment file, Rates tab) — and a critic will read that as coding intensity. Three verifiable facts frame the honest answer. First, the validation regime is national, and Puerto Rico is inside it. CMS’s RADV audits test whether diagnoses submitted for payment are supported in the enrollee’s medical record; the 2023 final rule (88 FR 6643) authorizes extrapolated recoveries beginning with payment year 2018 with no FFS adjuster, and in May 2025 CMS moved from auditing ~60 contracts a year to every eligible MA contract annually, targeting the PY2018–2024 backlog by 2026. Second, PR contracts already appear in the published audit record — a Humana Puerto Rico plan was among the 90 PY2011–2013 RADV audits released in 2022, and HHS-OIG’s 2024 audit of MMM Healthcare (H4003) found 108 of 688 sampled diagnosis groups unsupported and recommended a $165,312 enrollee-level refund, consistent with the rule’s pre-2018 policy. Third, coding intensity is a nationwide phenomenon, not a Puerto Rico oneMedPAC estimated MA risk scores roughly 16% above an FFS-equivalent nationally for 2025, an estimate its March 2026 report revised downward under the v28 model phase-in. One caveat carried openly: in September 2025 a federal district court vacated the 2023 rule on procedural grounds (Humana v. Becerra, N.D. Tex.); CMS has appealed to the Fifth Circuit and continues the expanded audits and enrollee-level recoveries while extrapolation is litigated. The defensible claim is therefore not that Puerto Rico’s scores are proven clean — it is that they sit inside the same medical-record validation machinery as every mainland contract, while the island’s dual-heavy demographics (Duals tab) and measured chronic-condition prevalence (Circularity tab) supply documented clinical reasons for above-average acuity.

[19] 88 FR 6643 (Feb 1, 2023) · CMS press release, May 21, 2025 · HHS-OIG A-04-20-07090 (Aug 2024) · KFF Health News (Nov 2022) · MedPAC March 2025 Report to the Congress, ch. 11 · Humana v. Becerra (N.D. Tex., Sept 25, 2025), appeal pending. Full verification register [19], Circularity tab.

The policy position: benchmark parity with the US Virgin Islands

Apply to Puerto Rico the same benchmark treatment CMS already applies to the US Virgin Islands.

This is the position, stated in full. It is not a request for mainland rates. It is not a request for any special adjustment, demonstration, or carve-out. It is the application of an existing CMS payment level to a directly comparable jurisdiction: adjacent geography, identical Medicare statute, identical program rules, and a shared regional health-care economy. The US Virgin Islands benchmark — $967 PMPM for CY2027 — is what CMS already pays in a Caribbean US territory under its own rate machinery. Puerto Rico's benchmark of $778 PMPM sits $189 PMPM below it. The discretion to close that gap is documented in CMS's own rate documents, twice over. First: the statute ranks benchmark quartiles only for counties in “the 50 States and the District of Columbia” (§1853(n)(2)), so every territory county's applicable percentage is assigned by administrative decision — “For the territories, CMS assigns an applicable percentage to each territory county based on where the territory county rate falls in the quartiles established for the 50 States and the District of Columbia” (CY2027 Advance Notice, Attachment II). Second: CMS has adjusted Puerto Rico's own FFS inputs by discretion every year since the CY2017 Rate Announcement — computing PR rates from beneficiaries with both Parts A and B and correcting for the nationwide propensity of zero-claims beneficiaries. The authority is not hypothetical; CMS exercises it annually.

The comparison removes every confounder typically raised in benchmark debates. Both jurisdictions are US territories. Both operate under the same statutory framework. Both serve Caribbean populations through the same federal program. CMS’s own Geographic Variation file prices their care within 2.4% of each other — $6,437 versus $6,597 per FFS beneficiary (actual dollars, 2024) — while the benchmarks sit 19.5% apart. Yet the lower-paid market — Puerto Rico — demonstrably outperforms the higher-paid one, and every state in the program, on penetration, Star Ratings, and drug-cost efficiency. No payment-policy rationale has been articulated that justifies a $189 PMPM differential between two adjacent territories under the same statute, in which the lower rate is assigned to the stronger performer. Absent such a rationale, benchmark parity with the US Virgin Islands is the defensible floor — the narrowest correction available, anchored entirely in CMS's own precedent.

This framing also defines the burden of proof. The question before CMS is not whether Puerto Rico merits an increase; it is whether the agency can defend treating Puerto Rico worse than the jurisdiction next door. If $967 is the right benchmark for the US Virgin Islands, the agency must articulate the basis on which $778 is right for Puerto Rico. CMS has faced a version of this question before: in the CY2018 Rate Announcement it declined a proposal to use the USVI rate as a proxy for Puerto Rico, maintaining that Puerto Rico's own FFS data was sufficient to set accurate rates. That defense is precisely what this console tests — the measurement base behind it, now roughly 4% of Puerto Rico's Medicare-eligible population and shrinking, is the circularity documented throughout. The data in this console indicates no such basis exists.

CMS's own cost data vs CMS's own benchmarks — Puerto Rico as a share of USVI

How to read it: the green bar is CMS's Geographic Variation file pricing Puerto Rico's care at 97.6% of USVI — $6,437 vs $6,597 per FFS beneficiary (actual, non-price-standardized dollars, 2024; in standardized dollars the ordering reverses, $6,685 vs $6,126 — by either measure PR's measured cost is equal or higher). The red bar is what the benchmark pays: 80.5%. Both cost figures ride on thin FFS bases — PR 26,979 lives, USVI 12,643 — which is the Circularity tab's argument, and it cuts against the benchmark, not for it.

Sources — costs: Medicare Geographic Variation PUF, TOT_MDCR_PYMT_PC (actual), state level, 2024 · benchmarks: CY2027 ratebook ($778 / $967 PMPM).

Puerto Rico as a national benchmark

Puerto Rico should be viewed not as a peripheral exception within Medicare Advantage policy, but as a potential national benchmark for how a mature, high-performing MA ecosystem functions when beneficiaries overwhelmingly participate in coordinated care and remain concentrated in highly rated plans. With a stabilized benchmark methodology, Puerto Rico can serve as a national showcase demonstrating that high enrollment, elite quality, strong compliance, and lower overall Medicare spending can coexist within a properly aligned MA system.

Why stabilization matters

Puerto Rico's performance has been achieved despite severe benchmark compression — not because that compression is sustainable. Over time, the current benchmark structure creates pressures that are increasingly difficult to absorb: margin compression, reduced provider reimbursement flexibility, physician-migration and specialist-access risk, workforce recruitment challenges, and diminished reserve and investment capacity. These pressures do not produce immediate collapse; they gradually weaken the financial and operational flexibility required to sustain elite performance. The objective now is to ensure the benchmark methodology does not undermine the long-term sustainability of a proven success.

The strategic choice

In a truly performance-based Medicare Advantage system, Puerto Rico would not rank near the bottom of the benchmark structure. It would stand as one of the clearest national demonstrations of how Medicare Advantage can simultaneously improve quality, increase participation, maintain compliance, and manage costs at scale. The choice facing federal policymakers is therefore not whether to rescue a failing market. It is whether to:

  • extend the benchmark standard CMS already applies to the US Virgin Islands, stabilizing one of the most efficient, highest-performing Medicare Advantage markets in the country and letting it serve as the national benchmark; or
  • allow continued structural benchmark compression to gradually weaken the very model CMS is working to build nationwide.

The evidence strongly supports the first approach.

Sources CMS CY2027 MA Ratebook · 2026 Part C & D Star Ratings · Monthly Enrollment by CPSC (August 2026). Enrollment-weighted figures reflect Puerto Rico MA contracts only (standalone Part D excluded).

The hospital wage index and the October 2027 cliff

FY2028 watch

The next structural decision point for Puerto Rico’s Medicare payments is the FY2028 IPPS rulemaking (proposed rule expected ~April 2027; final ~August 2027; effective October 1, 2027). This page states the verified facts and the two administrative requests before CMS. Sources follow each figure.

Puerto Rico’s hospitals carry the lowest Medicare area wage index in the nation — and the temporary protection that has cushioned them since the courts struck down CMS’s low-wage-index policy is in its final approved year. What CMS decides in the FY2028 rule determines whether that slide stops — or keeps compounding at the capped 5% a year (42 CFR §412.64(h)(7)) toward a cumulative ≈15.5% inpatient payment cut — and, through the ratebook’s claim-repricing machinery, what happens to every municipio’s Medicare Advantage benchmark from CY2029 forward.

On August 5, 2026, Gov. González-Colón welcomed a fourteen-Member letter led by Rep. Greg Murphy asking HHS to establish, beginning in 2028, a minimum MA benchmark equal to 70% of the national AGA — the level the U.S. Virgin Islands already receives.

The Murphy letter, Aug. 3, 2026 (PDF) · PRFAA release, Aug. 5, 2026 · full record in The paper trail below

The verified position, FY2027

  • Official wage indexes: San Juan–Bayamón–Caguas 0.3500, Guayama 0.3654, Ponce 0.3303, with Aguadilla, Arecibo and Mayagüez held at 0.3278 by the statutory imputed floor (ARPA §9831). The lowest area anywhere else in the nation is rural Mississippi at 0.7112. Source: FY2027 IPPS final rule (CMS-1849-F, 91 FR 49570, Aug. 4, 2026), Table 3.
  • What hospitals are actually paid on is a post-Bridgeport transitional exception — a payment-equivalent wage index of ≈0.5259 (85.74% of FY2024), declining 5% per year. All 55 hospitals carrying an FY2027 transition value are in Puerto Rico. Source: 91 FR 49570; FY2027 Table 2.
  • If nothing replaces the protection, Medicare inpatient payments to the island’s 52 IPPS hospitals settle ≈15.5% lower (the payment multiplier (0.62×WI+0.38) evaluated at 0.35 versus 0.5259) — reached not in one day but in 5%-a-year wage-index steps (42 CFR §412.64(h)(7)): a payment cut of roughly 2% compounding every October 1 for about eight years. Computation from the statutory 62% labor-related share, 42 U.S.C. 1395ww(d)(3)(E)(ii).
  • The political front is moving. On February 25, 2026, eighteen Members of Congress led by Rep. Pablo José Hernández asked HHS and CMS, in comments on the CY2027 MA Advance Notice, to set a minimum AGA factor of 0.70 for Puerto Rico under SSA §1876(a)(4) — noting benchmarks ≈41% below the national average, 674,000 seniors, and 94% MA enrollment (the letter). And on August 5, 2026, the Governor welcomed the fourteen-Member Murphy letter carrying the same ask into the 2028 rate cycle — featured in the callout at the top of this tab. An AGA of 0.70 would sit above every one of Puerto Rico’s 78 municipios today (CY2027 Rate Calculation Data: median 0.504, highest 0.650). The full five-document record — both congressional letters, the Governor’s formal comment, the healthcare community’s MOU, and the Puerto Rico Republican Party’s letter — is collected in The paper trail below, each linked to its primary source.
  • The wage arithmetic underneath: Puerto Rico hospital average hourly wages run $14.37–$31.77 against a ≈$60 national average. The four highest-wage hospitals on the island are the public institutions of Centro Médico — ASEM ($31.77), Centro Cardiovascular ($26.10), Centro Comprensivo del Cáncer ($24.03), Hospital Universitario ($24.02). Source: FY2027 Table 2, Average Hourly Wage FY2027 column.
  • Because CMS reprices all five historical claims years to current wage indexes when building the MA ratebook, the wage-index outcome of FY2028 reaches municipio benchmarks in the CY2029 ratebook (January 1, 2029) — all at once, no phase-in. Source: CY2027 Advance Notice (Jan. 26, 2026), repricing methodology.

The “15.5% cliff” was real — it just takes eight years

Early framings of the FY2028 problem described a one-day 15.5% payment cut. The number was right; the timeline was not. Because the wage index applies only to the 62% labor-related share of the IPPS operating payment (42 U.S.C. §1886(d)(3)(E)(ii)), each −5% wage-index step costs about 2% of payment — and compounding every remaining step of the glidepath from the FY2027 median (0.5259) down to the computed index (≈0.3500) yields a cumulative operating-payment cut of ≈15.4%. The “overnight cliff” figure was this same cumulative effect, compressed into one day. Spread over eight years at ≈2% per year, it is the quiet bleed — and via the AGA repricing, each step reaches the MA benchmarks one calendar year later.

Fiscal yearWage index (median, modeled)Payment change vs. prior year
FY20280.4996−2.31%
FY20290.4746−2.25%
FY20300.4509−2.18%
FY20310.4284−2.12%
FY20320.4070−2.06%
FY20330.3866−2.00%
FY20340.3673−1.93%
FY2035≈0.3500 (floor)−1.77%

Modeled at −5%/year from the FY2027 median transition value (0.5259, CMS-1849-F Table 2) until it meets the computed index (0.3500, held constant — the final step is partial). Payment factor = 0.62 × wage index + 0.38 (operating base payment only; capital, DSH, uncompensated care and outliers follow separate formulas). Cumulative FY2027→FY2035: 0.5970 / 0.7061 = −15.4%. Computed Sept. 1, 2026.

The paper trail — one number, five documents

How to read it: in six months, four separate constituencies — the island’s Democratic-aligned Resident Commissioner bloc, the Governor, the island’s healthcare community, the local Republican Party, and House Republicans — independently put the same ask on the federal record: a national MA benchmark floor at 0.70 of the national geographic adjustment, the level the U.S. Virgin Islands already receives, under CMS’s existing authority (SSA §1876(a)(4)), starting with the 2028 rate cycle. Every document below is the primary source, hosted on this site.

  1. Feb. 25, 2026 — Rep. Pablo José Hernández + 17 Members of Congress (bipartisan) to HHS/CMS, commenting on the CY2027 MA Advance Notice: set a minimum AGA of 0.70; maintain the zero-claimant and A&B-enrollment adjustments; extend MSP, LIS and LTSS. Letter (PDF) · house.gov original
  2. Feb. 25, 2026 — Gov. Jenniffer González-Colón to Secretary Kennedy (docket CMS-2026-0034-0001), same day: the benchmark rests on a ≈29,000-person FFS remnant that is “neither representative nor statistically reliable”; requests a national MA benchmark floor of no less than 0.70, with interim bid-instruction fixes if the floor cannot land in 2027. Letter (PDF)
  3. May 18, 2026 — Memorandum of Understanding, Puerto Rico Healthcare Community (public and private stakeholders): seeks an administrative solution in the 2028 rate cycle — “a national minimum payment at 0.70 geographic adjustment, using the current rate for the U.S. Virgin Islands as the baseline” — plus the wage-index fix through rulemaking, and the Medicaid agenda (the Sept. 30, 2027 funding cliff, MSP matching funds, an LTSS pathway, holding FMAP at 76%). MOU (PDF)
  4. June 23, 2026 — Republican Party of Puerto Rico (State Chairman Ángel M. Cintrón) to Secretary Kennedy, backing the Governor: a National MA Benchmark Floor beginning with the 2028 payment year at the USVI level, ≈70% of the national AGA — noting benchmarks ≈41% below the national average and ≈20% below the USVI, and that the island is “being penalized because of its MA program’s overwhelming success.” Letter (PDF)
  5. Aug. 3, 2026 — Rep. Greg Murphy, M.D. + 13 House Republicans (Fitzpatrick, Bacon, Kelly, Sessions, Radewagen, Giménez, Salazar, Díaz-Balart, Moore, Bilirakis, Moylan, Carter, Tenney) to Secretary Kennedy: establish for 2028 a National Minimum MA Payment Level under SSA §1876(a)(4) — a benchmark floor at 70% of the national average, consistent with the USVI — citing the ≈29,000-beneficiary (3%) FFS sample and a gap “widening annually since 2012.” Letter (PDF) · PRFAA release

Documents supplied to and hosted by this console (epac-hub.github.io/CMS-Console/letters/); quotations transcribed from the PDFs. Independent verification on this tab: the ≈29,000 FFS figure matches the live measurement-base series on The Circularity Trap (27,275 as of March 2026; 26,497 by May 2026 — retrieved Sept. 1, 2026); the USVI ≈0.70 level matches the CY2026 rate file (St. Croix 0.67352, St. John 0.70552); the ≈41%-below-national and 20%-below-USVI gaps are consistent with the CY2026 plan-payment analysis on The Territories’ Perspective. Press coverage: EIN Presswire · El Vocero.

The two requests before CMS

  • Extend the transitional payment exception into FY2028 and subsequent years. CMS wrote in the FY2027 final rule that it “may consider this in future rulemaking” after commenters asked for exactly this, “particularly in Puerto Rico.” The American Hospital Association has separately urged that the transition be non-budget-neutral (AHA comments, June 9, 2026).
  • Hold the rural floor computation harmless. Commit that the rural Puerto Rico wage index used for the BBA 1997 §4410 floor is never computed below the contiguous-average methodology of the FY2008 final rule (72 FR 47323).

The public calendar

  • Sept 1 (annual) — MGCRB reclassification applications due, 13 months before the fiscal year.
  • Dec 31, 2026 — Medicare-Dependent Hospital program and low-volume adjustment expire absent congressional extension (CAA 2026 §§6201–6202).
  • ~Late Jan 2027 — CY2028 MA Advance Notice; FY2028 preliminary wage index file.
  • ~April 2027 — FY2028 IPPS proposed rule display; §412.103 reclassification approvals must land within 60 days of display to count in the FY2028 wage index; ~60-day public comment period follows. On the FY2027 rule, only 4 of 982 comments came from Puerto Rico institutions (docket CMS-2026-1256).
  • ~Aug 1, 2027 — FY2028 final rule. Oct 1, 2027 — FY2028 begins. Jan 1, 2029 — CY2029 benchmarks carry the outcome.

The legislative track runs in parallel: H.R. 8109 (119th Cong.) would codify the low-wage-index policy — restoring a ≈0.60 wage index permanently — and CMS has stated that a cost-of-living adjustment for the territories “would require legislation.” (91 FR 49570.)

This page tracks a live rulemaking cycle and is updated as CMS publishes; figures above are frozen to the FY2027 final rule vintage of August 4, 2026.

What the administration has done — and what it can still do

CY2027 rates are final (announced April 6, 2026). The next lever is the CY2028 rate cycle: the Advance Notice arrives in early 2027 with a public comment window, and the final CY2028 Rate Announcement follows by the first Monday of April 2027. The record below separates what CMS has already proven it can do administratively from what it says requires Congress — in its own words, from twelve years of published Rate Announcements, every one linked and page-cited.

The outcome · Territory stabilization at a 0.70 cost index · CY2028

In the CY2028 Rate Announcement, CMS adopts a uniform Territory Stabilization methodology: where a jurisdiction’s FFS population fails representativeness — measurable triggers this console documents, such as FFS falling below 5% of A+B beneficiaries (PR: 3.75% by May 2026, and falling) or a majority of rate cells sitting under CMS’s own 1,000-enrollee credibility line (PR: 69 of 78) — the effective cost index used in rate development is developed by similar-area estimation and shall be no less than 0.70 of national. One rule, all three territories: Guam (≈0.88) untouched, USVI (≈0.68) moves marginally, Puerto Rico (≈0.50) is repaired.

Why it is administratively permissible — and what it delivers

Because it is drafted as estimation methodology, never as a benchmark floor: (1) SSA §1876(a)(4) authorizes the Secretary to estimate per-capita cost “based upon an adequate sample… in a geographic area served by an eligible organization or in a similar area, with appropriate adjustments to assure actuarial equivalence” — where the local sample is inadequate, the statute itself names the remedy, and USVI (near-identical raw cost) is the natural similar area; (2) the zero-claims adjustment is ten years of precedent for Secretary-directed corrections to PR’s FFS experience; (3) CMS’s own characterization of the PR methodology (CY2020, reprised verbatim CY2022 and CY2023) — “our exercise in discretion for the data used to develop the estimate for one geographic area” — is the agency’s own statement of this authority. What is not permissible is a floor labeled a floor: CMS has refused that formulation on the record (“limited discretion to… applying floors for specific locales,” CY2027), and the D.C. Circuit’s vacatur of the IPPS low-wage-index policy (Bridgeport Hosp. v. Becerra, 108 F.4th 882 (D.C. Cir. 2024)) shows what happens to administratively-invented floors. The formulation is the difference between the outcome and a court loss.

Modeled against CMS’s own CY2027 rate fileValue
PR average benchmark today (CY2027)$778
PR average under 0.70 stabilization, uncapped$1,045
Municipios where the statutory pre-ACA cap binds and trims the fix37 of 78
PR average under 0.70 stabilization, after statutory caps$997 (+28%)
USVI average benchmark (the parity target)$967 (within ≈3%)
US mainland average (still far above the stabilized PR)$1,313 (PR remains 24% below)

The statute polices its own outcome. Every benchmark remains capped at the pre-ACA trajectory — the ceiling that already sets Guam’s and half of USVI’s rates — so the stabilization cannot overshoot: it lands Puerto Rico at USVI parity and stops. Verbatim, SSA §1853(n)(4), 42 U.S.C. §1395w-23(n)(4): “In no case shall the blended benchmark amount for an area for a year (determined taking into account subsection (o)) be greater than the applicable amount that would (but for the application of this subsection) be determined under subsection (k)(1) for the area for the year.” Because the cap applies after quality bonuses, nothing pierces it; CMS implements it in Risk2027.csv (“All 2027 rates are capped at 2027 Pre-ACA rate”), and this console’s model honors it — the caps trim the uncapped $1,045 to $997.

Who does what, by when: comments making this ask, with this record, in the CY2028 Advance Notice window (≈late January – early March 2027) · the §1853(d)(4) island-wide payment-area request filed by the Governor by February 1, 2027 as the structural companion · plans and the congressional delegation carrying the same single formulation. CY2027 is final; CY2028 is the actionable year.

Modeled figures are this console’s computation from CMS CY2027 Rate Calculation Data (FFS rate, pre-ACA cap, and 0%-bonus columns; USPCC $1,297.74; quartile 1.15): effective index = max(current, 0.70); benchmark = min(index × USPCC × 1.15, pre-ACA cap); averaged unweighted across 78 municipios — re-run live on the published file: uncapped $1,044.68, capped $996.53, 37 of 78 trimmed. Construction note, disclosed: the simplified formula yields $744 at current indices versus the published $777.98 (the published rate embeds GME/IME/KAC/VA-DOD factors), so the modeled $997 is the counterfactual under the same simplified construction — conservative relative to the full formula. Authorities: 42 U.S.C. §1395mm(a)(4); CY2020 Rate Announcement p. 24 (reprised CY2022 p. 41, CY2023 p. 33); CY2027 Rate Announcement pp. 46–47; Bridgeport Hosp. v. Becerra, 108 F.4th 882 (D.C. Cir. 2024). Verification tiers: register [22], The Circularity Trap.

The record — twelve cycles of the ask, in CMS’s own words (CY2016–CY2027)

Every quote below is from the final Rate Announcement for the year shown — each linked to the published PDF, opening at the first cited page; all twelve links verified live by this console (Aug. 28, 2026), and the CY2020/CY2027 quotes re-checked verbatim against the fetched PDFs. The request has been on the administrative record continuously: a minimum benchmark since CY2016, an explicit “AGA floor” since CY2019, the specific 0.70 figure since CY2020 — restated every year, adopted in none, and never once answered on the terms of §1876(a)(4)’s similar-area clause.

CycleThe ask (as summarized by CMS)CMS’s answerType
CY2016 pp. 22–23“establish a minimum benchmark level for the lowest cost counties in Puerto Rico… at 85% or 90% of the lowest average MA benchmarks among states”“we do not believe the approach suggested by these comments would be permissible under statute”Statutory only
CY2017 pp. 22, 25“establish a minimum benchmark level for counties in Puerto Rico”; “use another jurisdiction as a proxy”“the FFS data in Puerto Rico is sufficient for establishing accurate MA benchmarks as well as consistent with the statutory requirements”Statutory + data-sufficiency
CY2018 pp. 29–31no rebasing reductions; ESRD: “using the US Virgin Islands rate as a proxy”“We believe that the FFS data in Puerto Rico is sufficient for establishing accurate MA benchmarks”Data-sufficiency
CY2019 pp. 25–26“establishing an AGA floor, using a proxy benchmark, and/or applying a hold harmless minimum benchmark” — the first explicit “AGA floor”“We will continue to analyze these issues and consider whether any refinements to the methodology may be warranted in future years”Pure deferral
CY2020 pp. 29, 33the first explicit 0.70: “apply an AGA of 0.70 or apply a nationwide average AGA”; ESRD: “use the rates of US Virgin Islands”“the law requires that Medicare Advantage benchmarks be based on a county’s average Medicare FFS per-capita cost, and there is no evidence that FFS costs in Puerto Rico are higher”The dual formula debuts
CY2021 p. 35“establishing an AGA floor/proxy (e.g., applying an AGA of 0.70 or a nationwide average AGA)” — citing CMS’s own USVI GPCI proxy as FFS-side precedentsame dual formulaStatutory + evidentiary
CY2022 pp. 43–44“establishing an AGA minimum/floor (e.g., applying an AGA of 0.70)”; the benchmark is “39.6 percent below the national average”same dual formulaStatutory + evidentiary
CY2023 pp. 34–35“establishing an AGA minimum/floor (e.g., applying an AGA of 0.70) or applying a hold harmless minimum benchmark”same dual formulaStatutory + evidentiary
CY2024 pp. 53–55“establishing a minimum AGA of 0.70 for Puerto Rico” — with the full §1876(a)(4) theory: the statute “expressly permits the Secretary to use data from ‘a geographic area served by an eligible organization or . . . a similar area’”, USVI as the similar areaFFS floor precedents “are based on statutory provisions that are neither applicable to the MA program nor provide a direct analog to the provisions in section 1853” — the similar-area clause itself: unaddressedStatutory, sharpened; clause ignored
CY2025 pp. 60–63“A large number of commenters requested that we consider establishing a minimum AGA of 0.70 for Puerto Rico (similar to the AGA level in the US Virgin Islands)”“CMS has limited discretion to incorporate targeted adjustments or exceptions”; “we have not seen evidence to suggest that Medicare FFS costs in another jurisdiction are a reliable proxy”“Limited discretion” debuts
CY2026 pp. 52–53“establish a minimum AGA of 0.7 either as a national MA benchmark floor or specifically for Puerto Rico”same two formulationsStatutory + evidentiary
CY2027 pp. 46–47“use a proxy rate for Puerto Rico, establish a minimum floor benchmark (e.g., at USVI rates), and establish a minimum floor AGA of 0.7”“limited discretion to incorporate targeted adjustments or exceptions, such as applying floors for specific locales”Sharpest statutory form

What the record shows — three things. (1) The reasoning migrated. Bare statutory refusal (CY2016–17) → “the FFS data in Puerto Rico is sufficient” (CY2017–18) → pure deferral (CY2019) → the settled statute-plus-no-evidence dual formula (CY2020–24) → “limited discretion” (CY2025–27). Across those same years the agency described the identical authority as an affirmative “exercise in discretion” whenever it was granting Puerto Rico’s data adjustments (CY2020 p. 24, reprised verbatim CY2022 p. 41 and CY2023 p. 33) — and as “limited discretion” precisely when floors were pressed hardest. (2) The similar-area clause has never been construed. The full §1876(a)(4) theory was put to CMS squarely once — CY2024, USVI as the similar area — and the response answered the fee-for-service floor precedents while never quoting, construing, or even mentioning the clause. Twelve years of responses; zero engagement with the statutory text at the center of the request. (3) The ask went congressional. In February 2026 — inside the CY2027 comment window — Resident Commissioner Pablo José Hernández and 17 other Members pressed the USVI-level minimum (the letter, hosted on this site), and the Medicare Advantage Integrity Act of 2025 (H.R. 6031, introduced November 13, 2025, bipartisan) would legislate the floor CMS says it lacks discretion to adopt. Every refusal in the table rejects a floor; none forecloses the estimation-methodology path — exactly the lane the CY2028 outcome above is drafted in.

All quotes: CMS final Rate Announcements CY2016–CY2027, links above (congress.gov loads in a browser; blocks automated checks — bill existence confirmed via archive snapshot). The five-letter 2026 advocacy record, with the Governor’s and Rep. Murphy’s letters, lives on FY2028 Watch (“The paper trail”).

Re-verified against the primary files — August 30, 2026

The 5%-per-year step-down is exact and universal. For all 54 Puerto Rico IPPS hospitals with a complete record across CMS’s own rule tables, the applied wage index fell by exactly 5.0% each year — San Juan: 0.6134 (FY2024) → 0.5827 → 0.5536 → 0.5259 (FY2027), cumulative −14.26% (0.95³ − 1). Roughly 7.9 more years remain until the transition meets the computed index (≈0.35): the grind runs through ≈FY2035, at ≈−2.3% of payment per year (the index applies to the 62% labor-related share, 42 U.S.C. §1886(d)(3)(E)(ii)).

The AGA is already slipping. PR mean AGA: 0.50842 (CY2025) → 0.49596 (CY2026) → 0.49875 (CY2027) — 52 of 78 municipios lower in CY2027 than in CY2025; Vieques fell 5.1% in one year (0.60441 → 0.57359).

And each wage-index cut reaches the benchmarks one calendar year later. The AGA rests on five years of FFS claims (CY2027 ratebook: claims 2020–2024), but the base period is repriced to the latest Medicare FFS pricing parameters before the AGAs are computed — so the cuts flow through almost immediately, not diluted across the window. The mapping (per HDI actuarial review, Sept. 2026): the FY2025 −5% reached the CY2026 benchmarks; the FY2026 −9.75% cumulative reached CY2027; the FY2027 −14.26% will be fully carried by the CY2028 benchmarks, and each further 5% step lands the following year through the early 2030s. That is the arithmetic case for acting in the CY2028 Advance Notice window — the cut it will carry is already locked in.

CMS is on notice in its own record. The CY2027 Rate Announcement (p. 46) logs the warning: “the annual changes in FFS wage index, including the changes to the low wage index policy, will contribute to declines in the MA rates for Puerto Rico.” CMS’s stated answer — the five-year average — delays the decline; it does not prevent it.

Verified 2026-08-30, hospital-by-hospital and municipio-by-municipio, from: FY2025 IPPS tables incl. CMS-1808-IFC · FY2026 tables (CMS-1833-F) · FY2027 tables (CMS-1849-F, linked above) · CY2025/CY2026/CY2027 Rate Calculation Data · 42 CFR §412.64(h)(7) · CMS Medicare Monthly Enrollment API (March 2026 pinned row unchanged: 27,275 of 705,363, 3.87%; newest month May 2026: 26,497 of 706,629, 3.75% — retrieved Sept. 1, 2026).

Proven — four PR-specific actions CMS already took without Congress

  1. CY2012 — the A+B-only claims base (proposed February 2011; fully implemented CY2014). Because PR beneficiaries must opt in to Part B — unlike automatic mainland enrollment — CMS restricted PR’s FFS rate calculation to beneficiaries with both Part A and Part B, finding their standardized costs “on average 5% higher” (CY2012 Advance Notice, pp. 13–14). PR is the only jurisdiction measured this way; the convention is printed in the FFS data files themselves.
  2. CY2017 — the zero-claims adjustment. “The Secretary has directed the Office of the Actuary to adjust the fee-for-service experience for beneficiaries enrolled in Puerto Rico to reflect the propensity of zero dollar claimants nationwide” (CY2017 Rate Announcement, pp. 24–25) — 13.9% of PR’s A+B FFS beneficiaries generate zero claims vs 6.1% nationally (CY2027 RA), so OACT reweights the cohort to the national proportion. Worth +4.1% to +4.7% every year since (CY2027: +4.4%), a Trust Fund impact CMS put at ≈$320M/yr (CY2023). Re-adopted annually — never made permanent.
  3. CY2018 — the double-bonus reinterpretation. CMS “reevaluated our interpretation of Section 1853(o)(3)(B)” (CY2018 RA, p. 3), making 52 of 78 PR municipios double-bonus counties for star-rated plans — zero exist in USVI or Guam.
  4. CY2020–CY2025 — a territory-tailored CMMI model. Puerto Rico participated in the MA VBID model for six years (Triple-S/GuideWell and MCS/MHH among participants), including a territory-specific design accommodation: socioeconomic targeting was defined, “in US territories,” by dual-eligible status rather than LIS — because the LIS statute excludes the territories.

These are the proof of concept: when CMS concluded PR’s data misrepresented PR’s costs, it adjusted the data — administratively, under existing estimation authority. One warning flag: the CY2026 Rate Announcement added that CMS will “reassess the need for ongoing special adjustments” — the existing concessions are re-adopted one year at a time, and keeping them is itself a standing agenda item for every comment cycle.

Available now — administrative actions for the CY2028 cycle

  1. Fix the standardization artifact (the largest lever). The AGA’s division by the FFS risk score converts PR’s and USVI’s near-identical raw costs (≈48% vs ≈49% of national, unweighted) into a 24% payment gap. CMS can adopt a representativeness or coding-completeness adjustment to the standardization step for jurisdictions whose FFS remnant is a small, non-representative residue of the population being priced — the exact species of correction as the CY2017 zero-claims adjustment. CMS’s own words establish the authority, written about Puerto Rico itself: “Our exercise in discretion for the data used to develop the estimate for one geographic area, based on circumstances unique to that area, illustrates how there is more than one way to develop a reasonable and reliable adjusted average per capita cost estimate for purposes of the MA statute” (CY2020 RA, p. 24; repeated verbatim CY2022 p. 41, CY2023 p. 33 — quote re-verified against the PDF by this console, Aug. 28, 2026).
  2. Consolidate Puerto Rico into a single island-wide payment area — a dormant statutory right, never used by anyone. Under SSA §1853(d)(4), 42 U.S.C. §1395w-23(d)(4) (regulation at 42 CFR §422.304(e)), upon written request of a State’s chief executive filed by February 1, the Secretary “shall make a geographic adjustment… to a single statewide Medicare+Choice payment area.” “State” expressly includes Puerto Rico — the Governor qualifies, and CMS’s 1998 rule confirms approval is not discretionary (“Section 1853(d)(3) directs us to approve a Governor’s request,” 63 FR 35004). CMS itself noted in 2005 that “no State has availed itself of this option since its enactment in 1998” (70 FR 4588). Guam is already priced island-wide; Puerto Rico is priced from 78 municipio cells, 69 below CMS’s own 1,000-enrollee credibility line. Honest scope note: the adjustment is budget-neutral within the state — it repairs fragmentation, small-cell volatility, and suppression, not the aggregate level. It is the structural cleanup; item 1 is the money.
  3. Use the “similar area” estimation authority already in the statute. §1876(a)(4) lets the Secretary estimate per-capita cost “in a geographic area served by an eligible organization or in a similar area, with appropriate adjustments to assure actuarial equivalence.” Where a ≈27,000-person remnant cannot support a credible estimate, the statute itself names the alternative CMS says it lacks. The clause is genuinely unaddressed ground: in twelve years of responses CMS has never quoted or construed it — including CY2024, the one cycle in which commenters presented it squarely.
  4. Design a successor territory model at CMMI. §1115A directs the Innovation Center toward models addressing “a defined population for which there are deficits in care” — Puerto Rico’s 300,000+ federally invisible duals are such a population, and the precedent exists: PR participated in VBID for six years with a territory-specific accommodation. VBID was terminated for excess Trust Fund costs, so any successor must be structured to survive §1115A’s cost test — but the authority, and the PR precedent, are on the books.
  5. The calendar. The CY2028 Advance Notice is due by early February 2027 (realistically late January: the CY2027 notice landed January 26) with a statutory 30-day comment window — roughly late January through early March 2027; the final CY2028 Rate Announcement lands by Monday, April 5, 2027. That window is where items 1 and 3 must be argued; a §1853(d)(4) payment-area request for CY2028 must be filed by February 1, 2027.

Requires Congress — only for permanence; the CY2028 outcome does not wait for it

Read the refusal record correctly: CMS closed the floor door — it never closed the estimation door. What CMS refused, twelve consecutive cycles: a benchmark floor labeled a floor, USVI-parity by fiat, cross-jurisdiction proxy rates — most recently CY2027 (pp. 46–47): “section 1853 of the Act prescribes the general approach that FFS per capita costs be used in developing MA rates and CMS has limited discretion to incorporate targeted adjustments or exceptions, such as applying floors for specific locales… we have not seen evidence to suggest that Medicare FFS costs in another jurisdiction are a reliable proxy.” What CMS never refused: similar-area estimation under §1876(a)(4) — never quoted, never construed, never answered, even when squarely presented (CY2024). The tension between “limited discretion” (when floors are pressed) and “exercise in discretion” (when data fixes are granted) is not a contradiction to lament — it is a map. It marks where the authority lives: the estimation inputs. The Outcome panel’s 0.70 stabilization is drafted exactly there, and its evidentiary record answers the one thing CMS says it has not seen.

What Congress adds — permanence, and the parts no agency can reach: a statutory territory benchmark floor (already pending: H.R. 6031, the Medicare Advantage Integrity Act of 2025, introduced November 13, 2025); extension of the Medicare Savings Programs and the Part D Low-Income Subsidy to Puerto Rico (the statutory root of the duals invisibility and the $0.00 LIS column documented on this console); and Medicaid financing parity. The sequence is the strategy: CMS stabilizes in CY2028 under authority it already holds; Congress makes it permanent. Neither needs to wait for the other.

Methodology & primary sources for this tab: CMS Rate Calculation Data and Ratebooks CY2018–CY2027 · CMS FFS county claims files CY2016–CY2024 · CMS Advance Notices and Rate Announcements CY2015–CY2027 (all twelve final-announcement links re-verified live Aug. 28, 2026; CY2020 and CY2027 quotes re-checked verbatim against the fetched PDFs) · CMS Medicare Monthly Enrollment (data.cms.gov) · U.S. Census Bureau · MedPAC March 2017 Report ch. 13 · GAO-06-75 (gao.gov; loads in a browser, blocks automated checks). Companion-console figures ported at content parity; the Risk2027 engine-room figures (USPCC $1,297.74, AGA levels, caps, the 0.70 model) were re-run live on the published files by this console — provenance and verification tiers in register [22], The Circularity Trap. PR FFS-share updated to this console’s live value (3.75%, May 2026, retrieved Sept. 1, 2026).

Provenance & reproducibility

Data & sources

Where every number comes from and how it is built. Lineage maps each CMS source to the warehouse table to the finding it supports; Inventory lists all 58 source files with origin URLs and SHA-256 checksums; SQL holds the reproducible DuckDB queries; Architecture documents the five-phase pipeline.

CMS source → warehouse table → finding

Data lineage

Click a category to drill in. Click a node to inspect what it is and where it lives.

View · Filter ·
All sources
Tap any node to inspect.

Every source file traced to origin

Source inventory

SHA-256 fingerprint and CMS URL for every artifact.

Year Category File Size Source

What's in the warehouse

TableRowsDescription

DuckDB · read-only · reproducible

SQL lab

Copy these into DuckDB to reproduce every finding.

PR vs US mainland average rate by year
SELECT report_year, AVG(CASE WHEN state = 'PR' THEN payment_rate END) AS pr_avg, AVG(CASE WHEN state NOT IN ('PR','VI') THEN payment_rate END) AS us_avg FROM ma_ratebooks_county WHERE rate_type = 'county_capitation' AND bonus_tier = '0%' GROUP BY report_year ORDER BY report_year;
FFS per-capita PR vs US
SELECT report_year, CASE WHEN state IN ('PR','PUERTO RICO') THEN 'PR' ELSE 'US' END AS region, AVG(TRY_CAST(part_a_total_per_capita AS DOUBLE) + TRY_CAST(part_b_total_per_capita AS DOUBLE)) AS ffs_pcm FROM ma_ffs_expenditure WHERE source_sheet IN ('FFS18','FFS19','FFS20','FFS21','FFS22','FFS23','FFS24') GROUP BY report_year, region ORDER BY report_year;
PR market HHI by year
WITH shares AS ( SELECT report_year, contract_number, SUM(enrollment_num) / SUM(SUM(enrollment_num)) OVER (PARTITION BY report_year) AS share FROM ma_enrollment_by_cpsc WHERE state_code = 'PR' -- one snapshot per year: the latest month available -- (2025 was published as April; others March/May) AND (report_year, report_month) IN ( SELECT report_year, MAX(report_month) FROM ma_enrollment_by_cpsc WHERE state_code = 'PR' GROUP BY report_year) GROUP BY report_year, contract_number ) SELECT report_year, SUM(POWER(share*100, 2)) AS hhi FROM shares GROUP BY report_year ORDER BY report_year;

Five phases · drift-tolerant by design

Pipeline architecture

From CMS discovery to DuckDB warehouse to this console.

PhaseModuleWhat it does
Iphase1_discover_and_download Crawls CMS landing pages from seeds; resolves links, sub-pages, year-stamped ZIPs; downloads with retry/cache; records SHA-256.
IIphase2_extract_and_parse Universal extraction — ZIP/CSV/XLSX/XLS/PDF/JSON. Year-independent column matching; header detection; numeric coercion; state-code normalization.
IIIphase3_build_database Loads parquet outputs into DuckDB. Mixed parquet+CSV directory support. Builds analytical views.
IVphase4_quality_control Validates row counts, year ranges, null distributions; produces QC report.
Vphase5_dashboard Generates this console and its JSON payload.

Drift-tolerant parsing

CMS schemas evolve every year (2018 uses "AL"; 2025 uses "ALABAMA"; bonus columns embed the year in their header). The parser handles this through year-independent column signatures, header-row scoring, bidirectional state normalization, and tolerant numeric coercion.