In December 2025, the CMS Innovation Center released proposed rules for two models—GLOBE (Global Benchmark for Efficient Drug Pricing) for Medicare Part B and GUARD (Guarding U.S. Medicare Against Rising Drug Costs) for Medicare Part D.
While the programmes target different benefit designs (Part B vs Part D), their shared mechanism is to compute an international benchmark from “economically comparable” countries and use it as an alternative reference point in calculating incremental manufacturer liability beyond the inflation rebate baseline.
The MFN “math” is not fundamentally different. The real differences sit in coverage scope and implementation mechanics—Part B’s ASP-based provider payments versus Part D’s rebate-driven net price environment.
What GLOBE and GUARD have in common
Both models use the same reference country basket and benchmark methodology—the differences lie in how that benchmark is applied within each Medicare benefit.
1. Reference country selection
CMS proposes a systematic methodology for identifying the 19 reference countries used in both GLOBE and GUARD. The aim is to select countries that are economically similar to the United States, have reasonably comparable purchasing power, and generally have existing international drug pricing information available.
To qualify as a reference country, a non-U.S. OECD member must meet both criteria (based on CIA World Factbook data as of October 1, 2025):
Based on these criteria, CMS has identified 19 reference countries from the OECD membership which would remain stable throughout the 5-year model period.
The 19 GLOBE and GUARD Reference Countries
Click on a country marker to see its GDP data, PPP adjuster, and how they use international reference pricing
Click on a blue marker to view country details
Notably, the reference countries themselves vary widely in their use of international reference pricing. As the map data shows, some operate formal IRP systems that directly set or cap prices based on other countries, others use international prices informally as benchmarks during negotiations, and some do not use IRP at all.
Circular referencing risk: Because many of the 19 reference countries already reference each other’s prices, the U.S. benchmark will be indirectly shaped by a web of interconnected pricing decisions. A price change in one country can ripple through others—and now, potentially, back to the U.S. This creates feedback loops that manufacturers will need to model carefully when forecasting exposure.
2. The shared MFN benchmark architecture
CMS proposes a two-track benchmark design in both GLOBE (Part B) and GUARD (Part D). Manufacturers can choose between a default benchmark or submit additional data for a potentially higher benchmark.
CMS derives a benchmark from publicly available international pricing data (ex-manufacturer/list prices), applies PPP adjustments, and holds this benchmark constant once set.
Manufacturers may voluntarily submit international net pricing data. CMS calculates a volume-weighted, PPP-adjusted benchmark that can be updated periodically.
Both methods rely on the same GDP (PPP) adjuster to normalise prices across countries with different economic resources. CMS defines this adjuster as U.S. GDP (PPP) per capita divided by the reference country’s GDP (PPP) per capita—but with a critical design feature: the adjuster has a floor of 1.000. This means that prices from countries with higher per-capita GDP than the U.S. (such as Switzerland, Norway, or Ireland) are never adjusted downward. For a deeper look at the PPP adjustment mechanics and other regulatory nuances, see our analysis of hidden details in GLOBE & GUARD.
How GLOBE and GUARD differ in practice
GLOBE and GUARD share the same MFN benchmark backbone, but they diverge on the questions that matter operationally: which drugs are pulled in and how the benchmark is applied inside each benefit.
1) Drug coverage and selection
| USP DC Category | Associated Therapeutic Area(s)* |
|---|---|
| Antigout Agents | Rheumatology |
| Antineoplastics | Oncology |
| Blood Products and Modifiers | Immunology, Oncology |
| Central Nervous System Agents | Immunology |
| Immunological Agents | Immunology, Rheumatology, Oncology |
| Metabolic Bone Disease Agents | Endocrinology, Rheumatology |
| Ophthalmic Agents | Ophthalmology |
*This is not an exhaustive list of therapeutic areas that correspond to the USP DC categories but illustrates how the named USP DC categories are associated with the 5 therapeutic areas of oncology, endocrinology, immunology, rheumatology, and ophthalmology.
Why this matters for portfolio planning: even when the MFN benchmark method is similar, inclusion depends on the benefit and its inflation rebate perimeter. A product can be “MFN-exposed” in Part B, Part D, both, or neither—depending on utilisation channel mix, exclusions (especially IRA MFP), and whether an international analogue can be cleanly mapped to the US unit basis.
2) Implementation mechanics and rebate structure
Once a drug qualifies, the models diverge most sharply in how MFN is wired into the payment system. Part B is ASP- and billing-unit driven; Part D is a plan/PBM net-price environment where net price is reconstructed from WAC and reported concessions.
GLOBE mechanics (Part B)
GUARD mechanics (Part D)
How GLOBE, GUARD, and GENEROUS Compare
These proposals are part of a broader CMS push toward international reference pricing. GENEROUS—a voluntary Medicaid model already active since January 2026—tests the concept through supplemental rebates.
| GLOBE | GUARD | GENEROUS | |
|---|---|---|---|
| Stage | Proposed (comments invited by Feb 23, 2026) | Active (enrolment underway) | |
| Programme | Medicare Part B | Medicare Part D | Medicaid |
| Participation | Mandatory (for qualifying drugs) | Voluntary (manufacturers + states) | |
| Geographic scope | Randomised geographies (ZCTAs) | National framework (state opt-in) | |
| Reference countries | 19 OECD countries (AUS, AUT, BEL, CAN, CZE, DNK, FRA, DEU, IRL, ISR, ITA, JPN, NLD, NOR, KOR, ESP, SWE, CHE, GBR) | 8 countries (CAN, DNK, FRA, DEU, ITA, JPN, CHE, GBR) | |
| Country criteria | OECD + per capita GDP (PPP) ≥60% of U.S. + aggregate GDP (PPP) ≥$400bn | Fixed 8-country set | |
| Benchmark method | PPP-adjusted; two pathways (lowest list price or volume-weighted net); higher benchmark used | PPP-adjusted; second-lowest net price in 8-country set | |
| Start date | Oct 1, 2026 | Jan 1, 2027 | Jan 1, 2026 |
| Duration | 5-year performance period (reconciliation to 2033) | 5 years (to Dec 2030) | |
The Bottom Line
GLOBE and GUARD are best seen as a paired Medicare experiment: one MFN concept, executed through two different programme architectures. For pricing and market access teams, the technical question isn’t only the international benchmark—it’s how eligibility, unit definitions, and the Part B vs Part D data environment shape the size and timing of incremental liability.
The comment period closed on February 23, 2026. Until CMS publishes final rules, manufacturers should treat the proposals as a near-term planning scenario and be ready to update assumptions as the details are finalized.
Contact Symaptics to discuss strategic implications and scenario modelling for your portfolio.
Related reading
Need structured MFN scenario modelling?
MFN-style policies are increasingly technical: benchmarks, unit definitions, timing, and programme interactions matter. A transparent modelling approach helps keep assumptions consistent and audit-ready across teams.