Rating Actions & Surveillance Policy

How MARA Rating Company maintains, reviews, updates, and publicly accounts for its ratings — including the ones that were wrong.

Purpose

A rating issued once and never revisited is a snapshot, not a signal. This policy describes how MARA ratings are kept current when evidence changes, how the method behind them is version-controlled, and how the full performance record — successes and failures alike — is disclosed every year. The objective is a risk signal that stays stable, comparable, and defensible over time, and a track record anyone can check.

Rating Action Types

• Affirmed — rating confirmed after review; no change in band.
• Upgrade — rating moved to a higher band following materially improved evidence or reimbursement outlook.
• Downgrade — rating moved to a lower band following materially adverse evidence or reimbursement developments.
• Under Review — rating placed under review pending evaluation of a material event.
• Withdrawn — rating withdrawn only if minimum evidence requirements are no longer met, or the rating is more than five years old. Withdrawal is never used to remove an unfavourable outcome from the record.

Surveillance Framework

Every MARA rating is subject to:

• Event-driven surveillance — triggered by material clinical, economic, or policy developments.
• Periodic review — at least annually, to confirm continued alignment with the evidence and the HTA context.

Material Event Triggers (Non-Exhaustive)

• New pivotal clinical data
• New HTA decisions or payer policy updates
• Label expansions or restrictions
• Material pricing or reimbursement developments
• Significant safety findings
• Major changes in standard of care
• Documented methodological corrections

Process & Governance Controls

• Ratings are determined through structured committee judgment. No single analyst determines outcomes.
• Evidence is restricted to public, verifiable sources.
• Committee minutes are recorded.
• Issuers may correct factual errors, but cannot direct rating outcomes.

Method Versioning & Change Control

A rating is only comparable over time if the method behind it changes in the open. MARA’s methodology carries a version number (currently v1.2.1, February 2026), and every change follows the same documented path:

• Scheduled review. The Expert Committee reviews the prior year’s performance each April. Recalibration is triggered by pre-defined thresholds — a false-positive rate above 3% or overall accuracy below 85% — not by discretion.
• Versioned publication. Any change to domain definitions, weighting logic, or threshold bands produces a new methodology version, dated and published. The version in force at the time of each rating is recorded with that rating.
• No silent revision. The method is never adjusted retroactively, and never at the request of a rated company. A rating issued under v1.2.1 remains a v1.2.1 rating; it is not restated under a later method to improve its apparent accuracy.

Ratings are kept current under the surveillance rules above. The method is kept current under this section. The two are governed separately because they answer different questions: whether the evidence changed, and whether the yardstick changed.

Validation Disclosure Policy

A rating that is only shown when it was right cannot be tested. MARA publishes the full performance record of its ratings once a year, on a fixed schedule, in a fixed format, regardless of what that record shows.

What is disclosed

• Annual Results Report — published every June, covering the prior calendar year in full.
• Year-level performance metrics — accuracy, false-positive rate, false-negative rate, and Area Under the ROC Curve (AUC).
• Every rating issued in the reporting period. No rating is excluded from the calculation because its outcome was unfavourable to MARA.

How a miss is defined

• False positive — a rating of A or higher assigned to an asset that did not go on to secure reimbursement. This is the costliest error: capital committed on the strength of a signal that did not hold.
• False negative — a rating below the A threshold assigned to an asset that did secure reimbursement. A missed upside rather than a capital loss — disclosed on the same basis.

Both are counted. Neither is omitted from the published figures to improve them.

Five-Year Track Record (current rule set)

Across 2020–2025, MARA’s cumulative precision against NICE decisions is 97%: 264 of 272 assets rated A or above went on to achieve reimbursement. The remaining 8 are false positives — a cumulative false-positive rate of 2.9% — disclosed in the table by year, not removed from it.

The table also shows the two years in which MARA’s own recalibration threshold was breached: 2021 and 2024, when the false-positive rate exceeded 3%. Those years triggered recalibration under the Methodology. They were not edited out of the record.

As calibration against G-BA, HAS, and ICER reaches the same sample depth as the NICE dataset, this disclosure standard extends to those markets on the same fixed schedule.

What this policy does not permit

• Ratings are not withdrawn to remove an unfavourable outcome from the record. Withdrawal is limited to the conditions defined under Rating Action Types above.
• The reporting period, the metrics disclosed, and the recalibration thresholds are fixed in advance and are not adjusted after a result is known.
• A published rating remains on the public MARA Ratings List after issuance, whether or not the subsequent HTA decision matched it.

Why this is disclosed, not summarised

An investment committee, a board, or a lender does not need to be told that MARA is accurate. It needs to be able to check. A track record that reports only its successes is a claim. A track record that discloses every miss, on a schedule the issuer does not control, is evidence.

Relationship to the Independent Market-Access Check Standard

MARA has published a standard describing what any independent market-access check should include — six criteria covering independence, consistency of method, historical testing, inspectable reasoning, an openly published track record, and documented currency. The standard is provider-neutral: any assessment meeting the six criteria satisfies it.

This page documents how MARA holds itself to two of those criteria:

• A track record published openly, including the mistakes — governed by the Validation Disclosure Policy above.
• Kept current, under documented rules — governed by the Surveillance Framework, Material Event Triggers, and Method Versioning & Change Control sections above.

The remaining criteria — independence, consistency of method, historical calibration, and evidence-linked reasoning — are documented in the MARA Rating Methodology and the Rating Committee Charter.