Additionality and permanence — the two tests every credit must pass

1. Definitions and standards basis

Additionality and permanence are the two integrity tests that determine whether a tonne of CO₂e claimed as offset is real. They are defined separately in the methodologies, but in practice a credit failing either is non-fungible with one that passes both.

Additionality is defined under the Verra VCS Standard v4.7 §3.13 as the requirement that a project activity is not the baseline scenario — i.e. the GHG reductions or removals would not have occurred in the absence of carbon finance. The test is operationalised through the VT0001 Tool for the Demonstration and Assessment of Additionality (inherited from CDM EB39 Annex 10), which runs a four-step sequence: identification of alternatives, investment analysis, barrier analysis, and common practice analysis. Gold Standard applies an equivalent test under GS4GG Principles & Requirements §4.1. The ICVCM Core Carbon Principles (published March 2023, Assessment Framework v1.1) codify additionality as CCP 4 and require the methodology to demonstrate it conservatively.

Permanence is the requirement that sequestered carbon remain out of the atmosphere over a defined accounting horizon. VCS AFOLU projects use a minimum 40-year crediting period and a non-permanence risk buffer determined by the AFOLU Non-Permanence Risk Tool, with buffer contributions typically 10–25% of issued volume. The EU Carbon Removals and Carbon Farming Regulation (CRCF, Regulation (EU) 2024/3012, entered into force 26 December 2024) defines four activity classes — permanent removals, carbon farming, carbon farming removals, and carbon storage in products — with monitoring liability extending to at least 5 years for carbon farming and the full storage duration for permanent removals (≥ several centuries for DACCS/BECCS/mineralisation).

2. Mechanism in technical detail

Additionality is mechanically demonstrated through a financial counterfactual. For a project to clear VCS investment analysis under VT0001, the developer presents an IRR or NPV calculation excluding carbon revenue and shows it falls below a benchmark — typically the weighted average cost of capital, a government bond yield plus equity premium, or sectoral hurdle rates from the host country. Sensitivity analysis across ±10% of key parameters (capex, opex, output price) must not flip the result. Where investment analysis is unsuitable (e.g. zero-revenue projects), barrier analysis substitutes: institutional, technological, or local-ecological barriers must be documented with primary evidence.

Permanence accounting differs by carbon pool. For a REDD+ project under VM0007 or the consolidated VM0048 (Reducing Emissions from Deforestation, 2023), reversal risk is scored across internal (project management, financial viability), external (land tenure, political), and natural (fire, pest, climate) categories. The aggregated risk percentage is withheld as buffer credits in the AFOLU Pooled Buffer Account. Reversals trigger automatic cancellation from the buffer; if the buffer is exhausted, the project loses crediting eligibility.

For engineered removals (DACCS, enhanced weathering, biochar under VM0044), permanence is accounted through monitored storage media — geological reservoirs with leakage monitoring per ISO 27914, or stable mineral phases. Biochar uses a decay function based on H/C_org molar ratio; values below 0.4 yield a 100-year permanence factor of ~0.8 under EBC and IPCC 2019 Refinement guidance.

3. Worked example

Consider Verra project ID VCS 2569 — Katingan Mentaya peatland REDD+, Indonesia, methodology VM0007 REDD+MF v1.6. Vintage 2019 issuance: ~5.4 Mt CO₂e gross, with a non-permanence buffer withholding of approximately 19.4% per the AFOLU risk tool. A typical retired serial range looks like 10000-484929501-484929700-VCU-039-MER-ID-14-2569-01012019-31122019-0, decomposing as registry-issuance ID, unit block start/end, unit type (VCU), country code, project ID, vintage period, and crediting period number.

Additionality for VCS 2569 rests on the legal baseline: the concession was zoned for industrial acacia plantation under Indonesian forestry law, with comparable concessions converted at documented rates. Investment analysis showed plantation IRR exceeding the carbon-financed conservation IRR absent VCU revenue. Common practice analysis referenced regional concession conversion rates of ~70% over the prior decade.

4. Irish-specific considerations

Ireland has effectively no domestic compliance market for voluntary credits — the EU ETS covers ~100 installations and aviation, but offsets are not fungible with EUAs since Phase 3. Demand-side, Irish corporates report under the Climate Action Act 2021, the EU CSRD (transposed via SI 336/2024), and increasingly under SBTi targets which restrict offsets to residual Scope 1/3 emissions and require like-for-like removal credits from 2030 under the Net-Zero Standard v2.

Supply-side, Irish-origin credits are scarce. The Department of Agriculture's Land Use Review Phase 2 and Bord na Móna's peatland rehabilitation programme (~33,000 ha) could in principle generate credits under VM0036 (Rewetting Drained Temperate Peatlands) or the EU CRCF carbon-farming pathway, but no Irish-issued VCU or GS-VER currently trades in volume. The CRCF certification methodologies for peatland rewetting and agroforestry are still in development by the Commission's expert group as of 2025.

5. IMPT position

IMPT retires one tonne CO₂e per hotel booking, funded from booking commission rather than passed to the guest. We retire only credits from methodologies that pass current ICVCM CCP assessment or that we have independently mapped against CCPs 4 (additionality), 5 (permanence) and 6 (robust quantification). The retirement is executed on-chain against a tokenised representation of a Verra VCU or Gold Standard VER, with the underlying serial number written to the retirement event and resolvable through the originating registry account. Buyers receive the registry-side serial — not just the on-chain hash — so audit traces back to the source registry, not to us.

We weight the portfolio toward credits with explicit buffer pool contributions, vintage ≤ 5 years, and methodologies revised after 2020. Pre-2016 vintages and unbuffered avoidance credits are excluded.

6. Common mistakes by Irish buyers

• Treating avoidance and removal as fungible. A VM0007 avoidance VCU and a VM0044 biochar removal VCU are not interchangeable under SBTi Net-Zero or under CRCF accounting.
• Ignoring vintage drift. Buying 2012-vintage renewables credits to offset 2024 emissions fails the GHG Protocol Scope 2 Quality Criteria and most CSRD assurance tests.
• No serial-number verification. Buyers accept a PDF certificate without confirming the serial in the Verra or Gold Standard public registry shows status Retired against their account.
• Double counting under Article 6. Purchasing credits from a host country that has not issued a corresponding adjustment letter risks the same tonne being claimed in the host NDC.
• Conflating insetting with offsetting. Supply-chain insetting under GHG Protocol Scope 3 Category 1 is not a retirement event and should not be reported as an offset.

7. Verification and audit pattern

The integrity chain runs: project developer → VVB (ISO 14065-accredited, e.g. SCS Global, Aster, EPIC) → registry. Validation under ISO 14064-2 confirms the project design; verification confirms ex-post emission reductions against the monitoring report. Audit evidence a buyer should hold:

• Project Description (PD) and most recent Monitoring Report
• Validation and Verification Reports from the VVB, with VVB accreditation scope
• Issuance record with vintage and methodology version
• Retirement record with serial range, retirement date, and beneficiary
• For removals: buffer pool contribution percentage and reversal monitoring schedule

For IMPT-retired credits, the on-chain retirement transaction hash plus the registry serial number provides a two-sided proof: cryptographic on the token side, registry-authoritative on the credit side. Either alone is insufficient — only the pair closes the audit loop.