Like-for-Like in NDCs
On-going and preliminary results - Feedback welcome !
Embedded Files
From Emissions to Equivalence: The Missing Logic of ‘Like for Like’ in National Climate Mitigation Policy
From Emissions to Equivalence: The Missing Logic of ‘Like for Like’ in National Climate Mitigation Policy
Ongoing work with Injy Johnstone.
Context
Context
Current emissions accounting frameworks treat greenhouse gases as fungible through CO₂e metrics, yet emerging climate science demonstrates the need for a “like for like” approach that distinguishes between fossil and biogenic carbon in achieving net zero.
The like for like approach describes how fossil CO₂ functions as a long-lived stock pollutant requiring equally durable geological carbon removal to be neutralised in contrast to biogenic carbon which can be neutralised using biogenic carbon removal.
Drawing a distinction between the two is an essential foundation for credible greenhouse gas accounting. While like for like has begun to be applied at the corporate level it is currently missing from national climate mitigation policies. As a result critical differences between fossil and biogenic emissions pathway, risks and opportunities for neutralisation pose a risk of misaligning net zero outcomes with actual climatic impacts.
Approach
Approach
Desk research related to the scientific underpinnings and practical applications of the like for like principle.
Assessment of latest national inventory data of Annex One countries, identifying volumes of fossil vs biogenic (LULUCF) emissions, and trends over time to identify geologic offset ratio (unavailable yet on this webpage).
LLM enabled assessment of NDCs to identify whether countries are starting to integrate like for like in the form of :
The identification of residual emissions and hard to abate sectors.
The matching of these sectors to types of removals.
Additionally we also look at how NDCs reflect the intent to use Article 6 or not, and the intent to develop or invest in CDR.
Explore the raw results of the LLM NDC analysis below :
Explore the raw results of the LLM NDC analysis below :
Key result : only 3 countries apply a like-for-like matching of hard-to-abate sectors with geologic CDR.
Key result : only 3 countries apply a like-for-like matching of hard-to-abate sectors with geologic CDR.
The NDCs are currently consistent with a growing CDR gap.
Method
Method
Extraction of NDC text snippets based on keyword search
Extraction of NDC text snippets based on keyword search
Before sending the NDCs to the Large Language Model (LLM), we selected text snippets based on keywords classified in 4 themes: Article 6, biogenic emissions, removals, and carbon accounting. The keywords are available in the Python script below.
# ── keywords.py ── Shared keyword lists for NDC analysis# Import in any analysis notebook: from keywords import ARTICLE6, BIOGENIC, ...
ARTICLE6 = [ # English "article 6", "art. 6", "art 6", "cooperative approach", #"cooperative approaches", "corresponding adjustment", #"corresponding adjustments", "itmo", "internationally transferred mitigation outcome", # French "article 6", "approche coopérative", "ajustement correspondant", "ajustements correspondants", # Spanish "artículo 6", "enfoque cooperativo", "ajuste correspondiente",]
BIOGENIC = [ # English "biogenic emission", #"biogenic emissions", "biologic emission", #"biologic emissions", "land sector emission", "land-sector emission", "afolu", "land use emission", "land-use emission", "LULUCF", "agriculture forestry and other land use", "agriculture, forestry and other land use", "methane emission", #"methane emissions", "livestock emission", #"livestock emissions", "land-use carbon flux", "land use carbon flux", # French "émission biogénique", "émissions biogéniques", "émission biologique", "émissions biologiques", "secteur des terres", "émissions de méthane", "agriculture forêts et autres", # Spanish "emisión biogénica", "emisiones biogénicas", "emisiones del sector tierra", "metano", "ganadería",]
REMOVALS = [ # English "removal", "cdr", #carbon dioxide removal "negative emission", "removal", #"engineered removal", "engineered removals", #"nature-based removal", "nature based removal", #"removals durability", #"removals permanence", #"removals accounting", #"permanent removal", "permanent removals", "sequestration", #"geological sequestration", #"biological sequestration", "daccs", "direct air capture", "beccs", "bioenergy with carbon capture", "carbon sink",
# French "élimination",# du dioxyde de carbone", "élimination du CO2", "élimination du carbone", "absorption", # du CO2", "absorption du carbone", "absorption de carbone", "absorption de CO2", "absorptions nettes", "émission négative", "émissions négatives", "séquestration", #"séquestration géologique", #"séquestration biologique", "captage et stockage", "puits de carbone", #"éliminations permanentes", # Spanish "eliminación de dióxido de carbono", "emisión negativa", "emisiones negativas", "secuestro", #"secuestro geológico", #"secuestro biológico", "sumideros de carbono",]
ACCOUNTING_like4like = [ # English "hard-to-abate", "hard to abate", "like for like", "like-for-like", "emissions balancing", "emission balancing", "mitigation hierarchy", "residual emission", "residual emissions", # French "similaire pour similaire", "équivalence des émissions", "hiérarchie d'atténuation", "émissions résiduelles", # Spanish "equivalencia de emisiones", "jerarquía de mitigación", "emisiones residuales"]
ACCOUNTING_general = [ # English "neutralisation", "neutralization", "offsetting", "carbon offset", "net zero", "carbon neutrality", "climate neutrality", # French "neutralisation", "compensation carbone", "neutralité carbone", # Spanish "neutralización", "compensación de carbono", "neutralidad climática"]
# ── Combined dictionary for easy iteration ────────────────────────────────ALL_THEMES = { "article6": ARTICLE6, "biogenic": BIOGENIC, "removals": REMOVALS, "accounting_like4like": ACCOUNTING_like4like, "accounting_general": ACCOUNTING_general,}
ALL_KW = ARTICLE6 + BIOGENIC + REMOVALS + ACCOUNTING_like4like + ACCOUNTING_general
ARTICLE6 = [ # English "article 6", "art. 6", "art 6", "cooperative approach", #"cooperative approaches", "corresponding adjustment", #"corresponding adjustments", "itmo", "internationally transferred mitigation outcome", # French "article 6", "approche coopérative", "ajustement correspondant", "ajustements correspondants", # Spanish "artículo 6", "enfoque cooperativo", "ajuste correspondiente",]
BIOGENIC = [ # English "biogenic emission", #"biogenic emissions", "biologic emission", #"biologic emissions", "land sector emission", "land-sector emission", "afolu", "land use emission", "land-use emission", "LULUCF", "agriculture forestry and other land use", "agriculture, forestry and other land use", "methane emission", #"methane emissions", "livestock emission", #"livestock emissions", "land-use carbon flux", "land use carbon flux", # French "émission biogénique", "émissions biogéniques", "émission biologique", "émissions biologiques", "secteur des terres", "émissions de méthane", "agriculture forêts et autres", # Spanish "emisión biogénica", "emisiones biogénicas", "emisiones del sector tierra", "metano", "ganadería",]
REMOVALS = [ # English "removal", "cdr", #carbon dioxide removal "negative emission", "removal", #"engineered removal", "engineered removals", #"nature-based removal", "nature based removal", #"removals durability", #"removals permanence", #"removals accounting", #"permanent removal", "permanent removals", "sequestration", #"geological sequestration", #"biological sequestration", "daccs", "direct air capture", "beccs", "bioenergy with carbon capture", "carbon sink",
# French "élimination",# du dioxyde de carbone", "élimination du CO2", "élimination du carbone", "absorption", # du CO2", "absorption du carbone", "absorption de carbone", "absorption de CO2", "absorptions nettes", "émission négative", "émissions négatives", "séquestration", #"séquestration géologique", #"séquestration biologique", "captage et stockage", "puits de carbone", #"éliminations permanentes", # Spanish "eliminación de dióxido de carbono", "emisión negativa", "emisiones negativas", "secuestro", #"secuestro geológico", #"secuestro biológico", "sumideros de carbono",]
ACCOUNTING_like4like = [ # English "hard-to-abate", "hard to abate", "like for like", "like-for-like", "emissions balancing", "emission balancing", "mitigation hierarchy", "residual emission", "residual emissions", # French "similaire pour similaire", "équivalence des émissions", "hiérarchie d'atténuation", "émissions résiduelles", # Spanish "equivalencia de emisiones", "jerarquía de mitigación", "emisiones residuales"]
ACCOUNTING_general = [ # English "neutralisation", "neutralization", "offsetting", "carbon offset", "net zero", "carbon neutrality", "climate neutrality", # French "neutralisation", "compensation carbone", "neutralité carbone", # Spanish "neutralización", "compensación de carbono", "neutralidad climática"]
# ── Combined dictionary for easy iteration ────────────────────────────────ALL_THEMES = { "article6": ARTICLE6, "biogenic": BIOGENIC, "removals": REMOVALS, "accounting_like4like": ACCOUNTING_like4like, "accounting_general": ACCOUNTING_general,}
ALL_KW = ARTICLE6 + BIOGENIC + REMOVALS + ACCOUNTING_like4like + ACCOUNTING_general
LLM prompt to generate the output
LLM prompt to generate the output
With Claude Sonnet 4-6
SYSTEM = """You are an expert climate policy analyst specialising in the Paris Agreement, carbon markets, and carbon dioxide removal. You will receive excerpts from a country's NDC grouped by theme. Answer each question based only on the provided excerpts. Reply ONLY with valid JSON — no preamble, no markdown fences."""
PROMPT_TEMPLATE = """---
### INSTRUCTIONS
Read all chunks carefully before answering. Base your answers only on what is explicitly stated or clearly implied in the text. Do not infer or hallucinate. If a field cannot be determined from the text, use null. For every non-null value, populate the corresponding _evidence field with a verbatim quote from the source text (up to 50 words). Preserve the exact wording. If no direct quote is available, write null. For each theme, only use the text excerpts associated with that theme. If no excerpts are available, reply "not_mentioned" or null to all questions. ---
### OUTPUT FORMAT
Return only a valid JSON object. No explanation, no markdown, no preamble.
{ "country": "<country name>", "document_date": "<YYYY-MM or YYYY>",
"article6": { "intent": "<one of: yes | no | considering | not_mentioned>", "intent_evidence": "<verbatim quote up to 50 words or null>", "role_buyer": "<one of: yes | no | possible | null>", "role_buyer_evidence": "<verbatim quote up to 50 words or null>", "role_seller": "<one of: yes | no | possible | null>", "role_seller_evidence": "<verbatim quote up to 50 words or null>", "mechanisms_mentioned": ["<e.g. Article 6.2, Article 6.4, Article 6.8>"], "other_mechanisms_mentioned": ["<e.g. voluntary carbon market, JCM, CORSIA, etc>"], "notes": "<one sentence max or null>" },
"cdr": { "cdr_considered": "<one of: yes | no | not_mentioned>", "cdr_considered_evidence": "<verbatim quote up to 50 words or null>", "cdr_quantitative_target": "<one of: yes | no | not_mentioned>", "cdr_quantitative_target_evidence": "<verbatim quote up to 50 words or null>", "cdr_nature_based": "<one of: yes | no | not_mentioned>", "cdr_nature_based_types": ["<e.g. afforestation, blue carbon, soil carbon, REDD+, biochar>"], "cdr_nature_based_evidence": "<verbatim quote up to 50 words or null>", "cdr_engineered": "<one of: yes | no | not_mentioned>", "cdr_engineered_types": ["<e.g. DACCS, BECCS, enhanced weathering >"], "cdr_engineered_evidence": "<verbatim quote up to 50 words or null>", "domestic_cdr": "<one of: yes | no | not_mentioned>", "domestic_cdr_evidence": "<verbatim quote up to 50 words or null>", "international_cdr": "<one of: yes | no | not_mentioned>", "international_cdr_evidence": "<verbatim quote up to 50 words or null>", "permanence_addressed": "<one of: yes | no | not_mentioned>", "permanence_evidence": "<verbatim quote up to 50 words or null>", "notes": "<one sentence max or null>" },
"like_for_like": { "residual_emissions_identified": "<one of: yes | no | not_mentioned>", "residual_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned>", "residual_emissions_evidence": "<verbatim quote up to 50 words or null>", "hard_to_abate_sectors": ["<sector names explicitly identified>"], "hard_to_abate_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned>", "hard_to_abate_sector_type_biologic": "<one of: yes | no | not_mentioned>", "hard_to_abate_sector_type_geologic": "<one of: yes | no | not_mentioned>", "hard_to_abate_sectors_evidence": "<verbatim quote up to 50 words or null>", "offsetting_considered": "<one of: yes | no | not_mentioned>", "offsetting_evidence": "<verbatim quote up to 50 words or null>", "sector_offset_matching": "<one of: explicit | implicit | none | not_mentioned>", "sector_offset_l4l_removals": "<one of: yes | partly | no | not_mentioned>", "sector_offset_matching_evidence": "<verbatim quote up to 50 words or null>", "matched_pairs": [ { "sector": "<sector name>", "sector_type": "<biologic | geologic | mixed>", "sector_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned — USE reduced_emissions for CCS on industrial/fossil sectors, never removals>", "offset_nature": "<one of: removals | reduced_emissions | avoided_emissions | mixed | not_mentioned>", "offset_type": "<unspecified | biologic | geologic | other | none>", "removal_detail": "<specific method if offset_nature is removals, e.g. BECCS, DACCS, afforestation — null otherwise>", "match_explicitness": "<explicit | implicit>", "match_evidence": "<verbatim quote up to 50 words or null>", "notes": "<one sentence max or null>", } ] }}
---
### DEFINITIONS
- Buyer (Article 6): country intends to use ITMOs to count toward its own NDC target.- Seller (Article 6): country generates and transfers mitigation outcomes to another country.- IMPORTANT — Article 6 intent: a conditional NDC depending on international financial support or technology transfer does NOT qualify as Article 6 intent unless carbon markets or ITMOs are explicitly named. Only answer "yes" or "considering" if Article 6, ITMOs, cooperative approaches, or carbon market mechanisms are explicitly referenced.- CDR/removals: intentional human activities that remove CO2 from the atmosphere and store it durably. CCS alone is NOT CDR — it does not remove CO2 from the atmosphere. SAF is not CDR — CO2 is re-emitted through combustion.- Nature-based CDR: forests, soils, wetlands, blue carbon, REDD+, agroforestry. Not all LULUCF/AFOLU activities qualify — storage must be durable. Deforestation reduction is NOT removal.- Engineered CDR: DACCS, BECCS, enhanced weathering, carbon mineralization, waste incineration with CCS. SAF is mitigation, not removal.- CCS on geologic sectors (industry, fossil energy): is emissions reduction, NOT CDR, because no CO2 is removed from the atmosphere. CCS is CDR only when applied to bioenergy, direct air capture, or waste incineration.- Domestic CDR: removal within national borders counted domestically.- International CDR: removal credited via ITMOs, Article 6.4, or voluntary carbon markets.- Biologic sector: residual emissions from agriculture, forestry, land use change.- Geologic sector: residual emissions from industry, cement, aviation, fossil energy.- Like-for-like matching: GEOLOGIC match = a fossil/industrial/process sector paired with engineered removal (DACCS, BECCS, geological CCS on bioenergy/waste/DAC). BIOLOGIC match = an agricultural/land-use/biogenic sector paired with nature-based removal (afforestation, soil carbon, blue carbon, REDD+). MISMATCH = biologic sector paired with engineered removal, or geologic sector paired with nature-based removal.- sector_offset_l4l_removals values: "yes" — all identified residual sectors are matched with removals of the same type (geologic↔engineered, biologic↔nature-based) "partly" — some sectors are matched by type, others use mixed/unspecified removals or non-removal offsets "no" — sectors and removals are present but none are matched by type "not_mentioned" — insufficient information to determine matching logic
### SELF-CONSISTENCY CHECK: Before returning, verify that cdr_nature_based_types, cdr_engineered_types, and offset_nature are consistent with the definitions:
- If removal_detail is "CCS on industry", "CCS on fossil fuels", or any CCS applied to a geologic sector → offset_nature MUST be "reduced_emissions", never "removals"- If you wrote "emissions reduction" or "not CDR" in the notes field → offset_nature MUST be "reduced_emissions"---
### INPUT
Country: COUNTRY_PLACEHOLDERDocument date: DATE_PLACEHOLDER
Text chunks:CHUNKS_PLACEHOLDER"""
PROMPT_TEMPLATE = """---
### INSTRUCTIONS
Read all chunks carefully before answering. Base your answers only on what is explicitly stated or clearly implied in the text. Do not infer or hallucinate. If a field cannot be determined from the text, use null. For every non-null value, populate the corresponding _evidence field with a verbatim quote from the source text (up to 50 words). Preserve the exact wording. If no direct quote is available, write null. For each theme, only use the text excerpts associated with that theme. If no excerpts are available, reply "not_mentioned" or null to all questions. ---
### OUTPUT FORMAT
Return only a valid JSON object. No explanation, no markdown, no preamble.
{ "country": "<country name>", "document_date": "<YYYY-MM or YYYY>",
"article6": { "intent": "<one of: yes | no | considering | not_mentioned>", "intent_evidence": "<verbatim quote up to 50 words or null>", "role_buyer": "<one of: yes | no | possible | null>", "role_buyer_evidence": "<verbatim quote up to 50 words or null>", "role_seller": "<one of: yes | no | possible | null>", "role_seller_evidence": "<verbatim quote up to 50 words or null>", "mechanisms_mentioned": ["<e.g. Article 6.2, Article 6.4, Article 6.8>"], "other_mechanisms_mentioned": ["<e.g. voluntary carbon market, JCM, CORSIA, etc>"], "notes": "<one sentence max or null>" },
"cdr": { "cdr_considered": "<one of: yes | no | not_mentioned>", "cdr_considered_evidence": "<verbatim quote up to 50 words or null>", "cdr_quantitative_target": "<one of: yes | no | not_mentioned>", "cdr_quantitative_target_evidence": "<verbatim quote up to 50 words or null>", "cdr_nature_based": "<one of: yes | no | not_mentioned>", "cdr_nature_based_types": ["<e.g. afforestation, blue carbon, soil carbon, REDD+, biochar>"], "cdr_nature_based_evidence": "<verbatim quote up to 50 words or null>", "cdr_engineered": "<one of: yes | no | not_mentioned>", "cdr_engineered_types": ["<e.g. DACCS, BECCS, enhanced weathering >"], "cdr_engineered_evidence": "<verbatim quote up to 50 words or null>", "domestic_cdr": "<one of: yes | no | not_mentioned>", "domestic_cdr_evidence": "<verbatim quote up to 50 words or null>", "international_cdr": "<one of: yes | no | not_mentioned>", "international_cdr_evidence": "<verbatim quote up to 50 words or null>", "permanence_addressed": "<one of: yes | no | not_mentioned>", "permanence_evidence": "<verbatim quote up to 50 words or null>", "notes": "<one sentence max or null>" },
"like_for_like": { "residual_emissions_identified": "<one of: yes | no | not_mentioned>", "residual_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned>", "residual_emissions_evidence": "<verbatim quote up to 50 words or null>", "hard_to_abate_sectors": ["<sector names explicitly identified>"], "hard_to_abate_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned>", "hard_to_abate_sector_type_biologic": "<one of: yes | no | not_mentioned>", "hard_to_abate_sector_type_geologic": "<one of: yes | no | not_mentioned>", "hard_to_abate_sectors_evidence": "<verbatim quote up to 50 words or null>", "offsetting_considered": "<one of: yes | no | not_mentioned>", "offsetting_evidence": "<verbatim quote up to 50 words or null>", "sector_offset_matching": "<one of: explicit | implicit | none | not_mentioned>", "sector_offset_l4l_removals": "<one of: yes | partly | no | not_mentioned>", "sector_offset_matching_evidence": "<verbatim quote up to 50 words or null>", "matched_pairs": [ { "sector": "<sector name>", "sector_type": "<biologic | geologic | mixed>", "sector_emissions_quantitative_assessment": "<one of: yes | no | not_mentioned — USE reduced_emissions for CCS on industrial/fossil sectors, never removals>", "offset_nature": "<one of: removals | reduced_emissions | avoided_emissions | mixed | not_mentioned>", "offset_type": "<unspecified | biologic | geologic | other | none>", "removal_detail": "<specific method if offset_nature is removals, e.g. BECCS, DACCS, afforestation — null otherwise>", "match_explicitness": "<explicit | implicit>", "match_evidence": "<verbatim quote up to 50 words or null>", "notes": "<one sentence max or null>", } ] }}
---
### DEFINITIONS
- Buyer (Article 6): country intends to use ITMOs to count toward its own NDC target.- Seller (Article 6): country generates and transfers mitigation outcomes to another country.- IMPORTANT — Article 6 intent: a conditional NDC depending on international financial support or technology transfer does NOT qualify as Article 6 intent unless carbon markets or ITMOs are explicitly named. Only answer "yes" or "considering" if Article 6, ITMOs, cooperative approaches, or carbon market mechanisms are explicitly referenced.- CDR/removals: intentional human activities that remove CO2 from the atmosphere and store it durably. CCS alone is NOT CDR — it does not remove CO2 from the atmosphere. SAF is not CDR — CO2 is re-emitted through combustion.- Nature-based CDR: forests, soils, wetlands, blue carbon, REDD+, agroforestry. Not all LULUCF/AFOLU activities qualify — storage must be durable. Deforestation reduction is NOT removal.- Engineered CDR: DACCS, BECCS, enhanced weathering, carbon mineralization, waste incineration with CCS. SAF is mitigation, not removal.- CCS on geologic sectors (industry, fossil energy): is emissions reduction, NOT CDR, because no CO2 is removed from the atmosphere. CCS is CDR only when applied to bioenergy, direct air capture, or waste incineration.- Domestic CDR: removal within national borders counted domestically.- International CDR: removal credited via ITMOs, Article 6.4, or voluntary carbon markets.- Biologic sector: residual emissions from agriculture, forestry, land use change.- Geologic sector: residual emissions from industry, cement, aviation, fossil energy.- Like-for-like matching: GEOLOGIC match = a fossil/industrial/process sector paired with engineered removal (DACCS, BECCS, geological CCS on bioenergy/waste/DAC). BIOLOGIC match = an agricultural/land-use/biogenic sector paired with nature-based removal (afforestation, soil carbon, blue carbon, REDD+). MISMATCH = biologic sector paired with engineered removal, or geologic sector paired with nature-based removal.- sector_offset_l4l_removals values: "yes" — all identified residual sectors are matched with removals of the same type (geologic↔engineered, biologic↔nature-based) "partly" — some sectors are matched by type, others use mixed/unspecified removals or non-removal offsets "no" — sectors and removals are present but none are matched by type "not_mentioned" — insufficient information to determine matching logic
### SELF-CONSISTENCY CHECK: Before returning, verify that cdr_nature_based_types, cdr_engineered_types, and offset_nature are consistent with the definitions:
- If removal_detail is "CCS on industry", "CCS on fossil fuels", or any CCS applied to a geologic sector → offset_nature MUST be "reduced_emissions", never "removals"- If you wrote "emissions reduction" or "not CDR" in the notes field → offset_nature MUST be "reduced_emissions"---
### INPUT
Country: COUNTRY_PLACEHOLDERDocument date: DATE_PLACEHOLDER
Text chunks:CHUNKS_PLACEHOLDER"""
Page updated
Google Sites
Report abuse