The Cow Credit Fraud: How Biogenic Methane, a Broken Metric, and a Sickening Feed Additive Built a Carbon Market House of Cards

Money does not only grow on trees, it is created out of thin air

There is a carbon credit product on the market right now — actively sold, actively retired against fossil CO₂ emissions — that rests on a foundation of incomplete science, an IPCC metric error, industry-funded research, unmeasurable outcomes, and a feed additive that made 68% of Danish dairy herds sick when it was first deployed at commercial scale.

Each of those failures is bad on its own. Together, they describe something more serious: a fraud architecture where no single actor is obviously culpable, because the structure itself is the mechanism.

Let's build it from the bottom up.

The Cycle the IPCC Doesn't Account For Properly

Start with basic carbon biology, because it's where the entire problem originates.

When a cow eats grass, it is consuming cellulose — plant material built from CO₂ that the grass pulled out of the atmosphere via photosynthesis. The cow's rumen microbes ferment that cellulose, and as a metabolic byproduct, the cow belches methane (CH₄). That methane stays in the atmosphere for approximately 12 years, then oxidises via hydroxyl radicals back into CO₂ and water vapour. That CO₂ is then available for grass to absorb again.

The carbon atoms in that methane are recycled atmospheric carbon. They were in the air. They went into a plant. They went into a cow. They came out as methane. Twelve years later, they're back in the air as CO₂, ready to go into another plant.

No new carbon enters the system. At steady state — a herd that is not growing — the biogenic methane cycle is closed.

This is not a fringe argument. It is described in peer-reviewed literature, acknowledged by the IPCC itself, and contrasted explicitly with fossil methane — which releases carbon that has been locked in geological storage for hundreds of millions of years, permanently adding new carbon to the active atmosphere cycle.

The difference matters profoundly:

A stable herd with stable emissions is not adding to the warming burden. It is recycling it.

Where the IPCC Got It Wrong: GWP100

The IPCC assigns every greenhouse gas a Global Warming Potential (GWP) — a measure of how much heat it traps relative to CO₂ over a given time horizon, typically 100 years.

Biogenic methane in IPCC AR6: GWP100 = 27.

This means that under the official accounting framework, one tonne of methane from a dairy cow is counted as equivalent to 27 tonnes of CO₂. Every carbon registry — Verra, Gold Standard, ACR, CAR — uses this figure as the foundation of their livestock methane methodologies.

The problem is structural. GWP100 was designed for stock gases — gases that accumulate in the atmosphere, like CO₂. Methane is a flow gas. It enters the atmosphere, does its warming, and exits. At steady-state emissions, no new warming is being added beyond what is already baked in from the current atmospheric burden.

Oxford researchers formalised this distinction in a metric called GWP* (GWP-star). The principle is that the CO₂-equivalence of a short-lived gas should be determined primarily by changes in its emission rate, not by its absolute rate. Under GWP*:

• A herd with stable methane emissions → zero additional warming contribution
• A herd with declining methane emissions → net cooling effect
• A herd with increasing methane emissions → warming contribution

A 2025 peer-reviewed study modelling FAO livestock scenarios found that GWP100 overestimates the temperature effect of constant livestock methane by approximately 0.04°C by 2050 compared to GWP* estimates. Under a decreasing emissions scenario, GWP100 implies continued temperature increase while GWP* correctly shows stabilisation and cooling.

The IPCC itself, in AR5, stated: "There is no scientific argument for selecting 100 years compared with other choices." They acknowledge the limitation. They have not fixed the methodology.

To be precise about what the error is: the IPCC did not make a mistake by including biogenic methane as a greenhouse gas — it is one, while it's in the atmosphere. The error is applying the same equivalence framework to a flow gas as to a stock gas, producing systematic overestimation of the warming contribution of stable or declining livestock herds. This is documented in the peer-reviewed literature. It has not entered registry methodology.

The Carbon Credit Product Built on This Foundation

Here is what the credit product looks like:

A cattle operation enrolls in a Verra or Gold Standard methane reduction programme. The project developer selects a methodology — typically one built on DSM-Firmenich-funded efficacy studies — that assigns a baseline methane emission factor per animal per day. The farm then feeds cows Bovaer (active ingredient: 3-nitrooxypropanol, 3-NOP), a rumen enzyme inhibitor designed to suppress methanogenesis.

The claimed reduction: 20–30% of enteric methane emissions.

That reduction figure is multiplied by:

• Number of animals
• Days of treatment
• GWP100 = 27 (the flawed metric)
• Converted to CO₂e tonnes

The result is a quantity of carbon credits, which are sold to industrial emitters to offset their fossil CO₂ emissions.

Every node in that calculation chain has a problem.

The Measurement Problem: There Is No Ex-Post Verification

Here is the most structurally important failure, and the one least discussed:

It is not possible to measure actual methane reduction from enteric fermentation at commercial farm scale.

The only rigorous measurement method is the respiration chamber — an enclosed room with precision sensors that can capture total gas output from individual animals. These are expensive laboratory infrastructure. They are not deployable across commercial herds.

What happens instead: a short-term controlled trial (typically 12–16 weeks) in a respiration chamber, under controlled feed conditions, produces an efficacy figure. That figure — let's say 25% reduction — is then applied as a universal emission factor across entire herds, for the full crediting period, often extending to 2030 and beyond.

This is not measurement-based accounting. It is assumption-based accounting that borrows the vocabulary of measurement.

A third-party auditor reviewing one of these projects cannot:

• Independently measure actual methane output from the herd
• Verify that Bovaer was administered consistently to every animal every day
• Confirm that feed composition remained stable (which directly affects efficacy)
• Test whether the lab-derived emission factor applies to this specific herd under these specific conditions

What the auditor can do is verify that the project developer followed the methodology. Verification becomes process compliance, not outcome verification. The methodology is the problem — confirming adherence to it confirms nothing about the real-world climate outcome.

The Research Conflict: Who Paid for the Science?

Bovaer is manufactured by DSM-Firmenich, a Swiss-Dutch multinational. The efficacy data underpinning registry methodologies — the 20–30% methane reduction figures — comes predominantly from studies funded or co-funded by DSM.

This is the same structural problem documented in supplier-specific emission factors across Scope 3 carbon accounting: the entity with the largest financial interest in a high efficacy figure is the entity financing the research that produces that figure. The sampling is not random. The incentive runs in one direction. The output re-enters the system as objective baseline.

An independent commercial-scale replication study producing a figure of 8% reduction, or demonstrating adaptation of the rumen microbiome that erodes efficacy over time, would not invalidate credits already retired. The financial instruments are one-way.

No Long-Term Data: The Permanence Assumption Has No Empirical Basis

Bovaer received EU regulatory approval in 2022. Commercial use began in 2022–2023. Crediting periods issued now extend to 2030 and beyond.

The scientific questions that have never been answered at commercial scale:

• Does the rumen microbiome adapt over time, developing compensatory fermentation pathways that restore methane output?
• Does efficacy vary with feed quality, season, stress, disease burden — and if so, how?
• What happens to other fermentation byproducts when one pathway is chemically suppressed?
• Are there cumulative health effects from chronic 3-NOP exposure?

These are not hypothetical. They are the standard questions you would ask before issuing a financial instrument against a multi-year emissions reduction claim.

Denmark: The First Real-World Test, and What It Found

In January 2025, Denmark became the first country to mandate Bovaer use — requiring all dairy farms with over 50 cows to feed it for at least 80 days per year, under threat of heavy fines. Most farmers began October 1, 2025.

Within two weeks, reports began arriving.

A SEGES Innovation survey of 1,641 conventional Danish dairy farms with over 50 cows received 551 responses:

• 68% reported lower milk yield
• 66% reported reduced feed intake
• 59% reported both simultaneously
• 349 herds reported increased digestive and metabolic disorders — reduced rumination, diarrhoea, atypical milk fever, and fever

Individual farmer accounts were more stark. One farmer with 1,000 cows reported a full-herd diarrhoea event within days. Another — a veterinarian who milks 580 cows — documented an explosion of digital dermatitis cases, stopped feeding Bovaer, saw recovery within two days, restarted to comply with the law, and watched the same symptoms return.

"I wouldn't recommend it to anyone, not in one million years," he said.

Sweden stopped its Bovaer trials. Norway halted its programme. Denmark quietly allowed sick-cow opt-outs.

DSM-Firmenich's response: "There remains absolutely no evidence that Bovaer is a cause of any problems."

Note the epistemological move: the company sets the evidentiary standard as "direct causal linkage," which is difficult to establish in complex biological systems, while 68% of farms using a single new compound at a single new time report the same suite of symptoms, with recovery upon cessation and relapse upon reintroduction.

Also note: only now, three years after regulatory approval, is Aarhus University conducting the first real-world on-farm study. The approval, the methodology, and the credits preceded the evidence.

The Complete Architecture

Assemble all the layers:

Layer 1 — Cycle science: A stable herd at steady-state emissions adds no new carbon to the atmosphere. The biogenic cycle is closed.

Layer 2 — Metric error: IPCC GWP100 applies a stock-gas framework to a flow gas, systematically overestimating the warming contribution of stable or declining livestock herds. Peer-reviewed literature documents this. Registry methodology has not incorporated the correction.

Layer 3 — No ex-post measurement: Credits are calculated from modeled emission factors derived from short-term lab studies. Commercial-scale measurement is physically impossible with current technology. Verification confirms process, not outcome.

Layer 4 — Conflicted research: The efficacy figures underpinning those models come predominantly from manufacturer-funded studies.

Layer 5 — No long-term data: Crediting periods extending to 2030+ rest on efficacy assumptions that have no multi-year empirical basis.

Layer 6 — Denmark: The first mandatory commercial-scale deployment produced a 66–68% adverse event rate. The gap between controlled lab conditions and real-world commercial deployment — which the credit methodology assumes away — is real, significant, and measurable.

Layer 7 — Asymmetric offset claim: Industrial emitters burning fossil fuels — permanently adding ancient geological carbon to the active atmosphere — retire these credits against that activity, claiming equivalence. The offset claim rests on a flawed metric, an unmeasurable reduction, conflicted research, and an additive that damaged two-thirds of herds when deployed at scale.

Who Benefits from the Impossibility of Questioning This

The registries have a financial interest in high credit volumes. DSM-Firmenich has an interest in both additive sales and the carbon market the methodology creates. Industrial emitters have an interest in cheap, abundant offsets that allow business-as-usual operations. Governments with large livestock sectors have an interest in appearing to act on agricultural emissions without requiring structural change.

The researchers who developed GWP* were doing legitimate science. The structure that prevents their correction from entering registry methodology is not scientific — it is institutional.

The pattern is not new. REDD+ forest carbon followed the same trajectory: an elegant concept, real science, legitimate monitoring challenges, financial instruments issued ahead of verification capacity, credits retired that never represented real reductions, a Guardian investigation, belated acknowledgment. The architecture was visible to people who understood verification. The institutional incentives prevented correction until the reputational damage made it unavoidable.

We are earlier in that cycle with livestock methane feed additive credits. The Denmark situation is not a side story. It is the empirical confirmation, at commercial scale, of the gap that the methodology had always assumed away.

The credits already retired will not be clawed back. The companies that bought them will not restate their net-zero claims. The farmers whose cows got sick were the only ones in this chain who had no financial interest in the outcome — and they were the ones who paid.

Tags: carbon markets, biogenic methane, GWP*, Bovaer, enteric fermentation, carbon credits, livestock, Denmark, IPCC, carbon accounting, verification integrity