The Carbon Footprint of Leather: A Critical Review of Roncevich et al. (2026)

A new reassessment puts bovine leather's carbon footprint at twelve times that of "vegan leather" alternatives. The number driving that gap is an allocation figure the newest primary data does not support.

At a glance

• Roncevich et al. (2026) allocate 8.1% of farm-stage emissions to hides, a weighted average of 6.8% (Italy), 8% (India), and 14% (North America), replacing PEFCR's published default of 3.5%.

• UNIDO's 2026 Guidelines for Assessing the Environmental Footprint of Leather, built on 299 primary-data studies across 87 tanneries in 26 countries, sets that allocation at 1.50% instead, over five times lower.

• Mass-based allocation barely moved under the same new data (7% to 7.33%); only the economic allocation, the input Roncevich raised, moved down.

• The paper's country export weights come from a trade category that mixes raw hides with tanned and finished leather. Trade Data Monitor puts the US at 27% of global wet blue and salted hide exports by value, well above the paper's 10.6% "North America" figure.

• A separate US-specific hide study puts the US economic allocation at just 1%.

Roncevich, Hayek, and Hinestroza published “The Carbon Footprint of Leather: A Comprehensive Reassessment Using Global Livestock Data and Meta-Analysis” this year, concluding that a kilogram of bovine leather produces 187.1 kg of CO₂-equivalent emissions, compared with 14.9 kg for the plant-based and synthetic materials it markets as "vegan leather." That twelvefold gap rests on one input more than any other: an 8.1% economic allocation of farm-stage emissions to hides, which the paper builds by weighting country-level ratios for Italy (6.8%), India (8%), and North America (14%) against each country's reported share of global raw hide exports. This replaces the European Union's published PEFCR default of 3.5%, which the paper cites for comparison but does not use.

That number has three problems worth knowing before repeating it. The trade data behind the country weights measures something other than what the paper says it measures. The newest primary data on hide allocation, published after the paper went to press, points in the opposite direction from where the paper takes the number. And the allocation figure is only one of several methodological choices in the paper, each of which pushes the result the same way. The first two are covered next; the rest follow further down.

Where the Export-Share Weights Come From, and Why They Don't Hold Up

The paper's country weights (18.4% for Italy, 2.7% for India, 10.6% for North America) come from a single cited source: the Observatory of Economic Complexity's trade profile for "Raw Hides & Skin (Non-Fur)." That category is not raw hides. It is the entire Harmonized System Chapter 41, which groups untanned hides (HS 4101–4103) together with tanned, crust, and finished leather (HS 4104–4107, 4112–4115).

Italy's trade record shows what that mixing does to the numbers. In 2023, Italy exported roughly $128 million worth of actual raw bovine and equine hides under HS 4101, against roughly $726 million in imports of the same code. Italy is a net importer of raw hides by a factor of nearly six, which tracks with its tanning industry running largely on imported hide stock. An 18.4% global export share is not plausible for a country in that position. It is far more plausible for a country that exports large volumes of tanned and finished leather, which is exactly what Italy does. The paper's Italy figure is almost certainly leather-export value counted as hide-export value.

That has a second effect: it understates the United States' actual position. Trade Data Monitor data on wet blue and salted hides, using export value and volume as reported from the import side by the rest of the world's trading partners, puts the US at 27% of global trade by value and 21% by volume, roughly double to two-and-a-half times the paper's 10.6% "North America" figure on its own.

What the Newer Data Says

Set the trade-data problem aside and the allocation number still moves the wrong way once newer studies are brought in. The 3.5% PEFCR default the paper cites traces back to a single-country pilot study, run in the Netherlands using 2009 and 2014–2015 data. UNIDO's own 2026 guidelines say plainly that this default "should not be adopted... at global scale." In its place, UNIDO recommends an economic allocation of 1.50%, drawn from 299 primary-data studies across 87 tanneries in 26 countries collected between 2022 and 2024 (Section 6.7.3.1.3 of its Guidelines for Assessing the Environmental Footprint of Leather).

A separate, US-specific hide study commissioned by the Leather & Hide Council of America (Resilience Services PLLC and Spin360, 2024) arrives at a similar place from a different direction. Rather than a global production-weighted average, it models continental US cattle operations directly, using the USDA-validated Integrated Farm System Model, and calculates a US economic allocation of 1% (a mass allocation of 4.8%, against the same 3.5% PEFCR benchmark).

The Leather Working Group's own 2024 LCA, drawing on primary data from 28 finished-leather and 32 wet-blue manufacturers across 16 to 18 countries, lands in the same range: an average economic allocation of 1.8% and a mass fraction of 7.8%, which its authors likewise note is meaningfully lower than the PEFCR defaults. Counting Roncevich's own PEFCR baseline, that makes four figures on the record for hide's economic allocation, 1%, 1.5%, 1.8%, and PEFCR's 3.5%, none within reach of the paper's 8.1%.

Mass-based allocation, notably, barely moved under this same body of new data: PEFCR's 7% becomes UNIDO's 7.33%. Only the economic allocation, the figure Roncevich raised to 8.1%, turns out to move the other direction under current primary data. Two independent efforts, a 299-study global review and a dedicated US model, now put that number between 1% and 1.5%. Neither comes close to 8.1%.

“The newest, most representative data available, 299 tannery studies across 26 countries, puts the hide allocation at 1.5%. This paper uses 8.1%. The correction runs backward.”

Notably, the paper does not reference the Leather Working Group anywhere in its text, and cites Leather Naturally exactly once, as the web host of an LHCA marketing infographic it disputes rather than as an LCA data source. Its farming-phase benchmark instead relies on FAO's GLEAM 3 model, built on a 2015 baseline, combined with a meta-analysis of a small set of peer-reviewed leather LCAs, several of them over a decade old.

Industry LCAs built from primary tannery data report considerably lower figures than the paper's. At the paper's own stated leather density of 0.92 kg/m², its 187.1 kg CO₂-e/kg headline converts to roughly 172 kg CO₂-e/m². The Leather Working Group's 2024 industry LCA, covering 50 products across 60 tanneries in 16 to 18 countries, reports 22.48 kg CO₂-e per square meter of finished leather. A separate study affiliated with Leather Naturally (Brugnoli et al. 2025, Discover Sustainability), built from 56 LCA studies across 16 facilities in 11 countries, reports 22.0 kg CO₂-e per kilogram of leather at a declared unit of 1.0632 m². Both figures land well over seven times below the paper's per-area equivalent.

Independent Economic Research Points the Same Direction

A separate line of evidence, using entirely different data and methods, reaches the same conclusion by a different route. Brester and Swanser (2021), in two studies commissioned by the Leather & Hide Council of America, used Granger causality tests and reduced-form price regressions to ask whether hide values are actually an economic driver of cattle production, in the United States and in Brazil. In both countries, they found no statistical evidence that hide prices directly affect the number of cattle produced. The indirect effect, working through cattle prices, is real but small: fed steer prices in the US respond to hide prices with an elasticity of just 0.13, and Brazilian fat cattle prices respond with an elasticity of 0.15. Carried through to production, a 10% increase in hide prices raises US cattle production by an estimated 0.31%, and Brazilian production by 0.36%, translating into inventory increases of roughly 0.17% and 0.12%, respectively.

The same US study also tracks hide value as a share of total cattle byproduct value from 1995 through 2019. Hides represented about half of byproduct value through 2016, but a declining share after that, while total byproduct value itself has averaged only about 10 to 11% of a live animal's value. On this measure, hides are a modest, and by the paper's own study period shrinking, fraction of what a slaughtered animal is worth, an odd basis for an allocation input meant to be moving upward.

A Comparison the Methodology Doesn't Support

ISO 14044 restricts comparative assertions between LCA studies unless system boundaries, functional units, and production scope are aligned (Clause 4.2.3.7). When the comparison is between substitutable materials, like leather and its alternatives, LCA guidance offers two legitimate paths: expand the system boundaries to cover the full product lifecycle, or use a consequential model built to capture substitution effects. Roncevich et al. takes neither path. It combines cradle-to-gate attributional studies built with different scopes and assumptions, and compares the outputs directly.

The US hide study cited above shows what observing that discipline looks like in practice. Its authors state plainly that the results are not intended to support comparative assertions of environmental superiority, because the model is attributional: built to describe one production system's impacts, not to capture what would happen if that system were displaced by something else. Roncevich et al. is built on the same attributional, cradle-to-gate footing but skips that caveat, and uses its output to rank leather against synthetic and plant-based alternatives directly.

What the Monte Carlo Step Actually Shows

Monte Carlo simulation is a legitimate, well-established tool for a specific job: where the underlying data is incomplete, it lets a researcher estimate a probable distribution rather than lean on a single point estimate. Applied to raw underlying data, it strengthens a study. Roncevich et al. applies it differently. It runs the simulation on LCA outputs that are already aggregated averages drawn from other studies, and does not disclose how it handled unequal sample sizes, cross-correlations, error propagation, or resolution level across those source papers. Averaging figures that are themselves already averages can skew the combined result in either direction. Without that disclosure, a reader has no way to tell whether the simulation reduced uncertainty or compounded it.

One Density Figure for Two Variable Materials

The paper assumes fixed densities of 0.92 kg/m² for leather and 0.72 kg/m² for vegan alternatives, though both vary considerably with thickness and construction. Leather runs roughly 0.49 to 2.1 kg/m²; vegan alternatives run roughly 0.28 to 1.5 kg/m². This is why most material LCAs declare the functional unit in square meters rather than kilograms. A single fixed density understates that variability, and introduces the same kind of uncertainty the paper's Monte Carlo step was meant to resolve.

Where Allocation Should Be Measured

Economic allocation is supposed to use co-product value at the point of separation, the slaughterhouse, not a downstream, value-added price. Roncevich et al. uses export values instead, which carry hide-merchant margin and do not appear to be adjusted for hide wastage. An analogy shows the scale of the problem: aluminum sells for about $3.20/kg at the foundry gate, but pricing it at the roughly $1,495/kg implied by a $74 million Falcon 9 rocket (about 90% aluminum by mass) shows what happens when a downstream value stands in for a point-of-separation value.

An Uncounted Co-Product

The paper converts raw hide to finished leather at 6.37 kg of raw hide per kg of leather, the average of four ratios (4.63, 5, 7.7, 8.15) it discloses in its own supplementary data. That conversion does not appear to account for the drop split generated during processing, a co-product that itself requires allocation between upstream and core production. The issue is well documented elsewhere: a 2008 case study on Puma's footwear carbon footprint (Barling, FHTW Berlin) found that an initial evaluation built on full-grain leather had to be revised once it became clear the material actually used was split leather, precisely because the two carry different co-product allocations (Barling, 2008).

Conclusion

The paper's 187.1 kg CO₂-e/kg figure rests on five choices, and each one moves the result in the same direction: an economic allocation that the newest primary data revises downward rather than upward, a direct comparison between studies that were never built to be compared, an undisclosed Monte Carlo methodology, fixed material densities that erase real variability, and a hide-to-leather conversion that appears to leave out the drop split. None of this is hidden. It is documented in the paper's own methods section. Taken together, though, these choices do not support the conclusion that leather's carbon footprint is twelve times higher than the materials it is compared against.

One further note, separate from the methodology: the paper's use of "vegan leather" describes a term with no scientific definition and no legal standing in at least three jurisdictions. Portugal (Decree-Law No. 3/2022), Italy (Decree-Law No. 68/2020), and Brazil (Law No. 4,888/1965) all reserve "leather" terminology for animal-derived material. That loose usage is a small thing on its own, but it is consistent with the lack of definitional rigor found in the rest of the paper's methodology.

Sources

• Roncevich, M. K.; Hayek, M. N.; Hinestroza, J. P. "The Carbon Footprint of Leather: A Comprehensive Reassessment Using Global Livestock Data and Meta-Analysis." ACS Sustainable Chem. Eng. 2026, 14 (27), 12580–12589.

• United Nations Industrial Development Organization. Guidelines for Assessing the Environmental Footprint of Leather. Vienna, 2026, Section 6.7.3.1.3.

• Rosa-Giglio, P. D.; Ioannidis, I.; Gonzalez-Quijano, G.; Fontanella, A. Product Environmental Footprint Category Rules (PEFCR) for Leather, 2018.

• International Organization for Standardization. ISO 14044:2006, Environmental management — Life cycle assessment — Requirements and guidelines, Clause 4.2.3.7.

• Portugal, Decree-Law No. 3/2022 (Jan. 4, 2022); Italy, Decree-Law No. 68 (June 9, 2020); Brazil, Law No. 4,888 (Dec. 9, 1965).

• Leather Working Group. Measuring the Environmental Impact of Leather: Life Cycle Assessment. Prepared by Spin360, 2024.

• Brugnoli, F.; Sena, K.; Zugno, L.; Oggioni, A. "A global study on the Life Cycle Assessment (LCA) of the modern cow leather industry." Discover Sustainability 2025, 6, 80. https://doi.org/10.1007/s43621-025-00798-6

• Trade data, HS 4101, Italy, 2021–2023: Trendeconomy / UN Comtrade.

• Trade Data Monitor. Export value and volume, wet blue and salted hides, by reporting country, import-side (mirror) data.

• Resilience Services PLLC; Spin360. Lifecycle Assessment of Leather from American Hides. LCA Technical Report, prepared for the Leather & Hide Council of America, Sept. 15, 2024, Rev. 0, Table ES3.

• Brester, G. W.; Swanser, K. Quantifying the Relationship Between U.S. Cattle Hide Prices/Value and U.S. Cattle Production. Research report, Leather and Hide Council of America, Feb. 4, 2021.

• Brester, G. W.; Swanser, K. Quantifying the Relationship Between Brazilian Cattle Hide Prices and Brazilian Cattle Production. Research report, Leather and Hide Council of America, Aug. 24, 2021.

• Barling, R. L. Carbon Footprint of a Shoe Produced by an International Enterprise with an Implemented Corporate Social Responsibility Strategy, Taking PUMA AG as Example. FHTW Berlin, 2008.

Senior Technical Advisor, Institute for Data Integrity. Karl began his tanning career at the Leather Industries Research Institute (LIRI, South Africa) and has managed tannery production and consulted on tannery process problems since 1997. He has taught and practiced mass-balance and life-cycle analysis of leather and allied materials since 2001 and 2015, respectively.