By Amanda Symm, Communications, LHCA
Brands treat Higg MSI scores as settled fact when they compare raw materials. Textile Exchange just showed how shaky that ground can be.
Two tools sit behind many brand-facing carbon claims for fibers. Ecoinvent is a database, the actual warehouse of measured, modeled data on how materials and chemicals get made, that many carbon-accounting tools draw from. Higg MSI (Materials Sustainability Index) is built on top of it: a scoring system brands use to compare raw materials' environmental impact, feeding directly into sourcing decisions and public sustainability claims. If ecoinvent is missing a real production pathway, any tool built on it, Higg MSI included, is scoring that material on incomplete information, without anyone downstream necessarily knowing it.
That's what Textile Exchange found in its own polyester data, released in June 2026. Textile Exchange is the nonprofit standards body much of the apparel industry already relies on, not a leather industry group, and it's partway through a systematic effort to build better raw-material LCAs (Life Cycle Assessments) across fibers using a broader methodology it calls LCA+, which extends past standard carbon accounting into biodiversity, soil health, animal welfare, and social impact. It published a cotton LCA in March 2026, the polyester LCA in June, and a cashmere LCA in July. Leather hide is next: Textile Exchange's own published research roadmap now confirms studies on leather hide, nylon, wool, and mohair are due throughout 2026 and into 2027, no firmer date than that has been set yet, but the material is officially on the list. When the industry's own standards body turns this same methodology on leather, whatever it finds will carry real weight, for leather or against it. The polyester gap below is a preview of how that process plays out.
The gap Textile Exchange found sits in one chemical: ethylene glycol, sold industrially as MEG (Monoethylene Glycol), a key input to virgin PET (Polyethylene Terephthalate), the plastic that becomes most virgin polyester and much of the feedstock for chemically recycled polyester too. Ecoinvent's standard model assumes MEG comes from oil or natural gas. In China, which produces most of the world's polyester, a growing share now comes from coal instead. Coal-to-MEG production has scaled up fast enough that China's dependence on imported MEG fell from 67% in 2013 to 29% in 2023. Ecoinvent has no model for the coal route at all. Textile Exchange had to pull figures from an outside academic paper just to test what including it would do to the numbers.
The answer: coal-based MEG carries a climate impact up to 92% higher than the oil-or-gas route ecoinvent assumes. Once that's factored through to finished polyester, the products modeled in the study saw their carbon footprint climb 15 to 21%, and up to 32% for the Chinese recyclers most reliant on virgin MEG as an input. The study is also cradle-to-gate only, meaning it stops at the factory gate and never accounts for what happens after a garment is worn and discarded, so polyester's durability, or lack of it, doesn't factor into these numbers either way.
Textile Exchange is careful to say this data isn't meant for fiber-to-fiber comparisons, and that caveat is worth taking seriously. But the practical takeaway for LHCA members holds regardless: a tool brands already use to justify sourcing decisions and public sustainability claims was built on a database that didn't, until now, reflect how a substantial and growing share of its most common comparison fiber actually gets made. That's the same structural weakness IDI (the Institute for Data Integrity) was built to find in the tools used to score leather, and it's a preview of the scrutiny leather's own LCA is likely to face once Textile Exchange gets there.
Ecoinvent has said it's evaluating whether to incorporate the coal-based pathway into future updates. Until it does, every brand claim built on top of Higg MSI for polyester is missing part of the picture, a reminder that a certified number isn't automatically a complete one.