Why Bangladesh Loses More Cotton Than Norway's Best Recycling System Can Ever Recover

Manufacturing a cotton t-shirt in Bangladesh is, in material terms, a process of radical reduction. Of the 197gm of raw cotton fibre that enter the production system, only 110gm survive as a finished garment. By the time a completed garment leaves the factory, 86gm have been lost across four successive stages: yarn spinning, fabric production, wet processing, and garment cutting, consumed by the ordinary mechanics of industrial textile manufacturing. The finished t-shirt embodies just 56% of the cotton that was required to make it.

Yarn production is by far the largest single point of attrition, accounting for 49gm of loss per t-shirt: a 25% process loss rate driven by the removal of short fibres and impurities during carding and combing, by damaged or quality-rejected yarn, and by the inherent inefficiencies of opening bales and preparing fibre for high-speed circular knitting. A further 17gm disappears in wet processing (12%), where faulty and unevenly dyed fabric is the primary culprit. Cutting and sewing—apparel production—loses another 17gm (13%) through fabric offcuts and quality nonconformance, while fabric production itself contributes a relatively minor 3gm (2%).

That comparison is what the arithmetic comes down to. The 86gm lost before the t-shirt reaches a consumer is larger than the 33gm that Norway's best-practice end-of-life system recovers after the consumer is done with it. Pre-consumer waste, in other words, outweighs post-consumer recovery by a factor of more than two and a half. Yet the policy architecture being assembled in Brussels is oriented almost entirely toward what happens after the garment is discarded.

The discrepancy between factory-level data and the figures embedded in global material flow assessments compounds the problem. The Ellen MacArthur Foundation's widely cited analysis estimates production-stage losses at 12%, less than a third of the 44% found when factory-level empirical data from Bangladeshi manufacturing facilities is applied. That gap is not a measurement error. Global material flow assessments typically draw on national accounts, official statistical surveys, and industry reports, sources that do not capture the operational losses of a specific product line in a specific factory. The paper, by contrast, draws on recent factory-based empirical research covering a large number of manufacturing facilities and explicitly quantifying process losses at each stage. The result is a pre-consumer waste figure that is more representative of what actually happens on the factory floor, and considerably more damaging to the assumptions underpinning EU circular textile targets.

The t-shirt that eventually reaches a Norwegian consumer has already been diminished before it is ever worn. What the post-consumer system inherits is 110gm, not 197gm of recoverable cotton, and the ceiling on what any downstream recycling scenario can achieve has been constrained before a single garment reaches a charity bin.

The separately collected t-shirt begins its post-consumer journey in reasonable shape. It has been pulled from the residual waste stream, pre-sorted at a Norwegian facility, and loaded onto a lorry heading for a largescale sorting centre in Vilnius. From there, the recyclable fraction will travel by sea to Panipat, India, where mechanical recycling, shredding and fibre opening, will return it to a state that can re-enter yarn production. This is the route that represents the current ceiling of EU textile waste management. It is also a route that loses most of what it handles.

The attrition begins at the household bin. Norway's separate collection rate for textiles sits at 40%, which means that 60 of every 100 discarded t-shirts never enter this system at all. They go directly into the residual waste stream and are incinerated. Of the 40 that are separately collected, a further 1.2% are rejected at pre-sorting in Norway as unsuitable for further processing. From there, the remaining garments make the lorry journey to Lithuania, where the Vilnius sorting facility rejects a further 6.1% as unrecyclable. What survives this sequence is shipped to Panipat, where mechanical recycling introduces a 20% process loss of its own, fibres lost to the shredding and opening drums that convert garments back into loose fibre. At the end of this chain, 33gm of fibre are recovered from the 197gm that started the first loop, just 17% of what went in.

Raising the collection rate does not solve the problem. The paper models a scenario in which the rate climbs to 90%, the level implied by the EU Waste Framework Directive's 2025 mandate for separate textile collection. Under those conditions, material recovery rises to 37%. That is a meaningful improvement, but it still falls well short of half the original fibre input, and it is achieved by assuming that the sorting and recycling infrastructure downstream can absorb the increased volume without degradation in performance. At scale, the binding constraint is what happens to garments once they are collected, not the collection rate itself.

The reuse market, which normally absorbs the highest-quality separately collected garments and offers the most significant environmental benefits of any end-of-life pathway, is under growing pressure. Mandatory separate collection across EU member states is expected to increase the volume of poor-quality fast-fashion items entering the collected stream, garments that cannot be resold in Europe and are instead exported in mixed bales to markets in Ghana and other recipient countries, where they frequently end up landfilled or in waterways. The system built to recover value from discarded textiles is simultaneously being flooded with material it cannot usefully process.

Mechanical recycling itself has a ceiling that collection policy cannot shift. Cotton fibres shorten with each shredding cycle, and yarn blends containing more than 30% recycled cotton face quality constraints that current spinning technology cannot fully resolve. The paper treats all recovered fibres as reusable in the second production loop, an assumption that implicitly requires future improvements in recycling and spinning technology. Without those advances, a portion of the recovered fibre would be downcycled into lower-value applications rather than reintroduced into t-shirt production. The 17% baseline figure, already modest, may overstate what the system currently delivers in practice.

The upper ceiling of recovery, even under idealised conditions where pre-consumer waste in Bangladesh is also recycled, reaches 44%. Nothing in the current system's architecture pushes it higher.

Four interventions were modelled against the baseline two-loop system, each applied independently so that the effect of a single change could be isolated and measured. Ranked by return, the results point in one direction—and the intervention with the largest environmental gains does not involve a European consumer, a European collection container, or a European sorting facility. It happens in a Bangladeshi factory, before the garment exists.

Pre-consumer waste minimisation, defined as reducing process losses to 13% in yarn production, 5% in wet processing, and 10% in apparel production based on best-practice efficiency observed across Bangladeshi manufacturing facilities, cuts primary fibre demand by 22%. Less cotton enters the system, less material is processed at every subsequent stage, and less waste accumulates across both loops. The environmental returns are the largest of any scenario modelled: a 10% reduction in global warming potential, and reductions of 20–25% across eutrophication, ecotoxicity, land use, and water consumption. Cotton cultivation dominates the environmental profile of a t-shirt across all impact categories except climate, accounting for 96% of land use impact and more than 73% of water consumption across the two loops. Any intervention that reduces the quantity of virgin cotton required yields outsized returns precisely because it acts on the most environmentally costly input in the entire system.

Pre-consumer waste recycling, diverting production offcuts from landfill to recycling rather than reducing the waste generated, achieves the highest material recovery of any scenario at 44%, and delivers environmental improvements that fall between the higher separate collection and waste minimisation scenarios. The mechanism is different but the lever is the same: both act on primary fibre demand, substituting recycled input for virgin cotton. Higher separate collection, by contrast, operates entirely downstream of the cultivation stage. Raising the collection rate to 90% yields roughly a 2% reduction in global warming potential and around 10% reductions in other impact categories. The green energy scenario, which models a shift to 30% renewable electricity in Bangladeshi manufacturing, cuts global warming potential by 9% but produces no measurable gains in the other four impact categories, because those categories are dominated by cotton cultivation rather than energy use in production.

The gap between what upstream intervention delivers and what downstream intervention delivers goes deeper than measurement variance—it reflects where the environmental mass of the system actually sits. Wet processing alone accounts for 27% of cumulative global warming potential across two loops, and primary fibre production dominates land use and water consumption so thoroughly that no amount of improved collection, sorting, or recycling in Europe can meaningfully shift those numbers. The only interventions that can are those that displace virgin cotton, and both originate in Bangladesh, not Brussels.

The policy consequences have yet to be fully reckoned with. The ESPR targets brands and products on recyclability and recycled content but does not impose explicit reporting requirements on production-stage waste generated outside the EU, which means that if the upstream loss rate is underestimated by a factor of three or more in the evidence base informing those targets, what exactly are the targets measuring? Not the system as it operates, only the part of it that is visible from Brussels. Pre-consumer waste in Bangladesh is not systematically measured, not required to be disclosed, and not subject to EU recovery performance indicators. A substantial share of fabric offcuts and rejected garments moves through informal grey-market channels where traceability is absent, labour conditions are often exploitative, and recycling infrastructure is negligible. This material does not enter any formal recovery system. It does not appear in European reporting. It simply disappears, and European policy, as currently framed, has no mechanism to find it.

The scenario results are model-based upper bounds, not operational targets. Real-world recovery is lower than the baseline 17% because garments diverted to reuse, resale, or repair are unavailable for recycling, and the model assumes idealised routing that does not reflect the mixed-bale export reality of current collection systems. But the directional finding holds across the full sensitivity range tested. The textile system's material losses are predominantly upstream. Its environmental impacts are dominated by cotton cultivation. And the interventions with the largest returns are those that reduce primary fibre demand, not those that improve what Europe does with garments after consumers are finished with them.

The Numbers Point Elsewhere

The EU's 2030 textile targets assume post-consumer collection is where the problem lives. The fibre flow arithmetic of a single Norwegian t-shirt says otherwise. Even at 90% separate collection, upstream losses consume more cotton than any end-of-life system can recover. The real question is whether Europe is willing to extend its reach to factory floors in Bangladesh, where most of the fibre is actually lost. Somewhere in Panipat, a shredding drum is turning 33gm back into something spinnable. The other 164gm are gone.