Europe's plastic packaging can go circular. Its import dependence would move rather than disappear.
Reuse and lighter packaging could cut demand for virgin PET by a third. But meeting recycled-content targets means importing recyclate Europe cannot produce for itself.
Status: results from Deliverable D8.1, not yet approved by the European Commission and subject to revision.
European industry depends on what it can buy from the rest of the world, and disruption far away feeds through into production costs at home. Circular economy measures are increasingly proposed as an answer: if Europe used less material, and used it more than once, it would need to buy less from elsewhere.
This is the argument we explore in this study by asking a specific question: what effect can circularity strategies actually have on EU trade flows and industrial resilience?
We tested that question on PET — the plastic in drinks bottles and food trays. It is an unusually clean test, because in Europe PET is essentially a packaging polymer: packaging accounts for around 99% of EU PET demand. Circular measures here act on almost the entire system, not a corner of it.
That matters, because the companion study on steel in passenger cars found the opposite: cars were too small a share of steel demand for even ambitious circularity to shift Europe's import position. PET is not subject to that limitation.
What we modelled
Three scenarios, each building on the last, covering all forms of PET packaging in EU27, UK, Norway and Switzerland (EU27+3) from 2018 to 2050:
- R1 — recycling. The baseline: single-use packaging only, with EU packaging law's collection, sorting and recycled-content targets met. Recycled content in beverage bottles rises from 30% in 2030 to 65% in 2040.
- R2 — recycling plus reuse. Adds reusable packaging. Reusable beverage bottles rise from 1% of the market in 2022 to 50% by 2050, supported by deposit return schemes, with trays and non-beverage bottles following at lower rates.
- R3 — recycling, reuse and reduce. Adds lightweighting: the lighter designs already available in the early 2020s become the norm by 2030.
Underneath all three is the same assumption about how much packaging Europe wants — a steady 1% annual growth. The scenarios differ only in how that demand is met.
These are what-if scenarios. They describe what would follow if the measures were adopted on that schedule — not what is expected to happen.
What would have to change for this to happen
The scenario narratives came out of a stakeholder workshop in Brussels in December 2025.
R2 assumes deposit return systems for reusable bottles develop alongside those for single-use, and that refill-at-home and refill-on-the-go schemes keep expanding. Reusable trays appear only where plausible — mono-material trays that need no glue seal, such as berry punnets and takeaway containers. Flexible packaging and film never become meaningfully reusable.
R3 is deliberately modest. Packaging has been getting lighter for twenty years, driven by manufacturers' costs rather than policy, and that trend appears to have plateaued in the late 2010s — higher recycled content and reusability both work against further weight reduction. So R3 assumes no breakthrough, only that today's lighter designs become standard.
Reuse means using more plastic in order to use less
The first result is counterintuitive. A reusable bottle is built to survive: at 103g it weighs nearly two and a half times a 42g single-use bottle, and reusable trays carry a similar penalty. Switching to reuse therefore increases the total tonnage of PET in circulation.
In 2050, total PET need reaches 6.6 Mt under R1 but 10.2 Mt under R2 — 55% higher. Yet R2's demand for virgin PET is 20% lower, and for recycled PET 24% lower. The extra material is not consumed; it circulates. Around half of R2's 2050 need is met by polymer already in use.
Reusable beverage bottles drive most of this. By 2030 the PET reused in them is seven times the virgin and recycled polymer needed to make new ones — 0.7 Mt against 0.1 Mt — and that ratio holds to 2050, when 4.3 Mt comes from reuse against 0.6 Mt from primary and secondary sources.
Adding lightweighting in R3 dampens the picture. Total need falls to 7.8 Mt in 2050, a quarter below R2, while virgin demand drops further to 1.7 Mt — cumulatively across 2018–2050, lightweighting alone accounts for about 20% less primary PET. The composition is much the same as R2: roughly 50% reused, 30% recycled, 20% virgin, against 72% virgin in 2022.
Loading chart...
Loading chart...
Recycled-content targets run out of domestic material
The second result is more awkward. Recycling reduces reliance on virgin polymer, but unlike reuse or lightweighting it does not reduce how much material must be produced — recyclate still has to come from somewhere.
Until 2030, beverage bottle recycling produces a surplus other applications can absorb. After that the arithmetic turns: mandates keep climbing while recycling yields hit a ceiling. Even at a 90% sorting rate, reprocessing losses cap the overall yield for beverage bottles at about 66%.
The balance between recyclate demanded and produced turns negative by 2035 at the latest, reaching −0.8 Mt by 2050. At that point no packaging category can supply itself: trays are short by 0.5 Mt even assuming tray-to-tray recycling improves dramatically from a 2022 base where it barely existed, and even beverage bottles run a small deficit.
Closing that gap means importing recyclate, investing well beyond what these scenarios assume in tray collection and sorting, developing chemical recycling — or falling back on virgin PET.
Loading chart...
Where Europe's PET dependence actually sits
Europe barely trades finished packaging: gross imports run at about 10% of demand and, because exports are similar, net imports are only 2–3%. Bottles and trays are bulky and cheap to make locally.
The dependence is upstream, in the polymer itself. The EU is a persistent net importer of PET in primary form, with gross imports around 25% of total PET demand — 1.32 Mt in 2030 rising to 1.60 Mt in 2050 — against exports of only 0.17–0.22 Mt. Net primary imports climb from 1.14 Mt to 1.38 Mt, a dependence of about 20–21% of demand throughout.
This is the same pattern the steel study found: the structural dependence sits at the material stage, not the product stage. It is therefore the stage at which circular measures would have to act.
The dependence ratio only moves if production is defended
Both circular scenarios cut PET demand on a widening basis. R2 reduces packaging demand by 0.24 Mt (about 4%) in 2030 and 1.48 Mt (22%) by 2050. R3, where lightweighting compounds reuse, cuts 1.40 Mt (26%) in 2030 and 2.55 Mt (39%) by 2050.
Yet under the default trade assumption, Europe's dependence barely changes. Imports and domestic production contract together, roughly in proportion to their existing shares — and here, unlike the steel case, exports fall too, because the domestic supply behind them shrinks. Absolute imports do drop: under R3, net primary imports fall to 0.86 Mt in 2050 against 1.38 Mt in the baseline. But net import dependence stays close to 21% of demand in every scenario. Europe buys less because it needs less, not because it depends less.
A genuine reduction appears only under the protective variant, where domestic primary production is held at its baseline path and the whole demand reduction is forced onto trade. Then the effect is substantial and lasting: under R2, net primary imports fall from 1.14 Mt to 0.90 Mt in 2030 and approach balance by 2050. Under R3, the EU becomes a net exporter of primary PET around 2030, reaching roughly 1.05 Mt of net exports by 2050 — against 1.14 Mt of net imports in the baseline.
This is the mirror image of the steel finding. There, imports moved almost one-for-one with demand, so a demand cut improved the trade balance automatically. For PET, the cut is shared between imports and domestic supply, so the balance only improves if domestic primary production is deliberately defended.
Loading chart...
The recyclate catch
One complication ties the two halves of this story together.
The global trade model (REMIND-MFA) cannot tell recyclate from virgin polymer — customs data do not separate them. But the EU material flow model (EU MFA) calculates Europe's recyclate needs directly, and finds the EU becoming a net importer of recyclate from the mid-2030s under every scenario, including the baseline: around 0.79 Mt by 2050 under R1, 0.83 Mt under R2, 0.72 Mt under R3.
This puts a hard limit on the strategic-autonomy case. Where domestic production is defended and the whole demand cut falls on imports, modelled primary PET imports drop to zero from 2033 under R3. But total imports cannot fall below what is needed to bring in the recyclate — and from 2033 under R3, the recyclate requirement already exceeds modelled primary imports. That variant is an idealised upper bound rather than an achievable outcome, unless collection, sorting and reprocessing capacity scale up in step.
What this means
PET was the favourable case — circular measures acting on a whole material system rather than a fraction of one. Even here, circularity changes what Europe imports more reliably than how much it depends on imports.
Reuse does the heavy lifting; recycling shifts the problem. Reuse and lightweighting genuinely reduce the material that must be produced. Recycled-content targets reduce reliance on virgin polymer but replace it with reliance on recyclate Europe cannot currently produce enough of. Circularity may relocate import dependence rather than remove it, unless secondary material capacity expands alongside the targets.
The trade gain is a policy choice, not an automatic consequence. Whether reduced demand shows up as reduced dependence or merely as reduced throughput depends on whether domestic production is defended. That is a decision about industrial policy, not a property of circularity.
Circularity indicators need care. Measured by Circular Material Use Rate — the share of recycled and reused material in total demand — R3 scores slightly worse than R2, despite using considerably less material overall. Ratio-based metrics can penalise the strategy that shrinks the denominator, and need reading alongside absolute flows.
Limitations
Recyclate trade is not modelled directly. Customs data do not distinguish recyclate from virgin polymer, so the trade scenarios cannot show how recycled-content targets reshape trade. The recyclate import need reported here comes from the material flow model instead.
No prices, no rebound. The analysis tracks physical material and trade flows only — no energy, no emissions, no economic behaviour. The lightweighting scenario ignores any rebound effect, where cost savings drive additional consumption that offsets the material saved.
Recycling is modelled within packaging types. Loops close within each application — bottle to bottle, tray to tray — with no market for trading recyclate between them. Before 2030 this may make the recyclate picture look better than reality, given the strain on high-quality bottle recyclate observed in the early 2020s.
Scenario uncertainty. Results depend on assumptions about technology adoption, policy implementation and future demand. Stakeholder validation strengthens their standing but does not remove the uncertainty.
How this was modelled
Two linked models were used.
The EU MFA module is a dynamic material flow analysis model covering steel, plastics and cement. For this case study the plastics sub-module was extensively rebuilt. It now tracks materials through mechanical recycling loops and reuse cycles simultaneously, with user-set limits on how many of each a given polymer in a given application can undergo — here, a maximum of 7 recycling loops (after which PET chain length degrades below bottle and tray quality) and 20 reuse cycles. Rather than assuming the effect of circularity on virgin demand, the model allocates reused products and recycled material against final demand and derives virgin polymer requirements as the remainder. It runs in three-month time steps, aggregated to annual figures for reporting.
Scope: PET only, across single-use and reusable beverage bottles, single-use and reusable non-beverage bottles, trays, flexible packaging and other formats. Region EU27+3 (EU27 plus the UK, Norway and Switzerland). Note: the steel companion study uses EU27+1 (EU plus the UK), following that module's regional definition.
REMIND-MFA is a global dynamic material flow model covering 12 world regions to 2100. It adopts the EU MFA's flows for the EU region rather than projecting them independently, then computes EU trade endogenously within a globally market-clearing trade model. Population and GDP follow SSP2; plastics history runs to 2019.
Two trade responses were modelled. Under demand-following trade, Europe's import share at each stage stays at recent historical levels, so lower demand contracts imports and domestic production in proportion. Under fixed-supply trade, applied at the primary/polymer stage, domestic production is held constant and trade absorbs the entire adjustment. Within the fixed-supply case, the split between cutting imports and expanding exports changes the composition of trade but not the net balance.
Data and citation
- Scenario results are available on IAM PARIS.
- Models: EU MFA · REMIND-MFA · Region: EU27, UK, Norway and Switzerland (EU27+3) · Years: 2027-2050 · Socioeconomic pathway: SSP2
- Full methodology: TRANSIENCE Deliverable D8.1 (will be soon available on the TRANSIENCE website)
- Licence: CC BY 4.0 — charts and text may be reused with attribution.
This explainer accompanies work carried out under the TRANSIENCE project, funded by the European Union. The underlying deliverable has not yet been approved by the European Commission and its findings may be revised.
The AI assistant Claude (Opus 5, Anthropic) was used to support the writing of this explainer. All findings, figures and interpretations derive from Deliverable D8.1 and have been reviewed and approved by the authors, who take full responsibility for the content.