The plastics treaty rules provide for a two-thirds vote as a last resort, and in 2023 the committee recorded that members dispute it

What is happening?

Two different numbers circulate for how much plastic is recycled, and they measure different things.

The OECD Global Plastics Outlook of February 2022 reports that of 353 Mt of plastic waste generated in 2019, 33 Mt was recycled, a share of 9 percent, while 19 percent was incinerated, almost 50 percent went to sanitary landfill and 22 percent to uncontrolled dumpsites. The same report gives a second figure on a different denominator: secondary plastics from recycling reached 29.1 Mt in 2019 against total production of 460 Mt, which the report states as 6 percent of total feedstock. The first number is the share of discarded plastic that gets recycled. The second is the share of newly made plastic that is made out of recycled material.

Neither number has been flat. On the OECD modelled series the world share of plastic waste recycled rose from 3.70 percent in 2000 to 9.29 percent in 2019, and the OECD restates the same measure as 9.5 percent for 2020. Secondary plastics rose from 6.8 Mt in 2000 to 29.1 Mt in 2019, while production over the same period rose from 234 Mt to 460 Mt.

What is flat is a projection rather than a record. In the OECD baseline published in October 2024, plastics use and waste generation rise 70 percent from 2020 levels by 2040, use going from 435 Mt to 736 Mt, and the recycled share stays unchanged at 6 percent of total plastics use, which is 41 Mt in 2040. Leakage to the environment rises by 50 percent, to 30 Mt in 2040. The 70 percent belongs to use and waste. The 50 percent belongs to pollution. They are separate figures in the same report and the OECD topic page still leads with the 50.

The instrument meant to change that trajectory has not been concluded. The fifth session of the intergovernmental negotiating committee, resumed in Geneva, adjourned early on 15 August 2025 without consensus on a text, with more than 1,400 delegates from 183 countries in the room, and the committee agreed to resume at a future date to be announced. The Chair of the committee stepped down in October 2025. The Bureau met on 30 October 2025 and set a one-day resumed session, which was held in Geneva on 7 February 2026 for organizational and administrative purposes only, elected a new Chair and held no substantive negotiations. Through 2026 the committee has run an intersessional programme, and as of 2026-08-19 no date or venue for the next negotiating session has been announced.

The rules the committee applies are not a consensus mandate. Rule 38(1) of the draft rules of procedure directs the committee to make every effort to reach agreement on matters of substance by consensus, and provides that where all efforts at consensus have been exhausted without agreement, a decision may be taken as a last resort by a two-thirds majority of the members present and voting. Adoption of those rules was deferred at the first session in late 2022 and they apply on a provisional basis. The bracketed text sits in rule 37, on the voting rights of regional economic integration organizations, not in rule 38. At the second session in 2023 the committee adopted an interpretive statement recording that there are differing views among members on rule 38(1), and that if it were invoked before the rules were formally adopted, members would recall that lack of agreement.

Whose problem is this?

The body that has to agree the text is the same body that decides whether its own tie-break rule is available, and nothing outside it sets a date for finishing.

Where does this problem end?

The boundary here is drawn around the absence of an agreed text and around a share that is not rising, and not around the question of what the text should contain.

A reader who wants a production cap and a reader who wants none of it can both agree on the dates and the tonnages set out below.

What is the state now, and what should it be?

Now

What it should be

The process states its own condition, and it is not a recycling number. UNEP describes the goal of the Geneva session as agreeing the text of the instrument and identifying unresolved issues that need further preparatory work ahead of a diplomatic conference. So the target state the negotiation itself uses is an agreed text, followed by a diplomatic conference at which the instrument opens for signature.

No body in the negotiation has set a recycling rate to reach. The OECD publishes a modelled figure that reads like one: ending plastic leakage by 2040 would require a global recycling rate of 42 percent in 2040, against 9.5 percent in 2020. That is a requirement inside a scenario and not a target that any state or the committee has adopted, and the same report holds the baseline recycled share at 6 percent through 2040.

How big is it?

Material flows. Production rose from 234 Mt in 2000 to 460 Mt in 2019. Waste rose from 156 Mt to 353 Mt over the same years. Of the 2019 waste, 33 Mt was recycled, 19 percent was incinerated, almost 50 percent went to sanitary landfill and 22 percent to uncontrolled dumpsites. Leakage to the environment was 22 Mt in 2019, and accumulated stocks stood at 109 Mt in rivers and 30 Mt in the ocean.

Regional shares on one basis and one year. On the OECD modelled series for 2019: India 13.3, China 12.8, Europe 12.4, world 9.29, Sub-Saharan Africa 5.7, Middle East and North Africa 5.4, United States 4.49 percent of plastic waste recycled.

Three published figures for one country, on three bases.

These three are not the same measurement and they are not comparable, because they use different definitions and different years and come from different compilers. The federal agency page carries 2018 data as its most recent, so the most recent official national figure covers 2018.

Projections, and their base years are not the same.

The four reports do not share a base year, so their tonnages are not points on one line and they are set out here report by report. Two of them carry a figure of 435 Mt for plastics use, one attaching it to 2019 and the other to 2020, and this document keeps those separate rather than treating them as one value. Production and use are also different quantities in these reports, 460 Mt and 435 Mt for 2019, and they are not merged here either.

Affected population. No count of people is derivable from these sources. They measure tonnages, shares and greenhouse gas emissions, and none of them converts any of those into a number of persons exposed or harmed.

Under what conditions does it arise?

1. The recycled share is a ratio and both of its terms grow. Between 2000 and 2019 secondary plastics went from 6.8 Mt to 29.1 Mt while production went from 234 Mt to 460 Mt. In the OECD baseline for 2040 the share stays at 6 percent of use while use rises 70 percent from 2020 levels, so the tonnage recycled rises to 41 Mt and the fraction does not move.

2. The negotiation has more than one fault line. The negotiating record of the Geneva session names the intractable issues as the scope of the mandate over sustainable production of plastic, financing together with a proposed donor categorization of parties with a capacity, global versus voluntary national measures on plastic products, chemicals of concern, and decision making itself. On the final day a small group worked on a package of production, finance, plastic products and decision making.

3. Two of those lines cross each other. Finance divides developed states from developing states and production divides producers from regulators, and those are not the same division of the room. A coalition assembled on one of them is not the coalition on the other.

4. The tie-break exists on paper and is unresolved in practice. Rule 38(1) provides a two-thirds majority as a last resort, the rules containing it were never formally adopted, and the committee has recorded that members disagree about whether that paragraph can be applied in the meantime.

5. Not agreeing needs no coalition. Four text iterations were on the table when the session adjourned, and the record shows the meeting closing abruptly on the morning of 15 August 2025 before observers could deliver statements, with no clarity as to next steps.

6. The route back to the assembly for a fresh mandate was open and was not used. The seventh session of the assembly concluded on 12 December 2025 with 11 resolutions, three decisions and a ministerial declaration, and the plastics instrument was not among them. The next assembly is set for 6 to 10 December 2027.

What has been tried?

What was found?

Why is it still unsolved?

Coordination failure — the tie-break written into the procedure the committee applies is itself in dispute, so the route that has operated has been agreement, and no agreement is required of anyone.

The first part is the rule. Rule 38(1) provides that once all efforts at consensus have been exhausted without agreement, a decision may be taken as a last resort by a two-thirds majority of members present and voting. That paragraph sits in a set of rules whose adoption was deferred at the first session in late 2022, so the rules apply provisionally, and at the second session in 2023 the committee adopted an interpretive statement recording that members hold differing views on that paragraph. So the practice of deciding only by consensus is sustained by an unresolved procedural dispute rather than by the mandate. Nothing in the record here settles the dispute, and that statement phrases any invocation of the paragraph as a future contingency rather than as an event that had occurred. What has operated in the meantime is the practice rather than the paragraph.

The second part is that the disagreement is not one axis, so a majority on one question is not a majority on another. Mandate scope over production, financing and the categorization of who pays, measures on plastic products, chemicals of concern and the decision rule itself were all live at the same time, and the group working through the final night carried four of them as a single package. Two states that are far apart on production can be on the same side on finance, and the result of that arithmetic is another meeting.

The third part is that no source here records a consequence attaching to delay. The session adjourned, the committee agreed to resume at a date to be announced, and an intersessional programme filled the following year. None of the sources below identifies a default text that takes effect if no agreement is reached, a body that gains the drafting power instead, or a deadline after which anything changes. A negotiation that has adjourned and a negotiation that has stopped look the same from outside and are not the same thing.

The fourth part is that the escalation route was open and was not used. The assembly that issued the mandate met in December 2025 and adopted eleven resolutions on other subjects. The instrument was not among them, and the next assembly is two years away.

The fifth part is material rather than procedural. The share stands still because recycling and production grow at close to the same rate, and not because recycling stopped growing. Recycled tonnage rose from 6.8 Mt to 29.1 Mt between 2000 and 2019 while production rose from 234 Mt to 460 Mt, which the OECD describes as a doubling, and the OECD baseline still holds the recycled share at 6 percent of use in 2040 while use rises 70 percent from 2020 levels. A policy that raises recycling without touching the denominator moves the tonnage and leaves the fraction where it is.

What observation would mean it is solved?

Candidates — (a) the committee agrees the text of the instrument (b) a diplomatic conference is held, the instrument opens for signature and enough states ratify for it to enter into force (c) the recycled share of plastics use or the annual leakage tonnage departs from the OECD baseline.

(a) is the condition the process itself names. UNEP describes the goal of the Geneva session as agreeing the text and identifying unresolved issues needing further preparatory work ahead of a diplomatic conference. It is also the observation that can be satisfied at the lowest price, because a text can be agreed with voluntary national measures, with the contested subjects deferred to later instruments, or with timelines long enough that nothing moves before 2040.

(b) counts an instrument rather than a material outcome. Entry into force binds the states that ratify. None of the sources here estimates how many of the 183 countries present in August 2025 would ratify, or whether the largest producers and the largest recipients of leakage would be on the same side of that line.

(c) is the only one that measures the thing the instrument exists to change, and it is the slowest. The OECD baseline puts the recycled share at 6 percent of use in 2040 and leakage at 30 Mt in that year, so a departure from either figure is observable against a published projection.

No body in the negotiation has set a recycling rate or a leakage tonnage to reach, so the third observation has to be read against a projection rather than against a target.

What is it connected to?

A secondary account records that the voting rules of the 1992 climate convention were left in brackets in 1991 and have not been adopted in the thirty-five years since, and that majority voting is available in the High Seas Treaty, in the Basel, Rotterdam and Stockholm conventions and in the Convention on Biological Diversity. The shape it points at is that an unadopted decision rule can persist for decades inside a multilateral body without anyone ever deciding that it should.

That account is graded medium because it describes the plastics rule less precisely than the note by the secretariat does, placing the brackets on the decision paragraph where the primary document places them on the rule about regional economic integration organizations. On the plastics facts the primary document governs.

The subjects that sit next to this one without an established relation are the petrochemical feedstock market, the international trade in plastic waste, and the design of financial mechanisms in multilateral environmental agreements. No source listed below settles a relation type or an evidence grade for any of the three.

Fills with researchone comparison is on the record and is set out below at medium grade. It is the only one, and it is a comparison of procedure rather than an established relation to another problem.

What these sources do not say

• No national plastic recycling rate for most states. The OECD modelled series carries the world, three countries and six regions. A national figure for any other state has to come from a different compiler on a different basis, and none of the sources here establishes that any of those bases is comparable with the OECD series.
• No date or venue for the resumed fifth session. The UNEP page carrying the intersessional programme was last updated on 2026-08-19 and gives dates only as far as a third informal in-person meeting inside the window from 25 January to 2 February 2027.
• No explanation, attributed to any party, of what would close the remaining differences. The record carries that the session adjourned without consensus, that four text iterations were on the table, and that the committee agreed to resume.
• No resolution of whether rule 38(1) can be applied. The committee recorded that members hold differing views and that the lack of agreement would be recalled if the paragraph were invoked before the rules are formally adopted. No source here records the paragraph being invoked, and none records the dispute being settled.
• No source states that a reduction in plastic production by one country is offset by an increase in another. The proposition circulates in discussion of emissions policy. None of the sources here makes it for plastics.
• No population or disease-burden figure for exposure to plastic pollution. These sources measure tonnages, shares and greenhouse gas emissions. None of them converts those into a count of persons exposed or harmed.
• No official United States figure for any year after 2018. The federal agency page carries 2018 data as its most recent, so the more recent numbers in circulation come from a modelled series and from an advocacy estimate rather than from the government.
• No base year for the 2050 projection. The policy highlights read here give the 2050 endpoints and the starting tonnages without naming the year those starting tonnages belong to.

See the evidence

The load-bearing figures here come from publishers opened directly. The OECD policy highlights documents carry the production, waste, recycling, leakage and emissions figures and the projections to 2040 and to 2060, and the OECD topic page repeats the 2040 leakage and emissions pair. The UNEP pages carry the outcome of the Geneva session, the one-day resumed session, the change of chair and the intersessional programme, and the note by the secretariat carries the rules of procedure, the deferral of their adoption and the interpretive statement. The federal agency page carries the official United States figure. Where the record is unsettled it is left visible rather than resolved: the four OECD reports do not share a base year and are set out report by report rather than merged, the 2050 projection was reached on a national ministry mirror because the OECD surface returns an automated refusal and is graded medium for that reason, the itemisation of the disputed subjects at Geneva rests on one specialist bulletin, and the procedural precedent from the climate convention rests on one secondary account that describes the plastics rule less precisely than the primary document does. Several of the hosts listed above return an automated refusal to unattended requests while serving the same pages to a browser, and that refusal is a property of the host rather than of the page. The evidence tier for this file is secondary because no source here is linked to an observation record, and not because the documents above went unread. This is a research-based definition, so observation_refs is empty and provenance_mode is press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below