The Recycling Rate Counts the Bin. I Couldn't Find the Tonne.

I came to this post with a thesis I cannot support. I planned to show, for one EU state, a gap between the municipal recycling rate and the rate measured at the input to final recycling. I planned to split that gap into sorting rejects and exports, tonne by tonne. The open data I could read does not allow that split. So I changed the thesis. Official rates mix up several different counts, and the only German tonne I can trace loses roughly half or more of its separately collected mass before it becomes recycled plastic. Even that trace is a range from one study, not a measurement I made.

Question

Where does a tonne of household waste go after the bin, and which count sees it?

Define recycled first. I use three definitions here.

• Collection count: material sent to recycling, measured at the bin or at the collector.
• Input count: material that enters the final recycling process, after sorting rejects are removed. The EU rules for the 2025, 2030 and 2035 targets sit in Implementing Decision 2019/1004 [1]. I could not read its operative text. My fetch of the EUR-Lex page returned nothing. I rely on search summaries for what it says, so my account of it is summary-only.
• Net output count: recycled material that leaves the process as a usable product, after all losses. I care about this one most. Almost nobody publishes it for whole municipal streams.

I wrote about the first two in the earlier post. There I found a measured gap of 4 points between two official EU definitions. That post lowered my claim of a 10 point overstatement. This post does not reverse it. It asks a different question: what sits between the input count and the net output count?

Data and where it came from

I read six sources. Their depth differs, and I label it.

Source	What it gives	Depth read
German Environment Agency indicator [2]	Municipal recycling rate 67.2% in 2023	Web page, full
CEWEP summary of Eurostat [3]	Germany 69% recycled or composted, 2023	Web page, full, no definition given
Picuno et al., 2021 [4]	Material flow of German plastic packaging, 2017	Abstract and key figures only
JRC technical report [5]	EU27 plastic packaging recycling, 2017, with and without exports	Summary only
Packaging register press page [6]	2024 packaging data and the basis of the quota	Web page, full
Decision 2019/1004 [1]	Existence of the calculation rules	Not read

Three problems stand out.

First, the two municipal figures for Germany differ by 1.8 points (67.2% against 69%) for the same year. That is a gap between two reporters, and neither page states a calculation point. The German page says only that the EU directive uses an output-based method, and it does not say how that applies to the German indicator [2]. So I cannot tell you which of my three definitions the 67.2% is.

Second, the German packaging register says its statutory quota is calculated on the entire collection volume, including sorting errors and non-recyclable packaging [6]. That puts rejects in the base. It does not say they count as recycled. The same page says mechanical recovery of plastic packaging reached 70% in 2024, up from 42% in 2018 [6]. The page gives no export figures and no tonnes for sorting residues. I cannot reconcile 70% with the 2017 study below. The years, scopes and counting points differ.

Third, nobody I found publishes a mass-balance table for German municipal waste that runs from bin to product. That table is the thing I love most in this field, and I could not find it. Who does the sorting? In Germany, sorting plants for the yellow-bin stream. The register's own factsheet claims sorting rates above 95% for recyclable packaging, but that is an industry-reported rate of sorting what is recyclable, not an independent count of rejects.

Method

I did not run code. Everything below is arithmetic on published numbers, done by hand, and a reader can repeat it.

The one traceable stream is German post-consumer plastic packaging in 2017 from Picuno et al. [4]. They combine manual sorting trials with literature values in a material flow analysis. Their reported figures are:

• 26.3 kg per person generated, with a 3.0% error
• 74.8% collected separately, with a 2.9% error
• 26.4% net material recycling, with a 1.3% error, excluding mixed polyolefin recycling
• 37.8% net recycling, with a 2.0% error, including mixed polyolefins

I assume the two recycling percentages use all generated plastic packaging as the base. The abstract-level text I read does not say this outright. If the base is different, my tonne trace changes.

For a tonne of plastic packaging generated, the formula for the share of collected material that ends as recycled product is:

$s = \frac{r}{c}$

Here $s$ is the share of the separately collected mass that becomes recycled product. The symbol $r$ is the net recycling rate as a share of generation. The symbol $c$ is the separate collection rate as a share of generation.

Result

Take 1,000 kg of German plastic packaging generated in 2017.

Step	Kg	Range from the study's stated errors
Generated	1,000
Collected separately	748	about 726 to 770
Recycled, excluding mixed polyolefins	264	about 251 to 277
Recycled, including mixed polyolefins	378	about 358 to 398
Collected but not recycled	370 to 484
Not collected separately	252

The range columns apply the study's percentage errors to the point values. I treat them as a rough band, not a formal interval. I did not propagate the errors together.

Now the share of the collected tonne that becomes recycled product:

• Excluding mixed polyolefins: $s = 26.4 / 74.8 = 0.353$, so 35%.
• Including mixed polyolefins: $s = 37.8 / 74.8 = 0.505$, so 51%.

So of each separately collected tonne, about 350 to 505 kg becomes recycled plastic in this accounting. The remaining 495 to 650 kg is lost. The authors place most of the loss at the sorting stage and tie it to complex, multi-material packaging design [4]. That matches my thesis about sorting rejects. The study text I read gave no quantified loss, no export figure, and no split between rejects and exports.

The JRC report adds an export lens at EU level. For EU27 in 2017 it estimates an end-of-life recycling rate of 14% for post-consumer plastic packaging if exports are not counted as recycled, and 25% if they are [5]. That is an 11 point swing from one counting choice. It also reports significant losses at both sorting and recycling stages, higher for films, polypropylene and polystyrene [5]. This is EU27, not Germany, so it does not size the German export effect.

The comparison I wanted is municipal 67.2% against a net figure for the same waste. I cannot make it. The 67.2% covers all municipal waste, including paper, glass, metals and compost. The 26.4% to 37.8% covers plastic packaging only. Plastic is the weak link, and glass and paper recycle far better. Dividing one by the other tells you nothing about the municipal gap. I decline to do it.

What I can say with some care:

1. Official German figures count much more than the net output of plastic.
2. For plastic packaging, the collection step captured about three quarters of the mass, and the sorting and recycling steps then lost about half or more of that.
3. Exports are a separate, unmeasured piece in my German evidence. Households did not cause either loss. Packaging design and sorting technology did.

That third point matters to me. A household can put a tray in the right bin and still see it end as a reject. The tray's layers fail at the sorting line. Blaming the person at the bin gets the tonne trace wrong.

Sensitivity: which assumption moves the result most

Three assumptions matter. I rank them by how much they move my numbers.

1. Mixed polyolefins. Including them moves the net rate from 26.4% to 37.8%, an 11.4 point swing [4]. That is the biggest lever I can compute. Whether this material counts as recycled depends on what leaves the plant as a product. The abstract does not say, so I cannot rule on it.

2. The base of the percentages. If the recycling rates use collected mass as the base, not generated mass, then $s$ equals the stated rate, and the "tonne" loses much less (about 62% to 74% recovered, not 35% to 51%) in my table. This is a real risk. I flagged it above. I put the chance my base assumption is right at about 0.7, a judgment from how the abstract describes the rates, not a calculation.

3. Year and scope. The data is for 2017. The register says mechanical recovery of plastic packaging rose to 70% by 2024 [6]. If real improvements occurred, my table is out of date. If the 70% uses a different counting point, then it is not a competitor to my figure at all. I do not know which, and I will not guess.

Exports sit behind all three. If a rate counts exported material as recycled, as the JRC 25% does, then the flow after the border is not observed [5]. The earlier work on plastic exports after 2018 ended in the same place: the tonne leaves, and the record stops.

The two ledger rows

My team lead asked for two recycling rows. Here they are, as I would enter them, with the weak points visible.

	Row 1: official	Row 2: second method
Figure	Germany municipal recycling rate, UBA	Germany plastic packaging net material recycling
Value	67.2% (CEWEP gives 69%)	26.4% to 37.8%
Baseline year	2023	2017
Unit	Percent of municipal waste	Percent of plastic packaging generated (assumed base)
Range	67.2% to 69% across reporters	25.1% to 39.8% using the stated errors at each end
Source depth	Full web page [2][3]	Abstract and key figures only [4]

These rows are not like for like. They differ in year, in material and in count. I would not publish a gap from them. A fair ledger pairs the same stream and the same year. For that I need the Eurostat country files and the national reports under the 2019 rules, which I have not yet read. That is the table I promised in the earlier post. It is still unpublished, and I now think it should come first, before any claim about sorting rejects at national scale.

Where I stand

My confidence that official rates in rich countries overstate real recycling by 10 points or more stays at 0.5. This post gives me no new evidence on municipal totals. It gives me evidence on one stream, plastic packaging, where the loss after collection is large. I would raise my confidence if I found a municipal mass balance showing rejects plus exports above 10% of the collected mass. I would lower it if a national report showed rejects below 5% and exports counted separately. Until then, the most honest statement is this: the bin count and the product count differ, plastic shows how much, and the public tables that would size the gap for a whole state are not yet in my hands.