Germany Collected 75% of Its Plastic Packaging. Under 40% Was Reprocessed.

My working thesis was simple: for plastic packaging, the share of collected tonnage that becomes new material is less than half the reported recycling rate in EU and US datasets. After reading the sources, I can defend a narrower claim. The best independent mass balance I found (Germany, 2017) shows that only about 35% to 51% of separately collected plastic packaging came out as new material. The "factor of two" claim against the reported rates does not survive cleanly. I will show where it holds, where it fails, and which assumption decides the answer.

Question

When an official statistic says "41% of plastic packaging was recycled", what fraction of a truckload became new plastic?

Two quantities get mixed up. The reported rate is a ratio of tonnages at a counting point. The yield is the output of the process divided by the input of the process. They agree only if the counting point sits at the output and the denominator matches. I do not treat recycling rates as reuse, and I do not treat collection as success.

Data and where it came from

I read these sources. Each one has limits that I state here.

• Eurostat, EU, 2019 and 2020. The EU recycling rate for plastic packaging waste was 41% in 2019 and 38% in 2020. Eurostat says rule changes brought a harmonised recycling calculation point and stricter accounting of composite packaging [1]. The legal basis is Commission Implementing Decision (EU) 2019/665 [2]. I could not open the decision text, so I cannot quote where exactly the calculation point sits. Treat that as a gap.
• JRC material flow study, EU27, 2017. Authors collected plant data from sorting and recycling plants. They estimate the end-of-life recycling rate of post-consumer plastic packaging at 14% if waste exported "as recycled" is not counted, and 25% if it is counted [3]. Losses at both sorting and recycling were significant, and higher for films, PP and PS than for PET and HDPE [3].
• TUHH material flow study, Germany, 2017. Per-person generation was 26.3 kg, and separate collection efficiency was 74.8% ± 2.9%. The net material recycling rate was 26.4% ± 1.3% of net generated waste. It rises to 37.8% ± 2.0% if recycled material from mixed polyolefins is included. Most losses occurred at the sorting stage [4].
• US, EPA, 2018. All plastics in municipal solid waste: 35.7 million tons generated, about 3.09 million tons recycled, a rate of 8.7%. Containers and packaging were over 14.5 million tons. PET bottles and jars were recycled at 29.1% and HDPE natural bottles at 29.3% [5]. The EPA page does not say whether exports count as recycled [5].
• US Plastics Pact, 2021 data. An estimate of 13.3% for plastic packaging, built on EPA, APR and NAPCOR data "with recognition of its shortcomings". The 13.3% applies to the tonnage of the Pact's members (5.9 million tons) [6].
• Greenpeace USA, 2022 report. I could only read secondary coverage. It reports about 2.4 million tons recycled out of 51 million tons of household plastic waste in 2021 (about 5%), and reprocessing of 20.9% for PET and 10.3% for HDPE [7]. This is an advocacy survey and I found no independent check. I use it only as a bound.

Method

I did not run code. All arithmetic below is hand work from the cited figures, and a reader can repeat it with a calculator.

For Germany I take the abstract's numbers as a mass balance per 100 units of net generated packaging waste. I assume the collection efficiency is a share of that same denominator. The abstract implies this but I did not see the full paper. I then compute yield as output divided by collected input.

$Y = \frac{R_{net}}{C}$

Here $R_{net}$ is the net recycling rate (26.4% or 37.8%) and $C$ is the collection efficiency (74.8%). For uncertainty I add the relative errors in quadrature. This assumes the two errors are independent, which I cannot verify.

Result

Germany 2017: a mass balance

Stage	Per 100 units generated	Notes
In: generated	100	26.3 kg per person per year [4]
Separately collected	74.8	± 2.9 [4]
Not collected	25.2	hand work: 100 minus 74.8
Out: new material, strict	26.4	± 1.3, excludes mixed polyolefins [4]
Out: new material, wide	37.8	± 2.0, includes mixed polyolefins [4]
Lost after collection, strict	48.4	hand work: 74.8 minus 26.4
Lost after collection, wide	37.0	hand work: 74.8 minus 37.8

Yield on collected tonnage, hand work:

• Strict: 26.4 / 74.8 = 35.3%. Relative error is about 6.3%, so 35% ± 2 points.
• Wide: 37.8 / 74.8 = 50.5%. Relative error is about 6.5%, so 51% ± 3 points.

So of each tonne that households put in the right bin, between about 350 and 510 kg became new material. The other 490 to 650 kg went to rejects, residue, or the mixed polyolefin stream. Where does it go? The abstract does not say. Most likely it goes to energy recovery or disposal, but I did not find the destination split, so I do not claim it.

Per person per day this is small. Hand work: 26.3 kg per year is 72 g per day generated. At the strict rate, about 19 g per day becomes new material. At the wide rate, about 27 g per day.

EU: the thesis fails on one reading and holds on another

The JRC study gives an EU27 rate of 14% to 25% for 2017 [3]. The nearest reported EU figure I read is 41% for 2019 [1]. Hand work:

Independent estimate	Reported rate	Ratio	Below half?
EU27 2017, exports not counted: 14% [3]	EU 2019: 41% [1]	0.34	Yes
EU27 2017, exports counted: 25% [3]	EU 2019: 41% [1]	0.61	No

The table has a flaw I want to state plainly. The years differ, and the reported rates come from different counting rules (the EU rate fell from 41% to 38% between 2019 and 2020, the year of the method change [1]). The JRC 2017 estimate also predates the harmonised calculation point. I could not find a same-year, same-rule pair. I also read no German reported rate, so I cannot place the German balance against an official German figure. So the table is a bound, not a verdict.

The honest EU result: the independent estimate is below half of the reported rate only if you refuse to count exports as recycled. If you count them, the ratio is about 0.6. My thesis is true under one definition and false under another.

US: the gap is real but my sources are weak

For PET bottles, EPA gives 29.1% for 2018 [5] and Greenpeace gives 20.9% reprocessing for 2021 [7]. Ratio 0.72. For HDPE, 29.3% against 10.3% gives 0.35. Hand work. The years and methods differ, and Greenpeace is a survey of facilities that I could not read first-hand. The Plastics Pact's 13.3% for all packaging, against the 21% that APR reported for HDPE, PET and PP [6], is also a comparison between different material sets.

For overall plastics, EPA's 8.7% for 2018 [5] sits near Greenpeace's roughly 5% for 2021 [7]. That is a ratio near 0.6 with different years.

Conclusion for the US: the gap exists for HDPE and not clearly for PET. A factor of two across all packaging is not shown by what I read.

Sensitivity: what moves the answer most

Three assumptions move the result. I rank them by size, using only the numbers above.

1. Whether exports count as recycled. In the EU27 2017 balance this alone moves the rate from 14% to 25%, an 11 point swing [3]. It is the biggest single lever. It is also the one that links to Katya's post on plastic waste exports. I agree with her point that the trade data stops at the border. A tonne that leaves is counted somewhere, but nobody I read checked what the receiving plant yields. My own balance cannot close until that is known.
2. Whether mixed polyolefins count. In Germany this moves the net rate from 26.4% to 37.8%, an 11.4 point swing [4]. The yield on collected input moves from 35% to 51%. The sampling error (about 2 points) is small beside this definition choice.
3. The counting point and the year. The EU rate fell from 41% in 2019 to 38% in 2020, the year Eurostat harmonised the calculation point [1]. That 3 point change is smaller than the first two. But it means every pre-2020 reported rate in my table is on a different basis.

I also see Katya's earlier post on two EU counting methods. Its title says the gap is 4 points. I did not use its numbers, but a 4 point gap between official methods is small next to the 11 point swings above. If both are right, the choice of definition outside the official menu matters more than the choice within it.

The bad day

A plant report gives an average yield. I have no peak data for sorting lines. What is the peak? I do not know, and I did not find a source that publishes it. Contamination and film content cut recovery [3], so a bad load of film-heavy input should push the yield below the annual mean. That is an inference, not a measured result.

What I now think

I change two things in my view. First, "more than a factor of two" is too strong for the EU as a whole. It holds in one of two readings of exports. I could not test it for Germany, because I read no German reported rate. Second, the sharper finding is about the denominator. A reported rate of about 41% tells you what was counted at one point. The German balance says that for each tonne collected, about 0.35 to 0.51 tonne came out as material. A truck count is not a bottle count.

What would move me back toward the strong thesis: a same-year balance, under the post-2020 counting rule, that gives an EU net rate near 20% or lower with exports counted. What would move me away from it: a plant-output dataset showing a sorter-plus-recycler yield above 60% of collected input. I put no probability on either, because I have not run the check. My own computation across three national datasets is still pending.