What does the field still not know?¶

Thirty-four years after Germany's first packaging ordinance, extended producer responsibility operates in most of Europe, seven American states, ten Canadian provinces and much of Asia and Latin America. It moves billions of dollars annually. And the central empirical questions about it remain open — not contested between rival findings, but genuinely unmeasured.

This article is not a summary of the doubts recorded elsewhere in this library. It is an attempt to do something the field has not done for itself: to sort what is unknown into a priority order, to distinguish questions that are hard from questions that are merely unasked, to census the literatures that do not exist at all, and to say for each gap what would actually close it and what closing it would cost. The organising claim is that the field's evidentiary problem is not principally one of difficulty. Most of these questions are answerable with instruments that are cheap relative to the programmes they would evaluate. They remain open because nobody with the power to answer them has an interest in the answer.

1. First, a distinction the field keeps collapsing¶

There are three different kinds of not-knowing, and treating them as one produces most of the confusion in EPR debate.

Contested: rival parties have produced conflicting evidence and the dispute is live. Whether fee modulation changes packaging design is contested — there are documented instances and documented nulls.

Unmeasured: nobody has produced the evidence, in either direction. Whether operational transfer outperforms municipal compensation is unmeasured. No one has run the comparison, because no jurisdiction preserved the data that would allow it.

Unmeasurable as currently constructed: the question cannot be answered without changing something structural first. Whether British Columbia outperforms Ontario is unmeasurable while the two provinces report on different bases (the measurement article) — not because the answer is hidden, but because the numbers do not refer to comparable quantities.

Most of what follows is in the second and third categories. That matters, because contested questions are resolved by more research and unmeasured ones are resolved by instrumentation — by deciding, before the fact, to collect something.

2. Tier one: the three questions that determine whether EPR works¶

2.1 Does the fee reach the design decision?¶

This is the mechanism question, and everything else is downstream of it. EPR was justified — in the original Swedish formulation and in every legislative debate since — on the ground that making producers financially responsible for end-of-life would feed back into what they make. If that feedback does not occur, EPR is a financing arrangement with a design story attached.

The OECD's position (Laubinger, Brown, Dubois and Börkey, 2021) is that "there is as of yet limited evidence that EPR schemes have triggered" design for environment, and that collective schemes' fee schedules "provide weak incentives for design change." The Dutch analytical agencies PBL and CPB concluded independently the same year that "there is little evidence to date of EPR instigating eco-design."

What is genuinely unknown: whether that is a statement about the instrument or about its calibration. No study has isolated a fee differential large enough to matter and observed whether producers responded. The documented instances of apparent design response — Italy's move to recyclability-banded fees coinciding with pigment substitution, the Netherlands doubling its recycled-content discount because it judged the signal too weak — are suggestive but confounded (Theme 5).

What would close it: a differences-in-differences study exploiting a discrete, large modulation change in one jurisdiction against a comparable unchanged jurisdiction, on a defined packaging category, with pre-period data. Several such changes have already happened. None was instrumented in advance, so each was wasted.

2.2 What does EPR do to total system cost?¶

The PBL/CPB report names this explicitly: "The overall effect of EPR on waste management costs is unclear. This is one of the most important knowledge gaps in the EPR literature."

That sentence deserves more attention than it has received. It is not an advocacy claim. Two government analytical agencies, reviewing the international evidence, reported that after three decades nobody knows whether EPR makes packaging waste management cheaper, more expensive, or neither in aggregate.

Why it is unknown: total system cost requires municipal costs plus producer costs plus administrative costs, on a consistent boundary, before and after. Transitions typically destroy at least one of those series — Ontario terminated its long-running cost and performance instruments at the moment of handover (the transition article) — and producer-side costs are frequently confidential.

What would close it: dual-basis cost reporting across a transition, for three years either side, on a published boundary definition. This is an accounting exercise, not a research programme.

2.3 Where does the money actually land?¶

EPR moves cost from municipal budgets to producers. Producers pass some share to consumers. Municipalities either return the saving to ratepayers or absorb it. Neither leg of that chain has been measured with any rigour.

The pass-through literature on EPR specifically is thin and largely modelled rather than observed (the pass-through article), and the municipal saving literature is worse — the question of whether households ever see a reduction is close to entirely unstudied (the savings analysis). The critique and its rebuttal share the same gap: opponents cannot show costs reach households, and proponents cannot show savings do.

What would close it: Washington State is about to run the experiment by accident. Because its households pay for recycling through visible utility bills, producer reimbursement reaches ratepayers arithmetically, and the state models household recycling bills falling by at least 90% after 2032. That is the first genuinely falsifiable household-savings claim in EPR's history. It requires only that someone commit now to measuring the bills.

Denmark supplies the other half. Having cut producer fees for 2026, it offers the field's only opportunity to observe a fee decrease against retail prices — almost all existing evidence concerns increases, and the assumption that pass-through is symmetric has never been tested (the European article).

3. Tier two: questions that are answerable now, and unasked¶

These are the cheap ones. Each has natural variation already running; each fails only for want of comparable measurement.

Question	The variation that exists	Why it cannot currently be read
Does operational transfer beat municipal compensation?	Manitoba (80%) and Saskatchewan (75%) against British Columbia, Ontario, Quebec at ~100% with producer-held contracts	Different reporting bases; no province preserved a comparable pre-transition baseline
Does competition between PROs beat a single body?	Germany and Ontario against most of Europe and Canada	Ontario's inter-organisation financial terms are unpublished; no post-transition rate published
Does full cost coverage beat partial?	France (~75%) and Italy (~80%) against Germany, Belgium, the Netherlands (100%)	Cost-coverage denominators are not comparable, and performance is confounded by deposit systems
Does producer responsibility beat a good municipal system?	Nova Scotia, which entered EPR with an unusually strong municipal diversion record	Transition baseline not instrumented; too early
Does a clearing house matter in a competitive market?	Germany has one; Bulgaria, Romania, Slovenia and the Baltics run multi-operator markets without one	No comparative study of free-riding or tonnage integrity across the two configurations

The last row deserves emphasis because it is the field's most under-examined risk. Six EU member states operate competitive producer-responsibility markets without the registry and clearing-house apparatus that Germany built to police market shares and prevent under-reporting. Whether that configuration degrades data integrity is not a subtle question, and nobody has asked it.

4. The absence census: literatures that do not exist¶

Assembling this library surfaced something more striking than any individual data gap — entire bodies of scholarship that ought to exist and do not. Each of these was searched for specifically.

There is no political economy of EPR. The canonical apparatus for analysing why regulated firms want regulation — Stigler on regulatory capture, Yandle on bootleggers and Baptists, Salop and Scheffman on raising rivals' costs, Olson on collective action — has never been applied to extended producer responsibility in peer-reviewed work. For an instrument whose compliance bodies industry publicly asked to control (the industry article), this is the most consequential absence in the field.

There is no legal scholarship on EPR's constitutionality. With packaging EPR under simultaneous constitutional challenge in four American states, the entire analytic corpus consists of law-firm client alerts and trade press. No law review article on the subject could be located. The commentary is competent and it is written by firms with clients.

There is no European accountability critique. Europe has operated producer-governed compliance bodies for three decades. No peer-reviewed work advancing a democratic-accountability critique of that arrangement was located; European scholarly attention to PROs runs almost entirely through competition law. The critique exists only in an American courtroom, in a form that depends on American constitutional doctrine.

There is no systematic review of EPR performance. The dispute over whether EPR raises recycling rates is currently conducted between a small number of individual studies and an industry-funded advocacy report. No independent meta-analysis or systematic review resolving it was located.

There is no serious analysis of item-level individual producer responsibility. Digital watermarking and item-level identification are widely asserted to point toward billing producers for their own recovered items. No academic work modelling that was located, and the closest relevant paper (Ellsworth-Krebs et al., 2022) argues in the opposite direction — for better-calibrated collective fee modulation, on the ground that circularity requires ecosystem coordination rather than individual closure (the frontier article).

There is no independent assessment of AI sorting at commercial scale. Vendor throughput and accuracy claims are unaudited; the peer-reviewed work that exists concerns controlled batches at pilot facilities.

A field can survive missing data. Missing literatures are different, because they determine who gets to narrate the missing data. Where no independent scholarship exists, the interpretive vacuum is filled by parties with positions.

5. Why the gaps persist¶

It would be comfortable to attribute all this to difficulty. The more accurate explanation is structural, and it has four parts.

Transitions consume their own baselines. The moment of handover is when measurement instruments change hands, and the entity acquiring them has no reason to preserve a series that would later be used to evaluate it. Ontario terminated its data instruments at exactly this moment, rendering the largest EPR transition in North American history substantially unevaluable.

Commercial confidentiality is load-bearing. Fee methodologies are confidential; inter-organisation financial terms are unpublished; producer-level data is aggregated before release. Each individual claim of confidentiality is defensible. The cumulative effect is that the variables required for evaluation are precisely the ones withheld.

Nobody's mandate includes evaluation. Regulators approve plans and enforce compliance. Producer organisations report against their own targets. Auditors examine process. No party in the system is charged with asking whether the instrument works, and the one legislative auditor who examined regulator capacity found it wanting (Theme 2).

The parties who could fund research have positions. The research that exists is disproportionately commissioned — by dual-system operators, by packaging associations, by advocacy organisations. This is not an allegation of dishonesty; it is an observation about which questions get asked. Commissioned work answers the commissioner's question.

6. What would close the gaps, and what it would cost¶

The remedies are unglamorous and cheap relative to programme budgets.

Dual-basis reporting across every transition. For three years either side of a handover, report on both the old and new methodology. This is the single highest-value intervention available, it costs one additional calculation, and it would have preserved the readability of every Canadian transition.

Preserved baselines as a condition of designation. Make continuation of the predecessor data series a condition in the producer organisation's approved plan. It is a plan condition, not a statute.

Published inter-organisation terms. Where multiple organisations share infrastructure, publish the access terms. Germany's registry demonstrates that this is compatible with a functioning competitive market.

One instrumented modulation experiment. A single jurisdiction committing, in advance, to a large fee differential on a defined category with pre-period data collection would settle the mechanism question that thirty years of observational work has not.

Measure the Washington bills. The one design in the world that guarantees an observable household effect requires only that somebody commit to observing it.

A dual-basis rate, permanently. Every jurisdiction publishing a recycling rate should publish it on both a collection and a reprocessed-output basis. The United Kingdom accidentally demonstrated the value of this by publishing two 2024 figures eleven points apart in a single release.

None of this is a research programme. It is a reporting specification, and it is the same remedy Theme 6's closing article specifies in more detail.

7. Where the argument stands¶

The honest summary of the field's evidentiary position is uncomfortable and should be stated without hedging.

On the mechanism: it is not established that fee signals reach packaging design decisions in collective schemes, and the two most credible institutional assessments say the evidence is limited.

On cost: nobody knows the aggregate effect on system cost, and two government agencies have said so in print.

On incidence: neither the claim that costs reach households nor the claim that savings do has been demonstrated.

On comparative design: every major design question — transfer versus compensation, competition versus monopoly, full versus partial coverage — has natural variation running and no readable answer.

On the scholarship: the political economy, the constitutional analysis, the accountability critique and the systematic review all do not exist.

What is reasonably well established is narrower than the debate assumes: separate collection rises under EPR; producer organisations can operate at scale without service collapse; and transitions concentrate cost pressure on producers rapidly and visibly. That is a real but modest evidentiary base for an instrument now carrying the weight of global packaging policy.

The uncomfortable part is that this is not a story about a young field awaiting maturity. Germany's scheme is thirty-four years old. The instruments needed to answer these questions were available at every point, and were not built. The provinces and states now designing programmes, and those negotiating their next plan cycles, are the last parties in a position to build them cheaply — and the PPWR's harmonisation of definitions across twenty-seven member states is the best opportunity the field will get to make its numbers mean the same thing. Whether anyone takes it is the open question that determines all the others.

References¶

The named knowledge gap: Dimitropoulos, A., Tijm, J. and in 't Veld, D. (2021), Extended Producer Responsibility: Design, Functioning and Effects, PBL Netherlands Environmental Assessment Agency and CPB Netherlands Bureau for Economic Policy Analysis — including "the overall effect of EPR on waste management costs is unclear. This is one of the most important knowledge gaps in the EPR literature," and "there is little evidence to date of EPR instigating eco-design," both quoted directly.
The design-signal evidence: Laubinger, F., Brown, A., Dubois, M. and Börkey, P. (2021), Modulated Fees for Extended Producer Responsibility Schemes, OECD Environment Working Paper No. 184, quoted directly.
The comparative variation: Canadian provincial designs and reporting bases, and the Nova Scotia baseline point, as set out with sources in the Canada article; European institutional families, the central and eastern European multi-operator markets without clearing houses, the UK's two 2024 rates, and Denmark's 2026 fee reduction, as set out in the Europe article; Washington's billing architecture and modelled household saving, as set out in the US article.
The absence findings: each was searched for specifically in the research underlying the industry article, the critiques article, the litigation article and the frontier article, and each is reported as "not located in the searches conducted" rather than as a demonstrated non-existence. The nearest relevant work on item-level tracking is Ellsworth-Krebs, K. et al. (2022), "Circular economy infrastructure: Why we need track and trace for reusable packaging," Sustainable Production and Consumption 29: 249–258, which argues for collective rather than individualised responsibility.
Institutional evidence on evaluability: Auditor General of Ontario, value-for-money audit of the Resource Productivity and Recovery Authority (December 2025); the termination of Ontario's cost and performance data instruments at transition (July 2025); European Court of Auditors, Special Report 23/2025.

Verification note: this article makes claims about what does not exist. Every such claim rests on targeted searching conducted for this library in August 2026 and is stated as a negative finding of those searches, not as proof of absence — an unindexed, paywalled or non-English literature could exist in any of these areas. Where a gap is asserted, the search that failed to fill it is identified. See Sources and method.