Is EPR regressive? The distributional evidence and its gaps¶
If packaging EPR costs pass substantially into grocery prices — as the theory and analogue evidence reviewed in the preceding article suggest — then the follow-on question is who, across the income distribution, bears them. "EPR is a regressive hidden tax" is now a standing frame in US and Canadian opposition; "EPR relieves regressive municipal charges" is the standard reply. Both claims contain true mechanisms, and the striking fact about this most heavily contested of questions is that no published income-decile incidence study of packaging EPR exists anywhere. This article assembles what is actually known: the regressive mechanism and its arithmetic; the equally regressive counterfactual that most regressivity claims ignore; the missing empirical parameter (packaging intensity by income) on which the whole calculation turns; the better-documented deposit-return sub-case; and the one billing design that would make the distributional outcome observable rather than argued.
1. The regressive mechanism is real¶
The mechanism requires no exotic economics. An EPR fee embedded in grocery prices operates like a small consumption charge on food and household staples, and consumption charges on necessities are regressive on income: lower-income households spend a larger share of income on food — roughly 15% in the poorest UK quintile against an 11% average, on the standard national expenditure data as compiled by the Food and Drink Federation — so an equal percentage increment on grocery prices takes a larger share of a poor household's income than a rich one's.
Three refinements from the tax-incidence literature discipline the claim without dissolving it. Measured against expenditure rather than income, regressivity shrinks — the standard result for consumption levies, since measured income understates the resources of many low-income households. The magnitude bounds the stakes: at fees of 0.4–2% of product price and food at 11–15% of budgets, the fee increment is a small fraction of one percent of household spending — regressive in sign, modest in size, on any published arithmetic. And salience matters distributionally: an embedded fee is invisible, so no household can avoid it through attention or effort — unlike a visible deposit, which the diligent redeem and the constrained forfeit (Section 4).
The industry-side modelling adds a distributional claim worth recording with its provenance: the New York cost study by Calvin Lakhan of York University (released by the state business council) reports that households earning under $40,000 buy roughly 20% more prepackaged goods than higher-income households — which, if representative, would steepen the regressivity beyond the food-share arithmetic. The claim points at exactly the right parameter; its evidentiary basis is the study's own market data rather than a published consumption survey, and no independent estimate of packaging intensity by income exists to check it against (Section 3).
2. The counterfactual is also regressive — and this is the unstated half of every claim¶
Regressivity is a comparison, and the comparison is not between EPR and nothing. It is between EPR and the financing it displaces: municipal waste and recycling services funded from property taxes, flat utility fees, or general revenues. That baseline has its own distributional profile, and it is not progressive.
Property-tax and flat-fee waste financing charges households without regard to income or waste generation; it subsidises high-waste households at the expense of low-waste ones; it is invisible to renters, who pay it through rents without seeing it; and a flat annual charge is more regressive than a proportional price increment, since it takes a strictly larger share of a smaller income. When EPR shifts recycling costs from that base onto product prices, the distributional question is the net of two regressive instruments: a regressive price increment minus the incidence of the displaced municipal charge.
The sign of that net is theoretically ambiguous and empirically unmeasured. A low-income household that consumes little packaged product but paid the same flat waste fee as everyone else plausibly gains from the swap; a large low-income family buying heavily packaged staples in a municipality that never reduces its rates plausibly loses. No study anywhere computes the net. This is the single most important unstated caveat in every regressivity claim on both sides — the opposition's "regressive hidden tax" ignores the regressive charge being displaced, and the advocacy's "relieves municipal burdens" assumes a household-level relief that has never been documented to arrive (the municipal-savings article).
3. The missing parameter¶
The calculation that would settle the question needs one number the literature does not contain: packaging intensity per dollar of spending, by income decile — how much fee-bearing packaging accompanies each income group's consumption basket.
The components exist separately. Household expenditure surveys give spending composition by decile; fee schedules give per-material rates; market data can map products to packaging. Nobody has joined them. The distributional profile could plausibly run either way: lower-income baskets skew toward some packaging-intensive categories (processed and shelf-stable foods — the direction of the Lakhan claim) but also toward categories with modest packaging fees per dollar (bulk staples), while higher-income baskets carry their own packaging-heavy segments (prepared foods, delivery, single-serve formats). Assertions about which effect dominates are, at present, assertions.
The same gap disables the offset side. Whether municipal waste-financing incidence falls more heavily on low-income households than the fee increment does requires knowing both distributions; only fragments of the first (property-tax incidence literature) and none of the second exist for any EPR jurisdiction.
| Component of the net calculation | Status |
|---|---|
| Food/grocery share of budget by decile | Measured (standard national statistics) |
| Fee as share of product price | Measured (0.4–2%; per-item studies) |
| Pass-through rate | Unmeasured for EPR; high on theory and analogues |
| Packaging intensity by income decile | Unmeasured anywhere |
| Incidence of displaced municipal financing | Fragmentary |
| Municipal savings reaching households | Undocumented in every examined case |
| The net distributional effect | Never computed, any jurisdiction |
4. The deposit-return sub-case: better documented, cuts both ways¶
The one corner of this terrain with an official distributional assessment is deposit-return, and it illustrates both edges of the problem.
Scotland's Fairer Scotland Duty assessment of its planned deposit system quantified the burden mechanism: the deposit float — money tied up between purchase and redemption — amounts to roughly £1.40–1.80 per week for the bottom two income deciles, a real liquidity cost for constrained households. The regressive edge is non-redemption: a household that cannot conveniently return containers — no car, no nearby return point, mobility or time constraints — pays the deposit as a tax. Deposits are the visible, effort-contingent version of packaging charges: avoidable through effort that is itself unequally distributed.
The counter-mechanism is equally documented in the North American debates: wherever informal collectors ("canners") gather unredeemed containers, forfeited deposits function as a transfer to some of the lowest-income participants in the economy — a framing prominent in Michigan and California. The DRS case is thus a miniature of the whole question: the same instrument is regressive through one mechanism and progressive through another, and the net depends on local redemption infrastructure and who forfeits to whom. It also carries the general lesson: distributional outcomes are design outcomes — return-point density and float size are choices, not properties of the instrument.
5. The political career of the regressivity claim¶
Because the net is unmeasured, the regressivity claim functions politically as a frame rather than a finding, and its career is worth documenting as part of the state of the debate.
In the United States, "regressive hidden tax" is standing vocabulary: the Tax Foundation maintains the framing across state EPR debates; Americans for Tax Reform's 2026 commentary titled the instrument "a regressive hidden tax Americans will soon know well"; and the industry cost studies supply the distributional garnish (the New York modelling's under-$40,000 consumption claim). In Canada, the Charlebois grocery-inflation estimates — published without disclosed methodology (the pass-through article) — were adopted into federal Conservative messaging as "hidden food taxes." In the United Kingdom, the Food and Drink Federation's framing bundles packaging EPR with deposit return into a £160-per-year policy package equal to "2% of gross income for the lowest earners" — an aggregation across instruments that the government's own £48 EPR-specific figure puts in proportion.
Three features of this career are analytically notable. The claims are almost always sign-only — correct that the increment is regressive, silent on magnitude and on the displaced charge. The aggregation move (bundling EPR with deposits, packaging taxes and other measures) manufactures larger household figures than any single instrument supports. And the counter-messaging has conceded the frame rather than contesting it — no government or scheme has published the decile analysis that would replace assertion with measurement, leaving the field to whoever asserts most vividly. The regressivity debate, in short, is being conducted entirely with the building blocks of Section 1 and none of the arithmetic of Section 3 — on both sides.
6. Design margins that change the answer¶
Because the net is unmeasured, the practically useful question is which design choices move it.
Billing architecture. Washington State is the exception that will eventually produce evidence: households there pay visible utility bills for recycling collection, so producer reimbursement flows arithmetically to ratepayers — the state models household recycling bills falling by at least 90% as reimbursement ramps to 2032. Where the offset arrives automatically on a bill, the swap's household side becomes observable and enforceable; where savings land in municipal general funds, it becomes a fiscal choice that, in every documented case, has gone the other way. Distributionally, a visible utility-bill rebate is worth more to low-income households than an invisible property-tax abatement that never happens.
Material and category structure of fees. Fees concentrated on packaging formats prevalent in low-income baskets are more regressive than the same revenue raised across formats; no fee-setter currently examines schedules through this lens, and nothing in any statute requires it.
Exemptions. Small-producer and category exemptions (Theme 3) shift costs among products with distributional consequences nobody has traced.
Visible display. Itemising the fee at the till (mandated, permitted or banned across US states — the fee-or-tax article) does not change incidence, but it changes salience — and the salience literature implies visible charges provoke avoidance behaviour that invisible ones do not, with its own distributional texture.
7. Where the argument stands¶
Stated with the confidence the evidence supports:
| Proposition | Confidence |
|---|---|
| The fee increment on groceries is regressive in sign (income measure) | High — mechanism arithmetic on measured budget shares |
| Its magnitude is small (well under 1% of household spending) | High — fee-share and per-item measurements |
| The displaced municipal financing is also regressive | High — standard incidence of property taxes and flat fees |
| The net effect of the swap is known | No — never computed anywhere |
| Low-income baskets are more packaging-intensive | Unverified — one interested-party estimate, no independent measurement |
| Municipal savings reach low-income households | Undocumented in every examined case; Washington's mechanism is the live test |
The honest synthesis is uncomfortable for both rhetorical camps. The "regressive hidden tax" framing asserts the sign correctly, inflates the magnitude, and ignores the equally regressive charge being displaced. The "EPR relieves regressive municipal burdens" framing describes a mechanism that has never been shown to complete its journey to any household. The truthful sentence supported by current evidence is narrower than either: packaging EPR replaces one regressive financing instrument with another, smaller-per-household one, and whether any income group ends up net better or worse off depends on parameters — packaging intensity by income, and the fiscal behaviour of municipalities — that no one has measured.
For a field now setting fees in the billions annually, the research agenda is correspondingly precise: an income-decile incidence study joining expenditure surveys to fee schedules; a packaging-intensity-by-income measurement; and a household-level tracing of the offset in the one jurisdiction (Washington) whose billing design makes it observable. Each is feasible with existing methods; none has been funded; and until they exist, every confident distributional claim in this debate — in either direction — is running ahead of the evidence.
References¶
- Food and Drink Federation, Eating into Household Budgets — food budget shares by income quintile (compiled from UK Office for National Statistics expenditure data); an industry source using standard national statistics, identified as such.
- RRS / EPI-Lorax for Oregon DEQ (2020) — per-item fee magnitudes; Joltreau, E. (2022), Environmental and Resource Economics 83(3) — fee shares of price; UK impact-assessment figures — the magnitude anchors in Section 1.
- Lakhan, C. (2025). New York EPR impact study, Circular Innovation Hub, York University (released by the Business Council of New York State) — the under-$40,000 packaging-consumption claim, identified with its provenance.
- Chetty, R., Looney, A. & Kroft, K. (2009). "Salience and Taxation." American Economic Review 99(4) — the salience mechanism.
- Scottish Government, Fairer Scotland Duty assessment and equality impact assessment for the Scottish deposit return scheme — the float-burden quantification and non-redemption analysis.
- Washington Department of Ecology (July 2026) — the utility-bill reimbursement mechanism and modelled household-bill reduction, cited as the observable-design case.
- Toronto 2026 budget documentation and the municipal-absorption record — see the municipal-savings article and sources therein.
- Standard incidence literature on consumption taxes and property-tax/flat-fee financing — the income-versus-expenditure measurement point and the counterfactual's regressivity.
Verification note: the central claims of this article are absence findings — no income-decile incidence study, no packaging-intensity-by-income measurement, no computed net — verified by targeted search and stated as the article's principal conclusions rather than as rhetorical devices. See Sources and method.