EPR Fee Forecasting: How to Build a Number Finance Can Defend

Every autumn the same email goes out at a few hundred consumer brands. Subject line: EPR number for the plan. It lands with whoever runs sustainability, an answer comes back inside two days, and one figure goes into the annual operating plan. No range. No source for the rates behind it. No note on which programs have actually invoiced anybody yet.

I have sat in the review meeting where that figure gets picked up. The question from the controller is never "is this right." It is "where did this come from." And the honest answer, most of the time, is a spreadsheet somebody assembled in a hurry from a mix of real invoices, draft schedules and educated guesses.

Let's set the record straight on the two things that actually matter here, because neither one is the thing teams worry about.

First: the number being wrong is survivable. Every forecast in your plan is wrong to some degree and finance is entirely comfortable with that. What is not survivable is a number nobody can take apart when it moves. A point estimate with no provenance cannot be defended, cannot be corrected intelligently, and cannot be explained to an auditor twelve months later. That is the failure. Not the variance.

Second: this is a finance line. Extended producer responsibility fees, the charges regulators levy on companies for the packaging they place on the market, are an operating cost driven by commercial volumes. They are not a sustainability project with a budget stapled to it. The buck stops with finance. Period.

So how do you forecast EPR fees properly? Build the number bottom up from packaging weights and sales volumes. Price it with rates labeled by how certain each one is. Then route the accounting questions to the people qualified to answer them. That is the whole method. The rest of this is how each piece works.

One number hides three decisions somebody already made for you

The single figure is not just imprecise. It quietly resolves three open questions, and it resolves all of them without telling you.

The obligation and the invoice live in different periods. Most programs base fees on packaging placed on the market during a reporting period, then invoice after the reports are filed and reconciled. The activity that creates the cost can sit several quarters ahead of the cash going out. One number cannot carry both timelines, so it picks one and stays silent about the choice.

Rate certainty varies enormously by program. Some programs have invoiced real producers, so actual rates exist. Others have published draft schedules that may still change. A few have nothing beyond the statute: no schedule, no timeline, nothing to price against. Blend those three into one figure and you have hidden exactly the uncertainty a finance function exists to manage.

This is not a software vendor's framing of the problem. When PackUK published illustrative year two fees for the UK scheme, Mike Revell, executive chair of the Foodservice Packaging Association, welcomed the early publication and warned that without that transparency the scheme risks operating as a "retrospective tax." He then listed the gaps that stop a producer pricing anything with confidence: how local authority net costs get calculated and allocated by material, and how recycled material market values feed into future fee setting. A trade body chair is describing Tier 2 and Tier 3 uncertainty there, in plain language, from outside the compliance software category. If the inputs behind a rate are not public, no amount of modeling on your side makes the output certain.

The drivers sit outside your department. Tonnage depends on packaging specs and sales volumes. The split across states and countries depends on where product ships. Eco modulation, the mechanism that discounts or surcharges fees according to how recyclable the packaging is, depends on design characteristics finance holds no view of. Your forecast is only as good as the data feeding it, and none of that data starts in your systems.

Tier every rate, and make the tier travel with the number

This is the part I would push into your plan this cycle if you change nothing else. Every rate assumption carries one of three tiers, and the tier stays attached to the number all the way into the plan and out the other side into the variance review.

Tier 3 is the one people fight me on. The instinct is to put something in the cell, because an empty cell looks like you did not do the work. Resist it. A rate you made up reads exactly like a rate you sourced once it is three rows deep in a model, and the person who inherits that model in eighteen months will have no way to tell them apart.

The strongest argument for tiering does not come from me either. Circular Action Alliance submitted California's inaugural program plan in June 2026 carrying a five year budget projected as a range, USD 9.3 billion to USD 17.2 billion, funded entirely through fees assessed on producers, with the 2027 figure alone spanning USD 1.26 billion to USD 1.87 billion. Read that again. The organization that will calculate your invoice published a range close to double its own low end, on a plan that closed public comment on August 14, 2026 and goes to CalRecycle in October. If the body setting the fees declines to commit to a point estimate, an operating plan carrying one is making a stronger claim than the PRO does.

The tiers stopped being theoretical about a year ago

For a while this framework was housekeeping, because almost every US program sat in Tier 3 and the distinction was academic. That changed fast.

Oregon's producer responsibility organization issued its first round of EPR invoices in July 2025 and a second round in January 2026. Colorado's base fees went live in January 2026 under HB 22-1355, making it the second US state charging producers. In the UK, PackUK issued first pEPR invoices in October 2025, and its modulation policy applies the first recyclability adjusted fees to the 2026 to 2027 financial year, calculated from packaging supplied in 2025. California started issuing early fee invoices this month, with full program fees from January 2027.

Real invoices exist. Which means Tier 1 exists, and separating it from a draft schedule finally buys you something. Meanwhile Washington enacted its program and has been finalizing its detailed fee schedule and reporting templates, which is a textbook Tier 3 line: real obligation, real tonnage, no rate to price it with yet.

If you want the mechanics behind any one of those, we have written up Oregon's first invoicing cycle, the UK base fee formula, and how eco modulation moves fees across jurisdictions.

Build it in four layers, material by material

Volumes come first, straight from the sales plan. That matters more than it sounds. When the fee line shares its demand assumptions with everything else in the plan, a volume variance later has one explanation instead of two competing ones and an argument between two teams.

Material mix converts units into tonnage using component weights from the current specification versions. This layer is where data quality bites hardest. A bottle that was lightweighted in May but never updated in the compliance record flows straight through your model as an overstated fee, every unit, all year. We took that failure apart in detail in the two grams nobody reported.

PackUK put the same finding in writing. Its published operational plan for 2026 to 2027 records that during the first assessment year, producer resubmissions and significant shifts in obligated tonnages exposed financial risk and drove the scheme administrator toward stronger recalculation processes and a structured schedule for producer recalculations. Read that from the producer side rather than the administrator side. Tonnage moved after the numbers were filed, and it moved enough that the body running the scheme redesigned its processes around the movement.

Jurisdiction allocation splits the tonnage by where product is placed on the market, using the same logic your compliance filings use. Not similar logic. The same logic. Two allocation methods produce two answers and one very long meeting. The data engineering behind defensible allocations is worth reading before you pick a method.

Rates come last, with any eco modulation assumption stated on its own line rather than buried inside a blended figure. Packaging design decisions move that lever. Finance should be able to watch it move.

On the tonnage input specifically, compliance can pull actuals rather than estimates. Packgine's Inventory Compliance Alerts track packaging tonnage by material and jurisdiction through the ERP or WMS connection and raise alerts at 50, 80 and 100 percent of each program threshold, alongside deadline alerts at 60, 30 and 7 days. For a finance team that converts a year end surprise into a mid year signal that the plan's volume assumptions have drifted. That is the difference between explaining a variance and discovering one.

One material, one program, start to finish

Here is the chain for a Tier 2 program, which is the fiddliest case.

Units from the sales plan, multiplied by the component weight on the current spec version, gives you tonnes. The placement split allocates those tonnes to the jurisdiction. The draft schedule's rate band prices them. The eco modulation assumption adjusts the result, stated separately so anyone can see what it contributed. What comes out is a range whose width equals the band, carrying a Tier 2 label and the date of the schedule it came from.

You will have noticed that no producer rate figure appears anywhere in this post. The California program budget above is a published total, not a price per tonne anybody will invoice you, and I have kept that distinction sharp on purpose. It is the same discipline I am asking you to apply. No producer rate schedule I would rely on for this planning cycle carries a date inside it, and a rate without a dated source is precisely how the undefendable point estimate got into your plan in the first place. I am not going to hand you one dressed up as analysis.

When the first invoice lands, split the variance four ways

From the first real invoice onward, decompose forecast against actual into four causes. Each one gets a name attached to it.

Volume: units differed from plan. That belongs to demand planning. Mix and weight: tonnage per unit differed, which is almost always a spec data problem rather than a forecasting one, and belongs to whoever owns packaging data. Allocation: the jurisdiction split differed, and belongs to whoever owns market attribution. Rate: the applied rate or the modulation differed, and that one is yours.

An invoice that confirms a Tier 2 assumption promotes it to Tier 1 for the next cycle. Run this for four quarters and the fee line stops being an annual argument and starts being an estimate that converges. That is the entire payoff, and it only works if the tiers were labeled going in.

One more thing on ownership, because it is where this usually breaks. Finance owns the consolidated number, but the inputs cannot come from finance. The arrangement that works is a standing data contract: compliance delivers tonnage by material and jurisdiction on a fixed cadence, each delivery carrying a note on data quality, and finance delivers the volume plan. The annual email asking somebody for "the EPR number" is not a data contract. It is how a point estimate with no provenance reached your plan.

Where I stop, and your auditors start

Whether amounts are recognized when packaging ships or when the invoice lands, how they are classified, and what gets disclosed as estimation uncertainty are accounting policy questions under your applicable framework. I do not answer them and neither should your FP&A team. What the tier summary gives your controller and your auditors is the briefing they need to answer them well: what is known, what is drafted, what is structural, and where each figure came from.

Nothing here is accounting, tax or legal advice.

Do this before the plan locks

Replace the point estimate with the tiered range. Book the accrual and disclosure conversation with your auditors now, tier summary attached, rather than in February when the first invoice arrives and nobody can explain the gap. Stand up the quarterly variance review with the four owners named out loud.

The question in front of finance is not what the fees will be. Nobody knows that, including me. It is whether the plan carries a number you can still defend when it moves.

Send us one jurisdiction. Connect the tonnage feed, and your plan gets a fee input that updates from live data instead of from an email in October. Check your obligation status with the EPR checker or book a working session.

Frequently asked questions

How do you forecast EPR fees?

Build the forecast bottom up in four layers: sales volumes, material mix converted to tonnage using current spec weights, jurisdiction allocation using your filing logic, then rates. Label every rate with a confidence tier and keep the tier attached to the number.

Which programs are actually invoicing EPR fees right now?

Oregon has invoiced since July 2025, Colorado since January 2026, and the UK issued first pEPR invoices in October 2025. California began early fee invoicing in August 2026 with full program fees from January 2027. Several other enacted US programs still have no published rate to price against.

Should sustainability or finance own the EPR fee forecast?

Finance owns it, because these fees are an operating cost driven by commercial volumes. The inputs come from compliance and packaging teams on a fixed cadence, but the consolidated number and its defense belong to finance.

Why not just use one EPR fee number in the plan?

Because a single figure blends three different levels of certainty and silently resolves the timing question between obligation and invoice. When it moves, nobody can say which input moved, so the correction becomes a negotiation instead of an update. The producer responsibility organization running California published its own five year program budget as a range, not a point estimate. Your plan should not be more confident than the body setting the fees.

When should EPR fees be accrued?

That is an accounting policy question under your applicable framework, and it belongs to your controller and your auditors. Bring them the tier summary showing what is invoiced, what is drafted and what is structural, and they can answer it properly.

Sources

• Oregon Department of Environmental Quality, Plastic Pollution and Recycling Modernization Act: https://www.oregon.gov/deq/recycling/pages/modernizing-oregons-recycling-system.aspx
• Colorado Department of Public Health and Environment, Producer Responsibility Program (HB 22-1355): https://cdphe.colorado.gov/hm/epr-program
• PackUK, Packaging Extended Producer Responsibility Producer Disposal Fees Modulation Statement, 28 June 2025: https://assets.publishing.service.gov.uk/media/685ea3c362b2e559cbd75387/Packaging_Extended_Producer_Responsibility__pEPR__Producer_Disposal_Fees_Modulation.pdf
• PackUK Operational Plan 2026 to 2027, GOV.UK, updated 2 March 2026: https://www.gov.uk/government/publications/packuk-operational-plan/packuk-operational-plan-2026-to-2027
• CalRecycle, Packaging Extended Producer Responsibility: https://www2.calrecycle.ca.gov/Packaging-EPR/
• Circular Action Alliance California Program Plan figures as reported in DLA Piper, California's EPR Program Plan: Costs, compliance, and design considerations for producers, 24 June 2026: https://www.dlapiper.com/en-us/insights/publications/2026/06/californias-epr-program-plan
• Mike Revell, Foodservice Packaging Association, quoted in Packaging Insights on PackUK year two illustrative fees: https://www.packaginginsights.com/news/uk-packaging-epr-amendments-costs-352306.html

About the author

Ajay Vasanthakumar, Marketing Director at gCurv Technologies. He works with consumer brands and importers on packaging compliance data, covering US state EPR reporting, UK pEPR, and EU PPWR conformity evidence.