EPA Blew a 1.76 Billion Hole in the RFS, Then Promised to Patch It the Same Afternoon
The US EPA ruled on 34 small refinery exemption petitions for the 2025 compliance year on 31 August, exempting 1.76 billion renewable identification numbers across 29 refineries. That is roughly 78% above the 990 million RINs the agency assumed when it set record blending volumes in March. In the same action it said it will propose reallocating 100% of the gap between projected and actual exempted volumes into the 2026 and 2027 obligations before the end of October.
Take those two sentences separately and you get opposite trades. Take them together and you get the actual story, which is more interesting than either.
What was decided
Of the 34 petitions, 18 received full exemptions, 11 received partial exemptions at 50%, three were denied and two were found ineligible. The resulting 1.76 billion RINs sits at the very top of the 1.2 to 1.8 billion range the market had been trading on for the past fortnight.
Alongside it, the agency confirmed a direct final rule extending the 2025 compliance date by 30 days, to 1 October. That is the procedural piece that had been sitting in regulatory review since 26 August, and its purpose is mechanical: give the market time to absorb the additional credits before compliance is due.
The number that changes the read
Anyone who spent August watching this file had 990 million as the benchmark, because that is what EPA's own March rule assumed 2025 exemptions would come to. Coming in at 1.76 billion is an overshoot of about 770 million RINs.
On the framework the market was using a week ago, that is the bear case. The reason it has not played out as a straightforward collapse is the reallocation.
EPA's March rule already carried forward 70% of exempted 2023-25 obligations, which on a 990 million assumption embedded roughly 693 million RINs into the 2026 and 2027 standards. What the agency has now proposed is to reallocate the full difference between projected and actual — the 770 million overshoot — rather than 70% of it.
Run those together and, on our calculation, something close to 1.46 billion of the 1.76 billion exempted volume would end up reallocated to non-exempt refiners: about 83% of the total. That is a very different picture from 1.76 billion of demand simply disappearing, and it is why trade bodies that called the exemptions unjustified still described the reallocation pathway as offering a route to no net loss in renewable fuel demand.
Where the caution belongs
Three things stop this being a clean bullish resolution.
The first is that a proposal is not a rule. EPA has said it will propose reallocation before the end of October. Proposals get comment periods, get litigated and get changed. Between now and a final rule, the exempted volume is real and the offset is a commitment.
The second is category mix. The demand that matters to vegetable oil sits in the biomass-based diesel and advanced pools, not in conventional ethanol credits. A reallocation that restores headline RIN volume without restoring the advanced and biomass-based diesel components would leave the soyoil bid weaker than the aggregate suggests. That breakdown is the detail worth reading in the decision documents rather than the press coverage.
The third is timing mismatch. The exemption is immediate; the reallocation lands in 2026 and 2027 obligations. Physical blending decisions and feedstock purchasing happen in between. A gap between when demand is removed and when it is restored is exactly the window in which crush margins and vegetable oil spreads move.
The read for India
India buys roughly 60% of its edible oil from abroad, so the transmission runs straight through to landed cost.
The near-term direction is still helpful. Palm has already given back its August rally, with the benchmark Bursa contract closing 28 August at 4,890 ringgit after touching 5,018 a week earlier, and the sell-off that started with the exemption rumours has not been reversed by the decision itself. Import cover is comfortable: July arrivals were the strongest in ten months at 1.481 million tonnes, and cumulative imports over the first nine months of the 2025-26 oil year reached 11.923 million tonnes against 11.346 million a year earlier.
The medium-term point is the one buyers should sit with. If the reallocation is proposed and finalised as described, the structural US feedstock bid for 2027 is largely intact, and the past fortnight's weakness looks more like a repricing of policy uncertainty than a change in demand. Buyers who extend cover deep into 2027 on the assumption that this decision permanently shrank US vegetable oil demand may be trading a headline that has already been half-retracted.
Watch three things through October: whether the reallocation proposal actually appears on schedule, how much of it lands in the biomass-based diesel and advanced categories, and whether D4 credit values recover from the levels they reached while the market was pricing the worst case.
Forward-looking views here are analysis, not investment advice.
