Corrections issued 25 August 2026 — three, and the first is material to positioning. UK ETS supply: the deficit this desk published since 1 July was a quarter too large — an amendment in force on 1 July was not ingested. Washington: this note carried no Washington section despite a citable mark on our own tape. Publication gap and stale marks: this note was written as if the desk had reported yesterday after a five-day silence. · All corrections
CCA — mark: 28.81 (May 2026 joint auction, current-vintage settlement price).
What moved: Nothing overnight in the auction tape; the live item is CARB's amendment package, expected to become effective September 1, 2026 — one week out.
Why: The updates remove approximately 118 million allowances from annual budgets between 2027 and 2030, attributed to the 2022 GHG inventory update, with the cap declining 11% per year for the remainder of this decade ; separately, the updates create a new Manufacturing Decarbonization Incentive . Our forward rows already carry WCI in deficit (-12.0 Mt 2027, deepening to -42.5 Mt by 2030) on the pre-amendment statutory table.
Desk view (judgment): Bullish CCA on a 6–12 month horizon; supply contraction binds at vintage-2027 allocation, not spot, so near-dated upside is a grind, not a gap.
RGA — mark: 35.0 (2026-06-03, Auction 72 clearing price).
What moved: No new RGGI development in the last 24 hours.
Why: Standing driver is the Third Program Review ten-state trajectory (69.8M in 2027, ~-8.5M/yr), against a ~67M private bank that cushions the flip to deficit we model in 2029.
Desk view (judgment): Neutral-to-constructive into the 9 September auction; the bank caps the squeeze this year.
Alberta TIER — mark: 95.0 (2025-05, TIER fund price, frozen since May 2025).
What moved: No TIER development overnight. (Yesterday's AUC data-centre ruling is a utility matter, not a TIER item.)
Why: The May 15 Canada–Alberta Implementation Agreement schedule holds $95 in 2026 and $100 for 2027–2029, with the credit-price floor regulation owed by 31 December 2026.
Desk view (judgment): Neutral; the floor regulation, not compliance flow, is the repricer.
RINs — marks: D6 2.37 and D4 2.41 (2026-06-04, traded RIN price; EIA Today in Energy, Bloomberg data). No fresher print on our tape; direction below is sourced, not levelled.
What moved: Prices for U.S. ethanol blending credits plunged Monday to their lowest levels in more than four months after EPA extended a September 1 compliance deadline for refiners and ruled on long-pending small refinery exemption requests ; EPA said it will rule on SRE petitions by the end of August .
Why: Industry and analyst estimates put the exemptions at 1.2–1.8 billion RINs freed for 2025 compliance — a demand deletion, not a supply add.
Desk view (judgment): Bearish D6 and D4 into the end-August rulings; risk is two-way if waiver volumes disappoint.
CA LCFS — mark: 76.39 (2026-08-16, CARB weekly snapshot, volume-weighted average of all non-zero transfers).
What moved: No fresher weekly print carried on our marks this morning.
Why: Cross-market read-through: a RIN collapse strips revenue from the renewable-diesel stack, and RD is the marginal LCFS credit generator in California — less RD pull west is credit-generation negative, i.e. tightening at the margin.
Desk view (judgment): Mildly bullish LCFS credits on a multi-month view; flat today.
Canada CFR / BC — mark: BC LCFS 135.8 (2026-07, Monthly average credit transfer price). No fresh CFR or BC development; standing read is a firm bid with no new catalyst.
No citable REC mark on our tape. Overnight, only Korea's plan to expand renewable supply for chipmakers (Chosunbiz, 25 Aug) — offshore of US RPS demand. Churchill Falls (17 Aug) is standing context.
Desk view (judgment): Neutral across US compliance RECs; no fresh RPS catalyst in the last 24 hours.
No new SBTi development: the Corporate Net-Zero Standard v2.0 is standing context (published 11 June 2026), with credit-use implementation guidance still expected by end-2026 — the largest single VCM demand variable. On removals, Puro Standard General Rules v4.4 was approved 07 May 2026 ; our monitors flag a document change on that file, which we treat as administrative until a second source shows substance. No fresh Article 6/GCC item. No citable CORC mark on our tape.
Desk view (judgment): Neutral removals; guidance timing, not supply, sets the next leg.
EUA — mark: 83.45 (2026-08-25, Latest EEX auction clearing price).
What moved: This morning's primary clearing is the fresh print; no policy change overnight.
Why: Rows run on the Directive (EU) 2023/959 LRF — surplus +113.7 Mt in 2027 narrowing to a 2029 deficit. COM(2026) 616 remains a proposal, not law.
Desk view (judgment): Constructive; structural tightening beats soft near-term demand.
UKA — mark: 58.68 (2026-07, Official monthly average UKA settlement price (Dec futures), CCM table, monthly-final observation), +2.9% vs prior same-basis mark.
What moved: Nothing new overnight.
Why: SI 2020/1265 trajectory leaves the UK in deficit (-8.7 Mt 2027, widening); linkage talks are the swing factor.
Desk view (judgment): Bullish on structure, neutral tactically pending linkage news.
NZU — mark: 54.77 (2026-08-13, NZU secondary spot, close observation), -2.2%. Auctions failing below the NZD 71 floor leave supply near zero against ~-33.8 Mt deficits. Judgment: bullish structurally, soft tape.
ACCU — mark: 37.5 (2026-05-15, Generic ACCU spot, CER QCMR March quarter 2026). Safeguard deficit deepens on the legislated 4.9%/yr decline. Judgment: neutral near term.
Correction, stated out loud — issued 25 August 2026, after publication
1. This note was written as if the desk had published yesterday. It had not. The last note subscribers received was 2026-08-20-evening. This note went out after a five-day silence, and it uses "overnight" or "in the last 24 hours" seven times — including "No new RGGI development in the last 24 hours" and "No TIER development overnight." Those statements are about one day. They were offered where a statement about five days was owed. Nothing in them was checked against the gap window, so they should be read as unverified for 20–25 August, not as evidence that the period was quiet. Four notes were drafted during that window and discarded for reasons unrelated to their market content; that is a production failure of this desk, not a quiet market, and the note should have said so.
2. Several marks are materially older than the note's framing implies. They are real prints, correctly dated in the text, but presented beside a same-day EUA clearing without their age being stated:
mark as of age at publication RGGI (RGA) 35.0 2026-06-03, Auction 72 83 days CCA 28.81 2026-05-20, WCI Joint Auction 47 97 days ACCU 37.5 2026-05-15, CER QCMR 102 days Alberta TIER 95.0 2025-05, fund price 467 days RGGI in particular: an 83-day-old auction clearing carried into a paragraph headed "no new development in the last 24 hours" invites the reader to treat 35.0 as a current level. It is the last auction print and nothing more. Auction 73 is 9 September; there is no fresher clearing, and this desk holds no citable secondary RGGI mark.
3. The NZU line's "-2.2%" is not a daily move. It measures 54.77 (2026-08-13) against 56.0 (2026-07-23). Both carry the same source and the same basis in the desk's own store —
theecanmole/PDDL, "NZU secondary spot" — so the pairing is legitimate, but its two endpoints are 21 days apart. Printed unqualified beside genuine same-session changes, it reads as recent action. It is three weeks of drift.4. What is NOT being corrected: the Alberta floor date. The note states the credit-price floor regulation is "owed by 31 December 2026." That was queried and has been verified against the Implementation Agreement and re-confirmed — it stands. Two dates are in play and both are correct: the minimum transfer price takes effect in 2030 ($60 in 2030 rising to $110 in 2040), and Alberta intends to enact the Price Floor regulation by 31 December 2026 in order to enable that 2030 start. They are a deadline and an effective date, not competing versions of one date. No separate end-2026 GGPPA amendment deadline appears in the Agreement. Sources: Prime Minister of Canada backgrounder, 15 May 2026 and Osler.
The generator now reads its own publication history and is required to cover any gap explicitly; the day-over-day block now carries the span of every delta. Neither fix changes this note, which is why the record is corrected here rather than edited.
5. Two errors in the first issue of this correction, corrected here.
The CCA age was wrong. The table above first said 102 days. It is 97. The 102 came from reading the display label "May 2026 joint auction" and assuming mid-month — the exact defect this correction was written to describe, committed inside the correction itself. The mark is now pinned to 2026-05-20, the date WCI Joint Auction 47 was held. That date, not the 27 May results release, is the anchor: the mark's basis is "current-vintage settlement price," which is what the auction cleared at, and a settlement price is a property of the auction event — the release is when this desk could learn it. Verified against the CARB/WCI summary results for Auction 47 (held 2026-05-20, released 2026-05-27, current vintage settled 28.81). The same seven-day pattern holds for Auction 48: held 19 August, results 26 August.
The NZU sourcing sentence was too flat. This correction said 56.0 and 54.77 share a source. That is true. The 19 August evening note called 56.0 an "older separately-sourced scrape mark," which reads as a contradiction. Diffed against the store: both statements are true of different pairs. 56.0 is separately sourced from the 55.6 secondary spot the 19 August note was comparing it against; it shares a source with 54.77, which is the pairing at issue here. Neither figure was wrong. What was wrong is the 19 August note's phrasing, which stated a relationship between two marks as though it were a property of one — and this correction restated it without saying which pair it meant. Both are corrected here rather than left to be read as a conflict.
6. What this note should have named as tomorrow's catalysts, and did not. The outlook section listed no dated events. It should have carried: WCI Joint Auction 48 results, released 26 August — tomorrow, the first sale before the 1 September CARB amendment takes effect; the CARB LCFS weekly credit transfer report, published this session; and the CFTC Commitments of Traders report on Friday 28 August. All three were on the desk's own calendar and none reached the note, because the morning generator never read that calendar. It does now.
7. UK ETS supply — the deficit we published was a quarter too large
The cause is an amendment we did not ingest. On 1 July 2026, SI 2026/392 — the Greenhouse Gas Emissions Trading Scheme (Extension to Maritime Activities) Order — amended Table B of art. 22 of SI 2020/1265, raising the Base column as maritime activities joined the scheme. Our forward rows continued to carry the pre-amendment figures for fifty-five days.
year supply carried amended Table B balance carried correct balance 2027 70.1 72.317 −8.7 −6.5 2028 53.5 55.618 −22.9 −20.8 2029 50.9 52.965 −23.2 −21.1 2030 49.3 51.284 −22.6 −20.6 Across 2027–2030 the cumulative deficit narrows from −77.4 Mt to −69.0 Mt — 8.4 Mt, or 10.8%, less short than we published. The 2027 figure, the one that appeared in notes, was overstated by a quarter.
What that changed in print. Three published notes carried the figure, and each attributed it to the statute by name:
- 19 August, morning — "Bullish on a deficit throughout (−8.7 Mt 2027); linkage headlines are the swing factor."
- 20 August, evening — "Fundamentals stay deficit throughout (2027 balance −8.7 on SI 2020/1265)."
- 25 August, morning — "SI 2020/1265 trajectory leaves the UK in deficit (−8.7 Mt 2027, widening)."
Each cited the right instrument and the wrong version of it. Being specific about the legal basis is what made the error look verified.
What survives and what does not. The direction stands: the UK is in structural deficit across the whole window on the amended table too, and the 2028 step-down is legislated rather than modelled. The desk view — bullish on structure, linkage the swing factor — does not change. What was wrong is magnitude, and it was wrong in the direction that flattered the view: we published a tighter market than the law describes. A reader sizing a position off the 2027 number was working with a deficit a quarter larger than the statute supports.
How it survived fifty-five days, stated plainly because it is the more useful part. The desk validates its forward rows daily against a table of legislated figures. Both the rows and that table are maintained by hand, and both carried the same pre-amendment numbers — so the check compared two copies of one mistake and reported the market OK every day. The desk's own price store had already recorded the amended figures in a separate file. We held the right numbers and the wrong numbers at the same time and nothing compared them.
That is fixed at the mechanism, not the instance. Figures appearing verbatim in a legal instrument are now transcribed into a separate anchor file with the citation and the table row, and the daily check compares the model against the statute rather than against itself. UK ETS was the first market anchored and the mismatch surfaced on the first run. The corrected rows are live on the forward balances page, and the change journal records SI 2026/392 as the named cause.
8. Washington — this note carried no Washington section, and it should have
The desk's tape holds a citable primary print for Washington Cap-and-Invest — USD 64.56, the current-vintage settlement from Auction #14, held 3 June 2026 — and this note does not mention Washington or WCA once. That is not a judgment call about newsworthiness. An instrument with a citable mark on our own tape gets a line, or the omission gets stated.
Washington is not quiet. Our forward layer carries it as a widening deficit across the whole window — supply 44.3 / 39.5 / 34.8 / 30.0 Mt against demand 56.7 / 55.9 / 55.0 / 54.2 Mt for 2027–2030, a balance of −12.4 / −16.4 / −20.2 / −24.2 Mt. Those rows sit on the current statutory slope precisely because Ecology's post-2026 budget rule, due 1 October 2026, has not landed. A market in structural deficit with its supply trajectory pending a rule five weeks out is not one to leave unmentioned.
The mark's age, stated rather than implied. 64.56 is 83 days old at this note's publication. It is the last settlement print and nothing more.
This is a pattern, not a slip. The last four morning notes — 19, 21, 24 and 25 August — mention Washington zero times. The 18 August morning note covered it, and the evening notes have covered it throughout. The morning note specifically stopped carrying a market the evening note still does.
Why, mechanically. Section 1a of this note is headed "Compliance carbon — WCI/CCA, RGGI, Alberta TIER." Washington is not in that scope, so there was nowhere for it to go. The forward block does carry it, under the code
wa-ci, while the tape names itWashington (WCA)— one instrument under two labels, in two inputs, and a section heading that admitted neither. The morning note's section structure is now generated from the tape rather than fixed in the prompt: an instrument carrying a citable mark gets a section or an explicit statement of why there is nothing to say.No figure in this note was wrong. What was missing was a market.