Our forecast, tested and corrected in public
The build-out pays for both Canada and the United States. The trade war doesn't.
Intended or not, the pressure of the last two years is making Canada build. The US gets more supply and more secure supply either way; counted as income, the build-out pays Americans a little but the trade war costs them more, so the US comes out ahead in 2034 only if nearly all the tariff damage is gone. Canada gets a bigger resource base and the capacity to sell elsewhere. By 2034, it comes out ahead unless most of today's tariffs are still in place; but the decade as a whole leaves it behind, and the longer the tariffs last, the less of it comes back.
That is a forecast, not a finding: it has published odds and could turn out wrong. Our first odds for it counted only the US side; we corrected them in public. This journal tests it as events unfold, tracking what is actually being built, where it goes and who pays, and ties every claim to public receipts.
Canada’s crude oil exports
thousand barrels a day (kb/d), yearly average
- 3,947to the US, in 20253,663 in 2019
- 445to other countries, in 202595 in 2019
By the numbers
- 688dated events on the record, 65% of them actions or documents rather than statements
- 8/17tracked projects carrying output south to the US / to tidewater, where other buyers can be reached (a selection, not every project in Canada)
- 13 of 27tracked laws and agreements actually in force (passing is not the same as switching on)
- 22pre-registered predictions, 1 settled so far
- 102corrections logged in public, with the old and new wording
In the research database
The public site shows what has been reviewed and used in articles. Behind it sits a larger research database, released piece by piece, after review, as the articles draw on it.
- 53,703data points across 916 time series of trade, prices and flows
- 3,553sources collected
- 1,311dated events logged
- 5,929judgments of events against each competing reading
- 108 / 52projects / laws and agreements tracked
- 98open items on the watch list
Latest articles
Who wins, who pays, part 12
The tariffs spared the raw stuff
Every US tariff in force on Canadian goods spares energy, potash, most ores and gold, and taxes processed goods at 10% to 50%. Canada's exports have tilted toward the spared goods, but much of the shift in value is prices: at 2024 prices, manufactured goods slipped only from 36.2% to 35.7% of exports. Steel, aluminum, trucks and lumber pay most, in Ontario, Quebec and British Columbia, though each has other causes too. The raw-materials exemptions apply to every country the tariffs cover, which counts against reading them as aimed at Canada.
Who wins, who pays, part 11
Soft before, hard after
We matched 41 actions taken by five winning governments after their votes against the platforms they ran on: 14 of 22 routine actions had been clearly promised, against 2 of 19 controversial ones, and none of the 13 that created new powers or tax breaks or reversed a rule. Quebec's Parti Québécois promised no fast-track power, so any fast-track or set-aside bill it tables by about December 29 counts as a hit. BC's Conservatives put deadlines, a 120-day approval process and DRIPA repeal in writing, so only powers they left out or kept vague count, by about January 17. We publish the result either way.
Method, part 2
What our build odds rest on
Our forecast's odds that an approved or postponed project gets built come from a count of past Canadian resource and energy projects: since 2014, about 29% of approved projects and 4% of postponed ones were in service within eight years, against 88% of those under construction. The rules were written before any project was coded, and a second coder checked a sample blind. The count can't say why most projects die, and opposition shows no overall effect. Rulings made during coding, three model choices the rules left open and five outcomes the records couldn't settle are set out here.
Tail risk, part 4
Price, not barrels
Venezuelan crude is back on the US Gulf Coast, and Gulf Coast refiners took the least Canadian crude since 2014 in August 2026. But total US imports from Canada stayed between 3.8 and 4.2 million barrels a day all year: the barrels moved, they were not replaced. The test we set on October 1, before the data, is not met. Canada pays through price: the discount on its heavy crude widened, mostly in step with higher oil prices through September, the latest monthly figure. Alberta producers and the Alberta treasury give up part of the windfall; US refiners, mostly in the Midwest, buy cheaper.
Where the outcome stands
Our current estimate of where the build-out lands, as shares that add up to 100%. Estimates, not measurements; as of .
- O0/O1Little or no extra reaching the USEither no real speed-up, or faster building with the extra output going mostly to other buyers. No separate estimate for each; this is what is left over.4%together
- O2Faster, and more reaches the USCanada builds faster and more volume reaches the US, even if the US share of Canada’s exports falls. Canada also gains routes to other buyers.70%odds
- O3More to the US, on locked-in termsAs O2, plus preferential terms for the US written down: trade-deal clauses, supply guarantees or US ownership.26%odds
And was it meant?
The journal’s claims are about outcomes. Whether Washington foresaw or wanted this is treated separately, as a labelled thought experiment tested against rival readings, not as the author’s belief. Of 220 events judged both ways so far, 147 fit the self-interest and the foresight readings equally.
The receipts
- Timeline688 events, month by month, with their sources
- Claims79 claims, each with the evidence it rests on
- ProjectsWhat is being built and which way it points
- LawsPassed is not switched on: every stage, dated
- DataTrade, prices and flows, charted from the source data
- PredictionsRegistered in advance, scored in public
- CorrectionsEverything changed, and why
- MethodActions over statements, and what “non-discriminating” means