2026-09 — full dated evidence record
Original readings, including their limits and then-future triggers. These are historical assessments.
Latest → · Corrections record →
Evidence update · Material change
CMHC’s August metropolitan tables report 15,729 starts, 16,899 gross completions and 360,554 units under construction at month end. July recorded 16,374 starts, 18,797 completions and 362,059 under construction.
More on this: Can governments still solve housing shortages? →
Why it stands: Both matching August workbooks were obtained and qualified September 30; their issuer publication date is September 27. June and July extracts remain intact.
The existing housing pages now distinguish August construction milestones from the separate national permits and modelled housing benchmarks.
Trigger recorded for this reading: A successor source or correction that changes the qualified reading.
Limit: Unadjusted metropolitan counts, not national net additions. Ottawa and Gatineau are separate among 42 area rows. Monthly movements do not establish a trend, cancellations or closure of the housing shortage.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Correction · Material change
The corrected register comparison has four added report URLs, not three, and one removed. The newly reviewed public reports disclose JBL operating conditions, unresolved project-scope and service-date disagreements, equipment phasing, and a withdrawn historical proposal.
More on this: Claims on the grid: what Ontario's connection record proves about its largest new loads — and why the register they sit in cannot be added up →
Why it stands: The earlier comparison omitted the report on the newly added K1/K2 application. All four added reports and the removed Pointe Aux Roches reference were reviewed against their public pages on September 30.
Report review reveals operating conditions and unresolved scope/date disagreements that application status alone cannot establish.
Trigger recorded for this reading: A successor source or correction that changes the qualified reading.
Limit: September 30 is the review/correction date, not a new register vintage or report issue date. The underlying register vintages remain August 14 and September 10; no energized capacity or current validity for a withdrawn proposal is inferred.
Reads onto the Brief it reads from
Evidence update · Material change
Canada generated 634,612.2 GWh in the twelve months to July 2026, up 7,054.9 GWh, or 1.1%, from the preceding twelve months. The latest matched annual window is now higher; the earlier April window was slightly lower than a year earlier.
More on this: Canada's electricity expansion: the projected pace and the delivered record →
Why it stands: Statistics Canada’s September 29 release adds May, June and July without revising the shared generation observations. The current national and provincial comparisons move to July. Installed capacity still ends in 2024 and the CER Energy Future scenario files are unchanged. The CER cross-border panel advances separately to June, the latest month its captured file contains.
The newer window changes the near-term delivered-generation reading. It does not establish the sustained pace required by the official scenarios or explain why output changed. Generation, installed capacity, electricity trade and connection applications remain different measures.
Trigger recorded for this reading: Another monthly generation or electricity-trade observation, a new annual installed-capacity year, or a new CER scenario edition.
Limit: Monthly generation totals are not adjusted for weather. One rolling annual increase is not proof of a durable future pace. CER and Statistics Canada cross-border totals are compared on matching June windows and kept on separate issuer bases; the CER file’s exact publication date is not established.
Reads onto the Big Question it bears on · the Brief it reads from
Evidence update · Material change
June has 938,951 active employer businesses, 1,247 more than revised May and 622 more than a year earlier, a 0.07% annual increase. July corporate credit is $2.407 trillion: 4.85% higher than a year earlier but about 0.1% below revised June.
More on this: Business dynamism: employers, credit and financial distress →
Why it stands: The September 28 employer release adds June and revises earlier estimates, including May’s active count from 936,959 to 937,704. The September 18 credit release adds July; the selected series revise January–June 2026. June credit rises from $2.398491 trillion to $2.408873 trillion, leaving July $2.347 billion lower. Insolvency and lending-survey source bytes remain unchanged, through July and 2026 Q2 respectively.
The employer total remains close to its year-earlier level while openings and closures show substantial movement. Using the unrevised June credit value would give the wrong monthly direction. The updated charts and retained editions make that distinction inspectable.
Trigger recorded for this reading: The next employer edition, the August corporate-credit release scheduled for October 20, or a new insolvency workbook or lending survey.
Limit: Employer counts are seasonally adjusted. A closure means a business stopped employing people, not necessarily a permanent exit or bankruptcy. Corporate credit, employer businesses, lender opinions and insolvency filings cover different populations and cannot establish a common failure rate or cause.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor · the Brief it reads from
Evidence update · Material change
Ontario’s September connection register has 2,074 records and 75 qualifying large-load applications, up from 2,062 and 71 in August. Across the whole register, three applications move from Active to Complete, one from Complete to Withdrawn, and eleven entries change from Part 1 SIA Comp to Complete.
More on this: Claims on the grid: what Ontario's connection record proves about its largest new loads — and why the register they sit in cannot be added up →
Why it stands: The IESO dates the newer list September 10; the capture was September 29. The same large-load rule identifies four entrants and no exits: all four are new register IDs, not existing records reclassified by field edits. The whole register adds twelve records and removes none. Among the qualifying large-load applications, twenty have a posted reference and fifty-five do not.
The register provides a newer account of application membership and status. The eleven Part 1 SIA Comp to Complete changes are reported separately; the register does not state whether they reflect further assessment work or relabelling. Only one of the 15 transitions belongs to the current large-load group, an Active-to-Complete move. None measures newly energized capacity or delivered power.
Trigger recorded for this reading: The next attended register capture, or separate qualification of newly posted or changed assessment documents.
Limit: The large-load subset includes load-class applications stating at least 75 MW and excludes sizes expressed only in MVA. Engineering interpretations retain their August 4 cutoff; newly posted or changed assessment references have not been qualified here as engineering evidence.
Reads onto the Brief it reads from
Checked, unchanged · Material reading
The infrastructure reading is unchanged after a fresh source check. HICC’s project file is identical in every byte to the July and September captures, with reported project ends still no later than March 2025. The 2025 capital and asset-life estimates are also unchanged.
More on this: Infrastructure delivery: from commitments to the capital base →
Why it stands: September 29 retrievals matched the retained project, dictionary, annual capital and asset-life source bytes. The annual Statistics Canada tables retain their June 12 publication date. A new checking date does not create a newer project observation.
The site can reaffirm what this record establishes, but cannot report fresh project delivery from an unchanged file. The distinction between funded entries, reported completion and improved public services remains unresolved.
Trigger recorded for this reading: A changed HICC project file with comparable identifiers, or the next annual infrastructure accounts and asset-life release.
Limit: The HICC file includes studies and plans as well as physical assets. Approved contributions are not cash spending, eligible costs are not final costs, and identical files do not show that construction stopped. Asset-age estimates do not establish engineering condition.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor · the Brief it reads from
Unresolved · Material reading
CMHC’s detailed geography and intended-market publication pages displayed a September 27, 2026 publication date when checked on September 29, 2026. The matching workbooks have not yet been obtained and reconciled, so this panel retains the last qualified July estimates. The newer package’s reference period and numerical changes remain unverified.
More on this: Can governments still solve housing shortages? →
Why it stands: Both official publication pages display a newer date. The download selectors did not yield a qualified geography/intended-market workbook pair; ordinary direct requests returned HTTP 403. Access failure is not evidence of unchanged values.
The Housing question and Capacity monitor keep their July metropolitan estimates, explicitly dated, until a matching successor pair passes source and total checks.
Trigger recorded for this reading: Acquire the matching geography and intended-market workbooks from the official publication controls; establish their reference month, coverage notes and totals.
Limit: Page metadata cannot establish reference-period values. These gross metropolitan counts do not measure national net additions or occupancy.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Analysis update · Material change
The United States import bans reach their legal effective date on September 29. Three proclamations cover specific Canadian alcoholic beverages, whey and molasses products, non-alcoholic beer and motorcycles in the tariff line for engines over 800 cc. These are listed products, not whole industries. Two further proclamations changed the goods subject to earlier 50 percent duties from September 15 and made those amended duties apply alongside Section 232 duties. The trade reading remains on July, before these September actions.
More on this: The Tariff Ledger →
Why it stands: Five proclamations signed September 8 and published in the Federal Register September 14 were captured with their complete annexes and assessed on September 29. The evidence date is their Federal Register publication date. This is a late reading of already published documents, not a new September 29 trade observation. The scope-change effective date is September 15; the import-exclusion effective date is September 29 at 12:01 a.m. eastern time.
For the specific excluded products, the legal rule changes from an additional tariff to an import prohibition. Other products enter or leave the amended tariff lists. That change in legal treatment warrants a fresh reading, while the aggregate economic effect remains unmeasured here.
Trigger recorded for this reading: The August merchandise-trade release scheduled for October 6, or an official amendment, implementation notice or court decision changing the cited legal provisions.
Limit: The annexes define narrow tariff lines and packaging limits. Covered goods imported before September 29 but not yet entered for consumption or withdrawn from warehouse for consumption retain the earlier 50 percent duty. This record has not observed blocked shipments, duties assessed or collected, an affected-trade-value total, or a causal effect on trade, employment, prices or GDP. The dairy prohibition references a separate scope document this reading has not established; no sixth proclamation is inferred. A complete reconciliation against the current HTSUS is not claimed.
Reads onto the Big Question it bears on · the US Exposure Monitor · the Brief it reads from
Evidence update · Material change
Canada's July 1 population estimate is 41,798,407, up 80,343 from revised April 1 and 189,425 from revised July 1, 2025. Growth is 0.5% over the year. Non-permanent residents number 2,779,774, or 6.65% of the population; their stock fell by 18,875 over the quarter. The same release puts the July 1 median age at 40.9 and the share aged 65 and over at 20.0%.
More on this: Canadian Capacity Monitor →
Why it stands: The September 23 population releases add July 1 stocks, Q2 migration and natural-increase components, and the annual age profile. Statistics Canada also revised earlier estimates; Countability retained its prior inputs and compared every shared observation before recomputing the existing series.
The observed population is growing slowly. The NPR share is above this record's straight-line path to the stated end-2027 target, changing its pace reading; that path is an accounting convention, not a forecast. People added per annualized housing start remain lower than a year earlier, but the former negative-pressure and record-low claims no longer hold. The pace change is already present in the revised April estimates; it is not caused by the subsequent quarterly decline in the NPR count.
Trigger recorded for this reading: The next quarterly demographic release or a revision changing population, migration components or target pace.
Limit: July 1 stocks, Q2 flows and the annual July 1 age profile have different reference conventions. These are preliminary postcensal estimates based on the 2021 Census, not 2026 Census population counts. NPR administrative estimates remain revisable. Population and housing measures describe co-movement, not causation.
Reads onto the Big Question it bears on · the Big Question it bears on · the Canadian Capacity Monitor
Issuer revision · Material change
Population revisions lower the Q2 real-GDP-per-person gain to 0.6%, from the previously published 1.6%. Output per worker is now the largest positive term; the revised population grew rather than fell.
More on this: The anatomy of Canadian living standards →
Why it stands: The September 23 demographic release revises April 1, 2026 population from 41,417,056 to 41,718,064 and also revises the year-earlier denominator. On the same 2025 Q2–2026 Q2 window, real GDP remains +1.1%, while population changes from −0.45% to +0.54%. GDP per person changes from $60,944 to $60,504 and from +1.59% to +0.59%.
This source revision supersedes the earlier current claim that population fell and that population structure was the largest positive accounting term. In the identity, population structure contributes +0.22 log percentage points rather than +1.21; output per worker remains +0.80 and is now the largest positive term. The September 2 and September 15 readings remain dated historical evidence.
Trigger recorded for this reading: The next aligned quarterly GDP, population, productivity or hours observation or revision.
Limit: This is a revised Q2 reading, not a Q3 GDP observation. The existing convention pairs Q2 GDP with April 1 population and quarter-start monthly labour inputs. Log-change contributions describe arithmetic and do not identify causes. Business-sector and whole-economy coverage remain different.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor · the Brief it reads from
Evidence update · Minor change
New-home prices fell 0.1% in August and were 2.0% below August 2025. The national New Housing Price Index is 120.4, down from 120.5 in July. The existing housing-price comparison now includes August.
More on this: Can governments still solve housing shortages? →
Why it stands: Statistics Canada released the August New Housing Price Index on September 17. The registered series adds one observation and contains no changed shared-month values in this capture.
New-home prices remain below their year-earlier level, alongside August all-items consumer inflation of 3.0%; the two indices describe different baskets.
Trigger recorded for this reading: The September New Housing Price Index or a source correction.
Limit: The index covers contractor selling prices for new single, semi-detached and row houses, excluding sales taxes. It is not a resale-home index, a measure of all housing costs, or seasonally adjusted.
Reads onto the Big Question it bears on
Correction · Minor change
The residential construction investment series was labelled in 2017 dollars. Statistics Canada's source table uses a 2023 price basis. The label is corrected without rescaling any observation. The separate July update adds a month and incorporates the issuer's revisions to April, May and June.
More on this: Can governments still solve housing shortages? →
Why it stands: The issuer's June and July 2026 investment releases both identify table 34-10-0293 as the successor on a 2023 constant-dollar basis. The held record already uses that table and the same seasonally adjusted residential-investment vector.
The amounts describe construction volume at the source's 2023 price basis. They are not nominal spending and should not be compared as dollar levels with a series on a different price basis.
Trigger recorded for this reading: A source-definition change or the next investment release, scheduled for October 21, 2026.
Limit: This is a unit-label correction, not a conversion of observations from one price base to another. Construction investment does not measure completed or occupied homes.
Reads onto the Big Question it bears on
Evidence update · Material change
July building permits authorized 24,022 new dwelling units, 19.8% fewer than a year earlier on the unadjusted count. The trailing twelve-month total was 297,088, down 3.6% from the preceding twelve months. Residential construction investment was $14.643 billion in constant 2023 dollars, seasonally adjusted, 4.9% above July 2025. These series measure different stages and cannot establish how many homes were completed or occupied.
More on this: Can governments still solve housing shortages? →
Why it stands: Statistics Canada released July permits on September 16 and July construction investment on September 21. The captures also revise June permit units and April, May and June investment. The construction price-basis label is corrected separately; no observations were rescaled for that correction.
The permit count is lower over matching annual windows while measured residential construction volume is higher. That contrast describes authorizations and activity; it does not establish conversion rates, project delays or causation.
Trigger recorded for this reading: The August permits release on October 14 or August construction investment on October 21, or revisions to these series.
Limit: Permit counts are unadjusted, so month-to-month changes are not interpreted as seasonally controlled momentum. Constant-dollar investment is a volume measure rather than nominal spending. Permits, starts, construction spending, completions and occupancy are distinct; the detailed CMHC metropolitan completion panel remains on July.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Checked, unchanged · Minor reading
The job vacancy rate was 2.8% in July, unchanged from June and a year earlier. Employers reported about 501,000 vacancies; Statistics Canada describes both the monthly and annual comparisons as little changed.
More on this: Canadian Capacity Monitor →
Why it stands: The September 24 release adds July and revises six earlier vacancy-count observations. The held vacancy-rate observations are unchanged. The new month does not change this bounded reading of aggregate vacancy demand.
The latest release gives little evidence of a new direction in aggregate vacancy demand. It does not establish how easily an individual applicant can find work.
Trigger recorded for this reading: The next payroll and vacancy release, scheduled for October 29, or a source revision.
Limit: These are seasonally adjusted monthly vacancy estimates. A raw arithmetic difference does not establish statistical significance. The rate uses occupied-plus-vacant jobs as its denominator and is not an unemployment rate; a simple monthly average is not the separate quarterly vacancy survey statistic.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Evidence update · Minor change
August housing starts were nearly flat at 229,046 units at an annualized rate, against a revised 229,360 in July. The six-month average fell from 247,446 to 244,149. These Canada-wide rates describe construction beginning, not homes completed.
More on this: Canadian Capacity Monitor →
Why it stands: Statistics Canada’s September 17 mirror of CMHC’s September 16 release adds August and revises May from 254,178 to 254,597, June from 240,773 to 240,486, and July from 229,074 to 229,360. The rolling window drops August 2006 and the derived 2007 Q3 population-per-start point.
The national starts chart advances to August. The detailed metropolitan delivery panel remains on July; the national net-additions gap remains unresolved.
Trigger recorded for this reading: September starts or revisions to the monthly record.
Limit: Annualized rates are not actual monthly counts. The monthly rate and six-month average are distinct measures; neither establishes national net additions or whether the housing shortage is closing. August construction headlines cover centres of 50,000+; they do not replace this panel’s detailed metropolitan edition.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Evidence update · Material change
August is fully covered: 14,144.03 truck entries/day, including 1,778.42 US-plated (12.57%). July: 14,329.90 entries/day, including 1,811.55 US-plated (12.64%). August has more weekend days; entries rose within both weekday and weekend groups. Plate shares rose on weekdays but fell on weekends; these averages do not establish a trade trend.
More on this: US Exposure Monitor →
Why it stands: The September 14 capture added 31 August days and rolled off 31 older days; overlapping daily counts were unchanged. The September 16 source check found no update.
Coverage reaches August 31. Trade values stay on July; tariff coverage remains unmeasured.
Trigger recorded for this reading: New daily observations or revisions; no next release date is established.
Limit: Counts are unadjusted; plates do not identify cargo origin or duties paid. August straddles the United States duties' effective date and precedes Canada's counter-tariffs. No tariff effect is inferred.
Reads onto the US Exposure Monitor
Evidence update · Material change
August CPI rose 3.0 per cent over the year, matching July's rounded rate, while the unadjusted index fell 0.1 per cent during the month. Rent inflation rose from 2.5 to 2.8 per cent and remained below the headline for a second month; its gap narrowed from −0.5 to −0.2 percentage points. Average hourly wage growth slowed from 2.8 to 2.0 per cent, widening the rounded wage–price gap from −0.2 to −1.0 percentage point. CPI-trim is 1.9 per cent, CPI-median 2.0 and CPI-common 2.6.
More on this: The price of everything: one headline rate, and the components underneath it →
Why it stands: Reproduced the August observations from the governed index and wage records. The September 14 capture added August to all twenty-seven CPI series; six earlier CPI-common values were revised. Headline and component retained observations were unchanged.
The headline is steady while rent's own inflation rate rises and measured wage growth slows. Two observations below the headline do not establish a lasting change in rent inflation.
Trigger recorded for this reading: September CPI is scheduled for October 19, 2026, according to the issuer’s August release.
Limit: Component rates are not weighted contributions. The wage gap is a difference of rounded annual rates, not a real-wage index or a household estimate. Core measures remain revisable.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor · the Brief it reads from
Analysis update · Material change
The four-quarter Q2 2026 living-standards headline remains +1.6 per cent. New productivity and hours observations fill the previously unavailable output-per-worker split: output per hour contributes +1.19 percentage points, measured hours per employed person −0.29, and the scope reconciliation −0.10, against the +0.80-point output-per-worker total.
More on this: The anatomy of Canadian living standards →
Why it stands: The September 3 productivity release, captured in the Domain refresh, extends the business-sector productivity and hours records to Q2. The existing identity and log-change formulas are unchanged.
The missing accounting split is now observed, but business-sector and whole-economy coverage remain different. These terms locate arithmetic, not causes.
Trigger recorded for this reading: The next aligned quarterly observation or revision to GDP, population, productivity or hours.
Limit: These are contributions in a log-change identity, not ordinary percentage changes. The reconciliation is derived and is not an hours effect.
Reads onto the Big Question it bears on · the Brief it reads from
Analysis update · Material change
Current metropolitan housing completions are now part of the record. The national shortage remains unresolved: these gross construction counts do not measure national net additions.
More on this question: Can governments still solve housing shortages? →
Why it stands: CMHC’s direct June and July Excel tables provide a reproducible source beyond the three held Statistics Canada completion series, which still end in 2022. The metropolitan totals reconcile across the construction and intended-market tables.
The housing evidence gap is narrower. We can now show starts, unfinished construction and completions on a stated metropolitan basis, with intended-market detail. National benchmark-compliance percentages are replaced by separate institutional context.
Trigger recorded for this reading: The next CMHC edition, a revision to a held edition, clarified survey coverage, or a comparable national net-additions measure.
Limit: The broader provincial totals conflict with the published coverage notes. This reading uses metropolitan tables only. It does not infer cancellations, occupancy, linked construction cohorts or a trend from two unadjusted months.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Evidence update · Material change
Statistics Canada's August 28 release revises the federal deficit for fiscal 2025-26 on the National Accounts basis from $44.823 billion to $47.636 billion, an increase of $2.813 billion. The revised annual estimate is 1.5 per cent of the GDP measure for the same four quarters. Separately, April-June 2026, the first quarter of fiscal 2026-27, records net borrowing at a seasonally adjusted annual rate of $57.964 billion, narrower than the revised $61.700 billion in January-March 2026, the last quarter of fiscal 2025-26. That new quarter is outside the year being revised. None of these is an audited Public Accounts settlement.
More on this question: Does the federal budget land where it was promised? →
Why it stands: Compared preserved issuer evidence against the prior reading; new observations and revisions are counted separately.
Read the new period and prior comparison on their stated basis; preserve the previous dated reading.
Trigger recorded for this reading: The next issuer release or maintained capture for this product.
Limit: National Accounts and Public Accounts remain separate. Annual ratios use the same four quarters in numerator and denominator; SAAR quarterly levels are not fiscal-year totals.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor · the Brief it reads from
Evidence update · Material change
Between the IESO's July 10 and August 14, 2026 list vintages, the whole register grows from 2,044 to 2,062 records: 18 added and none removed. Under the same large-load rule -- load-class applications stating at least 75 MW, excluding sizes stated only in MVA -- membership rises from 67 to 71, with four entries and no exits. Across the whole register, 24 status strings change: 6 move to Withdrawn (four from Active and two from On-Hold), 4 move from Active to Complete, and 3 move from Complete back to Active. One of the 24 changes merely corrects 'Actice' to 'Active'. These are register states, not proof that six projects were cancelled or four projects now supply power. The issuer dates the newer list August 14; Countability captured it on September 2.
Why it stands: Compared preserved issuer evidence against the prior reading; new observations and revisions are counted separately.
Read the new period and prior comparison on their stated basis; preserve the previous dated reading.
Trigger recorded for this reading: The next issuer release or maintained capture for this product.
Limit: The historical engineering chapter retains its August 4 evidence cutoff. Cycle-2 reference links are identified, but their contents have not been newly adjudicated.
Reads onto the Brief it reads from
Evidence update · Material change
Employment fell by 41,700 in August 2026. The unemployment rate held at 6.4 per cent, the employment rate slipped from 60.9 to 60.8 per cent, and youth unemployment rose from 12.6 to 12.9 per cent. Over the twelve months to August, employment nevertheless rose by 216,500 against 187,000 additional people aged 15 and over in the Labour Force Survey population estimate: 1.16 additional employed people per additional person, up from 1.00 over the year to July. The higher annual ratio does not mean August employment improved. August 2025 was a local employment low: its 61,700 decline from July 2025 is larger than this August's decline, raising the trailing annual employment gain by 20,000 despite the weaker current month; the annual population gain also narrowed from 197,300 to 187,000. These are different windows, and the ratio does not explain participation, hours, or the full-time/part-time mix. Re-running the existing descriptive detectors with August evidence places the employment level shift around September 2024, one month later than before, and the nine-indicator labour cluster in January-July 2024 instead of December 2023-June 2024. These moving-window findings locate changes in measured levels; they do not establish a cause. Earlier dated readings retain the windows and evidence they originally used.
More on this question: Can Canada absorb its own population growth? →
Why it stands: Compared preserved issuer evidence against the prior reading; new observations and revisions are counted separately.
Read the new period and prior comparison on their stated basis; preserve the previous dated reading.
Trigger recorded for this reading: The next issuer release or maintained capture for this product.
Limit: Employment counts employed people, not jobs; the annual ratio compares changes in the age-15-and-over population and does not establish stronger current-month hiring.
Reads onto the Big Question it bears on · the Canadian Capacity Monitor
Evidence update · Material change
Third reading, on the July 2026 trade record, and the share set a record. Over the twelve months through July 2026 the United States bought 68.50 per cent of Canada's goods exports, against 74.08 per cent a year earlier -- a one-year fall of 5.58 percentage points. That window is the lowest United States share in the twenty-nine-year monthly record, and unlike the previous low it stands alone: the next-lowest window, ending April 2026, reads 68.99 per cent, half a percentage point higher. The single month is sharper than the window. Exports to the United States fell 6.6 per cent from June to July, ending a run of five straight monthly rises, while exports to every other country together rose 7.4 per cent to the highest monthly level this record holds. In July, 33.7 per cent of Canada's goods exports went somewhere other than the United States. The reallocation is still narrow: 52.7 per cent of the twelve-month rest-of-world rise is the United Kingdom alone, and on the customs record the all-countries unwrought gold, silver and platinum-group-metals series is up 43.9 per cent window on window. Motor vehicles found almost no other buyer -- a 9.7 per cent fall in United States-bound sales against a rest-of-world gain worth 5.6 cents of every dollar lost.
More on this: Canada's exports and the United States: the record break, and where the difference went →
Why it stands: Statistics Canada published the July 2026 international merchandise trade record on 3 September 2026 at 08:30, its own recorded release time, verified in the issuer's own cube metadata before any figure was read and committed as evidence. Both trade Resolutions were advanced to the 2026-09-04 capture under an explicit supersession chain: 48 publications re-pinned and 92 values re-derived from the newly archived bytes, each publication's series coordinate proven against its own already-published June figures before it was moved. The June and May vintages are retained on the record, superseded and unrewritten.
The break deepened rather than held. Through the spring the share sat within a tenth of a point of a low it kept not quite setting; July set it outright, by half a point, and did so through the United States side falling rather than through the rest of the world merely rising. What has not changed is the shape of the reallocation: outside the bullion channel it stays modest, and about seven of every ten export dollars still come from one customer.
Trigger recorded for this reading: The August 2026 trade record, which the issuer's own July release article states is scheduled for 6 October 2026. That August reference month is the first with any day on or after the date the United States duties became legally effective, but only its last ten days fall on or after that date -- it is only partly after the duties, not entirely after them.
Limit: All values are nominal, and partner shares are gross shipment values, not domestic value added; nothing here separates price from volume. Every monthly value behind the window remains revisable, and this release itself restated June. July 2026 ends before 22 August 2026, the date the United States duties became legally effective under the proclamations, and before 8 September 2026, the date Canada's counter-tariffs become legally effective. No month in this reading covers a single day under either measure, so no effect of either can be observed in it, and none is claimed. This record has not yet observed duties assessed or collected on shipments.
Reads onto the Big Question it bears on · the US Exposure Monitor · the Brief it reads from
Evidence update · Material change
The living-standards question has its second reading, and the answer changed. Real GDP per person rose 1.6 per cent over the four quarters to April-June 2026, to $60,944 in chained 2017 dollars. That is the largest four-quarter gain since the four quarters ending July-September 2022, and it leaves the per-person level 0.1 per cent below the record set in that same quarter. Of the 262 quarters this record holds a per-person figure for, going back to 1961, only two stand at or above today's level: July-September 2022 and April-June 2026. The first reading, taken in July, said real GDP per person was barely moving and that its small gain was entirely compositional. That is no longer true. What is still true is that the gain is not mainly a production story: real GDP grew 1.1 per cent over those four quarters while the population fell 0.5 per cent, so a smaller denominator supplied part of the rise. Splitting the total the way this record always splits it, population structure contributed +1.21 percentage points of the +1.57-point total, output per worker +0.80, the employment rate +0.02, and participation -0.46. Output per worker adding is a change: it subtracted in each of the two windows before this one, and last added over the four quarters to July-September 2025.
More on this: The anatomy of Canadian living standards →
Why it stands: Statistics Canada published the April-June 2026 National Accounts on 28 August 2026 at 08:30 Eastern, the issuer's own recorded release time. Countability refetched the two GDP series it holds from that table rather than copying figures out of the release, and classified the result against its own committed evidence before writing this sentence: two new observations, two restatements of January-March 2026, no period lost, and 522 shared observations that came back carrying exactly the values they carried before.
The record's own reading of the flagship question moves from 'growth is a population statement, not a productivity statement' to something more mixed: output per worker is adding again, but a falling population is still contributing more than any other single term. One quarter does not make a turn.
Trigger recorded for this reading: The business-sector labour-productivity release for April-June 2026, which would show whether output per worker gained from producing more per hour or from working more hours; then the next quarterly GDP release or population estimate that moves any term.
Limit: The output-per-worker term cannot be split for this window. Statistics Canada's business-sector labour-productivity and hours-worked table still ends at January-March 2026, one quarter behind the GDP record, so output per hour, hours per employed person, and the narrower national-accounts scope reconciliation are all unavailable for the four quarters to April-June 2026. This record reports them as unavailable and does not estimate them; earlier windows keep the three-part split. Per-person figures divide a National Accounts aggregate by Statistics Canada's quarterly population estimate, which is itself revisable, and the two come from different survey frameworks reconciled at the ratio. The identity locates where a change shows up in measurement; it never says why. This reading was taken on 2 September 2026, five days after the release it reads. Adapted from Statistics Canada, table 36-10-0104-01, 2026; this does not constitute an endorsement by Statistics Canada of this product.
Reads onto the Big Question it bears on · the Brief it reads from
Issuer revision · Material change
The same release restated a quarter this record had already published, and this is what the restatement did. Statistics Canada raised real GDP for January-March 2026 by $2,924 million, from $2,500,680 million to $2,503,604 million in chained 2017 dollars -- a rise of 0.12 per cent -- and raised the same quarter at current prices by $6,904 million, from $3,321,588 million to $3,328,492 million, a rise of 0.21 per cent. Nothing else in the table moved: 522 shared observations were compared and 520 came back identical. Four published readings changed as a result, none of them by an amount a reader would have guessed from the size of the restatement. First, real GDP for January-March 2026 measured against the same quarter a year earlier changed sign, from -0.05 per cent to +0.06 per cent: on the evidence published in May the economy was very slightly smaller than a year before, and on the evidence published in August it was very slightly larger. Second, the four-quarter change in real GDP per person ending in that quarter -- the reading this record published in July -- moved from +0.19 to +0.31 percentage points, before the new quarter was counted at all. Third, on the defence-spending question, the four quarters of nominal GDP spanning fiscal 2025-26 now average $3,271,267 million instead of $3,269,541 million, so the same $62,713 million of defence spending falls $2,712 million short of two per cent of GDP rather than $2,678 million; the recomputed ratio reads 1.92 per cent either way, so what that comparison concludes is unchanged. Fourth, on the fiscal question, Public Accounts net debt measured against GDP moved from 44.9 to 44.8 per cent -- the debt figure did not move, the denominator did. Countability's forecast scorecard also moved: measured against the restated quarter, the Bank of Canada's July 2026 projection for that quarter is about three and a half times further from the mark than it was against the quarter as first published, while its January and April 2026 projections are closer.
More on this question: Can Canada raise living standards, not just its population? →
Why it stands: Statistics Canada publishes the quarterly national accounts as revisable and restates recent quarters as later source data arrives; the cube carried no correction footnote, so this is an ordinary scheduled restatement, not an issuer correction. Countability compared the issuer's bytes against its own committed bytes twice -- once against what the issuer was serving on 2 September 2026, once against the evidence committed to this record before the release -- and both comparisons agree exactly on what moved.
This is the first time this record has published what a restatement did to its own readings rather than simply absorbing one. The useful lesson is that a restatement too small to notice in the level -- one part in a thousand -- was large enough to flip the sign of a published year-over-year reading that happened to sit close to zero, and large enough to move three other published figures on three different questions. Where a reading sits near a boundary, the size of a revision matters far less than where the reading sits.
Trigger recorded for this reading: The next quarterly National Accounts release, which by the issuer's own practice may restate these quarters again.
Limit: This states what one restatement did to figures this record publishes. It is not a measure of how much Statistics Canada's national accounts revise in general: that would need a series of vintages, and this record holds one earlier vintage of this table, not a history of them. The issuer's own real-time vintage table for expenditure-based GDP is not in this record. No cause is assigned to the restatement; the issuer's revision practice is stated as the issuer states it. The 44.8 per cent and 1.92 per cent figures are recomputations of existing published comparisons against a moved denominator, not new claims about debt or defence spending. Adapted from Statistics Canada, table 36-10-0104-01, 2026; this does not constitute an endorsement by Statistics Canada of this product.
Reads onto the Big Question it bears on · the Brief it reads from · the Brief it reads from · the Brief it reads from
Checked, unchanged · Material reading
The capital question was re-checked against the release it was waiting for, and not one of its figures moved. Capital stock per worker is still down 3.7 per cent since 2019 and 5.7 per cent from the record's 2014 peak; real business investment per worker is still down 7.2 per cent; the investment share of GDP still barely moved, from 20.6 per cent in 2014 to 18.6 per cent in 2025. The new quarter is real and is recorded: business gross fixed capital formation ran at $427,991 million in chained 2017 dollars, seasonally adjusted at annual rates, and $618,156 million at current prices, in April-June 2026.
More on this question: Is Canada rebuilding the capital behind each worker? →
Why it stands: Every figure this question publishes is a calendar-year average or a year-end level, and the newest calendar year it publishes is 2025. The release added one quarter, April-June 2026, which sits inside a calendar year that is not finished and therefore inside no published window, and restated one quarter, January-March 2026, which sits outside every published window too. Countability refetched all four source tables behind this question and compared them observation by observation with the archived bytes its figures are pinned to. The two annual records -- the capital-stock record and the multifactor-productivity record -- returned no change of any kind. The two quarterly records changed only in the two quarters just named.
An unchanged reading after a real release is a result, not a gap: the release that could most plausibly have moved this question did not move it. The first calendar year that can extend these windows is 2026, and that needs all four of its quarters.
Trigger recorded for this reading: The remaining two quarters of 2026, which together with these would complete a calendar year; the next annual capital-stock release; or the next multifactor-productivity release.
Limit: Unchanged means unchanged after a complete re-fetch and an observation-by-observation comparison against the archived bytes, not unchecked. The two quarterly figures quoted here are stated as the freshest published quarter and are deliberately not folded into any per-worker or share figure, because a single quarter is not a calendar year. The capital-stock record still ends at 2024 and the multifactor-productivity record at 2023; that lag is this question's own standing evidence gap and this release did nothing to close it. Adapted from Statistics Canada, tables 36-10-0104-01, 36-10-0108-01, 36-10-0097-01 and 36-10-0208-01, 2026; this does not constitute an endorsement by Statistics Canada of this product.
Reads onto the Big Question it bears on · the Brief it reads from
Copyright © 2004-2022 Independent Electricity System Operator, all rights reserved. This information is subject to the Terms of Use set out in the IESO’s website (www.ieso.ca)