2026-08 — full dated evidence record

In July, the United States moved to put a 50 per cent tariff on lists of Canadian goods -- three separate proclamations, each with an effective date of August 19. The U.S. then delayed that effective date by three days, to August 22 -- the proclamation that did this says senior U.S. officials reported that Canada had promised to change the practices at issue, not a finding the proclamation makes in its own voice. But no deal was reached. On August 21, the Prime Minister said last-minute changes in the U.S. position were unfair, and put the reliability of any deal in doubt. He suspended trade talks and called Canada's negotiators home. The duties became legally effective the next day, August 22, under the proclamations, on roughly $28 billion of Canadian goods. This record has not yet observed duties assessed or collected on shipments. The government's own figure doesn't say which country's dollars. Canada said it would match the tariffs dollar for dollar. The day after that, the Prime Minister said Canada's own measures would begin 'the Tuesday after Labour Day.' This record does not turn that into a calendar date, and at the time, no rate or full list of goods had been given. The most recent trade figures on this record are from June, before any of this happened.

More on this: The Tariff Ledger →

Why it stands: This reading rests on six primary documents: three U.S. proclamations signed July 20, 2026; the proclamation that delayed their start; and the Prime Minister's own statements on August 21 and August 22, 2026. All six are archived here with their own content hashes, and read into a new Resolution that leaves this record's 2025 tariff entry untouched. The U.S. Exposure Brief was updated when this reading was published, to carry it next to its separate, unchanged trade reading.

The exposure this record already measures now sits under a new legal authority: a flat 50 per cent extra rate on certain products of Canada, as each proclamation's own Annex II sets out. The transcribed operative text names no exemption for goods that would otherwise qualify under USMCA. Canada has said it will retaliate, starting 'the Tuesday after Labour Day' in the Prime Minister's own words -- but at this point no rate or full product list had been published yet. Whether or how the trade share moves is a question for the next trade release, not this one.

Trigger recorded for this reading: The July 2026 trade record, at Statistics Canada's next scheduled monthly release. Canada's own retaliation rate, scope, and effective date were published August 25 -- see the newer reading above.

Limit: None of the three proclamations states a total dollar value for the goods newly subject to its duty. The 'roughly $28 billion' figure is the Government of Canada's own statement, in a currency it does not designate, and it is not reconciled here against any U.S. figure. At the time, Canada's own retaliatory tariffs had no published rate and no full product list in either of the two Canadian primary sources this record had captured by then. The Prime Minister's own words placed the timing at 'the Tuesday after Labour Day' -- this record did not convert that into a calendar date. Each proclamation also names an Annex I once, as a possible exception to how these duties stack on other duties; this record has not read what Annex I contains. This record does not attribute any trade movement to any action, and none has been measured yet. Update: Canada published its own rate, scope, and effective date on August 25 -- see the newer reading above.

Reads onto the Big Question it bears on · the Brief it reads from

Countability's first dated reading on Canada's opioid-toxicity death record. The Public Health Agency of Canada's released file reports 5,630 apparent opioid toxicity deaths in Canada in 2025. That is 1,696 fewer than the 7,326 it reports for 2024 -- a fall of 23.2 per cent as currently reported. But 5,630 is not a settled count. The agency's own technical notes say that data resting on open investigations “are considered preliminary and subject to change”, and that an investigation “can range from approximately three to twenty-four months”. Measured back from the file's own cut-off of 12 May 2026, that window still reaches across every month of 2025. The file itself says none of this. None of its eleven columns flags an observation as preliminary. No cell in its 30,316 rows carries the words “preliminary”, “provisional” or “subject to change”. So a reader holding only the machine-readable file cannot tell which of its numbers are still moving.

Why it stands: Every count here is re-derived from the Open Government Licence data file itself, preserved by Countability on 10 August 2026 and read from the preserved copy: 30,316 rows, of which the 700 that carry overall apparent opioid toxicity death counts are the slice used. Each quotation is checked against preserved bytes too — the technical notes were fetched and preserved on 18 August 2026 for that purpose, because they are not part of the data file and are the only one of the four documents held here that writes the provisionality statement and the Quebec definition down. It is the publisher's own README, released inside the data package, that points a reader from the file to those notes.

A large real fall and a reporting lag look identical in these bytes, so anyone treating 5,630 as a settled 2025 total is treating as final a number the publisher has not called final. Two further properties of the same file travel with anyone who uses it for anything. The file does mark suppression where it applies — 3,999 of its 30,316 values read “Suppr.” in the column that carries the numbers — but it nowhere reconciles what that marking does to its own totals, and it carries no provisionality flag at all. Its national annual totals exceed the sum of its own four published quarters in every one of the ten years on file, by 121 deaths in total, and all 121 trace to exactly two provinces — Prince Edward Island 70 and Newfoundland and Labrador 51 — with zero residual across the other eleven jurisdictions. The technical-notes page states the general point, that “quarterly totals for Canada may not equal the annual totals due to suppressed data for some provinces and territories with low numbers of deaths”, and names those two provinces among those suppressing counts between one and four in quarterly data; the size of the discrepancy, and its decomposition to those two provinces alone, are this record's own derivation from the file, and appear in none of the four documents held here. And Quebec's quarterly count roughly doubles between the last quarter of 2023 and the first of 2024, from 81 to 160, and stays there: the notes give the reason, and it is a definitional change — Quebec's 2016-to-2023 figures “include deaths with completed investigations only”, while data “for 2024 onwards from Quebec include presumed unintentional deaths with ongoing investigations”. Both are computable in a few lines from the released file.

Trigger recorded for this reading: The next quarterly update of the file — the first occasion on which 2025's figures can be compared against the versions published here.

Limit: Every figure here is the publisher's own reported count, never a Countability estimate, and the 2025 figures are not final. This record does not say which of them will move or in which direction: the publisher states no direction, and none is inferred. The comparison is deaths counted in 2025 against deaths counted in 2024 on the file as it now stands, and a revision to either year would change it. The publisher's own README instructs that “comparisons over time and between provinces and territories should be interpreted with caution”, because differences may exist in case identification and reporting between jurisdictions, and that instruction is honoured here: no jurisdiction is ranked against another, and Quebec's figures before and after the start of 2024 are not comparable with each other on the publisher's own account. Suppressed cells are excluded from every calculation and are never treated as zero, so no total here is complete for the provinces that suppress. This reading is about counts and about what the file does and does not tell its reader; it says nothing about causes, about any policy, or about anyone's response to the crisis these numbers record. Four documents are in custody for it: the licensed data file, released as a package that also carries the publisher's own README; the publisher's technical-notes page; the machine-readable column description the publisher's data service publishes for its opioids table, which names that table's dimensions and carries no notes and no values; and the open-data catalogue record the file is released under, which is where its quarterly frequency is stated. Every statement here about what is or is not written down is a statement about those four and about nothing else the publisher maintains. Contains information licensed under the Open Government Licence – Canada; the technical notes are the publisher's own page, quoted with attribution and not reproduced.

Countability's first dated reading on Canada's conservation commitment -- and it is a subtraction. Canada's stated target, in Environment and Climate Change Canada's own words on its conserved-areas indicator page, is “to conserve 25% of its lands and 25% of its oceans by 2025, and 30% of each by 2030”. The same department's own measurement of the same thing, in its Canadian Protected and Conserved Areas Database summary table dated December 2025, is 14.0 per cent of Canada's land territory conserved, and 15.5 per cent of its marine territory. Against the 2030 target, that is a gap of 16.0 percentage points on land -- 1,594,621 square kilometres -- and 14.5 points at sea, or 831,037 square kilometres. The 25 per cent milestone's own stated date has now passed; measured against that milestone at the same reference date, the gaps are 11.0 points on land and 9.5 at sea. Nothing here says whether either target will be met. Both percentages are the publisher's own, and so is the 2030 gap -- its own indicator page states that distance. What this record adds is the same subtraction against the milestone whose date has passed, plus the square-kilometre size behind each of the 2030 gaps.

Why it stands: Both halves are read from bytes this record holds. The percentages are the publisher's own and reproduce from the publisher's own square-kilometre columns in the same file: 1,400,780 divided by 9,984,670 is 14.03 per cent, and 893,963 divided by 5,750,000 is 15.55, matching the published 14.0 and 15.5. The target sentence is quoted from the indicator page, which is itself a declared resource of the Open Government Licence dataset “Canada's conserved areas” — this record fetched and preserved that dataset's catalogue record on 18 August 2026 in order to rest the quotation on the licence rather than on an argument about quotation length.

The indicator page carries the target sentence, the same two measured percentages with their square-kilometre magnitudes, and the publisher's own statement of the distance to the 2030 target, which it writes as “14.5% below” for marine areas and “16.0% below” for terrestrial ones — the same two figures reported above as percentage points. The summary table carries the December 2025 measurement itself, jurisdiction by jurisdiction, with the baseline areas each percentage is taken over, which is what lets the percentages be reproduced rather than accepted. What this record adds to them is the subtraction against the 25 per cent milestone — 11.0 points on land and 9.5 at sea — which neither document states. That milestone is the sharper of the two, because its date has already passed: that gap is contemporaneous rather than forward-looking.

Trigger recorded for this reading: The next annual vintage of the publisher's protected-and-conserved-areas statistics table.

Limit: This is one dated snapshot set against a stated target, and nothing more. It is deliberately not a trend. The publisher's own account is that other effective area-based conservation measures “were formally recognized as a policy approach in 2017 and 2018” and were added to the database then “even though the measure(s) may have been established in previous years”, that areas “are accounted for in the year they were recognized and not in the year they were established”, and that for those reasons “comparisons with previous reports should be made with caution” — so no series is published here and none of these figures should be read against an earlier report's. The publisher's own downloadable annual series for this indicator was acquired alongside this reading and carries an earlier reference date than the page it sits beside: its own note says its data are current as of 31 December 2024, where the page's say 31 December 2025, and its square-kilometre figures differ from the page's in every one of the thirty-five years the two share. That is why no long-run trend is published here at all. The percentages are the publisher's best available estimates of a spatial quantity, the marine baseline is its own “approximately 5 750 000” square kilometres, and the provincial and territorial rows are not a league table — the publisher states that its own provincial and territorial totals carry artefacts at those borders, accounted for only nationally. Contains information licensed under the Open Government Licence – Canada.

Countability's first dated reading over a century and a half of federal debt. Finances of the Nation publishes one file covering federal government debt from Confederation onward. The copy this record holds carries 156 consecutive annual observations of net federal debt as a share of GDP -- one for every year from 1867 to fiscal year 2021/2022, none missing. The last observation is 43.71 per cent of GDP, at fiscal year 2021/2022: that year ended 31 March 2022, not the present. The highest reading anywhere in the run is 110.31 per cent, at fiscal year 1945/46. The lowest is 11.60 per cent, at 1912/13. This is a reference spine, not something this record follows on a schedule: as of a 10 August 2026 read of the metadata the file's host repository keeps for it, it was last modified on 6 January 2023, and its series ends four fiscal years before this reading was written.

Why it stands: The series is read from the publisher's own file, preserved by Countability on 10 August 2026 and read from the preserved copy. The share-of-GDP normalisation is the publisher's own column, not a Countability calculation; the four figures above are that column's last value, its maximum, its minimum and its observation count, computed from the file. The endpoint is stated three times over in bytes this record holds — the file's own last year, its own last fiscal year, and the publisher's own dataset page, which says the data run “from 1867 to March 31, 2022”.

A contemporary fiscal number means something different placed against 1946 than placed against 2019, and until now this record had no way to place one at all. What a spine like this supports is exactly that: a dated endpoint against a dated maximum and a dated minimum, on one publisher's single consistent basis. What it does not support is any statement about today — and the endpoint labelling is doing real work here, because a chart of this file captioned “to present” would misdate its own last point by four fiscal years.

Trigger recorded for this reading: A newer vintage of the publisher's file, which is not scheduled.

Limit: The endpoint is fiscal year 2021/2022 and must never be restated as “current” or “present”. The publisher's own title for the series, in a small companion file read in the same 10 August 2026 acquisition and recorded then rather than preserved here, calls it “1867 to present”; the dataset page that is preserved here dates the same run to 31 March 2022, and that is the date this reading uses. Only the share-of-GDP column is used. The file's real-dollar and real-per-capita columns are deliberately not used and not quoted: their published labels read “Real 2019”, while the file's own arithmetic puts them in 2022 dollars — real equals nominal in 2022 and nowhere else — and nothing in the documents held here resolves that discrepancy, so those two columns are set aside rather than reconciled. The denominator behind “share of GDP” is the publisher's own and is not identified in the bytes held here; for a series anchored in 1867 the nineteenth-century denominator is necessarily a historical reconstruction, so the early values are the publisher's construction and are cited as such rather than as measurements. Two modification dates sit behind the sentence above, and only one of them is in custody. The publisher's dataset page preserved here carries its own modification stamp in its metadata — a page-level date of 6 January 2023, which is a property of bytes this record holds. The data file's own modification date is not: it comes from the file listing kept by the repository the publisher hosts the file on — the Open Science Framework, file 8gxwc — read on 10 August 2026 and recorded then, and never preserved, so it is cited as that read rather than as a current fact. The two land on the same day and are not the same measurement: a page can be edited without the file moving, and the file can move without the page being touched. What is held here is the file itself, hashed and matched against the digest that same read reported for it, and the publisher's dataset page. Nothing here is a forecast, a sustainability judgement, or a comparison against any other country or any other publisher's debt measure, and no Countability reading is built on top of it yet. Source: Finances of the Nation, licensed under Creative Commons Attribution 2.0; the figures quoted are the publisher's own.

Ontario's 2024 municipal Financial Information Return changed between two copies Countability kept -- and it changed in three ways at once. Of the 402 municipalities present in the copy kept on 25 July 2026, exactly two have figures that moved by the copy kept on 11 August. Port Hope's return moves a 175,000-dollar amount between two development-charges reserve-fund lines: in two columns, the amount moves off the line the province labels “Other” and onto the line it labels “Highways (Roads and Structures)”, while in a third column the same amount moves the other way. South Bruce's return moves along its asset-retirement-obligation chain: accretion expense and total asset retirement obligations each rise by 4,181 dollars, from 37,380 to 41,561. The third change is not a restatement at all: three municipalities -- Douro-Dummer, Highlands East and Atikokan -- appear in the file for the first time, with no rows at all in the earlier copy. That takes the file from 402 filers to 405.

Why it stands: Ontario publishes this filing year at a single address and replaces the file there; nothing this record holds — not the file, not the catalogue record captured with it — is an erratum or a dated archive of the copy replaced, and whether Ontario publishes either elsewhere was not checked. So the only way to see that a filing year has moved is to hold two copies of it and compare them. Countability keeps one at every capture. The comparison is made on the return's own cell addresses — the schedule-line-column codes the province itself assigns — so the identity is the issuer's, not one this record invented, and both copies were checked against the hashes recorded when they were captured before any figure was taken from them.

Port Hope's change is one reporting line added, two removed and four cells changed — one of the four being the filer's own date stamp, which moved from 20 April 2026 to 6 August. South Bruce's is one added, five removed and seven changed, its stamp moving from 6 July to 7 August, and a 5,981-dollar figure the earlier copy reported on the line “Prior period adjustment” now appears on the line “Write-downs of tangible capital assets”, with restated beginning-of-year net financial assets falling by that same 5,981. The three newly present filers bring 1,220, 1,185 and 1,925 reported lines with them. This is the second closed Ontario filing year this record has watched move, and it shows two things the first did not: a file can change by gaining filers rather than by restating figures, and a single filer's change can be a small annotated adjustment rather than a large correction. Anyone still holding the July copy of the 2024 return holds a file that is three municipalities short and two filers out of date, and nothing inside the replacement file itself announces either.

Trigger recorded for this reading: The next weekly capture of the Financial Information Return.

Limit: This states what changed and by how much. It does not state why, and nothing here should be read as saying: a correction, a reclassification, an audit adjustment and a late filing all look identical in these bytes. The only statements of intent in the file are the filers' own — their last-updated stamps, and, on one of the lines South Bruce removed, that filer's own annotation “ARO accretion due to estimate change” — and they are quoted as the filers' words, never adopted as this record's explanation. Line labels are the return's own throughout: a line the province labels “Other” is reported here as “Other” and is not re-described. “Two of 402” counts only the filers that were already in the earlier copy and whose figures moved; the 402 is that copy's own filer count and is not Ontario's municipality count. Five municipalities differ between the two copies in all — the two whose figures moved, and the three that were not in the earlier copy at all — and those three are counted separately for that reason and are never described as restatements. The three newly present filers are new to this file, which is not a statement about when they filed, and no claim is made about whether Ontario announced any of this. Countability publishes no figure derived from this source. Contains information licensed under the Open Government Licence – Ontario.

July’s Consumer Price Index is now in the record, and it rearranges the price story this record has been telling. Prices overall rose 3.0 per cent over the year to July, up from 2.8 per cent in June. The index itself rose 0.5 per cent on the month. Rent inflation fell to 2.5 per cent — its lowest reading since November 2021. For the first time since January 2023, rent rose more slowly over the year than prices overall: the gap between rent and the headline rate was +0.7 of a percentage point in June, and is now −0.5. Energy moved fastest over the year: gasoline stood 25.7 per cent above a year earlier, and energy as a whole was up 16.6 per cent. May’s readings were higher still, at 33.2 and 22.2, so July is a re-acceleration inside a volatile year, not a peak. These are the fastest-moving of the twenty-two price series this record holds from the headline table, not measured shares of the headline number; this record holds no basket weights. The Bank of Canada’s preferred core measures sit below the headline: 1.9 per cent (CPI-trim), 2.0 per cent (CPI-median), and 2.7 per cent (CPI-common). Measured pay rose 2.8 per cent over the same year, 0.2 of a point behind prices.

More on this question: Can governments still solve housing shortages? →

Why it stands: Statistics Canada published July’s index on 17 August, and Countability refetched every price series it holds rather than copying figures out of the release — twenty-seven series now, expanded this month from two so the headline can be decomposed rather than quoted. For the headline table, all 239 months already on file came back carrying exactly the value they carried before, and the frozen June record still reproduces to the digit, so July is a new observation and not a restatement of older ones. The one exception is the issuer’s own and is by design: Statistics Canada restated June’s CPI-trim from 1.8 to 1.9 per cent, and six historical CPI-common readings, because those measures are published as revisable — trim and median rest on seasonally adjusted inputs, and common’s model is re-estimated over its whole history each month. That is reported here as a restatement, not as an error.

The rent reading behind the housing question changes in character rather than merely in size: for the first time in this record’s run of published readings, rent is not outrunning overall prices. The level shift the record locates in early 2025 still stands, and the maintained correlation between rent inflation and net non-permanent-resident inflows now reads against a rent series that decelerated from 3.5 to 2.5 per cent in a single month. Whether July is a turn or a wobble is exactly what the next readings decide; nothing here settles it.

Trigger recorded for this reading: The next monthly index — in particular whether a second month shows rent inflation at or below the headline.

Limit: One month is one month, and every direction named here is a single reading old. “Rent” is the rent component of the price index — what tenants across existing and new leases are paying — not asking rents on newly listed units. The year-over-year comparisons span Statistics Canada’s May 2026 update to the index’s basket — the three most recent months, July included, sit on the updated basket and the other nine predate it — and a basket effect must never be read as an economic movement. Component comparisons rank rates of change, not contributions: this record holds no basket weights, so it cannot say how much of the 3.0 came from where. The energy figures compare against weak months a year earlier; a base effect is not a trend. No comparison here reaches earlier than the record’s claim floor of July 2006; the held window itself now begins in August 2006, one month later, its oldest month rolling off with each release. Core measures are the issuer’s published figures, revisable by design. Adapted from Statistics Canada, table 18-10-0256-01, 2026; this does not constitute an endorsement by Statistics Canada of this product.

Reads onto the Big Question it bears on

Ontario republished the 2023 municipal Financial Information Return -- a filing year that closed three years ago -- and one filer's numbers changed. Stratford's return now differs from the version published in July: 581 reported lines carry different values, and a further 78 lines were added and 77 removed. Its equity in tangible capital assets is 47.1 million dollars higher than the figure previously published. Its Ontario conditional grants are 11.5 million dollars lower than the figure previously published. A prior period adjustment of 5.8 million dollars now appears where none was reported before. The filer's own last-updated stamp moved from 22 October 2025 to 4 August 2026. Of the 436 filers present in the file, Stratford is the only one whose figures moved at all.

Why it stands: Countability keeps a copy of this file each time it is published, so it can compare two versions of the same closed year line by line. The comparison reads the filing's own content — member names, uncompressed sizes and checksums — rather than the archive's packaging timestamps, so a routine rebuild of the file cannot look like a restatement. Without the two kept copies there would be nothing to compare: the published file is replaced in place, and the version it replaced is not offered anywhere.

A closed filing year is not necessarily a settled one. Anyone who took the July version of Stratford's 2023 return as final now holds figures the issuer has since replaced, and the replacement is not announced in the file. This is a demonstration of why the copies are kept, not a finding about Ontario's municipal finances or about this municipality.

Trigger recorded for this reading: The next weekly check of the Financial Information Return.

Limit: This states that these cells changed and by how much. It does not state WHY, and nothing here should be read as one: a correction, a reclassification, an audit adjustment and a late consolidation all look identical in the bytes, and which it was is the filer's business. The only statement of intent in the file is the filer's own last-updated date, which is why it is quoted. "1 of 436" counts the filers present in this file, which is not Ontario's municipality count. Countability publishes no figure derived from this source; this reports a change in someone else's record, not a change in its own. Whether Ontario issued any notice of the republication was not checked, and no absence of notice is claimed. Contains information licensed under the Open Government Licence – Ontario.

The unemployment rate was 6.4 per cent in July 2026 -- its lowest reading since July 2024. It has now fallen for three months running: from 6.9 per cent in April, to 6.6, to 6.5, to 6.4. For people aged 25 to 54 it was 5.5 per cent. The share of everyone aged 15 and over in work was 60.9 per cent.

Why it stands: Statistics Canada released the July 2026 Labour Force Survey at 8:30 a.m. Eastern on 7 August 2026. This record captured the table on 10 August 2026 and every figure here is read from that capture, not from the issuer's summary.

Two years is the exact span: the last month at or below 6.4 per cent was July 2024. Over the twelve months to July 2026 the employment rate rose two tenths of a point while the participation rate — the share of the population working or looking for work — fell one tenth, to 65.1 per cent. On those two figures the share of people in work rose without the share taking part rising.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: These are seasonally adjusted estimates from a monthly household survey and each carries sampling error; a one-tenth-of-a-point move is small against it, and three such moves in a row is a description of the series, not a trend established here. A falling unemployment rate is consistent both with more people finding work and with fewer people looking for it, and these figures on their own do not separate the two.

Reads onto the Canadian Capacity Monitor

Youth unemployment was 12.6 per cent in July 2026 -- the lowest reading since February 2024, when it was 12.1 per cent. The share of people aged 15 to 24 in work was 55.4 per cent. No month since May 2024 has been that high; May 2024 was 55.4 per cent exactly, so July 2026 matches it rather than beating it. A year earlier, those two figures were 14.5 and 53.7 per cent.

Why it stands: Statistics Canada released the July 2026 Labour Force Survey at 8:30 a.m. Eastern on 7 August 2026. This record captured the table on 10 August 2026 and every figure here is read from that capture, not from the issuer's summary.

The Monitor's entry-level module watches one thing: the gap between youth and prime-age unemployment against its own historical band. That gap was 7.1 percentage points in July 2026, against 8.7 points in July 2025 and 8.7 in July 2024 — the first of those three Julys in which it narrowed. Both parts of the gap fell over the year; the youth part fell further.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: Age fifteen to twenty-four is used throughout as a stated stand-in for someone starting a working life, which it is not precisely: it counts students working part time and misses anyone starting a first career later. Two measures moving together over twelve months is not a verdict on whether entry to work has become easier, and nothing here says why either moved.

Reads onto the Canadian Capacity Monitor

All of the past year's net employment growth, and more, was full-time. Over the twelve months to July 2026 full-time employment rose by 205,700 while part-time employment fell by 9,200, for a net gain of 196,500. Part-time work went from 18.38 to 18.16 per cent of all employment.

More on this question: Can Canada absorb its own population growth? →

Why it stands: Statistics Canada released the July 2026 Labour Force Survey at 8:30 a.m. Eastern on 7 August 2026. This record captured the table on 10 August 2026 and every figure here is read from that capture, not from the issuer's summary.

The absorption reading published beside this one gives the size of the twelve-month employment gain and states explicitly that it says nothing about the full-time and part-time mix. This is that mix, on the same window and the same survey: the net gain is smaller than the full-time gain because part-time work shrank.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: Full-time and part-time are the survey's own classifications, resting on usual hours at the main job. They say nothing about pay, permanence, or whether the hours are the ones the worker wanted. These are net changes in levels: they hide how many people moved between the two, and they are differences of survey estimates, each with its own sampling error.

Reads onto the Big Question it bears on

Employment growth is not what pushed the absorption ratio close to one. Over the twelve months to July 2026 the ratio reached 0.996. It got there because the number it divides by got much smaller: the year's gain in the Labour Force Survey's working-age population estimate fell from 715,000 to 197,300 — 72.4 per cent smaller. Employment grew 196,500 over that year. Over the twelve months to July 2025 it grew 293,200. So the ratio moved from 0.41 to almost 1.00 while the annual employment gain itself got smaller.

More on this question: Can Canada absorb its own population growth? →

Why it stands: The reading published on 11 August compared the July window against the June window, where the employment gain rose and the population gain fell — two moves that push the ratio the same way, upward. Against the same month a year earlier both gains are smaller than they were, and the ratio still rises. That is the comparison a reader needs before taking a ratio near one as news about employment growth.

A ratio at or near one says employment growth is keeping pace with growth in the working-age population. It does not say the pace of employment growth rose. On this evidence the annual employment gain is about a third smaller than a year ago and the population gain is about three-quarters smaller. Both readings are true at once and the ratio alone cannot tell them apart.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: Both terms are twelve-month differences of monthly household survey levels, so both are volatile. Employment here is a count of people in work at each date, so a twelve-month change in it is a net change in that count and never a count of people hired. This compares two twelve-month windows a year apart and describes arithmetic only: nothing here establishes why either term moved, and the working-age population estimate is the survey's own, not a population count.

Reads onto the Big Question it bears on · the Canadian Capacity Monitor

The July release added a month and restated nothing this record can see. This record compared its copy of the labour table from before the release with its copy from after. All 3,491 monthly values the two copies share are identical. Each of the ten series gained exactly one new month, and of the earlier months held on both sides, not one moved.

Why it stands: The comparison was recomputed value by value from this record's own stored copies of the table taken either side of the 7 August release, rather than inferred from a file being different. Statistics Canada's own description of the table carries no correction note.

Restating months already published is ordinary in official statistics, and this record counts it as a different kind of event from new data. Here there was none, so every movement in the labour readings published beside this one comes from the new month alone rather than from the past being redrawn.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: This covers the ten national, seasonally adjusted series this record draws from the table. It says nothing about the rest of the table, about unadjusted estimates, or about the survey's annual seasonal-adjustment revision, which lands at a different time of year. Seven of the ten series are held only for the most recent 240 months, so a restatement older than that window would not be visible here.

Reads onto the Canadian Capacity Monitor

One more month of labour data changes the absorption answer. Over the twelve months to July 2026, employment rose by 196,500 while the Labour Force Survey's working-age population estimate rose by 197,300. That is 1.00 jobs added for each additional working-age person, as published -- 0.996 before rounding. On the June window this record read 0.45. Employment growth has gone from covering under half of working-age population growth to coming within half a per cent of matching it.

More on this question: Can Canada absorb its own population growth? →

Why it stands: Statistics Canada's Labour Force Survey table now runs to July 2026 in this record's own capture, taken 2026-08-10. Two things moved at once and both push the same way: the twelve-month employment gain almost doubled, from 99,000 to 196,500, and the twelve-month working-age population gain fell, from 219,500 to 197,300.

A ratio at or above one means employment growth is keeping pace with growth in the working-age population; below one means it is not. This is a return towards parity rather than a first: the ratio has been at or above one in 42 of the 229 months this record holds, most recently in January 2023, and it has not quite reached one now either — the published 1.00 is a rounding of 0.996. It says nothing about who holds the added jobs, about hours, or about full-time and part-time mix.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: This is a ratio of two twelve-month changes, both from the same monthly household survey, and it is volatile because both parts are differences. None of the movement here came from the issuer restating earlier months — recomputed on the same June window with the newly captured data, the ratio is unchanged at 0.45, so the entire move is the new month.

Reads onto the Big Question it bears on · the Canadian Capacity Monitor

Immigration, Refugees and Citizenship Canada's monthly records now run to May 2026. Three of the four monthly flows this record follows are falling less steeply than a month earlier, and the fourth is rising faster. Study permits are down 12.4 per cent on the previous twelve months, against 16.6 per cent in the April window. Work permits went from 3.3 per cent up to 9.8 per cent up. Permanent-resident admissions are down 15.6 per cent, against 17.1 per cent. Asylum claims are down 34.9 per cent, against 35.2 per cent.

More on this question: Can Canada absorb its own population growth? →

Why it stands: Each figure compares the latest twelve months with the twelve before it, so one new month enters each window and one leaves. For work permits the swap is unusually large: May 2025 contributed 60,075 permits and May 2026 contributed 91,705, and that single substitution accounts for the entire 31,630 rise in the twelve-month total. For study permits the same swap runs 16,565 to 24,095.

Three of the four flows are still well below where they were a year earlier; what changed is the size of the shortfall, not its direction. Only work permits are above their year-earlier level. These are flows — documents becoming effective, or claims made in the month — and not counts of people resident in Canada.

Trigger recorded for this reading: The next monthly release of the same records.

Limit: Values are rounded to the nearest five at source. A rolling twelve-month comparison is sensitive to one month at each end, which is exactly what happened here, so a large move in the rate is not by itself evidence of a change in trend.

Reads onto the Big Question it bears on

Alongside the new month, Immigration, Refugees and Citizenship Canada restated 153 previously published values across its five monthly series, some reaching back to 2015. Those restatements barely moved the headline year-over-year figures: 0.11 points on study permits, 0.16 points on work permits, and 0.10 points on asylum claims -- each pulling the opposite way from the improvement reported above -- and 0.01 points on permanent-resident admissions, pulling the same way.

More on this question: Can Canada absorb its own population growth? →

Why it stands: This record keeps the figures it published last month, so it can recompute the same twelve-month window from the newly published data and compare. That isolates what the issuer changed about the past from what the new month added. Two Statistics Canada series were restated in the same update — job vacancies and residential investment, 9 values between them — and are not part of these four figures.

The improvement in these flows is not an artefact of the past being rewritten. For three of the four, the restatements pushed the opposite way; for the fourth they pushed the same way by a hundredth of a point against a one-and-a-half-point move. It is worth stating plainly because the opposite conclusion is the natural one to reach when a published rate moves and the source has also revised its history.

Trigger recorded for this reading: The next monthly release of the same records.

Limit: This measures the effect of restatement on four national twelve-month rates only. Most of the restated values fall in months outside the current comparison windows, where the effect on a rate computed from the latest twelve months is small by construction; the effect on any single month, or on a province, may not be.

Reads onto the Big Question it bears on

The housing supply readings are unchanged -- and that is a finding, not something skipped. This record refreshed all 148 of its source files for this update. Housing starts, units under construction, and the quarterly and annual completions series came back with every value identical to what this record already held.

More on this question: Can governments still solve housing shortages? →

Why it stands: Housing starts still end at the second quarter of 2026 and official completions still end at 2022, exactly as before. Nothing in the refresh moved them. Two other series on this domain did move and are not covered by this reading: the New Home Price Index gained a June 2026 month, and residential investment gained a May 2026 month and had three earlier months restated.

The share of the modelled building requirement that actual starts are running at is the same figure it was; so is the gap between how fresh the starts series is and how stale the completions series is. A reader who saw a new update date on this record should not read it as new evidence on housing supply.

Trigger recorded for this reading: The next quarterly housing starts release, and any release that ends the completions stall.

Limit: This covers the supply series only — starts, units under construction, and completions. The monthly completions series for larger centres dropped its oldest 2006 observation as its fetch window rolled forward, which changes no published figure and does not end its stall at 2022. This is not a statement about house prices, about residential investment, or about series this record does not track. The price index in particular did move: its long-run record no longer holds, and the level shift located in it is measured slightly differently than a month ago.

Reads onto the Big Question it bears on

Nine labour indicators shifted level together, inside one six-month window from December 2023 to June 2024. That published finding still stands, and the count is still nine -- but which nine has changed. With the new July data, total employment's own level shift now dates from August 2024, outside that window, so it drops out of the group. That indicator's shift is not published as its own separate reading, so where it was visible through the group, it is now not reported at all. Because total employment left the group, the measure of employment added per working-age person added now counts in its own right. It had been set aside before, because the series it is an arithmetic transform of was then in the group.

Why it stands: The group is assembled by taking every labour indicator whose level shift falls inside the same six-month window. One more month of data moved where the shift in total employment is located, from May 2024 to August 2024, which put it outside the window.

The conclusion is unchanged — a cluster of labour indicators turned together over Dec 2023 to Jun 2024 — but the reader should know the membership is data-dependent and moved this month, and that one indicator has dropped out of the reported picture rather than moving elsewhere in it. Two readings can report the same count for different reasons, which is what happened here.

Trigger recorded for this reading: The next monthly Labour Force Survey.

Limit: Locating a level shift is not a statement about cause, and a group of indicators turning together is not evidence that one moved another. Members may share surveys, definitions, and economic mechanisms.

Reads onto the Canadian Capacity Monitor

Second reading, on the June 2026 trade record. Over the twelve months through June 2026, the United States bought 69.07 per cent of Canada's goods exports, against 74.48 per cent a year earlier. That is a one-year fall of 5.41 percentage points -- shallower than the 5.78-point fall the May window showed, and well short of the record's fastest fall, 7.16 points, in the window ending March 2026. The share sits just above its record low: the March and April 2026 windows jointly hold that low, at 69.01 per cent, only 0.004 points apart. U.S.-bound exports have now risen for five straight months.

More on this question: Can Canada sell beyond the United States? →

Why it stands: Statistics Canada published the June 2026 international merchandise trade record on August 4, 2026, on the schedule its own May release stated. Both trade Resolutions were advanced to the 2026-08-04 capture under an explicit supersession chain: 47 publications re-pinned and 92 values re-derived from the newly archived bytes, with the May vintage retained unchanged on the record.

The break is neither deepening nor unwinding. The share has sat within a tenth of a point of its record low in three of the last four windows -- May 2026 is 0.10 points above it -- while the U.S.-bound level rises, so the one-year fall is now narrowing mostly because the base month it is measured against was itself depressed. The reallocation outside the bullion channel remains modest: about two-thirds of the rest-of-world rise is still the metals section.

Trigger recorded for this reading: The July 2026 trade record, at Statistics Canada's next scheduled monthly release.

Limit: All values are nominal, and partner shares are gross shipment values, not domestic value added. The record low is a near-tie inside the issuer's own revision movement, so which single month holds it can change with any revision. This reading is computed from the issuer's data tables; the issuer's own June narrative release was not captured this reading and is not quoted.

Reads onto the Big Question it bears on · the Brief it reads from