Is Canada rebuilding the capital behind each worker?
Capital per worker and output per hour describe different parts of productive capacity. Reading the official records together can show whether investment and productivity are improving, while keeping their scopes, dates and limits visible.
On this page
Tools, technology and structures matter for productive capacity. The latest reading leaves annual capital-per-worker declines unchanged and advances the business-sector productivity comparison. Their different scopes and dates prevent a direct claim that one caused the other.
The current reading
Second dated reading, and it says the same thing as the first. The release this question was waiting on -- the April-June 2026 National Accounts -- arrived on 28 August 2026 and was read on 2 September 2026. Not one figure in the reading above moved. Every figure here is a calendar-year average or a year-end level, and the release added a single quarter inside a calendar year that is not yet complete. Countability re-fetched all four source tables and compared them observation by observation with the archived bytes these figures are pinned to: the annual capital-stock and multifactor-productivity records returned zero changes, and the two quarterly records changed only in April-June 2026, which is new, and January-March 2026, which the issuer restated. Neither quarter is inside any window this reading publishes. An unchanged reading after a real release is a result, not a gap.
The current reading: non-residential capital stock per worker fell 3.7 per cent from 2019 to 2024, and 5.7 per cent from the record's 2014 peak; real business investment per worker fell 7.2 per cent from 2019 to 2025. The investment share of GDP barely moved over the same span. Canada's workforce has grown faster than its roughly steady investment share of a growing economy -- so capital per worker has been falling even without a spending collapse, and the fall is concentrated in machinery and equipment specifically.
Basis: Level and share arithmetic over official published series, each pinned from archived source bytes and re-derived automatically; the ratios locate where the capital record stands, never why it moved.
- Non-residential capital stock per worker, computed by the Brief's disclosed formula from the geometric-net-stock and employment records, at its 2014 peak, at 2019, and at the latest published year.
- Real business investment per worker, computed the same way, from 2019 to the latest complete calendar year.
- The investment share of GDP at its 2014 peak against the latest complete calendar year, essentially unchanged.
Next: The July-September 2026 National Accounts release, expected in late November 2026 on the issuer's observed cadence (provisional -- no capture behind this reading pins the date); then the next annual capital-stock and multifactor-productivity releases. Trigger: Any release that materially moves the stock-per-worker or investment-per-worker levels, or a revision that changes a prior window's reading.
Limit: The capital-stock figure lags the investment-flow figure by roughly two years; per-worker ratios divide a National Accounts dollar aggregate by a Labour Force Survey employment count without an hours adjustment; no figure attributes the productivity slowdown to the capital record or any other specific cause. Expiry: Expires at the next quarterly National Accounts release; updated or explicitly re-affirmed then.
The Brief behind this reading → · See the Receipt — sources, method and limits →
Where institutions disagree
What cannot yet be known
How much of the current productivity slowdown does the capital record itself explain, on Statistics Canada's own official growth-accounting decomposition?
What is missing: Statistics Canada's multifactor-productivity programme (table 36-10-0208-01) is the only official source that decomposes measured labour-productivity growth into a capital-intensity contribution, a labour-composition contribution, and multifactor productivity, on one consistent accounting framework. It is published annually, roughly two and a half years behind the current quarter (2023 is the latest available year as of this Brief's evidence cutoff) — too stale to decompose the 2024-2026 stretch of the productivity slowdown this Brief's stock and flow readings cover.
Consequence: The Capital Behind Each Worker Brief can state whether measured capital per worker (stock) and real business investment per worker (flow) are rising or falling, on the issuer's own levels — and it does. It cannot state how much of the post-2019 productivity slowdown the capital record itself explains, because the one official measure built to answer that question does not yet cover the years in question. The Brief therefore reads the capital and productivity records side by side, on their own bases, and does not construct its own decomposition to bridge the gap.
The projection record
This record shows what an institution published and how its figures compare across versions or with later outcomes. It does not rank institutions or grade their forecasting ability. See the full public Resolution record →
The capital record behind each worker: non-residential capital stock and business investment, by asset class, 1961 (flow) / 1961 (stock) through 2026 Q1 / 2024
Canada's non-residential capital stock (geometric end-year net stock, total all industries, chained (2017) dollars, table 36-10-0097-01, annual through 2024) and business gross fixed capital formation by asset class (chained (2017) dollars and current prices, table 36-10-0108-01, quarterly through Q1 2026; one member cross-cited from table 36-10-0104-01), as Statistics Canada publishes them, beside Statistics Canada's own employment record (table 14-10-0287-01, seasonally adjusted, monthly) as the per-worker denominator -- the record behind whether Canada is rebuilding the productive capital behind each worker.
comparable · difference 118001.0 · same issuer. Numeric difference: the later year-end level minus the earlier, in millions of Canadian dollars; percent difference relative to the earlier level. Both are the same published stock series.
comparable · difference 247120.0 · same issuer. Numeric difference: the later year-end level minus the earlier, in millions of Canadian dollars; percent difference relative to the earlier level.
comparable · difference 9540.25 · same issuer. Numeric difference: the later calendar-year average minus the earlier, in millions of Canadian dollars; percent difference relative to the earlier level. Both windows are averages of the same published quarterly SAAR series.
comparable · difference 192895.0 · same issuer. Numeric difference: the later calendar-year average minus the earlier, in millions of Canadian dollars; percent difference relative to the earlier level.
- 'Total all industries' on the stock table (36-10-0097-01) is a total-economy aggregate -- it includes the government sector and non-profit institutions serving households alongside business industries. The flow record (36-10-0108-01) and the per-worker comparisons on this page use business gross fixed capital formation specifically. The two are never mixed inside a single ratio: the stock-per-worker reading and the flow-per-worker reading are stated as two separate scenarios, each internally consistent about its own scope.
- All levels on this page are real (chained (2017) dollars) except where explicitly marked 'current prices' for the investment-share-of-GDP reading, which is conventionally measured in nominal terms (both numerator and denominator inflate together). No figure here mixes the two bases inside one ratio.
- The annual capital-stock record (2024) lags the quarterly flow record (2026 Q1) by roughly two years -- a structural publishing gap in the official record, not a Countability delay; a published evidence gap names the same lag on the multifactor-productivity side.
What a first estimate of Canadian investment is worth: the capital expenditures survey's own release archive, 2015 through February 2026
Canada's first published number for a year's business investment appears in February of that year, before the year has happened. It comes from the Annual Capital and Repair Expenditures Survey, which asks organisations what they intend to spend; the figure is replaced a year later by a preliminary actual and a year after that by an actual. Table 34-10-0278-01 keeps every one of those publications as a separate Release member, so the intentions figure and the figure the same reference year carries today can be read side by side out of one table. This entry pins that path for national capital expenditures, all industries -- how far the first estimate stood from the settled one, in which direction, and how many publications it took to stop moving. It does not adjust any figure, does not derive a correction, and treats no revision as a statement about the issuer.
comparable · difference 21715.4 · same issuer. Numeric difference: reference year 2019 as Release member 13 gives it, minus the same reference year as Release member 5 first gave it, in millions of Canadian dollars. One series, one reference year, one issuer, one survey; what differs is the release the figure is read at, and with it the information the issuer had.
comparable · difference -26602.7 · same issuer. Numeric difference: reference year 2020 as Release member 13 gives it, minus the intentions figure Release member 6 published on 27 February 2020, in millions of Canadian dollars. Of the nine reference years whose whole path this archive holds, this is the one that moved furthest downward.
comparable · difference -32865.4 · same issuer. Numeric difference: reference year 2020 at the August 11, 2020 release minus the same reference year at the February 27, 2020 release, in millions of Canadian dollars. Both readings are of the same reference year in the same table, five and a half months apart. The step is larger than the reference year's whole settled movement, because the two later releases returned part of it.
comparable · difference 32426.9 · same issuer. Numeric difference: reference year 2023 as Release member 13 gives it, minus the intentions figure Release member 10 published on 28 February 2023, in millions of Canadian dollars. Of the nine reference years whose whole path this archive holds, this is the one that moved furthest upward in dollars.
comparable · difference 15481.3 · same issuer. Numeric difference: reference year 2024 as Release member 13 gives it, minus the intentions figure Release member 11 published on 27 February 2024, in millions of Canadian dollars. This reference year has just reached the publication at which every completed year in the archive stopped moving.
- This survey is not the national accounts. Table 34-10-0278-01 is the release archive of the Annual Capital and Repair Expenditures Survey, which measures capital expenditures on non-residential tangible assets. It is a different instrument from the quarterly national-accounts business investment series the Capital Behind Each Worker Brief reads elsewhere, and the two are never merged, differenced, or substituted here. What this archive supports is a statement about how much a first published estimate of Canadian investment moves -- not a restatement of any national-accounts figure.
- The national-accounts investment series has no release archive at all. Statistics Canada publishes real-time release tables for some programmes and not others; there is none for the quarterly business gross fixed capital formation series, so the revision behaviour recorded here cannot be checked against that series and is not claimed to describe it.
- There is no correction factor here, and none can be derived from this entry. Across the nine reference years whose complete path this archive holds -- 2015 through 2023, each first published as an in-year intentions figure -- five settled above their first estimate and four below, and the mean movement in absolute terms is 5.8 per cent. A first estimate cannot be adjusted toward its settled value by any fixed amount or direction the record supports.
Business-sector labour productivity through 2026 Q2, with the official capital-intensity decomposition beside it
The latest held quarterly publication reaches 2026 Q2 (September 2026 release). Earlier era comparisons retain their original endpoints and captured editions; the annual capital-intensity decomposition remains a separately dated measure.
comparable · difference 16.906 · same issuer. Numeric difference: the later index level minus the earlier, in index points; the percentage difference is the total percent change over the full ten-year window (not annualized). The average annual pace over the window, computed as average annual log growth (the standard national-accounts convention for a multi-period index), is +2.15 percent per year -- stated in this entry's note, not as a separate schema field.
comparable · difference 6.758 · same issuer. Numeric difference in index points; the percentage difference is the total percent change over the full ten-year window (not annualized). Average annual log growth: +0.75 percent per year, stated in this entry's note.
comparable · difference 8.832 · same issuer. Numeric difference in index points; the percentage difference is the total percent change over the full 8.5-year window (not annualized). Average annual log growth: +1.06 percent per year, stated in this entry's note.
comparable · difference 1.217 · same issuer. Numeric difference in index points; the percentage difference is the total percent change over the full 6.25-year window (not annualized). Average annual log growth: +0.19 percent per year, stated in this entry's note -- roughly one-eleventh the 1990s pace.
comparable · difference 1.117 · same issuer. Numeric difference in index points; the percentage difference is the total percent change over the four-year window. This index measures a different quantity, on a different basis, than the stock/flow per-worker levels on the capital-record entry -- the two are never combined into one figure; this comparison states only what Statistics Canada's own official decomposition shows, on its own terms, through its own latest available year.
comparable · difference 2.418 · same issuer. Later index minus earlier index; percentage difference is the total change over 6.5 years. Average annual log growth is 100 × ln(later/earlier) ÷ 6.5. The earlier boundary is unchanged in the new source vintage.
comparable · difference 1.028 · same issuer. Numeric difference is the later index minus the earlier; percentage difference is 100 × (later / earlier − 1), not annualized.
- The four productivity eras are bounded at the record's own visible turning points (a decade after the record begins; two further decade marks; the pre-pandemic peak quarter), not at arbitrary round years -- 2019 Q4 is used, not 2020, because it is the last quarter before pandemic-era distortion enters the seasonally-adjusted series.
- Average annual log growth (the standard convention for comparing multi-period index growth across windows of different lengths) is stated in each comparison's note; the recorded difference and percentage-difference figures carry the total, non-annualized change over each full window.
- The capital-intensity-contribution index (36-10-0208-01) is Statistics Canada's own official decomposition of measured labour-productivity growth into its sources; it is published annually and currently lags the current quarter by roughly two and a half years -- named in a published evidence gap -- the boundary this Resolution's context reading hits. No figure on this page infers what the 2024-2026 contribution would show.
What would change the record
- Statistics Canada publishing the multifactor-productivity decomposition (table 36-10-0208-01 or a successor) for reference years 2024 and 2025, closing the lag to within a year or two of the current quarter.
This page brings together the evidence, institutional disagreements and gaps behind a continuing question. Where the record supports an answer, it appears with its date, sources, limits, Brief and Receipt. New evidence can change that answer.
Follow these updates · RSS feed
This page holds the current public record for this question. For a more specific question, see Work with Countability.
Work with Countability