ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
READ ME — Arts Operating Funding in Canada
2
A combined dataset of public operating grants and appropriations to arts organisations, from federal, provincial, territorial and municipal funders.
3
4
WHAT THIS IS
5
One combined record of public OPERATING funding to arts organisations in Canada, built from the funders' own published recipient lists, annual reports, and audited statements. Each funder's sheet shows its organisations down the rows and years across the columns, so every cell answers the same question: what did this funder pay this organisation in this year? It covers federal bodies (Canada Council, Canadian Heritage, Telefilm, and direct appropriations), every provincial and territorial arts council or ministry that publishes recipients, provincial gaming and lottery instruments, and the arts-funding programmes of some thirty municipalities. As of 5 October 2026: 53 funder sheets, 12,600 organisation rows, 7,565 distinct organisations once cross-linked (counted as cross-linked groups, not rows), $12,523,902,750 recorded, series running from 1991 (Winnipeg, calendar year) to 2026-27.
6
7
WHAT IT IS FOR
8
Questions no single funder can answer: how public money to the arts is distributed across organisations, ages, regions and levels of government; how organisations move between funders over time; and what has changed. It records operating support only — recurring, unrestricted, cash, not double-counted. Project grants, capital funding, COVID emergency programs and in-kind support are out of scope except where a sheet's own scope note says otherwise.
9
10
WHERE IT CAME FROM
11
This file is the public edition of a research workbook maintained since 2026. Everything in it comes from public sources, and every funder sheet's first row says exactly which source, and whether its totals were reconciled to a figure the funder itself published (the 'Source register' sheet collects this for the whole file). The dataset was compiled, cross-checked and documented with the assistance of Claude (Anthropic's AI assistant); every scope decision, identity ruling and founding-year adjudication was reviewed and decided by the author, who is responsible for the contents. The research workbook also holds the full audit trail — the working logs behind every cross-link and founding-year decision, and a Source register column naming the project documents behind each funder's row, which this edition does not carry — all of it available on request.
12
13
HOW TO FIND YOUR WAY
14
The 'Directory' sheet lists every sheet with the funder's full name, level of government, years covered, and organisation count. Data sheets are named for the funder ('CCA', 'OAC', 'Calgary', 'MCC' ...) plus 'Directly Funded (operating)' for institutions paid by appropriation rather than grant; the remaining sheets are the file's own rules and registers, described below.
15
16
HOW A FUNDER SHEET IS LAID OUT
17
Row 1 is the title and carries the sheet's most important caveats — read it first. Row 2 is the header. Rows 3 onward are organisations, one per row, up to a row labelled TOTAL. Below TOTAL sit the sheet's SCOPE note (what is in, what was left out, and why) and any MEMO rows — nothing below TOTAL is an organisation. Standard columns: Organisation, Founding Year (with a source column beside it), the year columns, Total ($), Years Funded, Notes, and Also funded by. Nineteen sheets also carry an 'Exclude from arts analysis' column, described under Conventions.
18
19
READING THE YEAR COLUMNS
20
A header like '2023 ($)' is that funder's year as the funder reports it — calendar year for most cities, a fiscal year for councils and ministries; the title says which. A blank cell means NO AMOUNT RECORDED, which is not the same as NOT FUNDED: where the distinction matters, a '[funder name] coverage codes' sheet says, for each organisation-year, whether the amount is known (A), the organisation was confirmed unfunded (Z), funded but unpublished (C), unknown (U), impossible because the organisation did not yet exist (X), or inferred (I) — six funders have one (CALQ, New Brunswick (THC), Nova Scotia (OACO), Ville de Québec, Winnipeg and Kitchener), where the published record left real ambiguity. Do not lean on the Total ($) and Years Funded columns: on some sheets the Total is a live =SUM formula across that row's year cells, on others a stored value that was correct when written. Either way, when you analyse, sum the year cells yourself. A negative year cell is a repayment the funder recorded; it nets against that organisation's grants in the same series.
21
22
FOUNDING YEARS
23
Every organisation carries a founding year where one could be established, in one of five forms explained on the 'Founding year forms' sheet: a bare number (known), 'by 1972' (a ceiling: the organisation existed by then), '1972 (uncertain)', 'c. 1972', or '1970s'. The rule is ACTIVITY BEATS LEGAL ENTITY: the year is when the organisation's continuous activity began, not when its current corporation was registered, and a merged organisation takes the older year; where only a registration date could be found, the row carries 'by' that year rather than asserting it as the founding. The evidence for each year is in the source column beside it. 12,416 of the 12,600 organisation rows are dated (10,082 bare years, 1,060 ceilings, 935 uncertain, 327 circa, 12 decades); 184 rows have no year and their notes say what was tried. Two limits to know. First, a small number of rows are still under review where the grantee is a society or foundation formed later than the institution it operates or funds — a museum and the society that now holds its money, a city- or university-held gallery and the body that applies on its behalf — and both dates are defensible; their notes say HELD, and the year shown is the one the evidence supports pending that ruling. Second, founding years are not uniformly sourced, and the workbook says so per row. The research workbook's Founding year evidence register holds 2,110 entries against the 12,416 dated rows in this file, and that is a boundary rather than a failure: the register is forward-only by design, recording years settled since 30 August 2026 rather than back-filling what came in with the original compilation. The sheet to read with most care is 'BC Gaming', whose founding years are the least sourced in the file; that sheet's row 1 describes the checks run on them and their results. Treat any unsourced year as unverified.
24
25
ONE ORGANISATION, MANY SHEETS
26
About four in ten organisations appear on more than one funder sheet — and together they hold two-thirds of the dollars in this file — under names that often differ in spelling, language, article or corporate suffix. The 'Also funded by' column on each row declares the other sheets that hold the same organisation, and the 'Cross-link aliases' sheet records the name pairs that had to be matched by hand (alias rows are written in both directions — a consumer counting undirected pairs should halve the row count; one declaration may point at more than one row on a target sheet). 'Ruled distinct' is the negative register: pairs of similar names that have been adjudicated as different organisations and must never be merged. Two registers new in this version refine identity further: 'Lineage register' lists predecessor/successor pairs, where one organisation continued another under a new name or form, so that a count by founding year can flag a merged lineage instead of counting it twice; 'Shared producer register' lists pairs of event-title rows (a city's festival grants, say) that name the same producing body. If you analyse this data, build organisation identity from these declarations, never from name-matching alone — for example, Canada has more than one Artspace.
27
28
TERMS USED IN THE NOTES
29
Row notes and register entries use four words of the workbook's own. A TRANCHE is one batch of rows written into a sheet as a single verified import, numbered in sequence across the project, so a row note can say which import put the row there and when. A PASS is one sweep over the workbook doing a single kind of work — a research pass, a cross-link pass, a correction pass; a pass may touch many sheets, where a tranche adds data to one. A BATTERY CONTROL is one of the automated checks the workbook is tested against before any version is released; each is numbered and tests one property — that no organisation's founding year contradicts itself across sheets (control 9), that the parsers exclude exactly the sheets the register says they exclude (control 22), that no row is dated after its own first grant (control 11). 'The battery' is all of them run together; a release requires zero problems, and where a control reports something deliberate rather than broken, the exception is declared on the sheet the control reads. THE STEWARD is the workbook's single writer: every correction enters through one hand with the battery run before and after, which is why a row note can say when a cell changed, what it held before, and on what evidence. Notes also cite the workbook's revision numbers (r1, r2 ... ) and dated packs of research; these are the audit trail's own references and need no further decoding.
30
WHO THE NOTES NAME. Row notes and register entries that say a year, a scope or a fold was "ruled" name the person who ruled it. Where they say MARK they mean Mark Hopkins, who compiled this workbook and holds the adjudication decisions; where they say THE STEWARD they mean the single writing process described above.
31
32
CONVENTIONS TO KNOW BEFORE ANALYSING
33
Exclude everything at or below each sheet's TOTAL row before summing. Match rows by organisation name, never by row position. Money stays on the row of the body that received it, even if the activity later moved. Where a funder pays on a cycle rather than annually (Alberta's casino proceeds every two to three years; Canadian Heritage's multi-year agreements booked in the signing year), a single year under-represents that funder — average across its cycle. Where a sheet's basis differs from full-year awards (Québec's MCC records paid-to-date snapshots; see 'MCC basis'), scale only at the aggregate, never per cell. If you recount this file's headline figures: a row counts as an organisation if it sits above a sheet's TOTAL row, and a column counts as money if its header carries both a year and a '$' — under that rule, two independent implementations reproduce the totals above to the cent.
34
THE 'EXCLUDE FROM ARTS ANALYSIS' COLUMN. Some funders' operating programmes pay bodies that are not arts organisations — agricultural fairs, sport and recreation clubs, genealogical societies, social-service agencies, individual artists. Those rows stay in the file because the funder published them in the programme, and 128 of them carry a reason in the 'Exclude from arts analysis' column ($38,990,007 across all years). The headline figures above count them, under the recount rule just stated. The author's own analysis, and the essay published with this dataset, leave them out: on that basis the file holds 12,472 rows and $12,484,912,743, and the 2022–2024 cross-section is 5,158 organisations receiving $956.9 million a year, of which 146 (2.8%) receive half. If you reproduce those figures, honour the flag; if you want the programme as the funder published it, ignore it. Either is a defensible count — say which you used.
35
36
KNOWN LIMITS
37
Coverage is what the funders published: some cities are not here, some series start late or stop early, and no sheet is complete beyond its own scope note. Read a sheet's row 1 and the 'Source register' before trusting its totals. Where an organisation's 'Also funded by' declaration names a sheet of the research workbook that is not part of this public edition, that token has been removed from this file so that no reference dangles; the research workbook holds the complete declaration set. Row notes and register entries also mention, by name, working sheets of the research workbook that this edition does not carry — 408 cells in this file do so, almost all of them naming a per-funder 'cross-link log', the working sheet on which a name match was argued out before its conclusion was recorded on 'Cross-link aliases' or 'Ruled distinct'; the rest name the research workbook's sheet register, its founding-year evidence register, a duplicate-review sheet, or two sheets held back because their contents fall outside this file's definition of operating funding ('MCC (out of scope)', which one Source register row describes, and 'CALQ official totals'). Those notes are left exactly as written rather than edited for this edition; the sheets they name are available on request.
38
39
WHAT CHANGED SINCE VERSION 1.0
40
Version 1.0 was cut on 15 September 2026 from research master a9d32d36 (53 funder sheets, 12,549 rows, $11,720,157,105). Between the two versions: five institutions paid by appropriation rather than grant were added to 'Directly Funded (operating)' — the Conservatoire de musique et d'art dramatique du Québec, the Art Gallery of Nova Scotia, the Yukon Arts Centre, TCU Place in Saskatoon, and Toronto's TO Live, the last carried as four rows so that its pre-merger constituents (the Hummingbird Centre, the St. Lawrence Centre and the North York Performing Arts Centre) keep their own years — and that sheet's series was widened to 1993–2026, which with back-filled years for existing rows adds $816 million to that sheet (about $31 million of it the Centre In The Square's money, moved rather than new); Kitchener's Centre In The Square moved from the city's grant sheet to 'Directly Funded (operating)' as a municipal agency subsidy, with a 'Kitchener coverage codes' sheet recording what the move left blank; the Alberta Foundation for the Arts' 2025-26 year was imported; sixteen pairs of rows that were one organisation carried twice were joined on 'Cross-link aliases' or ruled distinct; the 'Exclude from arts analysis' column grew from 107 to 128 flagged rows, chiefly Halifax's sport and community-event grants; the founding-year research campaign continued (ceilings written as 'by YYYY' rose from 329 rows to over a thousand as registration-only dates were re-expressed, the undated rows fell from 464 to under two hundred, and the evidence register grew from 997 to over two thousand entries); 'Lineage register' and 'Shared producer register' were added; and the frozen header rows of three register sheets that could not be scrolled past in Google Sheets were unwrapped. One correction to version 1.0's own Directory: 'Nova Scotia (OACO)' is a provincial programme and is now listed as such. Apart from one Halifax 2024-25 cell corrected by $2,500 against the municipality's own list, no money figure published in version 1.0 changed; the differences are additions, moves, folds and re-expressions, each recorded in the row's notes.
41
42
VERSION, CITATION, LICENCE, CORRECTIONS
43
This is version 1.1, generated 5 October 2026 from the research master (494c6463). Later versions will be published in the same place as this file and will carry a higher version number here; check this line before citing. Please cite it as: Hopkins, M. (2026). Arts Operating Funding in Canada, v1.1 [dataset]. LICENCE: this compilation is released under Creative Commons Attribution 4.0 International (CC BY 4.0) — use, share and build on this dataset freely, with attribution. The underlying figures are facts drawn from the funders' own publications and are not claimed; contains information licensed under the Open Government Licence – Canada and equivalent provincial and municipal open licences, and all figures originate with the funders named on each sheet. Corrections and questions are welcome at mark.c.hopkins@gmail.com — if you spot an error, say which sheet and row, and it will be fixed in the master and flow into the next public version.
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100