Methodology

The underlying data is the Monthly Business Openings and Closures (MBOC) program, published by Statistics Canada under Table 33-10-0270-01. The series is labelled experimental, which is StatCan's classification for high-frequency datasets that draw on novel methodological choices, but it has been published on a stable monthly cadence since January 2015 with consistent definitions.

The data is seasonally adjusted. The unit of measurement is the establishment (a physical location with active payroll), not the firm (which may comprise multiple establishments). The underlying universe is drawn from the federal Business Register cross-referenced against T4 payroll filings.

A business is counted as an opening in the reference month if it had no payroll in any of the prior twelve months and has payroll in the reference month. A closing is the inverse: payroll in any of the prior twelve months and no payroll in the reference month. The twelve-month look-back distinguishes genuine openings and closings from seasonal payroll gaps.

The MBOC program publishes eight distinct business-dynamics measures per (geography, sector, month) record: opening businesses, closing businesses, entrants (first-time openings with no prior payroll history at all), exits (permanent closures), continuing businesses (active in both reference and prior month), reopening businesses, temporary closures, and active businesses (the stock count). The Maple & Bay tracker captures all eight in the JSON dataset, surfacing the headline pair (openings, closings) and the net change in the dashboard while making the other six available through the data download.

StatCan publishes MBOC data at three geographic levels: national Canada total, the ten provinces plus three territories, and a set of census metropolitan areas (CMAs). The Maple & Bay tracker filters to the national plus the thirteen provinces and territories. The CMA-level data is in the source CSV and could be added in a future release; the current tracker holds it back to keep the dashboard footprint manageable.

The dashboard surfaces the eighteen top-level NAICS sectors plus the business-sector total. The source data ships every level of the NAICS hierarchy (down to specific 6-digit industries) and a set of tourism-specific composite breakdowns; the tracker filters to top-level codes to keep the per-page complexity reasonable for general readers. Researchers needing finer-grained sector data should pull directly from the StatCan source.

One quirk to note: the MBOC series groups Finance and insurance (NAICS 52) together with Management of companies and enterprises (NAICS 55) into a single composite line ("52, 55"). This is a StatCan publishing choice driven by sample size in NAICS 55. The Maple & Bay tracker preserves the composite as published.

MBOC is published with an approximately three-month lag. The data for reference month M is typically published in month M+3. This is the time StatCan needs to receive, process, and validate the underlying T4 payroll filings. The lag is structural and shorter than most labour-market or productivity series; the Labour Force Survey publishes monthly with a one-month lag, but the LFS measures household-side employment rather than establishment counts.

StatCan revises prior months as new T4 data arrives. Revisions are most material in the most recent three to six months of published history and smaller further back. The tracker uses whichever release of the data was most recent at the time of the last ingestion run. The scheduled monthly agent re-runs the full ingestion on the twenty-eighth of each month, which replaces both the new month and any revisions to prior months.

Statistics Canada applies confidentiality suppression at the small-cell level. When a (province, sector, month) cell contains fewer than a threshold number of businesses, the value is suppressed and ships as a null in the source CSV. The Maple & Bay tracker renders suppressed values as an em-rule character (—) in tables and as a gap in time-series charts.

Suppression is most visible in the small territories (Yukon, Northwest Territories, Nunavut) and in small sectors (utilities, mining, agriculture). National totals are not affected. Provincial totals for the larger provinces are not affected. The suppression is a feature of the source data, not an editorial choice by Maple & Bay; suppressed cells cannot be reconstructed.

The dashboard's editorial layer is limited and explicit. The tracker derives net change as a simple subtraction (openings minus closings) and renders negative net change in red rather than green. The 24-month rolling window on the time-series chart is an editorial pick on visualization horizon; the underlying dataset extends back to January 2015.

The CFIB Business Barometer overlay is a separate publicly available index sourced from the Canadian Federation of Independent Business. It updates monthly as a PDF release and is hand-curated into the tracker dataset on each release. The breakeven line at 50 on the CFIB chart is the value above which CFIB itself defines sentiment as net optimistic.

The editorial paragraph published with each monthly release reflects Maple & Bay's reading of the print. It is generated as a first draft by the scheduled remote agent that handles ingestion and reviewed by a human editor before merging. The paragraph identifies the largest-magnitude provincial and sectoral movers and notes trend continuity or breaks against the prior twelve months.

A scheduled remote agent re-runs the full ingestion on the twenty-eighth of each month at 12:00 UTC (6:00 AM Edmonton). The agent fetches the StatCan bulk CSV, regenerates the JSON dataset, drafts an editorial paragraph for the new month, and opens a pull request for human review before publishing.

If StatCan has not yet published a new release at the scheduled run time, the agent opens a heartbeat pull request with no dataset changes and a one-line note. The next month's run picks up the missed release.

The full dataset is downloadable at /tracker/business-formation/data.json. The file includes both the source metadata (period bounds, ingestion timestamp, attribution) and every record at full sector and province resolution.