NFederal Vendor OwnershipOpen Nshipyard

Federal procurement, ownership-resolved

Who really owns the companies getting federal contracts?

Federal contract data lists vendor names as typed by more than a hundred reporting offices. The same company appears under many spellings, and nothing connects a vendor to its corporate record. This project resolves vendors to corporation numbers from Corporations Canada and measures what the raw files cannot: concentration of contract dollars, Canadian versus foreign ownership, and how much of it we can prove.

The spine in numbers

1,099,940

contract records in scope

$463.2B

total contract value

140,525

canonical vendors resolved

5,706

fuzzy merges logged

1,572,149

corporations in the merged registry

34.5%

of contract dollars matched to a corporation

The question nobody answers with data

Vendor normalization exists. Ownership does not.

A federal vendor directory already normalizes vendor names. That tells you who got paid. It does not tell you who owns the vendor, whether the money stayed in Canada, or how many contract dollars flow to one corporate family. The Buy Canadian Procurement Policy Framework took effect on December 16, 2025, and the Auditor General traced corporate networks by hand to audit one consulting case. This project builds the repeatable evidence base: vendor to corporation number, corporation to registered country, dollars to owners.

See the findings

From the showcase

How concentrated are federal contract dollars?

The top 10 canonical vendors absorb a share of contract value far beyond their count. When one company appears under many spellings, its real weight hides. Resolution brings it into the open.

How much federal contract money stays in Canada?

Of the contract dollars matched to a corporation record, the Canadian-registered share is shown below. The unmatched remainder is labelled unknown, not guessed. That boundary is the honest line in every chart on this page.

How messy were the vendor names?

Raw vendor strings collapse through normalization and fuzzy merging into canonical vendors. The funnel shows how much of the apparent vendor count was duplication.

How it was built

Five repairs before the first chart

Vendor names were normalized across 100+ reporting entities, fuzzy merges were resolved at a documented threshold, the CanadaBuys June 2023 schema break was crossed with a column crosswalk, Corporations Canada's four split files were merged into one table keyed on corporation number, and ambiguous rows were quarantined instead of guessed. Every merge is logged. Every rule is in the methodology.