The Canadian Tax Journal has published Tax Filing as State Administrative Infrastructure: Governance Gaps in Canada’s Tax-Based Benefit System, by Gillian Petit, Lindsay M. Tedds, and Jennifer Robson. The argument is that the personal income tax return has quietly become shared public infrastructure, relied on by every order of government to deliver social policy, while nobody has been made responsible for keeping it working.
Filing a tax return is no longer only a matter of tax compliance. It is the standardized administrative gateway to income-tested programs across the federation: the Canada child benefit, the GST/HST credit, and the Canada workers benefit at one end, and at the other, the Canada disability benefit, the guaranteed income supplement, the dental care plan, provincial housing and child care subsidies, pharmacare, and even a municipal transit pass in Calgary. More than 100 programs draw on CRA income verification alone. Where a person does not file, eligibility cannot be established, and benefits are delayed, made burdensome, or lost outright.
The article argues that this makes tax administration a form of infrastructure in a sense Canada does not currently recognize: not roads or buildings, but the durable legal authorities, data practices, and administrative processes through which governments implement policy at scale. Intangible administrative systems sit outside national accounting conventions for infrastructure, which is part of why the dependency has gone unmanaged.
The consequences fall where the stakes are highest. Up to 12 percent of working-age Canadians do not file, rising to as much as 20 percent among people in families below the market basket measure poverty line, and federal data put non-filing as high as 17 percent for those with incomes under $20,000. The Canada Learning Bond is the sharpest illustration: takeup reaches only 10 percent of children in households with incomes under $10,000, against 24 percent in the $40,000 to $49,999 range, and roughly 14,000 children a year miss out because a parent does not file. The CRA identifies potentially eligible children and passes the list to ESDC, but the two do not run joint outreach, and when payments stop, ESDC cannot tell whether income rose or a parent simply stopped filing. A similar pattern shows up in income assistance: 14.7 percent of social assistance recipients issued a T5007 slip, about 290,930 people, did not file in 2021.
Existing intergovernmental machinery was not built for this. Tax collection agreements harmonize tax bases and authorize CRA administration; the income verification program permits one-way, consent-based sharing from the CRA outward. Neither assigns anyone responsibility for filing supports, error resolution, or takeup, and the one-way structure means programs cannot signal back to the CRA that someone is very likely eligible but missing from the system.
The timing is deliberate. Automatic tax filing begins phasing in with the 2026 taxation year, and it will close part of the gap. But it is designed around people who already have a recent filing history and a simple tax situation, and it does nothing about the fragmentation of responsibility underneath. Automation improves access where the national tax agency administers the benefit directly. Where it does not, automation can make access more opaque rather than less.
What the article proposes is a tax-filing and benefit access accord: a light-touch intergovernmental framework layered onto the instruments that already exist, rather than replacing them. It does not centralize benefit delivery, expand entitlements, or alter constitutional responsibilities for program design. It establishes shared responsibility for the performance of the gateway itself, through four functions: shared responsibility for filing supports; standardized, purpose-limited, consent-based two-way data protocols; integrated service design working through trusted community intermediaries; and incremental implementation with public performance reporting, beginning with pilots where access gaps are already well documented.
Signatories would be governments rather than individual programs, with the federal side represented by the CRA and the Department of Finance, and provinces and territories by central agencies such as finance ministries or treasury boards. Because the framework touches sensitive personal and financial information, the article grounds it in purpose limitation, informed consent, transparency, auditability, and life-cycle management of data use.
This piece is one half of a pairing in the same Policy Forum. Its companion article, by Selvia Arshad, Gillian Petit, and Lindsay M. Tedds, supplies the welfare-based framework for evaluating tax administration reform. Together they set out how to govern the infrastructure and how to judge whether changes to it are worth making.
Three points carry beyond the specific proposal. First, access failures in Canadian social policy are often infrastructure failures rather than program design failures, which is why simplifying an individual program’s rules can leave takeup roughly where it was. Second, the tax system has repeatedly been adapted to new policy purposes, from the guaranteed income supplement through the pandemic benefits, while the governance framework around it has not been adapted in step. Third, filing supports remain thin relative to the load they carry: fewer than 70,000 Canadians use the invitation-based phone filing service in a given year, and the community volunteer income tax program completes 800,000 to 900,000 returns on volunteer labour that is difficult to scale further.