Overview

Abstract (English)

Governments transfer resources to low-income families both directly, with social assistance payments, and indirectly, through the tax system. In Canada, several important benefits aimed at low-income families are delivered through the tax system, but only to those who file tax returns. Furthermore, the tax system is frequently used to verify eligibility for a wide range of income-tested public goods and services. As we pointed out in a previous article (Robson & Schwartz, 2020) up to 12 percent of Canadians do not complete tax returns. Those who do not file are disproportionately those with low incomes and many miss out on cash benefits and public programs for which they are eligible.One way to get those benefits to their intended recipients would be for the Canada Revenue Agency (CRA) to absorb more of the administrative burden of initiating and completing tax returns on behalf of low-income individuals and families. To do so, however, given the current Canadian tax system, might require a wider range of third-party information or a set of administrative changes. In this paper, we estimate how many families have simple returns that the CRA could readily complete with information it already collects from third parties. Our calculations are based on the family-level variables constructed by Statistics Canada for the Longitudinal Administrative Databank (LAD), a 20% sample of all tax returns.We conclude that the CRA, if so directed, could complete returns for two-thirds of the families with income from provincial social assistance programs. And we estimate that collecting third-party information from new sources or instituting alternative administrative procedures could substantially increase the number of returns that the CRA could complete. We conclude there is a reasonable pathway for the tax agency in Canada to do much more to reduce the administrative burden imposed by the current tax filing system and a strong case for it to do so.

Abstract (French)

Please note that abstracts only appear in the language of the publication and might not have a translation.