Supplementary Submission on the Underground Economy

1. Two different measures: the underground economy and the tax gap

At its first meeting, the committee heard testimony from Statistics Canada on its estimate of the underground economy, but many of the senators’ questions concerned tax compliance: who is not paying, why, and what would change that. This section unpacks the difference. It sets out what Statistics Canada’s estimate measures and does not, what the CRA’s tax gap estimates measure, and how the two relate.

1.1 Statistics Canada’s underground economy estimate: what it is

Gross domestic product (GDP) is the value of all final goods and services produced in Canada in a year. It can be measured three ways that should give the same answer: by adding up the value added by each industry, by adding up spending on final goods and services by households, businesses, governments and buyers abroad, or by adding up the incomes earned in production, such as wages and profits. GDP counts production, not every transaction. The resale of a used car adds only the dealer’s margin, and transfers such as gifts or benefit payments add nothing. This is why underground activity can be partly in GDP already: a household’s spending on a cash renovation may be captured in spending data even when the contractor’s income is missing from income data.

Statistics Canada’s estimate asks how much of that production is missing from GDP because it is concealed or informal. Its purpose is to make sure GDP is complete, not to measure tax compliance. It follows the OECD’s handbook on measuring the non-observed economy (OECD, 2002) and is built within the national accounts (Luo et al., 2024). Statistics Canada estimates underground economic activity at $72.4 billion in 2023, or 2.5 per cent of GDP (Statistics Canada, 2025). It covers legal production that is concealed, such as unrecorded sales, undeclared tips, unreported rent and cash construction work; legal production by unregistered operators, such as in-home child care, household services, personal care and roadside farm sales; and three illegal markets, unlicensed alcohol, tobacco and cannabis (Luo et al., 2024). In short, it is largely undeclared activity in legal markets, measured as missing production rather than missing income (this responds to a question from Senator C. Deacon). The OECD’s handbook, together with the European Union’s requirement that member states make GDP exhaustive, sets the international standard for this kind of measurement (this responds to a question from Senator Fridhandler).

1.2 What its design means for the committee’s questions

Several features of the method determine what the estimate can and cannot tell the committee (Luo et al., 2024).

• It is a ceiling, not a best guess. Each component is estimated as the largest amount of underground activity that is plausible given other data. The $72.4 billion is an upper bound on missing production, not a measure of what the government loses, and no margin of error is published (this responds to a question from Senator C. Deacon about how much the estimate may understate the underground economy).
• It assumes underground activity happens only in small businesses, those with revenue under $2 million in 1992 dollars. Large firms are outside the estimate by design, so it cannot address concerns about large players; the closest evidence is the CRA’s estimate of the large corporation gap, discussed below (this responds to a question from Senator Pate).
• It counts underreporting only on sales to households. Underreporting between businesses does not change GDP, because it is already embedded in the price the final customer pays. It does change tax: when one business deducts a payment that the receiving business never reports, tax is lost with no effect on GDP. Business-to-business platforms and unreported contractor fees are invisible to this measure for that reason (this responds to a question from Senator Fridhandler).
• Rates of underreporting are assumptions, not measurements. Each category of household spending is assigned a rate, for example zero for regulated purchases such as electricity and 17 per cent for taxis. A cash discount offered at a nail salon enters the estimate through an assumed rate for personal care, not through any observation of how often it happens. Likewise, the 11 per cent fall in underground residential construction in 2023 reflects less renovation spending; the method cannot detect a change in compliance (this responds to questions from Senator Yussuff).
• CRA audit data are used, but only to distribute the total across industries, not to measure its size.

1.3 What it is not

• It is not a measure of tax lost. Much of the spending in question is already in GDP; what is missing is the matching income. As Statistics Canada’s witness told Senator Pupatello, “we are talking about GDP. We are not necessarily talking about the tax base” (Standing Senate Committee on Banking, Commerce and the Economy, 2026). The $3 billion to $5 billion adjustment for tips that Statistics Canada described is an adjustment to GDP, not an estimate of tax revenue forgone (this responds to a question from Senator Pupatello; see also section 4.5).
• It leaves out activity that matters for tax but not for GDP. A business reselling used goods through sites such as Kijiji and eBay has taxable income even though the goods add almost nothing to GDP; since January 2024, platforms must report such sellers to the CRA (this responds to a question from Senator Fridhandler). Drug trafficking, human trafficking, money laundering, barter and cross-border shopping are excluded (Luo et al., 2024), and profit shifting by multinationals is outside it altogether.
• It does not identify who underreports or why, because it is built from aggregates and assumed rates rather than information on taxpayers.
• It cannot evaluate policy. Quebec’s mandatory billing system increased restaurants’ reported sales by 5.8 to 9.8 per cent (Boyer & d’Astous, 2023), yet Quebec continues to rank near the top of the provincial estimates, because the assumed rates do not change when behaviour does.

1.4 How to read the numbers

• Stability is a feature of the method. The estimate has stayed between about 2 and 3 per cent of GDP in every year since 1992 (Statistics Canada, 2016, 2025), moving mainly with the construction cycle and the size of the economy. Statistics Canada told the committee that it hopes to update the estimate in 2027 with funding from the CRA, which funds the account (Standing Senate Committee on Banking, Commerce and the Economy, 2026). For the same reasons, a new release is unlikely to change the picture, and it will not show who is not complying, why, or what would work (this responds to a question from Senator Yussuff).
• Provincial rankings reflect industry mix, not honesty. Provinces with large construction, farm and hospitality sectors rank high; Alberta and Newfoundland and Labrador rank low partly because oil and gas production enlarges their GDP. No margins of error are published, so differences of a few tenths of a percentage point should not be read as meaningful (this responds to a question from Senator Fridhandler). Misclassified workers illustrate the point. The B.C. building trades estimated in 2019 that about 14,000 construction workers were improperly classified as independent contractors. Misclassified workers who still report their income do not show up in the underground economy (this responds to a question from Senator Yussuff). Misclassification does not change GDP, because the same houses are built either way; it only moves the workers’ pay from employee compensation to self-employment income in the accounts. A province with widespread misclassification would therefore rank no differently. For tax and for workers it matters a great deal: income moves from reporting by the employer to self-reporting, where underreporting is far more common, the employer stops paying its share of CPP contributions and EI premiums, and the worker loses EI coverage and employment protections (see also section 4.3).
• Industry detail is limited. As Statistics Canada’s witnesses noted, some industries are combined into broad groups, such as finance, insurance, real estate and leasing, so the estimate cannot answer sector-specific questions such as those about fisheries or agriculture (this responds to questions from Senators Fridhandler and C. Deacon). Based on the published description of Statistics Canada’s methods, some of what senators raised about fisheries likely would not appear in the estimate even with more detail. The catch is production and is counted in GDP whoever holds the licence. Payments to lease or control a fishing licence or quota are generally treated in the national accounts as rent for the use of a natural resource, and the sale of a licence is the transfer of an asset; neither is production. For tax, both matter. Lease payments are taxable income to the licence holder, the sale of a licence can give rise to a taxable gain, and side arrangements paid in cash, which have grown up around restrictions on who may hold licences and quotas, are precisely the kind of income no third party reports.

1.5 Why it is not a measure of tax compliance

Tax compliance asks a different question: of the tax legally owed, how much is paid, by whom, and what changes that. A simple example shows the difference. A restaurant that keeps $100 of cash sales off its books reduces measured income by $100, and Statistics Canada’s estimate tries to add that production back. For tax, the same act means unpaid GST/HST on the sale, unpaid income tax on the profit, and possibly cash wages with no EI or CPP contributions. A business that claims a $100 expense it never incurred changes nothing in GDP, yet costs the same tax. Statistics Canada’s own witnesses told the committee that it is “more of an agency like the CRA that looks at tax compliance” (Standing Senate Committee on Banking, Commerce and the Economy, 2026). The committee’s questions about cash payments, tips, building permits, misclassified workers and platforms are questions about tax compliance. They require different evidence, drawn from tax administration, tax law and the economics of compliance rather than from national accounting.

1.6 The CRA’s tax gap estimates

The tax gap is the measure that does address compliance. It is the difference between the tax that would be collected if every taxpayer met their obligations in full and the tax actually paid. The CRA’s second overall report, published in March 2026, estimates a gross federal tax gap of $59.5 billion for tax year 2022, or 16.0 per cent of federal tax revenue, and a net gap of $34.7 billion, or 9.3 per cent, after audits, reassessments and collections (CRA, 2026). Within that total, the report estimates:

• An underground economy component of the personal income tax gap of $9.8 billion, or 4.7 per cent of personal income tax revenue. About 80 per cent of it relates to compensation paid to employees and about 20 per cent to self-employment income.
• A corporate reporting gap of $22.6 billion, of which $17.2 billion is attributed to large corporations and $5.4 billion to small and medium-sized enterprises.

The report is a significant step forward in transparency. Its components, however, are built in different ways (CRA, 2026):

• The underground economy component is derived from Statistics Canada’s estimates, not observed in tax data.
• The small and medium-sized enterprise estimate rests on the CRA’s last random audit of these businesses, for tax year 2011, in which 37.6 per cent of audited businesses were assessed additional federal tax. Later years are projected forward using growth in reported tax and growth in Statistics Canada’s underground economy series.
• The large corporation estimate comes from risk-based audits, which target files the CRA considers most likely to be non-compliant, adjusted statistically to represent the population.

1.7 How the two measures relate

The two estimates are not independent confirmation of each other. Where the CRA’s figures concern the underground economy and small business, they rest largely on Statistics Canada’s series and an audit now fifteen years old. The decline in the CRA’s recent small business estimates reflects slower growth in Statistics Canada’s series relative to reported income rather than new audit evidence. Neither measure can show which taxpayers are not complying, why, or whether a reform works. The table summarizes the differences.

The CRA’s headline gross gap of 16.0 per cent of federal tax revenue is similar to the United States’ projected gross gap for tax year 2022 of $696 billion, about 15 per cent of tax owed, with a voluntary compliance rate of 85 per cent (Internal Revenue Service, 2024). The levels should not be compared directly, because the two estimates cover different taxes (Canada’s includes the GST/HST, and the United States has no federal sales tax) and are built differently. The more telling comparison is the effort behind them. The U.S. estimates rest on recurring random audits through the National Research Program and are broken down by type of tax and by non-filing, underreporting and underpayment (Internal Revenue Service, 2022, 2024), and the United Kingdom publishes estimates every year from its random enquiry programs (HM Revenue & Customs, 2026). Canada’s estimate for small and medium-sized businesses rests on a single random audit from tax year 2011.

2. Macro estimates of the underground economy and micro estimates of tax compliance

The research on this subject has moved in the past thirty years from macro to micro methods, for the reason the committee encountered in the first panel.

Macro estimates infer the size of the underground economy from aggregate data. Besides the national accounts approach used by Statistics Canada, they include models based on the demand for currency and statistical models that infer a hidden economy from its presumed causes and indicators. I began my career estimating the Canadian underground economy with methods of this kind (Giles & Tedds, 2002).

These research estimates are not estimates of what is missing from GDP. They generally aim at a broader concept: income or activity hidden from government, and especially from the tax authority, whether or not it is already counted in GDP. Definitions also vary from study to study and across countries. In 2023 the International Labour Organization adopted statistical standards that define informality from the worker’s side, by whether work carries legal and social protection such as pension contributions and employment insurance (International Labour Organization, 2023), a definition that separates someone doing cash cleaning to pay the rent from a business concealing sales (this responds to a question from Senator Pate). The results are usually expressed as a percentage of GDP to give a sense of scale, which invites two misunderstandings: that GDP is understated by that amount, and that the figures can be compared directly with Statistics Canada’s estimate. My own macro estimates of a broadly defined Canadian underground economy, for example, rose from 7.5 per cent of GDP in 1976 to about 15.3 per cent in 2001 (Tedds, 2005), several times Statistics Canada’s figure, because they measured something different. Much of the debate over whether Canada’s underground economy is 2.5 per cent of GDP or several times that comes from comparing estimates of different things. Macro research estimates also vary widely with their assumptions, which is one reason the OECD and Statistics Canada do not rely on them for the national accounts (Luo et al., 2024). Like Statistics Canada’s estimate, none of them can say who is not reporting, why, or whether a policy works.

Micro estimates measure compliance directly among taxpayers and transactions. The main approaches are:

• Random audits. Tax authorities audit a random sample of returns line by line and project the results to the population. The United States’ National Research Program shows a net misreporting rate of about 1 per cent on wages subject to withholding and about 55 per cent on income with little or no third-party reporting (Internal Revenue Service, 2022). The United Kingdom publishes annual estimates from random enquiry programs (HM Revenue & Customs, 2026). Canada’s most recent random audit of small and medium-sized businesses covered tax year 2011 (CRA, 2026).
• Randomized experiments. In Denmark, randomized audits and audit-threat letters showed that evasion is close to zero on third-party-reported income and substantial on self-reported income (Kleven et al., 2011).
• Spending and income comparisons. Households that report similar incomes but spend differently reveal underreporting. This approach was developed in the United Kingdom and applied to Canada’s self-employed (Pissarides & Weber, 1989; Schuetze, 2002). My own work extended it to allow underreporting to vary with income and found that the gap between true and reported self-employment income is largest for households at the lower end of the self-employment income distribution (Tedds, 2010).
• Natural experiments in administration. Changes in reporting or remittance rules show what compliance was before and after. Quebec’s restaurant billing system (Boyer & d’Astous, 2023) and platform remittance of lodging taxes in the United States (Bibler et al., forthcoming) are examples, as is Norway’s decision in 2001 to make tax records easily searchable online, after which reported income rose, particularly among the self-employed (Bø et al., 2015). Using the introduction of the GST in 1991 as a natural experiment, I found no statistically significant change in income underreporting by self-employed households after the tax was introduced (Tedds, 2005, 2010), in contrast to earlier macro estimates that pointed to a post-GST increase in the underground economy (Spiro, 1993).
• Bunching at thresholds. Unusual numbers of businesses reporting sales just below a threshold, such as the $30,000 GST/HST small supplier threshold, reveal a response to it.

All of these require access to administrative tax and audit data. Access to these data is uneven. Through its Research Data Centres, Statistics Canada makes available the Longitudinal Administrative Databank, a 20 per cent sample of individual tax filers, which supports many kinds of research. Compliance research usually needs more: full population files such as the T1 Family File, and business tax data from corporate (T2) returns and the GST/HST, which are considerably harder to obtain. The CRA’s audit and compliance data are not available to researchers at all. That is why my brief recommends restoring a published random audit program and secure researcher access to the CRA’s compliance data. Without them, Canada can estimate roughly how large the underground economy is but cannot answer the questions this committee is asking.

3. Information that exists but does not connect

Questions from several senators turned out to be about the same problem. The information needed to see underground activity often already exists, but it sits in separate systems that do not connect, held by different organizations, levels of government and private intermediaries, each using it for its own purpose. This is the point I made in my opening remarks about treating tax administration as state infrastructure: the system that collects taxes and delivers benefits runs on the same information and should be designed as one system, with governance to match (Petit et al., 2026). Examples from the senators’ questions include:

• Building permits. Permits record renovation work, the largest single source of underground activity, and reach the property assessor: in Ontario, the Municipal Property Assessment Corporation processed 176,000 permits in 2025 (Municipal Property Assessment Corporation, 2025). They do not reach the tax authority (this responds to a question from Senator Varone).
• Fishing licences. Fisheries and Oceans Canada holds licence records, buyers report fishers’ earnings on T4F slips, and processors keep purchase records, but licence lease payments and payments in kind are reported nowhere (this responds to a question from Senator C. Deacon).
• Workers’ status. Whether a worker is an employee is decided by the CRA only when someone requests a ruling, payments to contractors that would reveal patterns have gone largely unreported since 2011, and provincial employment standards and workers’ compensation systems apply their own tests to their own data (this responds to a question from Senator Yussuff; see also section 4.3).
• Financial records. Card payments and e-transfers leave records with banks and payment processors that never reach the tax authority; unlike the United States, where payment processors report merchants’ receipts to the IRS on Form 1099-K, Canada has no general requirement of this kind. At the same time, these institutions already report large cash transactions and suspicious transactions to the Financial Transactions and Reports Analysis Centre of Canada (FINTRAC) for anti-money-laundering purposes. The two systems connect poorly: the Auditor General found in 2004 that FINTRAC and the CRA had not agreed on criteria for identifying money laundering related to tax evasion (Office of the Auditor General of Canada, 2004), and its audit tabled on October 29, 2026, found that FINTRAC takes an average of about four months to pass intelligence to law enforcement and that half the time the information arrives too late to be useful (Duggan, 2026). The Office of the Auditor General estimates that between $45 billion and $113 billion is laundered through Canada each year (Office of the Auditor General of Canada, n.d.) (this responds to questions from Senators Pupatello, Yussuff and Pate).
• Platforms. The CRA has sought information from one platform at a time rather than requiring all platforms to report on the same terms (this responds to questions from Senators Fridhandler and C. Deacon).
• Federal and provincial governments. The CRA collects personal income tax for every province except Quebec, and provinces lose their own income and sales tax and workers’ compensation premiums on the same unreported income, but there is no shared approach to compliance outside Quebec (this responds to questions from Senator Yussuff; see also section 4.1).
• Payroll and benefits. Wages are reported annually through payroll slips, while eligibility for income-tested benefits is assessed in separate files, which is why real-time payroll reporting through ePayroll matters for both compliance and benefit delivery.

Where jurisdictions have connected this information, they have used it to deliver benefits as well as to secure compliance, which is what makes it a carrot as well as a stick. In the United Kingdom, payroll information that employers send to the tax authority is passed automatically to the Department for Work and Pensions, which uses it to calculate Universal Credit each month without the worker having to report earnings, and employers report once for both purposes (GOV.UK, n.d.). In Australia, payroll data reported to the tax office through Single Touch Payroll is used by Services Australia to pre-fill the employment income that benefit recipients must report (Services Australia, n.d.). Canada already does this in one place: since 2009, the CRA’s Automated Benefits Application has allowed parents to apply for the Canada Child Benefit and related benefits when they register a birth with their province, using information the province already collects (CRA, 2015). These systems work only if the underlying data are accurate and there are clear ways to correct errors, which is why the governance of shared information matters as much as the connection itself (Petit et al., 2026).

In each case the first step is less about collecting new information than about connecting what already exists, under clear rules about who may use it and for what.

4. Responses to senators’ questions

This section responds to every question senators asked at the meeting, whether it was put to Statistics Canada or to me, in the order in which senators first spoke. Questions answered in sections 1 and 3 are marked there and are not repeated here.

4.1 The Chair, Senator Gignac

Countries with small underground economies, and whether Quebec has worked on the issue (asked of Statistics Canada)

Statistics Canada rightly cautioned that differences in concepts and methods make international rankings unreliable (Standing Senate Committee on Banking, Commerce and the Economy, 2026). What high-compliance countries share is visibility: extensive third-party reporting, pre-filled returns and little use of cash. Within Canada, Quebec has gone furthest.

Quebec has worked on the underground economy more than any other jurisdiction in Canada, across the full range of tools discussed at the meeting. Because several senators’ questions touched on Quebec, its measures are gathered here.

• Certified sales recording. Restaurants have been required since 2011 to record every sale through a certified system that reports to Revenu Québec, a requirement since extended to passenger transportation. Reported restaurant sales rose by 5.8 to 9.8 per cent (Boyer & d’Astous, 2023).
• Tips through payroll. Employees in regulated establishments such as restaurants, bars and hotels must declare all their tips to their employer, who must accept the declaration and run the full amount through payroll. Where declared tips fall below 8 per cent of an employee’s tippable sales, the employer must allocate the difference, and a lower rate can be requested where tips are typically lower (Revenu Québec, n.d.-a). Declared tips count toward Employment Insurance and Quebec Parental Insurance Plan benefits, vacation pay and statutory holiday pay, and under the Act respecting labour standards tips are the exclusive property of the employees who receive them and employers cannot require employees to pay credit card costs (Revenu Québec, 2016). The same design gives the tax authority visibility, extends social insurance to tip income and protects workers against employers keeping tips.
• Compliance as a condition of contracting. Contractors on public construction contracts must hold an attestation from Revenu Québec that their tax affairs are in order, and persons involved in transactions penalized under Quebec’s general anti-avoidance rule can be listed in the registry of enterprises ineligible for public contracts (RSM Canada, 2019).
• Aggressive tax planning. Since 2009 Quebec has required mandatory disclosure of certain aggressive tax planning transactions, with a preventive disclosure option for taxpayers who come forward (Revenu Québec, n.d.-b), and since 2019 it has required disclosure of nominee agreements (RSM Canada, 2019). The federal general anti-avoidance rule was strengthened only in 2024, when amendments lowered the threshold for an avoidance transaction, made a significant lack of economic substance an explicit consideration in deciding whether a transaction is abusive and introduced a penalty of 25 per cent of the tax benefit (PwC Canada, 2023).
• Beneficial ownership. Since March 31, 2023, the identity of companies’ ultimate beneficiaries has been publicly accessible through the Quebec enterprise register, the first province to do so (Borden Ladner Gervais, 2023). The federal registry of individuals with significant control covers only federally incorporated companies, so Quebec’s registry goes further in addressing concerns about numbered companies (this responds to a question from Senator Pate).
• Renovation tax credits. The ÉcoRénov, LogiRénov and RénoVert credits required work by a qualified contractor, and renovation contracts must show the contractor’s Régie du bâtiment licence and GST and QST numbers (Régie du bâtiment du Québec, n.d.).
• Joint enforcement. The ACCES construction committee brings together the Ministry of Labour, the Régie du bâtiment, Revenu Québec, the Commission de la construction du Québec and the workplace standards and safety commission to share information and conduct joint site visits (Commission de la construction du Québec, 2023), against construction-sector tax losses that Quebec’s Ministry of Finance has estimated at $1.5 billion a year (Commission de la construction du Québec, 2019). A separate committee, ACCEF, has coordinated action against economic and financial crime with tax implications since 2010 (Bolduc, 2011).

Quebec also evaluates: the evidence on its billing system comes from anonymized tax data. Because Revenu Québec administers the GST in Quebec, the federal government already shares in the results.

There is no comparable federal-provincial mechanism. Because the CRA collects personal income tax for every province except Quebec, and provinces lose their own income and sales tax and workers’ compensation premiums on the same unreported income, collaboration should centre on the tax administrations rather than the statistical agencies, beginning with a federal-provincial working group modelled on ACCES construction (suggestion 4 in the summary). This also responds to Senator Yussuff’s questions about collaboration with the provinces and which province has been most active.

4.2 Senator Varone

Workers who participate in both the formal and underground economy, and whether tax policy or information should come first (asked of Statistics Canada and of me)

No one tracks this group directly. The information that could identify them exists in pieces: tax returns show who has both employment and self-employment income, union benefit and pension plans record hours worked, and building permits record renovation work. None of it is linked for this purpose.

On whether tax policy or information should come first, the evidence points to information. The worker the Senator described is not typically affected by benefit clawbacks. The gain from cash work comes from avoiding income tax, both the employee and employer shares of CPP contributions that apply to self-employment income, and the GST/HST the customer would otherwise pay. That gain is shared with the homeowner through a lower cash price. Lower rates would narrow the gap but not close it, and where income is reported by a third party, evasion is close to zero even at high tax rates (Kleven et al., 2011). Because the homeowner shares the gain, the most direct levers act on the customer’s side of the transaction: receipt-based renovation credits (see the response to Senator Fridhandler) and connecting building permit information with tax administration (see section 3).

4.3 Senator Yussuff

Misclassification of construction workers (asked of Statistics Canada)

Whether a worker is an employee or an independent contractor is a legal question with a well-established test in tax law, concerning control, ownership of tools, the chance of profit and risk of loss, and integration into the payer’s business (Wiebe Door Services Ltd. v. M.N.R., 1986; 671122 Ontario Ltd. v. Sagaz Industries Canada Inc., 2001). Sections 1.4 and 3 explain why misclassification does not affect GDP and how the information that would reveal it is scattered. Disputes over status are usually resolved only when a worker requests a CPP or EI ruling, often after losing a job. Enforcing the existing requirement to report fees for services would allow the CRA to identify repeated patterns before a worker has to come forward. Provincial employment standards and workers’ compensation tests do not always match the federal test, which adds to the scope for misclassification.

4.4 Senator Fridhandler

How barter is tracked (asked of Statistics Canada)

In the Senator’s example, a lawyer preparing wills in exchange for a finished basement, both parties have taxable income equal to the fair value of what they received, because the exchange is part of their businesses. Barter between individuals in a business context is income in law, but when neither side reports it, there is no record for anyone to see, and Statistics Canada does not include barter in its estimate (Luo et al., 2024). Where barter runs through organized exchanges, there is an intermediary with records. The United States requires barter exchanges to report their members’ transactions to the IRS; as far as I am aware, Canada has no equivalent requirement. The line that matters for tax is between barter in the course of a business, which is income at the fair value of what is received, and exchange outside any business, such as neighbours helping with a harvest or a barn raising, which is not (CRA, 1982). Much of the exchange within close-knit communities is of the second kind, while their business sales count as income whether paid in cash or in kind (this also responds to Senator C. Deacon’s question about Amish communities).

The payer’s side, and receipt lotteries (asked of me)

When the payer is a business, the payer is the natural point of reporting. Employers report wages on T4 slips, construction businesses report payments to subcontractors on T5018 slips, and businesses are required to report fees paid for services. Enforcing that last requirement, and denying a deduction for payments that have not been reported, would give businesses their own reason to comply.

I undertook to identify the countries that use receipt lotteries. They include Malta (since 1997), Slovakia (2013) and Portugal (2014), with Taiwan operating a receipt lottery since the 1950s (Fooken et al., 2015). Greece introduced one in 2017 (Avalara, 2017), and Poland, Lithuania, Latvia and Italy have followed (Avalara, 2019). In Brazil, São Paulo’s program gave consumers rebates and lottery entries for receipts carrying their tax number, which increased the revenue firms reported (Naritomi, 2019). Lotteries depend on receipts that the tax authority can verify; in Canada that would first require certified point-of-sale systems of the kind Quebec uses in restaurants.

Canada already has a form of receipt incentive that suits its current system: tax credits that require a receipt from the supplier. The 2009 Home Renovation Tax Credit required receipts for eligible work but was designed as temporary stimulus (Department of Finance Canada, 2009), and I am not aware of a rigorous evaluation of its effect on cash work. Quebec’s renovation credits went further by requiring that the work be done by a qualified contractor (see the response to the Chair). A time-limited, receipt-based credit in residential renovation, the sector that accounts for about a third of underground activity in Statistics Canada’s estimates, with an evaluation built in from the start, would test whether consumer incentives can bring cash work into view in Canada. This is the basis for suggestion 3 in the summary.

Gaps by sector, and the platform economy (asked of me)

The pattern is that the CRA acts on one platform at a time and does not generalize the lesson. It pursued eBay’s high-volume sellers through the courts in the mid-2000s (eBay Canada Ltd. v. M.N.R., 2008) and Shopify’s merchants in 2023 (Canada (National Revenue) v. Shopify Inc., 2025). The principle these cases point to, that platforms facilitating transactions should report on the people they pay, was applied generally only with the 2024 platform reporting rules, nearly two decades after eBay. Even those rules cover defined activities: personal services, property rental, sales of goods and vehicle rental. Payments to content creators, and services supplied by businesses rather than individuals, appear to fall outside them, and payment services that are not marketplaces may not be covered at all. The second suggestion in the summary proposes closing those gaps across all platforms.

New research shows what happens when the obligation moves to the platform. Bibler, Grigolon, Teltser and Tremblay (forthcoming) study agreements under which Airbnb collects and remits local lodging taxes on behalf of guests in 24 U.S. metropolitan areas. Before these agreements, when hosts were responsible, taxes were paid on roughly zero to 1.6 per cent of Airbnb transactions. After them, compliance was effectively complete. Compliance among professional hosts with five or more listings was higher before the agreements, between 37 and 59 per cent, but still far from complete.

Canada made the equivalent change for the GST/HST on July 1, 2021. Accommodation platform operators are now responsible for collecting and remitting GST/HST on short-term accommodation supplied by property owners who are not registered (CRA, n.d.). No evaluation of this change, or of the 2024 platform reporting rules, has been published. The data to do so exist, and the methods in Bibler et al. could be applied directly.

4.5 Senator Pupatello

The cost of exempting tips from tax, and how much tip income goes untaxed (asked of Statistics Canada)

For tax purposes, what matters is how tips are paid. Tips controlled by the employer go through payroll and are reported, taxed and counted for EI and CPP. Tips paid directly to the worker depend on self-reporting and are the ones most likely to go unreported.

Estimating the cost of exempting tips would start from an estimate of total tips, because Canada does not report tips separately; it would then model the tax forgone on declared tips using microsimulation, and finally account for behaviour, particularly employers relabelling wages as tips. Because many tipped workers pay little income tax, the benefit of an exemption would go disproportionately to higher-earning tipped workers, and other low-wage workers would receive nothing.

Quebec’s approach, described in the response to the Chair, shows how tips can be brought into view: running them through payroll gives the tax authority visibility, extends Employment Insurance and parental insurance coverage to tip income and protects workers against employers keeping tips. It also provides the best Canadian data on tip income.

The United States offers a contrast. Its 2025 deduction for qualified tips applies only to tips reported on information returns or on the worker’s own return (Internal Revenue Service, 2025). If Canada were to consider an exemption, conditioning it on tips running through payroll would bring tip income into view and extend EI and pension coverage to it.

4.6 Senator C. Deacon

Clustering of businesses just under the $30,000 threshold (asked of Statistics Canada)

Statistics Canada does not look for clustering of businesses with sales just under $30,000, the GST/HST small supplier threshold below which businesses need not register, but this kind of clustering, known as bunching, can be measured with tax data. In my own earlier analysis of individual tax data for 1999, people with self-employment income show a sharp concentration of taxable income just below $29,590, where the federal income tax rate then rose from 17 to 26 per cent. Wage earners, whose income is reported by employers, have little ability to respond this way; the self-employed do. This is older work, but the mechanism has not changed: income that is self-reported can respond to tax rules in ways that third-party-reported income cannot.

Figure 1. Distribution of taxable income among individuals with self-employment income, 1999

Note: Kernel density of taxable income for individuals with positive self-employment income, income range $25,000 to $45,000. The vertical line marks $29,590, the 1999 threshold between the 17 and 26 per cent federal income tax rates. Source: Author’s earlier analysis of Revenue Canada individual tax data.

Research on bunching is considerably more advanced in the United States and Scandinavia, where researchers have access to tax records. Using U.S. tax return data, Saez (2010) found clear bunching at the first kink of the Earned Income Tax Credit, concentrated among the self-employed, and Chetty et al. (2013) used the full population of U.S. tax returns to show that this bunching is driven largely by self-reported self-employment income. Danish population tax registers have supported similar work (Chetty et al., 2011). Testing for clustering at the GST/HST threshold in Canada would require business revenue data, which are difficult for researchers to obtain, and distinguishing genuine changes in activity from underreporting would require the CRA’s audit and compliance data. That the most direct Canadian evidence available is decades old reflects the limits on access to these data.

Incentives and the ease of the tax system (asked of me)

Two pieces of evidence add to my answer at the meeting.

• Pre-filled returns and integrated filing. Close to 90 per cent of tax administrations surveyed by the OECD pre-fill personal income tax returns with information they already hold (OECD, 2025). Canada pre-fills only partially, and taxpayers must still prepare and file a return.
• Report once. Australia’s Standard Business Reporting builds government reporting rules into business software so that information is “recorded once, reported to many” (Australian Government, 2012, p. iv). An evaluation found that early savings were overstated and voluntary uptake was slow (Evans et al., 2012), which supports a common standard built into certified software rather than a voluntary option.

4.7 Senator Loffreda

Making compliance easier without adding red tape, and the role of complexity and effective tax rates (asked of me)

Two design features add to my answer at the meeting. When Quebec introduced mandatory billing in restaurants, it subsidized the purchase of the required equipment (Ryan, 2011), which reduced the cost to compliant businesses. And new reporting should replace existing filings, such as year-end information slips and separate Records of Employment, rather than add to them.

On the Senator’s question about complexity and combined effective tax rates, which I did not address at the meeting:

The highest effective marginal rates on additional earnings in Canada are often faced by lower-income working families rather than high earners. An extra dollar of reported earnings can increase income tax and EI and CPP contributions while reducing the Canada Workers Benefit, the Canada Child Benefit, the Canada Groceries and Essentials Benefit and provincial benefits at the same time. Unreported cash work avoids all of these. Research commissioned by the B.C. Expert Panel on Basic Income illustrates the scale. Combining federal and provincial income taxes, refundable credits and income assistance, Milligan (2020) finds participation tax rates, the share of earnings lost to taxes and reduced benefits when someone moves into work, of 50 to 75 per cent for some British Columbians earning around $20,000, and above 75 per cent around $25,000 for people receiving disability assistance. Marginal effective rates reach 100 per cent over ranges where assistance is reduced dollar for dollar with earnings. His calculations use the Canadian Tax and Credit Simulator, which can produce comparable estimates for other provinces. Smoother benefit phase-outs, and real-time payroll that lets benefits adjust promptly, would make reporting additional work more worthwhile. Much of the measured response to taxation reflects changes in reported income rather than in hours worked (Saez et al., 2012).

Figure 2. Effective tax rates for a single person with no children receiving B.C. disability assistance

Note: The marginal effective tax rate (left) is the share of an additional dollar of earnings lost to taxes and reduced benefits; the participation tax rate (right) is the share of total earnings lost when moving from no work into work. Rates combine federal and provincial income taxes, refundable and non-refundable credits and income assistance. This case is among the starkest and is shown as an illustration. Source: Milligan (2020), Figure 3, bottom panels.

Milligan (2020) concludes that the pattern is complex enough that simplifying how the pieces of the tax and benefit system interact should be a goal of any reform.

References