The number is $10.3 billion.
Nearly 774,000 Canadians, we are told, still owe the federal government that amount in pandemic benefit overpayments.
It is a wonderfully authoritative number. Not $10 billion. Not approximately $10 billion. $10.3 billion. That extra $300 million is reassuring. Somebody has obviously counted.
But almost nobody asks the obvious question: How do we know?
I don’t doubt that CRA has $10.3 billion recorded as owing. The question is how we get from CRA has classified $10.3 billion as owing to Canadians owe CRA $10.3 billion.
Those statements look identical. They aren’t.
During the pandemic, government deliberately got money out quickly. CERB relied largely on attestation. Verification came later, sometimes years later, when people were asked to reconstruct their earnings and produce documentation acceptable to CRA.
If CRA wasn’t satisfied, a payment became an overpayment, an overpayment became a debt, and a debt became an account receivable.
Add the receivables together and voilà: $10.3 billion.
But consider what happens to real people inside this machinery.
In Khairy, the problem involved proving earnings in a cash economy. A person can receive cash, issue invoices, pay expenses and spend the remainder without depositing every dollar in a bank. Yet documentary verification should not privilege bank deposits over other evidence of income. The absence of a deposit is not proof that the income never existed. If that distinction disappears, the government’s preferred method of proving income becomes its de facto definition of income.
In Bishop, the difficulty concerned records from a business that had ceased operating. Years after the pandemic, obtaining documentation from a defunct business could be difficult or impossible. But the disappearance of a business’s records does not establish that the employment or earnings never existed.
Then there is Amanda Coscarelli whose documented earnings fell less than eight dollars short of the $5,000 threshold. Eight dollars. If the statutory threshold was genuinely missed, CRA may have been applying the law correctly. But it is an extraordinary illustration of how a program intended to provide emergency support can produce an enormous repayment demand over a difference smaller than the price of lunch. Administrative precision is not necessarily good public policy.
Finally, consider honorarium income. Government’s public-facing guidance led applicants to believe certain payments qualified, while the legislation contained a narrower definition of eligible income. When the discrepancy emerged, the recipient faced the consequences of relying on government information. The institution that supplied the guidance did not necessarily accept responsibility for the misunderstanding.
These four situations are not legally identical. Some concern evidence, another a rigid statutory threshold, and another the difference between official guidance and legislation. Nor does every successful court challenge establish eligibility.
But together they reveal something the $10.3 billion cannot tell us:
How much of the total represents established ineligibility, how much rests on disputed evidence, and how much reflects the rules and interpretations government chose to apply?
We tend to imagine that a number is either right or wrong. There is a third possibility. It can be perfectly accurate while failing to establish what we think it establishes.
Suppose I devise a test, apply it to a million people and determine that 100,000 failed. I can accurately announce that 100,000 people failed my test.
But watch the language change.
Failed the test becomes ineligible. Ineligible becomes received money they weren’t entitled to. Eventually somebody attaches a dollar sign.
Nothing necessarily involves dishonesty. An administrative determination has simply become a statement about reality.
There is another irony. Federal accounts have carried enormous allowances for doubtful collection against pandemic-benefit receivables. This doesn’t invalidate the debts; doubtful collection and legal entitlement are different questions. But it does show that even government’s own accounting treats the gross total with qualifications that seldom survive the headlines.
Government produces the number. Journalists report it. Researchers cite it. Politicians repeat it. Advocates argue about it.
Eventually the number stops being a claim and becomes a premise.
We no longer ask whether $10.3 billion is properly owing. We ask how aggressively CRA should collect it, whether repayment periods should be extended and how much will ultimately be recovered.
I have called this decision-based evidence-making, a term developed by Shillington and Yalnizyan: the less celebrated cousin of evidence-based decision-making.
We normally imagine government gathering evidence before reaching conclusions. But government can also establish categories, decide what evidence satisfies them, classify people accordingly and then present the resulting administrative data as independent confirmation of reality.
There will be spreadsheets. There will be tables. There may even be decimal points.
Nothing says objective truth quite like 17.3%.
None of this means everyone assessed a pandemic-benefit debt was eligible. Clearly some people received benefits to which they were not entitled. The difficulty is that the published total doesn’t distinguish those cases from decisions affected by questionable evidence requirements, missing records or government guidance that did not accurately communicate the law.
And we have no reliable public accounting of those distinctions.
The lesson extends beyond CERB. Governments count homelessness, welfare caseloads, hospital waits, immigration backlogs and tax debts. We need those numbers. But administrative statistics tell us what systems have counted according to rules somebody designed.
Sometimes that is exactly what we want to know. Sometimes it isn’t.
So, the most useful question about $10.3 billion isn’t whether CRA can produce a spreadsheet that adds up to $10.3 billion. I expect it can.
The question is what exactly CRA counted because perhaps the most dangerous numbers in public policy aren’t the ones that are wrong.
They are the ones that are perfectly accurate measures of something we forgot to question.