A new way to measure tax reforms: Looking beyond cost and revenue

When governments consider reforming how the tax system runs, whether it be new online portals, automatic filing, tighter audits, or leaner call centres, the question that usually decides whether a reform goes ahead is some version of: will it save money or bring in more revenue? That question matters. But it leaves out most of what a tax system actually does to the people who depend on it.

In a new Policy Forum article in the Canadian Tax Journal, we argue that judging tax administration only by its cost and revenue gives an incomplete and sometimes misleading picture of whether a reform is a good idea. We propose a way to measure what a reform is actually worth: its social return per administrative dollar spent. We call it the marginal value of public funds for tax administration, or MVPF-TAdmin.

The yardstick we use measures the wrong things

Administrative performance is traditionally tracked through operational numbers: cost per return processed, audit yield, processing times, compliance rates. These tell you how efficiently the machine is running. They tell you almost nothing about whether a change leaves tax filers better or worse off.

Every administrative choice—how returns are filed, how audits are targeted, how people reach a human at the Canada Revenue Agency (CRA)—shapes who can actually access the system and how fair they judge it to be. The CRA’s own recent track record makes the point. The 2025 federal budget commits the agency to expanding automatic filing, improving digital portals, and streamlining compliance. In the same period, the auditor general flagged long wait times, low accuracy on general tax questions, and access problems at CRA contact centres. A reform can look efficient on a spreadsheet while shutting people out.

Think of a store that replaces its cashiers with self-checkouts. It cuts staffing costs, so by a narrow efficiency metric it’s a clear win. But the work didn’t disappear. It shifted onto the customers who must now figure out the fruit and vegetable codes, and the ones who can’t manage the machines end up in a long line at the single remaining human-operated till. The savings to the store are easy to count. The cost, borne by the people, isn’t.

Three dimensions worth measuring

Our argument is that reforms of tax-related administrative design create and redistribute welfare, and a serious evaluation of any reform has to capture that. We propose a framework called the marginal value of public funds for tax administration (MVPF-TAdmin). This framework integrates three evaluative dimensions:

• What the reform does to administrative costs and to compliance and revenue—the things conventional metrics already capture.
• Who bears the time, literacy, and access burdens, and who gains access to benefits. For example, moving from mixed paper and digital access to a digital-only portal may not affect or even improve access to tax filing to persons with good internet and digital literacy; however, it may simultaneously raise barriers for those without them.
• How the reform affects trust, transparency, and people’s sense that they’ve been treated fairly—which in turn shapes whether they engage with the system at all.

Putting these together reframes the central question from “Does this reform save money?” to “Does this reform increase inclusive social welfare once fiscal, distributional, and trust effects are all counted?”

A single ratio: social value per administrative dollar

The MVPF-TAdmin is a tool adapted from welfare economics. (Want a plain-language overview of the MVPF approach this builds on? See our Maytree overview.) It views reforms to tax administration as investments that generate social return. It is calculated by aggregating the welfare gains and losses across all potential tax filers and dividing that by the net fiscal cost to government. That is, the numerator adds up the gains and losses tax filers experience through changes in time, access, accuracy and fairness. The denominator is what the reform actually costs the public purse after accounting for knock-on effects such as higher benefit takeup or reduced downstream spending.

A ratio above one means that more social welfare per administrative dollar is generated while a ratio below one means the opposite. Analysts can also weight the numerator to reflect that, for example, an hour of saved hassle means more for a low-income senior than for a high-income professional.

This is a transparent, consistent way to compare reform options on efficiency, equity, and legitimacy, rather than on cost alone.

A test case: automatic federal benefits

To show how this works, we apply it as a proof-of-concept to automatic federal benefits (AFB). Beginning in the 2026 taxation year, the CRA will automatically file returns for a first group of roughly 50,000 low-income people with simple tax situations and no tax owing. This is the leading edge of a broader shift toward prefilled returns for as many as one million Canadians who would otherwise not file.

The problem AFB targets is real and stubborn: each year, 10 to 12 percent of Canadians don’t file a return, and as a result forgo benefits they are entitled to. Non-filers are disproportionately low-income, younger, Indigenous, recent immigrants, and people in rural or northern communities. Why those access gaps persist—and how governments might close them—is the focus of another Policy Forum article which examines tax filing as shared state administrative infrastructure and proposes an intergovernmental tax-filing and benefit access accord.

The parliamentary budget officer estimates that a fully phased-in automatic benefits system could result in roughly $1.6 to $1.8 billion a year in additional benefits to about a million low-income people, mainly through better takeup of the Canada child benefit, the Canada workers benefit, and the GST/HST credit (renamed the Canada Groceries and Essentials Benefit in July 2026).

That makes automatic federal benefits a natural test of whether a reform that expands benefit access in fact generates welfare gains that cost-based evaluation simply can’t see.

The same reform, four very different outcomes

We examine automatic federal benefits through four stylized scenarios to show that the same headline reform can produce very different social returns depending on how it’s designed and delivered (summarized in the chart below).

In the base case, you count only the direct dollars in and out—increase in benefits paid to tax filers, and cost spent on administration (i.e., setting up the technology and pre-filling the tax returns). This is consistent with conventional fiscal costing practices. Using this approach, we show that the MVPF-TAdmin ratio comes out around one: roughly a wash in welfare terms. This ratio of one means the reform looks neither worth doing nor worth avoiding. But it does not incorporate equity and legitimacy and shows the limit of what cost-only accounting can detect.

The well-designed case is the mirror image. We assume that in such a case, the reform reaches nearly everyone eligible, runs with little friction, and its ripple effects—more economic security means less spending on crisis supports and more taxable consumption—offset almost all of its cost. At the same time, the access, fairness, and trust gains the base case ignored now enter the numerator, weighted toward the people who benefit most. With the numerator rising and the denominator shrinking toward zero, the MVPF-TAdmin ratio climbs sharply. This ratio means the reform looks worth doing—but it marks the best the reform can achieve. That number is an idealized upper bound rather than a forecast; what it really shows is how much value a cost-only lens was leaving on the table.

The poorly designed case is the one that advises caution. Here the rollout of automatic federal benefits is poorly executed in that it excludes exactly the people it was meant to help: prefilled returns don’t reach many non-filers because of weak outreach, spotty broadband, or no bank account for direct deposit. Takeup stalls, so few of the benefits flow. The damage is in the numerator. The filers who can’t navigate the new system are pushed into more time, more hassle, and more frustration—burdens that fall hardest on the low-income and marginalized filers the reform was supposed to reach and so weigh heavily once equity is counted. The MVPF-TAdmin  ratio is below zero, meaning that the reform looks actively worth avoiding. It destroys more welfare than it costs, even as conventional accounting, seeing only the saving, would wave it through as efficient.

The conservative design case sits between these extremes, and it is the one anchored to the parliamentary budget officer’s central estimates rather than to a best or worst case. Here take up is strong but not universal, behavioural responses are modest, and the ripple effects onto other programs are real but partial—enough to offset some of the fiscal cost without erasing it. Counting those gains alongside the equity and trust improvements the base case ignored, the MVPF-TAdmin ratio comes out around 2.4. This ratio well above one means the reform looks worth doing. And because it rests on cautious, evidence-based assumptions rather than the optimism of the well-designed case, it is the most realistic of the four.

Why a “cheap” reform can be the worst one

Set these four cases side by side and the real trap comes into focus. The poorly designed rollout of automatic federal benefits is not only the worst outcome for the people it was meant to help, but on the government’s books it is also the cheapest. Benefits that never reach non-filers are benefits the government never pays out, so weak execution shows up as a low price tag rather than as failure. A reform can therefore look like a bargain on a cost-per-return basis—fewer staff, lower processing costs, lower transfers—while quietly destroying social value by pushing complexity and access barriers onto low-income, low-connectivity taxpayers. The MVPF-TAdmin lens catches what a cost-only view misses: a reform can be cheap because it is failing and making the country worse off.

This is the core strength of the framework. When access depends on whether people can use a new system, the design details aren’t an implementation footnote. They are where most of the social value is won or lost.

Aligning evaluation with what we value

None of this means abandoning fiscal discipline. Administrative resources are scarce, and cost still counts. The argument is that cost should sit inside a fuller account of public value, not stand in for it.

Practically, that means policy makers, the CRA, and fiscal review bodies should supplement cost-per-return and yield figures with standardized measures of taxpayer burden, access, and trust—some of which can be built today from existing data, others phased in over time. The value of the MVPF-TAdmin framework lies less in any single number than in shaping how reforms are judged as they are piloted, scaled, and revised over the coming years, so that modernization improves access and fairness alongside efficiency.

This post draws on our Policy Forum article, “Beyond Cost and Yield—An MVPF-TAdmin Framework for Evaluating Tax Administration Reforms Through Public Value, Equity, and Welfare,” in the Canadian Tax Journal (2026) 74:2.