A Blueprint for Ending Poverty
A practical, evidence-based path — every plank backed by a real pilot, study, or program, including the parts that didn’t go cleanly.
Poverty isn’t a mystery. It’s not a character flaw, a permanent feature of human nature, or an unsolvable puzzle — it’s the predictable output of a specific set of policy choices, and policy choices can be changed. Most of what plagues a society downstream — theft, violent crime, addiction, poor mental health, strained emergency services — traces back to it in some measure. That doesn’t mean poverty explains everything, but it explains an enormous amount, and treating it as the root problem rather than the background noise changes what “fixing things” actually looks like.
This isn’t a hypothetical exercise, either. Canada already has direct proof that the number moves when it’s targeted deliberately: under the federal Poverty Reduction Act, the government’s own tracking using the Market Basket Measure (Canada’s official poverty line, built from the actual cost of a basic basket of food, shelter, clothing, and transportation in each region) showed the poverty rate fall from 12.1% to 9.5% between 2015 and 2017 alone — about 825,000 fewer people, with child poverty down roughly 30% over the same stretch.1 That’s dated data now and it deserves updating rather than treating as current, but the point it proves doesn’t expire: poverty in a wealthy country isn’t fixed at some natural rate. It moves when policy leans on it, and it’s measurable when it does.
What follows is a blueprint, not a slogan — a sequence of moves that reinforce each other, each one backed by a real pilot, study, or program somewhere in the world, including the parts that didn’t go as cleanly as advocates on either side like to claim. Real evidence is messier than a talking point, and a blueprint worth following has to survive that mess.
Every number here is attributed. Figures come from government statistical agencies, intergovernmental bodies, university research teams, and published economics — principally Canada’s Market Basket Measure reporting, the OECD, the University of Washington’s Seattle Minimum Wage Study, and the published results of the Finnish, Ontarian, Kenyan and Alaskan cash-transfer programs.
Where an advocacy group’s number is used, it is labelled as one and paired with an institutional estimate. The $492 billion tax-avoidance figure is a Tax Justice Network number; it sits next to the OECD’s own, more conservative $100–240 billion range, and the gap between them is discussed rather than averaged away.
The evidence that complicates this blueprint is inside the blueprint. Seattle’s hours-worked losses, Utah’s failure to scale, Ontario’s small follow-up sample, Alaska’s substance-related spike and Kenya’s unresolved long-run question all appear in the sections that make the case — and again, together, in section 8.
This is a policy argument, not a party platform. Every mechanism proposed already operates somewhere in the OECD today, under governments of every stripe.
Before the Comments Start
A piece like this draws a predictable set of reactions, and it’s worth answering them directly instead of pretending they won’t show up.
“This is just socialism dressed up.”
Nothing here nationalizes an industry, abolishes private property, or replaces markets with central planning. Every mechanism below — enforcing existing tax law, indexing an already-existing minimum wage, running a work-linked tax credit like the EITC, funding supportive housing — already operates inside Canada, the U.S., and most of the OECD today, under governments across the political spectrum. Closing a tax loophole isn’t seizing the means of production; it’s collecting a tax that’s already legally owed.
“How exactly do we pay for this?”
Two ways, both spelled out below rather than left vague: recovering revenue that’s already being generated but currently escaping offshore (the OECD’s own, more conservative estimate puts that at $100–240 billion a year globally, separate from the larger $492 billion figure cited further down), and reallocating money the safety net already spends — 91–99% of which already reaches beneficiaries directly — more efficiently by cutting overlap and benefit cliffs, rather than piling new spending on top of the old system.
“People will just stop working if you hand them money.”
This is the single most tested claim in this whole piece, and the data doesn’t support it: in Ontario’s pilot, most recipients kept working; in the U.S., a $1,000 increase in the Earned Income Tax Credit raised employment by 7.3 percentage points rather than lowering it, because the credit is structured to reward work instead of replacing it. The detailed numbers are further down, not just asserted here.
“That $492 billion tax haven number is just activist propaganda.”
Fair to be skeptical of any single advocacy group’s figure — which is exactly why the tax section below leads with a second, independent number: the OECD, an intergovernmental body of the world’s wealthiest economies rather than an advocacy nonprofit, separately estimates global corporate tax avoidance at $100–240 billion a year on its own, more conservative methodology. Even at the low end of that range, it’s still larger than most national anti-poverty budgets.
“People are poor because of their own bad choices, not because of some system.”
Individual choices matter and nothing here pretends otherwise — but the evidence above isn’t about individual choices, it’s about what happens to people who are already working. The EITC’s employment gains show up specifically among people already trying to work more; Ontario’s pilot found the same. A blueprint aimed at making work pay, closing revenue leaks, and housing people stably isn’t an argument that no one is ever responsible for their own situation — it’s an argument that the system surrounding even a responsible, hard-working person can still be stacked against them, and that part is fixable regardless of what anyone believes about the rest.
“Just tax the rich. Just tax corporations. Done.”
Half agreed — and it’s a bigger lever than most people who say it realize. A coordinated 2% minimum effective tax on the roughly 3,000 people worth over a billion dollars would raise an estimated $200–250 billion a year, on top of the profit-shifting money above; extending it to fortunes over $100 million adds $100–140 billion more. That proposal isn’t fringe — it was commissioned by the G20’s Brazilian presidency, and it’s built on the exact same top-up mechanism as the 15% global corporate minimum already discussed below.17 For context, billionaires currently pay an effective rate of about 0.3% of their wealth.
But “tax the rich” as a slogan skips the part that decides whether it works: coordination and design. Twelve OECD countries had net wealth taxes in 1990; four do now. The OECD’s own review blames narrow bases, avoidance, administrative cost, and capital flight.1819 Norway raised its rate to 1.1% in 2022 and the results are still being fought over — critics count tens of billions in wealth walking out the door, while Norway’s finance ministry projects wealth-tax revenue up from 27 to 34 billion kroner over the same period.21 Both of those are true.
Meanwhile the cleanest study on flight — UK administrative data covering a 2017 reform that ended a major offshore exemption — found emigration rose only temporarily, and the wealthy people who stayed reported around 50% more income and tax than before.20
So: yes, tax the very wealthy more, and there’s a serious internationally-designed mechanism for doing it. But do it the way the corporate minimum is being done — coordinated across jurisdictions, on a broad base, with the exit routes closed first — because a rate that one country raises alone is a rate the mobile can simply move away from. And even the high-end estimate doesn’t fund everything. It’s a plank, not the whole floor.
1. Measure It Like You Mean It
A strategy without a public, audited number is a press release.
Before anything else, an anti-poverty strategy needs a number it’s accountable to, published on a fixed schedule, that nobody gets to quietly redefine when the results are inconvenient. Canada already built this tool — the Market Basket Measure — and the 2019 Poverty Reduction Act set explicit, public targets against it: a 20% reduction by 2020 and a 50% reduction by 2030, both measured against 2015 levels.2 Having that yardstick already in place is a genuine advantage most countries don’t have, and it’s worth more than a symbolic declaration, because a declaration doesn’t tell you whether you’re succeeding — a published, audited number does.
The number moves when policy leans on it
Canada’s official poverty rate on the Market Basket Measure, against the targets written into law
2. Stop the Leaks Before Adding the Fixes
Before spending a single new dollar, close the gap through which existing money is already leaving the system.
The Tax Justice Network’s 2024 accounting puts the global loss from corporate profit-shifting and hidden offshore wealth at roughly $492 billion a year — split between about $348 billion in corporate tax avoidance and $145 billion in individual offshore wealth-hiding.3 That’s the advocacy-sector figure, and it deserves the skepticism any single interest group’s number deserves — so it’s worth pairing with the OECD’s own, more conservative institutional estimate, which puts global base erosion and profit shifting at $100–240 billion a year, or roughly 4–10% of global corporate tax revenue.4 The two estimates use different methodologies and land in different places, which is normal for this kind of research — but even the low end of the OECD’s own range is money that’s already being generated inside these economies and is simply exiting through a side door rather than money that needs to be created from nothing.
Two estimates, one side door
Annual global tax revenue lost to profit-shifting and offshore wealth — advocacy total vs. institutional range
This isn’t just a talking point anymore, either — it’s partway built. In October 2021, more than 130 countries agreed through the OECD to a 15% global minimum corporate tax (Pillar Two) on multinationals with revenue over €750 million. As of mid-2025, roughly 65 countries have finalized or drafted legislation to implement it, and the EU began applying it in 2024.5 The United States has taken a partial, parallel path instead of adopting the model rules outright, which is exactly the kind of coordination gap that lets capital route around the rule rather than through it — proof that this lever works only as far as enforcement actually reaches, and that reach is still a live fight, not a finished one.
The individual version of this same mechanism is already drafted. In June 2024, at the request of the G20’s Brazilian presidency, economist Gabriel Zucman published a blueprint for a coordinated 2% minimum effective tax on individuals worth over $1 billion — roughly 3,000 people globally, currently paying an effective rate of about 0.3% of their wealth — estimated to raise $200–250 billion a year, with another $100–140 billion available if the threshold drops to $100 million.17 It borrows its architecture directly from Pillar Two: a top-up applied only where the floor isn’t already met, with each country left to decide whether it gets there through income tax, a levy on unrealized gains, or something else. And it has inherited Pillar Two’s central problem too — the United States dismissed it and Germany was cool to it, which is the same coordination gap that lets capital route around a rule rather than through it. The lever exists. Whether it gets pulled is a political question, not a technical one.
3. Set a Wage Floor That Means Something — and Design It Carefully
An un-indexed minimum wage isn’t a floor; it’s a historical artifact that erodes every year.
A minimum wage that isn’t indexed to real, local cost of living isn’t a floor; it’s a historical artifact that erodes every year until someone bothers to update it. But the evidence on how to do this well is more nuanced than either side of the debate usually admits, and it’s worth looking at directly rather than picking the half that’s convenient.
The most rigorously studied case is Seattle’s minimum wage increase to $13/hour, tracked by a team at the University of Washington using actual hours-worked data rather than survey estimates. Workers earning under $19/hour saw wages rise 3.4% — but average hours worked fell 7.0%. For workers who already had the job before the increase, the wage gain outpaced the hours loss, netting them about $12 more per week on average. But the same research found a decline in hiring of low-wage workers who didn’t already have a foothold in the local labor market — meaning the policy helped incumbent workers more than it helped people trying to get in the door.6
Seattle: who the raise reached, and who it didn’t
Measured effects of the increase to $13/hour on workers earning under $19/hour
Not “minimum wage hikes work” and not “minimum wage hikes backfire” — the design matters as much as the number: phased increases, regional cost-of-living indexing instead of one national figure, and attention to entry-level hiring effects, not just the topline wage.
4. House People First, but Pair It With Actual Supply
Chronic homelessness is expensive to leave unsolved — and the fix has a known failure mode.
In Denver, cycling one person through jail and emergency safety-net services cost the city almost $4,000 in just 90 days; scaled across roughly 250 people cycling through jail and homelessness repeatedly, that added up to about $7.3 million a year for the city. In Los Angeles, people experiencing homelessness accounted for $65.5 million in jail costs in a single fiscal year.7 On the other side of the ledger, New York City’s FUSE supportive-housing program cost about $23,000 per person per year but generated roughly $16,000 in annual savings per person across reduced jail, shelter, and emergency healthcare costs — covering about two-thirds of its own cost.8
| Ledger item | Figure | What it counts |
|---|---|---|
| Denver, per person | ~$4,000 / 90 days | Jail plus emergency safety-net services for one person cycling through both |
| Denver, citywide | ~$7.3M / year | Roughly 250 people cycling repeatedly through jail and homelessness |
| Los Angeles | $65.5M / year | Jail costs alone, attributable to people experiencing homelessness, one fiscal year |
| NYC FUSE — cost | ~$23,000 / person / yr | Full cost of a supportive housing placement |
| NYC FUSE — offset | ~$16,000 / person / yr | Reduced jail, shelter and emergency healthcare use — about two-thirds of the cost |
“Housing First” — giving someone stable housing without first requiring sobriety or program compliance — is the model behind those savings, and it has real evidence behind it.9 But it’s not a standalone silver bullet, and Utah’s often-cited early success is the clearest cautionary tale: later, fuller evaluations found the initial dramatic drop in chronic homelessness was harder to sustain once the program scaled up without enough actual housing supply and support staffing to match demand.10
Housing First works as one layer of a stack, not as a complete strategy on its own — it needs to be paired with genuinely building more housing, not just funding vouchers for housing that doesn’t exist yet.
5. Put Cash Directly in People’s Hands — and Report Exactly What It Does and Doesn’t Do
The plank with the deepest evidence base in the world — and the evidence tells several honest stories, not one clean one.
Finland ran a national basic income trial and published unfiltered results: recipients getting €560/month reported significantly better mental health, lower stress, and more confidence in their own future than the control group — and no statistically significant increase in employment.11 Ontario ran a similar pilot from 2017–2019 before cancelling it early for political rather than evidentiary reasons; in the smaller follow-up survey of participants, 79% reported improved physical wellbeing and 83% reported improved mental wellbeing, roughly half reported reduced alcohol and tobacco use, and only 17% of recipients stopped working altogether — most kept working, and a large share of those who stopped went back to school.12
Four programs, four different answers
What direct cash reliably did, what it didn’t do, and where the evidence is still open
GiveDirectly’s twelve-year universal basic income study in rural Kenya, the largest and longest-running experiment of its kind, found that a lump-sum payment structure outperformed the same total money paid monthly — lump-sum recipients started 19% more businesses with 80% higher revenue than monthly recipients — while causing no measurable local inflation, because the spending spread across many different goods rather than driving up the price of any one thing.13 Researchers are still tracking whether that lump-sum advantage holds up long-term or fades after five or six years, and that open question matters for how any real program should be designed.
And on the specific fear that direct cash fuels crime: a rigorous study of Alaska’s annual Permanent Fund Dividend — a universal, unconditional cash payment every Alaskan receives — found property crime fell about 8% in the four weeks after each payment, while substance-related incidents rose roughly 14% the day after payment and stayed about 10% elevated over the following month. Netted out over a full year, researchers concluded the overall effect on crime was negligible in either direction, and specifically that “crime-related concerns of a universal cash transfer program may be unwarranted” — while also flagging that spreading large payments into smaller, more frequent ones could reduce the substance-related spike without losing the property-crime benefit.14
Put together, this is not a case for cash transfers as a magic fix, and it’s not a case against them either. It’s evidence that direct cash reliably improves health and psychological wellbeing, that its effect on employment and local economies depends heavily on how it’s structured (lump sum vs. monthly, universal vs. targeted), and that the loudest fear attached to it — that it funds crime and idleness — isn’t well supported by the actual data.
6. Redesign the Safety Net Around Take-Up, Not Around Chasing “Waste”
The overhead myth doesn’t survive its own numbers. The real problem is friction and benefit cliffs.
A common claim is that welfare spending is eaten up by administrative bureaucracy. That claim doesn’t hold up under its own weight: a Center on Budget and Policy Priorities analysis of major U.S. low-income programs found 91–99% of federal spending on programs like Medicaid, SNAP, housing vouchers, and the Earned Income Tax Credit reaches beneficiaries directly, with administrative costs eating up single-digit percentages at most.15 The real problem isn’t overhead dollars — it’s that stacking multiple separate means-tested programs, each with its own paperwork, income tests, and renewal schedule, creates enough friction that eligible people don’t receive benefits they qualify for, and creates “benefit cliffs” where earning slightly more income can cause someone to lose several programs at once, punishing the exact behavior — working more — the system is supposed to encourage.
The Earned Income Tax Credit is the strongest real-world proof that a well-designed, work-linked cash transfer avoids that trap: research by economists Hilary Hoynes and Ankur Patel found that a $1,000 increase in EITC generosity produced both a 9.4-percentage-point reduction in after-tax poverty among affected families and a 7.3-percentage-point increase in employment.16 The credit is structured so that working more increases the benefit rather than shrinking it, which is precisely the design lesson worth generalizing across the rest of the system. Consolidating overlapping programs into fewer, simpler income-linked transfers — modeled on what already works about the EITC — is less about recovering wasted dollars and more about making sure the dollars already being spent actually reach the people they’re meant for, without punishing them for trying to work their way further out.
7. Run It Like an Engineering Problem, Not a Moral Crusade
Pilot it, measure it honestly, scale what holds up.
Every piece above is measurable: tax revenue recovered, wage floor versus regional cost of living, chronic homelessness rate, health and stress outcomes, program take-up rates, cost per person served. A serious anti-poverty strategy publishes its numbers on the same public schedule Canada already uses for the Market Basket Measure, funds what the data shows is working, and is willing to say plainly when a pilot underperforms rather than quietly burying it. That discipline — pilot it, measure it honestly, scale what holds up — is the same standard any competent organization applies to anything else it’s serious about fixing.
8. Where the Evidence Still Argues With Itself
A blueprint that’s honest about its evidence has to be honest about where that evidence disagrees.
Every plank above rests on a different strength of evidence, and pretending otherwise is how blueprints get discredited by their own footnotes. Here is the whole thing graded against itself, in one place, before the closing argument.
| Plank | Evidence | The complication |
|---|---|---|
| Work-linked cash (EITC) | Strong | Large U.S. samples, consistent findings on both poverty and employment |
| Direct cash & wellbeing | Strong | Consistent across Finland, Ontario and Kenya; employment effects vary by design |
| Supportive housing costs | Strong | Municipal cost accounting is well documented; FUSE offsets about two-thirds of cost |
| Minimum wage design | Mixed | Real gains for incumbents, real hiring cost for entrants — one city, one increase |
| Housing First at scale | Mixed | Utah’s early success did not survive scaling without matching housing supply |
| Ontario basic income | Mixed | Small follow-up survey; pilot cancelled before long-run data existed |
| Kenya lump-sum advantage | Open | Unknown whether the advantage holds past five or six years |
| Global tax-loss totals | Contested | Advocacy and institutional estimates differ by a factor of two to five |
| Billionaire minimum tax (2%) | Open | Revenue modelled, never implemented; depends entirely on multi-country coordination that hasn’t happened |
| National wealth taxes | Contested | 12 OECD countries in 1990, 4 today; Norway’s revenue rose while its wealthiest left — both sides cite it |
Housing First’s flagship success story got harder to replicate at scale once Utah tried to grow it past its original pilot size. Seattle’s minimum wage data shows real gains for existing workers alongside a real cost for people trying to enter the workforce. Ontario’s basic income results come from a small follow-up survey of a much larger pilot group, cut short before it could produce the kind of long-run data Finland’s trial did. Kenya’s lump-sum cash advantage might not last past five or six years — nobody knows yet. And Alaska’s universal cash transfer produced a real, if temporary, uptick in substance-related incidents alongside its drop in property crime, which is a genuine cost even if the net annual effect washed out to roughly zero.
None of that is a reason to abandon the approach. It’s the reason to run it the way engineers run anything with real stakes attached: pilot before scaling, publish the results that complicate the story along with the ones that support it, and let the next version of the plan be built from what actually happened rather than from how the last version was supposed to go.
9. Sources
Numbered to match the reference marks in the text. Where a permanent publisher link was not verifiable at time of writing, the source is named without one rather than guessed at.
- Government of Canada — Canada’s Poverty Reduction Strategy, and subsequent progress reporting against the Market Basket Measure.
- Poverty Reduction Act, S.C. 2019 — Justice Laws Website.
- Tax Justice Network — The State of Tax Justice 2024. Advocacy-sector estimate; used here alongside note 4.
- OECD — Base Erosion and Profit Shifting (BEPS) overview and revenue-loss estimates.
- Tax Foundation — Global Tax Agreement: Pillar Two details and implementation tracking.
- University of Washington, Evans School of Public Policy & Governance — Seattle Minimum Wage Study.
- Urban Institute — Five Charts That Explain the Homelessness-Jail Cycle.
- NYC FUSE (Frequent User Service Enhancement) supportive-housing evaluation — per-person cost and offset figures.
- National Low Income Housing Coalition — The Case for Housing First.
- The Independent Institute — “The Rise and Fall of ‘Housing First’ in Utah.” A critical re-evaluation; read alongside note 9.
- Kela (Finnish Social Insurance Institution) basic income experiment, final results, as summarised by the World Economic Forum.
- Ontario Basic Income Pilot participant follow-up survey, reported in coverage of the cancelled trial (New Atlas and Canadian outlets).
- GiveDirectly twelve-year universal basic income study, rural Kenya, as reported by NPR.
- Institute of Social and Economic Research, University of Alaska Anchorage — study of the Permanent Fund Dividend’s effect on crime.
- Center on Budget and Policy Priorities — analysis of federal low-income program spending reaching beneficiaries rather than administration.
- Tax Policy Center — How does the EITC affect poor families?, summarising Hoynes & Patel.
- Gabriel Zucman — A Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals, commissioned by the G20 Brazilian presidency (2024).
- OECD — The Role and Design of Net Wealth Taxes in the OECD (2018).
- Sarah Perret — “Why were most wealth taxes abandoned and is this time different?”, Fiscal Studies (2021).
- Advani, Burgherr & Summers — Taxation and Migration by the Super-Rich, CESifo Working Paper 11870 (2025).
- Norwegian Ministry of Finance wealth-tax revenue projections, as reported by AFP, October 2025.
Figures are reported as published by their sources at time of writing. Where a figure is dated — Canada’s 2015–2017 poverty tracking — or contested — the Tax Justice Network and OECD tax-loss estimates — that status is stated in the body text and in the evidence table in section 8, rather than smoothed over. The charts encode only figures named in the text; no value has been interpolated or estimated for the sake of a cleaner line.
