Universal Basic Abundance

This is the second brief in Category IX, which is new to this map. The categories before it ask what can be built; Category IX asks how the output is produced, claimed, distributed and counted. Within it, this slot is the instrument slot. Post Scarcity Economics holds the theoretical question of whether scarcity can stop binding; Wealth Distribution Systems holds the mechanisms that move stocks of claims. This brief holds one specific proposal for a world in which scarcity still binds: an unconditional payment, and the question of whether it can be made universal and afforded. A basic income is a rationing device. If the strong post-scarcity claim were true, this slot would have no subject.

Established Unconditional cash is unusual in social policy: it has actually been tested, repeatedly, by randomised trial. Across five decades, four continents, samples from a few hundred to over twenty thousand, by researchers with opposed priors and with several studies pre-registered. There is more randomised evidence on this than on almost any other macro-scale proposal in this category — and it is mostly ignored in public argument in both directions, with advocates citing pilot results and skipping labour supply, and opponents citing the 1970s labour-supply findings and skipping fifty years of nulls and positives since.

Established One framing correction governs everything below. “Universal” and “unconditional” are separable, and almost no study tests both. Alaska's dividend is universal and unconditional but tiny. Kenya's is unconditional and near-universal within villages but geographically bounded. Finland's was unconditional but restricted to the unemployed. The US and Stockton studies were unconditional but targeted by income. No study anywhere has tested a payment that is simultaneously universal, unconditional, permanent and large enough to live on. Hoynes and Rothstein put it flatly: the ongoing pilot studies “will do little to resolve the major outstanding questions.” Every number in this brief should be read against that limit.

Established Kenya is the largest and longest trial, and it is the strongest evidence that transfers do not induce idleness. Banerjee, Faye, Krueger, Niehaus and Suri randomised 295 villages across Bomet and Siaya counties, roughly 23,000 treated adults, into four arms: control (100 villages, about 11,000 people); a twelve-year transfer of $0.75 a day, about $1.88 PPP (44 villages); the same daily amount for two years (80 villages); and a one-off lump sum of about $500, roughly $1,250 PPP (71 villages). At two years there was no significant effect on total hours worked in any arm. There was a large occupational shift: wage employment fell — long-term arm −99.85 hours, p<0.05 — and was offset by self-employment.

Established The Kenyan economic results are large and the nulls inside them are load-bearing. Enterprise creation rose +9.93 enterprises, about +14%, in the long-term arm, with revenues +41% and net revenues +52%; the lump-sum arm produced +14.67 enterprises with effects at least as large as the twelve-year arm on most metrics; the two-year arm's +3.39 was not significant. All three arms significantly reduced standard depression measures and life satisfaction improved, largest in the short-term arm. Consumption in the long-term arm rose 5% and not significantly, though protein consumption rose 15–25% and food variety improved significantly in all arms. And there were no significant effects on children's anthropometric outcomes in the long-term or lump-sum arms. Two of the paper's authors co-founded the implementing NGO, which makes the reported nulls harder to discount, not easier.

Established The pandemic follow-up adds the durability finding, and it is unwelcome for the enterprise story. Hunger fell by 11 percentage points in the long-term arm against a control mean of 68%, and 5–6 points in the other arms; sick household members fell 3.6–5.7 points against a 44% control mean; depression improved in the two ongoing-transfer arms but not in the lump-sum arm. But the non-agricultural enterprises the transfers had created were wiped out by the shock — long-term-arm non-agricultural profits fell 54 percentage points, reversing the prior gains. Cash builds enterprise; enterprise built by cash is not shock-proof. Frontier The structural finding is not about work at all: holding money transferred roughly constant, the structure of the transfer changed outcomes substantially, with lump-sum matching or beating twelve-year on enterprise creation and mental-health effects running inverse to economic ones. That is a direct problem for any argument treating “unconditional cash” as one treatment.

Established The three-year United States trial is the most inconvenient result in the literature, and the interest runs against the finding. Vivalt, Rhodes, Bartik, Broockman and Miller ran an RCT in Texas and Illinois: 1,000 treated adults receiving $1,000 a month against 2,000 controls receiving $50, ages 21–40, household income at or below 300% of the federal poverty level, November 2020 to October 2023. It grew out of the OpenResearch effort — a funding environment sympathetic to basic income. It reported labour market participation down 4.2 percentage points (s.e. 1.7), hours worked down 1.3 a week (s.e. 0.6) in survey data and about 1.4 in time diaries, and individual income excluding the transfer down about $1,900 a year (s.e. roughly $900), 5–6% of the control mean. The recovered time went predominantly to leisure, with negligible increases in childcare, job search, exercise or self-improvement.

Established The companion consumption paper is where the balance-sheet result sits, and it includes an unexplained residual the authors state openly. Measured spending rose at least $310 a month — food +$67, rent +$52, car costs +$30, gifts and loans to family and charity +$22, a 25% increase — with alcohol and tobacco up about $13 a month, roughly 1% of the transfer. But net worth fell about $1,000, about $2,000 excluding real estate, total debt rose about $1,800, and net-worth increases above $5,700 can be ruled out. The authors' own decomposition of the roughly $950 net monthly transfer: about $310 to consumption, $280 to increased leisure, $30 to savings, −$60 to higher debt, and about $390 — roughly 40% — unexplained. Frontier A residual that large is a limit on how confidently any mechanism can be asserted, and the authors say so.

Established The health paper is the starkest, because it produces tight bounds rather than a null. Improvements in stress and food security appeared in year one only and were gone by year two. Pooled across the study, the authors can rule out mental-health improvements greater than 0.03 SD, physical-health improvements greater than 0.02 SD, and biomarker improvements greater than 0.01 SD across A1c, blood pressure, cholesterol and obesity. Healthcare utilisation rose and medical spending rose about $20 a month. Three years of $1,000 a month to low-income US adults produced more consumption, more leisure, more help given to family, no durable health improvement, and lower net worth. That is roughly what a standard income-effect model predicts. It is not what the advocacy literature describes, and it was produced by researchers with no incentive to find it.

Established Finland is the only statutory, nationwide, randomised basic income, and its design removes the self-selection problem that afflicts most pilots. Kela randomly selected 2,000 unemployed people from those receiving basic unemployment benefits in November 2016 and paid them €560 a month, tax-exempt and unconditionally, from 1 January 2017 to 31 December 2018; participation was compulsory for those selected. Frontier The employment effect was small and positive — roughly six additional days of employment over the November 2017 to October 2018 window on a base of about 78 — with no significant group-level effect in the first year and modest improvement for families with children in both. Established Wellbeing improved on multiple self-reported measures: life satisfaction, mental strain, depression, sadness, loneliness, perceived memory and concentration, perceived economic wellbeing, and trust in institutions and other people.

Established And the Finnish confound is stated by the government's own report, which is why quoting the employment result without it is quoting it wrongly. A new activation model introduced at the start of 2018 imposed more stringent entitlement conditions on unemployment benefits and bit asymmetrically on treatment and control. The second year's employment comparison is therefore contaminated in a direction that is hard to sign. Frontier Six days on a base of 78 is small relative to that confound, and Finland is properly read as neither a vindication nor a refutation.

Established Stockton is the most cited guaranteed-income result in US public debate, and it has two publications that say different things. SEED paid $500 a month for 24 months to 125 recipients — a 100-person core research sample — against 200 controls, from February 2019. The first-year report, funded by a foundation's evidence programme and describing its own analysis as “preliminary” and limited to “basic descriptive and inferential statistics”, reported full-time employment in the treatment group rising 28% to 40% against 32% to 37% in controls, income volatility 46.4% against 67.5%, the share able to cover a $400 emergency rising 25% to 52% against 25% to 28%, and significant improvement on the Kessler-10 distress scale, with under 1% of spending on tobacco and alcohol.

Established The peer-reviewed paper is the one to cite, and on employment it says something different. West and Castro in the Journal of Urban Health (April 2023) analysed 110 treatment and 198 control observations, retention 55% and 35%. Significant effects: income volatility 19% against 26% (p=0.039); Kessler-10 improvement (p=0.027); SF-36 pain, energy, emotional wellbeing and physical functioning (p between 0.006 and 0.037); ability to meet a $400 emergency (p<0.001). Employment was null: the paper states the employment trend data were not statistically significant, alongside the financial-wellbeing scale during the pandemic and second-year income volatility. Its own stated limitations are that limited power prevented subgroup analysis and attrition may have impaired detection.

Established The 1970s negative income tax experiments are the thing everybody forgot, and they agree with the modern US result. Five randomised experiments ran 1968–1980: New Jersey (1,216 enrolled, 983 analysed, urban two-parent), Rural Iowa and North Carolina (809 to 729), SIME/DIME in Seattle and Denver (about 4,800, running to 1980), Gary, Indiana (1,799 to 967, mostly Black single-parent households) and Mincome in Manitoba (about 1,300). Frontier Reported labour-supply reductions relative to controls: husbands 0.5%–9.0%, roughly 0.5 to 4 hours a week; wives 0%–27%, up to 166 hours a year; single mothers 15%–30%; averaged across the four US experiments, primary earners reduced work by roughly 5%–7.9%. These are the standard reported ranges, sensitive to the treatment of attrition and of self-reported income, and the summary they are drawn from is by a basic-income advocate reporting them against interest.

Frontier The non-labour findings from the same experiments are positive and much less quoted: improved elementary school attendance, higher test scores, longer schooling, fewer low-birthweight babies, higher food consumption and better dietary quality. Frontier And the single most consequential result in the history of this literature is one that did not replicate: SIME/DIME showed higher divorce rates in treatment groups; Mincome later found lower rates. The marital-dissolution finding is widely credited with killing US basic-income politics in the late 1970s. A policy proposal was buried by a result subsequent work failed to reproduce, which is a finding about how this evidence gets used rather than about cash.

Established Alaska is the only permanent, universal, unconditional payment on earth, and its labour-market effect is a precisely estimated zero. The Permanent Fund Dividend has paid every eligible resident annually since 1982, ranging from $331 in 1984 to $2,072 in 2015 — roughly $8,000 for a family of four in recent peak years. Jones and Marinescu, using synthetic control on CPS data 1977–2014, find an employment effect of 0.001 percentage points, 95% CI −0.031 to 0.032, with part-time work up 1.8 percentage points, a 17% rise, 95% CI 0.004 to 0.032, p=0.025. Frontier Their proposed mechanism is general equilibrium — the transfer stimulates consumption and hence labour demand, offsetting the income effect — supported by employment reductions appearing only in tradable sectors. That mechanism is an interpretation, not a measurement.

Frontier Alaska's distributional record is real and its educational record is a null. The University of Alaska Anchorage's institute reports that dividends reduced the state poverty rate by an estimated 2.5 to 4 percentage points annually since 1990, lifting about 25,000 Alaskans out of poverty in 2015, with the largest effects for children, Alaska Natives and rural residents — figures reported through a university news release about its own institute's work rather than through the underlying paper. Against that, Lerner's synthetic-control study of the dividend's effect on high school status completion among 18–24 year-olds found no detectable effect, p = 0.95, ruling out true effects above about 2%, with trans-Alaska pipeline migration confounding the period. Cash at Alaska's scale did not move educational attainment.

Established Ontario is where the binding constraint shows itself, and it is not economics. The pilot enrolled about 4,000 participants across Hamilton, Lindsay and Thunder Bay, paying roughly C$17,000 a year for individuals and C$24,000 for couples with benefits reduced 50 cents per dollar earned. An incoming provincial government cancelled the three-year, C$150-million programme; a legal challenge to the cancellation failed. Speculative The surviving evidence is a survey of 217 former recipients plus 40 interviews, run in partnership with an anti-poverty advocacy organisation: nearly three-quarters of working participants kept their jobs, 80% reported better health, 83% less stress and anxiety, 81% more self-confidence, 66% stronger family relationships. There is no control group and no randomised comparison surviving the cancellation. Those percentages are accurate descriptions of what respondents said and are not estimates of the pilot's effect.

Established On cost, Canada has the best public arithmetic, and the honest version is less favourable than the headline. The Parliamentary Budget Officer costs a national guaranteed basic income for 2025 at 75% of the low-income measure — C$21,903 for a single person, C$30,975 for a couple, plus C$7,355 for persons with disabilities. On a nuclear-family definition: C$107 billion gross, C$112.4 billion with behavioural response, poverty down 34%. On an economic-family definition: C$53 billion gross, C$57.0 billion with behaviour, poverty down 40%. Frontier Three things matter more than the headline. It is not universal — the PBO models a means-tested top-up that tapers, so it does not answer the universal question at all. The offsets are large and politically explosive, assuming elimination of roughly 25 federal and provincial credits and programmes plus reductions in the basic personal amount, which redistributes within the low-income population and creates losers among current beneficiaries. And the behavioural cost is small in the model — C$5bn on C$107bn, under 5%, from applied elasticities — against a measured 4.2-point participation fall in the one large modern RCT. Established The 2025 poverty-reduction estimate is also lower than the same office's 2021 estimate of 49%: a downward revision by the same model, where revisions in this literature usually go the other way.

Frontier The central open question is whether the two best-evidenced findings are compatible, and the candidate reconciliation is untested. Unconditional cash does not reduce labour supply in Kenya or Alaska; it does reduce it in the 1970s US experiments and in the 2020s US trial. The obvious reconciling variable is transfer size relative to income together with the thickness of the local labour market, and the pattern across Kenya, Alaska, Finland and the United States is monotone in transfer size. Speculative That is a hypothesis, not a finding. The design that would settle it is a dose-response trial varying transfer size within one setting, and nobody has run one.

Frontier Whether cash improves mental health durably is genuinely contested, and the contest is between study designs rather than between camps. Kenya, Finland and Stockton all report improvements. The US trial can bound pooled mental-health improvement above 0.03 SD and shows year-one gains gone by year two. Frontier On physical health the evidence leans negative: improvements above 0.02 SD, and biomarker improvements above 0.01 SD, are ruled out. What is missing is longer follow-up with pre-registered endpoints; the US study is the only one with bounds tight enough to be informative in either direction.

Frontier The structure-of-transfer question is the most surprising live result and the least explored. In Kenya, holding money roughly constant, a single lump sum matched or beat a twelve-year stream on enterprise creation and revenues, while mental-health effects ran inverse to economic ones — lump-sum recipients showed the smallest depression reduction despite the largest earnings growth. Frontier A minority position follows directly: lump sums may beat streams, which almost nobody advocates politically and which the arm comparison supports on the money delivered. Replication outside Kenya is the whole question.

Frontier A related and fringe-but-important hypothesis: the effect may run through certainty rather than money. Twelve-year-arm recipients saved and joined rotating savings groups at far higher rates than two-year-arm recipients receiving identical amounts, and Finland's clearest results were on perceived security rather than on income. Speculative The design that would test it varies announced duration while holding the amount fixed. Kenya does this partially and by accident; nobody has done it deliberately.

Speculative Whether unconditional cash is inflationary at scale is almost entirely unevidenced in either direction. The critical version — that the transfer is bid away in rents — is common in commentary and rests on nothing. The US trial found rent spending up $52 a month, but that is a within-recipient consumption change and not a market price effect. Frontier Village-level randomisation of the kind Kenya uses can in principle detect general-equilibrium price effects; a saturation design randomising at market level with price outcomes is the missing study, and it is the one that would matter most for a genuinely universal programme.

Frontier Whether a universal payment is affordable depends on a distributional argument that microsimulations answer differently. Hoynes and Rothstein's structural verdict is that a UBI would direct much larger shares of transfers to childless, non-elderly, non-disabled households than existing programmes, and much more to middle-income than to poor households, and that one large enough to raise transfers to low-income families would be “enormously expensive”. Frontier The counter-case is that consolidation makes it affordable. Both are modelled rather than measured, the models disagree, and no national implementation exists with published fiscal accounts to adjudicate them.

Frontier And the constraint that may dominate all the economics: political durability across an electoral cycle. Ontario cancelled a three-year C$150-million pilot mid-stream; Finland did not extend; Alaska's dividend has survived four decades and is the one tied to a resource fund rather than to general revenue. Speculative That correlation is suggestive and rests on a handful of cases. Comparative political-economy work on which unconditional programmes survive, and why, does not exist in usable form.

Established The binding evidential bottleneck is stated by the literature's own reviewers: none of the pilots tests the thing being proposed. Universality, unconditionality, permanence and adequacy are four properties, and no study has combined them. Alaska has universality, unconditionality and permanence without adequacy; the US trial had adequacy and unconditionality without universality or permanence; Finland had unconditionality and statutory backing without universality or adequacy. Every extrapolation from these studies to a universal basic income crosses at least one untested boundary.

Established Statistical power and attrition bite hardest exactly where the results are most quoted. The peer-reviewed Stockton analysis rests on 110 treated and 198 control observations with 55% and 35% retention, and the paper itself says limited power prevented subgroup analysis and that attrition may have impaired detection. Frontier A programme evaluated at that scale can support the effects it detected and cannot support a null being read as an absence, which is the correct reading of its employment result: not significant, not shown to be zero.

Frontier Almost every wellbeing outcome in this literature is self-reported, and a separate brief in this category owns the question of whether such measures are valid at all. Finland's mental-strain results, Kenya's life-satisfaction results and Stockton's Kessler-10 results are all reported here as the studies report them. The measurement critique — including whether ordinal reported scales can support the comparisons made of them — belongs to the Human Flourishing slot and is not adjudicated here.

Established Pilots cannot see general equilibrium, and that is where the interesting effects of a universal programme would be. Prices, wages, rents, labour demand and firm entry all respond to a transfer everybody receives; none of them can respond to a transfer 1,000 people receive. Frontier Alaska is the only setting where a general-equilibrium channel has even been proposed — the tradable-versus-non-tradable pattern behind the offsetting labour-demand story — and there it is an interpretation of a synthetic-control result rather than a measurement.

Established And the delivery machinery is a real constraint that the debate rarely mentions. Paying every resident requires an individual-level register of residence and income, a payment rail that reaches people without bank accounts, and an integration with the tax system that claws the payment back from those who do not need it. Alaska has all three because it built them for one payment a year; most countries do not have them at individual level. Universality is an administrative property before it is a fiscal one.

Established Nothing on this map produces a result this brief waits on, and no typed depends-on edge is claimed. The instrument does not wait on a discovery; it waits on a register that can pay everyone, a financing package that survives the losers it creates, a programme that outlives an electoral cycle, and a funder willing to pay for the trial nobody has run. All four are recorded as typed requirements below and all four are things a legislature or a philanthropist could choose to supply.

Frontier What it waits on from research is specific and mostly cheap. A dose-response design varying transfer size in one setting; a duration-varying design holding amount fixed; a saturation design randomising at market level with price outcomes; and pre-registered mental- and physical-health endpoints with follow-up beyond three years. Speculative None requires new methods. The reason none has been run is that pilots are funded to demonstrate rather than to discriminate, which is the most reliable structural fact about this literature.

Established The single most valuable experiment is a saturation design: randomise at the level of a labour and housing market, not at the level of a household. It is the only way to observe the effects that matter for a universal programme — rents, wages, prices, firm entry, labour demand — and it is the direct test of the inflation objection, which currently has no evidence on either side. Kenya's village-level randomisation is the closest existing design and its scale is too small for prices.

Frontier Second: a dose-response trial. Vary transfer size within one setting and one labour market. The claim that the labour-supply response scales with transfer size relative to income is the reconciliation the whole literature implicitly relies on, and it has never been tested directly — it is inferred from a cross-study pattern in which country, transfer size, labour market and decade all vary together.

Frontier Third: hold the amount fixed and vary the announced duration. Kenya's twelve-year and two-year arms differ in behaviour in ways the money alone does not explain, with far higher saving and rotating-savings-group participation in the long arm. If the mechanism is certainty rather than cash, that changes what an optimal programme looks like more than any cost estimate does.

Frontier Fourth: replicate the lump-sum-versus-stream comparison outside Kenya. The arm comparison there is the most surprising result in the modern literature and it currently rests on one country, one implementing NGO and one time period.

Established Fifth, and least glamorous: pre-register the health endpoints. The US trial is the only study that produced bounds rather than an unpowered null, and it did so because it was designed to. Every future pilot that reports a health improvement without a pre-specified endpoint and a power calculation adds nothing to what is already known.

Established The engineering of a universal payment is register, rail and taper, in that order. The register has to identify every resident individually and know where they live; the rail has to reach people without bank accounts, reliably, at a cadence the payment assumes; the taper has to reclaim the payment through the tax system from people who do not need it, without producing effective marginal rates that swamp the transfer. Ontario's pilot used a 50-cents-per-dollar reduction; the PBO's costing assumes a taper and is therefore costing a guaranteed income rather than a universal one.

Frontier The taper is where the affordability argument is actually settled and where it is most often finessed. A universal payment at a given benefit level costs dramatically more gross than a tapered one, and much of the increment is clawed back through tax from people who do not need it — so the net cost depends almost entirely on the tax schedule chosen, which is why gross-cost headlines in this debate are close to uninformative on their own. Established The PBO's C$107bn and C$53bn figures differ by a factor of two purely on the definition of a family unit, which is the same point in a different form.

Established And the disability supplement is the design detail that decides whether consolidation is defensible. The PBO's model adds C$7,355 for persons with disabilities on top of the base amount, precisely because a flat payment replacing categorical programmes leaves people with high needs worse off. Any consolidation-financed proposal has to price its supplements, and a proposal that does not is not costed.

The boundaries inside Category IX. This slot is a specific instrument: an unconditional payment, and the experimental record of how recipients respond to it. Post Scarcity Economics is the theoretical claim that scarcity itself can cease to bind — the automation-displacement literature lives there, and although it is frequently used as a motivation for basic income, none of the evidence about whether basic income works depends on it and the two must not be mixed. Wealth Distribution Systems covers mechanisms and their measured effects on stocks and flows: the financing side, wealth and inheritance taxation, and sovereign funds. Alaska appears in both briefs, correctly and for different reasons — here as an unconditional universal payment with an estimated labour-market effect, there as a sovereign-fund distribution mechanism. The Canadian cost figures are this brief's; wealth-tax elasticities are that one's.

Two further Category IX slots are being authored in parallel and this brief does not reach into either. Human Flourishing owns whether the self-reported wellbeing measures used throughout this evidence base are valid at all — a question that bears directly on every “life satisfaction improved” claim reported here, and which is deliberately left to it. Human Development Metrics owns composite indices; this brief uses poverty rates as outcome measures, not composite indices, and does not adjudicate what a poverty line is.

Elsewhere on this map: Future Public Administration, which owns the delivery machinery any universal payment runs on and the finding that evaluation is discretionary; Digital Citizenship, where the individual-level register question is a rights question rather than an administrative one; Institutional Design, where programme durability across electoral cycles is stated generally; Future Democracies and Future Federalism, which supply the political-durability and fiscal-tier constraints; Artificial General Intelligence, the current source of the displacement argument used to motivate this instrument; and Future Education Systems, one of the categorical programmes a consolidation-financed payment would replace.

Established The standing of the sources here is unusually consequential, and it runs in both directions. The Kenyan papers are working papers with two authors who co-founded the implementing NGO — and they report nulls on consumption and child anthropometrics. The US trial came out of a basic-income-sympathetic funding environment and reported a participation fall, a net-worth fall and tightly bounded health nulls, so its interest runs against its finding and its weight rises accordingly. The 1970s summary used here is by a basic-income advocate reporting the labour-supply reductions and the divorce finding against interest. The Ontario survey was run in partnership with an advocacy organisation and has no control group. The Finnish government summary is self-reporting and nonetheless states the activation-model confound plainly.

Established The one non-interested costing in the record is a legislative budget office, and it revised its own estimate downward. Canada's Parliamentary Budget Officer is non-partisan by statute; its 2025 poverty-reduction estimate of 34% on a nuclear-family basis is lower than its own 2021 figure of 49%, using the same model. In a literature where revisions usually move the other way, that is worth more than a headline.

Frontier The institutional constraint that decides the question is durability, and the evidence is a handful of cases pointing one way. Ontario's programme was cancelled by an incoming government and the legal challenge failed; Finland's was statutory, time-boxed and not extended; Alaska's has run for four decades and is the only one financed by a dedicated fund rather than by general revenue. Speculative The hypothesis that hypothecation to a fund is what makes an unconditional payment survive is supported by one case and contradicted by none, which is not the same as being supported.

Established The most consequential ethical point in this brief is that a labour-supply reduction is not automatically a cost. The US trial found participation down 4.2 points, hours down 1.3 a week, and the recovered time going predominantly to leisure. Increased leisure is a welfare gain, and treating it as a pure fiscal loss embeds a judgement about the value of paid work that the evidence does not supply. The same result read as a cost and read as a benefit is the same number; the disagreement is about what to count.

Established The temptation-goods objection is the one claim the evidence flatly refutes, in every setting tested. Alcohol and tobacco spending rose about $13 a month in the US trial, roughly 1% of the transfer; Stockton's recipients spent under 1% on tobacco and alcohol; Kenyan recipients built enterprises with revenues up 41%. The recurring political objection to unconditional cash has been tested repeatedly and has failed every time.

Frontier The serious distributional objection runs the other way, and it is about who loses under consolidation. A payment financed by eliminating roughly 25 credits and programmes, as the Canadian costing assumes, redistributes within the low-income population as well as towards it, and creates losers among current beneficiaries of targeted programmes — disproportionately people with high needs, which is why the disability supplement exists in the model. Frontier Hoynes and Rothstein's version for the United States is that a universal payment moves money towards childless, non-elderly, non-disabled and middle-income households. That is a distributional argument with modelled support and no experimental support, and it is the strongest argument against the instrument that does not depend on labour supply at all.

Established And there is an ethics of evidence here that this brief applies to itself. The Ontario percentages — 80% reporting better health, 83% less stress — are accurate reports of what 217 self-selected former recipients said in a retrospective online survey with no control group. Presenting them as programme effects, which is common, is not a small error: it converts testimony into causation. The respondents are not the problem; the citation practice is.

Established The terminal position is a declared tie, and both halves are measured. Unconditional cash produces real welfare gains: hunger down 11 points in Kenya's long arm, poverty down 2.5 to 4 points a year in Alaska, $310 a month of additional consumption at the bottom of the US income distribution, significant improvements in distress scores and emergency resilience in Stockton, and about 1% of transfers going to alcohol and tobacco. And it produces measurable labour-supply reductions and costs a great deal: 4.2 points of participation and 1.3 hours a week in the US trial, 5–7.9% among primary earners in the 1970s experiments, and a Canadian gross cost of C$107 billion for a programme that is not even universal. Both are supported. They are not in conflict, and a page that reported only one would be misreporting the record.

Frontier What follows is that the interesting question is no longer whether unconditional cash “works”. It is which structure, at which size, financed how, and surviving what. The Kenyan arm comparison shows structure changing outcomes at constant money; the cross-country pattern shows size plausibly changing the labour-supply sign; the Canadian costing shows financing deciding who loses; Ontario shows the electoral cycle deciding whether any of it runs long enough to measure. Speculative A programme optimised on all four has never been designed, let alone tested.

Frontier The civilisational claim usually attached to this instrument — that it is the answer to technological displacement — is the weakest part of the case and does not need to be part of it. The displacement forecasts have aged badly and the evidence about whether unconditional provision works does not depend on them. Established A basic income is defensible or not on its measured effects on poverty, consumption, health, work and public finances, all of which have been estimated. Tying it to a forecast that has already missed once makes the case weaker, not stronger.

Speculative And the long-run version worth taking seriously is the Alaskan one rather than the European one. The only unconditional universal payment that has survived four decades is hypothecated to a fund, small relative to income, and politically defended as a property right rather than as welfare. If durability is the binding constraint, the design that survives may be one the advocacy literature considers too small to matter — which would be an uncomfortable outcome for both sides of the argument.

These horizons track study completion, fiscal decisions and political durability rather than technology readiness:

• 10 yr: Frontier The dominant fact in this window is that the study everyone needs is not funded. Expect more targeted, time-limited guaranteed-income pilots reporting self-reported wellbeing gains on small samples, and expect the four-property gap — universal, unconditional, permanent, adequate — to remain unclosed. Frontier Expect the US three-year trial's results to be replicated in direction if any comparable rich-country trial runs, because they agree with the 1970s experiments. Established And expect Alaska to keep functioning, because it is hypothecated to a fund and its dividend is defended as a property right.
• 25 yr: Speculative A national implementation somewhere is plausible at this horizon and would be the first evidence that matters more than the existing trials combined, because it would produce published fiscal accounts and general-equilibrium outcomes. Speculative The most likely route is not a new universal payment but the incremental convergence of existing tapered transfers into something closer to a guaranteed income — which would answer the affordability question and leave the universality question exactly where it is. Handwave Claims that automation will force adoption by a stated date belong here and none carries a falsification condition; the last generation of such claims has already passed its horizon.
• 50 yr: Speculative At this range the binding variable is fiscal and demographic rather than experimental: whether a consolidation-financed payment can be sustained alongside the pension and health commitments already legislated in ageing rich countries. Handwave Every published projection at this horizon rests on labour-supply elasticities assumed rather than measured — the Canadian costing puts the behavioural cost under 5% of gross, and the one large modern measurement suggests the response sits at the larger end of the assumed range.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The longest continuous data point in the record is Alaska's forty-three years, which is a single case in a single resource-rich jurisdiction. Handwave Proposals for a global or planetary basic income financed by a common resource rent are coherent as designs and have no implementation, no costing anyone has defended, and no institution capable of levying the rent.

• Depends on Nothing on this map. No brief in this corpus produces a result this one waits on: the instrument has been tested by randomised trial repeatedly and what is missing is a design nobody has funded and machinery nobody has built. No typed depends-on edge is claimed.
• Requires (not on this map) Four constraints, none of them a research result, and the first is the one the debate skips. Universality is an administrative property before it is a fiscal one: paying every resident requires an individual-level register of residence and income, a payment rail reaching people without bank accounts, and a tax integration that reclaims the payment from those who do not need it. Alaska has all three because it built them to pay one dividend a year; most countries have nothing equivalent at individual level, and no pilot has had to build one. The financing package is the second, and the Canadian costing shows why: C$107 billion gross on a nuclear-family definition, C$112.4 billion with behavioural response, for a programme that is guaranteed rather than universal, and financed by eliminating roughly 25 federal and provincial credits plus reductions in the basic personal amount — which creates losers inside the low-income population and is why the model carries a C$7,355 disability supplement. Durability is the third and may dominate the economics entirely: Ontario's three-year, C$150-million pilot was cancelled mid-stream by an incoming government and the legal challenge failed; Finland's statutory experiment was not extended; the only unconditional universal payment to survive four decades is hypothecated to a resource fund. And the fourth is a buyer for the study that would settle the question — a saturation-scale trial, randomised at the level of a labour and housing market rather than a household, permanent or long-announced, at a transfer people could live on, with market prices and rents as outcomes. Nothing in that design is methodologically novel. It has not been run because pilots are funded to demonstrate rather than to discriminate, and no funder has yet paid for the version that could return the wrong answer.
• Enables A durable unconditional payment would change the premises of several briefs on this map rather than unblocking them — the labour-market assumptions behind automation-heavy infrastructure, the political feasibility of transitions with concentrated losers, the household balance sheets that housing and energy programmes assume. No typed enabling edge is claimed, because none of those relationships has been measured and asserting one from a pilot would be exactly the extrapolation this brief spends its length refusing.
• Adjacent Labour economics, which supplies the elasticities; development economics, which supplies the large randomisations; public finance, which supplies the costing; social policy, which supplies the consolidation arithmetic; and within this map Future Public Administration, Digital Citizenship, Institutional Design and Wealth Distribution Systems.

Established “Stockton showed guaranteed income increases employment.” The peer-reviewed analysis does not support this, and the correction matters because the claim is everywhere. The widely repeated +12-point full-time employment figure — 28% to 40% in treatment against 32% to 37% in controls — comes from the pilot's preliminary descriptive report, which describes its own analysis as preliminary and limited to basic descriptive and inferential statistics. The peer-reviewed paper in the Journal of Urban Health states that the employment trend data were not statistically significant, on 110 treated and 198 control observations with 55% and 35% retention, and flags that limited power prevented subgroup analysis and attrition may have impaired detection. Frontier That is a non-significant result on a small sample, not a demonstrated zero — the honest statement is that Stockton did not measure an employment effect either way, and the significant findings it did produce were on income volatility, psychological distress, several SF-36 scales and the ability to meet a $400 emergency.

Established “Finland showed basic income works” — and equally “Finland showed it failed”. The employment effect was about six days on a base of 78; the wellbeing gains were self-reported; and the government's own report states that the 2018 activation model imposed stricter conditions that bit asymmetrically on treatment and control, contaminating the second year in a direction that is hard to sign. Frontier Both readings quote a real number and omit the confound the publishing government stated plainly.

Speculative “Ontario proved people keep working and get healthier.” The pilot was cancelled before it could produce randomised results. What survives is a survey of 217 self-selected former recipients, plus 40 interviews, run in partnership with an anti-poverty advocacy organisation, with no control group. The reported percentages — 80% better health, 83% less stress, three-quarters keeping their jobs — are accurate summaries of what respondents said and are not estimates of the programme's effect. Established The finding Ontario actually delivers is the cancellation itself: the most important variable for whether unconditional provision can be made universal may be political durability rather than labour supply or cost.

Established “The labour-supply objection is a myth.” It is not, and this brief will not round it away. The 1970s experiments found primary earners reducing work by roughly 5%–7.9% on average across the four US trials, with wives up to 27% and single mothers 15%–30%; the 2020s US trial found participation down 4.2 percentage points and hours down 1.3 a week, with income excluding the transfer down about $1,900 a year. Established Fifty years apart, two independent evidence bases agree on direction and rough magnitude. What is contestable is whether that is a cost; what is not contestable is that it happens.

Established “Cash transfers get spent on alcohol and drugs.” Tested repeatedly, refuted every time. Alcohol and tobacco spending rose about $13 a month in the US trial, roughly 1% of a $1,000 transfer; Stockton reported under 1% of tracked spending on tobacco and alcohol; the largest measured US increases were food, rent, transport, and gifts and loans to family and charity, the last up 25%. Frontier This is the rare case where a political objection has been directly measured and has simply failed.

Speculative “Cash transfers break up families.” SIME/DIME found higher divorce rates in treatment groups and the finding is widely credited with ending US basic-income politics in the late 1970s. Mincome later found lower rates. The effect did not reproduce. Frontier No modern trial carries a pre-registered family-structure endpoint, so the claim is unresolved rather than refuted — but a policy buried by a non-replicating result is a fact about the politics of evidence, and it belongs in any account of why this literature looks the way it does.

Speculative “A universal payment would just be absorbed by rents.” Almost no direct evidence exists in either direction. The US trial found rent spending up $52 a month, which is a within-recipient consumption change and not a market price effect — a pilot paying 1,000 households cannot move a housing market. Frontier The objection is coherent, untested, and testable by a saturation design with price outcomes that nobody has funded. Recording it as unevidenced is not the same as dismissing it.

Established “Basic income has been tested.” The instrument has been tested; the proposal has not. None of the tested programmes was simultaneously universal, unconditional, permanent and large enough to live on. Alaska is universal, unconditional and permanent but tiny; Kenya's is bounded by village; Finland's was restricted to the unemployed; the US and Stockton studies were targeted by income and time-limited. Frontier Hoynes and Rothstein's assessment — that the ongoing pilots will do little to resolve the major outstanding questions — is the correct reading of an evidence base that is unusually good at answering questions nobody is arguing about.

Frontier And a misconception on the sceptical side: that the case rests on automation. It does not, and it is weaker when it does. The displacement forecasts that motivate much basic-income advocacy have aged badly — a 47% high-risk share forecast in 2013 against 6% employment growth in high-risk jobs over 2012–2019 — and none of the randomised evidence about how recipients respond to unconditional cash depends on them in any way. Established The measured case for and against this instrument stands on poverty, consumption, health, work and public finance, all of which have been estimated, several of them well.