Future Taxation Models

Tax design is the question of what a state can observe, value and collect — and from whom. The framing under test is that the tax base can be redesigned for a changed economy: a coordinated minimum on corporate profit, new bases for digital activity and data, corrective taxes on carbon and congestion, taxes on wealth, land and automation.

The evidence is unusually good for a policy subject, because tax changes are dated, legislated and scored. What it returns is a two-part answer. The redesign is technically possible and has repeatedly been done. The global minimum tax exists and raised revenue in its first year; carbon pricing raised over $100 billion in 2024 across 80 instruments covering about 28% of global emissions; a congestion charge created a base that did not exist in 2024 and collected over half a billion dollars in twelve months.

What has not been demonstrated is that a redesign survives. The minimum tax was carved out for its largest participant within eighteen months. Canada's digital services tax was rescinded three days before first collection. Net wealth taxes went from about fourteen OECD countries in 1990 to fewer than five by 2015. And British Columbia repealed the most-studied carbon tax in the world on 1 April 2025 — seventeen years after enactment, at a fiscal cost of CA$1.99 billion a year, with three independent studies showing it cut gasoline consumption by 7–11% — for the stated reason that it had become divisive.

And it kept the invisible one. The output-based pricing system for large emitters, which no voter sees, remains in force. That pairing carries the general constraint this brief argues for, and the flag on it should be read carefully: a tax cannot be simultaneously invisible enough to survive and salient enough to change behaviour. That is an inference drawn here from two established findings, not a result anyone has published, and it is flagged speculative wherever it appears. The durability of the tax expenditures that finance long-horizon capital is a related question owned by Civilization Scale Investment and is not covered here.

Established The largest redesign ever attempted has now reported its first year, and — uniquely for this subject — there is an outturn to set against a forecast. The OECD's January 2024 updated impact assessment estimated annual global revenue gains of US$155–192 billion, or 6.5%–8.1% of current global corporate income tax revenues, with total shifted profit falling by about 50%, the share of low-taxed multinational profit falling from 36% to 7% of global profits — about an 80% reduction — after the substance-based income exclusion transition, and effective tax rate differentials between investment hubs and other jurisdictions halving from 14 to 7 percentage points. About two-thirds of the gain comes from top-up taxes and one-third indirectly from reduced profit shifting. Those figures reach this brief through a Big Four firm that sells compliance services for the regime, so mark the interest and note that the numbers are the OECD's.

Frontier The outturn is a fraction of the forecast, and the direction of the interest is what makes it citable. The OECD has published a 2026 economic impact assessment presenting preliminary evidence from the first year of implementation. The figures circulating from it, via a think tank with a standing position against taxation, are €79–109 billion — $91.2–125.8bn — raised in 2024, described as about a third of what was projected in 2023, and 2.4%–3.4% of global corporate revenue against 6.5%–8.1% projected. Established Mark the interest carefully, in both directions. The reporting institute has an interest in reporting that a tax raises less than promised; but the source it reports is the body that made the original projection publishing a downward revision against its own forecast. Interest running against the finding raises its weight: this is a forecasting institution reporting its own shortfall. The prudent treatment is to flag the specific numbers frontier and the direction established, and to say that the OECD document should be consulted before the figures are treated as settled. Frontier The institute's own argument — that early models were optimistic by construction and that future revenue would require directly taxing the investment and jobs the substance-based exclusion currently exempts — is advocacy, and the first two points are substantive enough to test rather than dismiss. Handwave Its claim that compliance costs run to the billions is asserted without a figure.

Established And then the coalition that produced the minimum tax exempted its largest member. On 28 June 2025 the G7, under Canada's presidency, issued a statement proposing a side-by-side system under which the income inclusion rule and the undertaxed profits rule would no longer apply to US-parented multinationals — with qualified domestic minimum top-up taxes still applying, and the exemption limited to US ultimate parent entities, so that US subsidiaries of, say, Canadian groups do not benefit. Simultaneously Canada rescinded its Digital Services Tax Act, halting a collection scheduled for 30 June 2025; and the proposed retaliatory measure under a new Internal Revenue Code section was omitted from the US Senate's substitute amendment released the same day, contingent on reciprocal commitments. Frontier This is the central fact about tax-base redesign and it is not primarily an economic fact. A minimum tax whose purpose was to remove the option of not paying it now contains an exemption for the jurisdiction with the most affected multinationals, obtained in exchange for the withdrawal of a threatened retaliatory tax. The binding constraint on international tax-base redesign is demonstrated to be the veto of the largest jurisdiction, not the design of the rules.

Established Digital services taxes had real, forecastable revenue and were abandoned for a trade reason. Canada's Parliamentary Budget Officer estimated the tax would raise CA$7.2 billion over five years, at 3% of Canadian digital services revenues above CA$20 million for groups with worldwide revenue of €750 million or more, covering online marketplaces, online advertising, social media and user data services; because the tax is deductible against taxable income, the effective rate is 2.59%. The officer flagged a high degree of uncertainty from limited financial disclosure and noted that businesses would likely adjust their services and prices in response. Three days before first collection, the tax was rescinded.

Handwave And here this brief carries a gap on the page rather than papering over it. The incidence question is the interesting one and no verified pass-through estimate is held. The Parliamentary Budget Officer explicitly did not address ultimate incidence. The relevant academic work — an incidence study of digital service taxes existing as a CESifo working paper — could not be retrieved for this brief, so no pass-through number is asserted, and the widespread claim that digital services taxes are borne by sellers and consumers rather than by platforms is flagged handwave as it stands here, pending a fetched source. Speculative The structural argument for pass-through is nonetheless worth stating: a tax on gross revenue from a platform with a near-monopoly position in a national market, applied to a subset of firms above a threshold, has no obvious reason to be borne by shareholders. That is a theoretical expectation, not a measurement, and treating it as one is the most common error in commentary on this instrument.

Established Carbon taxation has the strongest behavioural evidence in the whole of tax policy. Work on British Columbia's carbon tax finds gasoline consumption reduced by about 10.7%, or 6.7% once cross-border purchasing is accounted for. Disaggregated: in Vancouver, a one-cent increase in the carbon tax produces a 2.4% decrease in gasoline consumption; in mid-sized cities about 2%; in rural and northern areas, no statistically significant response. Earlier provincial-level work found 11.3% and 7%, and cross-border leakage tests leave the estimate robust at 1.3% reduction per cent of tax. Three independent studies converging on a 7–11% range is as strong as behavioural tax evidence gets. The reporting institute advocates carbon pricing and its framing is advocacy; the underlying estimates are academic.

Frontier The salience finding inside that evidence is the substantive one. The elasticity with respect to the carbon tax substantially exceeds the elasticity with respect to an equivalent market price change — consumers respond more to a visible tax than to an invisible price. That is a design finding with wide implications: the effectiveness of a corrective tax depends on its salience, which means the politically easiest design — hidden, upstream, collected from firms — is the least effective one. Established And the tax was repealed anyway. British Columbia eliminated the consumer carbon tax on 1 April 2025, taking it from roughly 17 cents per litre of fuel and 15 cents per cubic metre of natural gas to zero, at an estimated fiscal cost of CA$1.99 billion in the coming fiscal year, with the climate action tax credit that offset its regressivity also cancelled. The output-based system for large emitters remains. The Premier's stated reason was that the tax had become divisive at a time when unity was needed. The world's most-studied, best-evidenced, revenue-neutral carbon tax — the case that appears in every textbook — was repealed seventeen years after enactment for reasons of political salience, and the invisible half of the same policy survived.

Established Globally, carbon pricing is nonetheless the largest genuinely new revenue base, and it is small. The World Bank reports carbon pricing revenues exceeding $100 billion in 2024, from 80 instruments in operation — a net increase of five over the year — covering approximately 28% of global greenhouse gas emissions, with over half the revenue allocated to environment, infrastructure and development. Established $100 billion is roughly 0.1% of global GDP. It is real, it is growing, and it is small.

Established Wealth taxation has an administrative record and it is mostly a record of repeal. Individual net wealth taxes fell from about 14 OECD countries in 1990 to fewer than 5 by 2015, raising generally less than 1% of total tax revenue — with Switzerland the exception at around 4% — on narrow bases with widely varying design. Frontier The behavioural evidence is dominated by mobility, and the cleanest identification is a natural experiment at a scale that does not settle the question. Iacono and Smedsvik exploit the Norwegian municipality of Bø cutting its marginal wealth tax rate from 0.85% to 0.35% from 2021, finding a 66.6% increase in average taxable wealth per percentage point of rate reduction — 71.6% among wealth taxpayers, against about 43% in comparable Swiss work — a 10.3% increase in the weighted mass of wealth taxpayers per percentage point, and, decisively, the share of total wealth owned by movers into the municipality rising from about 1% before the reform to 71% in 2020 and about 20% in 2021. Municipal income and wealth tax revenue rose 51.68% from 2020 to 2021. Established The critical caveat, which interested secondary reporting tends to drop: this is intra-national mobility between Norwegian municipalities, which is far cheaper than emigration. It establishes that wealth is highly mobile across a low-friction border. It does not establish the elasticity of wealth to a national wealth tax, and using it to argue against one is an extrapolation — which is precisely what the anti-wealth-tax summary of the same paper does.

Frontier The counter-position has a concrete proposal and a number. Zucman's blueprint for Brazil's G20 presidency proposes a 2% minimum annual tax on the wealth of about 3,000 billionaires, structured as a top-up so that only those paying less than 2% of wealth in income tax would owe anything, estimated to raise US$200–250 billion a year, on the premise that effective tax rates converge toward zero at the top of the wealth distribution in some European countries. It is a serious costed proposal from an economist with a declared advocacy position, commissioned by a G20 presidency. Frontier The honest position on wealth taxation: the theory is clean, the administrative record is poor, the best identification is at a scale that does not settle the question, and the two leading positions are held by researchers with declared views on opposite sides.

Established Meanwhile the workhorse raises a fifth of everything and nobody argues about it. The OECD's consumption tax series reports VAT generating 20.8% of total tax revenue in OECD countries in 2022, consumption tax revenue stable at 9.9% of GDP, and the average standard VAT rate rising slightly to 19.3% in 2024, with consumption-tax-to-GDP ratios falling in 12 of 38 members over 2020–2022. Those figures reach this brief through a trade publication reporting the OECD report. Established That one-fifth of all OECD tax revenue comes from an instrument generating almost no policy debate is the most under-discussed fact in tax design. Every proposal in this brief competes for a share of revenue that is a small fraction of what VAT already raises without controversy. Speculative And the reason is the constraint this brief argues for. VAT is broad, mechanical, hard to avoid, collected by businesses rather than assessed on persons, and largely invisible at the point of payment — and every property that makes it politically durable is a property the carbon-tax salience finding identifies as behaviourally weak. A tax cannot be simultaneously invisible enough to survive and salient enough to change behaviour. That is an inference drawn here from two established findings; it is not a result anyone has published, and it is flagged accordingly wherever it recurs.

Frontier Land value taxation has the strongest theoretical case in public finance, and the first good identification splits it. The theory is that a tax on unimproved land value is non-distortionary because the supply of land is fixed, and that it is fully capitalised into land prices and borne by landowners. Nielsson, Wroblewski and Yding exploit Denmark's 2007 municipal reform, which forced merging municipalities with different land tax rates onto unified rates set by a government formula, creating persistent variation between neighbouring towns. They find a precise zero effect of land taxes on residential home prices, with 95% confidence intervals ruling out full capitalisation at discount rates below 8% against contemporary estimates of 1.9–5.3%; that the burden of land taxes is shared with tenants and future purchasers, contrary to the standard incidence prediction, which — with over 60% of individuals in lower income deciles being non-property owners — makes land taxes significantly more regressive than standard models imply; and null effects on housing development, homeownership rates and residential mobility, so limited efficiency consequences, though older homeowners did sort away from high-tax areas. Frontier The pattern is the recurring one: the instrument with the best theoretical case has the thinnest implementation record, and the first serious identification keeps the efficiency claim and rejects the incidence claim. That is more interesting than either confirmation or rejection, and it is very recent, so treat it as one study rather than as a literature.

Frontier Robot and automation taxes are widely proposed, essentially untried, and the one quasi-experiment is instructive. Acemoglu, Manera and Restrepo document that the US tax code taxes labour at 25.5% — up to 33.5% including implicit taxes from means-tested programmes — against equipment and software capital at 10% in the 2010s before the 2017 reform and 5% after, down from about 20% in 2000, with roughly half the decline attributable to more generous depreciation allowances. They estimate that moving to an optimal tax system would reduce automated tasks by 4.1% and raise employment by 4.02%; that automation taxes alone would raise employment by 1.14% at an optimal automation tax of 10.15%; and that a combined reform of lower capital taxes plus automation taxes raises employment by 1.59%. It is a model-based welfare calculation with clearly stated parameters in a peer-reviewed policy venue. Frontier The one real-world quasi-experiment cuts the other way on the outcome that matters. South Korea in 2018 reduced its tax credit for automation from 7% to 3% for large firms — the closest thing to a robot tax any country has enacted. Affected industries reduced robot installations by 28% relative to their Japanese counterparts, while the employment effect was statistically insignificant on average, with only financially unconstrained firms shifting from automation to hiring. Established Mark the interest, because it is the most informative thing in this section. The organisation reporting it argues that taxing robots stifles innovation and discourages adoption of a productivity-enhancing technology — an interested party opposing robot taxes, reporting a study showing the policy worked on its direct margin and failed on its purpose. The mechanism works and the objective is not achieved, reported by an opponent, which raises its weight.

Frontier And one new base has a measured first year that matches intent. For a subject overwhelmingly composed of proposals, New York's congestion charge is a rare case with an outturn: 27 million fewer vehicles entering the zone, an 11% traffic reduction, about 73,000 fewer vehicles daily; travel time improvements of 51%, 24.7% and 29.3% at three crossings with morning-rush crossing speeds up 23%; $518 million collected through November 2025 with year-end projections above $550 million net; transit ridership up 7%; particulate matter down 22%, crashes down 7%, injuries down 8%, zone traffic fatalities down 40% through July; and a set of economic indicators — office leasing up 9.2%, city sales tax receipts up 6.3%, zone foot traffic up 3.4% — that have no counterfactual at all. Frontier Every figure is self-reported by the authority that runs the policy and receives its revenue. The traffic and revenue counts are mechanical and hard to misstate; the economic attributions should be discounted heavily. What matters for this brief is the revenue: roughly $550 million a year from a tax base that did not exist in 2024, in one city, with an elasticity of about 11% at the introduced price.

Frontier “Coordinated international minimum taxation can end the race to the bottom.” The official hypothesis, held by the OECD Inclusive Framework and most finance ministries. For: 2024 revenue of the order of €79–109bn is not nothing, and the structural estimate — low-taxed profit falling from 36% to 7% of global profits — is the mechanism working. Against: first-year revenue at roughly a third of forecast, and a G7 carve-out for US-parented groups within eighteen months of first application. What would settle it: five years of outturn data by jurisdiction, published.

Frontier “The binding constraint on tax-base redesign is jurisdictional veto, not economics.” This is the best-supported hypothesis in the brief. Evidence: a side-by-side agreement disapplying two of the regime's rules to the largest participant; a CA$7.2 billion tax rescinded three days before collection under threat of a retaliatory measure; and a projection whose decline tracks assumptions about universal adoption. What would settle it: a case where a large jurisdiction is bound against its will. There is none in this record.

Handwave “Digital services taxes are borne by consumers and sellers, not platforms.” The near-universal claim in industry and much policy commentary. This brief could not verify a pass-through estimate from a fetched source and therefore records the claim as unsupported-by-this-evidence rather than as false. The theoretical expectation favours pass-through. What would settle it: the existing incidence work, which exists as a working paper and should be obtained. Stating the gap is more useful than repeating the claim with a hedge.

Established “Carbon taxes work behaviourally.” Three independent British Columbia studies at 7–11% consumption reduction, with a rural non-response indicating the mechanism is substitution rather than income. Frontier “And salience is why.” The carbon-tax elasticity exceeds the market-price elasticity, which is a mechanism claim resting on one jurisdiction. What would settle it: replication of the salience result in another jurisdiction with a comparable natural experiment. Speculative “Corrective taxes are therefore politically self-limiting: the salience that makes them effective makes them repealable.” British Columbia repealed the best-evidenced carbon tax in the world for stated reasons of divisiveness while retaining the invisible output-based system, and VAT's durability at 20.8% of OECD revenue with negligible controversy is the mirror image. Two well-documented cases and a coherent mechanism do not make a tested proposition, and what would settle it — a base rate of repeal for salient against invisible taxes — has never been computed.

Frontier “Wealth taxes are administratively infeasible.” Held by most finance ministries by revealed preference and by the anti-tax literature explicitly. For: the repeal base rate, sub-1%-of-revenue yields, and the Bø mobility elasticities. Against: Switzerland raises around 4% of total tax revenue from one, continuously — a standing counterexample the “infeasible” claim has to explain and generally does not. What would settle it: a modern wealth tax with third-party asset reporting and mark-to-market valuation, which no country has tried at national scale. Speculative “A coordinated billionaire minimum would raise $200–250bn a year and is limited only by political will.” The arithmetic of 3,000 individuals and a 2% floor is straightforward. Against it: the global minimum corporate tax — a far easier coordination problem with an existing institutional home — delivered a third of its forecast in year one and then exempted its largest member. The best available evidence about the feasibility of a billionaire tax is the outturn of the corporate one.

Frontier “Land value taxation is the theoretically optimal tax.” The Georgist position, with two centuries of theoretical support and an implementation record confined to a few municipalities and partial systems. The first well-identified test finds zero price capitalisation and burden shared with tenants, which, if it replicates, removes the incidence half of the argument while leaving the efficiency half intact. What would settle it: replication in another jurisdiction with an exogenous rate change. The theory has just taken its first serious empirical hit and the field has not absorbed it.

Frontier “The tax base must shift from labour to capital because automation is eroding the labour base.” Held across the future-of-work literature, with a documented 25.5%-against-5% wedge and an estimate that correcting it raises employment 4.02%. Against: a 28% reduction in robot installations produced no significant average employment gain. Frontier “A robot tax is the wrong instrument even if the diagnosis is right; the answer is to stop subsidising capital rather than to tax it.” This is the authors' own preferred reading — their combined reform outperforms automation taxes alone — and it is a distinct hypothesis usually collapsed into the previous one. What would settle either: a national-scale automation tax with employment outcomes tracked. None exists.

Speculative “The genuinely new tax bases are corrective and local, not redistributive and global.” The pattern across the successful cases here: carbon pricing at over $100bn from 80 instruments, a congestion charge at about $550m in year one with an 11% traffic response — against a global minimum tax that delivered a third of forecast and a billionaire tax that exists as a proposal. The hypothesis is that tax-base redesign succeeds where the base is physically located, the behaviour is observable, and no other jurisdiction can be moved to. It is a pattern, not a tested claim. Speculative “Taxing data or attention is the coming base.” Widely proposed — data taxes, attention taxes, bit taxes — and, as far as this brief can establish, enacted nowhere. The digital services tax is the closest approach and it taxes revenue, not data. The valuation problem is unsolved: nobody has proposed a workable base definition for a quantity that is non-rival, zero-marginal-cost and priced only in bundles.

Established The first bottleneck is that a tax base is only as good as what the state can observe, and the newest proposed bases are the least observable. VAT works because every transaction leaves an invoice trail with a counterparty who wants the deduction. A wealth tax requires valuing unlisted assets held by people who can move them; a data tax requires pricing a non-rival good that is never sold separately; a land value tax requires separating unimproved land value from improvements. The ranking of instruments by theoretical elegance is close to the inverse of the ranking by observability, and that is not a coincidence.

Frontier The second is mobility, and the best available number measures the wrong border. A 66.6% increase in taxable wealth per percentage point of rate cut, with movers' share of wealth going from about 1% to 71% in a single year, is a spectacular elasticity — and it is measured across a municipal boundary inside one country. An internal move requires no change of language, employer, citizenship or school, so the estimate is an upper bound of enormous width on the national question, and both sides of the wealth-tax debate use it as though it were the national number.

Handwave The third is incidence, and this brief carries an explicit hole in it. No verified pass-through estimate for a digital services tax is held here. That means the most widely repeated claim about the instrument is, in this brief, an unsupported one, and the correct response is to name the missing paper rather than to hedge the claim. Frontier The land value tax case shows why it matters: the standard incidence prediction failed its first good test, with the burden shared with tenants and future purchasers, and over 60% of individuals in lower income deciles being non-owners. An instrument's incidence is an empirical question that theory has repeatedly got wrong.

Established The fourth is that the coordination technology does not exist. The minimum tax is the most sophisticated attempt ever made and it contains a carve-out obtained by threat within eighteen months. Frontier There is no mechanism in the record by which a large jurisdiction is bound against its will, which means every cross-border base is priced with an option to exit held by the largest participant — and that option has a value, which nobody computes when forecasting revenue.

Speculative And the fifth is the salience trade-off, which is this brief's own inference and is flagged as such. The evidence that a visible tax changes behaviour more than an equivalent invisible price is from one province. The evidence that visible taxes get repealed while invisible ones persist is from one province and one instrument class. Put together they suggest a design constraint with no escape route, and nobody has tested it as a proposition — which is exactly why it is offered here as a hypothesis with a flag rather than as a finding.

Established Nothing on this map produces a result this brief waits on. What it waits on is legislation, administration and one missing paper — none of which is a discovery. The typed requirements below record the two institutional constraints that actually bind: a base a successor legislature cannot repeal by ordinary statute, and the third-party reporting and valuation machinery a modern wealth tax would need.

Handwave One dependency is a specific document. The incidence of digital services taxes is the single most-asserted and least-verified claim in this subject, and the relevant academic work exists as a working paper that could not be retrieved for this brief. Until it is obtained, every statement here about who bears a digital services tax is a theoretical expectation and is labelled one.

Frontier Two seams run to briefs that exist and are cross-referenced rather than restated. The durability of tax expenditures financing long-horizon capital — a twenty-year credit programme repealed inside three — is evidence about capital commitment and belongs to Civilization Scale Investment; this brief does not cover it. And carried interest is the obvious seam with Future Capital Markets, which owns the fee as a claim on returns while this brief would own the rate at which it is taxed — and this brief has no fetched source on carried-interest tax treatment and therefore asserts nothing about it.

Established The highest-value experiment is publication rather than policy: five years of global minimum tax outturn by jurisdiction. The first year came in at roughly a third of forecast and the only public figures reach this brief through an interested secondary source. Jurisdiction-level outturn against jurisdiction-level projection would identify whether the shortfall is the substance-based exclusion, the transition rules, or adoption gaps — three explanations with completely different implications. The data exist inside revenue authorities.

Frontier Second: replicate the salience result outside British Columbia. The claim that the elasticity with respect to a visible carbon tax exceeds the elasticity with respect to an equivalent market price change rests on one province, and it is the mechanism on which this brief's central inference depends. A jurisdiction with a comparable natural experiment — a tax change and a price change of similar magnitude in the same period — would either generalise the finding or confine it. Speculative If it confines it, the salience trade-off collapses and the case for invisible corrective taxes improves considerably.

Frontier Third: replicate the Danish land value tax identification. One study finding a precise zero price effect and burden shared with tenants overturns two centuries of theoretical consensus on incidence, and one study should not. Any jurisdiction with an exogenous land tax rate change — a boundary reform, a formula-driven reassessment — supplies the design. This is the cheapest available test of the most theoretically celebrated instrument in public finance.

Frontier Fourth: run an automation tax at national scale with employment tracked prospectively. The evidence is a model saying the correction raises employment by up to 4% and a quasi-experiment saying a 28% fall in installations produced no significant average employment effect. Those two are in tension and neither is decisive. Speculative The informative design would separate the two hypotheses that are usually merged: an automation tax alone against a combined reform lowering capital taxes while taxing automation, which is what the model's own authors prefer.

Speculative Fifth, and the one that would settle the most: build a repeal base rate. Every tax instrument in this brief has an enactment date and, in several cases, a repeal date. Coding a large sample of tax instruments by salience — assessed on persons or collected from firms, itemised at point of sale or embedded in price — against survival time would convert this brief's central inference into a measured hazard rate. Handwave Nothing in the literature reviewed here does it, and until somebody does, the salience-durability trade-off remains a well-organised pattern rather than a result.

Established The mechanics of the minimum tax are what determine its yield, and they are worth stating plainly. The regime combines an income inclusion rule, an undertaxed profits backstop and qualified domestic minimum top-up taxes, with a substance-based income exclusion that carves out a return on tangible assets and payroll. Two-thirds of the projected gain comes from top-up taxes and one-third from reduced profit shifting — so a shortfall could mean less shifting than expected, more substance carved out than expected, or fewer jurisdictions applying the rules. Frontier The side-by-side arrangement disapplies the first two rules to US-parented groups while leaving domestic top-up taxes in place, which preserves the floor inside each jurisdiction and removes the cross-border enforcement that made it a minimum rather than a menu.

Established A digital services tax is engineered as a gross-revenue levy with two thresholds, and its arithmetic is small and precise. Three per cent of in-country digital services revenue above CA$20 million, for groups above €750 million worldwide, deductible against taxable income so that the effective rate is 2.59%, over online marketplaces, advertising, social media and user data. Speculative A gross-revenue base is chosen because profit is unobservable where it matters and revenue is not — which is a confession about the observability bottleneck rather than a design preference, and it is why the instrument is legally fragile and diplomatically expensive relative to what it raises.

Established The carbon design has two halves and they behave differently, which is the whole finding. A consumer carbon tax at roughly 17 cents per litre of fuel and 15 cents per cubic metre of gas is visible at every purchase and produced a measured 2.4% consumption reduction per cent of tax in the largest city, about 2% in mid-sized cities and no significant response in rural areas. An output-based pricing system for large emitters is settled between a regulator and a firm and is visible to nobody. The first was repealed at a fiscal cost of CA$1.99 billion a year; the second survived. Frontier The rural non-response is the mechanistic detail worth keeping: where substitution is unavailable, a corrective tax collects revenue and changes nothing, which makes it a regressive levy on the people least able to respond — and cancelling the offsetting credit alongside the tax removed the compensation for exactly that group.

Frontier A wealth tax's engineering problem is valuation, and the record is a valuation record. Sub-1%-of-revenue yields on narrow bases across the countries that kept one, against Switzerland at around 4%, suggests the difference is administrative rather than conceptual. Speculative The untried design is third-party asset reporting with mark-to-market valuation, which is how securities are already reported for income tax in several jurisdictions and would extend the same machinery to a stock rather than a flow. No country has tried it at national scale, so its yield and its avoidance response are both unknown.

Frontier And congestion pricing is the cleanest piece of engineering in the brief because its base is physically fenced. A charge on entry to a defined zone, collected by gantry, with an elasticity of about 11% at the introduced price, producing roughly $550 million a year and measured co-benefits — travel times up to 51% better at one crossing, particulates down 22%, zone fatalities down 40% through July. Established Nobody can move a Manhattan street to a lower-tax jurisdiction, which is the property every other instrument in this brief lacks, and it is the property the “corrective and local” hypothesis says is doing the work.

Within this map: Civilization Scale Investment, which owns the durability of tax expenditures financing long-horizon capital and to which the clean-energy credit repeal belongs; Future Capital Markets, which owns the fee where this brief would own the rate; Long-Term Institutions, which owns appraisal guidance and whether a fiscal rule binds; Future Federalism, which owns tax competition between tiers of the same state and supplies the mobility evidence in that setting; Future Public Administration, which owns the collection capacity every instrument here assumes; and Digital Citizenship, where the identity and reporting rails a data tax would need are discussed.

Outside it: public finance and optimal tax theory; international tax law, which supplies the minimum-tax machinery; environmental economics, which supplies the carbon elasticities; urban economics, which supplies the congestion evidence; and the empirical literature on tax salience, which is small relative to its importance and is where this brief's central inference would have to be tested.

Established Nearly every number in this brief comes from an interested party and the direction of the interest is load-bearing in both directions. Running against interest and therefore weighted up: the OECD publishing a downward revision against its own forecast, and an organisation opposing robot taxes reporting that a quasi-robot-tax cut installations by 28%. Running with interest and cited for figures rather than framing: a Big Four firm reporting the regime it sells compliance for, a think tank with standing opposition to taxation reporting a shortfall, a carbon-pricing advocate reporting carbon-price elasticities, a wealth-tax opponent extrapolating a municipal elasticity to a national question, a government press release announcing its own repeal, and a transport authority reporting the success of the policy that funds it.

Frontier The institutional requirement that binds hardest is that a tax base is repealable by ordinary legislation. Seventeen years of evidence, three independent studies and revenue neutrality did not protect the British Columbia carbon tax; a five-year CA$7.2 billion revenue line did not protect a digital services tax against a trade threat with three days' notice. No jurisdiction in this record has entrenched a tax base against ordinary repeal, and every revenue forecast in the subject is therefore conditional on a political variable nobody prices.

Speculative The second is the administrative machinery a modern wealth tax would need and no country has built. Third-party asset reporting and mark-to-market valuation exist for listed securities in several jurisdictions; extending them to unlisted holdings, trusts and non-financial assets is the difference between Switzerland's roughly 4% of revenue and everybody else's sub-1%. Frontier The claim that wealth taxes are infeasible is a claim about that machinery, not about the tax — and the standing counterexample of a country that raises 4% continuously is what the infeasibility argument has to explain and usually does not.

Frontier The third is coordination capacity, and its ceiling has now been demonstrated. The most institutionally supported coordination attempt in the history of tax — an inclusive framework, years of negotiation, model rules, a commentary and an administrative guidance stream — produced a side-by-side carve-out for the largest participant in exchange for the withdrawal of a retaliatory measure. Any proposal requiring more coordination than that one required is proposing something no institution has demonstrated it can do.

Established The distributional facts here are measured and they cut against the instruments' own stories. A land value tax was expected to be borne by landowners and the first good identification finds the burden shared with tenants and future purchasers, with over 60% of individuals in lower income deciles owning no property — which makes it significantly more regressive than standard models imply. A consumer carbon tax produced no statistically significant response in rural and northern areas, meaning those households paid it and did not benefit from the substitution it was designed to induce. Frontier And when the tax was repealed, the climate action credit that offset its regressivity was cancelled with it, which removes the compensation from precisely the group that had been paying without responding.

Speculative The salience trade-off has an uncomfortable ethical reading, and it should be stated because it is the whole point of the constraint. If an invisible tax is durable and a visible one is not, then a state that wants durable revenue is pushed toward instruments its citizens do not notice paying. That is a design pressure toward reduced fiscal transparency, arising not from bad faith but from an observed regularity about repeal. Frontier VAT at 20.8% of OECD tax revenue with negligible controversy is what that equilibrium looks like when it works.

Frontier Third: the veto finding has a distributional shape internationally. A minimum tax negotiated in an inclusive framework of well over a hundred jurisdictions was modified to accommodate one, and a middle-power jurisdiction withdrew a domestic revenue measure under threat of a retaliatory tax. Smaller jurisdictions bear the compliance cost of a regime whose enforcement asymmetry runs against them, and the revenue arithmetic they were shown when they joined assumed universal application.

Speculative And the automation case raises a question the model does not answer. Correcting a tax code that taxes labour at 25.5% and capital at 5% is defensible on efficiency grounds alone, independent of employment. But the one quasi-experiment shows that reducing automation does not by itself produce jobs — only financially unconstrained firms substituted toward hiring — which means an automation tax sold as employment policy may deliver a distributional transfer without the employment outcome it was sold on. Handwave Whether that is a failure or a mislabelling depends on which objective was the real one, and the literature does not distinguish them.

Established The terminal position is a two-part statement and both parts are needed. The tax base can be redesigned: a global minimum tax exists, carbon pricing raises over $100 billion a year from 80 instruments, and a congestion charge created a functioning base from nothing in a single year. What has not been demonstrated is that a redesign survives. The minimum tax was carved out for its largest participant in eighteen months; a digital services tax was rescinded three days before collection; net wealth taxes fell from fourteen OECD countries to fewer than five; and the best-evidenced carbon tax in the world was repealed after seventeen years.

Speculative The pattern in the survivors is the brief's central inference and carries its flag. The instruments that persist — VAT at a fifth of OECD revenue, and the output-based carbon system British Columbia retained — share the property that nobody notices paying them. Which is precisely the property the carbon-tax salience evidence identifies as behaviourally weakest. A tax cannot be simultaneously invisible enough to survive and salient enough to change behaviour. This is an inference from two established findings, not a published result, and it is the single most consequential thing in this brief if it is right.

Frontier The long-run implication for a changing economy is unattractive. If the durable bases are the invisible ones, then a state facing an eroding labour base has two options: extend the invisible bases, which works and is regressive by construction; or build new visible ones, which change behaviour and get repealed. Speculative The escape route, if there is one, is the corrective-and-local pattern — a base that is physically located, whose behaviour is observable, and from which no taxpayer can move — and the only clean example in this brief raises about $550 million in a city of eight million people.

Handwave And the biggest open space is the one with nothing in it. No jurisdiction has designed a tax base for data, attention or the output of automated systems. The proposals recur in every decade, the valuation problem is unsolved for a good that is non-rival, zero-marginal-cost and priced only in bundles, and the recurring single-tax proposals — a single land tax, a flat consumption tax, a financial transactions tax as sole revenue — all do their work by assertion at the same step: the assumption that the base is inelastic at the rate required to replace everything else. Speculative That step is where each of those proposals should be examined, and it is the step none of them examines.

These horizons track legislative calendars, first-collection dates and repeal risk rather than technology:

• 10 yr: Frontier The global minimum tax accumulates enough outturn to settle whether the first year was transition or structure, and the side-by-side arrangement either holds, spreads to other jurisdictions, or collapses — three outcomes with very different implications for every cross-border base. Established British Columbia's repeal produces a clean before-and-after on gasoline consumption in a jurisdiction with seventeen years of pre-period data, which is the best natural experiment this subject will get for a decade and is available to anyone who runs it. Frontier Expect at least one further congestion or road-pricing scheme in a large city, and expect wealth tax proposals to continue outnumbering wealth tax enactments.
• 25 yr: Speculative If the corrective-and-local hypothesis is right, the visible growth in tax design is in physically located bases — road, congestion, land, resource extraction — and the global coordination projects continue to underdeliver against forecast. Speculative A national wealth tax with third-party reporting and mark-to-market valuation is either tried somewhere, in which case the infeasibility claim is finally testable, or it is not and the record continues to be a record of repeal. Frontier The salience trade-off is either measured as a hazard rate or it stays an inference; nothing in the current literature suggests anyone is building the dataset.
• 50 yr: Speculative On the pessimistic reading, an eroding labour base is met by extending the invisible instruments, and effective tax systems become more regressive without any decision being taken to make them so. Speculative On the optimistic reading, a data or automation base is finally defined in a way an administration can assess, which requires solving a valuation problem nobody has yet framed properly. Handwave Both are extrapolations of a pattern across a handful of cases, and neither is a forecast.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The one durable regularity at this horizon is that tax systems have followed what states could observe — hearths, windows, trade at ports, wages at source, invoices — and that each shift followed a change in observability rather than in theory. Handwave That is a historical pattern rather than a base rate, and it says nothing about which of today's proposals will be the one.

• Depends on Nothing on this map. This brief waits on no result another brief produces: what it lacks is legislation, administrative machinery and one unretrieved incidence paper. No typed depends-on edge is claimed. Two seams run to briefs that exist and are cross-referenced rather than restated — Civilization Scale Investment owns the durability of tax expenditures financing long-horizon capital, and Future Capital Markets owns the fee where this brief would own the rate.
• Requires (not on this map) Three constraints, all fiscal machinery and none of them a research result. First, a tax base a successor legislature cannot repeal by ordinary statute. The record is a record of repeal: British Columbia eliminated the world's best-evidenced carbon tax on 1 April 2025 at a fiscal cost of CA$1.99 billion a year, seventeen years after enactment and with three independent studies showing a 7–11% gasoline consumption reduction, for the stated reason that it had become divisive — while retaining the invisible output-based system for large emitters; Canada rescinded a digital services tax scored at CA$7.2 billion over five years three days before first collection, under threat of a retaliatory measure; net wealth taxes fell from about fourteen OECD countries in 1990 to fewer than five by 2015; and a G7 side-by-side statement of 28 June 2025 disapplied the income inclusion and undertaxed profits rules to US-parented groups within eighteen months of first application. Second, third-party asset reporting and mark-to-market valuation for a national wealth tax: the difference between Switzerland's roughly 4% of total tax revenue and everybody else's sub-1% is administrative machinery, and no country has built the modern version at national scale, so the infeasibility claim rests on the record of narrow bases rather than on a tested design. Third, and stated as an explicit gap rather than papered over: a published pass-through estimate for a digital services tax. The near-universal claim that these taxes are borne by sellers and consumers rather than platforms is, in this brief, unverified — the relevant incidence working paper could not be retrieved — so the claim is carried as a theoretical expectation and flagged handwave. Each of the three is a choice somebody could make, and none has been made.
• Enables Every programme on this map that assumes a state can pay for it inherits this brief's finding that a base is repealable by ordinary legislation. No typed enabling edge is claimed, because what would enable a downstream claim is an entrenched revenue line rather than a result this brief could produce, and no jurisdiction in this record has one.
• Adjacent Public finance and optimal tax theory; international tax law; environmental economics, which supplies the carbon elasticities; urban economics, which supplies the congestion outturn; the empirical literature on tax salience, which is where this brief's central inference would have to be tested; and within this map Future Federalism, Future Public Administration and Long-Term Institutions.

Established “The global minimum tax has raised its forecast revenue.” First-year figures are reported at roughly a third of the 2023 projection — 2.4–3.4% of global corporate revenue against 6.5–8.1% projected — by the forecasting body itself, reaching this brief through a secondary source that opposes taxation. Frontier The direction is established and the specific numbers are not settled, and the honest handling is to say both rather than to pick whichever suits.

Established “The minimum tax removed the option of not participating.” A G7 side-by-side arrangement disapplies the income inclusion rule and the undertaxed profits rule to US-parented groups, obtained in exchange for the withdrawal of a threatened retaliatory measure, with domestic top-up taxes still applying. The floor survives inside each jurisdiction; the cross-border enforcement that made it a minimum rather than a menu does not.

Handwave “Digital services taxes are passed straight through to consumers.” This is the most confidently repeated claim in the subject and this brief holds no verified pass-through estimate. The theoretical expectation favours pass-through — a gross-revenue levy on a near-monopoly platform, applied above a threshold, has no obvious reason to be borne by shareholders — but a theoretical expectation is not a measurement, and the paper that would settle it exists and was not retrievable here. Speculative Naming the missing source is more useful than restating the claim with a qualifier.

Established “Carbon taxes are politically durable once their effectiveness is demonstrated.” The best-evidenced case in the world was repealed. Three independent studies, a 7–11% consumption reduction, seventeen years of operation and revenue neutrality did not protect it, and the offsetting credit for low-income households was cancelled alongside it. Frontier “And carbon pricing is a large revenue base.” Over $100 billion in 2024 from 80 instruments covering 28% of global emissions is roughly 0.1% of global GDP, against VAT's 20.8% of all OECD tax revenue.

Frontier “Wealth taxes are administratively impossible.” Switzerland raises around 4% of total tax revenue from one, continuously. What the record shows is that most countries stopped, not that none can, and the difference is third-party reporting and valuation machinery rather than an impossibility. Established “And the Norwegian evidence settles it.” It does not: it measures mobility across a near-frictionless internal border, where movers' share of wealth went from about 1% to 71% in a single year. Applying an intra-national elasticity to an emigration question is an extrapolation, and the interested summaries of that paper make it without flagging it.

Frontier “Land value taxes are borne by landowners and capitalised into land prices.” The first well-identified test finds a precise zero price effect, ruling out full capitalisation at discount rates below 8%, and burden shared with tenants and future purchasers. Established The efficiency half of the theory survives — null effects on development, homeownership and mobility — and the incidence half does not. That is one recent study rather than a literature, and it is the only well-identified evidence there is.

Frontier “A robot tax raises employment.” The one quasi-experiment cut robot installations by 28% relative to a foreign control and produced no significant average employment effect, with only financially unconstrained firms substituting toward hiring — reported by an organisation that opposes robot taxes, which is why the installation half is credible and the employment half is worth taking seriously too. Frontier The model evidence pointing the other way is a model, and its own authors prefer the combined reform — lower capital taxes plus automation taxes — over automation taxes alone, which is a distinct proposal usually collapsed into the first.

Speculative “Data or attention is the obvious next tax base.” Proposed in every decade and enacted nowhere. The digital services tax is the closest approach and it taxes revenue, not data. The valuation problem is unsolved: nobody has proposed a workable base definition for a quantity that is non-rival, has zero marginal cost and is priced only in bundles. Handwave “Replace all taxes with a single levy.” The single land tax, the flat consumption tax, the financial transactions tax as sole revenue source and the automated-payment-transaction tax all have real intellectual pedigree and each identifies a genuine defect in the existing system. What they share is the assumption that the base is inelastic at the required rate — and the required rate is the problem, since replacing VAT's fifth of OECD revenue plus income and payroll taxes from any single base implies rates far outside the range where the elasticity evidence was estimated. In every version encountered here, that step is asserted.

Speculative And the claim this brief itself makes deserves the same scepticism. That a tax cannot be simultaneously invisible enough to survive and salient enough to change behaviour is an inference from two established findings, drawn here, and not a published result. It organises a striking amount of the evidence — VAT's durability, the survival of an output-based system alongside the repeal of a consumer tax, the fate of a digital services tax nobody paid at a till — and it has never been tested as a proposition. Handwave If somebody computed a repeal hazard rate by salience and found no relationship, the central claim of this brief would be the thing that had been refuted.