A Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals

Why it matters, in detail. For a fixed-fraction group, the growth of mean wealth equals g(W/Y) + g(Y/H), not g(W/Y) + g(Y). The reported 7.1% is what one obtains by using total real GDP growth (3.0%/yr) instead of real GDP per household (~1.3%/yr), which overstates per-member growth by the ~1.65%/yr growth of membership. Corrected inputs: per-member real growth 5.2-5.4% (world GDP deflator, the report's own convention); implied gross return ~5.8%; net of the current 0.3% individual tax 5.5%, versus the report's 6-7% average pre-tax and 5-6% average net returns. The following claims lose their quantitative support: the pre-tax return differential (p. 18), its 'magnification' by the tax system, and the claim that 36-66% of the 'excess net return' is due to low taxes (fn 10). Under a 2% floor, the billionaire net return would be ~3.8% rather than 5.5%, i.e. below the average net return, not 'comparable' to it (p. 27). The 4.25-fold rise in top-0.0001% wealth relative to GDP is unaffected, so per-member growth still exceeds average wealth growth (~5.3% vs 3.2%). Author's best case: a PPP-based construction like WIR 2022 Table 4.1 (top 1/1 million, 6.9%/yr per adult, 1995-2021, versus 5.8-6.0% from Figure 4 at market exchange rates for the same years) might narrow the gap by about 1 pp. The report neither presents nor uses such a construction, and its comparators are at market exchange rates.

Evidence. Figure 4 label (p. 19): '1987-2024 real growth rates of: Average wealth of the top 0.0001%: 7.1% a year; Average income per adult: 1.3% a year; Average wealth per adult: 3.2% a year'. The label is identical in the original 24 June 2024 PDF and in the author's slides. The text repeats it: p. 10 'the average wealth of the top 0.0001% ... has increased by about 7% a year'; p. 20 'The average growth rate of wealth for this group ... is 7.1%'; p. 18 converts it into a 7.5% gross return. The group is defined as a fraction of households: Figure 4 note '1 household out of 1 million'; p. 19 '(adjusted for the size of the global population)'; fn 11 uses b/0.0001. Adjudicator digitization of the embedded chart (agreeing with auditor A's to within 0.01 pp) gives 3.151% of world GDP in 1987 and 13.403% in 2024 (x4.254). World Bank data (Oct 2026 vintage) give real world GDP x2.988 and adults 15+ x1.835. Growth of aggregate real group wealth = 7.11%/yr, matching the reported 7.1%. Per member: 5.37%/yr (adults 15+), 5.23% (WID adults 20+), and 5.34% using the report's own 1.3% income-per-adult growth. If 7.1% and 1.3% were both per-capita rates, Figure 4's ratio would have had to rise x7.85, not x4.25; reconciling 7.1% as a per-member rate requires zero household growth. Fn 14's nominal '9.2%' equals the nominal growth of aggregate group wealth (9.21% with 2023 GDP used for 2024). Auditor A's reconstruction from Forbes totals follows the stated method (re-run by the adjudicator) and fits Figure 4 with RMSE 0.19 pp. It gives 5.07-5.44%/yr per member and 6.8-7.2% in aggregate; with US-CPI deflation, per-member growth is 4.3-4.7%. The fixed-number alternative needs a Pareto coefficient of 1.25, outside the stated 1.3-1.4, and contradicts the text.

Why this severity. The summary's 'observed' 7.5% return becomes about 5.8% (about 5.0% with US-CPI deflation), a 23-33% reduction. The report's comparison with the 6-7% global average reverses: billionaires' implied pre-tax return is no longer higher than average. The revenue arithmetic (C1-C2) and the low-tax diagnosis (C3-C4) are unaffected.

Why it matters, in detail. The post-reform all-in rate is 22.2% + (1-h)(m - tau0)/rho. With rho = 9% and h = 10%, m = 2% gives 39%, matching the middle class. Against the report's own data (rho of about 7.1-7.4%), d(rate)/d(rho) is steep: 2% gives 43-44% (46% with no haircut), and parity requires m* of about 1.6-1.8%. The non-regressivity benchmark revenue is then 14.2tn x (m* - 0.3%), about $180-210bn with perfect enforcement, versus $242bn. If rho = 9% was obtained by grossing up the inflated R = 7.5% for business-level taxes, the F1 error propagates into Figures 5-7. The 11% ratio for centimillionaires (fn 17) is likewise unsourced; a lower ratio there lowers their current-tax estimate (1.2%) and raises extension revenue, i.e. the error runs in the conservative direction.

Evidence. Appendix B, p. 49: 'a ratio of pre-tax income to wealth of 9% for billionaires and 11% for centimillionaires ... in line with available evidence'. US: 0.678/9.18 = 7.39% (digitized; text values 'about 0.6%' and 'about 8%' give 7.5%). France: 0.122/1.78 = 6.85%, but the two French groups may differ (Fig. 2 is the income-ranked top 0.0002%). fn 7: tau = 0.3%, tau/R = 4%, so R = 7.5%. Figure 5: P50-90 40.0%, P90-95 40.9%. Adjudicator computations (scripts/05): post-reform billionaire rate 42.9% at 7.4% and 43.7% at 7.1% (Appendix B's 10% haircut). The floor equating billionaires with 40% is 1.76% / 1.70% (10% haircut) or 1.62% / 1.56% (no haircut). Revenue at that floor: $179-208bn with perfect enforcement and $144-166bn with a 20% loss. Gross-up test: billionaires' non-individual taxes average 18.6% of pre-tax income in Figs 1-2 (US 13.8, FR 25.6, NL 16.3). 7.5%/(1-0.186) = 9.2% and 5.77%/(1-0.186) = 7.1%.

Why this severity. Leaves the revenue of a 2% floor unchanged. It does undercut the normative calibration ('the rate that makes tax systems non-regressive', 'no higher' than ordinary taxpayers) and the reading of $200-250bn as the loss relative to non-regressivity, which falls by about 15-25% on the report's own implied ratio.

Why it matters, in detail. Appending a narrower tax concept at the top of an all-tax DINA profile biases the measured top-end decline upward by the excluded taxes' incidence on the top group. VAT and pension contributions are negligible for billionaires, but production taxes borne by the firms they control are not shown to be. Income-ranked top groups are proxies for wealth-ranked billionaires. In the US the gap with the bottom deciles is within 0.6-2.6 pp and rests on Saez-Zucman (2019b) top-400 triangulation, which is disputed (corpus us-wealth-taxes-12 to -15); auditor B notes that later administrative work by the same team supports the direction. The generalisation to 'regardless of ... countries' rests on 3-4 countries. None of this reverses C3, but the precise ratios and the universality claim are weaker than presented.

Evidence. Appendix A, p. 47. IPP Note 92 (June 2023), p. 4: 'nous n'incluons pas tous les impôts (ni la TVA, ni les impôts de production), ni de fait toute la distribution du revenu national'; p. 3: economic income EUR 1,278bn = 80% of primary factor incomes, and 'milliardaires' = top 0.0002% of tax households (75) by economic income; p. 5 (box 3): contributory pension and unemployment contributions excluded. Bach et al. (RFBerlin DP 70/25, 2025, section 3.2): about 85% of those 75 households are in the Challenges top 500 and about half in its top 50. Bruil et al. (2025 version): top 0.0001% = 14 individuals, 'roughly equal in size to the group of Dutch billionaires'. GTER 2024 workbook 4-progressivity.xlsx, sheet datafig4.1: French series labelled 'Bozio et al. (2020)', P0-10 46.0%, P90-95 53.8%, billionaires 26% (auditor B; checked by the adjudicator). Figure 1 (digitized; checked visually): France P99.99-99.999 ~38.0%, P99.999-99.9999 ~28.9%, billionaires ~27.4%; US billionaires 23.0%, P30-40 23.6%, P0-10 25.6%. p. 6: regressivity holds 'regardless of the specific tax design choices and enforcement practices of countries'.

Why this severity. The direction of the finding (personal taxes vanish and all-in rates fall at the very top) is well supported by administrative data in France and the Netherlands. However, the headline comparisons overstate harmonization and representativeness. On IPP's own like-for-like concept the French ratio is about 0.57 (26% vs 46% at the top 0.1%) rather than about 0.52, and the effect of excluded production taxes is unquantified.

The authors' response. Zucman, 'Comment taxer les ultrariches ? Cadrer le débat sur l'impôt plancher', Le Grand Continent, 2 Jul 2025: the taxes excluded by the IPP (VAT, pension contributions, residence and property taxes) are tiny relative to billionaires' income, and once reintegrated the average rate returns to about 52% while billionaires' rate barely rises. The reply does not address production taxes. The gap persists under alternative conventions (e.g. 45% net of transfers vs 26%).

Why it matters, in detail. tau(W) in Section 1.3 is defined net of business-level taxes (fn 6). Corporate taxes are about 14-26% of billionaires' pre-tax income (Figs 1-2), i.e. about 1.0-2.3% of wealth at income/wealth ratios of 7-9%. An all-tax capital-tax comparison with the 1.1% average would narrow or reverse the 'quarter' gap, as fn 10 concedes. The design choice not to credit corporate tax is normative (the floor plus corporate tax targets total-rate parity), but it is not disclosed where the headline appears.

Evidence. Summary (vi), p. 2 and p. 7: 'the current effective tax rate of billionaires, equivalent to 0.3% of their wealth'. The slides say 'Billionaire pay around 0.3% of their wealth in tax'. Figure 1: simple mean of US/FR/NL billionaire all-in rates = 22.2% of pre-tax income; times 7.1-9% income/wealth = 1.58-2.00% of wealth. p. 18 and fn 8: 0.3% vs 1.1% including residential property taxes. pp. 8 and 21 on double taxation. p. 20: the minimum counts only individual taxes, and corporate taxes are 'not considered'. fn 10 concedes that counting business-level taxes changes the interpretation.

Why this severity. The 0.3% is correct as defined and is the right baseline for a floor on individual taxes, so revenue is unaffected. But the summary's label contradicts the report's own definition and overstates billionaires' under-taxation by a factor of about 5-7 for readers of the headline. That is an important caveat missing from the summary.

The authors' response. Zucman, Le Grand Continent, 2 Jul 2025, and his 2025 book (as quoted by Allègre & Timbeau 2026), per corpus offshore-havens-18: not crediting corporate tax is deliberate. The 2% floor plus the corporate tax paid through firms brings total rates to roughly the average rate, and corporate tax rates have fallen since 2016.

Why it matters, in detail. The UTPR analogy and the extended-residence rule are plausible but undemonstrated. Their legality under tax treaties, EU free movement and constitutional ability-to-pay doctrine, and their administrative feasibility, are open questions the report flags itself. The open-ended 2% liability on emigrants is a stronger instrument than existing exit taxes and the French bill (5-10 years), which increases legal exposure. The summary's unconditional 'could be enforced successfully' should read as a hypothesis.

Evidence. Summary (ii), p. 2; p. 37 ('for a number of years'); p. 38 (formula; 'minimum standard only' variant: liable 'each year'); p. 39 ('Each scenario would deserve a comprehensive legal analysis that falls outside the scope of this report'); p. 40 ('may require renegotiating double tax treaties'; liability for local establishments). Context, second-hand: a Conseil d'État opinion read in the Assemblée nationale on 31 Oct 2025, on a French variant, reportedly found 10-year post-emigration taxation incompatible with TFEU art. 49 (corpus offshore-havens-21; not published). Avi-Yonah (Tax Notes International, 16 Sep 2024) concludes that a global wealth tax is not feasible (abstract only; corpus offshore-havens-24).

Why this severity. One of the summary's main conclusions rests on a design sketch rather than an analysis, and the revenue estimate presumes enforcement. If the backstop mechanisms proved legally or politically unworkable, revenue would depend on actual participation.

The authors' response. The report itself flags the need for legal analysis and inclusive international discussion (pp. 38-40). On the French variant, Zucman argued that the tax would bring the law into line with constitutional equality before taxation (X, reported by franceinfo, 3 Sep 2025; corpus offshore-havens-21).

Why it matters, in detail. The 25% x 60% / 9 = 1.67% catch-up arithmetic is correct only for a taxpayer whose unrealized gains are 60% of wealth and who does not elect illiquid treatment. Prepayment crediting and refunds make the BMIT partly a timing device rather than a final 2%-of-wealth levy. Under the standard's own requirement, an implementing country using the BMIT would need an explicit top-up, especially for private-company billionaires (about half of global billionaire wealth is unlisted, p. 34).

Evidence. US Treasury, General Explanations of the Administration's FY2025 Revenue Proposals, pp. 83-84: the minimum tax liability is 'treated as a prepayment available to be credited against subsequent taxes on realized capital gains'; refunds are paid 'such as after unrealized loss or charitable gift'; the decedent's estate is refunded; on illiquid taxpayers: 'Taxpayers who are treated as illiquid may elect to include only unrealized gain in tradeable assets ... deferral charge ... would not exceed ten percent of unrealized gains'; valuation of non-tradeable assets as described. Report p. 24: 'In sum, a "billionaire minimum income tax" in the United States would implement a 2%-of-wealth minimum tax standard in the next 9 years, and plausibly so after that.'

Why this severity. Misdescribes a cited source behind the 'flexible implementation' claim for the country holding 40% of billionaire wealth. It does not affect the global revenue arithmetic, which assumes the standard is met.

Why it matters, in detail. Fractile growth = within (incumbent return net of consumption, taxes and gifts) + between (re-ranking) + demography. Reading it as R - tau - c assumes the last two terms are zero. In Gomez's Forbes 400 decomposition they net to about +1 pp/yr, which would lower the implied return; unmodelled philanthropy and dynastic splitting would raise it. The 7.5% (or the corrected ~5.8%) is therefore a model-based proxy with a plausible error band of roughly +/-1 pp, not an observed return. Forward-looking uses (incentives, liquidity, stationarity) also inherit a period effect from 1987-2024 valuation gains.

Evidence. Summary (vi) calls it 'the observed pre-tax rate of return'; p. 18 derives it as growth + 0.3% + 0.1%; fn 18 says charitable giving is not modelled. For the Forbes 400, Gomez ('Decomposing the Growth of Top Wealth Shares', version submitted to Econometrica, introduction) decomposes the 3.9%/yr growth of the wealth share (1983-2017) into within-person growth of 3.0%, a between term of 2.5% (new fortunes displacing old ones) and a demography term of -1.5% (death and population growth). Fractile growth therefore exceeds incumbents' relative growth by about 1 pp a year in that setting. Fewer than 10% of the 1983 Forbes 400 were still on the list thirty years later (Gomez, introduction). The 1987-2024 window also spans a secular fall in real interest rates that raised asset valuations.

Why this severity. Adds unquantified error of either sign to the return figure in C5 and to the cost arguments in C8: composition effects push fractile growth above incumbents' returns, while philanthropy and division among heirs push the other way. It does not affect the revenue estimates.

Why it matters, in detail. Revenue = W x (m - tau0) x (1 - L). W is a market-value rich-list base, tau0 = 0.3% and L is in {0, 0.2}; the range therefore reflects only enforcement uncertainty. Semi-elasticities used in wealth-tax scoring imply L of about 11-29% for a 1.7 pp increment: 8 in Saez-Zucman; 13 in PWBM, phased in over five years; 7-17 for the UK over 4-8 years in Advani-Tarrant. Long-run estimates are much larger (Jakobsen et al. 2026: 0.54 intensive-margin loss). The administrable base for unlisted holdings may fall short of Forbes values, and the figure presumes participation by countries holding about a fifth of billionaire wealth outside the US and Europe. A defensible central first-year estimate is plausibly about $150-210bn, against the printed $193-242bn. Offsetting factors: taxpayer-level heterogeneity (fn 15), the likely decline of tau0 since 2016-18, and the conservative centimillionaire inputs.

Evidence. p. 25: the high end assumes the tax is 'well enforced'; the low end assumes 20% lost, justified as above 'the 15% typically used in the literature (e.g., Saez and Zucman, 2019a)'. p. 27: the estimates 'are for the first year of the tax'. fn 18: charitable giving not modelled. p. 34: about half of billionaire wealth is in private companies. Forbes release (2 Apr 2024): 2,781 billionaires worth $14.2tn; US $5.7tn, China $1.7tn, India $954bn. Behavioural benchmarks: for the 1.7 pp increase in individual taxes, PWBM's semi-elasticity of 13 implies a base loss of about 22% (corpus optimal-tax-behavior-17). Advani & Tarrant's UK elasticities of 7-17 imply 11-25% over 4-8 years (computed by auditor B; corpus optimal-tax-behavior-20). Jakobsen, Kleven, Kolsrud, Landais & Munoz, NBER w32153 (Jan 2026 revision, p. 4 and pp. 36-37) put long-run intensive-margin losses at 0.54 and total losses at 0.76 per mechanically raised dollar, using JJKZ (2020). The Feb 2024 version, which the report itself quotes, already computed that 0.24 of each krona is lost through migration alone (Sweden, top 2%; Section 7.2), mostly through other tax bases. Valuation: Norwegian tax values of unlisted shares have a median of 61% of sale prices (Skar et al. 2025; small firms, net-asset method). Estate-tax values of Forbes decedents are about half the Forbes estimates (Raub et al. 2010, via corpus us-wealth-taxes-11). Sensitivity (adjudicator): a 30% loss gives $169bn; 40% gives $145bn; a current tax rate of 0.6% instead of 0.3% gives $199bn at the high end.

Why this severity. Plausible alternative assumptions (losses of 15-30%, a 0-15% valuation shortfall, less than universal participation) lower the yield by roughly 10-40% below the printed range, but none is demonstrated for this specific coordinated, exemption-free design. Some report assumptions are conservative (fn 15 heterogeneity; 2016-18 tax/wealth ratios applied to 2024), and the order of magnitude survives.

The authors' response. Report: scores the first year only and discloses its assumptions (pp. 25-27, fn 18). On the French variant, Zucman argued that a tax with no exemptions leaves no room for avoidance (France 5, quoted by BFM TV, 16 Sep 2025). Landais, as reported by France 24, said the Danish evidence 'n'a en pratique rien à voir' with this design (corpus offshore-havens-17).

Why it matters, in detail. Liquidity depends on the lower tail of the cash-yield distribution and on borrowing or payment-in-kind options, not on mean total returns. Incentive effects depend on behavioural elasticities and on where taxpayers stand in the return distribution, neither of which is estimated. With the corrected ~5.8% gross return, the floor takes about a third of the average total return. For low-payout owners it takes a larger share of cash income, which raises the stakes of the omitted analysis.

Evidence. p. 21 ('Because the proposed rate (2% ...) is significantly below this rate of return ... liquidity issues should be limited'); p. 8 ('Adverse incentive effects are unlikely to be significant at this level of net-of-tax return'); fn 3 (Amazon and Tesla pay no dividends; Alphabet and Meta have low payout). French Senate report n° 689 (IV.B; corpus offshore-havens-20): AFEP told the rapporteur that more than half of SBF 120 firms pay dividend yields below 3%. The report mentions bank loans but gives no evidence on their cost or availability for this purpose. F1 correction: 5.77% - 2% = 3.77%.

Why this severity. Affects the summary-level assessment of economic costs, which the report presents as a key input to weighing the policy. The concern is real for a subset of taxpayers with high valuations and low cash yields, but no aggregate cost is demonstrated either way.

The authors' response. Zucman (France Digitale debate transcript in Challenges, 17 Sep 2025; corpus offshore-havens-20): taxpayers could pay in shares, and special payment arrangements are needed for rare edge cases that should not become an 'alibi'. The report mentions access to bank loans (p. 21).

Why it matters, in detail. The C2 range ($108-135bn) is robust to the source discrepancy (+6% with WIR's table). It is sensitive to the 11% ratio: the current rate tau0 = 11.3% x rho falls as rho falls, and revenue rises because the floor exceeds tau0. The report correctly labels these estimates as tentative.

Evidence. pp. 25-26 and fn 16-17. WIR 2022, Table 7.1 (p. 139): $100m-1b, 73,710 adults, $16,500bn; above $1bn, 2,750 adults, $13,070bn. Adjudicator re-run of sample3.dta (felixbajard/global-wealth-tax-simulator; iso WO, 2021): above $100m, n = 65,127, w = $28.141tn; above $1bn, w = $12.846tn; (28.141-12.846) x 14.2/12.846 = $16.906tn. fn 17: 'pre-tax income includes some income'. Mean of the P99.99-99.9999 income-tax rates (Fig. 2; Europe = mean of FR and NL; then mean with the US) = 11.27%; times 11% = 1.24%. At 7.5%, the current rate would be about 0.85% and the high-end extension revenue about $194bn.

Why this severity. A citation slip plus an undocumented parameter. The inputs used are reproducible and, if anything, make the centimillionaire estimate cautious.

Why it matters, in detail. Mostly rounding and referencing. The $14.4tn appears only in the figure note and does not enter Table 1 ($3.4bn difference at the high end). The beta misattribution affects only the comparator for average wealth growth.

Evidence. Figure 4 note p. 19 vs p. 24; pp. 15, 31, fn 14; Figures 2-3 (digitized, checked visually); WIR 2022 Chapter 3, p. 75-76 and Figure 3.1. Note on (e): a smaller rise in the wealth/income ratio would lower the 3.2% average-wealth comparator to about 2.2-2.3%, which widens rather than narrows the top-versus-average gap.

Why this severity. Presentation and attribution slips. None changes a conclusion, and (e) runs against the author's interest.

Why it matters, in detail. Figure 4 can be reconstructed only approximately (Forbes totals, World Bank aggregates, an assumed household series). The parameters that drive the non-regressivity calibration (rho = 9%, 11%) and the capital-tax comparison (1.1%) cannot be audited.

Evidence. Searches of the author and EU Tax Observatory sites and of the PDF links by both auditors found no replication files. Appendices A-B (pp. 47-49) give methods only qualitatively.

Why this severity. A documentation problem. It does not itself change a result, but it allowed the F1 error to go undetected and prevents verification of the F4 and F5 parameters.

Why it matters, in detail. The relevant Kesten-type condition concerns E[log((1 + R - tau - c)/(1 + g))] at the top together with demographic dilution, not the sign of R - tau. For billionaires with c of about 0.1%, the empirical conclusion that their wealth outgrows the economy holds (per-member ~5.3% vs ~3% GDP growth). The general claim, and the policy corollary that only tax or return changes can stabilise concentration, do not follow.

Evidence. pp. 16 and 20. Counterexample (auditor B; re-run by the adjudicator, scripts/06): r(W) = 0.01 + 0.02W/(1+W), consumption 5% of W, labour income 1. The map is a contraction (slope 0.967-0.980) and converges to W* = 49.02 from every start, with a non-degenerate stationary distribution under shocks. Benhabib & Bisin (2018), the review the report cites, treats birth/death mechanisms that restore stationarity.

Why this severity. A theoretical overstatement in a motivating passage. No quantitative result depends on it, and the documented four-fold rise of top wealth relative to GDP stands on its own.

Why it matters, in detail. Population adjustment lowers per-member growth but raises the aggregate growth of a fixed-fraction group, because the 1982 group is smaller than 400. The printed 10.6% sits near the aggregate measure, so it is a second instance of the F1 conflation, but with nominal values and unknown inputs it is not confirmed.

Evidence. p. 23 and fn 14. Adjudicator computation (scripts/04, US Census households from FRED TTLHH, Pareto a = 1.3-1.5): the group of 400 in 2023 corresponds to 254 households in 1982. Aggregate growth 10.23-10.36%/yr (Census households) or 10.32-10.48% (tax units growing x1.8); per member 9.02-9.15% or 8.75-8.91%. BMIT equivalent 25% x (g + 0.7%) = 2.73-2.79% (aggregate) vs 2.36-2.46% (per member). Annual Forbes 400 totals were not available, so the exact 10.6% could not be matched.

Why this severity. The conclusion that a 25% broad-income minimum would on average exceed 2% of wealth survives (about 2.4%); only the margin shrinks.