Gabriel Zucman · In O. Blanchard and D. Rodrik (eds.), Combating Inequality: Rethinking Government's Role, MIT Press 2021: 271-276

Can a Wealth Tax Work?

17 problems found: 1 major · 5 moderate · 11 minor

What the paper claims

European wealth taxes failed not because wealth taxation is unworkable but because of three 'distinctive—but preventable—flaws': governments tolerated tax competition (the rich could move abroad; 'The European Union has never put restrictions on tax competition'; 'Tax competition almost single-handedly killed European wealth taxes'), were weak against offshore evasion, and riddled the taxes with 'loopholes, exemptions, and deductions'. In France, Germany and Sweden the taxes were abolished by conservative governments; reading these repeals as proof that 'progressive taxation is inherently unworkable' is 'a superficial reading of European history'.
A US federal wealth tax on the Warren/Sanders model can be shielded from these pitfalls: citizenship-based taxation with exit taxes ('Europe embraces tax competition; the United States refuses it'), FATCA reporting by foreign banks (2010), very high exemption thresholds that weaken liquidity claims and lobbying for exemptions, and enforcement resources ('Tax avoidance is not a law of nature; it is a policy choice'). 'By systematically collecting data on wealth (including from foreign banks), by taxing all assets above a high wealth level at their market value, by refusing tax competition, and by giving the tax authority the resources to enforce the law', such a tax 'could collect sizable sums, increase the progressivity of the US tax system, and curb the rise of wealth concentration'.
In 2018 (after the Tax Cuts and Jobs Act, all taxes at all levels of government) the US tax system 'looks roughly like a flat tax: every social group pays roughly the same tax rate, close to 28%'; 'The only exception is the top 400 richest Americans, who according to these estimates pay 23%'. A wealth tax of 2% above $50 million and 3% above $1 billion would make 'the effective tax rate paid by the ultrawealthy ... double from 23% to 46%'. Other reforms (e.g. a higher top income-tax rate) would not be enough because billionaires realize little taxable income.
There is broad public support for taxing the rich more and for a wealth tax.

Every problem found

1majorconfirmed · previously known · version instability

The chapter's key number, that the 400 richest Americans paid 23% of their income in taxes in 2018, less than anyone else, was an early estimate made before full data existed. When the authors later updated their own figures (in 2020, before the chapter appeared, and again in 2022 and 2026), the same group came out at 27-29%, about what the average American pays. A separate 2025 study by the authors using actual tax records for the 400 wealthiest finds about 24%, so the claim may still hold, but the chapter gives no warning that its number was this uncertain.

Technical detail and evidence

Why it matters, in detail. Like-for-like comparison: DINA top 400 ranked by pre-tax income, pre-tax income plus pure-KG denominator, full population. The top-400 rate minus the all-adult rate is -4.8 pp (2019 vintage), -1.1 (Sept 2020), +0.6 (Feb 2022) and +0.6 (2026 data). Minus the bottom-half rate it is -1.9, +1.8, +3.7 and +4.7 pp. Most of the revision is corporate and business-property tax allocated to the top 400 (corporate component +2.1 to +3.1 pp). It follows upward NIPA revisions of 2018 corporate taxes (TOI's input was $220bn) and the recalibration of top wealth. The fall from the top 0.1% (33.7%) to the top 400 survives in every vintage (4.4-6.5 pp; 8.3 pp in the 2019 vintage). Holding the BPEA wealth-tax increment (+22.8 pp) fixed, a 27-29% base gives about 50-52% with the Warren tax, so a near-doubling survives. The wealth-ranked administrative estimate (23.8% vs 30.2%) has its own convention sensitivities. Per this project's BSYZ adjudication, the gap is 3.2-3.5 pp when social-insurance benefits are counted in income.

Evidence. Authors' files, sheet 'taxrates', 2018 values of taxtop400 / taxall / taxbot50 / taxtop0p1: TOI archive SZ2019/TablesFigures/PSZUpdates/PSZ2019Distributions.xlsx (book vintage) 23.04 / 27.82 / 24.97 / 31.35; PSZ2020AppendixTablesII(Distrib).xlsx (Sept 2020) 27.23 / 28.34 / 25.40 / 33.77; PSZ2022AppendixTablesII(Distrib).xlsx (Feb 2022) 29.31 / 28.74 / 25.65 / 33.72; SaezZucman2026Data.xlsx 29.44 / 28.84 / 24.71 / 34.34 (scripts/vintage_top400_2018.py; output/vintage_top400_2018.txt). The top-400 corporate-tax component goes 5.08 -> 7.21 -> 7.81 -> 8.21 points. The Sept-2020 vintage uses the same denominator file (taxrates.do imports taxratepeinc_kg.xlsx). The PSZ update log ('September 2020') says: 'Distributional series for 2017, 2018 and 2019 are provisional and subject to revision.' Crossref publication date of the chapter: 2021-02-02. TOI online appendix sec. 1.2: 2018 is projected from the aged 2012 PUF. TOI Table A3 copies 2017 values into 2018 for sales, property and payroll taxes. Balkir, Saez, Yagan & Zucman (2025, NBER w34170) p. 1 and Table 2 give 23.8% vs 30.2%. On p. 3 they say the 2019 estimate gives 'similar bottom-line results'.

Why this severity. In the authors' own like-for-like updates the headline level rises by 4.2-6.4 points (18-28%). The gap to the average moves from -4.8 points to -1.1, +0.6 and +0.6, so 'the only exception ... 23%' does not survive in their current series. Their administrative estimate for the wealth-ranked richest keeps a low level, so the claim is uncertain rather than refuted: major, not critical.

The authors' response. The September 2020 update notes flag the 2017-2019 distributional series as 'provisional and subject to revision'. Balkir et al. (2025, p. 3) present their 23.8% as delivering 'similar bottom-line results' to the 2019 estimate. No acknowledgment of the higher 2018 top-400 rates in the 2020-2026 releases was found.

2moderateconfirmed · newly identified here · documentation gap

To get 23%, the authors divide the taxes of the 400 by their income plus part of the stock-market profits they cashed in, although the chart's own axis says 'percentage of pretax income'. Dividing by income alone gives 25.3%, more than four lower-income groups pay, so 'the only exception' depends on an unstated accounting choice. Either way, on these estimates the 400 pay less than average and much less than other very rich groups.

Technical detail and evidence

Why it matters, in detail. Rates on pre-tax income only: top 400 25.29%, against P10-P50 23.5-24.5% and all adults 28.20%. P99.9-99.99 is 36.1% and P99.99-top 400 34.5%. The adjustment binds only in high-realization years and raises top denominators, lowering top rates. It is asymmetric: unrealized housing or pension gains of lower groups are not added. The Warren 'doubling' is robust to it (25.3% -> 50.4% on pre-tax income).

Evidence. Chapter pp. 273-274 (snippets: axis label 'Percentage of pretax income'; caption). TOI SZ2019/Programs/tax-by-gperc.do: line 2; lines 42-45 (pure_kg = max(0, frac-0.03)/frac x kg_agi; peinc_kg = peinc + pure_kg); line 72 (denom = peinc_kg). StataOutput/full2018.xlsx, top-400 row: tax $105.2m, pre-tax income $416.0m, denominator $456.6m, giving 23.04% vs 25.29%. Group rates on both denominators are in output/fig291_rates.txt (scripts/fig291_rates.py). TOI Table A3: pure gains = 2.05% of national income in 2018. BPEA 2019 p. 467. Washington Post chart (25 Oct 2019), note: 'Taxes from all sources, local to federal, divided by pre-tax income'. TOI online appendix sec. 1.1 explains the adjustment.

Why this severity. A defensible but discretionary convention that the chapter does not disclose. It moves the headline rate by 2.3 points (about 10%) and on its own reverses the 'only exception' ordering against four deciles. On either denominator the top 400 remain below the average and far below the top 0.1%.

The authors' response. TOI online appendix (sec. 1.1): taxes on realized gains are in the numerator, so leaving the gains out would make top rates spike in high-realization years. Only gains above 3% of national income are added; the authors attribute gains below that line to retained earnings.

3moderateconfirmed · previously known · methodology

Whether lower-income groups pay 'the same' rate as everyone depends on bookkeeping choices: the authors count tax credits for low-wage workers as government spending rather than tax refunds, count workers' Social Security contributions as taxes but leave them out of income, and drop the lowest-income tenth of adults from the chart. Counting the credits as refunds, as US government agencies do, or counting the contributions in income, puts the bottom half at about 21-23%, below the 400 richest. This does not affect how the very richest compare with the average or with the merely rich.

Technical detail and evidence

Why it matters, in detail. Credits netted (authors' simulator): bottom 50% -3.18 pp; P0-10 -7.2, P10-20 -6.1, P20-30 -4.5, P30-40 -2.4, P40-50 -0.7 pp; top 400 unchanged at 23.04%. The bottom-half minus top-400 gap goes from +1.17 to -2.0 pp. Payroll add-back (public micro-file): bottom 50% -1.66 pp, P0-P50 deciles -1.6 to -1.8 pp, middle 40% -1.88 pp, top 1% -0.22 pp, top-income record -0.01 pp; bottom half about 22.6% vs top 400 23.0% on the published profile. The credit effect is much smaller than Auditor A's '>=17 pp', a 2010 figure for a different population (Splinter). In the opposite direction, using the full population raises the bottom half by +0.76 pp. Corporate-incidence alternatives (25% on labour) lower the top-400 rate by about 1.3-2.5 pp, which works in the chapter's favour.

Evidence. tax-by-gperc.do line 52 (keep if peinc > minwage/2). TOI Table A1, 2018: 243.02m adults, vs 216.69m retained in full2018.xlsx. TOI online appendix p. 5 ('equivalent to winsorizing at roughly the 9th percentile'). taxsimulator2.xlsx (readme: 'revised on 1/20/2020'), sheet calculatorbig: column EM 'Trump (2018)' vs column FW 'Taxes-refundable credits (2018)', weighted by adults x income (scripts/simulator_credits.py; output/simulator_credits.txt). Implied credits: $91.7bn. Full-population bottom-half rate: 24.97% (PSZ2019Distributions 'taxrates'). Auditor B's break-even: $32.4bn of negative taxes for the bottom half would erase its 1.17-point gap over the top 400. Payroll: USdinaonline codebook (peinc is net of plcon, i.e. pension, DI and UI contributions); tax-by-gperc.do line 7 (govcontrib in the numerator); public-file port scripts/payroll_denominator.py -> output/payroll_denominator.txt (bottom 50% 23.36% -> 21.70%). Splinter (2020, NTJ online appendix, Fig. B7; 2010 data): the bottom-quintile rate falls by at least 17 points under agency conventions.

Why this severity. It changes the descriptive 'flat' claim below the top. Under the agency convention for credits, or a consistent payroll denominator, it also reverses the 'only exception' ordering against the bottom half. The top-end comparisons (top 400 vs the average and vs the top 0.1%) are unaffected.

The authors' response. Following national-accounts conventions, the authors treat refundable credits as transfers. The minimum-wage restriction avoids very high, uninformative rates for people living on transfer income (TOI online appendix, p. 5).

4moderateconfirmed · previously known · speculative assumption

The claim that a wealth tax would double the richest Americans' tax rate assumes that only 15% of their wealth would escape it, an assumption the chapter does not mention. With other reasonable assumptions the rate rises to about 41-44% instead of 46%, and to about 37% under the authors' own pessimistic scenario, so it would still rise a lot.

Technical detail and evidence

Why it matters, in detail. Increment = 22.83 pp x (1-h)/0.85. h = 0: 49.9%; 15%: 45.9%; 20.1%: 44.5%; 22.9%: 43.8%; 32.3%: 41.2%; 50%: 36.5%; 89%: 26.0%. A flat haircut understates erosion wherever marginal rates exceed 2%. The figure is a static first-year effect on 2018 wealth. Offsetting this, the income-ranked top 400's wealth/income ratio is about 0.57 times the wealth-ranked group's, so for the actual richest the increment could be 1.5-1.8 times larger.

Evidence. SaezZucman2019BPEAData.xlsx, DataFig5: G17 = 0.23042; P17 = 0.22834 ('Pasted from Appendx Table D1 on July 20 (3% rate above 1bn)'); D17 = C17 + P17 = 0.45875; Q17 = P17 x (1-0.89)/(1-0.15); header D2 '(& 15% avoidance)'. scripts/fig291_rates.py -> output/fig291_rates.txt. Saez-Zucman Warren letter (18 Jan 2019): 15% avoidance. BPEA 2019 Table 2, row 'Weak enforcement: 50% evasion rate'. PWBM (Dec 2019): semi-elasticity -13. BSYZ (2025) p. 11: taxes = 1.3% of wealth at a 23.8% rate. The recovered p. 274 text states the 23%-to-46% result with no avoidance qualifier.

Why this severity. Under defensible alternatives the result is still roughly a doubling (x1.8-1.9). It falls to x1.6 only under the authors' own weak-enforcement case. This is an undisclosed, optimistic assumption with limited demonstrated impact on the qualitative point.

The authors' response. Avoidance is a policy choice. With a comprehensive base, third-party reporting, FATCA, an exit tax and IRS funding, hiding can be held near 15% (Warren letter; BPEA 2019). The authors' 2019 reply to Summers & Sarin rejects the estate-tax analogy as implying implausibly high avoidance.

5moderatelikely · previously known · selective presentation

The chapter suggests Europe's wealth taxes were scrapped mainly by conservatives and because rich people could move abroad, but several were abolished by left-of-centre governments, and international experts also blamed high running costs and economic side-effects. Switzerland's wealth tax, which starts at modest fortunes, has survived and raises a lot of money. The chapter's explanation may be partly right, but it tells only the side of the story that supports its case.

Technical detail and evidence

Why it matters, in detail. Of the nine repeals or suspensions in 1994-2008 listed by the OECD, six took place under governments led by or including centre-left parties. The OECD's reasons include efficiency and cost-yield problems that a high threshold does not obviously solve, such as valuing illiquid assets. The 'single-handedly' claim is unsupported. Low-threshold taxes in Switzerland, Norway and Spain survive despite mobility, and measured responses have large non-migration components (Danish and Swiss evidence).

Evidence. OECD (2018), The Role and Design of Net Wealth Taxes in the OECD: p. 16 (repeal dates: Austria 1994, Denmark 1997, Germany 1997, Netherlands 2001, Finland, Iceland and Luxembourg 2006, Sweden 2007; Spain's 100% credit 2008); p. 17 (reasons for repeal); p. 18 (Swiss revenue explained by low thresholds and broad bases). BPEA 2019 p. 478. Chapter p. 272 (snippet: 'Tax competition almost single-handedly killed European wealth taxes'). Corpus optimal-tax-behavior-19 (Brulhart et al. 2022): mobility is about 24% of the Lucerne-Bern response.

Why this severity. It weakens the chapter's interpretive core (that the failures were due only to preventable choices) by omitting evidence that cuts the other way. It does not refute it: the three flaws are real and overlap with the OECD's capital-flight, narrow-base and evasion reasons.

The authors' response. Taxes 'are bound neither to fail nor to succeed', and Europe's failures reflected policy choices (Saez & Zucman, Washington Post, 25 Oct 2019; BPEA 2019). The three flaws overlap with the OECD's capital-flight and narrow-base reasons.

6moderateplausible · previously known · speculative assumption

The chapter argues that a US wealth tax would work for three reasons: America taxes its citizens wherever they live, it receives data from foreign banks, and the tax would only hit fortunes above $32-50 million. But no country has tried a tax of this design, valuing private companies is hard, and the US estate tax on large fortunes, which also starts only at high wealth, captures only a fraction of what the richest are thought to own. Saying the tax 'could' work is fair, but readers get little sense of how uncertain that is.

Technical detail and evidence

Why it matters, in detail. At a 2% rate, plausible base erosion runs from about 14% to 34% (Advani-Tarrant, 7-17 per point) and much higher under Swiss-type semi-elasticities, against the 15% assumed. PWBM's conventional score was about 28% below the campaign's. The estate-tax analogy implies far larger erosion, but its reconciliation is disputed. Private businesses, about one-third of top-0.1% wealth, have no tested valuation and enforcement design.

Evidence. Chapter pp. 271-273, recovered snippets: 'could be shielded', 'may protect these taxes', 'Tax avoidance is not a law of nature; it is a policy choice', 'could collect sizable sums'. Corpus checklist: 'Elasticity is a policy choice' without costed enforcement. Saez & Zucman op-ed (25 Oct 2019): the IRS 'is not exploiting this information as much as it could'. Kopczuk (2019), BPEA comment. Scheuer & Slemrod (2021), JEP 35(1). Summers & Sarin (Washington Post, Apr 2019). BSYZ (2025) p. 3. BPEA 2019 p. 482: private businesses are 'probably around one-third' of top-0.1% wealth. Corpus optimal-tax-behavior-19 and -20; us-wealth-taxes-17, -18 and -21.

Why this severity. The feasibility thesis is the chapter's answer to its title question. It is hedged ('could', 'may'), and the opposing evidence is itself contested, so this is a robustness concern with limited demonstrated impact rather than a refutation.

The authors' response. Enforcement is a policy choice, and most top-0.1% wealth is listed equity or otherwise third-party reported. Where market values are missing, the authors propose payment in shares or formula valuation (BPEA 2019). The estate tax, in their view, is eroded by specific loopholes that the Warren and Sanders designs avoid (reply to Summers & Sarin, 2019).

7minorconfirmed · newly identified here · selective presentation

The chapter says everyone pays about 28% except the richest 400. On the authors' own chart, several middle- and low-income groups pay nearly as little as the richest 400, and some very rich groups pay noticeably more than 28%.

Technical detail and evidence

Why it matters, in detail. Group rates span 23.0-33.2%; the top 400 are lowest but within 0.5 pp of P30-40 and 1.2 pp of the bottom half on the authors' own conventions.

Evidence. BPEA DataFig5, column G (reproduced in output/fig291_rates.txt); BPEA 2019 p. 467; Auditor B's R3 (P30-40 23.52%, P99.9-99.99 33.15%).

Why this severity. A framing simplification in a short essay; the numbers plotted are correct and the point concerns the top tail.

8minorconfirmed · newly identified here · other

The '400 richest' in the chapter's chart are actually the 400 people with the highest incomes, which is not the same list as the 400 wealthiest. For the truly wealthiest, a wealth tax would raise the tax rate even more than the chapter says, but the chapter's group and the later study's group are not the same people.

Technical detail and evidence

Why it matters, in detail. The income-ranked group's income/wealth ratio is about 1.77 times the wealth-ranked group's, so the Warren increment for the wealth-ranked top 400 would be about +34-41 pp rather than +23 pp. The match between TOI (23.0%) and BSYZ (23.8%) is a coincidence across groups and vintages, not a replication.

Evidence. tax-by-gperc.do lines 54-61 (2018 doubling of the top 400) and 76-84 (top 400 = cum >= total_pop - 400 by peinc); full2018.xlsx top-400 row (avg $416.0m, hweal $4.27bn, nb 514.7). BSYZ (2025) p. 3; p. 11 (taxes = 1.3% of wealth; rate 23.8%).

Why this severity. A labelling and documentation issue. Its main consequence is that the 46% understates, rather than overstates, the wealth tax's effect on the very richest.

9minorconfirmed · newly identified here · data error

The chapter says Germany abolished its wealth tax in 1995. It actually stopped in 1997, after a court ruled that the way it valued property was unfair.

Technical detail and evidence

Why it matters, in detail. Factual slip that also conflicts with the authors' own BPEA text.

Evidence. Chapter p. 271 (snippet); OECD (2018) p. 16; Saez & Zucman (2019) BPEA p. 478.

Why this severity. A two-year dating slip; the omitted cause is folded into F5.

10minorconfirmed · newly identified here · internal inconsistency

The chapter gives two different years for when France ended its wealth tax; both are defensible, but it should use one.

Technical detail and evidence

Why it matters, in detail. No effect on any argument.

Evidence. Chapter pp. 271, 272, 273 (snippets).

Why this severity. Cosmetic.

11minorconfirmed · newly identified here · overclaiming

The EU does limit some kinds of tax competition (for example, minimum sales-tax rates), just not on the personal taxes that matter here.

Technical detail and evidence

Why it matters, in detail. No effect on the conclusion.

Evidence. Chapter p. 272 (snippet); VAT Directive 2006/112/EC art. 97; Council Directive 2003/48/EC.

Why this severity. Overstatement in wording; the narrower true claim supports the chapter's argument equally well.

12minorconfirmed · newly identified here · misrepresentation of source

The chapter cites a study as finding that the richest Scandinavians hid close to 20% of their taxes offshore; the study's main estimate is about 25%.

Technical detail and evidence

Why it matters, in detail. 5 pp below the cited benchmark. The external validity of the Scandinavian concentration is itself questioned by Dutch amnesty evidence (offshore-havens-04).

Evidence. Chapter p. 272 (snippet); AJZ (2019) AER 109(6), abstract p. 2073 and pp. 2093-94; NBER WP 23772 (2017), sensitivity: 3.5% return -> 20%, 6% -> 35%.

Why this severity. Conservative slip (it understates evasion, which the chapter treats as a European flaw).

13minorconfirmed · newly identified here · overclaiming

Before 2018 European tax authorities could already ask Swiss banks for information on specific suspects, though it rarely helped; what started in 2018 was automatic sharing. The chapter's point, that hiding money in Switzerland used to be easy, still holds.

Technical detail and evidence

Why it matters, in detail. No effect on the conclusion. Weak on-request exchange plus withholding left most offshore wealth undeclared before automatic exchange (AJZ 2019; JZ 2014).

Evidence. Chapter p. 272 (snippet: 'fighting wealth concealment in tax havens. Until 2018, there was no exchange of information between banks in Switzerland ...'); Johannesen & Zucman (2014), AEJ: Economic Policy 6(1), p. 65 (2009 France-Switzerland treaty) and p. 87 (Savings Directive withholding and voluntary disclosure since mid-2005).

Why this severity. An overstatement of wording; the argument needs only the absence of automatic exchange.

14minorconfirmed · newly identified here · overclaiming

Moving from Paris to Brussels did not free someone from all French taxes; it freed mainly their financial investments.

Technical detail and evidence

Why it matters, in detail. None for the conclusion.

Evidence. Chapter p. 272 (snippet); OECD (2018) ch. 4, 'Taxed assets, exemptions and reliefs' (non-residents); Sénat, Rapport d'information n° 42 (2019-2020), p. 29, n. 2: for taxpayers domiciled outside France, art. 885 L CGI exempted financial investments only.

Why this severity. Rhetorical overstatement; no effect on the argument.

15minorconfirmed · newly identified here · other

Under Sanders's plan the tax would start at $32 million for couples but $16 million for single people.

Technical detail and evidence

Why it matters, in detail. None.

Evidence. Chapter p. 273 (snippet); Saez & Zucman, Sanders wealth-tax note (22 Sep 2019), p. 1 and table.

Why this severity. Imprecision with negligible effect on the argument.

16minorconfirmed · newly identified here · data error

France did give big tax breaks on business shares, but the 75% break the chapter describes was not limited to small and medium firms.

Technical detail and evidence

Why it matters, in detail. No effect on the argument. The largest ISF relief, the income-based cap (plafonnement, EUR 1.3bn of EUR 2.6bn costed), is not mentioned.

Evidence. Chapter p. 273 (snippet: '... taxed at lower rates, homes that were primary residences enjoyed a 30% deduction, and shares in small or medium-size enterprises benefited from a 75% exemption'). Sénat, commission des finances, Rapport d'information n° 42 (2019-2020), pp. 29-31, table of ISF tax expenditures: 'Exonération de 75% des parts d'entreprises faisant l'objet d'un engagement de stabilisation de l'actionnariat (pacte Dutreil)' (EUR 210m); 'Exonération partielle des titres détenus par les salariés et mandataires sociaux' (EUR 63m); 'Exonération des titres reçus en contrepartie de souscriptions au capital de PME' (EUR 22m); 'Réduction d'impôt de 50% ... jeunes PME (ISF-PME)' (EUR 660m). OECD (2018) p. 83: 30% main-residence allowance in France.

Why this severity. A descriptive inaccuracy in an illustrative list; the point that the base was riddled with reliefs stands.

17minorconfirmed · newly identified here · data error

Most European wealth taxes started at far lower fortunes, often EUR 60,000-250,000, than the '$1 million or so' the chapter says; only France and Spain set the bar that high.

Technical detail and evidence

Why it matters, in detail. Ten of the 12 OECD wealth taxes had single-person thresholds below EUR 500k and eight below EUR 170k, so the statement fits France (and Spain since 2011) only. The low-threshold taxes that survived (Switzerland, Norway) also cut against the lobbying narrative built on it (F5).

Evidence. Chapter p. 273 (snippets: 'In Europe, progressive wealth taxes typically exempt or used to exempt only the first $1 million or so in net wealth. This means that the "merely rich" owners of valuable houses and burgeoning businesses were liable for the tax.'). OECD (2018), The Role and Design of Net Wealth Taxes in the OECD, Table 4.2 (p. 81) and p. 80 ('In 2017, France had the highest exemption threshold').

Why this severity. A factual misstatement that understates how broad European taxes were. It does not flatter the chapter's contrast with $32-50m US thresholds, which is even starker on the true numbers.

Claims examined and rejected

Problems one auditor raised that did not survive checking.

No chapter-level issue can be established at any severity, because the full text was not obtained (0 issues). — The recovered text (most of the running text of pp. 271-274, about 1,670 words, including the whole quantitative passage on pp. 273-274, note 1 and the reference list) is enough to verify F1-F17 against the authors' files and sources. Limited access narrows the audit but does not justify an empty issue list.
Critiques of the flat avoidance haircut and of campaign revenue scores (us-wealth-taxes-17, -20; optimal-tax-behavior-17) should not be transferred to the chapter, which does not assert those numbers. — Rejected in part. The chapter's 23%-to-46% claim is exactly BPEA DataFig5's 'with Warren wealth tax (& 15% avoidance)' series, so the flat-haircut critique applies directly (F4). The campaign revenue totals indeed do not appear in the recovered text, and none is attributed to the chapter here.
Later administrative data (Balkir, Saez, Yagan & Zucman 2025: 23.8%) corroborate the 23% and largely rebut the 'projection/triangulation' critique (us-wealth-taxes-14). — Rejected as stated. BSYZ concerns a wealth-ranked group with a different income concept. The authors' like-for-like updates of the chapter's own series put the 2018 top-400 rate at 27.2% (Sept 2020), 29.3% (Feb 2022) and 29.4% (2026 data), about the average (F1). The projection critique is therefore borne out for the series the chapter used, even though the later administrative estimate supports a low level for the wealthiest.
Counting refundable credits lowers the bottom-quintile rate by at least 17 points (Splinter, 2010), 'enough to turn flat into progressive'. — Magnitude rejected and replaced; the issue survives as F3. For the 2018 profile the chapter plots, the authors' own simulator gives -7.2 (P0-10), -6.1 (P10-20) and -3.2 points (bottom half). The 17-point figure refers to 2010 and to a population that includes the very-low-income adults the chapter's profile drops.
Check passed: the chapter's French examples are more accurate than the op-ed's 'Shares in small or medium-size enterprises got a 75 percent exemption'; whether the chapter repeats the SME claim is unknown. — Superseded: the rest of the p. 273 sentence, now recovered, repeats the op-ed's wording ('shares in small or medium-size enterprises benefited from a 75% exemption'). The inaccuracy Auditor A identified in the op-ed therefore applies to the chapter (F16).
The chapter passage on citizenship-based taxation and exit taxes was not recovered (treated as op-ed/slides content only). — Superseded: the passage is on p. 272 ('... citizenship—a move that triggers exit taxes on unrealized capital gains today and could trigger exit taxes on the stock of wealth itself should a wealth tax exist, making renouncing citizenship costly in practice. Europe embraces tax competition; the United States refuses it.'). It is now part of C2; no issue arises from it.

Audit notes

What reproduced. Reproduced exactly from the authors' own 2019 files: the 23.04% top-400 rate, all Figure 29.1 bars (DataFig5 vs TOI full2018.xlsx, max 0.11 pp), the 45.9% Warren rate, the 28% macro rate (TOI Table A3: 28.30%), the Gallup 62% and the plan thresholds. Sensitivities computed here: pre-tax-income denominator 25.29% (four deciles lower); refundable credits netted in the authors' Jan-2020 simulator puts the bottom half at 21.04% (top 400 23.04%); adding OASDI/DI/UI contributions back to the denominator lowers the bottom half by 1.66 points on the public 2018 micro-file; avoidance alternatives give 41.2-44.5% (36.5% at 50%, 26.0% at 89%). Later vintages of the same series (2018, top 400 / all): 27.23 / 28.34 (Sept 2020), 29.31 / 28.74 (Feb 2022), 29.44 / 28.84 (2026 data). Current NIPA macro rate for 2018: 29.65%. Not reproduced: the Stata DINA pipeline (no Stata; authors' outputs used) and the IRS internal files behind the top-400 calibration, which are not public. The full chapter text was not available; most of the running text (about 1,670 stitched words) was recovered from Google Books snippets.

Limits of this audit. The full chapter text was not obtained: MIT Press Direct and De Gruyter are behind Cloudflare/WAF, and there is no open-access copy. Most of the running text of pp. 271-274 (about 1,670 words stitched from overlapping snippets) was recovered from Google Books search-inside snippets. Pp. 273-274 (the whole quantitative passage and figure labels), note 1 and the reference list are essentially complete. Short gaps remain on pp. 271-272 (a few clauses between recovered passages). Google Books throttled this IP repeatedly. The unrecovered clauses are on pp. 271-272, away from the quantitative passage, so a caveat on avoidance or provisional data there is unlikely but cannot be excluded. Search-inside results are not exhaustive, so a missing hit is weak evidence of absence. The authors' Stata pipeline was not re-run; their published outputs and workbooks were used, and the IRS internal files are not public. The later-vintage 'taxrates' series cover the full adult population and were compared like-for-like with the 2019 vintage of the same sheet. Balkir et al. (2025) was not re-audited here (see its own adjudication). The polls in note 1 were not re-verified. Legal points (French non-resident rules, EU law) were checked against OECD (2018), the French Senate's 2019 ISF report and the directives cited, not against national legal texts. Party affiliations of repealing governments are standard historical facts and were not checked against a primary source.