Taxing across Borders: Tracking Personal Wealth and Corporate Profits

Why it matters, in detail. Writing the worldwide ETR as s*t_h + (1-s)*t_n with s the Figure-3 haven share of the ETR denominator, the rise of s from 5.8% to 15.4% explains 1.9-3.0 pp of the 9.42 pp fall for t_h = 0-3% (at most 3.3 pp under a 35% counterfactual), i.e. about $45-70bn a year at 2013 profit levels, not the 6-8 pp ($140-190bn) implied by 'two-thirds or more'. Most of the decline is a fall in the implied non-haven rate (31% -> 23%), to which falling capital-gains realisations, FTC conventions and portfolio dividends contribute. With the non-duplicated haven share (F2) the composition effect is smaller still. Firm-level evidence (Dyreng et al. 2017) shows similar declines for purely domestic firms, inconsistent with a predominantly international explanation.

Evidence. (i) Paper's own series (workbook DataFig1-5: worldwide ETR col. AY; haven profits = Figure 3 haven share x foreign profit inflows): haven-booked profits rise from 5.8% to 15.4% of the ETR denominator between 1998 and 2013. Taxing them at 0-3% against the implied non-haven rate (31% in 1998, 23% in 2013), the composition effect is 1.9-3.0 pp of the 9.4 pp decline (20-31%); even if every haven dollar would otherwise pay the 35% statutory rate it is 3.1-3.3 pp (32-35%) (adjudicator script etr_decomposition.py; same result as auditor A). (ii) The paper's own level logic (tax bill lower by the haven share of profits, p.130) implies havens lowered the ETR by 2.0 pp in 1998 and 4.2 pp in 2013, i.e. 2.2 pp (24%) of the decline. (iii) Identifiable non-avoidance factors exceed the 2-4 pp allowance: lower taxed corporate capital gains 2.6-4.6 pp (F4), the FTC-only treatment of foreign taxes about 0.6 pp and foreign portfolio dividends in the denominator 0.5 pp (F4), Section 199 0.4 pp; so even as a residual avoidance would be at most about 35-55%. (iv) Taking the paper's 2-4 pp at face value, the residual is 57.5-78.8% of the exact 9.42 pp decline, so 'two-thirds or more' is not even its lower bound (auditor B). (v) External: Dyreng, Hanlon, Maydew and Thornock (2017, Journal of Financial Economics 124(3): 441-463; abstract and NTA 2014 version read) find that US firms' effective tax rates fell 'at approximately the same rate for both multinational and domestic firms' over 1988-2012 and that declining foreign statutory rates 'explain little'. Their data are firm financial statements, not NIPA aggregates.

Why this severity. The abstract's attribution falls from about two-thirds to roughly one-fifth to one-third on direct estimates (at most about one-half as a residual), far more than a quarter of the headline. Avoidance still accounts for roughly 2-3 points of the decline, so the qualitative claim that profit shifting lowered US firms' effective tax rate survives; the claim that it was the main cause does not.

Why it matters, in detail. DII includes equity-method income of directly owned affiliates, so lower-tier income from high-tax countries is booked to the first-tier haven holding. Removing equity income from BEA MOFA data lowers the six-haven share of foreign income from 55-58% to 36.5-38.9% (post-tax, 2011-13) or 29-31% (pre-tax). Haven-booked profits then equal about 10-12% of US corporate profits (paper: 17.4-20%), and the 1999-2013 increase in the haven share is about 8 pp, not 21 pp. The true value lies between BR-type and DII-type measures; Zucman's later co-authored work uses profit-type return (which excludes equity income), and BR report that haven shares under PTR and adjusted pre-tax income are similar.

Evidence. BEA majority-owned foreign affiliate income statements (Table II.D.1) separate own income from 'income from equity investments' in other affiliates. For the paper's six havens with its grouping (Ireland, Luxembourg, Netherlands, Switzerland, Singapore and all of 'Other Western Hemisphere', as in workbook TableS.2), equity income is 75-76% of net income (2011p: $457bn of $612bn; 2013r: $456bn of $601bn). The havens' share of all-country net income is 54.9% (2011p) and 57.6% (2013r), close to the paper's 55.8% DII share, but only 36.5% and 38.9% of own (non-duplicated) net income and 29.4% and 31.3% of own pre-tax income (Bermuda + UK Caribbean grouping: 35.1%/37.8% and 28.1%/30.2%) (adjudicator script bea_mofa_havens.py; auditors A and B report the same figures for their groupings). Blouin & Robinson (2020, Sec. 4.1, Fig. 3) obtain about 35% for 2013. Scaling Figure 3 by 38.9/55.8 gives haven-booked profits of about 12% of all US corporate profits in 2013 (about 10% on a pre-tax basis), against 17.4% computed, 18% in the text and 20% in the abstract. Trend: BR report equity income at 19% of affiliate net income in the 1980s and 65% in the 2010s; in the BEA 1999 benchmark (Table III.E.1) the six havens had 30.5% of own net income versus 40.2% of net income and 35.1% of DII (workbook), so the corrected haven share rose about 8 pp from 1999 to 2013 (30.5% -> 38.9%) against about 21 pp in the DII series (35.1% -> 55.8%) (adjudicator script bea_mofa_1999.py). Against the 1982-89 average DII-based share of all profits (2.9%, when equity income was small), a corrected 2013 level of about 12% is roughly a fourfold rise, not tenfold.

Why this severity. On BEA's own non-duplicated measures the headline level falls by roughly 30-50% (from 17-20% to about 10-12% of all profits) and the 'tenfold' rise to roughly three- to sixfold. Because BR's subtraction can remove some genuine shifting (Clausing 2020), the corrected figure is closer to a lower bound. The qualitative finding, a large and rising haven share, survives.

The authors' response. The paper cautions (p.129) that balance-of-payments data reveal 'mainly the location of the holding companies'. Torslov, Wier & Zucman (REStud 90(3), 2023, Sec. 4.2.1) accept that BEA 'net income' double counts and state that their own later estimates use profit-type return and FATS; the 2014/2015 haven-share figures were not revised. Clausing (2020, 'How Big is Profit Shifting?', SSRN 3503091) argues that subtracting equity income can strip out genuine foreign-to-foreign shifting.

Why it matters, in detail. R = s(1 - t_h/t_n). Paper inputs (s = 0.174, t_h = 0.03, t_n = 0.23) give R = 15% (10-13% with residual US tax on repatriated haven income). Non-duplicated inputs (s = 0.10-0.12, t_h = 0.095-0.106, t_n = 0.23-0.25) give R = 5-8%. The counterfactual that every haven dollar is shifted is an upper bound. The book's $130bn becomes about $80bn under BR's correction.

Evidence. (i) The reduction relative to taxing haven profits like other profits is s*(1 - t_h/t_n), not s. With the paper's own 2013 inputs (s = 17.4%, t_h = 3%, implied non-haven rate t_n about 23%) it is about 15%; counting US tax on the roughly 20% of haven profits that are repatriated (Fig. 4) lowers it to about 10-13%. (ii) With the non-duplicated haven share (F2: about 10-12% of profits) and haven tax rate (F7: 9.5-10.6% on own pre-tax income) it is about 5-8%. (iii) It assumes all haven-booked profit would otherwise be taxed at full rates, although havens host real activity (e.g. 222,000 affiliate employees in the Netherlands, 168,000 in Singapore and 98,000 in Ireland in 2011, workbook TableS.2) and micro estimates imply much less shifting (Dharmapala 2019). (iv) Zucman (2015, The Hidden Wealth of Nations, pp.105-108) restates the result as 'close to 20%, or $130 billion annually'; Blouin & Robinson (2020, Sec. 5.2.1) reproduce it as $650bn x 55% x 35% and obtain about $80bn with non-duplicated income. (v) Auditor B's stylised check: with a 20% haven share and a 30% ordinary rate, moving the haven rate from 3% to 10.6% cuts the saving from 18.0% to 12.9%.

Why this severity. The stated reduction is overstated by about a quarter even on the paper's own inputs and by a factor of roughly two to three once haven profits and haven taxes are measured on a consistent basis. A sizeable tax saving from havens remains.

The authors' response. None located; Blouin & Robinson (2020) note that the $130bn figure 'has never been retracted'.

Why it matters, in detail. NIPA taxes include tax on realized gains while profits exclude them, so ETR = t*(P_taxable + KG)/P. Gains/profits fell from 0.25 (1998) to about 0.08-0.14 (2013), which mechanically lowers the measured rate by about 2.6-4.6 pp depending on the effective tax rate on gains (empirical slope 0.23; average rate; 35%). The gains-adjusted worldwide ETR falls from 23.7% to about 17.7-18.5%. Decade averages: -11.0 pp unadjusted, -8.2 pp adjusted. FTC-based foreign taxes and portfolio dividends add about 1 pp of measured decline. This also shrinks the room left for avoidance in F1, though in percentage-point terms the avoidance estimate there is unchanged.

Evidence. BEA: 'Taxes on corporate income consists of taxes paid on corporate earnings ... These earnings include capital gains and other income excluded from PBT' (NIPA Handbook ch.13, Dec. 2024, p.13-6). Corporate gains (NIPA Table 7.16 'Gains, net of losses, from sale of property', workbook col. V) were 25.0% of NIPA profits in 1998; for 2013 the workbook extrapolates 13.6% (+2 pp a year after 2011), while IRS SOI Table 12 gives C-corporation gains of $218bn gross / $173bn net, 10.4% / 8.2% of NIPA profits. The measured ETR tracks gains: OLS slope of annual changes in the ETR on changes in gains/profits, 1985-2011, 0.23 (r = 0.64); e.g. 2000-02 the ETR fell 12 pp as gains fell from 37% to 11% of profits. Adjusted 1998-2013 declines: gains added to the denominator 5.2-6.0 pp; a 35% tax on gains removed from the numerator 3.8-5.5 pp; the empirical slope about 6-7 pp; versus 9.4 pp (adjudicator script etr_decomposition.py; auditor A: 4.8-6.0 pp). The workbook's own gains-inclusive US-paid rate (col. AX: 20.8% -> 14.3%) is not reported. With decade averages, which footnote 5 recommends, the decline is 11.0 pp unadjusted and 8.2 pp gains-adjusted. Smaller conventions: counting foreign taxes only as FTCs on repatriated income (BEA MOFA foreign income taxes $138.6bn vs $107.1bn FTC in 2013) adds about 0.6 pp to the 1999-2013 decline, and foreign portfolio dividends in the denominator ($30bn in 1999, $172bn in 2013) 0.5 pp; the claim that 'unrepatriated profits face almost no foreign taxes' (p.133) does not hold for non-haven retained earnings ($120bn in 2013). The decline is robust to data revision: current NIPA vintage plus IRS-reported FTC gives 28.1% -> 19.7% (-8.35 pp).

Why this severity. The single-year headline trend (9.4 pp, 'a third') is overstated by roughly 30-45% (an adjusted fall of about 5-7 pp) and the decade-average trend by about a quarter. The decline is real in every variant, and the author flags capital-gains volatility (fn.5), though not for the 1998 base year.

The authors' response. fn.5 (p.132): yearly data are volatile 'in particular because of year-to-year swings in capital gains realizations', so decade averages are preferable for long-run trends. The 1998 base-year effect is not discussed.

Why it matters, in detail. Loss = U*(r*J + m*K + L). The income term ($125bn) scales with the assumed taxable return r = 7% (a total return taxed annually); r = 4-5% removes $36-54bn. The inheritance term ($60bn) applies m = 3% and top statutory rates K with no exemptions; it implies a US estate-tax take of 1.2% of undeclared wealth a year, roughly eight to ten times the ratio of federal estate and gift taxes actually collected (about $21bn in 2013) to onshore top-0.1% wealth (roughly $15tn, i.e. 22% of household wealth as cited in fn.9). Undeclared share 90% vs 75-80% removes $21-32bn. A plausible central range is roughly $110-150bn, with $95-170bn under other defensible choices.

Evidence. Re-implementation reproduces $190.9bn = income tax $125.3bn + inheritance $59.6bn + wealth tax $6.0bn (adjudicator script revenue_loss.py; same as both auditors). (i) Return: DataTab1!N3 = 7% 'including capital gains' (note A29: raised from the book's 6%), applied to the whole undeclared stock although unrealised gains are not taxed annually and the paper's own appendix capitalises interest at 2%. r = 5% gives $155bn, 4% $137bn, 3% $119bn. Pellegrini, Sanelli & Tosti (2016, Banca d'Italia QEF 367, pp.27-28), with a similar stock ($6.1-7.1tn at end-2013) and instrument yields of 0.5-3.3%, put capital-income tax evasion at $20-42bn a year (at most $49bn in 2013) against Zucman's $127bn. (ii) Inheritance: DataTab1!O3 = 3% per year taxed at the top statutory rate (col. K), including 30% for Asia and 40% for Latin America and Africa, although China, India, Hong Kong, Singapore, Malaysia, Indonesia and Mexico had no inheritance tax in 2013 and spousal or charitable exemptions are ignored; setting those three regions to zero gives $167bn, halving the whole inheritance component $161bn. For the US the model implies $13.1bn a year of evaded estate tax, 63% of all federal estate and gift taxes actually collected in 2013 ($20.9bn, NIPA Table 5.11, series LA000372). (iii) Undeclared share 90% (F6). Combinations: the author's own 2017-edition parameters (25% declared, r = 6.5%) on the JEP inputs give $152bn; r = 5% with 80% undeclared $138bn; the same with half the inheritance component $111bn; r = 4% with 80% undeclared and half the inheritance component $95bn. (iv) The abstract's 'more than $200 billion' exceeds Table 1's $190bn. (v) The US 'at least $36 billion' (p.142) includes about $13bn of estate tax but is compared with the top 0.1%'s federal income tax bill alone.

Why this severity. Defensible alternative parameters lower the headline by roughly 20-50% (to about $95-170bn), more than a quarter, although a loss on the order of $100bn or more a year survives and some omissions (non-financial assets, evasion on the income first moved offshore) push the other way.

The authors' response. No direct reply to Pellegrini et al. located. Zucman's 2017 French-edition workbook (DataTab4) lowers the return to 6.5% and raises the declared share to 25% ($170bn on $8.7tn for 2016); the EU Tax Observatory's Global Tax Evasion Report 2024 shifts to estimating untaxed offshore wealth.

Why it matters, in detail. Revenue is linear in the undeclared share in taxing regions: 0.9 -> 0.8 scales the $190.9bn to $169.7bn; 0.75 gives $159bn. The global 80.8% is an artefact of coding zero-tax Gulf wealth as declared.

Evidence. p.141: 'about 80 percent of the wealth held by Europeans seems to be evading taxes ... On the assumption of a like basis for other tax havens, Table 1 provides estimates'; p.142: 'my assumption that 80 percent of offshore funds is undeclared'. Appendix (revised Dec. 2014): at most 22.5% of European wealth in Switzerland is declared or under EU-savings-directive withholding. DataTab1!H5:H25 = 10% declared for Europe, Asia, Latin America, Africa, the US, Canada and Russia; the aggregate 80.8% (H26) arises only because Gulf wealth (zero tax) is coded as 100% declared. With 80% undeclared the loss is $169.7bn (-11%); with 77.5% $164bn; with 75%, the author's own later choice, $159bn.

Why this severity. Making text and computation consistent lowers a core number by 11-17%, a noticeable but not transformative change.

The authors' response. Zucman's 2017 French-edition workbook (DataTab4, note A29) raises the declared share to 25% for all countries, 'based on latest AFC STD data (see ... December 2014 update to JEP appendix)'.

Why it matters, in detail. tax/net income = 3.0% (2011p) but tax/(net - equity income + tax) = 10.6% (2011p), 9.5% (2013r); non-haven own rate about 35%. 'Negligible' becomes 'low, about one-quarter of the non-haven rate'.

Evidence. Workbook TableS.2!K12 = $18.4bn / $611.9bn = 3.00% (the 21 inputs match BEA's 2011p file; auditor B). $456.7bn of the $611.9bn is equity income (75%). On own pre-tax income (net - equity + taxes = $173.5bn) the rate is 10.6% for 2011p and 9.5% for 2013r (Bermuda + UK Caribbean grouping: 10.0% and 8.8%), against about 35% for non-haven affiliates and 27-28% for all countries (adjudicator script bea_mofa_havens.py). For the DII-based 'haven profits' of Figure 3, which include lower-tier income taxed elsewhere, the relevant rate would be higher still.

Why this severity. The haven tax rate is understated about threefold; this matters mainly through the tax-bill estimate (F3). On any consistent base havens remain very low-tax.

The authors' response. Later Zucman co-authored work measures affiliates' effective tax rates as foreign income tax divided by profit-type return, which excludes equity income (Torslov, Wier & Zucman 2023, Figure 2 note); the 3% was not revised.

Why it matters, in detail. B13 = 0.08 x (0.85 x 80,000 x 1.32 / 0.94); growth 2008-13 = growth of the denominator (about +30%). The method cannot detect changes in the offshore share; only Swiss and Luxembourg data are direct. Absolute growth is plausible given wealth growth (FGMZ 2025: stable at about 7% of wealth, about 10% of world GDP).

Evidence. Table1!B13 = ROUND(0.08*J13, -2), with J13 = $95,489bn hard-coded from the 2013 book (Zucman2013LivreTableauxGraphiques.xlsx, Donnees-T2!H4 = 0.85 x 80,000 x 0.3 x 4.4 / (1 - 0.06): world GDP, a wealth-income ratio, a financial share and a gross-up, not data); the 8% (Donnees-T2!G5) is an input carried over from the QJE end-2008 estimate ($5.9tn). 7.6/5.9 = 1.29 is essentially the growth of the assumed world-wealth denominator (95.5/73.75 = 1.29), so the '28 percent' cannot reflect any change in offshore wealth relative to wealth, contrary to the framing on p.141 ('The growth is stronger in the emerging Asian centers ... so that globally, according to my estimate'). Wealth outside Switzerland (EUR 4,000bn) is a residual. Later work by the author's team (Faye, Godar, Moura & Zucman 2025, Sec. 3.1) finds offshore holdings 'fluctuated around 7 %' of household financial wealth over 2001-2023 'without a clear trend'; Pellegrini et al. (2016) find $6.1-7.1tn for 2013.

Why this severity. The level is plausible and independently corroborated, but its precision is overstated and the global growth figure offered as evidence that offshore wealth is rising is a construct: an important undisclosed caveat for C7.

Why it matters, in detail. US offshore wealth (EUR 918bn) is 97% residual allocation (EUR 888bn = 0.85 x 26.11% x EUR 4,000bn). Europe's share is understated by about 1 pp through double counting; regional denominators mix Flow of Funds, Eurostat and Credit Suisse net financial wealth.

Evidence. DataTab1: of the EUR 4,000bn held outside Switzerland, Europe is set at 25%, Asia 20%, the Gulf, Latin America, Africa and Russia at their Swiss shares, and North America receives the residual 26.11% (E22, hard-coded), split 85/15 between the US and Canada. The US is given EUR 30bn in Switzerland versus EUR 60bn for Canada, whereas SNB fiduciary deposits put US holdings above Canada's (1.6 times in 2013, 3.3 times in 2008) and the author's 2017 edition uses US 2/3, Canada 1/3. The US $1.2tn and $36bn, and the claim that the EU has 'about 30 times more wealth hidden in Switzerland than the United States' (p.143; about 17 times with the 2017 split), depend on these choices. Europe's denominator (Table1!J5 = 23,980) is the 2013 book's onshore-plus-offshore total (Donnees-T2!H9 = onshore 21,380 + offshore 2,600), so offshore wealth is counted twice: 9.9% instead of about 11.0% (auditor B). Regional onshore plus offshore wealth sums to $102.1tn against the $95.5tn world total behind the 8%. Small parent/child mismatches: North America $32.9bn vs US + Canada $32.0bn; Europe's countries EUR 2,020bn vs EUR 2,000bn.

Why this severity. Affects secondary (regional) results and the US policy comparison, with mixed direction; the totals are unaffected.

The authors' response. The 2017 French edition (DataTab1 notes) revises the US/Canada split (US 2/3 of North America, based on 2008-13 fiduciary deposits) and the regional shares.

Why it matters, in detail. All non-resident custody holders (including foreign funds, banks, insurers, sovereign funds) are mapped to households. Directly held private-customer portfolios fell 16-19% from 2009 to 2014 while institutional holdings rose 37%; fiduciary deposits fell 56%. The true household share is unknown (between Schmitz's direct share and Zucman's 100%).

Evidence. SNB annual custody statistics (cube bawebedomsecwja; adjudicator script snb_custody_holders.py, same figures as auditor A): foreign-issuer securities of non-resident private customers fell from CHF 659bn (end-2009) to CHF 534bn (end-2014), -19% (-16% in USD); commercial customers -43%; institutional investors +37% (CHF 967bn -> 1,328bn); the private share fell from 63% (1998) to 27% (2014). In Table S.1 the April 2009-June 2014 rise (+17% in USD, +14% in EUR) combines +36% in custody securities with a fall in fiduciary deposits from $435bn to $191bn. Zucman's 2013 annex (pp.5-6) calls the SNB sector split 'deeply misleading' because holding companies, trusts and nominees count as commercial or institutional, but it also states that the author treats Middle-East central banks and sovereign funds as private wealth. Schmitz (2021) finds only about 24% (2017) held directly by private customers; with a 55% household share of col. [2], June-2014 Swiss offshore wealth would be $1.53tn rather than $2.46tn (auditor B; illustrative). Faye, Godar, Moura & Zucman (2025, EU Tax Observatory WP 41) find Switzerland's share of global offshore wealth fell from almost half before 2008 to 22%.

Why this severity. The evidence that households' offshore wealth in Switzerland kept rising after 2009 becomes ambiguous, and the Swiss level is uncertain; because the global total is fixed at 8%, the Swiss level mainly affects the regional split rather than the $7.6tn.

The authors' response. Zucman (2013, La richesse cachee des nations, online annex pp.5-6) argues the SNB sectoral breakdown is 'deeply misleading'; Faye, Godar, Moura & Zucman (2025) use centre-specific household shares but still attribute the residual gap to households.

Why it matters, in detail. 17.4% -> 'about 20%' (+15%); growth multiple 4.8-9.0x depending on base; abstract revenue exceeds its own Table 1 by more than 5%.

Evidence. Haven share of all profits 17.4% in 2013 (workbook DataFig1-5!CR89) vs '18 percent' (p.129) and 'about 20 percent' (abstract, p.121); 20% was exceeded only in 2008 (21.5%). 'Tenfold since the 1980s': 9.0x the 1984 low (1.9%), 5.9x the 1982-89 average (2.9%), 4.8x 1982 (3.6%); Figure 3 starts in 1984 although Figures 2 and 4 (same data) start in 1982. Revenue: Table 1 $190bn, text 'about $200 billion', abstract 'more than $200 billion'. (Conversely, the '$200 billion' of extra 2013 tax at the 1998 rate is $225bn in the workbook.)

Why this severity. Rounding and framing differences of roughly 5-15% on levels and a base-year-dependent growth multiple; no effect on the qualitative conclusions beyond those graded in F2 and F5.

Why it matters, in detail. Foreign tax rate on pre-tax earnings >= 4.6% (not <= 3.1%); bound direction reversed.

Evidence. Microsoft FY2014 10-K: $29.6bn unrecognised deferred US income taxes 'or foreign withholding taxes' on $92.9bn of permanently reinvested (after-foreign-tax) earnings. With the deemed-paid credit and gross-up, 0.35(E + T) - T + W = 29.6 gives T = $4.5bn + 1.54W, i.e. foreign taxes of at least 4.6% of pre-tax earnings. Even in the paper's ungrossed logic, the withholding taxes included in the $29.6bn make 3.1% a lower, not an upper, bound. The earnings are 'mostly from ... Puerto Rico, Ireland, and Singapore'; Puerto Rico is outside the paper's six havens.

Why this severity. A small error in a single-firm illustration; it does not change the 'low tax' message.

Why it matters, in detail. None changes a headline: the largest numeric effects are Fed-profit treatment (+0.7 pp on the haven share, against the author's interest) and the retention label (45% vs 30% of all foreign profits).

Evidence. Google's IPO was in August 2004, not 2003 (p.125; Google 2004 10-K; the 2015 book corrects it). Figure 4 note (p.131) says the negative 2005 reinvested earnings mean 'the 2005 data point excludes repatriations from profits made prior to 2005'; per BEA (FAQ 166), negative reinvested earnings indicate dividends paid out of accumulated prior earnings, i.e. the point includes them. Figure 2 note: total DII 'about $500 billion' vs $439bn ($452bn with CCAdj). '80 percent ... (that is, 45 percent of all foreign profits)' is 44.8% of DI income, about 30% of all foreign profits. '9 percent' of US equities 'in 2013' (p.137): the series ends in 2012 (8.7%; figure note correct). 'EuroStoxx 500' (p.141) does not exist and the cited source (Adam 2014, p.8) reports -25% 'pour l'Eurostoxx', not -20%. fn.8 describes $350bn in Luxembourg as household wealth held 'through family offices and other intermediaries'; Adam (2014, p.8, fn.3) describes EUR 270bn of institutional clients ('insurance companies, family offices, management companies of investment fonds, pension fonds') outside the study's scope. The Section 199 deduction was 9% only from 2010 (phased in from 3% in 2005), not 'since 2004'. The Depository Trust Company (not 'Corporation') records participants' positions, not beneficial owners (p.137). Figure 3 does not remove Fed profits from the denominator despite the appendix's 'always' (removing them raises 17.4% to 18.1%). The 1930s decade average in Figure 1 averages 1930-1940. German reprint (Wirtschaft und Gesellschaft 2015): 'zwei Drittel' for 'one-third' of US tax revenues from capital taxes, and an HSBC fine of '1,2 Mrd. $' for $1.92bn.

Why this severity. Presentation, dating and source-description errors with no effect on any core number.