← All papers
Theresa Neef, Panayiotis Nicolaides, Lucas Chancel, Thomas Piketty, Gabriel Zucman · EU Tax Observatory / World Inequality Lab Note, March 2022
Effective Sanctions against Oligarchs and the Role of a European Asset Registry
13 problems found: 1 major · 4 moderate · 8 minor
What the paper claims
- Russia exhibits the highest wealth inequality in Europe.
- Much of Russians' wealth is held in offshore centres: NPZ (2018) 'estimate that about half of total Russian household wealth is held abroad'.
- Russia's wealthiest nationals conceal a large share of their wealth through tax havens (AJZ 2018).
- The group of rich Russians with offshore wealth is far larger than the ~680 sanctioned individuals.
- Offshore wealth is a global phenomenon: about 10% of world GDP, relatively stable through the 2000s and 2010s.
- A Task Force for Asset Ownership leading to a European Asset Registry would yield better-targeted sanctions and effective tools against money laundering, corruption and tax evasion.
- Supporting context: broad sanctions hurt the general population, and the existing EU transparency framework has gaps.
Every problem found
1majorconfirmed · newly identified here · misrepresentation of source
The note says about half of all Russian household wealth is held abroad. The study it cites says about half of Russians' financial savings (bank deposits, shares, bonds) are abroad. Once homes, land and businesses are counted, the share held abroad is about a quarter. Russians' offshore wealth is still very large, but the note doubles its share.
Technical detail and evidence
Why it matters, in detail. Denominator error. NPZ's 2015 offshore stock (0.864 NI) is 0.578 of onshore financial assets plus offshore, but only 0.236 of net personal wealth (3.667 NI including offshore). The range across NPZ's r = g -/+ 2% scenarios is 0.194-0.285; the 2007 share is 0.175. Other sources agree. AJZ's Russian top-share correction (the source of the note's Figure 2) uses an offshore stock of 13.5-18.8% of household wealth in 2000-09. WID's own offshore aggregate in the vintage the note used is 19-22% of household wealth including offshore (0.73 NNI out of 3.28-3.75 NNI, depending on whether business assets are counted; 2015). The overstatement is about 26 pp (factor about 2.1). The 85%-of-income figure in the same paragraph is right, so the stock is correct; only the share is wrong. The same financial-to-total shorthand occurs in Zucman (2014, JEP 28(4), p. 140: 'the fraction of wealth held abroad ... as much as 50 percent in Russia', from a Table 1 column headed 'Share of financial wealth held offshore'). It also appears in press coverage (NBC News, 2 Mar 2022: '52 percent of Russia's wealth was held outside the country').
Evidence. NPZ (2018) JOEI 16:189-223, p. 205: 'Russians own approximately as much financial wealth offshore as onshore, i.e. they own about 50% of total financial wealth offshore'. NPZ journal ESM1 Table A20 (2015, bn RUB): net personal wealth 269,017, offshore 63,397, onshore financial assets 46,334 -> offshore = 23.57% of personal wealth (Table A22 gives 0.2357), 57.8% of onshore-plus-offshore financial wealth, 86.4% of national income (Table A21). Lower/upper return scenarios (Table A33, r = g -/+ 2%): 19.35% / 28.48%. 2007: 17.5%. Script final/scripts/npz_offshore_share.py -> results/npz_offshore_share.txt; both auditors obtained 23.6% independently.
Why this severity. This is the note's headline magnitude for offshore wealth. It is overstated about twofold (about 24% versus 'about half'), well beyond the rubric's one-quarter threshold. The qualitative claim that much Russian wealth is offshore survives: a quarter of household wealth, or 85% of national income, is still exceptional.
2moderateconfirmed · previously known · overclaiming
No survey measures household wealth in Russia. The note's Russian figures were built by taking the average pattern of wealth in the US, China and France and stretching the very top to fit the Forbes billionaire list. The note does not say this. Russia still comes out as Europe's most unequal country under every reasonable assumption, and an independent estimate agrees. But the exact percentages, and the comparison with the US, are much less certain than the note suggests.
Technical detail and evidence
Why it matters, in detail. Russian top shares are identified only by about 100 Forbes billionaires plus parametric interpolation from P99; P0-P99 is borrowed from the USCNFR average. Around NPZ's 2015 benchmark, the published variants span -1.8/+3.4 pp (top 10%) and -3.6/+6.8 pp (top 1%). Applying these, together with the denominator issue (F3), to the note's 2021 values shrinks Russia's lead over the next European country. In the most adverse case it falls from 6.6/10.8 pp to about 2.6/2.8 pp (Turkey); the rank holds but the margin is model-dependent. The Russia-US top-10% ordering flips across variants (69.5-74.7 vs 72.9 in 2015), exactly as NPZ warn. Vintage instability is of the same order: live WID revises 2019 to 72.0/43.6 and 2021 to 76.3/49.6, against 74.1/47.7 for both years in the vintage used. Credit Suisse's estimate is produced separately, though also calibrated on rich lists, and puts Russia's top-1% share even higher (58%), supporting the ranking.
Evidence. WIL Technical Note 2021/18 (Neef, dated Nov 2021), PDF p. 2: 'the normalized wealth distributions of France, China, and the US (USCNFR) are joint and then top-corrected with Russian billionaire's data from Forbes List'; PDF p. 3: 'Russian citizens and residents were included'; PDF p. 4: 'Unfortunately, a wealth survey is currently not available for Russia'. NPZ (2018) p. 221 (quoted). WID 25-Oct-2021 extract: the top-10% implied by the top-1% share and the USCNFR shape is 74.39 vs WID 74.12 (2019/2021) and 74.40 vs 73.47 (2015) [scripts/ru_top10_mechanics.py]. Normalised thresholds (e.g. P99.98/average 195.149) and shares (74.12/47.69) are identical in 2019 and 2021 [scripts/wid2021_checks.py]. NPZ Gpinter variants for 2015: top 10% 69.5-74.7, top 1% 39.0-49.4 (benchmark 71.3/42.6), vs US 72.9/36.6 in the same vintage [scripts/npz_wealth_variants.py]. Ranking: Russia is first among 40 core European countries (next Hungary 67.3/33.5) and among 44 including TR/GE/AM/AZ (next Turkey 67.5/36.9). Live WID (Oct 2026) for 2021: Russia 76.3/49.6 vs Sweden 68.5 / Georgia 34.0 [scripts/wid_live_europe.py]. Most adverse combination (NPZ least-unequal variant plus the F3 correction): Russia 2021 about 70.1/39.7, still first [results/europe_rank_2015_vs_npz_variants.txt]. In 2015, NPZ's benchmark top-10% share (71.3) equals Ireland's (71.4) in the same vintage. External: Credit Suisse Global Wealth Report 2021, Table 3 (p. 24), puts Russia's top-1% share at 58.2% (2020), vs UK 23.1, France 22.1, Germany 29.1, Italy 22.2 and US 35.3. Kapeliushnikov (2020, HSE preprint WP3/2020/01, Appendix 1, pp. 54-55) makes the same criticism: NPZ's wealth estimates rest on a 'practically zero empirical basis', and the bottom 99% is imputed from the US, France and China.
Why this severity. This is an important undisclosed caveat. The headline levels are model outputs, uncertain by several points, and the comparison with the US that the note invites is not identified. But Russia's first rank in Europe, the claim the note actually makes, survives every published variant and later revisions. The core claim is therefore qualified, not overturned.
The authors' response. No response to these points located, and no reply to Kapeliushnikov. The authors' own sources acknowledge the fragility (NPZ 2018, p. 221; Neef, WIL TN 2021/18: 'better approximations than the USCNFR normalized distribution of wealth can be explored'), but the note omits it.
3moderateconfirmed · newly identified here · methodology
The note says about 20,000 Russians had more than EUR 10 million and about 50,000 more than EUR 5 million. Those numbers use 'purchasing-power' euros, which adjust for Russia's lower prices, and the note does not say so. In real euros at the exchange rate the counts are about 9,000 and 20,000. That is still far more people than the roughly 680 on the EU sanctions list, so the note's point stands, but the figures are more than twice too high for the purpose they serve.
Technical detail and evidence
Why it matters, in detail. PPP conversion (39.8 RUB/EUR) instead of market conversion (86.7 RUB/EUR) lowers the RUB value of each euro threshold by a factor of 2.18. Given the Pareto tail, that raises the count above a fixed euro threshold by 2.5-2.6x (23,338 vs 8,993; 50,498 vs 20,469). The note's rounded figures are 2.2x and 2.4x the market-rate counts. PPP euros measure domestic purchasing power, not the euro value of assets that could be frozen in the EU. At the March 2022 rates quoted in the note (117-129 RUB/EUR) the counts would be lower still. The counts also inherit the imputation (F2) and denominator (F3) uncertainty, rest on the 2019 distribution carried forward, and fall in the USD 10-100m range where NPZ (p. 221) say information is lacking.
Evidence. WID 25-Oct-2021 extract, Russia 2021: npopul992i = 113,279,408; xlceup999i = 39.783 RUB per PPP euro; xlceux999i = 86.673 RUB per euro. Log-linear interpolation of thweal992j gives, in PPP euros, 23,338 adults (0.0206%) above EUR 10m and 50,498 (0.0446%) above EUR 5m. In market euros it gives 8,993 (0.0079%) and 20,469 (0.0181%) [scripts/wid2021_checks.py -> results/wid2021_checks.txt]. These are identical to Auditor B and close to Auditor A (23,485/50,813; 8,998/20,997). At market rates 10,394 adults are above USD 10m. On 4 Mar 2022 Zucman told ProMarket that freezing 'all offshore holdings above $10 million' would 'affect about 10,000 to 20,000 Russians'.
Why this severity. The counts are overstated by a factor of 2.2-2.4 relative to market-rate euros, but they are a secondary, supporting statistic. The point they serve, that the relevant group is far larger than the ~680 sanctioned individuals, holds at market rates (about 9,000-20,500 adults).
4moderatelikely · newly identified here · internal inconsistency
The database version the note used divided the fortunes of the richest Russians by a national total that left out money held offshore and business assets. That makes the richest group's share look bigger than it is. By the underlying study's fuller method, the top 1% own about 43% of wealth rather than 48%. Russia is still the most unequal country in Europe.
Technical detail and evidence
Why it matters, in detail. Suppose the vintage's Russian very top is Forbes-anchored in levels. A denominator that omits offshore (0.73 NNI) and business assets (0.47 NNI) then inflates top shares by up to 3.67/2.55 = 1.44 at the very top. The effect is smaller further down, where the borrowed shape scales with mean wealth. Applying the 2015 NPZ/vintage ratios to 2021 gives about 72.0% (top 10%) and 43.3% (top 1%), instead of 74.1%/47.7%. Other methodological updates between NPZ (2017/18) and the 2021 vintage could also contribute, and the WID 2021 Russia code is not public. The mechanism is therefore likely rather than proven. Direction: it overstates C1's levels but not Russia's European rank (next: 67.5/36.9).
Evidence. WID 25-Oct-2021 extract, Russia 2015: mhweal999i/mnninc999i = 2.550; mhwhou 1.986 + mhwagr 0.097 + mhwfin 0.659 - mpwdeb 0.193 = 2.549; mhwoff 0.731; mhwbus 0.472. ahweal x adults / mhweal = 1.0000, so the distribution is anchored to mhweal. The vintage's own household wealth-income ratio wwealh999i is 3.709. NPZ ESM Table A21: personal wealth 3.667 NI including offshore 0.864. Share ratios from scripts/very_top_npz_vs_wid.py -> results/very_top_npz_vs_wid.txt: 2015 top 1% 46.88 vs 42.58 and top 0.01% 23.08 vs 16.86; 2007 top 1% 40.22 vs 35.96 and top 0.01% 17.36 vs 11.47. Live WID (Oct 2026) normalises to a larger total (mhweal 3.54x NNI in 2015) and gives 72.3/44.8 for 2015, between NPZ and the vintage.
Why this severity. It shifts the headline top-1% share by roughly 4-5 pp (about a tenth) and the top-10% share by about 2 pp. That is a noticeable change to the level of a core number, but well below the one-quarter threshold, and it leaves Russia's European rank unchanged.
5moderatelikely · previously known · speculative assumption
The figure that Russia's richest keep 60% of their wealth offshore is not measured from Russian data. It comes from assuming that Russia's richest hold the same share of their country's offshore money as Scandinavia's richest do, a pattern found in leaked bank data. Under other reasonable assumptions the share could be anywhere from about 40% to 90%, and the estimate describes the 2000s, not 2021. Rich Russians very likely keep a large part of their wealth offshore, but the exact percentages rest on assumptions and should not be read as measured facts.
Technical detail and evidence
Why it matters, in detail. The offshore fraction is q*O/(T*(D+O)), where O is NPZ's net-errors-and-omissions-based stock, T is NPZ's Forbes-based top share and q is the Scandinavian allocation. The fraction moves about 1.29 per unit of q, so a change of 0.2 in q moves it by about 26 pp. No Russian evidence on q exists. Concentration may well exceed Scandinavia's: Forbes billionaires alone hold wealth equal to about half of NPZ's offshore stock (NPZ p. 206). The Dutch value (q about 0.07, which would give 9%) is probably not transferable to Russia. Annual figure-method fractions range from 49% to 69% over 2000-08. The 2009 value exceeds 100%, a sign that q = 0.52 is at the edge of feasibility given NPZ's totals. The stock choice also matters: AJZ's BIS-based Russian stock is about 28% of the NEO stock in 2007. It is narrower in scope, however, and not coherent with NPZ's totals (see rejected claims).
Evidence. AJZ2017bData.xlsx: T.A11!G101:G110 hard-coded (0.121 ... 0.115 in 2007 ... 0.075 in 2009). T.A8!G108 = (T.A11!G108*(T.A6!G108+T.A9!T107) - T.A9b!I107*T.A9!T107)/T.A6!G108, with T.A9b!I = 0.5156 (=I$106, constant). T.A8!G110 = -0.0022. Recomputation [scripts/ajz_russia_top001.py] gives a consistent offshore fraction of 66.3% (2000-09) and 65.2% (2007). Holding NPZ's totals fixed and varying the allocation q: q = 0.30 -> 38.6%, 0.40 -> 51.5%, 0.60 -> 77.2%, 0.70 -> 90.1%. For the top 0.1%: 45.5% (figure method) / 54.4% (consistent) [scripts/ajz_fig2_top01.py]. 'One-half to two-thirds' therefore matches the 2000-09 model for groups between the top 0.1% and the top 0.01%, but no estimate exists for 2021. Leenders et al. (2023, JPubE 217:104785) find the top 0.01% own 7% of hidden wealth in Dutch data. AJZ (2018, p. 98): 'for all plausible distributions, the impact on our findings is second-order'.
Why this severity. The specific 60% and one-half-to-two-thirds figures depend on an unvalidated allocation parameter and are presented as findings about Russia now. For defensible Russian values of the parameter (about 0.3-0.7) the fraction runs from about 40% to 90%, so 'a large share' survives. The precise magnitudes, and their application to 2021, do not.
The authors' response. AJZ (2018, p. 98) report robustness tables (Online Appendix A9-A11) and state that 'for all plausible distributions, the impact on our findings is second-order'. AJZ (2019, AER) state that they do not claim their Scandinavian estimates hold everywhere. No response specific to this note located.
6minorconfirmed · newly identified here · internal inconsistency
The chart splits the richest group's wealth into onshore and offshore parts using two different yardsticks. Done consistently, the offshore part is a bit bigger than shown: about two-thirds rather than 60%. The error was copied from the source study, and it works against the note's own argument.
Technical detail and evidence
Why it matters, in detail. s_in = T/(D+O) and s_out = (T - qO)/D, so the plotted segment s_in - s_out is not the offshore component qO/(D+O). The bias rises with O/(D+O), which was 0.13-0.19 for Russia in 2000-09. For Russia it understates the offshore part of top wealth by about 6.5 pp; the bias is smaller for the other countries, and the cross-country ranking is unchanged.
Evidence. AJZ2017bData.xlsx: T.A6 is 'Total household wealth (excluding offshore wealth)'; T.A8 formula as in F5. DataF7 rows 7-9 are the Figure 2 bars, which reproduce exactly: Spain 2.46+1.02, UK 2.73+1.68, Scandinavia 4.07+0.71, France 3.67+1.57, USA 7.14+0.71, Russia 5.02+7.48. Consistent offshore component q*O/(D+O), averaged over 2000-09, = 8.29% [scripts/ajz_russia_top001.py, scripts/ajz_fig2_top01.py].
Why this severity. This is a labelling and presentation flaw that the note copied faithfully from its source. Correcting it raises the offshore fraction (66% rather than 60%), so it strengthens the note's claim rather than weakening it.
7minorconfirmed · newly identified here · internal inconsistency
The note gives a single year, 2007, for numbers that are actually ten-year averages. The 2007 figure itself is slightly lower (11.5% rather than 'more than 12%'). This does not change the picture.
Technical detail and evidence
Why it matters, in detail. 11.5% (2007) vs 12.5% (decade mean). Annual values range from 7.5% to 15.1% over 2000-09, so the choice of year matters little relative to that volatility.
Evidence. AJZ2017bData.xlsx: T.A11!G108 = 0.115 (2007); T.A11b!G99 = AVERAGE(T.A11!G101:G110) = 0.125. Figure 2 heading '(2000-2009)'. AJZ (2018) p. 98, Fig. 8 note: 'on average over the 2000-2009 period'. AJZ's text does not date the Russian figure to 2007.
Why this severity. A one-point dating and labelling slip. The size of the top share and of its offshore part are essentially unchanged.
8minorconfirmed · newly identified here · selective presentation
The note treats Russians' offshore wealth as money parked abroad, out of Russia's reach and within reach of EU sanctions. The study it relies on suggests much of this money went back into Russian companies through offshore shell firms. If so, less of it sits within reach of European sanctions than the note implies, although a register of owners would still help show who controls what.
Technical detail and evidence
Why it matters, in detail. NPZ's stock measures claims routed through offshore entities via unrecorded capital flight, not the location of the underlying assets. If much of it is round-tripped FDI or portfolio equity in Russian firms, the EU-located, freezable part is a fraction of the 0.86 NI. Freezing shares of EU-domiciled holding companies (e.g. in Cyprus) still has bite, but the underlying corporate assets remain in Russia. NPZ themselves say 'it is difficult to go further', so the effect is unquantified. The direction points toward overstating what EU sanctions can reach.
Evidence. NPZ (2018) p. 206 (quoted). NPZ p. 205: the estimate is cumulated net errors and omissions plus fictitious transactions, i.e. unrecorded capital flight. Note p. 2 (definition) and p. 3 ('therefore, not in roubles'). Codex report, section 'Assumptions And Robustness'.
Why this severity. No inequality number changes, and a beneficial-ownership registry would still help trace offshore shells that own Russian assets. But the note omits the source's own warning that much of the stock may be Russian assets routed through offshore entities rather than assets located abroad. That bears on how much wealth EU sanctions could actually reach.
9minorconfirmed · newly identified here · arithmetic or transcription
The note says the rouble lost 37% of its value against the euro. Using the note's own exchange rates, it lost about 27%; 37% is how much more a euro cost in roubles, which is a different measure. Depending on the rate used, the true fall was about 27-33%. This is background detail and does not affect the main argument.
Technical detail and evidence
Why it matters, in detail. Depreciation of the rouble should be measured as the fall in its euro value, 1 - e0/e1 (e = RUB per EUR), not e1/e0 - 1. Correct values lie between 27% and 33% depending on the source and date of the quote, so the stated 37% overstates it by 4-10 pp.
Evidence. Bank of Russia XML_daily_eng: EUR 86.13 (18.02), 128.95 (12.03); USD 116.75 (12.03). MOEX ISS EUR_RUB__TOM closes: 85.50 (15.02), 87.55 (18.02), 121.03 (11.03), 117.70 (16.03) [scripts/fx_rates.py -> results/fx_rates.txt]. Rouble depreciation: 27.4% on the note's own rates; 29.4% on MOEX 15.02 -> 11.03; 33.2% on official rates 18.02 -> 12.03.
Why this severity. This is a context figure only, and the order of magnitude (about a third, on official rates) is right. No core finding depends on it.
10minorconfirmed · newly identified here · arithmetic or transcription
The note's global figure for money in tax havens is about 5% lower than its own source's, apparently because of a currency conversion. It makes no difference to the argument.
Technical detail and evidence
Why it matters, in detail. EUR 7.9tn x about 1.05 USD/EUR (end-2016) = USD 8.3tn, whereas the source's USD 8.7tn implies about 1.10 (the 2016 average rate). Either way, offshore wealth is about 8% of global household financial wealth (Zucman 2017, Tab2) and about 11% of 2016 world GDP.
Evidence. Zucman2017LivreTableauxGraphiques.xlsx: Tab2 'Held offshore (in tax havens throughout the world): 8.7 trillion $ (8%)'; Tab2(FR) '7 900 milliards EUR'; Tab4 '$8,700 billion grand total taken from AJZ (2017) Appendix I'.
Why this severity. A 5% understatement of a context figure that arises from currency conversion. No conclusion changes.
11minorconfirmed · newly identified here · misrepresentation of source
The note says an EU law made the registers of who really owns companies and trusts public. That is true for companies but not for trusts, which only people with a 'legitimate interest' can consult. It is a small error in the background section.
Technical detail and evidence
Why it matters, in detail. The legitimate-interest standard for trusts is exactly the kind of access gap the registry proposal targets. The CJEU's 22 Nov 2022 judgment (joined cases C-37/20 and C-601/20), which invalidated general public access to company registers, postdates the note and is not counted against it.
Evidence. Directive (EU) 2018/843: Art. 1(15)(c), amending Art. 30(5); Art. 1(16)(d), amending Art. 31(4). Cached EUR-Lex text: sanctions-registry-2022/amld5.html.
Why this severity. A descriptive inaccuracy in the legal background. It does not affect the inequality findings and, if anything, strengthens the case for better registers.
12minorconfirmed · newly identified here · arithmetic or transcription
A few background facts are slightly off. By the time the note came out, the EU sanctions list had grown to about 880 people, most of them politicians rather than oligarchs. 'Roughly half' rather than 'at least half' of Russia's reserves were frozen, and one author's name is misspelled. None of this matters for the note's conclusions.
Technical detail and evidence
Why it matters, in detail. The sanctions-list count was about 200 persons out of date at publication. Its composition, mostly legislators and officials, makes the contrast with the 9,000-50,000 wealthy Russians of F4 even starker. 'At least half' versus about 47% of reserves is negligible.
Evidence. Commission sanctions page (cached commission-20220302.html): '...the list comprises a total of 680 individuals and 53 organisations' (28 Feb package) and '702 individuals and 53 organisations' (2 Mar). IP/22/1649 (9 Mar): '862 individuals and 53 entities'. EU Financial Sanctions File generated 2022-03-18, programme UKR: 881 persons and 62 entities. First listings: 23 Feb 359, 25 Feb 99, 28 Feb 26, 2 Mar 22, 9 Mar 160, 11 Mar 37, 15 Mar 15. 448 person entries mention the Duma and 160 the Federation Council [scripts/eu_list_composition.py]. Interfax, 13 Mar 2022 (https://interfax.com/newsroom/top-stories/76603/): Siluanov, 'roughly half ... approximately $640 billion, with about $300 billion' unusable.
Why this severity. These are background details. Correcting them leaves the argument unchanged; on the list's composition the correction strengthens it, since few listed persons were oligarchs.
13minorplausible · newly identified here · overclaiming
The note says Russians hide more wealth abroad because Russian wealth is so concentrated. The study it cites does not show this; it links offshore wealth more to oil wealth and political instability. This is a passing remark and does not affect the main numbers.
Technical detail and evidence
Why it matters, in detail. Within AJZ's framework, a more concentrated distribution does mechanically imply more offshore wealth, given a concentrated allocation of offshore wealth. But cross-country variation in offshore holdings is attributed by AJZ to resource rents and political instability, not to concentration as such. The 'due to' claim is therefore an assertion, not a finding.
Evidence. Note p. 4; AJZ (2018) p. 95, section 4.3; AJZ workbook, constant allocation T.A9b.
Why this severity. An interpretive sentence with no numerical consequence for the core findings.
Claims examined and rejected
Problems one auditor raised that did not survive checking.
- Badarinza & Ramadorai's 'about 85%' is London-specific but presented without the London qualifier (endnote xv). — The note's paraphrase ('about 85% of properties purchased by foreign entities are conducted through offshore centres') matches B&R. B&R write that 'at least 85% of residential real estate purchases by foreigners in London occurring through a corporation ... are routed through off-shore special purpose vehicles' (accepted manuscript, p. 1). The same endnote cites B&R 'for the London real estate market'. There is no misrepresentation.
- Endnote xxiii cites Johannesen & Zucman (2014) and Johannesen (2014), which study pre-CRS regimes, as evidence of CRS loopholes. — The endnote cites them as examples of 'behavioral reactions' to information-exchange initiatives (relocation to uncooperative havens), which is what they document. It cites Casi et al. (2020) for the CRS itself and does not claim the earlier papers evaluate the CRS.
- AJZ's 'stable' refers to about 8% of household financial wealth, while the GDP ratio moved between 8.4% and 11.6%, so 'relatively stable' (as a share of GDP) misstates the source. — 'Relatively stable' around 10% of world GDP is a fair summary of AJZ Table A1 (2001-2015 mean 9.6%, range 8.4-11.6%), and the 10% figure is in AJZ's abstract (p. 89). This is not a material misstatement.
- The note misreports the 12.03.2022 exchange rate: the official EUR rate was 128.95, and 117 matches the USD rate. — Only partly accepted (folded into F9 as unexplained). 117 is not the official EUR rate for that date, but market EUR/RUB quotes in the days around the note were 115-122 (MOEX close 117.70 on 16 March, the note's date). The note's source cannot be identified, so only the arithmetic error (37% vs 27%) is confirmed.
- With AJZ's own BIS-based Russian offshore stock, the top 0.01% hold 6.8% of household wealth with only 31-35% offshore (2007): a defensible alternative estimate. — Kept only as an illustration of stock sensitivity, not as an alternative estimate. AJZ (2018, p. 95) and NPZ (fn 24) describe the BIS-based stock as narrower in scope (financial assets of residents only); the NEO stock adds real assets abroad and emigrants' wealth, both relevant to sanctions. NPZ's top-0.01% share, which AJZ use, is built on the NEO-inclusive total, so substituting the BIS stock is internally incoherent. The result also depends on what is held fixed: 31% with onshore wealth fixed (reproduced), 16% with total top wealth fixed.
- The note's first author documented the imputation in WIL TN 2021/18 in November 2021, four months before the note, which aggravates the omission. — The date cannot be verified. The technical note is dated November 2021, but the earliest verifiable public copy is a PDF created 5 April 2022 (first Wayback capture 27 May 2022). The substantive point stands independently: NPZ (2018), co-authored by two of the note's authors, carries the caveat (F2).
- The vintage's implied top-0.001% wealth (2015: USD 488bn) exceeds total Forbes Russian billionaire wealth (USD 336bn), suggesting inflated top shares. — Not diagnostic. The top 0.001% (about 1,150 adults) includes the ~90 billionaires plus about 1,000 centi-millionaires, so its wealth should exceed the billionaires' total. The denominator effect is established instead by comparing shares with NPZ's benchmark (F3).
Audit notes
What reproduced. Reproduced from the 25-Oct-2021 WID extract shipped with the WIR 2022 data, the vintage the note most plausibly used. Russia, France and Poland top-10%/top-1% shares match exactly at the stated precision (74.12/47.69, 59.48/27.04, 61.83/30.32). The US (71.12/35.88) and China (67.80/30.46) match within 0.9 pp. Russia ranks first among 40-44 European countries, by 6.6-6.8 pp (top 10%) and 10.8-14.2 pp (top 1%), and is also first in live WID (Oct 2026). Reproduced: the 'about 20,000 / 50,000' counts, but only in PPP euros (23,338 / 50,498); at market rates they are 8,993 / 20,469. Reproduced from NPZ's ESM: offshore wealth = 86.4% of national income in 2015 (note: 85%), and the NPZ quotation (p. 191). Not reproduced: 'about half of total household wealth', which is 23.6% (19.4-28.5%); half holds only for financial wealth. Reproduced from the AJZ workbook: all six Figure 2 bars exactly, and the 60% (7.48/12.50). Not reproduced: 'in 2007 ... more than 12%' (the 2007 value is 11.5%). On a consistent denominator the offshore fraction is 66%; it ranges 39-90% across allocation assumptions. Reproduced: AJZ's 10% of world GDP; the NIESR figures (as deviations from baseline); the key-rate change; the 53 entities and 680 individuals as of 28 Feb. Not reproduced: USD 8.3tn (the source says 8.7tn) and the 37% depreciation (27% from the note's own rates). New computations: the vintage's top shares exceed NPZ's offshore-inclusive benchmark by 3% (top 10%) to 37-69% (top 0.01-0.001%), consistent with its onshore-only denominator; the top-10% share is mechanically implied by the top-1% share and the borrowed shape; and the 2021 distribution is 2019's carried forward. Scripts are in final/scripts (run_all.sh), logs in final/results.
Limits of this audit. The exact WID extract the authors used in March 2022 is not available. The 25-Oct-2021 extract shipped with the WIR 2022 data reproduces the Russian figures exactly, but the US/China top-1% values differ by up to 0.9 pp, so the website data may have differed slightly. The WID 2021 Russia code is not public, so the denominator mechanism in F3 is inferred from the anchoring of the distribution to mhweal and from comparisons with NPZ's benchmark and live WID, not recomputed. NPZ's Stata/Gpinter pipeline was not re-run; its published variant outputs were used. The data quality of the comparator countries' WID wealth series (several are survey-based or imputed) was not audited. The Credit Suisse cross-check uses a published table and is itself rich-list-calibrated for Russia. Russian offshore wealth has no independent validation: external estimates largely reuse NPZ's capital-flight method. Surak (2020), the Task Force for Coordinated Action analogy, the EU AML package and Unshell descriptions, and the EC feasibility-study timeline were not checked in detail. The reserves figure relies on a contemporaneous Interfax report. The policy proposal (C6) is normative; its feasibility and effectiveness, including that an EU-only registry would not reach assets in Switzerland, the UK or the UAE (as the note itself anticipates), were not evaluated. No author response to any of these issues was located.