Emmanuel Saez · In Le Cercle des économistes (ed.), Agir face aux dérèglements du monde, Paris: Odile Jacob, 2020, pp. 133-141

Lutter contre l'injustice sociale : une priorité

4 problems found: 1 moderate · 3 minor

What the paper claims

US income inequality, as measured by the Piketty-Saez long-run top income shares (the essay's example is the top 1% share), was very high at the start of the 20th century and fell a lot during the Great Depression and the Second World War.
Inequality started rising again only with Ronald Reagan's election in 1980 and the development of neoliberal policies, and has risen until today.
Today US inequality is back to a level comparable to that of a century ago.
Programmatic thesis, known only from the title, the section headings and the book preface's summary: fighting social injustice is a priority; after the progressive income tax, a wealth tax ('Après l'impôt progressif, un impôt sur les fortunes'); taxation is a societal choice ('L'impôt, un choix de société'); inequality must be tackled 'prioritairement grâce à l'outil fiscal'. The preface (p. 16) summarises Saez and Zucman jointly. NOT ASSESSED: text inaccessible.

Every problem found

1moderateplausible · previously known · external inconsistency

Economists agree that US income inequality has risen since about 1980 but disagree about how much. A leading alternative estimate finds a rise only about half as large as the author's, partly because a 1986 tax reform changed how income was reported. The essay doesn't mention this debate. Most data sets, including the critics' own, still show today's inequality back near its level of the early 1900s, so the essay's basic message stands.

Technical detail and evidence

Why it matters, in detail. Mechanisms: fiscal-income top shares embed tax-law-driven reporting (C-to-S conversions after TRA86); the tax-unit denominator (declining marriage raises tax-unit counts below the top); and, in DINA, allocation choices for untaxed income. All three push the measured post-1980 rise up relative to AS-type estimates. The 1979/80-2019 rise in the top 1% share is +4.4 pp (AS), +8.4 pp (PSZ) and +11.9 pp (PS fiscal incl. gains), a factor of 1.9-2.7. 40% of the fiscal rise excluding gains occurs in 1986-88. For C3 the level comparison survives within PS (21.96 vs 18.30), in DINA (18.7 vs 20.4) and in the critics' GMMS-linked series (17.8 in 2015 vs 19.5 in 1928; 1917-29 mean 15.4). It fails only when AS's 2019 level (13.8) is set against DINA's early-century mean (20.4), and no AS-consistent pre-1960 series exists to make that comparison like-for-like. The PSZ-AS dispute is open: see the PSZ comments of 2024 and 2026, and Gale, Sabelhaus & Thorpe (2023) per corpus us-income-21.

Evidence. Auten & Splinter (2024, JPE 132(7)), Table 1 and p. 2182: pretax top 1% share 9.4% (1979) to 13.8% (2019), '+4.4 pp ... about half the increase in fiscal income'; after tax +1.4 pp. Their largest 2019 adjustment is 'base groups on number of individuals', -2.9 pp in 2019 vs -1.5 pp in 1960. PS TabFig2018 Table A1: top 1% excl. gains 9.13% (1986) to 13.17% (1988), +4.04 pp, which is 40% of the 1980-2018 rise of +10.20 pp. DINA Oct 2020: +2.92 of +8.39 pp (35%). After 1988 the rise is still +5.21 pp excl. gains and +6.47 pp incl. gains (to 2018). Size of the rise across series: AS +4.4 pp (1979-2019), PSZ DINA +8.39 pp (1980-2019), PS fiscal incl. gains +11.94 pp (1980-2018). Level comparison (C3): PS fiscal 21.96% (2018) vs a 1913-29 mean of 18.30%. DINA pre-tax 18.73% vs 20.41%. WID 2026 19.04% (2019) and 20.73% (2024) vs 20.39%. GMMS (2022) Table 2, 1917-29: mean 15.39%, range 12.84-19.47%. GMMS Figure 12 linked series (their revision plus an Auten-Splinter denominator after 1960): 19.47% (1928), 8.20% (1970), 17.83% (2015). The Auten-Splinter 2017 series plotted there gives 14.12% for 2015, inside the GMMS 1917-29 range. Only the mixed-concept pairing of AS 2019 (13.8%) with DINA 1913-29 (20.41%) puts today about one-third lower. Scripts: ps_series.py, dina_series.py, wid_check.py, critics_series.py; logs in data/.

Why this severity. C2's direction is untouched: inequality rises in every series. C3 survives in nearly all consistent series, including the critics' own linked series. The dispute bears on magnitudes the essay does not state and on one cross-framework comparison. That makes it an important undisclosed caveat with limited demonstrated impact: moderate. For contrast, this project graded the same critique major for Saez & Zucman (2020, JEP), which states a quantified rise.

The authors' response. Piketty & Saez, reply to Reynolds (20 Dec 2006, https://eml.berkeley.edu/~saez/answer-WSJreynolds.pdf): reclassification cannot explain the rise, because the series including realized capital gains also doubled (10.0% in 1980 to 19.8% in 2004) and the top 1% W-2 wage share rose from 6.4% to 11.6%. PS (2003, pp. 3-4) had already acknowledged short-run TRA86 income shifting. Piketty, Saez & Zucman, 'Income Inequality in the United States: A Comment' (16 Sep 2024; revised 19 Aug 2026), argue that AS misallocate untaxed income, notably depreciation-exempt partnership and S-corporation income, and that 'after clarifying these assumptions ... the Auten and Splinter (2024) estimates become similar in level and trend' to PSZ. The dispute is unresolved.

2minorconfirmed · newly identified here · internal inconsistency

The essay says inequality began rising again with Reagan's election in 1980. By the author's own figures, top salaries were already pulling away in the 1970s, though top incomes overall really took off after 1980. It is a small dating slip that does not change the story.

Technical detail and evidence

Why it matters, in detail. Mechanism: series and endpoint choice. Fiscal top-1% shares trough in 1973 or 1976. The pre-1980 increase is 0.43 pp excl. gains (4.1% of the trough-to-2018 rise) and 1.16 pp incl. gains (8.9%). About 1.0 pp of the latter is the 1978-79 jump, consistent with a realization response to the capital-gains tax cut in the Revenue Act of 1978, enacted under Carter. Top 1% and top 0.1% wage shares rose 25% and 48% in 1970-80, 17-22% of their rise to 2011. In DINA (0.5%) and in Auten-Splinter (trough 1981) the 1980 dating is accurate. Direction: the essay overstates the sharpness of a 1980 break, which slightly strengthens the implied link to Reagan (see F3).

Evidence. TabFig2018 Table A1: top 1% excl. gains 7.74% (1973 trough) to 8.18% (1980), 4.1% of the trough-to-2018 rise. Table A3, incl. gains: 8.86% (1976 trough), 8.95% (1978), 9.96% (1979; the 1978-79 jump is consistent with a realization response to the Revenue Act of 1978's capital-gains tax cut), 10.02% (1980), so 8.9% of the rise predates 1980. Table B2 wages: top 1% 5.13% (1970) to 6.43% (1980), +25%; top 0.1% 1.06% to 1.57%, +48%; 22% and 17% of their rises to 2011 predate 1980. DINA Oct 2020 pre-tax top 1%: trough 10.30% (1976), 10.35% (1980), so 0.5% of the rise predates 1980. Auten-Splinter 2017 series (GMMS Fig. 12 data): trough 8.93% in 1981. Striking it Richer (10 Oct 2019), p. 4: 'dramatic increase in top income shares that has taken place since the 1970s'; p. 8: 'a significant fraction of the surge in top incomes since 1970 is due to an explosion of top wages and salaries'. PS (2003, QJE 118(1)), p. 31: 'wage inequality, measured by top fractile wage shares, starts to increase in the early 1970s'. Scripts: ps_series.py, dina_series.py, critics_series.py.

Why this severity. A dating imprecision of a few years. For the headline income-share series only 4-9% (fiscal) and under 1% (DINA) of the post-war rise predates 1980, and the national-income series turn in 1980-81, so the substance of C2 is unaffected. The dating is materially late only for top wage shares, which the essay does not cite.

3minorconfirmed · newly identified here · selective presentation

Before taxes, the richest 1% of Americans now get about as large a share of national income as a century ago. After taxes and government benefits their share is still noticeably smaller than in the 1920s, roughly 14-15% versus 18%. The essay mentions only the before-tax picture.

Technical detail and evidence

Why it matters, in detail. Concept choice: pre-tax (factor) income shares vs post-tax income after taxes and cash and in-kind transfers. With 1913-29 taxes and transfers small, early-century pre- and post-tax levels are close (top 1%: 20.4 vs 18.1). Today's tax-transfer system opens a 3.9-4.3 pp wedge (18.7-19.1 pre-tax vs 14.4-15.1 post-tax), so 'comparable to a century ago' holds pre-tax but not post-tax (-16% to -20% for the top 1%). The post-1980 rise is large on both concepts: post-tax top 1% +6.5 pp over 1980-2019.

Evidence. PSZ DINA Oct 2020 vintage (TC1, post-tax national income, equal-split adults): top 1% 1913-29 mean 18.10% (1920: 16.52%) vs 14.44% in 2019 (-20%); top 10% 42.90% vs 38.19% (-11%). Feb 2022 vintage: top 1% 18.08% vs 15.14% (-16%); top 10% 42.90% vs 38.99% (-9%). WID 2026 (sdiincj992, post-tax disposable income): top 1% 18.10% vs 15.11% (2019) and 14.90% (2024). Pre-tax shares, by contrast (TB1), are comparable: 20.41% vs 18.73% (Oct 2020) and 20.29% vs 19.08% (Feb 2022). The post-tax U-shape is also present: top 1% 7.92% (1980) to 14.44% (2019). Scripts: dina_series.py, wid_check.py.

Why this severity. C3 is literally correct for the measure the essay names, and the post-tax series shows the same U-shape. The omission concerns only the level comparison for disposable income: the top 1% is about 16-20% lower than in 1913-29 and the top 10% about 9-11% lower.

The authors' response. PSZ (2018, QJE 133(2)) publish and emphasise post-tax series. Piketty & Saez (2006 reply to Reynolds) wrote that the top 1% disposable-income share 'has most likely more than doubled since 1980'. No response specific to this essay.

4minorplausible · previously known · overclaiming

The essay implies that Reagan-era, pro-market policies caused the rise in inequality. That is a plausible view, but the author's own research says it cannot be proven, and other economists point to technology and global markets. Part of the measured rise also reflects a change in how income was reported to the tax authorities after the 1986 tax reform.

Technical detail and evidence

Why it matters, in detail. The fiscal series cannot identify causes. The essay's support is temporal coincidence with a 1980 break that is itself approximate (F2). Within the series, policy also affects measurement. Lower top rates after TRA86 moved business income onto Form 1040, adding about 4 pp to the top 1% share excl. gains in 1986-88 (40% of the 1980-2018 rise). That is a policy effect on reported income, not necessarily on economic inequality. Evidence for a policy channel exists elsewhere: cross-country correlations between top-rate cuts and top-share increases in the authors' work, and a rise that persists after 1988 (+5.2/+6.5 pp) and in DINA, which includes retained earnings. Competing explanations (technology, superstar and scale effects: Kaplan & Rauh 2013) are not excluded.

Evidence. PS (2003, QJE 118(1)), p. 3: 'Although our proposed interpretation for the observed trends seems plausible to us, we stress that we cannot prove that progressive taxation and social norms have indeed played the role we attribute to them.' Striking it Richer (10 Oct 2019), p. 8: 'A number of factors may help explain this increase in inequality, not only underlying technological changes but also the retreat of institutions developed during the New Deal and World War II - such as progressive tax policies, powerful unions, corporate provision of health and retirement benefits, and changing social norms regarding pay inequality.' PS (2003), p. 11: 'a significant part of the gain is concentrated in 1987 and 1988 just after the Tax Reform Act of 1986 which sharply cut the top marginal income tax rates'. TabFig2018 Table A1: 4.04 of the 10.20 pp rise in the top 1% share excl. gains over 1980-2018 occurs in 1986-88. Geloso et al. (2022, EJ, conclusion of the accepted manuscript): their gentler revised U-curve 'calls into question the posited causal link' with mid-century tax progressivity.

Why this severity. An interpretive gloss in an interview transcript validated by the author. It does not affect the descriptive claims C1-C3. The causal story is consistent with the authors' published cross-country evidence but is stated more firmly than that evidence, or the original paper's own caveat, supports.

The authors' response. The authors rest the interpretation on cross-country evidence that top shares surged mainly where top tax rates were cut (e.g. Piketty, Saez & Stantcheva 2014, AEJ:EP 6(1)). PS (2003, pp. 3-4) argue that 'most long-run inequality trends are the consequence of real economic change' rather than fiscal manipulation. No response specific to this essay.

Claims examined and rejected

Problems one auditor raised that did not survive checking.

Geloso, Magness, Moore & Schlosser (2022, EJ 132(647)) undermine C1: PS overstate pre-war concentration and the WWII compression, so the 1928-44 top-1% drop is 8.63 pp rather than 12.66 pp. Auditor A examined the same point in its R7. — Not an error of this essay. The revision lowers the early-century level and shifts the compression toward the Depression, but the essay names both the Depression and WWII and gives no magnitude. In GMMS the top 1% share still falls from 18.09% (1929) to 13.25% (1933) and 10.84% (1944), and GMMS state that their corrections do 'not alter the fact that there was a fall and then rise of inequality' (accepted manuscript, p. 34). Their lower 1917-29 levels (top 1% mean 15.39%) put today's PS level (21.96%) above them, which strengthens rather than weakens C3. The critique bites on works that draw a WWII-timing inference: this project's adjudication of Saez's 'Striking it Richer' updates grades it major for that reason. It does not bite on this sentence. Both auditors reached the same conclusion.
Supporting-workbook defect: Table A3 of TabFig2018.xls (and TabFig2018prel.xls) labels the 2018 row '2017' a second time (spreadsheet row 111). Auditor A noted the same slip as an incidental observation. — Confirmed (scripts/ps_series.py detects and repairs the duplicate), but it is a labelling slip in a separate data product. The values are correct (21.959% = data-Fig2 2018), the label is fixed in later vintages, and the chapter prints no number from it. It is not an issue of this chapter.
The IRS's own 2018 top-1% AGI share (20.93%, 18in01etr.xls TAB1) differs from the PS top-1% share (21.96%). — Different estimands, as auditor B itself stated: non-dependent returns and AGI on one side, all potential tax units and PS fiscal income on the other. Rebuilding the PS concept from IRS Table 1.4 gives 22.05% (auditor A's local-Pareto method, re-run here) or 20.97% (the adjudicator's cruder b-coefficient method). On any method the 2018 end-point is 21-22%, so there is no error.
The biographical statement 'Avec Thomas Piketty, rencontré au MIT' may be inaccurate. — Unresolved and immaterial. Piketty was an MIT assistant professor in 1993-95 (Piketty CV), and Saez's MIT PhD followed his 1996 DELTA degree (Saez CV). Two Berkeley-published profiles (California Magazine, 21 Aug 2014; UC Berkeley Research, 2023) date the start of the collaboration to Paris in 1995. No source rules out an earlier meeting at MIT. The chapter is a journalist's transcription of an interview, validated by the author (book acknowledgments), so the wording may be compressed. It has no bearing on any core finding and is not scored; auditor A also declined to score it.
Comparing Auten & Splinter's 2019 pretax top-1% share (13.8%) with PSZ's 1913-29 mean (20.4%) would put today about one-third below a century ago, contradicting C3. — Rejected as a cross-framework comparison: no AS-consistent pre-1960 series exists, and auditor A did not rely on it either. The critics' own juxtaposition, GMMS (2022) Figure 12, pairs a revised pre-1960 series with an Auten-Splinter denominator. It gives 19.47% (1928) against 17.83% (2015), and the Auten-Splinter 2017 national-income series (14.12% in 2015) lies inside the GMMS 1917-29 range (12.84-19.47%). The residual uncertainty is folded into F1.
The 1986-88 jump around TRA86 could explain away the post-1980 rise. — Tested and rejected by auditor B, and confirmed here. Excluding that window, the top 1% share still rises +5.21 pp (excl. gains) and +6.47 pp (incl. gains) from 1988 to 2018. DINA, which includes corporate retained earnings, also shows the rise. The jump bears on magnitude (F1); it does not refute C2.
Wealth-tax, tax-rate and incidence critiques (us-wealth-taxes-12, -13, -16, -18, -20, -21) may apply to the chapter's unread policy sections. — Not assessable. The section headings ('Après l'impôt progressif, un impôt sur les fortunes'; 'L'impôt, un choix de société') and the preface summary (p. 16) show that the chapter argues for capital and wealth taxation, but no specific figure or claim could be read. No issue is assigned on the strength of headings or authorship. Both auditors took the same position.

Audit notes

What reproduced. Reproduced: (i) every PS fiscal-income statistic behind C1-C3, recomputed from TabFig2018 (Feb 2020). Early-century values are identical in the Oct 2019 and June 2026 vintages; the 2018 top 1% incl. gains is 21.76/21.96/21.96% across the three. (ii) PSZ DINA pre- and post-tax statistics (Oct 2020 and Feb 2022 vintages) and WID 2026. (iii) The 2018 end-point from primary IRS SOI Table 1.4, using the PS income definition, tax-unit total and non-filer rule. Auditor A's local-Pareto script, re-run with its column mapping verified, gives 22.05% vs the published 21.96% (top 10% 50.33% vs 50.47%; top 0.1% 10.84% vs 10.84%). The adjudicator's cruder b-coefficient variant gives 20.97%. (iv) The critics' alternatives, read from their own files: GMMS Table 2 and Figure 12 data; AS 2024 Table 1. Every number used by either auditor matched to rounding. Not reproduced: the 1913-29 PS levels from the original SOI reports (GMMS's independent reconstruction was used instead); AS and PSZ national-income series from microdata (confidential); and the unreadable roughly 90% of the chapter.

Limits of this audit. Only about 10% of the chapter could be read: the 259-word opening paragraph, plus the four section headings, the book's acknowledgments and the preface's summary. The policy sections (US primaries, wealth tax, taxation as a societal choice, coronavirus), which make up most of pp. 133-141, are unaudited, and any figures in them (for example wealth-tax revenue or tax rates of the rich) were not seen. The chapter was not purchased (Cairn 'accès conditionnel', 5 EUR), and no open copy was found: Cairn returns HTTP 403, Google Books has no preview, the Internet Archive has no copy, and the publisher's preview stops at p. 23. The chapter is an interview conducted and transcribed by Isabelle Albaret and validated by the author (acknowledgments, front matter), so its wording may reflect the journalist. The 1913-29 PS levels were not rebuilt from the 1913-29 SOI reports; GMMS's published reconstruction, read from their replication workbook, served as the independent check. The Auten-Splinter and PSZ national-income series were not replicated, because they rest on confidential microdata. AS figures come from their published Table 1 and from the AS 2017 series as plotted in GMMS Figure 12. The tax-unit count in the IRS check was taken from PS Table A0. 'Rencontré au MIT' remains unresolved. The QJE-2018 DINA vintage values cited by auditor A were not re-extracted; the Oct 2020 and Feb 2022 vintages were.