Inequality: Are we really 'all in this together'?

Why it matters, in detail. On the briefing's own statistic (P90/P50 of full-time gross weekly earnings, NES/ASHE linked), the ratio rose from 1.649 to 1.971 (+0.32, +19.5%). The key point implies +1.4 (+87.5%), a 4.2-fold overstatement on a common starting value (4.35-fold using its own 1.6), and a 2013 level 52% too high. A charitable reading, that 'three' means the mean pay of the whole top decile relative to the median, does not rescue it. That statistic is plausibly of this order recently (Bell & Van Reenen 2014, Table 2: the top 10% take about 25% of the ASHE weekly wage bill, i.e. about 2.5 times mean pay). But it was already about twice mean pay in 1975 (19.8% share), so the 1.6 comparator is not like-for-like and the implied near-doubling is spurious.

Evidence. EA030 PDF p.2 (key point 2) vs printed p.4 and Figure 2. Chartbook of Economic Inequality, January-2015 vintage (UK sheet, 'Earnings at top decile as % median'): 164.91 (1978), 197.1 (2013). ONS ASHE 1997-2026 selected estimates, Table 5 (full-time gross weekly pay, April 2013): P90 GBP1,020.0 / P50 GBP517.4 = 1.971; all employees 909.8/415.3 = 2.19; full-time P90/P10 = 3.55; no standard percentile ratio equals 3. Adjudicator digitisation of Figure 2: 165.0 (1978) and 197.1 (2013). The error is repeated in the EA034 compilation (PDF pp.27 and 175; body on p.178 correct); the wage section is absent from the abridged LSE blog version of 2 May 2015. scripts/adjudicator_checks.py -> data/adjudicator_checks.json (wage_ratio).

Why this severity. The headline magnitude of a core finding, as stated in the key points, is wrong by far more than a quarter: the 2013 level is 52% too high and the 1978-2013 rise about 4.2-4.4 times too large. The direction (rising top-half wage inequality) survives and the correct figure is in the body, so the issue is not critical.

Why it matters, in detail. In both countries units are ranked by taxable capital income. On that basis the US top-1% share (59.0% in 2010; 63.5% on the passive-income definition) exceeds the UK's 39-40% by about 20-24 percentage points, or 1.5-1.6 times. Unit differences (US tax units vs UK individuals) and the inclusion of pass-through profits in the US measure cannot plausibly close a gap of this size, since the passive-only US series is even higher. The sentence therefore overstates UK capital-income concentration relative to the US; the UK level itself is unaffected.

Evidence. Saez-Zucman (2016) appendix Table B22 (top 1% of tax units ranked by taxable capital income, capital gains excluded): 34.0% (1978), 49.3% (1999), 59.0% (2010), 62.9% (2012). Table B22b (passive capital income: dividends, interest, rents, estate and trust income): 63.5% (2010), 67.9% (2012). The cited NBER WP 20625 says on p.8 that the top 0.1% alone had 33% in 2012, and on p.22 that the top-1% taxable capital-income share gained 29 points over 1978-2012. UK: Alvaredo-Atkinson-Morelli Table L1 gives 39.3% (WID WP 2017/2) or 39.9% (J. Public Econ. 2018 appendix) for 2010.

Why this severity. This is a clear factual misstatement of a secondary comparative claim, contradicted by the author's own cited data by about 20 percentage points. It does not change the UK figure (about 40%) or the conclusion that UK capital income is highly concentrated.

Why it matters, in detail. Pre-1997 values sit on the old (ESA95-based PZ) wealth basis and later values on the ESA 2010 balance sheet, so the series embeds a level shift of about 6-12% around 1997-98. Consistently linked, the 1970s-to-2013 increase is about +230-250 points rather than about +270 (from roughly 300% to 567%), i.e. 8-16% smaller. Part of the steeper late-1990s rise is genuine to the 2014 ONS vintage rather than an artefact of the splice. Later revisions (current WID: 482% in 2013) lower the endpoint further; that reflects instability in the data vintage rather than an error at the time.

Evidence. EA030 printed p.7 and Figure 4 (adjudicator digitisation, data/fig4_digitised.txt). The ratio of Figure 4 to PZ Table UK.1 beta averages 0.933 in 1970-96 and 1.048 in 1998-2010 (0.923 in 1996, 0.985 in 1997, 1.035 in 1998). ONS National Balance Sheet bulletin (20 Nov 2014): 'Estimates are available for the period 1997 to 2013.' Change 1996-99: Figure 4 +1.31, PZ +0.84, current WID +0.63. Re-linking pre-1997 values at the 1997 step (x1.066) gives a 1970s-to-2013 rise of 2.49 instead of the 2.69 shown (8% smaller); linking by the mean pre/post shift gives 2.32 (16% smaller). ONS NBS 2014 put household and NPISH net worth at GBP8,455bn at end-2013; the national income implied by Figure 4 with mid-year wealth (about GBP1,472bn) is plausible. Current WID (October 2026 bulk): 1970s mean 2.88, 2013 4.82, with its own 1994-95 break (3.92 to 3.35). scripts/adjudicator_checks.py (wealth_income_ratio).

Why this severity. This affects a secondary result. The direction (a large, housing-driven rise) is robust in every vintage. But the undisclosed splice inflates the rise by roughly 8-16% and visibly exaggerates the late-1990s jump, and the headline 2013 level is about 15% above today's revised estimate.

Why it matters, in detail. The 67-78% figure is finance's share of the 1999-2008 increase in the top-1% share of the wage bill, with earners ranked by pay. Applying it to the top-1% share of total fiscal income, where adults are ranked by total income and the denominator includes non-wage income, is not supported by BVR's tables, and it says nothing about 1979-99. The claim that the rise was 'mainly due to labour income' remains supported by other evidence (Alvaredo-Atkinson-Morelli 2018 appendix Fig. L1; Delestre, Kopczuk, Miller & Smith 2024).

Evidence. Bell & Van Reenen (2014), Economic Journal 124(574): F1-F21. The abstract reads: 'Rising bankers' bonuses accounted for two-thirds of the increase in the share of the top 1% after 1999'. Table 3 (SPI, 1999-2008): finance contributes 1.6 of the 2.4 pp rise in the top-1% share of the wage bill (67%); the ASHE annual-pay equivalent is 1.4 of 1.8 (78%). The Table 3 note restricts SPI to 'pay from employment net of benefits and foreign earnings'. The SPI sample (p.F9) is taxpayers aged 18-65 with pay above the PAYE threshold. The salary/bonus split (pp.F9-F10) begins in 2002.

Why this severity. The briefing hedges ('appears to be') and repeats the source's own compressed abstract wording. The qualitative point that finance pay drove much of the 1999-2008 top-end rise stands. The finance share of the all-income top-1% rise is not identified, so the impact on C1 is small and unquantified.

Why it matters, in detail. In the displayed CPI-base EUROMOD results, P90-95 gains about 0.2% and P95-100 loses about 0.7%. The text assigns these the other way round, slightly understating the loss at the very top. Citing panel (b) for panel (a) matters because the two comparators differ in the extent of losses (bottom half vs bottom 15% losing; aggregate neutral vs +0.9%).

Evidence. Hills (2015), CASE SPCC WP11, p.39: 'those in the penultimate twentieth broke even, and the top twentieth made a small loss on average'. Figure 9(a) on p.38 shows the poorest twentieth at about -2.7% and 'overall the changes were neutral'; Figure 9(b) shows the poorest at about -1.3% and an average gain of 0.9%. EA030 Figure 8 shows vingtile 19 at about +0.2%, vingtile 20 at about -0.7% and All at about +0.2% (visual reading; Auditor A's digitisation gives +0.20/-0.72/+0.16). Hills Table 5 (p.29) is 'Income Support levels in relation to poverty thresholds'; Hills Figure 7 (p.35) is 'Annualised rate of change in income by decile group'. The LSE blog version (2 May 2015) repeats the reversal and the 9(b) caption.

Why this severity. The reversal concerns only the top two twentieths. 'Mainly regressive' and the bottom-half losses are correctly reported for the panel actually shown. The mislabels are documentation errors.

Why it matters, in detail. Under the earnings-uprated base, which Hills describes as neutral towards inequality, losses are confined to the bottom three twentieths (the poorest about -1.3%) and the average household gains about 0.9%. Under the January-2010 base, before the 50p rate, the top twentieth loses about as much as the bottom. Combined with F6's reversal, the briefing's account of the top end leans towards the reading that the richest did fine.

Evidence. Hills (2015) pp.38-39: with Figure 9(b), households as a whole gain 0.9% and the changes are 'regressive until the very top'. Hills p.41: 'If the 2014-15 system was compared with the system in place in January 2010, as in the IFS analysis, then our analysis would also suggest that the top twentieth lost as much from direct tax and benefit changes as the bottom twentieth'. Also p.41: the DHS/Hills, IFS and HM Treasury analyses all find the changes regressive from the bottom tenth to the seventh or eighth tenth, and Hills calls the May-2010 base 'the more natural one'. Auditor B documented the same sensitivities from De Agostini, Hills and Sutherland (2014, pp.19-20) in its assumptions table.

Why this severity. 'Mainly regressive' is robust across comparators and baselines. Only the extent of bottom-half losses and the position of the very top depend on these choices, and the briefing's choice is defensible.

Why it matters, in detail. About 89% of remittance-basis claimants paid no charge in 2012-13 (5,000/46,700 = 10.7% paid). The omitted 2007-08 peak shows that the post-2008 fall in non-dom numbers coincided with the introduction of the charge.

Evidence. Channel 4 FactCheck, 8 April 2015 (EA030 footnote 1; archived copy of 14 April 2015 with HMRC/FOI tables): 83,000 non-doms in 1997/98, 'a high of 140,000 in 2007/08', 114,800 in 2012-13. Remittance-basis claims were 46,700 in 2012-13. Remittance-basis-charge payers were 5,410, 5,100, 5,600, 5,500 and 5,000 in 2008-09 to 2012-13; the article says 'Only about 5,000 non-doms a year pay these charges'. HM Treasury (6 April 2015) confirms the GBP30k/60k/90k schedule.

Why this severity. This is a factual slip in policy background. It understates rather than overstates the regime's generosity, so it does not favour the briefing's argument, and no inequality estimate depends on it.

Why it matters, in detail. The maximum couple threshold rises by GBP350,000 (+53.8%), not +100%. The gain is capped at GBP140,000 of tax per couple (40% of GBP350,000) and tapers away above GBP2m estates. The briefing's statement that the largest beneficiaries are GBP1-2m estates is correct.

Evidence. HMRC inheritance-tax policy note (2015), 'Current law': the nil-rate band is GBP325,000 and has been transferable between spouses since 9 October 2007, 'so that the nil-rate band can be up to GBP650,000'. Conservative manifesto 2015: 'increasing the effective Inheritance Tax threshold for married couples and civil partners to GBP1 million, with a new transferable main residence allowance of GBP175,000 per person'. 1,000,000/650,000 = 1.538.

Why this severity. This overstates the generosity of a policy proposal (+100% claimed vs +54%). The distributional direction, with gains concentrated on GBP1-2m estates, is right, and no core inequality finding depends on it.

Why it matters, in detail. Measured from 1980, the IFS/HBAI Gini rose about 8.0-8.4 points (about +31%), so 'around 10' overstates the change by about 20-25%. Measured from 1977-78 (23.9-24.0), the rise is about 9.7-10.1.

Evidence. Chartbook (January-2015 vintage), IFS Gini: 23.9 (1977), 24.0 (1978), 25.7 (1980), 34.1 (2011); the current Chartbook gives 33.66 for 2012. Adjudicator digitisation of Figure 5: 25.7 (1980) and 33.8 (2012), a rise of 8.1. The ONS ETB disposable-income Gini rises less (28.6 to 33.3, per Auditor A). Auditor B computed the same IFS values but did not grade the discrepancy. The LSE blog version repeats the sentence.

Why this severity. The qualitative claim, a large rise in disposable-income inequality concentrated in the 1980s, is intact. The overstatement (about 2 of 8 points) is a rounding-level imprecision in a narrative statement.

Why it matters, in detail. GBP100 of allowance equals GBP20 of basic-rate tax for a fully affected taxpayer. The benefit cap's effects are already inside the 2014-15 EUROMOD simulation shown, so listing it as a future change double-counts it as a forward-looking risk.

Evidence. HM Treasury, 'Tax changes coming into effect 6 April 2015' and 'Tax and tax credit rates and thresholds for 2015-16': personal allowance GBP10,600 (from GBP10,000). DWP press release of 20 December 2012: the cap applies from April 2013 and nationwide by summer 2013. Hills (2015) p.16 lists the GBP26,000 cap among reforms already made. The LSE blog version repeats both statements.

Why this severity. These are small factual and chronological errors in policy description. No inequality estimate in the briefing depends on them; Figure 8 ends in 2014-15.

Why it matters, in detail. Valuing in-kind services at cost and deducting indirect taxes moves the top/bottom quintile ratio from 5.6 (disposable cash income) to 3.8 (final income). A reader who takes 'net income' to mean disposable income, the concept used in Figures 5-6, would overstate cash redistribution by about 30%.

Evidence. ONS, The Effects of Taxes and Benefits on Household Income 2012/13 (26 June 2014), main points and Table 2: original income GBP81,300 vs GBP5,500 ('almost 15 times'); final income GBP59,900 vs GBP15,600 ('reduced to four-to-one'). Auditor B's computations from the same table: disposable income 63,628/11,434 = 5.56; post-tax 54,488/7,946 = 6.86. The EA030 p.8 text follows the ONS wording almost verbatim, without citation.

Why this severity. The numbers are correct for the ONS final-income concept, and the body says 'after all taxes and benefits'. The issue is an ambiguous label in the key point, not a numerical error.

Why it matters, in detail. (a) is ambiguous rather than wrong if 'fell in 2009-10' is read as 'between the 2009 and 2010 data points'. (b) reflects the lagged effect of the crisis on top incomes plus the reversal of forestalling. (c) is unsupported by the evidence shown and conflicts with the key points.

Evidence. WID/Chartbook top-1% share: 15.42 (2009) to 12.55 (2010). Hills (2015) p.36: inequality 'fell sharply between 2009-10 and 2010-11 ... and was then relatively flat over the following two years'. IFS Gini (Figure 5): 33.8 (2010), 34.1 (2011), 33.7-33.8 (2012). EA030 key point 5 and printed pp.2, 8-9, 11. The ONS ETB disposable Gini rose in 2012/13 and fell back in 2013/14 (Auditor A); the briefing does not present this. The LSE blog version repeats (a) and (c).

Why this severity. These are descriptive imprecisions about timing plus one unsupported sentence in the conclusions; no number or core conclusion depends on them.

Why it matters, in detail. None of these changes a reported magnitude. Items (3) and (8) are conceptual mislabels of the income definition: the series is fiscal income including taxable transfers and excluding capital gains.

Evidence. (1) Further reading cites Bell & Van Reenen (2014) as 'Bankers' Pay and Extreme Wage Inequality in the UK', Economic Journal 124(574): F1-387. The EJ article is 'Bankers and their bonuses', F1-F21; the cited title belongs to their 2010 CEP working paper, of which the EJ article is 'a revised version' (BVR 2014, title-page footnote). (2) Figure 1 is titled '1970 to 2011' but plots 2012: the digitised last point, 12.70, matches WID 2012. (3) Figure 1 is labelled 'pre-tax and transfer' income, although the UK fiscal series includes taxable transfers such as state pensions. (4) The Figure 7 note says 'spending from direct taxes, tax credits and benefits'; Hills Figure 1(b) is 'Real change in spending on social security and tax credits'. (5) Table 1 says 'Age 18-24' (as in Machin EA024) while the text says 'aged 25 and younger'. (6) 'Wealth and Asset Survey' should be 'Wealth and Assets Survey'. (7) 'Workers with an undergraduate degree': Wyness (EA026) says 'higher education'. (8) Key point 1 lists capital gains among the capital-income contributors, but the UK top-share series excludes capital gains (Chartbook header; noted by Auditor B). (9) The 15:1/4:1 passage and the Gini footnote follow the ONS ETB bulletin almost verbatim without citation.

Why this severity. These are presentation and documentation errors; none affects a number used in the argument.

Why it matters, in detail. The figure understates 1999-2004 by up to 1.1 pp. This makes the period look flat (12.5 to 12.9) rather than about 13.2-13.5 with a peak in 2000. The 1997-2007 rise is +3.4 pp on either vintage.

Evidence. Adjudicator digitisation of Figure 1 (data/fig1_digitised.txt) for 1999-2004: 12.50, 12.64, 12.69, 12.26, 12.11, 12.88. BVR (2014) Table 1: 12.5, 12.7, 12.7, 12.3, 12.1, 12.9, and 13.9 for 2009. Chartbook January-2015 vintage and current WID (sfiinci992): 13.24, 13.51, 13.39, 13.03, 13.24, 13.30, and 15.42 for 2009. All other years agree with WID within about 0.1 pp, apart from points on the steep 2009-10 segment.

Why this severity. No numerical claim in the text depends on 1999-2004; the 12.5-15.5% range and the long-run rise are unaffected.

Why it matters, in detail. In the published SPI-based series, individuals are ranked by investment income and the control total is 90% of national-accounts household rent, dividends and interest. On that series the late-1990s top-1% share averages about 29% and the 2010 share 39-40%, a rise of about 10 pp. The 2015 slides may have used a preliminary control total, so a misreport cannot be established without them. The control-total caveat (implied top-10% shares of 90-99% in 2007-09) suggests the levels in some years may be overstated.

Evidence. AAM, WID.world WP 2017/2, Appendix Table L1: 25.4 (1996), 28.4 (1997), 31.8 (1998), 31.8 (1999), 36.6 (2003), 39.3 (2010). AAM, J. Public Econ. 2018 online appendix, Table L1: 25.2, 28.4, 31.3, 31.3, 35.2 (2003), 39.9 (2010). AAM WP 2017/2 (p.75) warns that the national-accounts control total 'may be underestimated'. Neither auditor nor the adjudicator could locate the 9 March 2015 slides.

Why this severity. The 2010 level (about 40%) is right. The discrepancy concerns the starting point and, if anything, understates the rise (about +10 pp rather than +5 pp), which works against the briefing's argument.

Why it matters, in detail. With a like-for-like US figure of about 17.5-19%, the mid-point between France/Spain (about 8%) and the US is about 13-13.5%, still within the UK's 12.5-15.5% range. The 2012 US value excluding gains is a retiming-affected year (cf. corpus optimal-tax-behavior-21 on the 2013 US tax increase).

Evidence. Saez, TabFig2013prel (the vintage current in early 2015): US top 1% excluding capital gains 17.45 (2010), 17.47 (2011), 18.88 (2012), 17.54 (2013); including gains 19.86, 19.65, 22.83, 20.08. Chartbook January-2015 vintage (older US vintage): 19.34 for 2012 excluding gains. The Chartbook UK column header reads 'individuals, excluding capital gains; tax units before 1991'. France 7.91 (2009); Spain 8.24 (2010).

Why this severity. The qualitative ranking (the UK between continental Europe and the US) survives with any of the US figures; only the rounding of the US comparator is affected.

Why it matters, in detail. Fiscal-income top shares respond to rate changes through retiming and reclassification, the UK analogue of the US debate over the 1986 Tax Reform Act (us-income-01, us-income-17). The 1988 step and the 1990 unit break (at most about 1 pp) together bound the reform-related component at roughly 1-1.5 pp of a 9.5 pp rise.

Evidence. WID GB sfiinct992: 7.78 (1987) to 8.63 (1988), +0.85 pp, against a 1979-89 average of +0.27 pp a year. The 1989-90 rise of +1.13 pp spans the disclosed unit break. Bell & Van Reenen (2014) show the 1999-2008 rise is mirrored in ASHE payroll data, which are less exposed to such reporting margins.

Why this severity. Even attributing the whole 1988 excess to reporting would remove only about 0.6 pp of the roughly 9.5 pp rise over 1979-2007 (about 6%). Payroll data corroborate the later rise.