The CBI Economics Report: A Methodological Deconstruction

The Report's Context

In February 2025, the British Hair Consortium (BHC) - a coalition including the Hair & Barber Council - published "Securing the Future of UK Hairdressing & Beauty," a report by CBI Economics advocating for VAT reform.

H&BC's press release announcing the report stated:

This is significant. The report is not independent economic analysis that happened to reach conclusions favourable to H&BC. Rather, H&BC acknowledges being "an integral part of the drafting" - suggesting this is commissioned advocacy presented as economic research. The report's own footnotes, set out in the next section, bear this out.

The press release also introduces new financial claims not fully explained in the report itself:

• "A split-rate VAT model could be cost-neutral, or even positive with a gain of £0.75bn by 2030"
• "The gain could rise to £1.5bn if registration levels are restored to their historic levels"

These figures (£0.75bn gain, potentially £1.5bn) appear to be additional projections based on assumptions about preventing de-registrations and increasing registrations. The report's Figures 11 and 12 show the outputs; the assumption that a split rate would "at least halt VAT deregistration" (p.20) is asserted rather than evidenced.

The report makes dramatic claims about sector collapse and enormous tax losses. Its foreword, signed by Gareth Penn as "Chief Executive, Hair & Barber Council," presents these as established facts requiring urgent government intervention.

This analysis examines where the report's inputs came from, and whether its major claims withstand scrutiny against official government data.

What CBI Economics Was Given

The report carries the CBI Economics name, and that name does a great deal of work when the figures are repeated to MPs and journalists. It is therefore worth being precise about which parts of the analysis are CBI Economics' own work on public data, and which parts are inputs supplied by the commissioning body, its member organisations, or a single salon. The report is candid about this in its footnotes; it simply never draws the threads together. The table below does so, quoting the report's own page and footnote references.

Input	Where used	Stated source (report's own words)
Survey of salon owners underpinning the post-Budget scenarios	pp. 5, 7, 12, 23	"a comprehensive survey of 542 salon owners" (p.5) — conducted "as part of the project" with the BHC (p.7)
"Approximately 1,800 responses" post-Budget survey	pp. 5, 7, 9, 24–25	"the BHC's prior research, including a post-budget survey" (p.5); "a survey the BHC carried out following the Autumn Budget" (fn 10)
Labour costs = 60% of turnover; "typical labour %" of 54.2% in the salon-vs-retail tax comparison	pp. 3, 7, 8, 29; Figure 1	"Margaux Salon (2024) The 'Uberisation' of the UK Hair & Beauty Industry: Disguised Employment Practices and Unlawful Competition" (fn 1, fn 3) — a single salon's own paper
Fixed costs of £250,000 in the same comparison	p. 8, Figure 1; p. 29	"Assumes equal costs regardless of retail type. Estimate provided by BHC." (fn 6)
Sector borrowing, "zombie company" count and £266m deficit	p. 9	"A report prepared for the British Hair Consortium by Opus Business Advisory Group" (fn 11)
93:7 services-to-goods split driving the split-rate VAT modelling	p. 20, fn 23	"Data provided by Phorest Salon Software, used by over 2,500 salons in the UK"
Post-Budget employment scenario (−23.1% year one, −37.3% thereafter)	p. 12	"A survey of salon owners conducted for this project indicated…" — i.e. the 542 respondents' stated intentions
Post-Budget apprenticeship scenario (zero by 2027)	p. 23	"we surveyed respondents on how they would respond to the Budget in terms of hiring new apprentices"
Salon opening/closure counts	p. 25, fn 33	"Local Data Company analysis for Wella (2024)"
Historical employment series	pp. 10–13	ONS Annual Population Survey (public data)
Historical VAT-registered turnover series	p. 19	ONS Business Size, Activity and Location (public data)
VAT threshold bunching chart	p. 17	Office for Budget Responsibility (public data, "recreated by CBI Economics")
Historical apprenticeship starts, 2016–2023	p. 23, fn 28	Department for Education starts dataset (public data)

Two things follow. First, the four public-data inputs are exactly that — public — and every one of them is reproduced in our data archive, where anyone can check them. Second, every input that converts those series into a crisis — the labour percentage that makes salons look three times worse off than retail, the fixed-cost line, the debt figures, the behavioural assumptions that drive both forecasts to collapse — came from the client side. The honest description of the document is not "CBI Economics found that…" but "CBI Economics modelled forward the inputs its client supplied."

The report's closing page also carries this line, which is rarely quoted alongside the headline figures:

Gareth Penn's Foreword: The Opening Statement

Penn's foreword presents a series of stark claims that frame the entire report. Let's examine each:

Penn signs as "Chief Executive, Hair & Barber Council." The 1964 Act creates a "Hairdressing Council" with a "Registrar." The title "Chief Executive" suggests corporate executive authority rather than statutory officer of a voluntary register. The report's own "About the BHC" page (p.30) goes further, describing the Council as a "statutory authority" — in the same document that states, on page 9, that "registration is voluntary."

The statistical claims in Penn's foreword are examined below against the report's own stated methodology and against official government data.

Claim 1: The £2.4 Billion VAT Loss

The report's central claim is that VAT system flaws have resulted in "an estimated loss of £2.4 billion in VAT receipts alone." This is the report's headline figure, so it deserves the most careful treatment.

The Report's Actual Methodology

The report does provide a stated methodology (page 19, Figure 10). It is not simply asserted:

We have independently verified the 2009 and 2023 endpoints against the official ONS series for VAT-registered hairdressing and beauty enterprises (SIC 96020) and they check out. This figure is not fabricated, and a method is disclosed for how it was reached. The real problems with it are more specific than "no methodology" — they are set out below.

Problem 1: It Measures a Hypothetical Shortfall, Not an Actual Loss

The entire £11.8bn/£2.4bn figure is the gap between what turnover actually was and what it would have been had it tracked general consumer-price inflation. No money was removed from the Exchequer in the sense the report's language ("lost," "cost the Exchequer") implies — nothing was collected and then taken away. The claim rests on the assumption that turnover should have grown in line with CPI. But turnover reflects sector-specific pricing, service volume and business numbers — not the general price level — and there is no established economic reason a labour-intensive personal-services sector should track economy-wide CPI over a 14-year window that includes a financial-crisis recovery period, a well-documented shift toward self-employment, and a pandemic.

Problem 2: Gross Turnover × 20% Overstates Even the Hypothetical

Applying a flat 20% to the entire turnover gap treats it as though every pound of that hypothetical extra turnover would have flowed straight to the Treasury as VAT. It wouldn't. VAT-registered businesses reclaim input VAT on their costs, so net VAT actually remitted is output VAT minus input VAT — always less than 20% of gross turnover. This overstates any genuine fiscal read-through from the calculation, independent of whether the CPI-tracking premise itself is accepted.

A Separate, Genuinely Measured Figure: the Actual VAT Compliance Gap

A different question — how much VAT has genuinely gone uncollected due to non-compliance in this sector, as distinct from a hypothetical turnover shortfall — was separately put to HMRC via Freedom of Information request. Their response (FOI2024/00849) puts the actual VAT gap for hairdressing, 2009–2023, at £22.7 million. This is not a recalculation of the CBI report's £2.4bn using different data; it answers a genuinely different question (measured non-compliance, not a projected shortfall against a CPI trend). But it's worth setting the two side by side, because the £2.4bn figure has been publicly repeated — including outside this report — in ways that imply it represents this same kind of directly measured, actual loss. It doesn't.

The Response When Challenged

When the underlying data and time period behind the £2.4 billion figure were put to CBI Economics in writing, their response was notable:

The report's headline methodology is now public and can be assessed on its own terms (above). But CBI Economics has still not published the year-by-year data underlying the £11.8bn cumulative gap — only the endpoints and trend lines appear in the report — so the calculation cannot be fully independently reproduced from the published document alone.

Claim 2: The 93% Employment Fall

The report forecasts: "Employment in the sector will fall by 93%, meaning that the Government's Worker Rights Bill will have little impact" (Foreword, p.3; restated in the Executive Summary, p.5, as "Employment plummets to less than 6,500, a 93% reduction from 2024").

This claim has been analysed in detail elsewhere, but briefly:

The report measures the decline in traditional employees (from 86,800 to projected 6,468) and presents this as "employment fall." Their own Figure 5 shows total workforce at 231,000 in 2030, barely changed from 232,700 in 2024.

HMRC data shows total workforce at 398,000 (PAYE + self-employed), having grown 2.1% from 2018-19 to 2022-23.

A Forecast That Can Now Be Checked

The post-Budget scenario is not a vague warning. Page 12 states its first-year assumption precisely: following the October 2024 Budget, survey respondents indicated "reductions to salon-based employment of 23.1% in the first year and 37.3% thereafter," and the forecast assumes all of those workers switch to self-employment. The Budget measures took effect in April 2025. The ONS Annual Population Survey series the report itself uses — SOC codes 1253, 6221 and 6222, defined as "the sector" in footnote 14 (p.11) — now covers that first year, and the report's baseline reproduces from it: page 10 states "232,700 people employed in hairdressing and beauty salons in the UK," comprising "145,200 self-employed workers, or 63%… and 86,800 (37%) who were employees" (footnote 12, citing ONS APS, year ending September 2024). The SOC2020 occupation-by-employment-type table on Nomis gives 235,000 total and 85,400 employees for the same year (the difference is routine ONS revision). Page 11 gives the report's own long-run comparison: "In 2015, employment hit a high of over 300,000 but has since reduced to 232,700… a fall of approximately 23%," inferring that "69,400 workers have either left the profession or entered the hidden economy" (footnote 15) — figures which, now independently reproduced from Nomis (see "The Longer Run," below), check out almost exactly.

UK, SOC 1253 + 6221 + 6222	YE Sep 2024	YE Sep 2025	Change
Total employment	235,000	234,200	−0.3%
Employees	85,400	75,900	−11.1%
Self-employed (implied)	149,600	158,300	+5.8%
Report's post-Budget assumption for employees	86,800	≈ 66,700	−23.1%
of which: hairdressers and barbers (6221), employees	42,100	42,800	+1.7%
of which: beauticians (6222), employees	40,100	26,600	−33.7%

Source: ONS Annual Population Survey via Nomis, "regional – occupation (SOC2020) by sex by employment type," extracted 9 September 2026. Unit group 1253 employee cells are flagged by ONS as unreliable (small sample) and are included for completeness only.

On this year-to-September window, employee numbers fell by roughly half what the report assumed, and total employment barely moved. But this is only one of four ways of cutting the same twelve months, and it is the one that gives the smallest fall; the next section sets the others beside it. On total employment, the point matters: the report's own model expected total employment to be flat (it assumes every lost employee becomes self-employed), so the outturn is consistent with a continued shift in employment model and inconsistent with a collapse in the sector's workforce.

Two qualifications, which anyone quoting this section should carry with it. First, the APS confidence intervals at four-digit occupation level are wide — about ±12,000 on the employee total in each year — so on this window a fall of 9,500 is not statistically significant, and the survey cannot on its own rule out either "no change" or the report's −23%. Second, the composition on this window is unusual: the entire employee decline sits in beauticians (6222), whose employees fell by 13,500 while hairdressers and barbers (6221) rose by 700 (their total-employment figures, employees plus self-employed, moved by +18,400 and −23,800 respectively, which is why the split between the two codes should be treated with caution). That reading does not carry over to the other windows. On the January–December alignment the 6221 employee count is down 13.6% (45,700 to 39,500, an interval of about ±12,000, so not individually distinguishable from no change), and on the April–March and July–June alignments it is down 22.4% and 12.2%. "Hairdressing employees were unchanged" is therefore true of the year-to-September window only, and should not be quoted as a general finding.

Which Twelve Months? Four Windows, Four Answers

In July 2026 the BHC published a CBI Economics briefing, The Collapse of Employment and Apprenticeships in the Hair & Beauty Sector, stating that employees fell from 91,500 in 2024 to 70,100 in 2025 — a 23% fall that "almost exactly aligns" with the survey forecast. Both figures reproduce from the same Nomis table used above, but on calendar-year windows rather than the year-to-September window the original report used for its 86,800 baseline. An earlier version of this page described that switch as a retrospective choice that manufactured a larger fall. Having tested all four alignments of the twelve-month windows, that description was unfair: the calendar-year result is typical of the alignments, and the year-to-September window is the odd one out. The windows for the two years are set out below, followed by every alignment side by side.

Twelve-month window (UK, SOC 1253 + 6221 + 6222)	Total	Employees	95% interval on employees
Oct 2023 – Sep 2024 (basis of the February 2025 report's 86,800)	235,000	85,400	± ~12,400
Jan – Dec 2024 (baseline used in the July 2026 update)	233,100	91,500	± ~12,900
Oct 2024 – Sep 2025	234,200	75,900	± ~11,100
Jan – Dec 2025 (endpoint used in the July 2026 update)	230,200	70,100	± ~10,400

Source: ONS Annual Population Survey via Nomis, occupation (SOC2020) by employment type, extracted 9 September 2026 and re-checked against a fresh extract on 7 October 2026 (all four totals and employee counts reproduce). Totals are the Nomis total in employment, which includes a small "Other" category, so employees plus self-employed can be slightly lower; see the note under the longer-run table. Intervals combine the three occupation codes' published confidence intervals in quadrature. At UK level the 1253 (salon managers) employee cell is flagged unreliable by ONS (sample of 3–9) in the Oct 2023 – Sep 2024 window, the report's baseline; the 2025 windows are not flagged.

Alignment (UK, SOC 1253 + 6221 + 6222, employees)	Baseline year	Following year	Change	Approx. 95% interval on the fall
October – September (the report's own windows)	85,400	75,900	−11.1%	± ~16,600
January – December (the July 2026 update)	91,500	70,100	−23.4%	± ~16,600
April – March	103,500	76,200	−26.4%	± ~18,500
July – June	96,000	74,700	−22.2%	± ~17,700

Source: ONS Annual Population Survey via Nomis, occupation (SOC2020) by sex by employment type, UK, employees, extracts of 26 October 2026. Intervals are approximate: the published 95% intervals for the three occupation codes are combined in quadrature for each year, and then for the two years. Unit group 1253 cells are frequently flagged by ONS as unreliable. The July – June windows are the twelve months to June 2024 and June 2025.

The year-to-September comparison gives a fall of 11.1%, well short of its own interval. The other three give falls of 22.2%, 23.4% and 26.4%, and in each case the fall is larger than the approximate interval around it. The July 2026 update's January–December comparison is neither the largest fall nor the highest baseline: the April–March baseline (103,500) is higher, and it is the lowest of the four endpoints. The update then recomputed the survey forecast on its chosen baseline (91,500 × 0.769 = 70,363, the "BHC Post-2024 Budget Survey Forecast" marker on its Figure 1) so that forecast and outturn coincide; on the original report's baseline the forecast was about 66,700. The fair statement is that the survey-based forecast of a 23% first-year fall is within the range of what the four alignments show, and that the year-to-September window this page previously relied on understates the fall.

What this does not show is a collapse in the sector's workforce. Over the same alignments total employment moves much less than employees: −0.3% on the year-to-September windows and −1.2% on the calendar-year windows. On the July–June alignment the three-code total goes 241,400, 227,200, 246,200 across the twelve months to June 2024, 2025 and 2026, while self-employment goes 145,400, 152,500, 166,100. The employee count in the latest published windows has also partly recovered (75,000 for April 2025 – March 2026; 80,100 for July 2025 – June 2026), though these windows are as noisy as the rest. Of the 21,400 fall in employees on the January–December alignment, about 13,500 (63%) is beauticians (6222), 6,200 (29%) is hairdressers and barbers (6221) and 1,700 (8%) is salon managers (1253). The report and the update both describe this as "hairdressing and beauty", but hairdressers and barbers are not where most of it sits.

The table also shows how much of this is sampling noise. The two 2024 windows share nine of their twelve months and differ by 6,100 employees; the two 2025 windows share nine months and differ by 5,800. Across the eight consecutive windows from April 2021 – March 2022 to January – December 2025 the three-code employee count runs 95,100; 96,300; 87,100; 85,800; 103,500; 91,500; 76,200; 70,100, so the April 2023 – March 2024 baseline is an unusually high one and part of any fall measured from it is a return to the earlier range. Single-window comparisons are fragile in both directions, which is why none of the four should be quoted without the others. The administrative data is a different measure: HMRC's PAYE count of employees in SIC 96020 shows 186,000 for the tax year to March 2025 against 187,000 the year before. That count is not a test of the 2025 fall, because the employer National Insurance changes took effect in April 2025, after that tax year ended. It shows no collapse in payroll employees up to that point, and the comparison with the survey can only be made once HMRC figures for 2025/26 are available.

"The Largest Fall of Any Occupation": Does the Ranking Hold?

The BHC's July 2026 briefing (Table 3) ranks three-digit SOC occupations by the fall in employees and places hairdressing and related services (SOC 622, which is 6221 plus 6222) first, at −22%. Its method, as we read it, compares the average of the January–December 2023 and 2024 windows with January–December 2025. Reproduced from the same Nomis data, that method returns the BHC's figures.

Rank	SOC (3-digit)	Baseline (avg 2023, 2024)	2025	Change
1	622 (hairdressing and related services)	84,350	65,500	−22.3%
2	921	145,200	120,000	−17.4%
3	352	68,750	57,200	−16.8%
4	812	32,900	27,400	−16.7%
5	522	225,500	189,300	−16.1%

Source: ONS Annual Population Survey via Nomis, employees, UK, January–December windows, all three-digit SOC2020 occupations, extracts of 26 October 2026. Our ranking reproduces the BHC's Table 3 (−22%, −17%, −17%, −17%, −16%). These are central estimates with no confidence intervals; the intervals on three-digit cells of this size are substantial.

So the headline ranking is not an error: on the BHC's own method, hair and beauty employees did have the largest percentage fall. It is, however, sensitive to the baseline. Compare January–December 2024 with 2025 alone, and construction operatives (815) fall 23.8% against 23.0% for 622, putting hairdressing second, with 323 (−21.5%), 921 (−20.9%) and 522 (−18.2%) close behind. Compare 2023 with 2025 and 622 (−21.7%) drops to fourth, behind 352 (−30.2%), 356 (−24.1%) and 911 (−22.0%). Construction operatives are a volatile series (91,100 in 2023, 112,400 in 2024, 85,600 in 2025), which is exactly the kind of cell where the choice of baseline decides the ranking. Two further points go with it: the figure counts employees only, while self-employment in the same codes rose, and about 63% of the January–December fall in the three-code total is beauticians, so "hairdressers" overstates where it sits. A claim that hairdressing had one of the largest falls among occupations in 2025 is supported; a claim that it had "the highest of any profession" is supported only on one of three equally reasonable baselines.

The Longer Run: 2006–2026

Everything above tests one year against another. It's worth also asking what the same three occupation codes show over the two decades Nomis makes available, rather than only in the window either side of the report's own baseline — not because a longer run settles the Budget-year question, but because a "stable workforce" framing built on one or two years deserves the same scrutiny this analysis applies to everyone else's numbers.

UK, SOC 1253 + 6221 + 6222 (calendar year)	Total	Employees	Self-employed	Self-employed share
2006	238,600	131,400	107,200	44.9%
2009	252,800	138,300	114,500	45.3%
2012	263,700	134,400	129,300	49.0%
2015 (series peak, both total and employees)	309,300	154,200	155,100	50.1%
2018	289,300	129,000	160,300	55.4%
2021 (SOC2010 basis)	274,800	100,300	174,500	63.5%
2021 (SOC2020 basis, same period)	268,500	95,000	173,500	64.6%
2022	242,800	96,300	146,500	60.3%
2023	225,500	85,800	139,700	62.0%
2024	233,000	91,500	141,500	60.7%
2025 (calendar-year low)	229,500	70,100	159,400	69.5%
Apr 2025 – Mar 2026 (latest)	238,200	75,000	163,200	68.5%

Source: ONS Annual Population Survey via Nomis, "regional – occupation (SOC2010/SOC2020) by sex by employment type," extracted 17 September 2026, all periods available on each basis. SOC2010 covers 2006–2021, SOC2020 from 2021 onward; the two are shown side by side for 2021, the one overlap year. Full annual series and confidence intervals held in the source archive. Total in this table is employees plus self-employed. The Nomis total also includes a small "Other" category (about 100 in Jan–Dec 2024, 700 in Jan–Dec 2025 and 1,600 in Apr 2025 – Mar 2026), so these totals sit slightly below the Nomis totals used in the twelve-month window table above (233,100, 230,200 and 239,800).

Which Employment Measure?

The report's employee figure is the smallest of four available measures of the sector's workforce, and the only one that is survey-based. The other three are administrative counts held by HMRC and ONS for the industry (SIC 96020), obtained by Freedom of Information request and published in full in our business data archive.

Measure	Source and period	Employees	Self-employed
Household survey, three occupation codes (the report's basis)	ONS APS, YE Sep 2024	85,400	~149,600
Business register, all VAT/PAYE enterprises in SIC 9602	ONS Business Population Estimates, March 2025	161,261	—
Payroll administrative data, SIC 96020	HMRC PAYE RTI, 2024-25 (2023-24: 187,000)	186,000	—
Self-assessment tax returns, SIC 96020	HMRC Survey of Personal Incomes, 2023-24 (2021-22 and 2022-23: also 210,000)	—	210,000

The four measures are not directly reconcilable: the APS counts people by stated occupation; the BPE counts employment within registered enterprises; RTI counts PAYE employments and may count an individual more than once; the SPI counts individuals declaring self-employed income. They are shown together because they all describe the same industry and all point the same way.

Three things follow. First, the report's "salon-based employment" is not the salon's employment. The three occupation codes cover stylists, beauticians and salon managers; receptionists, assistants, apprentices coded to general trainee occupations, cleaners and administrative staff are on the salon's payroll, in HMRC's 186,000, and absent from the report's 86,800. Second, the report is not consistent about which base it uses: the 60% labour-cost figure it leads with is a share of the whole payroll, including every role the headcount excludes. Costs are measured on the wide base; the jobs forecast to disappear are measured on the narrow one. Third, the self-employed count from tax returns is 210,000 in each of the last three years available. It is not rising. The report's entire mechanism is an accelerating shift from employment into self-employment, and the administrative series that would show it, doesn't.

Claim 3: Zero Apprenticeships by 2027

The foreword states: "By 2027, there will be no apprenticeships left in the sector, resulting in a £3.2 billion loss in lifetime productivity benefits" (p.3).

How the Number Was Produced

Page 23 explains the method in two steps. First, a pre-Budget scenario: DfE historical starts for 2016 to 2023 (footnote 28) were "forecasted forward," producing a line that "will approach zero by 2029." Second, a post-Budget scenario: survey respondents were asked how they would respond to the Budget "in terms of hiring new apprentices," and the resulting "average reduction" was applied on top, bringing zero forward to 2027. In plain terms: a trend line drawn to the axis, then steepened by what 542 salon owners said they intended to do.

What the DfE Data Actually Shows

The figures below are hairdressing and barbering apprenticeship starts in England, built from the DfE standard codes ST0213, ST0214 and ST1273 plus the legacy frameworks they replaced. Definitions, the combined hair-and-beauty series, Scotland, and a dated corrections log are on the apprenticeship data page.

Academic year	Starts	Note
2015/16	16,080	Series start
2022/23	5,700
2023/24	6,760	Last full year in the report's history; up on the prior year
2024/25	5,250	First full year after the report. The April 2025 NI change applies only to the last four months (April–July 2025); 2025/26 is the first full year under it
2025/26 (Aug–Apr)	3,890	In-year, nine months; DfE "reported to date"; full-year figure due November 2026

Three observations.

• The decline is real. From 16,080 to 5,250 is a fall of 67.3% over the decade to 2024/25 (68.5% on the combined hair-and-beauty basis). Nothing in this analysis disputes that. But roughly 82% of that decline had already happened by 2019/20; the recent years are volatile around a much lower level, not a straight line downward.
• The forecast was made at the wrong moment. The last full year in the report's own history, 2023/24, was up on the year before. A trend line fitted through a series whose most recent movement is upward, and extended to zero, is a choice of method, not a finding.
• The forecast has now been tested and has missed. Eighteen months after publication, the first full academic year after the report, 2024/25, recorded 5,250 starts — down 22%, not on a path to zero within two years. That year includes only about four months after the April 2025 National Insurance change, so it is not a clean test of the Budget measures. The first nine months of 2025/26, the first year fully under them, show 3,890, a figure DfE will revise upward when late registrations are added. A decline that steep is a serious problem for the sector. It is not the report's line.

Reconciling the BHC's Apprenticeship Figures

In reply to our challenge, the BHC's Head of Compliance supplied apprenticeship starts for a wider series than the one used above: hairdressing, barbering and beauty therapy standards and frameworks combined, with nail services excluded. We have reconciled that series to the Department for Education tables released under FOI, and it matches exactly for 2015/16 to 2023/24. It is a legitimate series, and it is wider than the hairdressing-and-barbering series this page uses.

Academic year	Starts, hairdressing, barbering and beauty therapy	Note
2014/15	18,980	Supplied by the BHC; not in our FOI tables, so not independently verified here
2015/16	17,800
2016/17	17,040
2017/18	12,910
2018/19	11,320
2019/20	8,130
2020/21	7,170
2021/22	7,870
2022/23	6,320
2023/24	7,340
2024/25	5,720	Full year
2025/26 (in-year)	4,190	Incomplete year; cannot be compared with a full year

Source: BHC (via its Head of Compliance), reconciled to DfE apprenticeship starts tables supplied under FOI for 2015/16 to 2023/24. The 2024/25 and 2025/26 figures are as supplied by the BHC and have not been separately reconciled.

• The "70%" reproduces. 5,720 in 2024/25 against 18,980 in 2014/15 is a fall of 69.9%. It rests on a start year, 2014/15, that is not in the tables we hold; from 2015/16 the same series falls 67.9%, and on this page's narrower hairdressing-and-barbering series from 2015/16 the fall is 67.3%. The three figures differ by definition and start year, not by error.
• Figure 4 of the July 2026 briefing changes basis part-way through. The values printed on the chart for 2014/15 to 2018/19 (18,979; 17,796; 17,038; 12,904; 11,312) match the wider series to within 10 starts and do not match the hairdressing-and-barbering series (2015/16 is 17,796 on the chart against 16,080 on that series). From 2019/20 onwards every printed value matches the hairdressing-and-barbering series (Table A on our apprenticeship data page) and none matches the wider series: 7,190 for 2019/20 (7,200 on that series, against 8,130 on the wider one), 6,550 for 2020/21 and 7,020 for 2021/22 (both exact), then 5,700, 6,760 and 5,250 for 2022/23 to 2024/25 (exact, against 6,320, 7,340 and 5,720 on the wider series). The change of basis therefore falls between 2018/19 and 2019/20. The chart, headed "Apprenticeship data" and described in the text as hair and beauty apprenticeships, is sourced only to "Department for Education data (2026); British Hair Consortium analysis" and does not say the definition changes. Taken end to end the chart's own figures imply a fall of about 72.3% (5,250 against 18,980), not the 69.9% of a like-for-like comparison. We have not asked the BHC why, and they have not commented publicly.
• The Salon Employers Association's "84%" is 3,050 starts (the figure plotted for 2025/26, marked "incomplete" on Figure 4, and lower than the 3,890 reported to date to April 2026, so it appears to be an earlier in-year cut) against 18,980. The 3,050 is an incomplete in-year figure compared with a full year, which overstates the fall; on a like-for-like basis the full-year fall is about 70%.

The effect of this on the argument is limited but real. The decline in apprenticeship starts is large on every definition, about 67% to 70% over the decade to 2024/25. Most of it happened before the 2024 Budget: on the wider series, 2019/20 (8,130) is already 57% below 2014/15. What the reconciliation does not change is that the report's forecast of zero by 2027 has not materialised, and that 2024/25 includes only the last few months after the April 2025 National Insurance change.

Claim 4: The £14.4 Billion Productivity Loss

The report claims "£3.2 billion loss in lifetime productivity benefits, potentially rising to £14.4 billion by 2030."

The Report's Stated Method

An earlier version of this analysis described the method as undefined. It is not. Page 24 and footnote 30 set it out:

What Is Wrong With It

• It multiplies a counterfactual by a constant. The headcount is not people who existed and lost something; it is the difference between actual starts and a line drawn from 2016. Change the line and the "loss" changes. Under the post-Budget path the line reaches zero, so every start the sector would "otherwise" have had is counted as £67,000 lost — which is why the figure balloons from £3.2bn to £14.4bn.
• Starts are not completions. The DfE productivity figure is for people who complete a Level 2 apprenticeship. The report applies it to starts. Hairdressing achievement rates are well below 100%; the report's own footnote acknowledges it "assumed all starts refer to a Level 2 apprenticeship" but does not adjust for the proportion that never finish.
• It assumes every missing apprentice did nothing else. Footnote 30 says so directly. A 16-year-old who did not start a hairdressing apprenticeship in 2022 is assumed to have acquired no qualification and no productivity gain of any kind, anywhere.
• It is the same construction as the sector's other headline figures. A fixed per-head multiplier applied to a headcount, with the headcount doing all the work. The Salon Employers Association's published membership figures multiply to exactly £20,000 of "economic contribution" per salon at every point since 2022. The method is familiar.

When asked in writing to provide "details on the assumptions driving this projection, particularly in relation to workforce trends, economic inflation, and the potential long-term impacts on productivity," CBI Economics offered a phone call. The stated assumptions in footnote 30 are, in fairness, an answer to part of that question. They are also the reason the figure should not be quoted.

Claim 5: The Hidden Economy

The report references "a rising hidden economy" that will "further deprive the Exchequer of vital employment-related tax revenues."

When challenged to "detail how this was estimated and reconciled with HMRC data," CBI Economics offered a phone call.

The hidden economy is, by definition, unmeasured. Any estimate requires explaining:

• What methodology was used to estimate unmeasured activity
• What evidence suggests it is "rising"
• How much tax revenue this represents
• How this reconciles with official employment and VAT data

The report's own evidence for a rising hidden economy (pp. 11–12) is the 69,400 fall in total sector employment in the APS since 2015, from which it infers that workers "have either left the profession or entered the hidden economy." That is a disjunction, not a measurement. Adding unmeasured tax losses from unmeasured economic activity is convenient for inflating crisis figures, but impossible to verify.

Sample Size and Representativeness

The scenarios that drive the report's forecasts rely on a survey of 542 respondents. For context:

When challenged on "what measures were taken to ensure this sample is representative of the 48,000 UK salons," CBI Economics offered a phone call.

Questions about sample representativeness include:

• Self-selection bias: Did those responding have stronger views about VAT than non-respondents? The survey was conducted with the BHC, whose members were campaigning on VAT at the time.
• Geographic distribution: Does the sample match the regional distribution of salons?
• Business size: Are different size categories proportionally represented?
• Employment models: Are traditional employed salons vs self-employed models proportionally represented?
• Intentions vs behaviour: The forecasts convert what respondents said they would do into what the sector did. Stated intentions in a survey about a tax rise are not a reliable guide to subsequent hiring — and the outturn data in Claims 2 and 3 is now the test of that.

A 1.13% sample can be representative if properly stratified, but the methodology for ensuring representativeness hasn't been disclosed.

The VAT Reform Proposal: Who Actually Pays?

The report proposes "split-rate VAT" with services at 10% and goods at 20%. Penn's foreword states: "To be clear, this reform is not a tax cut."

However:

• Current VAT rate on salon services: 20%
• Proposed VAT rate on salon services: 10%
• This is, definitionally, a tax rate cut

The report models this rate cut against several VAT-registration scenarios (see its Figures 11 and 12). When challenged on how a rate cut would be funded, CBI Economics offered a phone call rather than a written answer.

The Threshold Reduction: Whose Proposal Is It?

Correction to earlier version: this section previously stated that the CBI report itself proposed lowering the VAT registration threshold from £90,000 to £30,000 as part of a "revenue neutral" mechanism. On review, the CBI report's own text only refers generally to the government "consider[ing] lowering the registration threshold to further induce registration and prevent disaggregation," without stating a specific figure.

The specific £30,000 figure comes from a different, related source: the Salon Employers Association (SEA) — one of the six BHC member organisations, and the body co-founded by Toby Dicker — whose own published campaign objectives state, as one of five aims, a call "to adopt a similar VAT model as Ireland, and pay a split rate of VAT... plus to campaign for a reduced VAT threshold of circa £30k." (Source: salonemployersassociation.co.uk, "Our key objective" / "Aims and ideas"; Wayback Machine capture held in our source archive.)

This is worth being precise about: it is not part of the CBI report's stated methodology, but it is the explicit, quantified position of a BHC member organisation actively campaigning for the same split-rate VAT reform the report models. The analysis below considers what a threshold cut of this kind, if adopted alongside the CBI-modelled rate cut, would mean in practice.

The Arithmetic of "Revenue Neutrality," If Both Elements Were Combined

If SEA's threshold proposal were adopted alongside the CBI-modelled rate cut, the combined package would be structured so that already-VAT-registered, typically larger salons receive a rate cut, funded in part by bringing a large number of currently-exempt, typically smaller salons into VAT registration for the first time.

Impact on Small Businesses, If This Package Were Adopted

For a salon currently turning over £40,000-£89,000 (below the current threshold, above SEA's proposed £30k threshold):

Aspect	Current (£90k threshold)	Under SEA's Proposed £30k Threshold
VAT Registration	Not required	Mandatory
VAT Added to Prices	No VAT charged	10% VAT added (at the proposed split rate)
Price Competitiveness	10-20% price advantage vs VAT-registered competitors	Advantage eliminated
Administrative Burden	No VAT returns	Quarterly VAT returns required
Cash Flow	No VAT payment delay	Must collect and remit VAT quarterly

Small businesses brought into registration under this scenario would face:

• Price increase requirement: Must add VAT to prices or absorb it in margins
• Competitive disadvantage: Lose current price advantage over larger, already-registered salons
• Administrative costs: VAT registration, accounting, quarterly returns
• Cash flow impact: Collecting VAT from customers, holding it, remitting quarterly
• Compliance risk: Penalties for late filing or errors

Who Would Benefit From This Combined Package?

Alternative Approaches Not Considered

Other potential VAT reforms that wouldn't shift costs onto small businesses in this way:

• Raise the threshold: Increase £90k to £120k+, giving more businesses VAT exemption
• Graduated approach: Sliding scale between threshold and full rate
• Input VAT on labour: Allow VAT-registered businesses to reclaim VAT on wages (like other inputs)
• Sector-specific threshold: Higher threshold for labour-intensive services

These alternatives would help small businesses rather than potentially funding a rate cut for larger ones. Neither the CBI report nor SEA's published campaign materials appear to set out analysis of alternatives along these lines.

Mandatory Registration Advocacy

Penn's foreword advocates for "a mandatory register of all hairdressing professionals," stating: "Without a register, VAT avoidance and disguised employment will continue unchecked" (p.4).

However, the Hairdressers (Registration) Act 1964 creates only a voluntary register. Section 2 states registration is for persons "who apply in the prescribed manner to be so registered." The report itself acknowledges this on page 9, footnote 9: "The Hairdressers Registration Act 1964 provides for a UK register of qualified hairdressers, but registration is voluntary."

Mandatory registration would require new primary legislation. Using a report on VAT reform to advocate for fundamentally changing the statutory basis of the Hair & Barber Council raises questions about the report's objectives.

The Pattern Across Claims

Claim	Report Says	Verification Shows	When Challenged
VAT Loss	£2.4 billion	Genuine methodology, but measures a hypothetical shortfall vs a CPI trend, not an actual loss; overstated further by applying the VAT rate to gross turnover	Phone call offered
Employment Fall	93% decline	Post-Budget year-on-year test: employees −11% on the report's own year-to-September window but −22% to −26% on the other three alignments (−23% on the BHC's calendar years), the fall exceeding its approximate interval on those three; total employment −0.3% to −1.2% on the two windows compared; about 63% of the January–December fall is beauticians, and hairdressers and barbers are flat only on the year-to-September window. Longer 2006–2026 run shows a real, decade-long contraction (mostly 2015–2021) and a steady employment-model shift — not a sudden post-Budget collapse	N/A (in report)
Apprenticeships	Zero by 2027	Trend line plus survey intentions; outturn 5,250 (2024/25) and 3,890 (2025/26 to April) — not on the report's path	N/A (in report)
Productivity Loss	£14.4 billion	Stated method: £67k × counterfactual headcount; assumes all starts complete, all Level 2, none take other training	Phone call offered
Hidden Economy	Rising tax losses	Inferred from a fall in APS employment; no estimation method	Phone call offered
Sample Size	542 respondents	Representativeness unverified; publicly restated as "2,000+"	Phone call offered
Key inputs	Presented as CBI Economics analysis	Labour %, fixed costs, debt data, services split and both surveys supplied by BHC, its members, or a single salon (see table above)	N/A (in footnotes)

Across these claims, the pattern is consistent: where a methodology is disclosed, it measures something narrower or more hypothetical than the language used to present it; where the inputs are disclosed, they come from the client; and where no methodology is disclosed, CBI Economics has declined to provide one in writing when asked.

What This Means for Policy

The report is being used to advocate for significant policy changes:

• Split-rate VAT (10% on services)
• A lower VAT registration threshold (per SEA's separately published campaign position)
• Mandatory professional registration
• Presented as urgent crisis requiring immediate intervention

Its figures have since been repeated in Budget submissions, in written evidence to Parliament, in MPs' constituency correspondence, and in social media campaigns by salon owners — frequently with the CBI Economics name attached as though it were the source of the inputs rather than the modeller of them. When policy advocacy is built on statistics that overstate what the underlying data supports, it raises fundamental questions:

• Are policymakers receiving accurate information about sector challenges?
• Is the case for intervention being overstated?
• Are legitimate sector issues being undermined by exaggerated claims?
• Who benefits if these policies are enacted based on an overstated case?

The Real Sector Challenges

This analysis doesn't claim the sector faces no challenges. Legitimate concerns exist:

• VAT threshold effects: Creating bunching and distortions
• Disguised employment: Tax compliance issues in chair rental models
• Business costs: Rising rent, rates, utilities affecting viability
• Skills development: A 67% fall in apprenticeship starts over a decade is a genuine pipeline problem, whatever the cause
• Employment model shifts: Implications for worker protections
• A real, decade-long contraction: Total workforce across these three occupation codes is around a quarter below its 2015 peak — a genuine long-run decline that predates, and is separate from, the report's post-Budget forecast

These deserve serious policy discussion. They don't require overstated crisis statistics to warrant attention.

When economic analysis is commissioned by organisations with clear policy objectives (VAT cuts, mandatory registration), built on inputs those organisations supply, and presents figures that overstate what the underlying data shows, it undermines the credibility of legitimate sector advocacy.