Skip to content

Publishing work and public data

When publishing employers' pay indicators move in different directions

Five of eleven employers in a fixed reporting panel show opposing changes across three pay indicators between the 2017 and 2025 snapshots. Six show movement closer to zero on all three. The distinction matters because a falling average pay gap, a changing median gap and a changing distribution across pay quarters answer different questions. None alone describes an employer's whole workforce, proves equal pay for equal work or measures an author's income.

Published
Share
Share this page
EmailCite & reuse this study
Key findings as text
Reusable imagesWhen publishing employers' pay indicators move in different directions — 11: Employers in the fixed main panel. Seventeen baseline entities were screened; eleven meet all eight-cycle identity, industry, size and data rules.Download sharing image (JPG) ↓

Key findings

11

Employers in the fixed main panel

Seventeen baseline entities were screened; eleven meet all eight-cycle identity, industry, size and data rules.

5 of 11

Opposing indicator movements

At least one absolute distance increased and another decreased; a reported tie is retained separately.

6 of 11

All three distances became smaller

Directional agreement does not establish equal pay, a universal progress verdict or unchanged workforce composition.

10

Employers in the nine-cycle sensitivity

Requiring the suspended 2019–20 report changes membership; the 2017 and 2025 endpoints stay the same.

A comparison built around reported employers

This study begins with the seventeen corporate employers whose 2017–18 records in the UK government's gender pay gap service included the exact industry code 58110, Book publishing. It follows company numbers and the service's employer identifiers, preserving leading zeros and reported names. Eleven employers satisfy the same identity, industry, size and data requirements in eight specified reporting cycles. Their eighty-eight employer-cycle observations form the main panel.

This is a deliberately bounded frame. An employer can report book publishing alongside other activities, and its pay figures cover the reported employer rather than an isolated books division. The Economist Newspaper Limited and Tate Enterprises therefore remain in the frame when they meet the frozen rules. Their presence does not establish that all their employees produce books. Conversely, an employer missing from this panel has not thereby ceased publishing or failed a workplace test.

The latest complete reporting cycle available when the data were captured was 2025–26. For these private corporate reports, that label concerns the snapshot on 5 April 2025, not wages earned throughout 2026. Earlier cycle labels are interpreted in the same way. The source files are retrospective administrative records downloaded together; older submissions may have been amended. We are comparing the historical figures present in that captured version of the service, not claiming to have preserved every original submission.

Three measures, three different questions

The mean hourly pay gap compares men's and women's average hourly pay, using men's average as the denominator. The median gap compares the two middle hourly-pay values. A positive reported gap places men's pay above women's on the relevant measure; a negative value reverses that ordering. Neither calculation matches employees doing the same job. Differences can reflect the distribution of occupations, hours, grades and other characteristics that these aggregate records do not identify.

The third indicator is the reported female share of the highest hourly-pay quarter minus the female share of the lowest quarter. If those shares are 60% and 80%, the difference is minus twenty percentage points. This is a contrast between two separately ranked pay groups. It is not the percentage of all women who occupy the top quarter, the proportion of women in management, or a comparison between men's and women's individual promotion chances.

The hourly figures and quarter shares concern full-pay relevant employees under the reporting framework. The headcount used to determine reporting obligations is a different construct. The source supplies broad employer-size bands, not the exact analytical sample sizes needed to combine these reports into a workforce-weighted estimate. We therefore give each employer one observation when describing the panel. No national publishing pay gap is reconstructed from the size bands.

Reading direction without confusing signs

The comparison rules were fixed before the middle-year files were acquired or pay changes calculated. For each indicator, we retain the signed 2017 and 2025 values, then calculate how its absolute distance from zero changed. A negative change means a smaller reported distance; a positive change means a larger one. For the quarter-share difference, approaching zero means that the two female shares became more similar. It does not establish equal pay or equal representation throughout the organisation.

This distinction is essential when an hourly gap is negative. Moving from minus two to minus seventeen is a decrease in the signed number but an increase in its distance from zero. A gap can also cross zero, changing which group has the higher reported pay, while finishing closer to zero than it began. The public endpoint table preserves both signs and magnitudes so that the direction classification cannot conceal these changes.

We call a three-indicator pattern opposing only when at least one distance increases and another decreases. All three decreasing, or all three increasing, is fully concordant. Three zeros would be unchanged. Other patterns containing zeros are tie-containing and unidirectional, rather than automatically opposing. An opposing pattern can itself contain a tie. All twenty-seven possible sign combinations appear in the downloadable results, including combinations with no observed employers. These are descriptive categories, not significance tests or a composite workplace score.

What the opposing patterns contain

Bloomsbury illustrates a straightforward disagreement between two hourly measures. Its reported mean gap moves from 23.30% to 17.57%, a reduction of 5.73 percentage points. Its median moves from 17.20% to 17.45%, an increase of 0.25 points. Meanwhile, the female top-minus-lower quarter difference moves from minus 25.00 to minus 17.39 points, reducing its distance from zero by 7.61 points. Quoting only the mean would omit the different movement in the median.

Pearson Education shows why a reported tie needs its own treatment. Its mean gap is 8% at both endpoints. Its median rises from 3% to 6%, while the quarter-share difference moves from minus ten to minus seven points. The mean's unchanged reported distance does not make the whole pattern unchanged: the other two measures move in opposing directions. Nor does the tie establish that the underlying, unrounded mean gap stayed exactly constant.

Macmillan Publishers International shows the importance of negative gaps. Its mean moves from minus 2% to minus 17%, increasing absolute distance by fifteen points. Its median moves from minus 34% to minus 31%, reducing distance by three points. Its quarter-share difference rises from nineteen to twenty-five points, increasing distance by six. Describing both hourly changes merely as increases or decreases would obscure that one moves toward zero and the other away from it.

Penguin Books provides a different complication. Its median gap crosses from minus 7.80% to plus 5.03%. The signed increase is 12.83 points, while absolute distance falls by 2.77 points. Its quarter-share difference also crosses zero, from plus 14.60 to minus 14.95 points, leaving a slightly larger absolute difference. Its mean gap rises from 6.70% to 15.89%. The distance classification is useful precisely because the original signed values remain beside it; a single arrow would not explain these crossings.

Tate Enterprises completes the five opposing cases. Its mean and median gaps increase in absolute distance, while its quarter-share difference moves closer to zero. These five examples do not establish five instances of unfair treatment. They identify five cases in which the fixed indicators support different directional descriptions of the same pair of reporting snapshots. The appropriate next step for a careful reader is to examine the employer's scope and explanatory material, rather than replace the missing workforce detail with a causal story.

What agreement does and does not establish

The six fully concordant employers are The Economist Newspaper Limited, Hachette UK Limited, HarperCollins Publishers Limited, John Wiley & Sons Limited, the company reported in 2025 as Springer Nature Limited, and The Random House Group Limited. Each has a smaller absolute endpoint value on all three indicators. This is agreement in direction, not proof that the gaps have disappeared or that every part of the organisation experienced the same change.

For example, Hachette's mean gap falls from 29.70% to 20.8% and its median from 24.70% to 15.2%. Both remain positive. HarperCollins' corresponding figures move from 16.10% to 11.4% and from 10.40% to 8.2%. These comparisons support the specific statement that the published distances narrowed. They do not identify the pay of a particular editor, establish the salaries available in a vacancy, or show whether people in equivalent roles received equal treatment.

Agreement also says nothing by itself about the path between endpoints. The eight-cycle panel retains the intervening observations so that a smooth progress narrative can be checked against actual reported snapshots. A line joining 2018 and 2020 spans two years because the suspended 2019 reporting cycle is omitted from the main panel. That interval is shown at its real width; no 2019 value is interpolated or disguised as another ordinary annual observation.

Identity is part of the result

The endpoint files each contain seventeen records carrying the book-publishing code, but only twelve company-number and employer-identifier pairs appear in both sets. Equal annual counts therefore do not imply that the same employers are being compared. The additional requirement for consistent coverage throughout the eight main cycles leaves eleven. A comparison that simply averages the seventeen endpoint records without linking entities would mix within-employer changes with changes in which employers appear.

Names alone would introduce another problem. Company number 00785998 and employer identifier 14802 appear as Macmillan Publishers Limited in the baseline and Springer Nature Limited in the latest report. The matching identifiers support the administrative link used here. They do not prove that ownership, activities, employees or reporting boundaries stayed constant. The public ledger preserves the reported name in each cycle instead of silently treating a familiar brand as a stable statistical unit.

Penguin Books Limited and The Random House Group Limited likewise remain separate reported legal entities in this analysis. Their records are not merged into an assumed publishing group total. Without exact employee counts and compatible payroll populations, averaging or combining their gaps would not recover a group's actual gap. This matters to readers who recognise group and imprint names more readily than corporate identifiers: the relevant question is which employer the source actually describes.

The exclusions remain visible

Six baseline employers do not enter the main panel. Butterworths and Elsevier continue to have records under their baseline identifiers but do not retain the exact book-publishing code in the later captured cycles. They fail this study's sector rule; the result is not a finding that they stopped publishing books. The strict code rule prevents an analyst from selectively retaining well-known companies after seeing their outcomes, but it also narrows what the panel represents.

Architectural Association Publications, Music Sales Group and Wolters Kluwer (UK) lack baseline-company records in specified later source files. Scholastic lacks records in three required middle cycles despite appearing at the endpoints. These are observations about the captured reporting files, not explanations for absence. They do not identify a closure, a threshold change, a corporate reorganisation or a reporting breach. The ledger keeps the affected cycle and the precise observed reason separate from any interpretation.

This visible attrition is part of the study's usefulness. A worker, editor or reporter can see whether a company disappeared because the exact industry token changed or because no baseline-company record was found. Neither case is repaired by substituting a similarly named employer. The excluded records remain in the audit trail, while the outcome comparison uses only the fixed qualifying panel. The resulting eleven employers are a selected set, not a representative sample of everyone employed in British publishing.

The suspended cycle changes membership

The government suspended enforcement and the expectation of reporting for the 2019–20 cycle during the pandemic. Requiring that cycle in every complete history would select employers that nevertheless submitted it. Our main comparison therefore uses the eight non-suspended cycles from 2017–18 through 2025–26. The later 2020–21 cycle is retained, with the separately documented extension to enforcement timing acknowledged. Service flags showing a late submission are not treated as evidence of misconduct, because later edits can also produce that flag.

Before calculating outcomes, we specified a sensitivity comparison requiring all nine cycles. Ten of the eleven main-panel employers qualify. Tate Enterprises lacks the suspended-cycle record and drops out. The sensitivity result is four opposing patterns and six fully concordant patterns. Both calculations still compare the 2017 and 2025 snapshots. The change from five opposing employers to four is consequently a change in panel membership, not a new estimate of a pandemic effect or a different endpoint period.

This distinction prevents a superficially reassuring completeness rule from doing hidden work. More annual records do not automatically create a less selective analysis when one year had a different reporting expectation. The eight-cycle and nine-cycle versions answer related but different coverage questions. Neither recovers the experiences of employers below the reporting threshold, non-reporting entities, freelance workers or people whose pay is outside the full-pay calculation.

An employer median is not an industry wage gap

The panel's unweighted median of reported mean gaps is 16.40% in 2017 and 13.8% in 2025. Its median of reported median gaps moves from 12.90% to 8.2%. These figures summarise the middle employer in a fixed set; they do not compare pooled men's and women's pay. A small qualifying employer and a much larger one each supply one reported percentage. The public data do not support substituting broad size-band midpoints for actual employee weights.

The unweighted median of the signed quarter-share differences moves from minus twelve to minus 13.6 percentage points. That panel summary must also be kept separate from the eleven employer-level changes. The employer occupying the middle position can differ between snapshots. A median of signed differences, a median of absolute differences and a count of entities moving closer to zero are different operations. We retain the employer rows so that readers need not mistake one for another.

The figures are not accompanied by sampling confidence intervals or claims of statistical significance. They are selected, self-reported administrative aggregates, with differing reported decimal precision and no specified sampling design for this panel. Calculations preserve the reported decimals. Small differences, including Bloomsbury's 0.25-point median movement, remain visible without being called statistically meaningful. Exact zeros refer to the published numbers, not a measured absence of underlying change.

A reference for better questions

Research on publisher pay gaps predates this study. Clark and Zuccala's 2025 work examines longstanding median gaps among science publishers, and employers already discuss pay-quarter representation in their own reports. This analysis claims no first discovery of publisher gaps or of quarter-share arithmetic. Its contribution is the reproducible combination of a fixed corporate frame, an explicit exclusion ledger and a joint reading of three indicators whose directions can differ.

A practical reading sequence begins with the company number and snapshot date, then checks the signed endpoints before interpreting the direction matrix. It asks whether the source covers the intended employer, whether missing cycles affect the comparison, and whether the three indicators tell the same directional story. An employer's accompanying explanation may clarify a change in activities or workforce composition, but this dataset alone cannot verify such a mechanism or attribute a measured amount to it.

Historical category consistency is another limit. The current government preparation guidance, updated in 2026, specifies how sex should be recorded. That current instruction does not establish how every employer classified every person in the earlier snapshots. We retain the source's male and female field labels without claiming that historical recording practices were independently verified. No individual employee information, contact identity or responsible-person name is included in the public evidence.

The downloadable records and code make the comparison inspectable rather than turning the five opposing patterns into a universal verdict. They support precise statements about named reported employers and captured snapshots. They leave questions about equal work, individual pay, promotions, organisational causes and the wider publishing workforce open wherever the source cannot answer them. That separation lets a reader use the information for a better-informed question without giving a single headline more authority than its underlying measure can carry.

Opposing changes in three absolute-distance indicators

Fixed eleven-employer panel; private corporate snapshots on 5 April 2017 and 5 April 2025. Hourly gap percentages use the male-pay denominator; changes are percentage points. Quarter difference compares female shares in top and lower ranked pay quarters. Signed values and crossings are retained; absolute-distance change is not a universal progress score.

Full chart and source
Opposing changes in three absolute-distance indicators. Fixed eleven-employer panel; private corporate snapshots on 5 April 2017 and 5 April 2025. Hourly gap percentages use the male-pay denominator; changes are percentage points. Quarter difference compares female shares in top and lower ranked pay quarters. Signed values and crossings are retained; absolute-distance change is not a universal progress score.

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

Signed endpoints and changes in absolute distance from zero
Reported employer in 2025Company numberMean gap 2017 (%)Mean gap 2025 (%)Change in absolute mean gap (pp)Median gap 2017 (%)Median gap 2025 (%)Change in absolute median gap (pp)Female top-minus-lower share 2017 (pp)Female top-minus-lower share 2025 (pp)Change in absolute quarter-share difference (pp)
BLOOMSBURY PUBLISHING PLC0198433623.3017.57−5.7317.2017.450.25−25.00−17.39−7.61
ECONOMIST NEWSPAPER LIMITED(THE)0023638332.5013.8−18.7029.5017.2−12.30−38.00−23.4−14.60
HACHETTE UK LIMITED0202017329.7020.8−8.9024.7015.2−9.50−25.90−18.5−7.40
HARPERCOLLINS PUBLISHERS LIMITEDSC02738916.1011.4−4.7010.408.2−2.20−7.00−3−4.00
JOHN WILEY & SONS LIMITED0064113221.1014.4−6.7021.5015.5−6.00−21.90−13.6−8.30
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED02063302−2.00−1715.00−34.00−31−3.0019.00256.00
SPRINGER NATURE LIMITED0078599818.0017.34−0.6612.909.17−3.73−22.40−19.41−2.99
PEARSON EDUCATION LIMITED008728288.0080.003.0063.00−10.00−7−3.00
PENGUIN BOOKS LIMITED008615906.7015.899.19−7.805.03−2.7714.60−14.950.35
TATE ENTERPRISES LTD031739751.804.212.413.007.614.61−12.001.9−10.10
THE RANDOM HOUSE GROUP LIMITED0095400916.407.7−8.7013.901.3−12.60−7.504.0−3.50

Fixed eleven-employer panel; private corporate snapshots on 5 April 2017 and 5 April 2025. Hourly gap percentages use the male-pay denominator; changes are percentage points. Quarter difference compares female shares in top and lower ranked pay quarters. Signed values and crossings are retained; absolute-distance change is not a universal progress score.

Fixed eleven-employer panel; private corporate snapshots on 5 April 2017 and 5 April 2025. Hourly gap percentages use the male-pay denominator; changes are percentage points. Quarter difference compares female shares in top and lower ranked pay quarters. Signed values and crossings are retained; absolute-distance change is not a universal progress score.

Download chart (SVG) ↓ · Download chart (PNG) ↓

Eight snapshots for the fixed eleven-employer panel

Reported corporate snapshots, not annual wage flows. The main panel excludes the suspended 2019–20 cycle; no 2019 value is imputed. Female shares concern each ranked hourly-pay quarter separately. Names are source-reported for that cycle; identical identifiers do not prove an unchanged workforce.

Full chart and source
Eight snapshots for the fixed eleven-employer panel. Reported corporate snapshots, not annual wage flows. The main panel excludes the suspended 2019–20 cycle; no 2019 value is imputed. Female shares concern each ranked hourly-pay quarter separately. Names are source-reported for that cycle; identical identifiers do not prove an unchanged workforce.

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

All eighty-eight observations in the fixed main panel
Reported employerCompany numberSnapshot dateMean hourly gap (%)Median hourly gap (%)Female share of top quarter (%)Female share of lower quarter (%)Top minus lower female share (pp)
Bloomsbury Publishing PLC019843362017-04-0523.3017.2058.1083.10−25.00
Bloomsbury Publishing PLC019843362018-04-0522.9021.7056.7081.00−24.30
BLOOMSBURY PUBLISHING PLC019843362020-04-0515.611.759.978.3−18.4
BLOOMSBURY PUBLISHING PLC019843362021-04-0519.314.86483.7−19.7
BLOOMSBURY PUBLISHING PLC019843362022-04-0519.220.562.5784.88−22.31
BLOOMSBURY PUBLISHING PLC019843362023-04-0516.512.762.983.33−20.43
BLOOMSBURY PUBLISHING PLC019843362024-04-0518.319.363.6884−20.32
BLOOMSBURY PUBLISHING PLC019843362025-04-0517.5717.4563.0880.47−17.39
ECONOMIST NEWSPAPER LIMITED(THE)002363832017-04-0532.5029.5024.0062.00−38.00
ECONOMIST NEWSPAPER LIMITED(THE)002363832018-04-0529.3029.2029.7066.90−37.20
ECONOMIST NEWSPAPER LIMITED(THE)002363832020-04-0522.119.330.856.0−25.2
ECONOMIST NEWSPAPER LIMITED(THE)002363832021-04-0518.722.435.262.0−26.8
ECONOMIST NEWSPAPER LIMITED(THE)002363832022-04-052120.43663−27
ECONOMIST NEWSPAPER LIMITED(THE)002363832023-04-0516.816.33861.4−23.4
ECONOMIST NEWSPAPER LIMITED(THE)002363832024-04-0517.319.438.264.3−26.1
ECONOMIST NEWSPAPER LIMITED(THE)002363832025-04-0513.817.238.561.9−23.4
HACHETTE UK LIMITED020201732017-04-0529.7024.7056.6082.50−25.90
HACHETTE UK LIMITED020201732018-04-0530.4020.1056.7079.20−22.50
HACHETTE UK LIMITED020201732020-04-0521.917.665.478.8−13.4
HACHETTE UK LIMITED020201732021-04-0521.518.164.079.0−15.0
HACHETTE UK LIMITED020201732022-04-0522.0017.7063.881.2−17.4
HACHETTE UK LIMITED020201732023-04-0522.91764.884.4−19.6
HACHETTE UK LIMITED020201732024-04-0522.217.364.583.2−18.7
HACHETTE UK LIMITED020201732025-04-0520.815.264.983.4−18.5
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892017-04-0516.1010.4057.0064.00−7.00
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892018-04-0516.5010.2055.0064.00−9.00
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892020-04-0515.38.861.075.0−14.0
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892021-04-0511.14.464.064.00.0
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892022-04-0510.82.96667−1
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892023-04-0511.94.266642
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892024-04-05113.56568−3
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892025-04-0511.48.26568−3
JOHN WILEY & SONS LIMITED006411322017-04-0521.1021.5044.2066.10−21.90
JOHN WILEY & SONS LIMITED006411322018-04-0523.1022.7046.0070.00−24.00
JOHN WILEY & SONS LIMITED006411322020-04-0520.126.442.069.0−27.0
JOHN WILEY & SONS LIMITED006411322021-04-0516.818.447.566.4−18.9
JOHN WILEY & SONS LIMITED006411322022-04-0520.422.14767−20
JOHN WILEY & SONS LIMITED006411322023-04-0518.918.147.567.0−19.5
JOHN WILEY & SONS LIMITED006411322024-04-0516.817.74868−20
JOHN WILEY & SONS LIMITED006411322025-04-0514.415.550.964.5−13.6
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022017-04-05−2.00−34.0065.0046.0019.00
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022018-04-05−5.00−28.0066.0041.0025.00
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022020-04-05−2−25634518
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022021-04-050.2−23.264.440.823.6
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022022-04-05−2−19655015
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022023-04-05−13−32684424
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022024-04-05−15−30684127
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022025-04-05−17−31714625
MACMILLAN PUBLISHERS LIMITED007859982017-04-0518.0012.9048.4070.80−22.40
SPRINGER NATURE LIMITED007859982018-04-0515.009.5049.4068.90−19.50
SPRINGER NATURE LIMITED007859982020-04-0515.17.362.060.91.1
SPRINGER NATURE LIMITED007859982021-04-0516.510.847.072.2−25.2
SPRINGER NATURE LIMITED007859982022-04-0516.674.445.5670−24.44
SPRINGER NATURE LIMITED007859982023-04-0518.169.9047.7470−22.26
SPRINGER NATURE LIMITED007859982024-04-0516.989.541.2867.19−25.91
SPRINGER NATURE LIMITED007859982025-04-0517.349.1747.8067.21−19.41
PEARSON EDUCATION LIMITED008728282017-04-058.003.0051.0061.00−10.00
PEARSON EDUCATION LIMITED008728282018-04-0515.009.0052.0065.00−13.00
PEARSON EDUCATION LIMITED008728282020-04-0511.08.051.061.0−10.0
PEARSON EDUCATION LIMITED008728282021-04-05985361−8
PEARSON EDUCATION LIMITED008728282022-04-05665460−6
PEARSON EDUCATION LIMITED008728282023-04-0511115367−14
PEARSON EDUCATION LIMITED008728282024-04-057.54.05460−6
PEARSON EDUCATION LIMITED008728282025-04-05865360−7
PENGUIN BOOKS LIMITED008615902017-04-056.70−7.8063.2048.6014.60
PENGUIN BOOKS LIMITED008615902018-04-0522.609.6064.3084.60−20.30
PENGUIN BOOKS LIMITED008615902020-04-0523.213.566.888.9−22.1
PENGUIN BOOKS LIMITED008615902021-04-0520.311.968.085.5−17.5
PENGUIN BOOKS LIMITED008615902022-04-0525.215.168.288.2−20.0
PENGUIN BOOKS LIMITED008615902023-04-0524.615.068.388.5−20.2
PENGUIN BOOKS LIMITED008615902024-04-0522.810.772.482.9−10.5
PENGUIN BOOKS LIMITED008615902025-04-0515.895.0370.3785.32−14.95
Tate Enterprises Ltd031739752017-04-051.803.0052.0064.00−12.00
Tate Enterprises Ltd031739752018-04-052.201.5056.0053.602.40
TATE ENTERPRISES LTD031739752020-04-054.81.248.440.48.0
TATE ENTERPRISES LTD031739752021-04-053.77.450.767.9−17.2
TATE ENTERPRISES LTD031739752022-04-056.69.851.6167.39−15.78
TATE ENTERPRISES LTD031739752023-04-0565.556.563−6.5
TATE ENTERPRISES LTD031739752024-04-056.2213.3853.6070.40−16.80
TATE ENTERPRISES LTD031739752025-04-054.217.6161.559.61.9
THE RANDOM HOUSE GROUP LIMITED009540092017-04-0516.4013.9045.4052.90−7.50
THE RANDOM HOUSE GROUP LIMITED009540092018-04-057.60−1.7050.3045.604.70
THE RANDOM HOUSE GROUP LIMITED009540092020-04-0511.80.953.645.48.2
THE RANDOM HOUSE GROUP LIMITED009540092021-04-059.42.555.358.0−2.7
THE RANDOM HOUSE GROUP LIMITED009540092022-04-059.5−0.357.455.42.0
THE RANDOM HOUSE GROUP LIMITED009540092023-04-056.6−6.059.355.93.4
THE RANDOM HOUSE GROUP LIMITED009540092024-04-052.7−5.861.355.45.9
THE RANDOM HOUSE GROUP LIMITED009540092025-04-057.71.367.663.64.0

Reported corporate snapshots, not annual wage flows. The main panel excludes the suspended 2019–20 cycle; no 2019 value is imputed. Female shares concern each ranked hourly-pay quarter separately. Names are source-reported for that cycle; identical identifiers do not prove an unchanged workforce.

Reported corporate snapshots, not annual wage flows. The main panel excludes the suspended 2019–20 cycle; no 2019 value is imputed. Female shares concern each ranked hourly-pay quarter separately. Names are source-reported for that cycle; identical identifiers do not prove an unchanged workforce.

Download chart (SVG) ↓ · Download chart (PNG) ↓

Data & downloads

Scroll horizontally to see all columns.

Signed endpoints and changes in absolute distance from zero
Reported employer in 2025Company numberMean gap 2017 (%)Mean gap 2025 (%)Change in absolute mean gap (pp)Median gap 2017 (%)Median gap 2025 (%)Change in absolute median gap (pp)Female top-minus-lower share 2017 (pp)Female top-minus-lower share 2025 (pp)Change in absolute quarter-share difference (pp)
BLOOMSBURY PUBLISHING PLC0198433623.3017.57−5.7317.2017.450.25−25.00−17.39−7.61
ECONOMIST NEWSPAPER LIMITED(THE)0023638332.5013.8−18.7029.5017.2−12.30−38.00−23.4−14.60
HACHETTE UK LIMITED0202017329.7020.8−8.9024.7015.2−9.50−25.90−18.5−7.40
HARPERCOLLINS PUBLISHERS LIMITEDSC02738916.1011.4−4.7010.408.2−2.20−7.00−3−4.00
JOHN WILEY & SONS LIMITED0064113221.1014.4−6.7021.5015.5−6.00−21.90−13.6−8.30
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED02063302−2.00−1715.00−34.00−31−3.0019.00256.00
SPRINGER NATURE LIMITED0078599818.0017.34−0.6612.909.17−3.73−22.40−19.41−2.99
PEARSON EDUCATION LIMITED008728288.0080.003.0063.00−10.00−7−3.00
PENGUIN BOOKS LIMITED008615906.7015.899.19−7.805.03−2.7714.60−14.950.35
TATE ENTERPRISES LTD031739751.804.212.413.007.614.61−12.001.9−10.10
THE RANDOM HOUSE GROUP LIMITED0095400916.407.7−8.7013.901.3−12.60−7.504.0−3.50

Fixed eleven-employer panel; private corporate snapshots on 5 April 2017 and 5 April 2025. Hourly gap percentages use the male-pay denominator; changes are percentage points. Quarter difference compares female shares in top and lower ranked pay quarters. Signed values and crossings are retained; absolute-distance change is not a universal progress score.

Scroll horizontally to see all columns.

All eighty-eight observations in the fixed main panel
Reported employerCompany numberSnapshot dateMean hourly gap (%)Median hourly gap (%)Female share of top quarter (%)Female share of lower quarter (%)Top minus lower female share (pp)
Bloomsbury Publishing PLC019843362017-04-0523.3017.2058.1083.10−25.00
Bloomsbury Publishing PLC019843362018-04-0522.9021.7056.7081.00−24.30
BLOOMSBURY PUBLISHING PLC019843362020-04-0515.611.759.978.3−18.4
BLOOMSBURY PUBLISHING PLC019843362021-04-0519.314.86483.7−19.7
BLOOMSBURY PUBLISHING PLC019843362022-04-0519.220.562.5784.88−22.31
BLOOMSBURY PUBLISHING PLC019843362023-04-0516.512.762.983.33−20.43
BLOOMSBURY PUBLISHING PLC019843362024-04-0518.319.363.6884−20.32
BLOOMSBURY PUBLISHING PLC019843362025-04-0517.5717.4563.0880.47−17.39
ECONOMIST NEWSPAPER LIMITED(THE)002363832017-04-0532.5029.5024.0062.00−38.00
ECONOMIST NEWSPAPER LIMITED(THE)002363832018-04-0529.3029.2029.7066.90−37.20
ECONOMIST NEWSPAPER LIMITED(THE)002363832020-04-0522.119.330.856.0−25.2
ECONOMIST NEWSPAPER LIMITED(THE)002363832021-04-0518.722.435.262.0−26.8
ECONOMIST NEWSPAPER LIMITED(THE)002363832022-04-052120.43663−27
ECONOMIST NEWSPAPER LIMITED(THE)002363832023-04-0516.816.33861.4−23.4
ECONOMIST NEWSPAPER LIMITED(THE)002363832024-04-0517.319.438.264.3−26.1
ECONOMIST NEWSPAPER LIMITED(THE)002363832025-04-0513.817.238.561.9−23.4
HACHETTE UK LIMITED020201732017-04-0529.7024.7056.6082.50−25.90
HACHETTE UK LIMITED020201732018-04-0530.4020.1056.7079.20−22.50
HACHETTE UK LIMITED020201732020-04-0521.917.665.478.8−13.4
HACHETTE UK LIMITED020201732021-04-0521.518.164.079.0−15.0
HACHETTE UK LIMITED020201732022-04-0522.0017.7063.881.2−17.4
HACHETTE UK LIMITED020201732023-04-0522.91764.884.4−19.6
HACHETTE UK LIMITED020201732024-04-0522.217.364.583.2−18.7
HACHETTE UK LIMITED020201732025-04-0520.815.264.983.4−18.5
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892017-04-0516.1010.4057.0064.00−7.00
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892018-04-0516.5010.2055.0064.00−9.00
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892020-04-0515.38.861.075.0−14.0
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892021-04-0511.14.464.064.00.0
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892022-04-0510.82.96667−1
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892023-04-0511.94.266642
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892024-04-05113.56568−3
HARPERCOLLINS PUBLISHERS LIMITEDSC0273892025-04-0511.48.26568−3
JOHN WILEY & SONS LIMITED006411322017-04-0521.1021.5044.2066.10−21.90
JOHN WILEY & SONS LIMITED006411322018-04-0523.1022.7046.0070.00−24.00
JOHN WILEY & SONS LIMITED006411322020-04-0520.126.442.069.0−27.0
JOHN WILEY & SONS LIMITED006411322021-04-0516.818.447.566.4−18.9
JOHN WILEY & SONS LIMITED006411322022-04-0520.422.14767−20
JOHN WILEY & SONS LIMITED006411322023-04-0518.918.147.567.0−19.5
JOHN WILEY & SONS LIMITED006411322024-04-0516.817.74868−20
JOHN WILEY & SONS LIMITED006411322025-04-0514.415.550.964.5−13.6
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022017-04-05−2.00−34.0065.0046.0019.00
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022018-04-05−5.00−28.0066.0041.0025.00
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022020-04-05−2−25634518
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022021-04-050.2−23.264.440.823.6
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022022-04-05−2−19655015
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022023-04-05−13−32684424
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022024-04-05−15−30684127
MACMILLAN PUBLISHERS INTERNATIONAL LIMITED020633022025-04-05−17−31714625
MACMILLAN PUBLISHERS LIMITED007859982017-04-0518.0012.9048.4070.80−22.40
SPRINGER NATURE LIMITED007859982018-04-0515.009.5049.4068.90−19.50
SPRINGER NATURE LIMITED007859982020-04-0515.17.362.060.91.1
SPRINGER NATURE LIMITED007859982021-04-0516.510.847.072.2−25.2
SPRINGER NATURE LIMITED007859982022-04-0516.674.445.5670−24.44
SPRINGER NATURE LIMITED007859982023-04-0518.169.9047.7470−22.26
SPRINGER NATURE LIMITED007859982024-04-0516.989.541.2867.19−25.91
SPRINGER NATURE LIMITED007859982025-04-0517.349.1747.8067.21−19.41
PEARSON EDUCATION LIMITED008728282017-04-058.003.0051.0061.00−10.00
PEARSON EDUCATION LIMITED008728282018-04-0515.009.0052.0065.00−13.00
PEARSON EDUCATION LIMITED008728282020-04-0511.08.051.061.0−10.0
PEARSON EDUCATION LIMITED008728282021-04-05985361−8
PEARSON EDUCATION LIMITED008728282022-04-05665460−6
PEARSON EDUCATION LIMITED008728282023-04-0511115367−14
PEARSON EDUCATION LIMITED008728282024-04-057.54.05460−6
PEARSON EDUCATION LIMITED008728282025-04-05865360−7
PENGUIN BOOKS LIMITED008615902017-04-056.70−7.8063.2048.6014.60
PENGUIN BOOKS LIMITED008615902018-04-0522.609.6064.3084.60−20.30
PENGUIN BOOKS LIMITED008615902020-04-0523.213.566.888.9−22.1
PENGUIN BOOKS LIMITED008615902021-04-0520.311.968.085.5−17.5
PENGUIN BOOKS LIMITED008615902022-04-0525.215.168.288.2−20.0
PENGUIN BOOKS LIMITED008615902023-04-0524.615.068.388.5−20.2
PENGUIN BOOKS LIMITED008615902024-04-0522.810.772.482.9−10.5
PENGUIN BOOKS LIMITED008615902025-04-0515.895.0370.3785.32−14.95
Tate Enterprises Ltd031739752017-04-051.803.0052.0064.00−12.00
Tate Enterprises Ltd031739752018-04-052.201.5056.0053.602.40
TATE ENTERPRISES LTD031739752020-04-054.81.248.440.48.0
TATE ENTERPRISES LTD031739752021-04-053.77.450.767.9−17.2
TATE ENTERPRISES LTD031739752022-04-056.69.851.6167.39−15.78
TATE ENTERPRISES LTD031739752023-04-0565.556.563−6.5
TATE ENTERPRISES LTD031739752024-04-056.2213.3853.6070.40−16.80
TATE ENTERPRISES LTD031739752025-04-054.217.6161.559.61.9
THE RANDOM HOUSE GROUP LIMITED009540092017-04-0516.4013.9045.4052.90−7.50
THE RANDOM HOUSE GROUP LIMITED009540092018-04-057.60−1.7050.3045.604.70
THE RANDOM HOUSE GROUP LIMITED009540092020-04-0511.80.953.645.48.2
THE RANDOM HOUSE GROUP LIMITED009540092021-04-059.42.555.358.0−2.7
THE RANDOM HOUSE GROUP LIMITED009540092022-04-059.5−0.357.455.42.0
THE RANDOM HOUSE GROUP LIMITED009540092023-04-056.6−6.059.355.93.4
THE RANDOM HOUSE GROUP LIMITED009540092024-04-052.7−5.861.355.45.9
THE RANDOM HOUSE GROUP LIMITED009540092025-04-057.71.367.663.64.0

Reported corporate snapshots, not annual wage flows. The main panel excludes the suspended 2019–20 cycle; no 2019 value is imputed. Female shares concern each ranked hourly-pay quarter separately. Names are source-reported for that cycle; identical identifiers do not prove an unchanged workforce.

Evidence & method

How we calculated it

  1. The baseline contains all seventeen corporate records with exact SIC 58110 in the 2017–18 source CSV. The main panel requires a unique unchanged CompanyNumber and EmployerId pair, the exact industry token, a reported size band of 250 or more, and valid required figures in every main cycle.
  2. The main cycles are 2017–18, 2018–19 and 2020–21 through 2025–26. The suspended 2019–20 cycle is used only in the prespecified nine-cycle sensitivity. For these corporate reports, cycle start years correspond to 5 April snapshots.
  3. For each indicator, compare abs(value2025) with abs(value2017). Retain signed endpoints and signed changes. Opposing patterns require both a positive and a negative absolute-distance change; ties and all twenty-seven sign combinations remain explicit.
  4. The third indicator is FemaleTopQuartile minus FemaleLowerQuartile, in percentage points. A smaller absolute difference means more similar female shares across those two ranked pay quarters; it does not mean 50/50 gender representation or equal pay for equal work.
  5. All calculations use Decimal and the reported source precision. Unweighted employer medians describe the fixed entities only. The downloadable code verifies all input fingerprints and reproduces five result files offline.

What the data cannot tell us

  • The sample is selected on baseline sector coding and consistent reporting; it does not represent all publishers, imprints, employees, authors, freelancers or independent businesses. The exact industry code can coexist with other sectors.
  • A constant company number and employer identifier do not establish unchanged employees, ownership or business scope. Historical records can be amended. Missing records and changed codes are kept separate from inferred closures or noncompliance.
  • Hourly gaps and quarter shares concern full-pay relevant employees, whereas threshold headcount uses a different employee construct. Broad size bands cannot supply valid employee weights or individual salaries.
  • Published male/female fields are retained without claiming verified historical recording consistency. Current 2026 guidance is not retroactively treated as evidence of how every earlier report was compiled.
  • No equal-pay diagnosis, individual promotion inference, causal explanation, legal recommendation, sampling confidence interval or statistical-significance claim is made. Exact ties concern reported values, not unrounded underlying quantities.

Original sources

  1. UK gender pay gap service: reporting year 2017 to 2018

    Official reporting-cycle 2017–18 corporate data, corresponding to the private-sector snapshot of 5 April 2017. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  2. UK gender pay gap service: reporting year 2018 to 2019

    Official reporting-cycle 2018–19 corporate data, corresponding to the private-sector snapshot of 5 April 2018. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  3. UK gender pay gap service: reporting year 2019 to 2020

    Official reporting-cycle 2019–20 corporate data, corresponding to the private-sector snapshot of 5 April 2019. Frozen full-file SHA and corporate-only extraction are recorded in the downloads. Suspended reporting cycle: used only for prespecified nine-cycle membership sensitivity, not main completeness.

    Accessed October 31, 2026

  4. UK gender pay gap service: reporting year 2020 to 2021

    Official reporting-cycle 2020–21 corporate data, corresponding to the private-sector snapshot of 5 April 2020. Frozen full-file SHA and corporate-only extraction are recorded in the downloads. The separately documented enforcement extension does not change the snapshot date.

    Accessed October 31, 2026

  5. UK gender pay gap service: reporting year 2021 to 2022

    Official reporting-cycle 2021–22 corporate data, corresponding to the private-sector snapshot of 5 April 2021. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  6. UK gender pay gap service: reporting year 2022 to 2023

    Official reporting-cycle 2022–23 corporate data, corresponding to the private-sector snapshot of 5 April 2022. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  7. UK gender pay gap service: reporting year 2023 to 2024

    Official reporting-cycle 2023–24 corporate data, corresponding to the private-sector snapshot of 5 April 2023. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  8. UK gender pay gap service: reporting year 2024 to 2025

    Official reporting-cycle 2024–25 corporate data, corresponding to the private-sector snapshot of 5 April 2024. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  9. UK gender pay gap service: reporting year 2025 to 2026

    Official reporting-cycle 2025–26 corporate data, corresponding to the private-sector snapshot of 5 April 2025. Frozen full-file SHA and corporate-only extraction are recorded in the downloads.

    Accessed October 31, 2026

  10. Download index and source glossary

    Confirms 2025–26 as the latest cycle whose reporting deadlines have passed; defines identifiers, industry codes, hourly gaps, quarter shares, size bands and the late flag; carries the OGLv3 notice.

    Accessed October 31, 2026

  11. Making your calculations

    Official calculation definitions. Hourly gaps and pay-quarter shares concern full-pay relevant employees; reported percentages may be rounded.

    Accessed October 31, 2026

  12. Preparing your data

    Current guidance and snapshot/employee definitions. Its 2026 sex-recording instructions are not treated as verification of historical classification practices.

    Accessed October 31, 2026

  13. Reporting and pandemic enforcement

    Documents the 2019–20 suspension and the additional enforcement time for 2020–21. This informs the frozen main-versus-sensitivity cycle rule.

    Accessed October 31, 2026

  14. Who needs to report

    Explains relevant-employee threshold headcount, separate legal entities and voluntary smaller-employer reporting. Threshold headcount is not the exact hourly-pay analytical population.

    Accessed October 31, 2026

  15. Companies House industry classification list

    Identifies exact SIC 58110 as Book publishing. A code may coexist with other activities; it does not isolate employees working exclusively on books.

    Accessed October 31, 2026

  16. Clark and Zuccala, 2025: prior publisher pay-gap research

    Bibliographic record and figure caption for Gender pay gaps and inequity at science publishers, DOI 10.1371/journal.pgph.0004673. Earlier publisher median trajectories are acknowledged; no first-ever claim or exhaustive full-text comparison is made.

    Accessed October 31, 2026

Cite & reuse this study

Alisson Bento. 2026. When publishing employers' pay indicators move in different directions. Original corporate-panel assembly and calculations from UK government gender pay gap reporting cycles 2017–18 to 2025–26. https://alissonbento.com/research/uk-book-publishing-employer-joint-pay-indicators-2017-2025/

Original analysis and code by Alisson Bento. Contains public sector information licensed under the Open Government Licence v3.0, except where otherwise stated by the source. Attribute the official data service and this analysis; no government endorsement or rights in excluded personal information, logos or employer prose are implied.

Corrections or editorial questions

Pass it on.

When publishing employers' pay indicators move in different directions | Alisson Bentoalissonbento.com

WhatsAppFacebookXTelegramRedditThreadsEmailSMS