Office for National Statistics, profile picture
Uploaded byOffice for National Statistics
PPTX, PDF63 views

Earnings Symposium Slidepack - 29 April 2025

The Office for National Statistics (ONS), the Wage and Employment Dynamics project (WED) and Administrative Data Research (ADR) UK ran the inaugural Earnings Symposium in collaboration with ESCoE. The symposium was dedicated to exploring the latest developments, methodologies, and applications relating to ONS earnings data and earnings statistics.

Download to read offline
1 / 185
2 / 185
3 / 185
4 / 185
5 / 185
6 / 185
7 / 185
8 / 185
9 / 185
10 / 185
11 / 185
12 / 185
13 / 185
14 / 185
15 / 185
16 / 185
17 / 185
18 / 185
19 / 185
20 / 185
21 / 185
22 / 185
23 / 185
24 / 185
25 / 185
26 / 185
27 / 185
28 / 185
29 / 185
30 / 185
31 / 185
32 / 185
33 / 185
34 / 185
35 / 185
36 / 185
37 / 185
38 / 185
39 / 185
40 / 185
41 / 185
42 / 185
43 / 185
44 / 185
45 / 185
46 / 185
47 / 185
48 / 185
49 / 185
50 / 185
51 / 185
52 / 185
53 / 185
54 / 185
55 / 185
56 / 185
57 / 185
58 / 185
59 / 185
60 / 185
61 / 185
62 / 185
63 / 185
64 / 185
65 / 185
66 / 185
67 / 185
68 / 185
69 / 185
70 / 185
71 / 185
72 / 185
73 / 185
74 / 185
75 / 185
76 / 185
77 / 185
78 / 185
79 / 185
80 / 185
81 / 185
82 / 185
83 / 185
84 / 185
85 / 185
86 / 185
87 / 185
88 / 185
89 / 185
90 / 185
91 / 185
92 / 185
93 / 185
94 / 185
95 / 185
96 / 185
97 / 185
98 / 185
99 / 185
100 / 185
101 / 185
102 / 185
103 / 185
104 / 185
105 / 185
106 / 185
107 / 185
108 / 185
109 / 185
110 / 185
111 / 185
112 / 185
113 / 185
114 / 185
115 / 185
116 / 185
117 / 185
118 / 185
119 / 185
120 / 185
121 / 185
122 / 185
123 / 185
124 / 185
125 / 185
126 / 185
127 / 185
128 / 185
129 / 185
130 / 185
131 / 185
132 / 185
133 / 185
134 / 185
135 / 185
136 / 185
137 / 185
138 / 185
139 / 185
140 / 185
141 / 185
142 / 185
143 / 185
144 / 185
145 / 185
146 / 185
147 / 185
148 / 185
149 / 185
150 / 185
151 / 185
152 / 185
153 / 185
154 / 185
155 / 185
156 / 185
157 / 185
158 / 185
159 / 185
160 / 185
161 / 185
162 / 185
163 / 185
164 / 185
165 / 185
166 / 185
167 / 185
168 / 185
169 / 185
170 / 185
171 / 185
172 / 185
173 / 185
174 / 185
175 / 185
176 / 185
177 / 185
178 / 185
179 / 185
180 / 185
181 / 185
182 / 185
183 / 185
184 / 185
185 / 185
Earnings Symposium 29 April 2025
Average Weekly Earnings: Outputs and Development Nicola White Head of AWE outputs and development Office for National Statistics
AWE: Data collection and production update • Average weekly earnings (AWE) is the lead monthly measure of average weekly earnings per employee. • It is calculated using information based on the Monthly Wages and Salaries Survey (MWSS), which samples around 9,000 employers, covering around 62% of all employees in Great Britain. • The sample is taken from the IDBR and the data are weighted by the IDBR. • Businesses under the size of 20 employees are excluded from the sample. • The survey response rate is generally around 83%.
AWE: Data collection and production update • Data collected on total wages, bonuses, arrears and employment • AWE for any given month is the ratio of estimated total pay for the whole economy, divided by the total number of employees paid. • AWE reflects changes to the composition of the workforce. • The “headline rates” of AWE are the changes in seasonally adjusted average weekly earnings, including and excluding bonuses, comparing the latest 3 months with the same 3 months in the previous year.
AWE: Data collection and production update Revisions policy: • first published as provisional, 6 to 7 weeks after the data month end • This provisional data is revised the following month to allow for late and amended data returns from respondents • exceptional revisions may be made if the impact is sufficiently significant • a change in methodology can lead to revisions to the entire historic time series • when seasonally adjusted data for a given month t is published, as well as t–1 being revised, t–2 is also revised, as well as the same 3 months in the previous year • the seasonal adjustment parameters are updated annually, in line with our policy on seasonal adjustment. This update can lead to revisions to the entire historic time series
How AWE collects and reports on arrears and bonus payments • Arrears specifically covers earnings arising from a backdated pay increase, not late payment of overtime or bonuses. • A bonus is a form of reward or recognition granted by an employer in addition to basic pay. • Both payments are reflected in estimates at the time they were paid, and not in the period they are awarded for. Therefore, back series are not revised.
How AWE collects and reports on arrears and bonus payments • When arrears or bonus payments are backdated, people who have left the business but are entitled to these back payments will be included in the number of employees that have received pay in that period. • This results in more employees being added to payroll for that month and will have an impact on the average pay as more employees will be included in the calculation. • The survey only requests one employee figure, so we are unable to split out those who have left the company and only eligible to the backpay and not the regular pay. • For the majority of time, the impact of this is minimal but for certain periods, where there has been a large backpay covering a long period, the calculation of average pay will be affected and is more accurately reflected in the following month’s data.
Continually compare AWE with RTI - discuss differences with HMRC % Single month mean % growth total pay (annual) Seasonally Adjusted Source Tab Earn01 and RT estimates (SA) Jan-21 M ar-21 M ay-21 Jul-21 Sep-21 N ov-21 Jan-22 M ar-22 M ay-22 Jul-22 Sep-22 N ov-22 Jan-23 M ar-23 M ay-23 Jul-23 Sep-23 N ov-23 Jan-24 M ar-24 M ay-24 Jul-24 Sep-24 N ov-24 Jan-25 0 2 4 6 8 10 12 RTI Total AWE Total AWE Total inc arrears
AWE exceptional revisions • Due to late and corrected data for a large business we have made revisions back to October 2020. • The revisions affected: the whole economy; private sector; services sector; wholesaling, retailing, hotels and restaurants sector; and the retail trade and repairs industry. • At the whole economy level, these revisions are generally small and within the range we would expect to see during seasonal adjustment reviews; as the estimates are broken down below the whole economy level, the revisions become larger.
AWE: Development work and plans Methodology work: • Annual Seasonal Adjustment review • currently reviewing the seasonal adjustment parameters as part of the annual review. • the review may lead to revisions to the entire historical AWE time series and will be implemented either in the May 2025 or June 2025 publication. • Sample optimisation review • Review sampling allocation • Optimise stratification / sample
AWE: Development work and plans Methodology work: • Under 20’s – current method / other options • create under 20 factors based on 2023r and 2024p ASHE and assess the impact on the AWE estimates • review the data sources and methods used in the under 20 factors to assess if we can improve them • Up-date uncertainty measures • Review outlier method • Review imputation method
AWE: Development work and plans • Review revisions policy • Continual coherence assessment in particular with the RTI but also other earnings datasets • Use of RTI that we have in the office
Annual Survey of Hours and Earnings (ASHE): Overview and Development Plans Lualhati Santiago Head of ASHE outputs and development Office for National Statistics
1. ASHE Overview
Key facts ASHE is our highest quality survey on individual-level employee earnings: • It is reliable (it is completed by employers). • It is timely (data collected in April, published in October). • It is large: the achieved sample for 2024 was 173,000.​
Data and sample (I): Overview Data: individual-level data on employee earnings (payments made to the employee by the employer) and the hours on which this pay was calculated. Sample: - 175k - allows for detailed analyses & breakdowns.
Data and sample (II): Selecting the sample Sample of employees in the UK selected by: 1. 1% simple random sample of NINOs (last 2 NI digits); 2. eligible NINOs are selected on the HMRC PAYE frame; 3. linked to ONS IDBR register to obtain businesses’ contact details - a form is sent and employers complete it.
Data and sample (III): Weights To correct for non-response biases, the data is weighted to the LFS. The ASHE weights are created as a combination of: 1. Design weight. 2. Calibration weight (uses the LFS).
Timeline of ASHE April May June July August September October Reference date Microdata ready Data collection, ingestion, processing & validation Survey close date Stats and Bulletins published (Key requirement: capture new NLW/NMW)
ASHE: published statistics • 3 bulletins • Employee earnings in the UK • Low and high pay in the UK • Gender pay gap in the UK • 800 tables with statistics • Microdata (SRS/IDS) Data included: - Current year (provisional data). - Previous year (revised data).
2. ASHE Development Work and plans
Why we are reviewing/developing ASHE • Methods and end-to-end system have been in place since the early 2000s and are due a review.​ • Data challenges since the pandemic (2020 onwards):​ •partly due to furlough and coronavirus-related impacts on the labour market,​ •but some remained post-pandemic.
Development work and plans: 5 workstreams • Methods • Processing system (end-to-end) • Data collection • User needs • Data coherence
Workstream 1: Methods • Data validation • Selective editing review • Validation and error prioritisation tools • Sampling, weighting and imputation review • Pension variables review
Workstream 2: Processing system (end-to-end) • SOC coding redevelopment • Data processing redevelopment • Table production redevelopment • More End-to-end holistic review
Workstream 3: Data collection • Transition to Electronic Questionnaire • Data collection systems: scoping redevelopment End-to-end holistic review
Workstreams 4 and 5: User needs & coherence • User needs: • Earnings User Group (across government) • Earnings Symposium • User consultations • Coherence: • Analytical assessments to compare earnings' sources Ongoing Feeds into workstreams 1-3
Timeline for ASHE developments 2024 2025 2026 2027 & beyond Data validation Dates TBC • Pension variables review • Transition to Electronic Questionnaire • Sampling, weighting and imputation review • Data collection systems redevelopment Data processing redevelopment SOC coding redevelopment Table production redevelopment
For questions, comments, and to learn more about ASHE and our plans, please get in touch: Earnings@ons.gov.uk
Plans for PAYE RTI and ONS Labour Market Statistics Tom Evans Analysis Lead Labour Market Transformation, ONS
Background • It has been a long-held goal of ONS to secure a regular supply of HMRC Pay As You Earn, Real-time Information data aka. PAYE RTI • In collaboration with HMRC, in recent years ONS has made increasing use of PAYE RTI statistics • Monthly Earnings and Employment from PAYE RTI Provides valuable insights on short-term movements in the labour market. Breakdowns by: Age, Industry (IDBR) and Geography (residence) • Wages and Salaries within the Annual National Accounts • Both of these cases see PAYE RTI used in isolation • Further benefits will arise from linking PAYE RTI with other survey and non- survey sources
Build a better understanding of existing statistics, with the potential to improve accuracy and precision Enable the production of new statistics and analysis to complement existing activity and provide new insights Broad intentions for PAYE RTI and data linkage
Existing Statistics and Analysis for example New Statistics and Analysis Improving understanding of bias Coherence and validation Survey developments Non-linkage outputs Linked Employer-Employee Dataset Intersections between labour market and other topics Expansion of existing uses for PAYE RTI Exploitation of PAYE RTI scale As ever, alignment with concepts and conceptual detail will need to be considered
Short-term surveys (Transformed) Labour Force Survey ((T)LFS) Sole source for headline labour market rates and levels in addition to a range of other labour market statistics Monthly Wages and Salaries Survey (MWSS) Underpins Average Weekly Earnings (AWE) estimates – lead measure of short-term changes in earnings Short-term Employment Surveys (STES) Main input into Workforce Jobs estimates – key comparator for (T)LFS estimates Structural Surveys Annual Survey of Hours and Earnings (ASHE) Detailed business survey on earnings and hours worked among employees (PAYE-based sample frame) – key input into NLW calculations Business Register and Employment Survey (BRES) Detailed business survey on the geographic and industry characteristics of employment. Only business survey which collects “Local Unit” information. Also used to update the Inter Departmental Business Register Labour Market datasets
Shorter-term plans LFS and TLFS
IDS Coherence Project RDMF – Demographic Index CNRLS Labour Market datasets – initial focus LFS TLFS HMRC PAYE RTI 2021 Census Clerical linkage – point-in-time Automated indexing, ongoing
Data linkage assurance Working with relevant areas within ONS to understand automated linkage processes such that we can understand how linkage approach and linkage quality impacts on analytical approach and findings Proof of Concept - Employee/Self-employed Assurance and Exploration Jan-M ar 2014 Jun-Aug 2014 N ov-Jan 2015 Apr-Jun 2015 Sep-N ov 2015 Feb-Apr 2016 Jul-Sep 2016 D ec-Feb 2017 M ay-Jul 2017 O ct-D ec 2017 M ar-M ay 2018 Aug-O ct 2018 Jan-M ar 20195 [r] Jun-Aug 2019 [r] N ov-Jan 2020 [r] Apr-Jun 2020 [r] Sep-N ov 2020 [r] Feb-Apr 2021 [r] Jul-Sep 2021 [r] D ec-Feb 2022 [r] M ay-Jul 2022 [r] O ct-D ec 2022 [r] M ar-M ay 2023 [r] Aug-O ct 2023 [r] Jan-M ar 2024 [r] Jun-Aug 2024 [r] 11.0% 11.5% 12.0% 12.5% 13.0% 13.5% 14.0% 14.5% 15.0% 15.5% Self-employment share (main job) Since the pandemic we have seen a marked and persistent decline in the share of people self-employed in their main job on the LFS, while the share is lower still on TLFS Linkage with PAYE RTI may offer new insights into whether there has been a more permanent change in how people self- classify on a survey vs. their tax definition
Developing understanding of bias Future Priorities Bias and coherence are live issues. PAYE RTI linkage could allow us to get closer to an external (albeit partial) benchmark for labour market status. By linking both sample and response files at the household level we can analyse the extent to which non-responding households contain one or more payrolled employees and how this has evolved over time This would build on a one-off analysis done in 2018
But what about earnings?
Short-term surveys (Transformed) Labour Force Survey ((T)LFS) Monthly Wages and Salaries Survey (MWSS) Short-term Employment Surveys (STES) Structural Surveys Annual Survey of Hours and Earnings (ASHE) Business Register and Employment Survey (BRES) Existing Statistics and Analysis Improving understanding of bias Coherence and validation Survey developments Non-linkage outputs Build a better understanding of existing statistics, with the potential to improve accuracy and precision Building upon existing aggregate checks, microdata will allow for more detailed coherence and validation checks Initial considerations likely to be around targeted question replacement, though need to be aware of conceptual/definitional differences
Wider opportunities “Pure” RTI outputs Monthly PAYE RTI statistics National Accounts inputs Linked analytical insights Linked Employer-Employee Dataset e.g. Productivity insights, career progression Characteristics analysis e.g. insights from linkage with Census and other surveys Exploiting PAYE RTI scale Enable the production of new statistics and analysis to complement existing activity and provide new insights As you can see, there is a large range of potential avenues to explore both within earnings statistics and across the wider labour market portfolio – in a resource constrained environment, coordination and prioritisation will be key. Welcome feedback on where priorities should lie
Wage and Employment Dynamics: Damian Whittard1 , Van Phan1 , Felix Ritchie1 , Anni Caden1 , Alex Bryson2 , John Forth3 , Carl Singleton4 , Rachel Scarfe4 1 University of the West of England, Bristol. 2 University College London, London. 3 City University, London. 4 University of Sterling, Stirling.
Acknowledgements and disclaimer • We gratefully acknowledge ADR UK (Administrative Data Research UK) and the Economic and Social Research Council (Grant No. ES/T013877/1 & ES/Y001184/1) for funding. • The work is based on analysis of the research-ready datasets from the WED Project: Office for National Statistics (2022) Annual Survey of Hours and Earnings linked to 2011 Census - England and Wales - https://doi.org/10.57906/80f7-te97 • The analysis was carried out in the Secure Research Service, part of the Office for National Statistics (ONS). • This work contains statistical data from ONS which is Crown Copyright. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.
1. The WED Project 2. ADR UK Research Fellowship • Working Towards an Environmentally Sustainable and Equitable Future? Introduction
WED Project: Aims Research-ready data • Definitive, clean, fully described data • Fully documented construction • weighting • Sustainable – life beyond the project as core dataset(s) • Training and information events Strategic research • Low pay • Job mobility • Pay gaps • Green jobs
Research-ready data HMRC data: 1. PAYE 2. Self Assessment 3. Migrant Workers Scan Review QA Link Document Add common personal IDs Strategic research Wage and Employment Spine wage progression low-wage labour markets ONS business register (Enhanced) ASHE 2011 Census WED Project: How it works
2022  ‘Enhanced’ ASHE  ASHE-Census 2011 2024  ASHE - PAYE & SA o ASHE - Migrant Scan Worker 2025 Expected 2026 o ASHE-Census 2021 WED Project: Progress to data
• WED Website • https://www.wagedynamics.com • Community of Interest group • https://www.wagedynamics.com/community-of-interests/ • Knowledge Hub • https://khub.net/group/wage-and-employment-dynamics-research-group/ Stakeholder Engagement: Supporting the Community
Working Towards an Environmentally Sustainable and Equitable Future? New Evidence on Green Jobs from Linked Administrative Data in the UK Damian Whittard Peter Bradley, Van Phan and Felix Ritchie Journal of Cleaner Production
Our contribution • Uses a new linked administrative dataset based on high quality earnings information to estimate the economic benefit of working in a green occupation. • New knowledge is presented about the attributes of those who work in green occupations and the characteristics of the jobs and the employers. • Furthering work on attitude-behaviour gaps, the study provides evidence that personal travel behaviours and green employment choices are not consistent • The research adds to the international literature of pay in green jobs, estimating a positive pay premium for England and Wales. • Provides an original contribution revealing that working in a green occupation can offset some of the inter-occupation pay gap, yet within these occupations, gender and ethnic pay gaps persist. • The study emphasises the need for inequalities to be captured by theory that attempts to understand and conceptualise the uptake of green jobs Introduction | Data | Methods | Results | Discussion
Motivation • Climate crisis and the environmental emergency • Huge economic opportunity (BEIS, 2021). • Green jobs at the heart of any transition • No universally accepted definition of green jobs (Bowen et al, 2018; Sulich 2020) • Lack of high-quality, large scale, longitudinal data (Skidmore, 2022) • Alternative measures of green jobs • Top down (Georgeson and Maslin, 2019; Eurostat, 2021; ) • Bottom-up (Bowen et al., 2018) • Occupations (Valero et al., 20121) • Tasks (Martin, J. and Monahan, E., 2022) • Exploit opportunity provided by linking administrative datasets • Employment (occupations) • Pay (premium or penalty) • Characteristics Introduction | Data | Methods | Results | Discussion
Data • Annual Survey of Hours and Earnings (ASHE) (Whittard et al, 2022) • 1% sample of all employees / longitudinal • Provided by their employer • Responses rate – approximately two-thirds • ASHE linked to Census (2011-2018) (See www.wagedynamics.com) • Personal and family characteristics for employees • Linkage results in 0.5 percent of the population in match year (2011) – circa 100,000 obs. • Attrition overtime - 76,000 observations in 2018 • Map US O*NET data on occupations (tasks based) (Dickerson and Morris, 2019) • O*NET identify 12 sectors contributing to the ‘green economy’ • Tasks were used to assess the level of occupational greening – three categories identified • Green new and emerging; green enhanced skills; green in demand • Circa 200 of 1,100 occupations assessed as green • Limitation - assumption tasks & occupations same in the UK and US • Binary and continuous (weighted) measure Introduction | Data | Methods | Results | Discussion
Figure 2: Share of green employment by gender and ethnicity (2018): binary measure (a) By gender (b) By ethnicity groups 32% 47% 18% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% Total Male Female 33% 31% 26% 24% 32% 0% 5% 10% 15% 20% 25% 30% 35% 40% White Mixed / multiple ethnic groups Asian / Asian British Black / African / Caribbean / Black British Other ethnic group Results
Results
Results: Regression Summary Results • Individuals are more likely to work in green occupations if they are white, male, full-time, not represented by a collective agreement, and work for an SME or foreign owned business • The study estimates a pay premium of between 10% (raw) and 4% (including all observable controls) • Personal travel behaviours and green employment choices are often inconsistent, but when they align this yields a pay dividend. • Green employment can partially mitigate inter-occupation pay gaps, however, the gaps persist within green occupations • Females appear particularly disadvantaged by domestic and childcare responsibilities. • There are also sector effects, with more traditional industries such as manufacturing and construction exhibiting entrenched gender biases Implications • The results highlight the need to integrate considerations of inequality into theoretical frameworks that aim to understand and conceptualise the uptake of green jobs. • There is an important role for policy to play if the green transition is to deliver social inclusivity alongside economic growth and job creation Introduction | Data | Methods | Results | Discussion
Thank You damian2.whittard@uwe.ac.uk
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus The Disability Pay Gap Within and Across Firms John Forth and Melanie Jones ONS Earnings Symposium 29th April 2025 Acknowledgements: This work was undertaken in the Office for National Statistics Secure Research Service using data from ONS and other owners and does not imply the endorsement of the ONS or other data owners. We are grateful to the ONS, Administrative Data Research UK and the Wage and Employment Dynamics project for creating the ASHE-Census 2011 dataset. John Forth gratefully acknowledges funding from the Wage and Employment Dynamics project under ESRC grant number ES/T013877/1.
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus • Disability Pay Gap (DPG) neglected relative to pay gaps for other protected characteristics • Little evidence on the role of the distribution of employees across firms to the DPG • This paper uses newly linked ASHE-Census data to separate the influence of the distribution of employees across firms from within-firm DPGs • Contemporary policy relevance given planned employer DPG reporting in the UK Motivation
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG is the percentage difference in average hourly pay between disabled and non-disabled employees • Sizeable (10-15%) and shows no sign of diminishing • Decompose DPG • Personal and job-related characteristics • Less than 50% explained (Jones et al., 2006) • Debates about identifying discrimination (DeLeire, 2001; Longhi et al., 2012) • Role of the firm • Schur et al. (2009) role of corporate culture on gaps in-work outcomes in the US • Jones and Latreille (2010) variation in within-workplace DPG using WERS 2004 Existing Literature
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus • Payroll information from ASHE 2011 • Multiple measures of pay (focus on total hourly earnings) • Personal characteristics from 2011 Census (74% of ASHE 2011 observations) • Disability (5.3%) • “Are your day-to-day activities limited because of a health problem or disability which has lasted, or is expected to last, at least 12 months? Include problems related to old age”. • Yes (limited a lot or limited a little). • Working-age (age 16-64) employees paid an adult rate, earnings unaffected by absence and basic weekly hours (1-99 hours). • Minimum of two employees within each firm • 78,037 jobs from 76,505 employees in 8,435 firms ASHE-Census
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus Table 1: Descriptive statistics on earnings (£/hour), by disability Basic rate Total pay Non-disabled Mean 13.92 14.53 N 73,815 73,815 Disabled Mean 12.46 12.95 N 4,222 4,222 DPG (%) Mean -10.51 -10.88
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG ( ) 𝜇 • Employee characteristics: sex and age (and age-squared), highest education, marital status, number of children, age of youngest child, ethnicity and UK born • Job-related characteristics: tenure, part-time, collective bargaining, second job and occupation (2010 SOC minor group) • Firm characteristics: firm size, industry sector (2007 SIC sections), sector and region • Within-firm DPG • Replace with firm identifiers • Heterogeneity e.g. gender, sector, firm size and over the wage distribution • Recentered influence function (RIF)-OLS earnings equations (Firpo and Pinto, 2016) • Extend established decomposition methods (Oaxaca, 1973; Blinder, 1973) Analysis
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus Raw Raw: within-firm Adjusted: employee and job characteristics Adjusted: employee, job and firm characteristics Adjusted: employee and job characteristics within-firm (1) (2) (3) (4) (5) Total pay -0.095*** -0.068*** -0.040*** -0.036*** -0.032*** (0.008) (0.007) (0.005) (0.005) (0.005) N 78,037 78,037 78,037 78,037 78,037 Adj R-squared 0.002 0.377 0.650 0.685 0.737 Notes: OLS regression coefficients, estimated from ASHE-Census 2011. DPG calculated as the difference in log points between the log hourly wages of disabled employees and non-disabled employees. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1. Table 2: Raw and adjusted DPG, across and within firms
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus Table 3: Decomposition of DPG Notes: Estimates based on OB decomposition methods, as set out in the text, applied to total hourly earnings from ASHE-Census 2011. Robust standard errors estimated via the delta method and reported in parentheses. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1. Components may not sum to raw DPG due to rounding errors.
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus Raw Raw: within-firm Adjusted: employee and job characteristics Adjusted: employee, job and firm characteristics Adjusted: emp. and job chars. within-firm (1) (2) (3) (4) (5) Disabled -0.118*** -0.025 -0.027* -0.023 0.002 (0.026) (0.021) (0.016) (0.015) (0.014) 250-999 employees 0.060*** 0.032*** 0.031*** (0.008) (0.005) (0.005) 1,000-4,999 employees 0.081*** 0.059*** 0.052*** (0.007) (0.005) (0.004) 5,000+ employees -0.015** 0.044*** 0.045*** (0.007) (0.004) (0.004) Disabled x 250-999 employees 0.027 -0.032 -0.002 -0.002 -0.022 (0.034) (0.028) (0.022) (0.021) (0.019) Disabled x 1,000-4,999 employees 0.035 -0.048* -0.020 -0.021 -0.047*** (0.031) (0.026) (0.019) (0.018) (0.017) Disabled x 5,000+ employees 0.027 -0.048** -0.015 -0.016 -0.036** (0.028) (0.023) (0.017) (0.016) (0.015) Adj R-squared 0.008 0.377 0.651 0.685 0.737 N 78,037 78,037 78,037 78,037 78,037 Table 4: Heterogeneity in the DPG by firm size Notes: OLS regression coefficients for total hourly earnings, estimated from ASHE-Census 2011. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1.
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus Figure 1: Raw and adjusted DPG, across and within firms, across the wage distribution Notes: Recentered influence function (RIF)-OLS regression coefficients for total hourly earnings, estimated from ASHE-Census 2011. Error bars show 95% confidence intervals.
The Public Value Business School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG and unexplained DPG – Mainly exist within firms – Within-firm DPGs are reinforced by the allocation of disabled employees across firms – Failure to account for firm allocation results in disability-related wage inequality being overestimated by 12-20% • Role for employer focus of DPG reporting – Large firms have larger within-firm DPG, suggests effective targeting of legislation – Value in reporting workforce disability composition Conclusions
Uncovering multiple employment with linked ASHE-HMRC data Dr Darja Reuschke Birmingham Business School Prof Tracey Warren Nottingham Business School
Multiple employment • Having more than one employment: • Dual/multiple job-holding • Hybrid entrepreneurship (employees with income from self-employment) • Existing studies • Based on survey data on ‘2nd job’ • Little differentiation between different forms of multiple employment • Little known about gender and multiple employment • Three hypotheses: • Financial constraints • Job insecurity • Non-monetary reasons (e.g. skills development)
• Using new administrative data sources to understand influencing factors on multiple employment • Differences between multiple job-holding vs mixing employee jobs with self- employment income • Gender differences • Research questions • Low wages? • Low working hours? • Regional labour market conditions? Objectives and research questions
• Office for National Statistics; His Majesty's Revenue and Customs, released 01 August 2024, ONS SRS Metadata Catalogue, dataset, Annual Survey of Hours and Earnings linked to PAYE and Self-Assessment data - GB, https://doi.org/10.57906/566k-5q15 • ASHE-GB: • Sample of 1% of employee jobs in GB • Main sample: PAYE jobs based on NI numbers • NI numbers remain in ASHE sample -> longitudinal element • Multiple (PAYE) jobs of the same NI number are sampled • Employer non-response • ASHE-HMRC Self-Assessment data: 2011-2017 • ASHE-HMRC Pay As You Earn (PAYE) data: 2014-2018 Data
ASHE-SA 2011-2017 • Link of self-employment income (profit) to ASHE individuals (by tax years) • Comparison of multiple jobs (in ASHE) vs mixing employee jobs & s/emp ASHE-PAYE 2014-2018 • ASHE survey reference period in April linked with April payslips • Identifying multiple ‘employment’ in PAYE data based on payslips with positive values & excluding pensions • Multiple employment of ASHE individuals based on April PAYE data > ASHE annual survey • Comparison of multiple jobs in annual ASHE vs multiple ‘employments’ in PAYE data • 21% of ASHE individuals aged 16-64 do not have payslips in PAYE data Linked longitudinal datasets
• Transition from one job/employment to multiple jobs/employment between subsequent years (t-1 to t) • Conditional fixed-effects logistic regressions by gender, 16-64-year-olds • Financial constraints of job: 1. Weekly earnings 2. Hourly wage & weekly basic hours 3. NMW groups: i) at or below NMW, ii) below 60% median, iii) 60% median + & weekly basic hours • Insecurity: • permanent/temporary job • Regional labour market (ITL1): • GDP per capita, male/female unemployment rate • Controls (incl. age, sector, year dummies, not shown below) Methods
Findings from ASHE-SA: Multiple jobs in ASHE vs employee job with income from self-employment
Multiple employee jobs in ASHE
Employees with income from self-employment
Findings from ASHE-PAYE: Starting multiple jobs using ASHE annual survey vs real time PAYE data
Multiple employments as employee in ASHE vs PAYE
• Multiple job-holding and mixing employee jobs with self-employment are different forms of multiple employment • Low working hours and resultant low weekly earnings consistent factors influencing multiple job-holding of women and men • Findings on low wages are mixed • Effect of low wages amplified in PAYE data • Potential under-coverage of multiple job-holding among those with low wages in ASHE • No evidence of job insecurity hypothesis (temporary jobs) • Mixing employee jobs with self-employment income may be better approached as hybrid entrepreneurship than multiple job-holding Key policy conclusion: Need of jobs with sufficient hours Summary and conclusion
Thank you! Darja Reuschke d.reuschke@bham.ac.uk Acknowledgement: This research is funded by an ADR UK fellowship (ES/Z503149/1) and conducted in collaboration with the UK WBG. I am grateful for the WED team, namely Damian Whittard, Prof Felix Ritchie, Dr Van Phan and Dr Carl Singleton, for their help. “This work was undertaken in the Office for National Statistics Secure Research Service using data from ONS and other owners and does not imply the endorsement of the ONS or other data owners.”
The use of ONS earnings data in decision- making at the Low Pay Commission Presentation to the ONS earnings symposium Tim Butcher 29 April 2025
Overview • Low Pay Commission and the National Minimum and National Living Wage • Determining the path: Benchmark; Actual wage growth; Forecasts • Analysing pay trends: AWE, RTI and ASHE • Descriptive analysis of structure of pay: bite, coverage and distribution • Research: external, independent and in-house • Conclusion 82
Our NLW recommendation must… 83 “ensure that the rate does not drop below two-thirds of UK median earnings for workers aged 21 and over” Take account of: “the impact on business, competitiveness, the labour market, the wider economy” Take account of: “the cost of living, including the expected annual trends in inflation between now and March 2026”
Our evidence-based approach 84 External research We make use of research produced by others including the IFS, Resolution Foundation and others Stakeholders We speak to employers, employees and their representatives, through visits and evidence sessions Internal analysis We undertake our own analysis of primary data sources, including econometric analysis International comparisons We make use of international data and research and convene events with our equivalents in other countries Consultation We invite written evidence submissions through an open, public consultation. Commissioned research Each spring we commission research from academics and other experts, who report their findings in the autumn.
LPC uses various ONS earnings data sources • The focus here is on the top three (with ASHE the main one) • Annual Survey of Hours and Earnings (annual) • Average Weekly Earnings (monthly) • RTI earnings data (monthly) published by ONS • (Labour Force Survey wage data – quarterly) • (National Accounts data on wages and salaries, and compensation of employees) 85
Determining the path 86
Estimating the path of median earnings • Since 2016, the LPC remit has referred to a target of the percentage of median earnings for the appropriate age group • We use October as the mid-point of the NLW year (April-March) as the target point • Step 1: the median earnings benchmark – Median of hourly earnings excluding overtime for those covered by the NLW • Step 2: actual wage growth since the benchmark – Change in AWE total pay (using a smoothed 12-month on 12-month average) • Step 3: forecast wage growth (to following Octobers) since the last actual wage growth – Wage forecasts from HM Treasury panel of independent forecasts plus OBR and the Bank of England (median of forecasts made in last 3 months) 87
Baseline median hourly pay (ASHE 2023 and 2024) Median hourly pay 21 and over 2023 provisional £15.98 2023 final £16.16 2024 provisional £17.19 2024 on 2023 wage growth 6.4% 2024 on provisional 2023 wage growth 7.6% 88
Large revisions to the baseline in 2023 had a knock-on to 2024 89 2021 Apr 2021 Oct 2022 Apr 2022 Oct 2023 Apr 2023 Oct 2024 Apr 14.00 14.50 15.00 15.50 16.00 16.50 17.00 17.50 14.26 14.92 16.16 17.19 14.28 14.9 15.98 Latest ASHE Month of projection Median of hourly earnings excluding overtime for those aged 21 and over (£) Source: LPC estimates using ONS data. Median of hourly earnings excluding overtime for those aged 21 and over from Annual Survey of Hours and Earnings, 2021-2024 provisional, 2021-2023 final.
From the baseline, we use a 12-month smoothed version of AWE to proxy ASHE median wage growth (from baseline to latest) 90 Source: LPC estimates using ONS data. Average weekly earnings total pay (KAB9), March 2021-February 2025. Note: LPC uses a smoothed 12-month on 12-month estimate of pay growth. Feb 03 Feb 05 Feb 07 Feb 09 Feb 11 Feb 13 Feb 15 Feb 17 Feb 19 Feb 21 Feb 23 Feb 25 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 10.0 AWE total pay Smoothed LPC total pay Change on a year ago (%)
But wage forecasts had been revised up since the spring 91 Source: Bank of England, HM Treasury and Office for Budget Responsibility. Forecasts of average wage growth (median of HMT panel of independent forecasts, August 2023-April 2025, BoE Monetary Policy Report August 2023-February 2025 and OBR March 2023-March 2025). Aug 2023 Oct 2023 Dec 2023 Feb 2024 Apr 2024 Jun 2024 Aug 2024 Oct 2024 Dec 2024 Feb 2025 Apr 2025 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 4.6 5.0 3.6 3.7 3.1 3.0 2.086919803629 32 2.027930825642 74 2024 HM Treasury panel 2024 Bank of England 2024 OBR 2025 HM Treasury panel 2025 Bank of England 2025 OBR Month of projection Annual wage growth (%)
What does all that mean? 92
The projected median in 2025 at the Retreat 2022 Apr 2022 Oct 2023 Apr 2023 Oct 2024 Apr 2024 Oct 2025 Apr 2025 Oct 14.50 15.00 15.50 16.00 16.50 17.00 17.50 18.00 18.50 19.00 14.92 16.16 17.19 17.63 17.99 18.32 Month of projection Projected median of hourly earnings excluding overtime for those aged 21 and over (£) Step 1: The Baseline ASHE Median Step 3: Wage projection 2 Wage forecasts (August 2024-October 2025) Step 2: Wage projection AWE (April 2024-August 2024) 93 Source: LPC, ONS, BoE, HMT and OBR. LPC calculations based on median of hourly earnings excluding overtime, ASHE, 2022-2024; AWE total pay (KAB9), 2021-2024, and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, BoE Monetary Policy Report Aug 2024 and OBR March 2024).
Evolution of the projected path of the NLW (2/3rds of median) 94 Path of the NLW for 2025 if kept at two-thirds of median 11.00 11.20 11.40 11.60 11.80 12.00 12.20 12.40 12.60 11.56 11.61 11.82 12.01 12.04 11.84 11.89 12.1 12.18 12.21 12.13 12.18 12.39 12.36 12.39 Lower range Central estimate Month of projection Projected NLW in 2025 (£) Source: LPC, ONS, BoE, HMT and OBR. LPC calculations based on median of hourly earnings excluding overtime, ASHE, 2022-2024; AWE total pay (KAB9), 2021-2024, and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, BoE Monetary Policy Report Aug 2024 and OBR March 2024).
Earnings growth remains above forecast in Q4 95 Source: LPC calculations based on ONS, Bank of England, and HM Treasury data. AWE total pay (KAB9), 2021-2024, median of pay growth from HMRC Real Time Information (Table 27, October 2024), and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, and BoE Monetary Policy Report Aug 2024). 2017 Dec 2018 Dec 2019 Dec 2020 Dec 2021 Dec 2022 Dec 2023 Dec 2024 Dec 2025 Dec 2026 Dec -2 -1 0 1 2 3 4 5 6 7 8 9 10 AWE total pay (nominal) RTI median of pay growth HM Treasury panel forecast AWE (Aug/Oct 24) Bank of England private sector AWE LPC smoothed AWE Annual growth in average weekly pay (%)
Analysing pay trends 96
Wage growth continues to be strong. RTI measures of pay growth remain above 5% (and AWE also similar) 97 Source: LPC calculations based on ONS. Growth in median pay (derived from Table 2); growth in average pay (derived from Table 3); and median of pay growth ((Table 27) from HMRC Real Time Information, seasonally adjusted, monthly, February 2017-February 2025. 2017 Feb 2018 Feb 2019 Feb 2020 Feb 2021 Feb 2022 Feb 2023 Feb 2024 Feb 2025 Feb -1 0 1 2 3 4 5 6 7 8 9 10 RTI median pay RTI mean pay RTI median of pay growth Change in pay on same month a year ago (%)
Over the last year, we needed to be aware of the impact of public sector bonuses but these have now dropped out of the annual comparisons 98 -4 -2 0 2 4 6 8 10 12 14 Private sector Public sector excluding finance Change on a year ago (%) Total pay Nov 16Nov 18Nov 20Nov 22Nov 24 0 10 20 30 40 50 60 70 Public sector excluding finance Private sector Average weekly bonus pay (£) Bonus pay -2 0 2 4 6 8 10 Private sector Public sector excluding finance Change on a year ago (%) Regular pay Source: LPC estimates using ONS data. Average weekly earnings total pay private sector (KAC4), public sector excluding financial services (KAD9); Average weekly earnings bonus pay private sector (KAF7), public sector excluding financial services (KAH3); Average weekly earnings regular pay private sector (KAJ2), public sector excluding financial services (KAK6), seasonally adjusted, November 2016-November 2024.
Real average weekly wages have picked up over the last two years but are still only back to their April 2021 levels 99 Apr 2021 Oct 2021 Apr 2022 Oct 2022 Apr 2023 Oct 2023 Apr 2024 Oct 2024 94 95 96 97 98 99 100 101 Real AWE total pay (CPI) Real AWE regular pay (CPI) Real AWE total pay (CPIH) Real AWE regular pay (CPIH) April 2021=100 Real average weekly pay index (April 2021=100) Source: LPC estimates using ONS data. Real average weekly earnings total pay (A3WX), real average earnings regular pay (A2FC), and Real average earnings using CPI (derived from X09), April 2020-February 2025.
But still not much higher than in 2008 on these measures 100 Source: LPC estimates using ONS data. Real average weekly earnings total pay (A3WX), real average earnings regular pay (A2FC), and Real average earnings using CPI (derived from X09), February 2001-February 2025. Feb 2001 Feb 2004 Feb 2007 Feb 2010 Feb 2013 Feb 2016 Feb 2019 Feb 2022 Feb 2025 400 425 450 475 500 525 550 Real (CPIH) AWE total pay Real (CPIH) AWE regular pay Real (CPI) AWE total pay Real Average Weekly Earnings (£)
Descriptive analysis of structure of pay 101
The bite of the minimum wage is the highest it has ever been, but it missed the target set for 2024 102 The UK’s adult minimum wage as a percent of median hourly pay (bite), UK, 1999-2024 Apr-99 Apr-04 Apr-09 Apr-14 Apr-19 Apr-24 40 45 50 55 60 65 70 25+ 21+ Eligible population 2020 target 2024 target NLW Bite (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 1999-2024. Note: LPC projections beyond April 2024 using AWE data and HM Treasury and Bank of England forecasts.
But, as at April 2024, the bite was above the 2/3rds target for all regions except London, Scotland and the South East 103 Bite of the NLW by region, NLW eligible population 0 20 40 60 80 100 2019 2023 2024 NLW Bite (%) 2024 target Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2019, 2023 and 2024.
Hourly wage growth was strong across the distribution in 2024, but strongest for low-paid workers 104 Growth in hourly wage by percentiles, 2024, UK Real hourly wage growth by percentile, 2024, 21+ population 1 9 17 25 33 41 49 57 65 73 81 89 97 -5 0 5 10 15 23+ 21-22 Percentile Growth (%) 1 9 17 25 33 41 49 57 65 73 81 89 97 -15 -10 -5 0 5 10 15 20 Real growth 2024 Real growth since 2019 Percentile Growth (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2019 and 2024.
Increases in hourly pay have fed through to weekly pay 105 1 (lowest paid) 2 3 4 5 6 7 8 9 10 (highest paid) 0 2 4 6 8 10 12 Hourly pay growth Weekly pay growth Hourly pay deciles Growth (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2024.
Despite the large increases in the NLW in recent years, coverage had not increased. However, coverage rose sharply in 2024 106 Number and share of jobs covered by the NMW/NLW, UK 2016-2024 Source: LPC analysis of ASHE, UK, low-pay weights. NLW coverage refers to workers aged 25 and over before 2021, and 23 and over from 2021 to 2023, and 21 and over for 2024, due to NLW eligibility change. 2016 2017 2018 2019 2020 2021 2022 2023 2024 0 600000 1200000 1800000 2400000 0 2 4 6 8 Covered NLW Covered Total Coverage rate NLW Coverage rate Total Persons covered (millions) Coverage rate (%)
The number of jobs paid within £1 of the NLW increased by 2 ppts, to be back at 2020 levels 107 Number of jobs by pay relative to the adult rate NMW/NLW, NLW eligible population Source: LPC analysis of ASHE, UK, low-pay weights. NLW eligible population refers to workers aged 25 and over before 2021, 23 and over from 2021 to 2023 and 21 and over for 2024, due to NLW eligibility change, excludes first year apprentices. 2016 2017 2018 2019 2020 2021 2022 2023 2024 0 2 4 6 8 10 12 14 16 18 20 Covered 6-50p above NMW/NLW 51p-£1 above NMW/NLW Share of employed popualtion (%)
Using the average wage growth of the 35-80th percentiles, we estimate that spillovers were responsible for up to a third of wage growth in 2024 108 Decomposition of hourly pay increase by percentile, UK, 21+ 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Counterfactual increase Increase required by NLW Spillover No spillover assumption Change in hourly pay (£) At the 5th percentile, people are paid £11.45 At the 13th percentile, people are paid £12.00 At the 24th percentile, people are paid £13.00 At the 35th percentile, people are paid £14.43 Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2024.
Research 109
Research • ONS earnings data have been used to conduct econometric analyses of the impact of the NMW/MLW • Externally commissioned research and independent research – Dube (2019) used ASHE in his review of minimum wages for HM Treasury – IFS (2021) developed the bunching analysis to identify minimum wage effects – IFS (2024) looked at effects of the National Living Wage on firms' wage structures – Dickens, Manning and Butcher (2012) looked at spillovers • In-house research – Extensions of the bunching analysis (more recent time periods) – Development and extensions of the geographic variations approach – Developing analysis to look at young people (age discontinuities) 110
Conclusions • ONS wage data is essential for our analyses of the National Minimum Wage and the National Living Wage (NMW/NLW) • The data help us to estimate the median and determine any target • They provide information on pay trends • They enable us to explore the structure of pay – hourly and weekly pay distributions • We can analyse the impact of the NMW/NLW on bite and coverage • We can look at those most affected by the NMW/NLW – Low-paying occupations, low-paying industries, low-pay areas, micro and small firms and very large ones, young people and older workers, women and part-time. • Pandemic (and recent data shortcomings) has made analysis much harder 111
Wage growth: insights from UK firm-level microdata Josh Martin Bank of England, King’s College London, Economic Statistics Centre of Excellence (ESCoE) Any views expressed are solely those of the author and so cannot be taken to represent those of the Bank of England or to state Bank of England policy. This paper should therefore not be reported as representing the views of the Bank of England or members of the Monetary Policy Committee, Financial Policy Committee or Prudential Regulation Committee. This paper uses ONS statistical research datasets. Outputs may not exactly reproduce National Statistics aggregates.
Data • Monthly Wages and Salaries Survey (MWSS) • Covers all industries & sectors; only samples units with 20+ employment • Largest units = census; smaller units = stratified random sample (by industry, size, legal status). Total sample 6,000 per month; response rate ≈ >80%. • Units stay in sample for multiple years, or indefinitely – can track over time • Survey is short: total [monthly] paybill, bonuses, arrears, number of employees • Regular pay = total minus bonuses and arrears • Access MWSS microdata via SRS • Annual files and monthly files; some inconsistencies and naming issues • Not yet available consistent with recent exceptional revision to AWE • Filtering: regular pay per employee; exclude all imputations and ONS-flagged outliers; no account of small firm adjustment
Construction of measures AWE AWE AWE AWE Oct 2018 Nov 2018 Dec 2018 Oct 2019 ? Survey responses: Regular pay and employment Aggregation: Total regular pay and total employment Calculation with aggregates: Average regular pay per … Firm-level growth rates Entry and exit from sample prevent growth rate calculation
Notes on interpretation • All charts relate to “regular pay” (excluding bonuses and arrears) • Will not match published AWE data • Nothing is seasonally adjusted here • Growth rates are mostly month-on-same-month-a-year-ago (i.e. annual growth, e.g. Dec 2019 / Dec 2018) or less frequently month- on-month (e.g. Dec 2019 / Nov 2019) • All series stop in 2019 – no comment on recent wage dynamics and inflation, and consistent microdata not available yet • All data for “whole economy” • Private sector in hidden slides (similar); industry breakdowns possible (ongoing)
Appropriately weighted MWSS data matches AWE fairly closely 200120012002200320042005200620072008200920102011201220122013201420152016201720182019 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Whole economy regular pay, annual growth (%) AWE Firm-level data tend to be higher, likely due to: 1. No “small firm” effect 2. Only continuing firms MWSS mean “AWE-weighted”
Alternative weightings tell slightly different stories, and give additional insights 200120012002200320042005200620072008200920102011201220122013201420152016201720182019 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Whole economy regular pay, annual growth (%) Mean “AWE-weighted” Mean employment-weighted Mean firm-weighted Firm-weighted typically higher aside from the downturn, suggesting more pro-cyclical growth in smaller and/or lower paying firms
2001200220042006200820102012201420162018 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Employment-weighted 2001200220042006200820102012201420162018 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Firm-weighted Mean growth consistently above median Median Mean Mean Median Whole economy regular pay, annual growth (%) Mean typically higher than median, suggesting a right- skewed distribution and inequality of wage growth
Median employment-weighted growth from MWSS correlates better with RTI than AWE 201507 201510 201601 201604 201607 201610 201701 201704 201707 201710 201801 201804 201807 201810 201901 201904 201907 201910 0.0 1.0 2.0 3.0 4.0 5.0 MWSS median employment-weighted AWE RTI median pay Correlation of annual growth AWE MWSS employment-weighted median RTI median AWE 1 0.86 0.76 MWSS employment-weighted median 0.86 1 0.88 RTI median 0.76 0.88 1
The distribution of annual growth in regular pay is wide, and moves around a lot <-10 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 >20 0 2 4 6 8 10 12 Density of annual growth in regular pay (%), whole economy, employment weighted 2001 2007 2019 2008-2018 2002-2006
In most months, there is very little month-on-month pay growth… <-10 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 >20 0 2 4 6 8 10 12 14 16 18 20 Density of month-on-month growth in regular pay (%), whole economy, employment weighted, pooled 2001-2019 January April Every other month December except January and April
Falls in average regular pay per employee are common across the dataset Period averages, alternative weightings Month on month Month on same month a year ago Unweighted 2001-2010 47% 28% 2011-2019 48% 34% AWE weighted 2001-2010 46% 22% 2011-2019 47% 30% Employment weighted 2001-2010 47% 23% 2011-2019 48% 31% Firm weighted Unweighted average by month Month on month Month on same month a year ago January 51% 31% February 48% 30% March 47% 31% April 41% 31% May 48% 31% June 47% 31% July 49% 31% August 49% 31% September 47% 30% October 47% 30% November 48% 30% Proportion of firms reporting a fall in their average regular pay
Applications and other remarks • Were the elevated levels of pay growth in recent years driven by higher pay growth amongst all firms or only some firms? • Can decompose aggregate wage growth into within-firm and between effects • Can see wage growth across the firm-level distribution • Can help to reconcile different measures of wages and earnings • MWSS employment-weighted median more similar to RTI median pay • Wide distribution of changes in average wages across firms, and surprising frequency of wage falls – likely partly reflects compositional changes within firm (e.g. hiring and firing) • But also likely some measurement error
Conclusions • Great scope for insights from microdata analysis of MWSS even without data linkage • Opportunities to improve MWSS microdata and AWE • Initial exploration gives sensible results: • Sample of ‘continuing’ firms tends to have higher annual wage growth than AWE • Firm-weighted average wage growth is more pro-cyclical, reflecting greater weight on smaller firms • Mean growth higher than median, reflecting right-skewed distribution • January and April are key months for wage growth
@resfoundation
128 Tax data lets us dive deeper into workers’ earnings @resfoundation • The dataset: a 1% sample of employees (and all their payslips) in PAYE system, 2014 to 2019 • 250,000 employees, 27m payslips • Link to survey data (ASHE) in April for more info. about workers & employers • Benefits: big sample; measured at maximum frequency • Limitations: no self-employment or other income; no household info.
129 Mean of absolute arc percentage annual change in real weekly earnings among 20-59-year-olds: UK Notes: Latest data points are 2023 (LFS), 2022 (BHPS/UKHLS), 2020 (ASHE/NESPD). Earnings are adjusted for CPI inflation. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of ISER, British Household Panel Survey; ISER, UK Household Longitudinal Study; ONS, Five-Quarter Longitudinal Labour Force Survey; ONS, Annual Survey of Hours and Earnings / New Earnings Survey Panel. @resfoundation Long-term: average volatility at annual frequency flat/falling
130 Mean of absolute arc percentage annual change in real weekly earnings among 20-59-year-olds: UK Drivers of lower ‘labour market’ volatility since 1990s: • Lower rates of entry to and exit from work • Lower use of temporary contracts Notes: Latest data points are 2023 (LFS), 2022 (BHPS/UKHLS), 2020 (ASHE/NESPD). Earnings are adjusted for CPI inflation. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of ISER, British Household Panel Survey; ISER, UK Household Longitudinal Study; ONS, Five-Quarter Longitudinal Labour Force Survey; ONS, Annual Survey of Hours and Earnings / New Earnings Survey Panel. @resfoundation Long-term: average volatility at annual frequency flat/falling
131 Distribution of the arc percentage change in real monthly earnings compared to the previous month among 20-59-year-olds working in both months: UK, 2014-15 to 2018-19 Notes: Earnings are deflated using CPIH. Results are pooled across all the months in dataset. For this figure, counts are based on arc-percentage change rounded to the nearest percentage point, which means ‘zero change’ in fact relates to arc-percentage changes between -0.5% and 0.5%, and ‘1-10% change’ in fact means changes between 0.5% and 10.5%, and equivalently for the other categories. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. @resfoundation Turning to the payslip data: earnings are same as previous month only in 4-in-10 months The average absolute change in earnings between months (among people working both months) was 15%. That’s similar in magnitude to what average households spend on food & clothing.
132 Stylised examples of earnings to illustrate ‘trajectory’ categories Definitions of trajectory categories: Extremely stable: all months within 5% of annual average Highly stable: all months within 10% of annual average Small blip: 10-25% from annual average Large blip: 25%+ from annual average Erratic: 4+ months where pay 25%+ from average These are artificial series drawn for illustration and do not represent actual data. @resfoundation Looking at separate months misses deep volatility faced by some: multiple large changes within year
133 Proportion of employees aged 20-59 and working all months in the year, by category of within-year earnings trajectory: UK, 2014-15 to 2018-19 Notes: Results are pooled across financial years. Analysis is based on real earnings, deflated using CPIH. Source: Analysis of HMRC PAYE dataset. @resfoundation Looking at separate months misses deep volatility faced by some: multiple large changes within year
134 Proportion of employees aged 20-59 and working all months in the year, by category of within-year earnings trajectory: UK, 2014-15 to 2018-19 Notes: Results are pooled across financial years. ‘Stable’ here includes the ‘Extremely stable’ and ‘highly stable’ categories as defined earlier. ‘Blips’ includes categories involving 1-2 small or large blips. Analysis is based on real earnings, deflated using CPIH. Source: Analysis of HMRC PAYE dataset and HMRC-ASHE PAYE dataset. @resfoundation ‘Erratic’ pay is most common among workers who are young, low-paid, in temporary jobs, working multiple jobs
135 Proportion of employees on a zero-hours contract (2018-2023), and average arc percentage change in real monthly earnings compared to the previous month among 20-59-year-olds (working both months): UK Notes: Industries not labelled are Manufacturing, Real estate, Professional services, Vehicle sales & repair, ICT, and 'Other' services. Size of bubble indicates the total employees in industry. Earnings are deflated using CPIH. Earnings volatility data is pooled across all months in dataset and zero-hours contract data is pooled across all LFS quarters where the variables are available across 2018-2023. Source: Analysis of HMRC-ASHE PAYE dataset; ONS, Labour Force Survey @resfoundation Volatile pay is associated with use of zero-hours contracts
136 Average arc percentage change in real monthly earnings, finance and insurance versus the rest of economy (workers working in both months): UK, 2014-15 to 2018- 19 Notes: Earnings are deflated using CPIH. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of HMRC-ASHE PAYE data. @resfoundation Not all volatility is bad: earnings volatility in some sectors driven by big bonuses
137 Average arc percentage change in real monthly earnings, finance and insurance versus the rest of economy (workers working in both months): UK, 2014-15 to 2018- 19 Every March, bonuses comprise more than half (55 per cent) of average earnings in Finance & insurance, versus 8 per cent for other workers Notes: Earnings are deflated using CPIH. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of HMRC-ASHE PAYE data. @resfoundation Not all volatility is bad: earnings volatility in some sectors driven by big bonuses
138 Anxiety relating to ‘unexpected changes to my hours of work’, by hourly pay quintile: UK, 2017 Recent qualitative research from NEST Insight found financial volatility can have: Financial costs: fees for missed bills; unable to benefit from cost-effective financial products which require lump sum payments. Psychological costs: stress of managing unpredictable incomes, constant need to monitor and adjust spending. Notes: Includes workers aged 20 to 65 only. Source: Analysis of UK Skills and Employment Survey. @resfoundation Unstable pay can have negative impacts
139 What can Government and employers do? @resfoundation Government: • ZHC reforms will help workers experiencing high volatility • Strengthen statutory sick pay, don’t just extend coverage • Universal Credit may amplify volatility, but solutions hard • Help low-paid workers build financial buffers Employers: • Pay staff at the frequency that most suits them
Unpacking Spatial Earnings Inequality in the UK: The role of People, Places and Industry Effects Richmond Egyei – Resolution Foundation and ESCoE Richmond.egyei@resolutionfoundation.org Richmond.Egyei@kcl.ac.uk
Acknowledgements/Disclaimer Acknowledgments: This work was funded by the Ministry of Housing, Communities and Local Government, through a grant to the Economic and Social Research Council, as part of a research programme initiated by the former Levelling Up Advisory Council, [Grant Number ES/Z000130/1]. Disclaimer: This is independent research and does not represent government policy. The statistical data used here are from the DfE, the ONS, HMRC, and HESA, and are Crown copyright and reproduced with the permission of the controller. The use of the data in this work does not imply the endorsement of the data providers in relation to the interpretation or analysis of the statistical data.
Overview T2 T3 T4 T1 Get data Perform Analysis Draft paper Submit Enrico Vanino – University of Sheffield Tasos Kitsos & Dalila Ribaudo – Aston University Richmond Egyei & Gregory Thwaites & Emily Fry – The Resolution Foundation
Overview What do we do? Leverage employer-employee data to find the drivers of spatial earning disparities in the UK during 2013-2020 How do we do it? AKM regressions at the individual level + variance decomposition analysis (Overman & Xu, 2022; Card et al., 2025) What do we find? place effects stronger than previously and portable; within industry variation explains most of area effects (rather than across areas industry composition); industry specific education pay premia largely drive place-based education wage premia Why is it important? Addressing spatial economic imbalances; policy choices (i.e. moving functions rather than moving industries)
Introduction Large and persistent spatial disparities both across and within UK regions in terms of income and economic opportunities (ONS, 2018; McCann, 2020).
Introduction
Introduction Large and persistent spatial disparities both across and within UK regions in terms of income and economic opportunities (ONS, 2018; McCann, 2020). Sub-national spatial disparities are substantial also among other OECD countries, with income, productivity, skills, and knowledge increasingly polarised in few large urban areas (Behrens and Robert-Nicoud, 2014; Eeckhout et al., 2014; Autor, 2019; Davis and Dingel, 2019). Reducing these regional disparities could address part of the country’s economic growth problems, ensuring more opportunities and prosperity for people across the country (ONS, 2021). Still limited understanding and robust evidence on size, nature and drivers of these regional inequalities.
Aims  Better understand what drives differences in local earnings, focusing on interactions in the distribution of three factors: workers, firms, and area characteristics across commuting zones (TTWAs).  Micro-level analysis of how much spatial imbalances in earnings across and within places could be explained by: • Workers’ education, skills, and location choices. • Local industrial structure, size, and productivity of firms. • Physical capital of places (infrastructure, public services, housing and amenities).
Literature  Distribution of skills (Diamond and Gaubert, 2022), and presence of urban wage premia (Moretti, 2010; Overman and Xu, 2022).  Local industrial structure (Mealy and Coyle, 2022; Card et al., 2025), firms’ size, tasks (Koster and Ozgen, 2021; Duranton and Puga, 2005) and productivity (Hart et al., 2020).  Spatial inequalities in availability of public infrastructure (Gibbons et al., 2024), housing (Hilber and Vermeulen, 2016; Cheshire, 2019) and amenities (Diamond, 2016).  Variance decomposition analysis to explain earnings imbalances across and within places (Gibbons et al., 2014; Overman and Xu, 2022; Card et al., 2025).
Data  Longitudinal Education Outcomes (LEO) links educational pathways and (early) labour market outcomes for 38 million individuals: • National Pupil Database school and early years data; • Higher Education Statistics Agency data; • Individual Learner Record data for colleges and apprenticeships; • HMRC and DWP data on earnings, benefits and tax data; • Inter-departmental Business Register on population of businesses.  Data cleaning and preparation resulting in 4 million individuals: • Workers living in England; • Between 22 and 36 years of age; • In full-time employment; • Observed from 2013 to 2020.
Analysis Figure 1: Median, bottom 10, and top 90 yearly earnings distribution across TTWAs in England in 2020.
Analysis Figure 2: Share top 20th percentile yearly earnings distribution across TTWAs in England in 2020.
Design 1. Estimate components of individuals’ earnings: workers’ observed and unobserved factors, employers’ (area) effect, residuals. a) Are the TTWA wage premia constant across space? 2. Calculate share of spatial variance in earnings driven by workers’ skills, industry composition, and other area characteristics. b) To what extent are TTWA wage premia due to composition adjusted local wage premia (i.e. within industry spatial pay variation), industrial composition (i.e. more industries that pay better) or a combination of the two?
Methods AKM 2WFE earnings regression to estimate contribution of employees’ characteristics (observed and unobserved) and employers’ characteristics: (1) • : individual log annual earnings; • : individual i fixed-effects; • : employer f fixed-effects; • : individual i time variant characteristics (age, experience, tenure, etc.); • : year t fixed-effects.
(1) (2) (3) (4) (5) Age 0.0261*** 0.0265*** 0.026*** 0.0264*** 0.0263*** (0.001) (0.0007) (0.0007) (0.0007) (0.0006) Experience 0.2752*** 0.2344*** 0.2340*** 0.2334*** 0.2354*** (0.028) (0.007) (0.007) (0.007) (0.007) Tenure 0.0405*** 0.0441*** 0.0471*** 0.0449*** 0.0456*** (0.0006) (0.001) (0.0008) (0.001) (0.0015) Exp.London 0.1157*** 0.1224*** 0.1142*** 0.1150*** (0.005) (0.005) (0.006) (0.0047) Exp. Major City 0.0288*** 0.0473*** 0.0290*** 0.0549*** (0.005) (0.004) (0.0055) (0.0036) Current Urban Large City London 2nd Tier City CU x Exp. 0.0046 0.0165*** -0.0139* (0.006) (0.003) (0.0056) CU x Tenure -0.0048* -0.0008 -0.005 (0.002) (0.002) (0.0034) CU x Exp. London -0.0099** -0.0012 -0.0127*** (0.002) (0.001) (0.0019) CU x Exp. Major City -0.0274*** -0.0225*** -0.0022 (0.0015) (0.002) (0.0014) Employee FE Y Y Y Y Y Employer FE Y Y Y Y Y Year FE Y Y Y Y Y No. Obs. 26,810,593 26,810,593 26,810,593 26,810,593 26,810,593 R-sq. 0.3192 0.3326 0.3371 0.3370 0.3371
Methods Locational wage premium associated with TTWA c as weighted average of employer effects in that TTWA: TTWA-by-Industry (SIC4) average wage premia:
Analysis  Use AKM model in equation 1 to decompose the variance in mean wage across TTWAs into the different components:
(1) (2) (3) (4) (5) Age 0.0261*** 0.0265*** 0.026*** 0.0264*** 0.0263*** (0.001) (0.0007) (0.0007) (0.0007) (0.0006) Experience 0.2752*** 0.2344*** 0.2340*** 0.2334*** 0.2354*** (0.028) (0.007) (0.007) (0.007) (0.007) Tenure 0.0405*** 0.0441*** 0.0471*** 0.0449*** 0.0456*** (0.0006) (0.001) (0.0008) (0.001) (0.0015) Exp.London 0.1157*** 0.1224*** 0.1142*** 0.1150*** (0.005) (0.005) (0.006) (0.0047) Exp. Major City 0.0288*** 0.0473*** 0.0290*** 0.0549*** (0.005) (0.004) (0.0055) (0.0036) Current Urban Large City London 2nd Tier City CU x Exp. 0.0046 0.0165*** -0.0139* (0.006) (0.003) (0.0056) CU x Tenure -0.0048* -0.0008 -0.005 (0.002) (0.002) (0.0034) CU x Exp. London -0.0099** -0.0012 -0.0127*** (0.002) (0.001) (0.0019) CU x Exp. Major City -0.0274*** -0.0225*** -0.0022 (0.0015) (0.002) (0.0014) Employee FE Y Y Y Y Y Employer FE Y Y Y Y Y Year FE Y Y Y Y Y No. Obs. 26,810,593 26,810,593 26,810,593 26,810,593 26,810,593 R-sq. 0.3192 0.3326 0.3371 0.3370 0.3371
Analysis (0) (1) (2) (3) Variance (Earnings) 0.010 0.010 0.010 0.010 Area Effect 0.34 0.373 0.316 0.328 Individual Fixed Effect 0.236 0.227 0.238 0.243 Individual Observables 0.000 0.009 0.017 0.154 Correlations (2 x covariance) Area effect, Individual FE 0.404 0.422 0.254 0.227 Area effect, Individual observables 0.000 -0.014 0.184 0.205 Individual FE, Individual observables 0.000 -0.003 -0.152 -0.147 Fixed Effects Y Y Y Y Employee Experience N Y Y Y Lab market experience N N Y Y Labour market x Urban N N N Y
Analysis Figure 3: Quintile distribution of Locational Wage Premium across TTWAs in England.
Methods Similarly, we calculate the variance in the area wage premia explained by the 3 industry components. In this way, we can decompose the TTWA wage premium in 3 industry components:
Analysis Area Effect (S.D) 0.037 Decomposition (variance share) Industry Earnings Premium 0.821 Industry Composition 0.005 Interaction Effect 0.016 cov(earnings premium, composition) 0.005 cov(earnings premium, interaction) 0. 136 cov(composition, interaction) 0.005 Table 3: TTWA Area Wage Premium variance decomposition.
Analysis Fig 4. Quintile distribution of TTWA wage premium components in England.
Analysis We repeat our main analysis focussing on returns to education across places To do this, we split our sample into college graduates and non- graduates and estimate the TTWA premium
Analysis Then we estimate the local wage premium by education group and decompose the difference in wages between our two groups
Analysis Earnings differences (S.D) 0. 308 Decomposition (variance share) Average TTWA education premium 0.172 Person effects between high and low education 0.59 Person time-varying effects 0.0655 cov(edu premium, person effects) 0.544 cov(edu premium, person time-varying) -0.123 cov(person effects, person time-varying) -0.249 Table : College Wage Premium variance decomposition.
Analysis And decompose TTWA education wage premia
Analysis Area Effect (S.D) 0.012 Decomposition (variance share) Education premium in TTWA 0.128 Industry Composition/more graduates in high paying industries 0.551 Sorting – clustering of workers in high or low paying industries 0.008 cov(edu premium, composition) 0.354 cov(edu premium, sorting) -0.032 cov(composition, sorting) -0.009 Table: TTWA Area Wage Premium variance decomposition.
Analysis Fig 6. Distribution of education wage premium and its decomposition
Conclusions Larger place effects are important for policymaking Within industry variations driving place effects point to different functions within industries across space. • College analysis corroborates this. Large urban experience follows workers. Next Steps: Real wages; what role for amenities/urban infrastructure; firm size & labour market characteristics
Conclusions Thank you richmond.egyei@kcl.ac.uk
Real Time Information (RTI) –The Impact of Health Interventions on Employment and Earnings Daniel Ayoubkhani, Emma Sharland, Charlotte Bermingham, Klaudia Rzepnicka, Isobel Ward, Hannah Bunk, & Vahé Nafilyan Health Research Group, Health and International Directorate Office for National Statistics
Background • The Get Britain Working White Paper identifies that reversing the increase in economic inactivity due to ill-health is a national priority • The Government have committed to expanding services such as NHS Talking Therapies to reduce mental ill-health and improve employment outcomes • Obesity is also a key priority for prevention for the government as it is a key risk factor for leading conditions that may be driving health-related economic inactivity • Kickstarting Economic Growth is also one of the new Labour Government's missions and this includes supporting people who are economically inactive due to ill-health back to work • Policy interest (from HMT, DHSC, NHSE) in which conditions are driving inactivity, and which interventions are most effective at increasing employment
Linked datasets held by ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2011 Census 2021 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures
Linked datasets held by ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2011 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures 97% linkage rate via Demographic Index (via 2011-2013 NHS Patient Registers and direct matching) 95% linkage rate via 2011-2013 NHS Patient Registers Estimated to cover 94% of the population of England & Wales
Linked datasets held by ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2021 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures 97% linkage rate via Demographic Index 96% linkage rate via NHS Personal Demographics Service Estimated to cover 97% of the population of England & Wales
Bariatric Surgery Aim: To estimate the average change in employee, pay and probability of paid employment attributable to bariatric surgery. This project was funded by the UK government’s Labour Markets Evaluation and Pilots Fund (2024 to 2025).​
Study population: • 40,662 individuals with a record of bariatric surgery in HES between 1 April 2014 and 31 December 2022, with an obesity diagnosis, and no prior record of bariatric surgery since 1 April 2009. • Individuals were resident in England and aged 25 to 64 on the date of bariatric surgery. • An unexposed sample of 49,921 individuals from the general population, age-sex stratified to match the exposed sample, who had not had bariatric surgery was also included. Methods: • We estimate the effect of having had bariatric surgery, at different time points after surgery, using fixed effects regression modelling. • We include fixed effects for individuals to account for confounding due to differences between individuals. • We include calendar time to account for changes in background economic conditions and age to account for changes in pay and likelihood of working over the working life. Bariatric surgery: Methods Data We use the 2011 and 2021 Censuses linked to: •Hospital Episode Statistics (HES) Admitted Patient Care (APC) records •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC)
Sustained increase in probability of employment from four months after surgery Little increase in pay among those in work from six months after surgery Initial decrease indicative of surgery recovery time Sustained increase in pay from six months after surgery Bariatric surgery: Key findings
NHS Talking Therapies Aim: To estimate the average change in employee, pay and probability of paid employment attributable to completing NHS Talking Therapies This project was funded by the Cabinet Office and HM Treasury Evaluation Task Force's Evaluation Accelerator Fund (2023 to 2024) and by the UK government's Labour Markets Evaluation and Pilots Fund (2024 to 2025).
Study population: • 842,127 individuals with; • a referral to NHSTT between 1 April 2016 and 31 March 2020 • attended at least one therapy session, • were at clinical caseness, • aged 25 to 60 years old at the time of referral • resident in England • Completed treatment group: 593,300 (70.5%) • Dropped out of treatment group: 248,827 (29.5%) Methods: • We estimate the effect of completing NHSTT, using fixed effects regression modelling. • We include fixed effects for individuals to account for confounding due to differences between individuals. • We include calendar time to account for changes in background economic conditions and age to account for changes in pay and likelihood of working over the working life. • Inverse Probability Weights (IPW) are used to balance the treatment and control groups NHSTT: Methods Data We use the 2011 Census linked to: •NHS Talking Therapies data (formerly IAPT) •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC)
NHSTT: Key findings We saw a maximum average increase in pay of £17 per month by year two post first therapy (time zero) For employment, we see sustained increases in employment from year 1, peaking at 1.5 p.p increase at year seven.
The “not working, seeking work” group have the biggest increases in the probability of being a paid employee following completion of treatment in NHSTT. This peaks at around 3.1 p.p. in year 4 and 5.
Endometriosis Aim: What is the impact of having a hospital diagnosis of endometriosis on women's labour market participation and progression? This project was funded by the UK government’s Labour Markets Evaluation and Pilots Fund (2024 to 2025).​
Study population: 55,290 women with a primary diagnosis of endometriosis in the Hospital Episode Statistics (HES) Admitted Patient Care (APC) data between 1 April 2016 and 31 December 2022 who were:  Enumerated in the 2011 Census  Linked to an NHS number and National Insurance number  Resident in England, as recorded in HES if available and the 2011 Census if not  Aged 25 to 54 years on the date of endometriosis diagnosis  Recorded as being female in both HES and the 2011 Census  Had no primary or secondary diagnosis of endometriosis in HES between 1 April 2009 and 31 March 2016 Method:  Fixed effects (within-person) regression modelling  We adjust for age, calendar time and births  Monthly pay was deflated to 2023 prices  Being a paid employee was defined as receiving a monthly pay greater than £0 Data We use the 2011 Census linked to: •Hospital Episode Statistics •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC) Endometriosis: Methods
Endometriosis: Key findings There was a sustained average decrease in pay. Monthly average pay among all women (both in paid work and not) reached £130 lower than pre-diagnosis levels four to five years after diagnosis. Monthly average pay among those in paid work reached £56 lower than pre-diagnosis levels four to five years after diagnosis. Monthly average pay among those in paid work reached £56 lower than pre-diagnosis levels four to five years after diagnosis.
Future plans
Work programme for 2024/25 NHS Talking Therapies (Publication: 9 Dec 2024) Bariatric surgery (Publication: 23 Oct 2024) NHS Diabetes Prevention Programme (Publication: Jul 2025) Endometriosis (Publication: 5 Feb 2026) MSK surgery (Publication: 3 March 2025) Adverse pregnancy events (Publication: Jun 2025) Major conditions (Publication: Jun 2025) Published: Scheduled for 2025: Infectious diseases (Publication: TBC 2025) NHS Talking Therapies (Benefits and Healthcare Utilisation) (Publication: TBC 2025) Long-term ill health and benefits receipt (Publication: 26 Feb 2026) Impact of waiting times (Publication: TBC 2025)

More Related Content

PPTX
Unpredictable Earnings: The volatility of pay packets and its impact on livin...
byResolutionFoundation
 
PPTX
Earnings Event
byOffice for National Statistics
 
PDF
Moore stephens consultants 6pp dl spring 2013
byLucienneP
 
PPTX
ONS household income statistics user event
byOffice for National Statistics
 
PDF
Why 2014 hasn’t been the year of the pay rise
byResolutionFoundation
 
PDF
Full slide-pack-final
byKROTOASA FOUNDATION
 
PDF
Pay trends 2015
bycherylrmorgan
 
PDF
Presentation by Matthew Whittaker
byResolutionFoundation
 
Unpredictable Earnings: The volatility of pay packets and its impact on livin...
byResolutionFoundation
 
Earnings Event
byOffice for National Statistics
 
Moore stephens consultants 6pp dl spring 2013
byLucienneP
 
ONS household income statistics user event
byOffice for National Statistics
 
Why 2014 hasn’t been the year of the pay rise
byResolutionFoundation
 
Full slide-pack-final
byKROTOASA FOUNDATION
 
Pay trends 2015
bycherylrmorgan
 
Presentation by Matthew Whittaker
byResolutionFoundation
 

Similar to Earnings Symposium Slidepack - 29 April 2025

PDF
Webinar - What's Next for Annual Reviews
byPayScale, Inc.
 
PPTX
Unleashing the power of sage payroll and hr
byAndyIrvine
 
PDF
2024 Labor Market Insights: How Emerging Trends Are Shaping the Workforce
byPayScale, Inc.
 
PDF
Webinar - How to Build Fair Compensation Policies That Support Pay Equity
byPayScale, Inc.
 
ODP
What is Real Time Information (RTI)?
byHudman Limited
 
PDF
“Job Quality, Labour Market Performance and Well-Being”_Hijzen_extending the ...
byStatsCommunications
 
PPTX
Handling Data in AS and A2 Economics
bytutor2u
 
PDF
Activity Comparing Industry Trends In Pay Rates Such As It, Fmcg
bysimply_coool
 
PDF
Employment law update - Browne Jacobson Exeter - 06 February 2020
byBrowne Jacobson LLP
 
DOC
13
byDreams Design
 
DOCX
Project report on compensation and benefits
bysukesh gowda
 
PDF
Performance equity global-gilead-20100627
byPERFORMENSATION
 
PDF
HMRC RTI changes checklist 2013
byKevsharkey
 
PPTX
Confidence in Handling Economic Data
byEton College
 
PDF
Regulations, Acts, and everything else you need to know about payroll legisla...
bySoftworld
 
PPTX
It’s the economy stupid
byThe Cloud School Ltd
 
PDF
January Labour Market Update
byKaylan Bland
 
PDF
Time to catch up? Living standards in the downturn and recovery
byResolutionFoundation
 
PDF
Plummer Parsons Chartered Accountants Mini Guide Insider June 2011
bynevillebeckhurst
 
PDF
June Labour Market Update
byKaylan Bland
 
Webinar - What's Next for Annual Reviews
byPayScale, Inc.
 
Unleashing the power of sage payroll and hr
byAndyIrvine
 
2024 Labor Market Insights: How Emerging Trends Are Shaping the Workforce
byPayScale, Inc.
 
Webinar - How to Build Fair Compensation Policies That Support Pay Equity
byPayScale, Inc.
 
What is Real Time Information (RTI)?
byHudman Limited
 
“Job Quality, Labour Market Performance and Well-Being”_Hijzen_extending the ...
byStatsCommunications
 
Handling Data in AS and A2 Economics
bytutor2u
 
Activity Comparing Industry Trends In Pay Rates Such As It, Fmcg
bysimply_coool
 
Employment law update - Browne Jacobson Exeter - 06 February 2020
byBrowne Jacobson LLP
 
13
byDreams Design
 
Project report on compensation and benefits
bysukesh gowda
 
Performance equity global-gilead-20100627
byPERFORMENSATION
 
HMRC RTI changes checklist 2013
byKevsharkey
 
Confidence in Handling Economic Data
byEton College
 
Regulations, Acts, and everything else you need to know about payroll legisla...
bySoftworld
 
It’s the economy stupid
byThe Cloud School Ltd
 
January Labour Market Update
byKaylan Bland
 
Time to catch up? Living standards in the downturn and recovery
byResolutionFoundation
 
Plummer Parsons Chartered Accountants Mini Guide Insider June 2011
bynevillebeckhurst
 
June Labour Market Update
byKaylan Bland
 

More from Office for National Statistics

PDF
ONS Economic Forum Slidepack – 21 July 2025
byOffice for National Statistics
 
PPTX
Growth by numbers: an introduction to GDP.pptx
byOffice for National Statistics
 
PDF
ONS Economic Forum - 22 September 2025 - Slidepack
byOffice for National Statistics
 
PPTX
ONS Economic Forum Slidepack – 19 May 2025.pptx
byOffice for National Statistics
 
PPTX
Measuring the economy slidepack - 30 September 2025 SlideShare.pptx
byOffice for National Statistics
 
PPTX
ONS Economic Forum Slidepack - 24 March 2025 (slideshare).pptx
byOffice for National Statistics
 
PPTX
SlideShare Annual crime and justice statistics forum - 7 November 2024
byOffice for National Statistics
 
PPTX
Bringing data to life: Artificial Intelligence and innovation - keeping human...
byOffice for National Statistics
 
PPTX
The truth behind the numbers: spotting statistical misuse
byOffice for National Statistics
 
PPTX
Reference Data Management Framework Overview Digital Booklet
byOffice for National Statistics
 
PPTX
Bringing data to life: How are your vitals? Exploring health by numbers
byOffice for National Statistics
 
PPTX
Bringing data to life | Bricks, mortar and data: understanding house and rent...
byOffice for National Statistics
 
PPTX
Bringing data to life - Crime webinar Accessible.pptx
byOffice for National Statistics
 
PPTX
Measuring what matters most: understanding national well-being
byOffice for National Statistics
 
PPTX
SlideShare ONS Economic Forum Slidepack - 27 January 2025
byOffice for National Statistics
 
PPTX
Global journeys: estimating international migration
byOffice for National Statistics
 
PPTX
Numbers of a nation: how we estimate population statistics | Accessible slides
byOffice for National Statistics
 
PPTX
A Quick Introduction to the Reference Data Management Framework
byOffice for National Statistics
 
PDF
ONS Economic Forum Slidepack - 30 September 2025
byOffice for National Statistics
 
PPTX
Connecting the dots: the benefit of linking data on health and wealth | Acces...
byOffice for National Statistics
 
ONS Economic Forum Slidepack – 21 July 2025
byOffice for National Statistics
 
Growth by numbers: an introduction to GDP.pptx
byOffice for National Statistics
 
ONS Economic Forum - 22 September 2025 - Slidepack
byOffice for National Statistics
 
ONS Economic Forum Slidepack – 19 May 2025.pptx
byOffice for National Statistics
 
Measuring the economy slidepack - 30 September 2025 SlideShare.pptx
byOffice for National Statistics
 
ONS Economic Forum Slidepack - 24 March 2025 (slideshare).pptx
byOffice for National Statistics
 
SlideShare Annual crime and justice statistics forum - 7 November 2024
byOffice for National Statistics
 
Bringing data to life: Artificial Intelligence and innovation - keeping human...
byOffice for National Statistics
 
The truth behind the numbers: spotting statistical misuse
byOffice for National Statistics
 
Reference Data Management Framework Overview Digital Booklet
byOffice for National Statistics
 
Bringing data to life: How are your vitals? Exploring health by numbers
byOffice for National Statistics
 
Bringing data to life | Bricks, mortar and data: understanding house and rent...
byOffice for National Statistics
 
Bringing data to life - Crime webinar Accessible.pptx
byOffice for National Statistics
 
Measuring what matters most: understanding national well-being
byOffice for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 27 January 2025
byOffice for National Statistics
 
Global journeys: estimating international migration
byOffice for National Statistics
 
Numbers of a nation: how we estimate population statistics | Accessible slides
byOffice for National Statistics
 
A Quick Introduction to the Reference Data Management Framework
byOffice for National Statistics
 
ONS Economic Forum Slidepack - 30 September 2025
byOffice for National Statistics
 
Connecting the dots: the benefit of linking data on health and wealth | Acces...
byOffice for National Statistics
 

Recently uploaded

PPTX
time value of money, investment criteria and project rating index.pptx
byYogesh Dixit
 
PPTX
National income definition and calculations.pptx
byshanthinavu
 
PPTX
CERI III Transparency in Public Sector Accounting -Prof Oyedokun.pptx
byGodwin Emmanuel Oyedokun MBA MSc PhD FCA FCTI FCNA CFE FFAR
 
PDF
COFACE - Encuesta 2025 riesgos en pagos corporativos en Francia. NOTA DE PREN...
byJaime Cubillo Fleming
 
PDF
FESE Capital Markets Fact Sheet 2025 Q3.pdf
bysecretariat4
 
PDF
Bladex Earnings Call Presentation 3Q2025
byBladex
 
DOCX
JPMorgan's Crypto Revolution: How Niufo Leads the Digital Asset Evolution
byNiufo
 
PDF
A PREDICTION FOR THE INTRODUCTION AND IMPLEMENTATION OF A NEW FINANCIAL SYSTE...
bynakasotoshimoto
 
PDF
Enhancing crypto Security for Enterprises & Exchanges in mena
byLiminal Custody
 
PDF
Understanding How Abhay Bhutada Foundation Fulfills SEBI ESG Criteria
byRaj Kumble
 
DOCX
Niufo Analysis: XRP Burn Activity Signals Potential Market Recovery
byNiufo
 
DOCX
Bitcoin Market Analysis: AHRFD Platform Insights on $115K Breakthrough
byAHRFD
 
DOCX
Is it legal to buy verified Binance accounts (Personal &amp .docx
bymarketing
 
PDF
Top 11 Best Site To Buy Snapchat Account in 2025.pdf
bymarketing
 
DOCX
Top 11 Best Site To Buy Snapchat Account in 2025.docx
bymarketing
 
PPTX
The Best Trust Converter Tools Application
bywalterfierz2
 
PDF
Real Assets as a Hedge Against Inflation
byCelestin Pepin
 
PDF
The Eight Faces of Weakness: Understanding Your Organization's Hidden Fragili...
byTheta Prime Institute, Cheyenne, Wyoming, USA.
 
PPTX
原版复刻英国密德萨斯大学毕业证(MDX毕业证书)成绩单信封在读证明可查学历认证
by南澳大利亚大学学历认证【Q微号:1954 292 140】研究生文凭(UniSA毕业证成绩单)
 
PPTX
SIR BABA KHAIL 04.pptxgghhhhhhhhhhhhhhhh
bysarimanjum55
 
time value of money, investment criteria and project rating index.pptx
byYogesh Dixit
 
National income definition and calculations.pptx
byshanthinavu
 
CERI III Transparency in Public Sector Accounting -Prof Oyedokun.pptx
byGodwin Emmanuel Oyedokun MBA MSc PhD FCA FCTI FCNA CFE FFAR
 
COFACE - Encuesta 2025 riesgos en pagos corporativos en Francia. NOTA DE PREN...
byJaime Cubillo Fleming
 
FESE Capital Markets Fact Sheet 2025 Q3.pdf
bysecretariat4
 
Bladex Earnings Call Presentation 3Q2025
byBladex
 
JPMorgan's Crypto Revolution: How Niufo Leads the Digital Asset Evolution
byNiufo
 
A PREDICTION FOR THE INTRODUCTION AND IMPLEMENTATION OF A NEW FINANCIAL SYSTE...
bynakasotoshimoto
 
Enhancing crypto Security for Enterprises & Exchanges in mena
byLiminal Custody
 
Understanding How Abhay Bhutada Foundation Fulfills SEBI ESG Criteria
byRaj Kumble
 
Niufo Analysis: XRP Burn Activity Signals Potential Market Recovery
byNiufo
 
Bitcoin Market Analysis: AHRFD Platform Insights on $115K Breakthrough
byAHRFD
 
Is it legal to buy verified Binance accounts (Personal &amp .docx
bymarketing
 
Top 11 Best Site To Buy Snapchat Account in 2025.pdf
bymarketing
 
Top 11 Best Site To Buy Snapchat Account in 2025.docx
bymarketing
 
The Best Trust Converter Tools Application
bywalterfierz2
 
Real Assets as a Hedge Against Inflation
byCelestin Pepin
 
The Eight Faces of Weakness: Understanding Your Organization's Hidden Fragili...
byTheta Prime Institute, Cheyenne, Wyoming, USA.
 
原版复刻英国密德萨斯大学毕业证(MDX毕业证书)成绩单信封在读证明可查学历认证
by南澳大利亚大学学历认证【Q微号:1954 292 140】研究生文凭(UniSA毕业证成绩单)
 
SIR BABA KHAIL 04.pptxgghhhhhhhhhhhhhhhh
bysarimanjum55
 

Earnings Symposium Slidepack - 29 April 2025

  • 1.
  • 2.
    Average Weekly Earnings: Outputsand Development Nicola White Head of AWE outputs and development Office for National Statistics
  • 3.
    AWE: Data collectionand production update • Average weekly earnings (AWE) is the lead monthly measure of average weekly earnings per employee. • It is calculated using information based on the Monthly Wages and Salaries Survey (MWSS), which samples around 9,000 employers, covering around 62% of all employees in Great Britain. • The sample is taken from the IDBR and the data are weighted by the IDBR. • Businesses under the size of 20 employees are excluded from the sample. • The survey response rate is generally around 83%.
  • 4.
    AWE: Data collectionand production update • Data collected on total wages, bonuses, arrears and employment • AWE for any given month is the ratio of estimated total pay for the whole economy, divided by the total number of employees paid. • AWE reflects changes to the composition of the workforce. • The “headline rates” of AWE are the changes in seasonally adjusted average weekly earnings, including and excluding bonuses, comparing the latest 3 months with the same 3 months in the previous year.
  • 5.
    AWE: Data collectionand production update Revisions policy: • first published as provisional, 6 to 7 weeks after the data month end • This provisional data is revised the following month to allow for late and amended data returns from respondents • exceptional revisions may be made if the impact is sufficiently significant • a change in methodology can lead to revisions to the entire historic time series • when seasonally adjusted data for a given month t is published, as well as t–1 being revised, t–2 is also revised, as well as the same 3 months in the previous year • the seasonal adjustment parameters are updated annually, in line with our policy on seasonal adjustment. This update can lead to revisions to the entire historic time series
  • 6.
    How AWE collectsand reports on arrears and bonus payments • Arrears specifically covers earnings arising from a backdated pay increase, not late payment of overtime or bonuses. • A bonus is a form of reward or recognition granted by an employer in addition to basic pay. • Both payments are reflected in estimates at the time they were paid, and not in the period they are awarded for. Therefore, back series are not revised.
  • 7.
    How AWE collectsand reports on arrears and bonus payments • When arrears or bonus payments are backdated, people who have left the business but are entitled to these back payments will be included in the number of employees that have received pay in that period. • This results in more employees being added to payroll for that month and will have an impact on the average pay as more employees will be included in the calculation. • The survey only requests one employee figure, so we are unable to split out those who have left the company and only eligible to the backpay and not the regular pay. • For the majority of time, the impact of this is minimal but for certain periods, where there has been a large backpay covering a long period, the calculation of average pay will be affected and is more accurately reflected in the following month’s data.
  • 8.
    Continually compare AWEwith RTI - discuss differences with HMRC % Single month mean % growth total pay (annual) Seasonally Adjusted Source Tab Earn01 and RT estimates (SA) Jan-21 M ar-21 M ay-21 Jul-21 Sep-21 N ov-21 Jan-22 M ar-22 M ay-22 Jul-22 Sep-22 N ov-22 Jan-23 M ar-23 M ay-23 Jul-23 Sep-23 N ov-23 Jan-24 M ar-24 M ay-24 Jul-24 Sep-24 N ov-24 Jan-25 0 2 4 6 8 10 12 RTI Total AWE Total AWE Total inc arrears
  • 9.
    AWE exceptional revisions •Due to late and corrected data for a large business we have made revisions back to October 2020. • The revisions affected: the whole economy; private sector; services sector; wholesaling, retailing, hotels and restaurants sector; and the retail trade and repairs industry. • At the whole economy level, these revisions are generally small and within the range we would expect to see during seasonal adjustment reviews; as the estimates are broken down below the whole economy level, the revisions become larger.
  • 10.
    AWE: Development workand plans Methodology work: • Annual Seasonal Adjustment review • currently reviewing the seasonal adjustment parameters as part of the annual review. • the review may lead to revisions to the entire historical AWE time series and will be implemented either in the May 2025 or June 2025 publication. • Sample optimisation review • Review sampling allocation • Optimise stratification / sample
  • 11.
    AWE: Development workand plans Methodology work: • Under 20’s – current method / other options • create under 20 factors based on 2023r and 2024p ASHE and assess the impact on the AWE estimates • review the data sources and methods used in the under 20 factors to assess if we can improve them • Up-date uncertainty measures • Review outlier method • Review imputation method
  • 12.
    AWE: Development workand plans • Review revisions policy • Continual coherence assessment in particular with the RTI but also other earnings datasets • Use of RTI that we have in the office
  • 13.
    Annual Survey ofHours and Earnings (ASHE): Overview and Development Plans Lualhati Santiago Head of ASHE outputs and development Office for National Statistics
  • 14.
  • 15.
    Key facts ASHE isour highest quality survey on individual-level employee earnings: • It is reliable (it is completed by employers). • It is timely (data collected in April, published in October). • It is large: the achieved sample for 2024 was 173,000.​
  • 16.
    Data and sample(I): Overview Data: individual-level data on employee earnings (payments made to the employee by the employer) and the hours on which this pay was calculated. Sample: - 175k - allows for detailed analyses & breakdowns.
  • 17.
    Data and sample(II): Selecting the sample Sample of employees in the UK selected by: 1. 1% simple random sample of NINOs (last 2 NI digits); 2. eligible NINOs are selected on the HMRC PAYE frame; 3. linked to ONS IDBR register to obtain businesses’ contact details - a form is sent and employers complete it.
  • 18.
    Data and sample(III): Weights To correct for non-response biases, the data is weighted to the LFS. The ASHE weights are created as a combination of: 1. Design weight. 2. Calibration weight (uses the LFS).
  • 19.
    Timeline of ASHE April May June July August September October Referencedate Microdata ready Data collection, ingestion, processing & validation Survey close date Stats and Bulletins published (Key requirement: capture new NLW/NMW)
  • 20.
    ASHE: published statistics •3 bulletins • Employee earnings in the UK • Low and high pay in the UK • Gender pay gap in the UK • 800 tables with statistics • Microdata (SRS/IDS) Data included: - Current year (provisional data). - Previous year (revised data).
  • 21.
  • 22.
    Why we arereviewing/developing ASHE • Methods and end-to-end system have been in place since the early 2000s and are due a review.​ • Data challenges since the pandemic (2020 onwards):​ •partly due to furlough and coronavirus-related impacts on the labour market,​ •but some remained post-pandemic.
  • 23.
    Development work andplans: 5 workstreams • Methods • Processing system (end-to-end) • Data collection • User needs • Data coherence
  • 24.
    Workstream 1: Methods •Data validation • Selective editing review • Validation and error prioritisation tools • Sampling, weighting and imputation review • Pension variables review
  • 25.
    Workstream 2: Processingsystem (end-to-end) • SOC coding redevelopment • Data processing redevelopment • Table production redevelopment • More End-to-end holistic review
  • 26.
    Workstream 3: Datacollection • Transition to Electronic Questionnaire • Data collection systems: scoping redevelopment End-to-end holistic review
  • 27.
    Workstreams 4 and5: User needs & coherence • User needs: • Earnings User Group (across government) • Earnings Symposium • User consultations • Coherence: • Analytical assessments to compare earnings' sources Ongoing Feeds into workstreams 1-3
  • 28.
    Timeline for ASHEdevelopments 2024 2025 2026 2027 & beyond Data validation Dates TBC • Pension variables review • Transition to Electronic Questionnaire • Sampling, weighting and imputation review • Data collection systems redevelopment Data processing redevelopment SOC coding redevelopment Table production redevelopment
  • 29.
    For questions, comments,and to learn more about ASHE and our plans, please get in touch: [email protected]
  • 30.
    Plans for PAYERTI and ONS Labour Market Statistics Tom Evans Analysis Lead Labour Market Transformation, ONS
  • 31.
    Background • It hasbeen a long-held goal of ONS to secure a regular supply of HMRC Pay As You Earn, Real-time Information data aka. PAYE RTI • In collaboration with HMRC, in recent years ONS has made increasing use of PAYE RTI statistics • Monthly Earnings and Employment from PAYE RTI Provides valuable insights on short-term movements in the labour market. Breakdowns by: Age, Industry (IDBR) and Geography (residence) • Wages and Salaries within the Annual National Accounts • Both of these cases see PAYE RTI used in isolation • Further benefits will arise from linking PAYE RTI with other survey and non- survey sources
  • 32.
    Build a betterunderstanding of existing statistics, with the potential to improve accuracy and precision Enable the production of new statistics and analysis to complement existing activity and provide new insights Broad intentions for PAYE RTI and data linkage
  • 33.
    Existing Statistics andAnalysis for example New Statistics and Analysis Improving understanding of bias Coherence and validation Survey developments Non-linkage outputs Linked Employer-Employee Dataset Intersections between labour market and other topics Expansion of existing uses for PAYE RTI Exploitation of PAYE RTI scale As ever, alignment with concepts and conceptual detail will need to be considered
  • 34.
    Short-term surveys (Transformed) LabourForce Survey ((T)LFS) Sole source for headline labour market rates and levels in addition to a range of other labour market statistics Monthly Wages and Salaries Survey (MWSS) Underpins Average Weekly Earnings (AWE) estimates – lead measure of short-term changes in earnings Short-term Employment Surveys (STES) Main input into Workforce Jobs estimates – key comparator for (T)LFS estimates Structural Surveys Annual Survey of Hours and Earnings (ASHE) Detailed business survey on earnings and hours worked among employees (PAYE-based sample frame) – key input into NLW calculations Business Register and Employment Survey (BRES) Detailed business survey on the geographic and industry characteristics of employment. Only business survey which collects “Local Unit” information. Also used to update the Inter Departmental Business Register Labour Market datasets
  • 35.
  • 36.
    IDS Coherence Project RDMF– Demographic Index CNRLS Labour Market datasets – initial focus LFS TLFS HMRC PAYE RTI 2021 Census Clerical linkage – point-in-time Automated indexing, ongoing
  • 37.
    Data linkage assurance Workingwith relevant areas within ONS to understand automated linkage processes such that we can understand how linkage approach and linkage quality impacts on analytical approach and findings Proof of Concept - Employee/Self-employed Assurance and Exploration Jan-M ar 2014 Jun-Aug 2014 N ov-Jan 2015 Apr-Jun 2015 Sep-N ov 2015 Feb-Apr 2016 Jul-Sep 2016 D ec-Feb 2017 M ay-Jul 2017 O ct-D ec 2017 M ar-M ay 2018 Aug-O ct 2018 Jan-M ar 20195 [r] Jun-Aug 2019 [r] N ov-Jan 2020 [r] Apr-Jun 2020 [r] Sep-N ov 2020 [r] Feb-Apr 2021 [r] Jul-Sep 2021 [r] D ec-Feb 2022 [r] M ay-Jul 2022 [r] O ct-D ec 2022 [r] M ar-M ay 2023 [r] Aug-O ct 2023 [r] Jan-M ar 2024 [r] Jun-Aug 2024 [r] 11.0% 11.5% 12.0% 12.5% 13.0% 13.5% 14.0% 14.5% 15.0% 15.5% Self-employment share (main job) Since the pandemic we have seen a marked and persistent decline in the share of people self-employed in their main job on the LFS, while the share is lower still on TLFS Linkage with PAYE RTI may offer new insights into whether there has been a more permanent change in how people self- classify on a survey vs. their tax definition
  • 38.
    Developing understanding ofbias Future Priorities Bias and coherence are live issues. PAYE RTI linkage could allow us to get closer to an external (albeit partial) benchmark for labour market status. By linking both sample and response files at the household level we can analyse the extent to which non-responding households contain one or more payrolled employees and how this has evolved over time This would build on a one-off analysis done in 2018
  • 39.
  • 40.
    Short-term surveys (Transformed) LabourForce Survey ((T)LFS) Monthly Wages and Salaries Survey (MWSS) Short-term Employment Surveys (STES) Structural Surveys Annual Survey of Hours and Earnings (ASHE) Business Register and Employment Survey (BRES) Existing Statistics and Analysis Improving understanding of bias Coherence and validation Survey developments Non-linkage outputs Build a better understanding of existing statistics, with the potential to improve accuracy and precision Building upon existing aggregate checks, microdata will allow for more detailed coherence and validation checks Initial considerations likely to be around targeted question replacement, though need to be aware of conceptual/definitional differences
  • 41.
    Wider opportunities “Pure” RTIoutputs Monthly PAYE RTI statistics National Accounts inputs Linked analytical insights Linked Employer-Employee Dataset e.g. Productivity insights, career progression Characteristics analysis e.g. insights from linkage with Census and other surveys Exploiting PAYE RTI scale Enable the production of new statistics and analysis to complement existing activity and provide new insights As you can see, there is a large range of potential avenues to explore both within earnings statistics and across the wider labour market portfolio – in a resource constrained environment, coordination and prioritisation will be key. Welcome feedback on where priorities should lie
  • 42.
    Wage and EmploymentDynamics: Damian Whittard1 , Van Phan1 , Felix Ritchie1 , Anni Caden1 , Alex Bryson2 , John Forth3 , Carl Singleton4 , Rachel Scarfe4 1 University of the West of England, Bristol. 2 University College London, London. 3 City University, London. 4 University of Sterling, Stirling.
  • 43.
    Acknowledgements and disclaimer •We gratefully acknowledge ADR UK (Administrative Data Research UK) and the Economic and Social Research Council (Grant No. ES/T013877/1 & ES/Y001184/1) for funding. • The work is based on analysis of the research-ready datasets from the WED Project: Office for National Statistics (2022) Annual Survey of Hours and Earnings linked to 2011 Census - England and Wales - https://doi.org/10.57906/80f7-te97 • The analysis was carried out in the Secure Research Service, part of the Office for National Statistics (ONS). • This work contains statistical data from ONS which is Crown Copyright. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.
  • 44.
    1. The WEDProject 2. ADR UK Research Fellowship • Working Towards an Environmentally Sustainable and Equitable Future? Introduction
  • 45.
    WED Project: Aims Research-readydata • Definitive, clean, fully described data • Fully documented construction • weighting • Sustainable – life beyond the project as core dataset(s) • Training and information events Strategic research • Low pay • Job mobility • Pay gaps • Green jobs
  • 46.
    Research-ready data HMRC data: 1.PAYE 2. Self Assessment 3. Migrant Workers Scan Review QA Link Document Add common personal IDs Strategic research Wage and Employment Spine wage progression low-wage labour markets ONS business register (Enhanced) ASHE 2011 Census WED Project: How it works
  • 47.
    2022  ‘Enhanced’ ASHE ASHE-Census 2011 2024  ASHE - PAYE & SA o ASHE - Migrant Scan Worker 2025 Expected 2026 o ASHE-Census 2021 WED Project: Progress to data
  • 48.
    • WED Website •https://www.wagedynamics.com • Community of Interest group • https://www.wagedynamics.com/community-of-interests/ • Knowledge Hub • https://khub.net/group/wage-and-employment-dynamics-research-group/ Stakeholder Engagement: Supporting the Community
  • 49.
    Working Towards anEnvironmentally Sustainable and Equitable Future? New Evidence on Green Jobs from Linked Administrative Data in the UK Damian Whittard Peter Bradley, Van Phan and Felix Ritchie Journal of Cleaner Production
  • 50.
    Our contribution • Usesa new linked administrative dataset based on high quality earnings information to estimate the economic benefit of working in a green occupation. • New knowledge is presented about the attributes of those who work in green occupations and the characteristics of the jobs and the employers. • Furthering work on attitude-behaviour gaps, the study provides evidence that personal travel behaviours and green employment choices are not consistent • The research adds to the international literature of pay in green jobs, estimating a positive pay premium for England and Wales. • Provides an original contribution revealing that working in a green occupation can offset some of the inter-occupation pay gap, yet within these occupations, gender and ethnic pay gaps persist. • The study emphasises the need for inequalities to be captured by theory that attempts to understand and conceptualise the uptake of green jobs Introduction | Data | Methods | Results | Discussion
  • 51.
    Motivation • Climate crisisand the environmental emergency • Huge economic opportunity (BEIS, 2021). • Green jobs at the heart of any transition • No universally accepted definition of green jobs (Bowen et al, 2018; Sulich 2020) • Lack of high-quality, large scale, longitudinal data (Skidmore, 2022) • Alternative measures of green jobs • Top down (Georgeson and Maslin, 2019; Eurostat, 2021; ) • Bottom-up (Bowen et al., 2018) • Occupations (Valero et al., 20121) • Tasks (Martin, J. and Monahan, E., 2022) • Exploit opportunity provided by linking administrative datasets • Employment (occupations) • Pay (premium or penalty) • Characteristics Introduction | Data | Methods | Results | Discussion
  • 52.
    Data • Annual Surveyof Hours and Earnings (ASHE) (Whittard et al, 2022) • 1% sample of all employees / longitudinal • Provided by their employer • Responses rate – approximately two-thirds • ASHE linked to Census (2011-2018) (See www.wagedynamics.com) • Personal and family characteristics for employees • Linkage results in 0.5 percent of the population in match year (2011) – circa 100,000 obs. • Attrition overtime - 76,000 observations in 2018 • Map US O*NET data on occupations (tasks based) (Dickerson and Morris, 2019) • O*NET identify 12 sectors contributing to the ‘green economy’ • Tasks were used to assess the level of occupational greening – three categories identified • Green new and emerging; green enhanced skills; green in demand • Circa 200 of 1,100 occupations assessed as green • Limitation - assumption tasks & occupations same in the UK and US • Binary and continuous (weighted) measure Introduction | Data | Methods | Results | Discussion
  • 53.
    Figure 2: Shareof green employment by gender and ethnicity (2018): binary measure (a) By gender (b) By ethnicity groups 32% 47% 18% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% Total Male Female 33% 31% 26% 24% 32% 0% 5% 10% 15% 20% 25% 30% 35% 40% White Mixed / multiple ethnic groups Asian / Asian British Black / African / Caribbean / Black British Other ethnic group Results
  • 54.
  • 55.
    Results: Regression Summary Results •Individuals are more likely to work in green occupations if they are white, male, full-time, not represented by a collective agreement, and work for an SME or foreign owned business • The study estimates a pay premium of between 10% (raw) and 4% (including all observable controls) • Personal travel behaviours and green employment choices are often inconsistent, but when they align this yields a pay dividend. • Green employment can partially mitigate inter-occupation pay gaps, however, the gaps persist within green occupations • Females appear particularly disadvantaged by domestic and childcare responsibilities. • There are also sector effects, with more traditional industries such as manufacturing and construction exhibiting entrenched gender biases Implications • The results highlight the need to integrate considerations of inequality into theoretical frameworks that aim to understand and conceptualise the uptake of green jobs. • There is an important role for policy to play if the green transition is to deliver social inclusivity alongside economic growth and job creation Introduction | Data | Methods | Results | Discussion
  • 56.
  • 57.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus The Disability Pay Gap Within and Across Firms John Forth and Melanie Jones ONS Earnings Symposium 29th April 2025 Acknowledgements: This work was undertaken in the Office for National Statistics Secure Research Service using data from ONS and other owners and does not imply the endorsement of the ONS or other data owners. We are grateful to the ONS, Administrative Data Research UK and the Wage and Employment Dynamics project for creating the ASHE-Census 2011 dataset. John Forth gratefully acknowledges funding from the Wage and Employment Dynamics project under ESRC grant number ES/T013877/1.
  • 58.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus • Disability Pay Gap (DPG) neglected relative to pay gaps for other protected characteristics • Little evidence on the role of the distribution of employees across firms to the DPG • This paper uses newly linked ASHE-Census data to separate the influence of the distribution of employees across firms from within-firm DPGs • Contemporary policy relevance given planned employer DPG reporting in the UK Motivation
  • 59.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG is the percentage difference in average hourly pay between disabled and non-disabled employees • Sizeable (10-15%) and shows no sign of diminishing • Decompose DPG • Personal and job-related characteristics • Less than 50% explained (Jones et al., 2006) • Debates about identifying discrimination (DeLeire, 2001; Longhi et al., 2012) • Role of the firm • Schur et al. (2009) role of corporate culture on gaps in-work outcomes in the US • Jones and Latreille (2010) variation in within-workplace DPG using WERS 2004 Existing Literature
  • 60.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus • Payroll information from ASHE 2011 • Multiple measures of pay (focus on total hourly earnings) • Personal characteristics from 2011 Census (74% of ASHE 2011 observations) • Disability (5.3%) • “Are your day-to-day activities limited because of a health problem or disability which has lasted, or is expected to last, at least 12 months? Include problems related to old age”. • Yes (limited a lot or limited a little). • Working-age (age 16-64) employees paid an adult rate, earnings unaffected by absence and basic weekly hours (1-99 hours). • Minimum of two employees within each firm • 78,037 jobs from 76,505 employees in 8,435 firms ASHE-Census
  • 61.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus Table 1: Descriptive statistics on earnings (£/hour), by disability Basic rate Total pay Non-disabled Mean 13.92 14.53 N 73,815 73,815 Disabled Mean 12.46 12.95 N 4,222 4,222 DPG (%) Mean -10.51 -10.88
  • 62.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG ( ) 𝜇 • Employee characteristics: sex and age (and age-squared), highest education, marital status, number of children, age of youngest child, ethnicity and UK born • Job-related characteristics: tenure, part-time, collective bargaining, second job and occupation (2010 SOC minor group) • Firm characteristics: firm size, industry sector (2007 SIC sections), sector and region • Within-firm DPG • Replace with firm identifiers • Heterogeneity e.g. gender, sector, firm size and over the wage distribution • Recentered influence function (RIF)-OLS earnings equations (Firpo and Pinto, 2016) • Extend established decomposition methods (Oaxaca, 1973; Blinder, 1973) Analysis
  • 63.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus Raw Raw: within-firm Adjusted: employee and job characteristics Adjusted: employee, job and firm characteristics Adjusted: employee and job characteristics within-firm (1) (2) (3) (4) (5) Total pay -0.095*** -0.068*** -0.040*** -0.036*** -0.032*** (0.008) (0.007) (0.005) (0.005) (0.005) N 78,037 78,037 78,037 78,037 78,037 Adj R-squared 0.002 0.377 0.650 0.685 0.737 Notes: OLS regression coefficients, estimated from ASHE-Census 2011. DPG calculated as the difference in log points between the log hourly wages of disabled employees and non-disabled employees. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1. Table 2: Raw and adjusted DPG, across and within firms
  • 64.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus Table 3: Decomposition of DPG Notes: Estimates based on OB decomposition methods, as set out in the text, applied to total hourly earnings from ASHE-Census 2011. Robust standard errors estimated via the delta method and reported in parentheses. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1. Components may not sum to raw DPG due to rounding errors.
  • 65.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus Raw Raw: within-firm Adjusted: employee and job characteristics Adjusted: employee, job and firm characteristics Adjusted: emp. and job chars. within-firm (1) (2) (3) (4) (5) Disabled -0.118*** -0.025 -0.027* -0.023 0.002 (0.026) (0.021) (0.016) (0.015) (0.014) 250-999 employees 0.060*** 0.032*** 0.031*** (0.008) (0.005) (0.005) 1,000-4,999 employees 0.081*** 0.059*** 0.052*** (0.007) (0.005) (0.004) 5,000+ employees -0.015** 0.044*** 0.045*** (0.007) (0.004) (0.004) Disabled x 250-999 employees 0.027 -0.032 -0.002 -0.002 -0.022 (0.034) (0.028) (0.022) (0.021) (0.019) Disabled x 1,000-4,999 employees 0.035 -0.048* -0.020 -0.021 -0.047*** (0.031) (0.026) (0.019) (0.018) (0.017) Disabled x 5,000+ employees 0.027 -0.048** -0.015 -0.016 -0.036** (0.028) (0.023) (0.017) (0.016) (0.015) Adj R-squared 0.008 0.377 0.651 0.685 0.737 N 78,037 78,037 78,037 78,037 78,037 Table 4: Heterogeneity in the DPG by firm size Notes: OLS regression coefficients for total hourly earnings, estimated from ASHE-Census 2011. Key to statistical significance: *** p<0.01; ** p<0.05; * p<0.1.
  • 66.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus Figure 1: Raw and adjusted DPG, across and within firms, across the wage distribution Notes: Recentered influence function (RIF)-OLS regression coefficients for total hourly earnings, estimated from ASHE-Census 2011. Error bars show 95% confidence intervals.
  • 67.
    The Public ValueBusiness School | Yr Ysgol Busnes Gwerth Cyhoeddus • DPG and unexplained DPG – Mainly exist within firms – Within-firm DPGs are reinforced by the allocation of disabled employees across firms – Failure to account for firm allocation results in disability-related wage inequality being overestimated by 12-20% • Role for employer focus of DPG reporting – Large firms have larger within-firm DPG, suggests effective targeting of legislation – Value in reporting workforce disability composition Conclusions
  • 68.
    Uncovering multiple employment withlinked ASHE-HMRC data Dr Darja Reuschke Birmingham Business School Prof Tracey Warren Nottingham Business School
  • 69.
    Multiple employment • Havingmore than one employment: • Dual/multiple job-holding • Hybrid entrepreneurship (employees with income from self-employment) • Existing studies • Based on survey data on ‘2nd job’ • Little differentiation between different forms of multiple employment • Little known about gender and multiple employment • Three hypotheses: • Financial constraints • Job insecurity • Non-monetary reasons (e.g. skills development)
  • 70.
    • Using newadministrative data sources to understand influencing factors on multiple employment • Differences between multiple job-holding vs mixing employee jobs with self- employment income • Gender differences • Research questions • Low wages? • Low working hours? • Regional labour market conditions? Objectives and research questions
  • 71.
    • Office forNational Statistics; His Majesty's Revenue and Customs, released 01 August 2024, ONS SRS Metadata Catalogue, dataset, Annual Survey of Hours and Earnings linked to PAYE and Self-Assessment data - GB, https://doi.org/10.57906/566k-5q15 • ASHE-GB: • Sample of 1% of employee jobs in GB • Main sample: PAYE jobs based on NI numbers • NI numbers remain in ASHE sample -> longitudinal element • Multiple (PAYE) jobs of the same NI number are sampled • Employer non-response • ASHE-HMRC Self-Assessment data: 2011-2017 • ASHE-HMRC Pay As You Earn (PAYE) data: 2014-2018 Data
  • 72.
    ASHE-SA 2011-2017 • Linkof self-employment income (profit) to ASHE individuals (by tax years) • Comparison of multiple jobs (in ASHE) vs mixing employee jobs & s/emp ASHE-PAYE 2014-2018 • ASHE survey reference period in April linked with April payslips • Identifying multiple ‘employment’ in PAYE data based on payslips with positive values & excluding pensions • Multiple employment of ASHE individuals based on April PAYE data > ASHE annual survey • Comparison of multiple jobs in annual ASHE vs multiple ‘employments’ in PAYE data • 21% of ASHE individuals aged 16-64 do not have payslips in PAYE data Linked longitudinal datasets
  • 73.
    • Transition fromone job/employment to multiple jobs/employment between subsequent years (t-1 to t) • Conditional fixed-effects logistic regressions by gender, 16-64-year-olds • Financial constraints of job: 1. Weekly earnings 2. Hourly wage & weekly basic hours 3. NMW groups: i) at or below NMW, ii) below 60% median, iii) 60% median + & weekly basic hours • Insecurity: • permanent/temporary job • Regional labour market (ITL1): • GDP per capita, male/female unemployment rate • Controls (incl. age, sector, year dummies, not shown below) Methods
  • 74.
    Findings from ASHE-SA: Multiplejobs in ASHE vs employee job with income from self-employment
  • 75.
  • 76.
    Employees with incomefrom self-employment
  • 77.
    Findings from ASHE-PAYE:Starting multiple jobs using ASHE annual survey vs real time PAYE data
  • 78.
    Multiple employments asemployee in ASHE vs PAYE
  • 79.
    • Multiple job-holdingand mixing employee jobs with self-employment are different forms of multiple employment • Low working hours and resultant low weekly earnings consistent factors influencing multiple job-holding of women and men • Findings on low wages are mixed • Effect of low wages amplified in PAYE data • Potential under-coverage of multiple job-holding among those with low wages in ASHE • No evidence of job insecurity hypothesis (temporary jobs) • Mixing employee jobs with self-employment income may be better approached as hybrid entrepreneurship than multiple job-holding Key policy conclusion: Need of jobs with sufficient hours Summary and conclusion
  • 80.
    Thank you! Darja Reuschke [email protected] Acknowledgement:This research is funded by an ADR UK fellowship (ES/Z503149/1) and conducted in collaboration with the UK WBG. I am grateful for the WED team, namely Damian Whittard, Prof Felix Ritchie, Dr Van Phan and Dr Carl Singleton, for their help. “This work was undertaken in the Office for National Statistics Secure Research Service using data from ONS and other owners and does not imply the endorsement of the ONS or other data owners.”
  • 81.
    The use ofONS earnings data in decision- making at the Low Pay Commission Presentation to the ONS earnings symposium Tim Butcher 29 April 2025
  • 82.
    Overview • Low PayCommission and the National Minimum and National Living Wage • Determining the path: Benchmark; Actual wage growth; Forecasts • Analysing pay trends: AWE, RTI and ASHE • Descriptive analysis of structure of pay: bite, coverage and distribution • Research: external, independent and in-house • Conclusion 82
  • 83.
    Our NLW recommendationmust… 83 “ensure that the rate does not drop below two-thirds of UK median earnings for workers aged 21 and over” Take account of: “the impact on business, competitiveness, the labour market, the wider economy” Take account of: “the cost of living, including the expected annual trends in inflation between now and March 2026”
  • 84.
    Our evidence-based approach 84 Externalresearch We make use of research produced by others including the IFS, Resolution Foundation and others Stakeholders We speak to employers, employees and their representatives, through visits and evidence sessions Internal analysis We undertake our own analysis of primary data sources, including econometric analysis International comparisons We make use of international data and research and convene events with our equivalents in other countries Consultation We invite written evidence submissions through an open, public consultation. Commissioned research Each spring we commission research from academics and other experts, who report their findings in the autumn.
  • 85.
    LPC uses variousONS earnings data sources • The focus here is on the top three (with ASHE the main one) • Annual Survey of Hours and Earnings (annual) • Average Weekly Earnings (monthly) • RTI earnings data (monthly) published by ONS • (Labour Force Survey wage data – quarterly) • (National Accounts data on wages and salaries, and compensation of employees) 85
  • 86.
  • 87.
    Estimating the pathof median earnings • Since 2016, the LPC remit has referred to a target of the percentage of median earnings for the appropriate age group • We use October as the mid-point of the NLW year (April-March) as the target point • Step 1: the median earnings benchmark – Median of hourly earnings excluding overtime for those covered by the NLW • Step 2: actual wage growth since the benchmark – Change in AWE total pay (using a smoothed 12-month on 12-month average) • Step 3: forecast wage growth (to following Octobers) since the last actual wage growth – Wage forecasts from HM Treasury panel of independent forecasts plus OBR and the Bank of England (median of forecasts made in last 3 months) 87
  • 88.
    Baseline median hourlypay (ASHE 2023 and 2024) Median hourly pay 21 and over 2023 provisional £15.98 2023 final £16.16 2024 provisional £17.19 2024 on 2023 wage growth 6.4% 2024 on provisional 2023 wage growth 7.6% 88
  • 89.
    Large revisions tothe baseline in 2023 had a knock-on to 2024 89 2021 Apr 2021 Oct 2022 Apr 2022 Oct 2023 Apr 2023 Oct 2024 Apr 14.00 14.50 15.00 15.50 16.00 16.50 17.00 17.50 14.26 14.92 16.16 17.19 14.28 14.9 15.98 Latest ASHE Month of projection Median of hourly earnings excluding overtime for those aged 21 and over (£) Source: LPC estimates using ONS data. Median of hourly earnings excluding overtime for those aged 21 and over from Annual Survey of Hours and Earnings, 2021-2024 provisional, 2021-2023 final.
  • 90.
    From the baseline,we use a 12-month smoothed version of AWE to proxy ASHE median wage growth (from baseline to latest) 90 Source: LPC estimates using ONS data. Average weekly earnings total pay (KAB9), March 2021-February 2025. Note: LPC uses a smoothed 12-month on 12-month estimate of pay growth. Feb 03 Feb 05 Feb 07 Feb 09 Feb 11 Feb 13 Feb 15 Feb 17 Feb 19 Feb 21 Feb 23 Feb 25 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 10.0 AWE total pay Smoothed LPC total pay Change on a year ago (%)
  • 91.
    But wage forecastshad been revised up since the spring 91 Source: Bank of England, HM Treasury and Office for Budget Responsibility. Forecasts of average wage growth (median of HMT panel of independent forecasts, August 2023-April 2025, BoE Monetary Policy Report August 2023-February 2025 and OBR March 2023-March 2025). Aug 2023 Oct 2023 Dec 2023 Feb 2024 Apr 2024 Jun 2024 Aug 2024 Oct 2024 Dec 2024 Feb 2025 Apr 2025 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 4.6 5.0 3.6 3.7 3.1 3.0 2.086919803629 32 2.027930825642 74 2024 HM Treasury panel 2024 Bank of England 2024 OBR 2025 HM Treasury panel 2025 Bank of England 2025 OBR Month of projection Annual wage growth (%)
  • 92.
    What does allthat mean? 92
  • 93.
    The projected medianin 2025 at the Retreat 2022 Apr 2022 Oct 2023 Apr 2023 Oct 2024 Apr 2024 Oct 2025 Apr 2025 Oct 14.50 15.00 15.50 16.00 16.50 17.00 17.50 18.00 18.50 19.00 14.92 16.16 17.19 17.63 17.99 18.32 Month of projection Projected median of hourly earnings excluding overtime for those aged 21 and over (£) Step 1: The Baseline ASHE Median Step 3: Wage projection 2 Wage forecasts (August 2024-October 2025) Step 2: Wage projection AWE (April 2024-August 2024) 93 Source: LPC, ONS, BoE, HMT and OBR. LPC calculations based on median of hourly earnings excluding overtime, ASHE, 2022-2024; AWE total pay (KAB9), 2021-2024, and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, BoE Monetary Policy Report Aug 2024 and OBR March 2024).
  • 94.
    Evolution of theprojected path of the NLW (2/3rds of median) 94 Path of the NLW for 2025 if kept at two-thirds of median 11.00 11.20 11.40 11.60 11.80 12.00 12.20 12.40 12.60 11.56 11.61 11.82 12.01 12.04 11.84 11.89 12.1 12.18 12.21 12.13 12.18 12.39 12.36 12.39 Lower range Central estimate Month of projection Projected NLW in 2025 (£) Source: LPC, ONS, BoE, HMT and OBR. LPC calculations based on median of hourly earnings excluding overtime, ASHE, 2022-2024; AWE total pay (KAB9), 2021-2024, and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, BoE Monetary Policy Report Aug 2024 and OBR March 2024).
  • 95.
    Earnings growth remainsabove forecast in Q4 95 Source: LPC calculations based on ONS, Bank of England, and HM Treasury data. AWE total pay (KAB9), 2021-2024, median of pay growth from HMRC Real Time Information (Table 27, October 2024), and forecasts of average wage growth (median of HMT panel of independent forecasts August and October 2024, and BoE Monetary Policy Report Aug 2024). 2017 Dec 2018 Dec 2019 Dec 2020 Dec 2021 Dec 2022 Dec 2023 Dec 2024 Dec 2025 Dec 2026 Dec -2 -1 0 1 2 3 4 5 6 7 8 9 10 AWE total pay (nominal) RTI median of pay growth HM Treasury panel forecast AWE (Aug/Oct 24) Bank of England private sector AWE LPC smoothed AWE Annual growth in average weekly pay (%)
  • 96.
  • 97.
    Wage growth continuesto be strong. RTI measures of pay growth remain above 5% (and AWE also similar) 97 Source: LPC calculations based on ONS. Growth in median pay (derived from Table 2); growth in average pay (derived from Table 3); and median of pay growth ((Table 27) from HMRC Real Time Information, seasonally adjusted, monthly, February 2017-February 2025. 2017 Feb 2018 Feb 2019 Feb 2020 Feb 2021 Feb 2022 Feb 2023 Feb 2024 Feb 2025 Feb -1 0 1 2 3 4 5 6 7 8 9 10 RTI median pay RTI mean pay RTI median of pay growth Change in pay on same month a year ago (%)
  • 98.
    Over the lastyear, we needed to be aware of the impact of public sector bonuses but these have now dropped out of the annual comparisons 98 -4 -2 0 2 4 6 8 10 12 14 Private sector Public sector excluding finance Change on a year ago (%) Total pay Nov 16Nov 18Nov 20Nov 22Nov 24 0 10 20 30 40 50 60 70 Public sector excluding finance Private sector Average weekly bonus pay (£) Bonus pay -2 0 2 4 6 8 10 Private sector Public sector excluding finance Change on a year ago (%) Regular pay Source: LPC estimates using ONS data. Average weekly earnings total pay private sector (KAC4), public sector excluding financial services (KAD9); Average weekly earnings bonus pay private sector (KAF7), public sector excluding financial services (KAH3); Average weekly earnings regular pay private sector (KAJ2), public sector excluding financial services (KAK6), seasonally adjusted, November 2016-November 2024.
  • 99.
    Real average weeklywages have picked up over the last two years but are still only back to their April 2021 levels 99 Apr 2021 Oct 2021 Apr 2022 Oct 2022 Apr 2023 Oct 2023 Apr 2024 Oct 2024 94 95 96 97 98 99 100 101 Real AWE total pay (CPI) Real AWE regular pay (CPI) Real AWE total pay (CPIH) Real AWE regular pay (CPIH) April 2021=100 Real average weekly pay index (April 2021=100) Source: LPC estimates using ONS data. Real average weekly earnings total pay (A3WX), real average earnings regular pay (A2FC), and Real average earnings using CPI (derived from X09), April 2020-February 2025.
  • 100.
    But still notmuch higher than in 2008 on these measures 100 Source: LPC estimates using ONS data. Real average weekly earnings total pay (A3WX), real average earnings regular pay (A2FC), and Real average earnings using CPI (derived from X09), February 2001-February 2025. Feb 2001 Feb 2004 Feb 2007 Feb 2010 Feb 2013 Feb 2016 Feb 2019 Feb 2022 Feb 2025 400 425 450 475 500 525 550 Real (CPIH) AWE total pay Real (CPIH) AWE regular pay Real (CPI) AWE total pay Real Average Weekly Earnings (£)
  • 101.
    Descriptive analysis ofstructure of pay 101
  • 102.
    The bite ofthe minimum wage is the highest it has ever been, but it missed the target set for 2024 102 The UK’s adult minimum wage as a percent of median hourly pay (bite), UK, 1999-2024 Apr-99 Apr-04 Apr-09 Apr-14 Apr-19 Apr-24 40 45 50 55 60 65 70 25+ 21+ Eligible population 2020 target 2024 target NLW Bite (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 1999-2024. Note: LPC projections beyond April 2024 using AWE data and HM Treasury and Bank of England forecasts.
  • 103.
    But, as atApril 2024, the bite was above the 2/3rds target for all regions except London, Scotland and the South East 103 Bite of the NLW by region, NLW eligible population 0 20 40 60 80 100 2019 2023 2024 NLW Bite (%) 2024 target Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2019, 2023 and 2024.
  • 104.
    Hourly wage growthwas strong across the distribution in 2024, but strongest for low-paid workers 104 Growth in hourly wage by percentiles, 2024, UK Real hourly wage growth by percentile, 2024, 21+ population 1 9 17 25 33 41 49 57 65 73 81 89 97 -5 0 5 10 15 23+ 21-22 Percentile Growth (%) 1 9 17 25 33 41 49 57 65 73 81 89 97 -15 -10 -5 0 5 10 15 20 Real growth 2024 Real growth since 2019 Percentile Growth (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2019 and 2024.
  • 105.
    Increases in hourlypay have fed through to weekly pay 105 1 (lowest paid) 2 3 4 5 6 7 8 9 10 (highest paid) 0 2 4 6 8 10 12 Hourly pay growth Weekly pay growth Hourly pay deciles Growth (%) Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2024.
  • 106.
    Despite the largeincreases in the NLW in recent years, coverage had not increased. However, coverage rose sharply in 2024 106 Number and share of jobs covered by the NMW/NLW, UK 2016-2024 Source: LPC analysis of ASHE, UK, low-pay weights. NLW coverage refers to workers aged 25 and over before 2021, and 23 and over from 2021 to 2023, and 21 and over for 2024, due to NLW eligibility change. 2016 2017 2018 2019 2020 2021 2022 2023 2024 0 600000 1200000 1800000 2400000 0 2 4 6 8 Covered NLW Covered Total Coverage rate NLW Coverage rate Total Persons covered (millions) Coverage rate (%)
  • 107.
    The number ofjobs paid within £1 of the NLW increased by 2 ppts, to be back at 2020 levels 107 Number of jobs by pay relative to the adult rate NMW/NLW, NLW eligible population Source: LPC analysis of ASHE, UK, low-pay weights. NLW eligible population refers to workers aged 25 and over before 2021, 23 and over from 2021 to 2023 and 21 and over for 2024, due to NLW eligibility change, excludes first year apprentices. 2016 2017 2018 2019 2020 2021 2022 2023 2024 0 2 4 6 8 10 12 14 16 18 20 Covered 6-50p above NMW/NLW 51p-£1 above NMW/NLW Share of employed popualtion (%)
  • 108.
    Using the averagewage growth of the 35-80th percentiles, we estimate that spillovers were responsible for up to a third of wage growth in 2024 108 Decomposition of hourly pay increase by percentile, UK, 21+ 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Counterfactual increase Increase required by NLW Spillover No spillover assumption Change in hourly pay (£) At the 5th percentile, people are paid £11.45 At the 13th percentile, people are paid £12.00 At the 24th percentile, people are paid £13.00 At the 35th percentile, people are paid £14.43 Source: LPC estimates using ONS data. Annual Survey of Hours and Earnings, 2024.
  • 109.
  • 110.
    Research • ONS earningsdata have been used to conduct econometric analyses of the impact of the NMW/MLW • Externally commissioned research and independent research – Dube (2019) used ASHE in his review of minimum wages for HM Treasury – IFS (2021) developed the bunching analysis to identify minimum wage effects – IFS (2024) looked at effects of the National Living Wage on firms' wage structures – Dickens, Manning and Butcher (2012) looked at spillovers • In-house research – Extensions of the bunching analysis (more recent time periods) – Development and extensions of the geographic variations approach – Developing analysis to look at young people (age discontinuities) 110
  • 111.
    Conclusions • ONS wagedata is essential for our analyses of the National Minimum Wage and the National Living Wage (NMW/NLW) • The data help us to estimate the median and determine any target • They provide information on pay trends • They enable us to explore the structure of pay – hourly and weekly pay distributions • We can analyse the impact of the NMW/NLW on bite and coverage • We can look at those most affected by the NMW/NLW – Low-paying occupations, low-paying industries, low-pay areas, micro and small firms and very large ones, young people and older workers, women and part-time. • Pandemic (and recent data shortcomings) has made analysis much harder 111
  • 112.
    Wage growth: insights fromUK firm-level microdata Josh Martin Bank of England, King’s College London, Economic Statistics Centre of Excellence (ESCoE) Any views expressed are solely those of the author and so cannot be taken to represent those of the Bank of England or to state Bank of England policy. This paper should therefore not be reported as representing the views of the Bank of England or members of the Monetary Policy Committee, Financial Policy Committee or Prudential Regulation Committee. This paper uses ONS statistical research datasets. Outputs may not exactly reproduce National Statistics aggregates.
  • 113.
    Data • Monthly Wagesand Salaries Survey (MWSS) • Covers all industries & sectors; only samples units with 20+ employment • Largest units = census; smaller units = stratified random sample (by industry, size, legal status). Total sample 6,000 per month; response rate ≈ >80%. • Units stay in sample for multiple years, or indefinitely – can track over time • Survey is short: total [monthly] paybill, bonuses, arrears, number of employees • Regular pay = total minus bonuses and arrears • Access MWSS microdata via SRS • Annual files and monthly files; some inconsistencies and naming issues • Not yet available consistent with recent exceptional revision to AWE • Filtering: regular pay per employee; exclude all imputations and ONS-flagged outliers; no account of small firm adjustment
  • 114.
    Construction of measures AWEAWE AWE AWE Oct 2018 Nov 2018 Dec 2018 Oct 2019 ? Survey responses: Regular pay and employment Aggregation: Total regular pay and total employment Calculation with aggregates: Average regular pay per … Firm-level growth rates Entry and exit from sample prevent growth rate calculation
  • 115.
    Notes on interpretation •All charts relate to “regular pay” (excluding bonuses and arrears) • Will not match published AWE data • Nothing is seasonally adjusted here • Growth rates are mostly month-on-same-month-a-year-ago (i.e. annual growth, e.g. Dec 2019 / Dec 2018) or less frequently month- on-month (e.g. Dec 2019 / Nov 2019) • All series stop in 2019 – no comment on recent wage dynamics and inflation, and consistent microdata not available yet • All data for “whole economy” • Private sector in hidden slides (similar); industry breakdowns possible (ongoing)
  • 116.
    Appropriately weighted MWSSdata matches AWE fairly closely 200120012002200320042005200620072008200920102011201220122013201420152016201720182019 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Whole economy regular pay, annual growth (%) AWE Firm-level data tend to be higher, likely due to: 1. No “small firm” effect 2. Only continuing firms MWSS mean “AWE-weighted”
  • 117.
    Alternative weightings tellslightly different stories, and give additional insights 200120012002200320042005200620072008200920102011201220122013201420152016201720182019 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Whole economy regular pay, annual growth (%) Mean “AWE-weighted” Mean employment-weighted Mean firm-weighted Firm-weighted typically higher aside from the downturn, suggesting more pro-cyclical growth in smaller and/or lower paying firms
  • 118.
    2001200220042006200820102012201420162018 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Employment-weighted 2001200220042006200820102012201420162018 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Firm-weighted Mean growth consistentlyabove median Median Mean Mean Median Whole economy regular pay, annual growth (%) Mean typically higher than median, suggesting a right- skewed distribution and inequality of wage growth
  • 119.
    Median employment-weighted growth fromMWSS correlates better with RTI than AWE 201507 201510 201601 201604 201607 201610 201701 201704 201707 201710 201801 201804 201807 201810 201901 201904 201907 201910 0.0 1.0 2.0 3.0 4.0 5.0 MWSS median employment-weighted AWE RTI median pay Correlation of annual growth AWE MWSS employment-weighted median RTI median AWE 1 0.86 0.76 MWSS employment-weighted median 0.86 1 0.88 RTI median 0.76 0.88 1
  • 120.
    The distribution ofannual growth in regular pay is wide, and moves around a lot <-10 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 >20 0 2 4 6 8 10 12 Density of annual growth in regular pay (%), whole economy, employment weighted 2001 2007 2019 2008-2018 2002-2006
  • 121.
    In most months,there is very little month-on-month pay growth… <-10 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 >20 0 2 4 6 8 10 12 14 16 18 20 Density of month-on-month growth in regular pay (%), whole economy, employment weighted, pooled 2001-2019 January April Every other month December except January and April
  • 122.
    Falls in averageregular pay per employee are common across the dataset Period averages, alternative weightings Month on month Month on same month a year ago Unweighted 2001-2010 47% 28% 2011-2019 48% 34% AWE weighted 2001-2010 46% 22% 2011-2019 47% 30% Employment weighted 2001-2010 47% 23% 2011-2019 48% 31% Firm weighted Unweighted average by month Month on month Month on same month a year ago January 51% 31% February 48% 30% March 47% 31% April 41% 31% May 48% 31% June 47% 31% July 49% 31% August 49% 31% September 47% 30% October 47% 30% November 48% 30% Proportion of firms reporting a fall in their average regular pay
  • 123.
    Applications and otherremarks • Were the elevated levels of pay growth in recent years driven by higher pay growth amongst all firms or only some firms? • Can decompose aggregate wage growth into within-firm and between effects • Can see wage growth across the firm-level distribution • Can help to reconcile different measures of wages and earnings • MWSS employment-weighted median more similar to RTI median pay • Wide distribution of changes in average wages across firms, and surprising frequency of wage falls – likely partly reflects compositional changes within firm (e.g. hiring and firing) • But also likely some measurement error
  • 124.
    Conclusions • Great scopefor insights from microdata analysis of MWSS even without data linkage • Opportunities to improve MWSS microdata and AWE • Initial exploration gives sensible results: • Sample of ‘continuing’ firms tends to have higher annual wage growth than AWE • Firm-weighted average wage growth is more pro-cyclical, reflecting greater weight on smaller firms • Mean growth higher than median, reflecting right-skewed distribution • January and April are key months for wage growth
  • 125.
  • 126.
    128 Tax data letsus dive deeper into workers’ earnings @resfoundation • The dataset: a 1% sample of employees (and all their payslips) in PAYE system, 2014 to 2019 • 250,000 employees, 27m payslips • Link to survey data (ASHE) in April for more info. about workers & employers • Benefits: big sample; measured at maximum frequency • Limitations: no self-employment or other income; no household info.
  • 127.
    129 Mean of absolutearc percentage annual change in real weekly earnings among 20-59-year-olds: UK Notes: Latest data points are 2023 (LFS), 2022 (BHPS/UKHLS), 2020 (ASHE/NESPD). Earnings are adjusted for CPI inflation. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of ISER, British Household Panel Survey; ISER, UK Household Longitudinal Study; ONS, Five-Quarter Longitudinal Labour Force Survey; ONS, Annual Survey of Hours and Earnings / New Earnings Survey Panel. @resfoundation Long-term: average volatility at annual frequency flat/falling
  • 128.
    130 Mean of absolutearc percentage annual change in real weekly earnings among 20-59-year-olds: UK Drivers of lower ‘labour market’ volatility since 1990s: • Lower rates of entry to and exit from work • Lower use of temporary contracts Notes: Latest data points are 2023 (LFS), 2022 (BHPS/UKHLS), 2020 (ASHE/NESPD). Earnings are adjusted for CPI inflation. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of ISER, British Household Panel Survey; ISER, UK Household Longitudinal Study; ONS, Five-Quarter Longitudinal Labour Force Survey; ONS, Annual Survey of Hours and Earnings / New Earnings Survey Panel. @resfoundation Long-term: average volatility at annual frequency flat/falling
  • 129.
    131 Distribution of thearc percentage change in real monthly earnings compared to the previous month among 20-59-year-olds working in both months: UK, 2014-15 to 2018-19 Notes: Earnings are deflated using CPIH. Results are pooled across all the months in dataset. For this figure, counts are based on arc-percentage change rounded to the nearest percentage point, which means ‘zero change’ in fact relates to arc-percentage changes between -0.5% and 0.5%, and ‘1-10% change’ in fact means changes between 0.5% and 10.5%, and equivalently for the other categories. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. @resfoundation Turning to the payslip data: earnings are same as previous month only in 4-in-10 months The average absolute change in earnings between months (among people working both months) was 15%. That’s similar in magnitude to what average households spend on food & clothing.
  • 130.
    132 Stylised examples ofearnings to illustrate ‘trajectory’ categories Definitions of trajectory categories: Extremely stable: all months within 5% of annual average Highly stable: all months within 10% of annual average Small blip: 10-25% from annual average Large blip: 25%+ from annual average Erratic: 4+ months where pay 25%+ from average These are artificial series drawn for illustration and do not represent actual data. @resfoundation Looking at separate months misses deep volatility faced by some: multiple large changes within year
  • 131.
    133 Proportion of employeesaged 20-59 and working all months in the year, by category of within-year earnings trajectory: UK, 2014-15 to 2018-19 Notes: Results are pooled across financial years. Analysis is based on real earnings, deflated using CPIH. Source: Analysis of HMRC PAYE dataset. @resfoundation Looking at separate months misses deep volatility faced by some: multiple large changes within year
  • 132.
    134 Proportion of employeesaged 20-59 and working all months in the year, by category of within-year earnings trajectory: UK, 2014-15 to 2018-19 Notes: Results are pooled across financial years. ‘Stable’ here includes the ‘Extremely stable’ and ‘highly stable’ categories as defined earlier. ‘Blips’ includes categories involving 1-2 small or large blips. Analysis is based on real earnings, deflated using CPIH. Source: Analysis of HMRC PAYE dataset and HMRC-ASHE PAYE dataset. @resfoundation ‘Erratic’ pay is most common among workers who are young, low-paid, in temporary jobs, working multiple jobs
  • 133.
    135 Proportion of employeeson a zero-hours contract (2018-2023), and average arc percentage change in real monthly earnings compared to the previous month among 20-59-year-olds (working both months): UK Notes: Industries not labelled are Manufacturing, Real estate, Professional services, Vehicle sales & repair, ICT, and 'Other' services. Size of bubble indicates the total employees in industry. Earnings are deflated using CPIH. Earnings volatility data is pooled across all months in dataset and zero-hours contract data is pooled across all LFS quarters where the variables are available across 2018-2023. Source: Analysis of HMRC-ASHE PAYE dataset; ONS, Labour Force Survey @resfoundation Volatile pay is associated with use of zero-hours contracts
  • 134.
    136 Average arc percentagechange in real monthly earnings, finance and insurance versus the rest of economy (workers working in both months): UK, 2014-15 to 2018- 19 Notes: Earnings are deflated using CPIH. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of HMRC-ASHE PAYE data. @resfoundation Not all volatility is bad: earnings volatility in some sectors driven by big bonuses
  • 135.
    137 Average arc percentagechange in real monthly earnings, finance and insurance versus the rest of economy (workers working in both months): UK, 2014-15 to 2018- 19 Every March, bonuses comprise more than half (55 per cent) of average earnings in Finance & insurance, versus 8 per cent for other workers Notes: Earnings are deflated using CPIH. Arc-percentage change uses average of both periods as denominator – similar to normal percentage change for low values. Source: Analysis of HMRC-ASHE PAYE data. @resfoundation Not all volatility is bad: earnings volatility in some sectors driven by big bonuses
  • 136.
    138 Anxiety relating to‘unexpected changes to my hours of work’, by hourly pay quintile: UK, 2017 Recent qualitative research from NEST Insight found financial volatility can have: Financial costs: fees for missed bills; unable to benefit from cost-effective financial products which require lump sum payments. Psychological costs: stress of managing unpredictable incomes, constant need to monitor and adjust spending. Notes: Includes workers aged 20 to 65 only. Source: Analysis of UK Skills and Employment Survey. @resfoundation Unstable pay can have negative impacts
  • 137.
    139 What can Governmentand employers do? @resfoundation Government: • ZHC reforms will help workers experiencing high volatility • Strengthen statutory sick pay, don’t just extend coverage • Universal Credit may amplify volatility, but solutions hard • Help low-paid workers build financial buffers Employers: • Pay staff at the frequency that most suits them
  • 138.
    Unpacking Spatial Earnings Inequalityin the UK: The role of People, Places and Industry Effects Richmond Egyei – Resolution Foundation and ESCoE [email protected] [email protected]
  • 139.
    Acknowledgements/Disclaimer Acknowledgments: This workwas funded by the Ministry of Housing, Communities and Local Government, through a grant to the Economic and Social Research Council, as part of a research programme initiated by the former Levelling Up Advisory Council, [Grant Number ES/Z000130/1]. Disclaimer: This is independent research and does not represent government policy. The statistical data used here are from the DfE, the ONS, HMRC, and HESA, and are Crown copyright and reproduced with the permission of the controller. The use of the data in this work does not imply the endorsement of the data providers in relation to the interpretation or analysis of the statistical data.
  • 140.
    Overview T2 T3 T4 T1 Getdata Perform Analysis Draft paper Submit Enrico Vanino – University of Sheffield Tasos Kitsos & Dalila Ribaudo – Aston University Richmond Egyei & Gregory Thwaites & Emily Fry – The Resolution Foundation
  • 141.
    Overview What do wedo? Leverage employer-employee data to find the drivers of spatial earning disparities in the UK during 2013-2020 How do we do it? AKM regressions at the individual level + variance decomposition analysis (Overman & Xu, 2022; Card et al., 2025) What do we find? place effects stronger than previously and portable; within industry variation explains most of area effects (rather than across areas industry composition); industry specific education pay premia largely drive place-based education wage premia Why is it important? Addressing spatial economic imbalances; policy choices (i.e. moving functions rather than moving industries)
  • 142.
    Introduction Large and persistentspatial disparities both across and within UK regions in terms of income and economic opportunities (ONS, 2018; McCann, 2020).
  • 143.
  • 144.
    Introduction Large and persistentspatial disparities both across and within UK regions in terms of income and economic opportunities (ONS, 2018; McCann, 2020). Sub-national spatial disparities are substantial also among other OECD countries, with income, productivity, skills, and knowledge increasingly polarised in few large urban areas (Behrens and Robert-Nicoud, 2014; Eeckhout et al., 2014; Autor, 2019; Davis and Dingel, 2019). Reducing these regional disparities could address part of the country’s economic growth problems, ensuring more opportunities and prosperity for people across the country (ONS, 2021). Still limited understanding and robust evidence on size, nature and drivers of these regional inequalities.
  • 145.
    Aims  Better understandwhat drives differences in local earnings, focusing on interactions in the distribution of three factors: workers, firms, and area characteristics across commuting zones (TTWAs).  Micro-level analysis of how much spatial imbalances in earnings across and within places could be explained by: • Workers’ education, skills, and location choices. • Local industrial structure, size, and productivity of firms. • Physical capital of places (infrastructure, public services, housing and amenities).
  • 146.
    Literature  Distribution ofskills (Diamond and Gaubert, 2022), and presence of urban wage premia (Moretti, 2010; Overman and Xu, 2022).  Local industrial structure (Mealy and Coyle, 2022; Card et al., 2025), firms’ size, tasks (Koster and Ozgen, 2021; Duranton and Puga, 2005) and productivity (Hart et al., 2020).  Spatial inequalities in availability of public infrastructure (Gibbons et al., 2024), housing (Hilber and Vermeulen, 2016; Cheshire, 2019) and amenities (Diamond, 2016).  Variance decomposition analysis to explain earnings imbalances across and within places (Gibbons et al., 2014; Overman and Xu, 2022; Card et al., 2025).
  • 147.
    Data  Longitudinal EducationOutcomes (LEO) links educational pathways and (early) labour market outcomes for 38 million individuals: • National Pupil Database school and early years data; • Higher Education Statistics Agency data; • Individual Learner Record data for colleges and apprenticeships; • HMRC and DWP data on earnings, benefits and tax data; • Inter-departmental Business Register on population of businesses.  Data cleaning and preparation resulting in 4 million individuals: • Workers living in England; • Between 22 and 36 years of age; • In full-time employment; • Observed from 2013 to 2020.
  • 148.
    Analysis Figure 1: Median,bottom 10, and top 90 yearly earnings distribution across TTWAs in England in 2020.
  • 149.
    Analysis Figure 2: Sharetop 20th percentile yearly earnings distribution across TTWAs in England in 2020.
  • 150.
    Design 1. Estimate componentsof individuals’ earnings: workers’ observed and unobserved factors, employers’ (area) effect, residuals. a) Are the TTWA wage premia constant across space? 2. Calculate share of spatial variance in earnings driven by workers’ skills, industry composition, and other area characteristics. b) To what extent are TTWA wage premia due to composition adjusted local wage premia (i.e. within industry spatial pay variation), industrial composition (i.e. more industries that pay better) or a combination of the two?
  • 151.
    Methods AKM 2WFE earningsregression to estimate contribution of employees’ characteristics (observed and unobserved) and employers’ characteristics: (1) • : individual log annual earnings; • : individual i fixed-effects; • : employer f fixed-effects; • : individual i time variant characteristics (age, experience, tenure, etc.); • : year t fixed-effects.
  • 152.
    (1) (2) (3)(4) (5) Age 0.0261*** 0.0265*** 0.026*** 0.0264*** 0.0263*** (0.001) (0.0007) (0.0007) (0.0007) (0.0006) Experience 0.2752*** 0.2344*** 0.2340*** 0.2334*** 0.2354*** (0.028) (0.007) (0.007) (0.007) (0.007) Tenure 0.0405*** 0.0441*** 0.0471*** 0.0449*** 0.0456*** (0.0006) (0.001) (0.0008) (0.001) (0.0015) Exp.London 0.1157*** 0.1224*** 0.1142*** 0.1150*** (0.005) (0.005) (0.006) (0.0047) Exp. Major City 0.0288*** 0.0473*** 0.0290*** 0.0549*** (0.005) (0.004) (0.0055) (0.0036) Current Urban Large City London 2nd Tier City CU x Exp. 0.0046 0.0165*** -0.0139* (0.006) (0.003) (0.0056) CU x Tenure -0.0048* -0.0008 -0.005 (0.002) (0.002) (0.0034) CU x Exp. London -0.0099** -0.0012 -0.0127*** (0.002) (0.001) (0.0019) CU x Exp. Major City -0.0274*** -0.0225*** -0.0022 (0.0015) (0.002) (0.0014) Employee FE Y Y Y Y Y Employer FE Y Y Y Y Y Year FE Y Y Y Y Y No. Obs. 26,810,593 26,810,593 26,810,593 26,810,593 26,810,593 R-sq. 0.3192 0.3326 0.3371 0.3370 0.3371
  • 153.
    Methods Locational wage premiumassociated with TTWA c as weighted average of employer effects in that TTWA: TTWA-by-Industry (SIC4) average wage premia:
  • 154.
    Analysis  Use AKMmodel in equation 1 to decompose the variance in mean wage across TTWAs into the different components:
  • 155.
    (1) (2) (3)(4) (5) Age 0.0261*** 0.0265*** 0.026*** 0.0264*** 0.0263*** (0.001) (0.0007) (0.0007) (0.0007) (0.0006) Experience 0.2752*** 0.2344*** 0.2340*** 0.2334*** 0.2354*** (0.028) (0.007) (0.007) (0.007) (0.007) Tenure 0.0405*** 0.0441*** 0.0471*** 0.0449*** 0.0456*** (0.0006) (0.001) (0.0008) (0.001) (0.0015) Exp.London 0.1157*** 0.1224*** 0.1142*** 0.1150*** (0.005) (0.005) (0.006) (0.0047) Exp. Major City 0.0288*** 0.0473*** 0.0290*** 0.0549*** (0.005) (0.004) (0.0055) (0.0036) Current Urban Large City London 2nd Tier City CU x Exp. 0.0046 0.0165*** -0.0139* (0.006) (0.003) (0.0056) CU x Tenure -0.0048* -0.0008 -0.005 (0.002) (0.002) (0.0034) CU x Exp. London -0.0099** -0.0012 -0.0127*** (0.002) (0.001) (0.0019) CU x Exp. Major City -0.0274*** -0.0225*** -0.0022 (0.0015) (0.002) (0.0014) Employee FE Y Y Y Y Y Employer FE Y Y Y Y Y Year FE Y Y Y Y Y No. Obs. 26,810,593 26,810,593 26,810,593 26,810,593 26,810,593 R-sq. 0.3192 0.3326 0.3371 0.3370 0.3371
  • 156.
    Analysis (0) (1) (2)(3) Variance (Earnings) 0.010 0.010 0.010 0.010 Area Effect 0.34 0.373 0.316 0.328 Individual Fixed Effect 0.236 0.227 0.238 0.243 Individual Observables 0.000 0.009 0.017 0.154 Correlations (2 x covariance) Area effect, Individual FE 0.404 0.422 0.254 0.227 Area effect, Individual observables 0.000 -0.014 0.184 0.205 Individual FE, Individual observables 0.000 -0.003 -0.152 -0.147 Fixed Effects Y Y Y Y Employee Experience N Y Y Y Lab market experience N N Y Y Labour market x Urban N N N Y
  • 157.
    Analysis Figure 3: Quintiledistribution of Locational Wage Premium across TTWAs in England.
  • 158.
    Methods Similarly, we calculatethe variance in the area wage premia explained by the 3 industry components. In this way, we can decompose the TTWA wage premium in 3 industry components:
  • 159.
    Analysis Area Effect (S.D)0.037 Decomposition (variance share) Industry Earnings Premium 0.821 Industry Composition 0.005 Interaction Effect 0.016 cov(earnings premium, composition) 0.005 cov(earnings premium, interaction) 0. 136 cov(composition, interaction) 0.005 Table 3: TTWA Area Wage Premium variance decomposition.
  • 160.
    Analysis Fig 4. Quintiledistribution of TTWA wage premium components in England.
  • 161.
    Analysis We repeat ourmain analysis focussing on returns to education across places To do this, we split our sample into college graduates and non- graduates and estimate the TTWA premium
  • 162.
    Analysis Then we estimatethe local wage premium by education group and decompose the difference in wages between our two groups
  • 163.
    Analysis Earnings differences (S.D)0. 308 Decomposition (variance share) Average TTWA education premium 0.172 Person effects between high and low education 0.59 Person time-varying effects 0.0655 cov(edu premium, person effects) 0.544 cov(edu premium, person time-varying) -0.123 cov(person effects, person time-varying) -0.249 Table : College Wage Premium variance decomposition.
  • 164.
    Analysis And decompose TTWAeducation wage premia
  • 165.
    Analysis Area Effect (S.D)0.012 Decomposition (variance share) Education premium in TTWA 0.128 Industry Composition/more graduates in high paying industries 0.551 Sorting – clustering of workers in high or low paying industries 0.008 cov(edu premium, composition) 0.354 cov(edu premium, sorting) -0.032 cov(composition, sorting) -0.009 Table: TTWA Area Wage Premium variance decomposition.
  • 166.
    Analysis Fig 6. Distributionof education wage premium and its decomposition
  • 167.
    Conclusions Larger place effectsare important for policymaking Within industry variations driving place effects point to different functions within industries across space. • College analysis corroborates this. Large urban experience follows workers. Next Steps: Real wages; what role for amenities/urban infrastructure; firm size & labour market characteristics
  • 168.
  • 169.
    Real Time Information (RTI)–The Impact of Health Interventions on Employment and Earnings Daniel Ayoubkhani, Emma Sharland, Charlotte Bermingham, Klaudia Rzepnicka, Isobel Ward, Hannah Bunk, & Vahé Nafilyan Health Research Group, Health and International Directorate Office for National Statistics
  • 170.
    Background • The GetBritain Working White Paper identifies that reversing the increase in economic inactivity due to ill-health is a national priority • The Government have committed to expanding services such as NHS Talking Therapies to reduce mental ill-health and improve employment outcomes • Obesity is also a key priority for prevention for the government as it is a key risk factor for leading conditions that may be driving health-related economic inactivity • Kickstarting Economic Growth is also one of the new Labour Government's missions and this includes supporting people who are economically inactive due to ill-health back to work • Policy interest (from HMT, DHSC, NHSE) in which conditions are driving inactivity, and which interventions are most effective at increasing employment
  • 171.
    Linked datasets heldby ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2011 Census 2021 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures
  • 172.
    Linked datasets heldby ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2011 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures 97% linkage rate via Demographic Index (via 2011-2013 NHS Patient Registers and direct matching) 95% linkage rate via 2011-2013 NHS Patient Registers Estimated to cover 94% of the population of England & Wales
  • 173.
    Linked datasets heldby ONS for work and health analysis NHS hospital records (Hospital Episode Statistics) ONS birth and death registrations NHS Taking Therapies records Tax records for employees (HMRC PAYE records) NHS Diabetes Prevention Programme records NHS general practice records (For COVID-19 research only) Labour market Population spine and socio- demographic characteristics Health 2021 Census Social security benefits (DWP) NHS maternity services records NHS elective care waiting list data Tax records for the self- employed (HMRC self-assessment) • Employment status • Monthly earnings and tax paid • Monthly benefits received • Age, sex, ethnicity, region • Education, employment, deprivation • Household size and composition • Date and cause of death • Appointment / admission / referral dates • Diagnoses and procedures 97% linkage rate via Demographic Index 96% linkage rate via NHS Personal Demographics Service Estimated to cover 97% of the population of England & Wales
  • 174.
    Bariatric Surgery Aim: Toestimate the average change in employee, pay and probability of paid employment attributable to bariatric surgery. This project was funded by the UK government’s Labour Markets Evaluation and Pilots Fund (2024 to 2025).​
  • 175.
    Study population: • 40,662individuals with a record of bariatric surgery in HES between 1 April 2014 and 31 December 2022, with an obesity diagnosis, and no prior record of bariatric surgery since 1 April 2009. • Individuals were resident in England and aged 25 to 64 on the date of bariatric surgery. • An unexposed sample of 49,921 individuals from the general population, age-sex stratified to match the exposed sample, who had not had bariatric surgery was also included. Methods: • We estimate the effect of having had bariatric surgery, at different time points after surgery, using fixed effects regression modelling. • We include fixed effects for individuals to account for confounding due to differences between individuals. • We include calendar time to account for changes in background economic conditions and age to account for changes in pay and likelihood of working over the working life. Bariatric surgery: Methods Data We use the 2011 and 2021 Censuses linked to: •Hospital Episode Statistics (HES) Admitted Patient Care (APC) records •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC)
  • 176.
    Sustained increase in probabilityof employment from four months after surgery Little increase in pay among those in work from six months after surgery Initial decrease indicative of surgery recovery time Sustained increase in pay from six months after surgery Bariatric surgery: Key findings
  • 177.
    NHS Talking Therapies Aim:To estimate the average change in employee, pay and probability of paid employment attributable to completing NHS Talking Therapies This project was funded by the Cabinet Office and HM Treasury Evaluation Task Force's Evaluation Accelerator Fund (2023 to 2024) and by the UK government's Labour Markets Evaluation and Pilots Fund (2024 to 2025).
  • 178.
    Study population: • 842,127individuals with; • a referral to NHSTT between 1 April 2016 and 31 March 2020 • attended at least one therapy session, • were at clinical caseness, • aged 25 to 60 years old at the time of referral • resident in England • Completed treatment group: 593,300 (70.5%) • Dropped out of treatment group: 248,827 (29.5%) Methods: • We estimate the effect of completing NHSTT, using fixed effects regression modelling. • We include fixed effects for individuals to account for confounding due to differences between individuals. • We include calendar time to account for changes in background economic conditions and age to account for changes in pay and likelihood of working over the working life. • Inverse Probability Weights (IPW) are used to balance the treatment and control groups NHSTT: Methods Data We use the 2011 Census linked to: •NHS Talking Therapies data (formerly IAPT) •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC)
  • 179.
    NHSTT: Key findings Wesaw a maximum average increase in pay of £17 per month by year two post first therapy (time zero) For employment, we see sustained increases in employment from year 1, peaking at 1.5 p.p increase at year seven.
  • 180.
    The “not working,seeking work” group have the biggest increases in the probability of being a paid employee following completion of treatment in NHSTT. This peaks at around 3.1 p.p. in year 4 and 5.
  • 181.
    Endometriosis Aim: What isthe impact of having a hospital diagnosis of endometriosis on women's labour market participation and progression? This project was funded by the UK government’s Labour Markets Evaluation and Pilots Fund (2024 to 2025).​
  • 182.
    Study population: 55,290 womenwith a primary diagnosis of endometriosis in the Hospital Episode Statistics (HES) Admitted Patient Care (APC) data between 1 April 2016 and 31 December 2022 who were:  Enumerated in the 2011 Census  Linked to an NHS number and National Insurance number  Resident in England, as recorded in HES if available and the 2011 Census if not  Aged 25 to 54 years on the date of endometriosis diagnosis  Recorded as being female in both HES and the 2011 Census  Had no primary or secondary diagnosis of endometriosis in HES between 1 April 2009 and 31 March 2016 Method:  Fixed effects (within-person) regression modelling  We adjust for age, calendar time and births  Monthly pay was deflated to 2023 prices  Being a paid employee was defined as receiving a monthly pay greater than £0 Data We use the 2011 Census linked to: •Hospital Episode Statistics •Office for National Statistics (ONS) death registrations •Pay As You Earn (PAYE) records from His Majesty’s Revenue and Customs (HMRC) Endometriosis: Methods
  • 183.
    Endometriosis: Key findingsThere was a sustained average decrease in pay. Monthly average pay among all women (both in paid work and not) reached £130 lower than pre-diagnosis levels four to five years after diagnosis. Monthly average pay among those in paid work reached £56 lower than pre-diagnosis levels four to five years after diagnosis. Monthly average pay among those in paid work reached £56 lower than pre-diagnosis levels four to five years after diagnosis.
  • 184.
  • 185.
    Work programme for2024/25 NHS Talking Therapies (Publication: 9 Dec 2024) Bariatric surgery (Publication: 23 Oct 2024) NHS Diabetes Prevention Programme (Publication: Jul 2025) Endometriosis (Publication: 5 Feb 2026) MSK surgery (Publication: 3 March 2025) Adverse pregnancy events (Publication: Jun 2025) Major conditions (Publication: Jun 2025) Published: Scheduled for 2025: Infectious diseases (Publication: TBC 2025) NHS Talking Therapies (Benefits and Healthcare Utilisation) (Publication: TBC 2025) Long-term ill health and benefits receipt (Publication: 26 Feb 2026) Impact of waiting times (Publication: TBC 2025)

Editor's Notes

  • #15 It is timely: - for a business survey of its size and characteristics, it’s our quickest. It is large: It allows for detailed analysis & to look at differences in levels and composition of earnings between groups, including earnings inequality.
  • #16 Mention key variables/themes.
  • #19 Data collection, ingestion, processing & validation: Reference date (surveys sent): April Key requirement: ensure new NMW/NLW are captured. Close date: end of August Microdata ready: mid-September Statistics & bulletins published: end of October
  • #27 This is iterative (ongoing means interative here).
  • #45 This information does exist but necessarily in one place, or in the form we would like WED project ambition is pull together these datasets – and not just acquire but clean and document to make a strategic research resource for the UK academic community
  • #46 Do we need this slide?
  • #83 Our remit has split Initially had the same approach applied to all rates The NLW meant two remits And the last one kind of meant three remit
  • #84 And this is how we meet that remit
  • #172 We have received funding in ONS to complete several evaluations of health conditions and interventions on labour market outcomes using administrative data.
  • #176 Bariatric surgery is an effective intervention to improve the health of people living with obesity and help prevent the development of obesity associated health conditions; the impact of this intervention on labour market outcomes is not well understood.
  • #178 Increases in pay has a sustained increases from 6-12 months after surgery to 5 years after Pay not increasing in those in paid employment at the time of surgery indicating the increases are being driven by people entering the workforce post-surgery. Initial dips are likely to be due to drops in pay whilst recovering from surgery. Following surgery, we see an increase in the probability of being a paid employee by just over 4 p.p. by year 5. The dip in this and the earnings in the first few months is likely to be due to recovering from surgery before returning to/starting work.
  • #179 NHS Talking Therapies is a service provided by the NHS for common mental health disorders such as anxiety and depression. One of the original aims of NHSTT when it was set up was an economic benefit, which prior to this analysis had not been evaluated at a national level. Smaller studies had seen a positive effect in the short term on employment following treatment in NHSTT.
  • #181 We continue to see significant increases until year 7 when the CI suggest a drop in statistical power.
  • #182 In this analysis we also stratified the results by different sociodemographic and treatment characteristics. All of which are available in the dataset. Here we show the breakdowns by employment status at the time of referral. This variable was taken from the NHSTT dataset. You can see here, that the “not working, seeking work” group have the biggest increases in the probability of being a paid employee following completion of treatment in NHSTT. This peaks at around 3.1 p.p. in year 4 and 5. We also see increases in the employed group, and then also in the not working, not seeking work group (economically inactive) although the increases are not significant by year 6 and 7. Further work is required to understand the not working, sick or disabled group as we see little change following the completion of NHSTT. It is possible, that these people have multiple conditions limiting their abilities to get back to work. We need to understand other comorbidities being experienced by people entering treatment in NHSTT to understand the reasons why this group may not be re-entering the workforce following treatment for their mental health.
  • #183 Economic inactivity due to poor health remains elevated compared to pre-pandemic levels. ​ Endometriosis can be a debilitating chronic condition with profound socioeconomic ramifications. This is the first population-level study to evaluate the impact of endometriosis on labour market outcomes in England.
  • #185 Increases in pay has a sustained increases from 6-12 months after surgery to 5 years after Pay not increasing in those in paid employment at the time of surgery indicating the increases are being driven by people entering the workforce post-surgery. Initial dips are likely to be due to drops in pay whilst recovering from surgery. Following surgery, we see an increase in the probability of being a paid employee by just over 4 p.p. by year 5. The dip in this and the earnings in the first few months is likely to be due to recovering from surgery before returning to/starting work. After being diagnosed with endometriosis, women earn less on average and are less likely to be in paid work, compared with the two years before their diagnosis. This suggests that, following diagnosis, women may be reducing their work hours, moving to lower-paid jobs, becoming self-employed, or receiving benefits. Further research is needed to understand the reasons behind these changes in pay and employee status post-diagnosis.    ONS are also working with academic partners to further explore the time between symptom presentation and diagnosis in women with endometriosis using primary care data linked to hospital data.
  • #187 Major conditions: CVD, respiratory, cancer, chronic kidney disease and MSK