Official Statistics

Workplace pension participation and savings trends statistics: Background information and methodology

Published 31 July 2025

Context of the statistics

Automatic Enrolment (AE) was introduced in 2012 to help address the decline in private pension saving and to make long-term saving the norm. It aims to increase workplace pension saving in the UK and forms part of a wider set of pension reforms designed to enable individuals to achieve financial security in retirement.

Eligibility of AE

AE mandates employers to provide a workplace pension for all workers meeting all of the following criteria:

  • earning more than £10,000 a year
  • are aged between 22 years and State Pension age (SPa) and are not already enrolled in a qualifying workplace pension

Note that throughout this report eligible employees are defined as employees who meet the AE age and earnings criteria each year, this includes employees already a member of a workplace pension scheme when AE was introduced.

Rollout of AE

AE commenced in 2012 and full implementation was completed in 2019. AE duties were brought in gradually over this time for employers, based on their size, and there was a phased increase of minimum contribution rates.

Staged implementation

The AE duties were staged in between October 2012 and February 2018 by employer size, starting in October 2012 with the largest employers based on PAYE scheme size, to the smallest in 2017. New PAYE schemes between April 2012 and September 2017 were staged last, between October 2017 and February 2018. New PAYE schemes from October 2017 have immediate AE duties.

Phasing

AE regulations require minimum contribution rates from the employer and in total (i.e. employer’s contributions, worker’s contributions and tax relief). Before 5 April 2018 these minimums were 2% of the worker’s qualifying earnings of which the employer must contribute at least 1% of earnings. Between 6 April 2018 and 5th April 2019, this rose to 5% total including at least 2% from employers, and on 6 April 2019 these rates rose again to a total of 8% including at least 3% from employers. The employee usually receives income tax relief on their contribution, usually at their marginal tax rate.

Minimum contributions based on qualifying earnings

Contribution (%)
Employer 3%
Employee 4%
Tax relief 1%
Total 8%

Purpose of the statistics

These are annual official statistics, first published in 2014.

The publication’s original aim was to complement the Automatic Enrolment Evaluation Reports by providing more detailed breakdowns of 2 key measures for evaluating the progress of AE implementation:

  • increasing the number of savers, by monitoring trends in workplace pension participation
  • increasing the amount of savings, by monitoring trends in workplace pension saving

Over time additional statistics have been added to the publication, enabling greater insight into pension saving trends. The latest publication provides annual statistics from 2009 on:

  • Workplace pension participation, with breakdowns by:
    • AE eligibility status
    • sector (private and public)
    • employer size
    • earnings
    • age
    • gender
    • working pattern
    • industry
    • occupation
    • region
    • parliamentary constituency
    • economic status (including eligible, non-eligible and self-employed)
    • disability
    • ethnicity
  • amount of savings and saving rates (with separate breakdowns for all employees and eligible employees):
    • total annual savings by sector, broken down by employer contributions, employee contributions and tax relief
    • median annual amounts saved by individuals, broken down by gender
    • median contribution rates for individuals, broken down by gender

Alongside this, timelier monitoring statistics are provided on:

  • quarterly opt-outs and stopping saving rates from 2014 to 2015 Q1
  • indexes of contribution amounts and active members from January 2020 based on data from a selected number of pension providers
  • monthly stopping saving rates from January 2020 based on data from a selected number of pension providers

Automatic Enrolment has increased the number of pension savers over the last 12 years. As a result, there is an increase in the number of people accessing a pension. Therefore, the publication includes additional details on the number and type of pensions being accessed.

Pension payments by product type, age bands and gender:

  • total payments
  • first time payments
  • payments in the same financial year that employment income was received

These statistics are available on an annual basis both as a statistical first release, a document that provides a summary of key statistics and findings, and data tables, which provide full breakdowns of these statistics along with sample sizes.

Statistics from 2003 to 2008 are available in previous publications.

Uses of the statistics

The statistics have a variety of uses including to:

  • monitor the key measures of AE of workplace pension participation and amount of savings
  • inform policy analysis and policy development
  • answer Parliamentary Questions and Freedom of Information requests
  • inform briefing and submissions
  • inform departmental costings
  • allow access to pension statistics for external users (for example the pensions industry, academic institutions, research organisations, media, members of the public)

Sources of the statistics

The statistics in this publication come from different data sources.

Annual Survey of Hours and Earnings (ASHE)

ASHE is conducted by the Office for National Statistics (ONS) and is a key source of information on workplace pensions in GB as it collects information on all types of workplace pension: occupational pension schemes, group personal pensions and group stakeholder pensions.

The survey results are used widely in order to analyse pension participation and to monitor the impacts of pension reforms.

Read the background information on this data source in the Annual Survey of Hours and Earnings (ASHE) methodology and guidance.

DWP Family Resources Survey (FRS)

The FRS is sponsored by the Department for Work and Pensions (DWP). The survey collects information on the income and circumstances of individuals living in a representative sample of private households in the United Kingdom.

The primary objective of the FRS is to provide DWP with information to inform the development, monitoring and evaluation of social welfare policy.

Detailed information is collected on respondents’ income from all sources including benefits, tax credits and pensions; housing tenure; caring needs and responsibilities; disability; expenditure on housing; education; childcare; family circumstances; child maintenance.

HMRC Real Time Information (RTI)

RTI is His Majesty’s Revenue and Customs (HMRC’s) reporting system for income taxed via Pay As You Earn (PAYE). Employers and pension providers are required to report to HMRC payments to employees, or recipients of occupational pensions, on or before each payment date where it is practical to do so. Within RTI, submissions relate to a payment to an employee, or occupational pension recipient. When they are submitted to HMRC by the PAYE scheme, they are contained within a Full Payment Submission (FPS). RTI includes information about the PAYE scheme, the employee (or occupational pension recipient), and the payment.

RTI only holds information on employments and pensions that are reported through the payroll reporting process. RTI is not designed to and does not include information about Self-Employment or pensions which are not paid via PAYE, such as State Pension and income from other sources that is seen as pension income by the individual.

PAYE scheme is not required for an employer if all of their employments are paid less than the Lower Earnings Limit of £125 per week or £542 per month (as of 2025 to 2026). If any single employee earns more than this, is in receipt of a pension, has another job, or receives expenses or benefits from the employer, the employer is required to report RTI for all employees. Some employers are also exempt from online payroll reporting, and therefore do not need to submit RTI electronically. This may be due to religious beliefs, where care services must be provided to the employer or a member of their family, or other reasons.

The published data using RTI is structured around three separate sets of data that are both derived from the PAYE schemes listed on RTI. These are:

  • monitoring of stopping saving
  • monitoring of private pension payments

Monitoring of stopping saving

The aim of this analysis is to identify the number of employments that have stopped paying pension contributions in each tax month. This is an update to stopping saving metrics previously published in the Automatic Enrolment Evaluation Report 2019.

The methodology has 3 steps:

  1. Identify spells of pension saving.
  2. Classify whether an employee has stopped pension saving in a particular month, and if yes, the reason for stopping.
  3. Break down the analysis with several classification variables.

This analysis was produced on a five tax month rolling period, referred to as the interest period. To estimate stoppages in a particular month, this interest period covers that month and the 4 following tax months. To calculate the number stopping saving for October 2022 for example, the interest period would include the months from October 2022 to February 2023.

To identify whether someone who was pension saving in the first month of the interest period through a particular employment stopped or continued saving, we look at their payment records in the interest period. The payment records in the last four months of the interest period will determine if we identify the employment as stopping pension saving in the first month of the interest period or if they continued to pension save. For example, if an employee who was saving in October did not make any pension contributions on any of their subsequent payments in November to February through that employment, we classify them as having stopped pension saving into that employment’s pension scheme in October.

The stopping saving rate for each month is calculated using this methodology, looking at the subsequent four months of payment records to identify whether an employee has stopped pension saving into that employment’s pension scheme in that month. As outlined in the limitations, it is possible that the employee may still be receiving employer pension contributions.

To identify spells of pension saving, we identify records that had employee pension contributions. If, on any payment date in a tax month, an employment was found to have employee pension contributions greater than £0 exactly, they were classed as paying employee pension contributions.

All records for an employment in the last 17 complete tax months at the point of analysis, including the interest period, were analysed to find spells of pension saving. This then identified tax months when the employment was making employee pension contributions. Of those identified as pension saving, only employments with employee pension contributions in the interest period were included in the analysis.

The volume of employments with employee pension contributions in each tax month of the interest period were calculated. This figure is used to estimate the level of participation in workplace pension saving per month. Of the most recent spell of pension saving, the last payment with employee pension contributions was identified.

Once the most recent employee pension contribution has been identified, we identify whether an employee has been classed as having stopped their pension saving. By looking at patterns of pay and employee pension contributions, we can infer pension savings behaviour within an employment and the reasons for it.

In this publication we monitor an employee as having stopped saving under the circumstance that it was an active decision; If an employment is still active and eligible, but stops contributing into a workplace pension, we classify this as an active decision to stop saving.

Data from private pension providers

DWP receives data from a selection of private pension providers who provide information on their number of active members, the amount of pension contribution from both employee and employer, and the number of members stopping saving.

12 private pension providers contribute their data to DWP, though not all of these providers provide information on all metrics. These providers account for the majority of the private pension market, though coverage is not comprehensive. Consequently, caution should be taken when applying any trends to the whole pension saving population.

The primary objective of collating the private pension provider data is to offer a timelier insight into pension participation trends as new data is gathered monthly. This has proved most useful in monitoring the success of AE in increasing and maintaining the number of savers throughout Covid-19 and periods of higher cost of living.

Definitions and terminology within the statistics

Annual Survey of Hours and Earnings (ASHE)

Term Definition
Eligible employee An eligible employee is an employee who meets the AE age and earnings criteria (age at least 22 and below SPa and earning above the AE earnings threshold each year, currently £10,000). This includes employees already a member of a workplace pension scheme when AE was introduced.
Ineligible employee An ineligible employee is an employee who fails to meet either of the AE age and earnings criteria (Ineligible if age at below 22 or above Spa, and earning below the AE earnings threshold each year, currently £10,000).
All employees All employees are all employees, irrespective of whether they meet AE age and earning criteria.
Eligible saver An eligible saver is an eligible employee who is saving into a workplace pension.
All savers All savers are all the employees who are saving into a workplace pension, irrespective of whether they meet AE age and earning criteria.

HMRC Real Time Information (RTI)

Term Definition
Earnings eligible In both the participation and contributions analysis, only employments that are earnings eligible are included. If an employment is listed relating to an individual between the age of 22 and state pension age, and has an annual income greater than £10,000 (daily taxable pay greater than £27.40) at the last payment date in the tax month, then the employment is classed as earnings eligible.

Private Pension Provider data

Term Definition
Opt Outs The percentage is the number of employees terminating their workplace pension within the opt-out period, divided by the number of new member enrolments.
Stopping Saving The percentage is the number of members making an active decision to cease saving, divided by total active members.

Comparisons between the statistics

ONS Funded occupational pension schemes in the UK

This publication included statistics of employee pension contributions from the Financial Survey of Pension Schemes. The latest version of this publication covers April to September 2024.

ONS Pension Wealth in Great Britain

This publication included trends in active private pension wealth from the Wealth and Assets Survey. The latest version of this publication covers household wealth (including pension wealth) from 2020 to 2022.

DWP Family Resources Survey

This publication included statistics on pension participation for working-age adults. The latest version of this publication covers the financial year 2023 to 2024.

Technical notes for accompanying tables

The data behind each of the charts using ONS ASHEHMRC RTI, DWP FRS and Private Pension Provider data included in the publication can be found in the accompanying tables.

Notes:

  • the analysis includes members of all workplace pension schemes: occupational pension schemes, group personal pensions (GPPs) and group stakeholder pensions (GSHPs).
  • all analysis is based on eligible employees (unless otherwise stated). Amounts saved calculated from the ONS ASHE data are adjusted to 2024 earnings terms using ONS Average Weekly Earnings (AWE) values. Gross annual earnings are derived using weekly pay, and no filter has been included for loss of pay in the pay period. The ONS Average Weekly Earnings Statistics, EARN01 (KAC3) series is used
  • the corresponding earnings thresholds have been used from 2012 onwards and deflated using ONS AWE between 2008 and 2011 to determine notional equivalent AE eligibility before 2012
  • State Pension age (SPa) began to increase during 2010. The age tables take account of this change and therefore SPa varies from 2011, these changes gave also been applied when selecting employees between 22 and SPa. See the State Pension age timetables for more information
  • these estimates use the Standard Industrial Classification (SIC) 2007 codes to identify industries
  • data up to 2011 is based on Standard Occupational Classification (SOC) 2000. From 2011 onwards, SOC 2010 is used, while from 2022 onwards SOC 2020 is used. Consequently, there are a couple of slight breaks in the series. Therefore, care should be taken when interpreting the full time series
  • in the amount saved tables, income tax relief on the employee contribution is calculated as the difference between the income tax due on observed earnings and the income tax that would be due if the employee contribution were treated as earnings
  • numbers have been suppressed where the sample size is very low to prevent risk of disclosure, and where the sample size is relatively low results have been marked to indicate that they will have a relatively high degree of uncertainty. Sample size cut-offs, and rounding rules used, are different for tables using ASHE data and tables using FRS data according to the relevant guidance for each dataset and consistency with other Official and National Statistics using these datasets. See the notes for each accompanying table
  • for the ASHE data utilised in iterations of this publication before 2023, ONS implemented validation checks for the pension saving variables. However, due to a scaling back of these checks by ONS, DWP have implemented a cleaning process from the 2023 publication onwards. For the latest July 2025 publication, an updated cleaning approach has been implemented following a thorough review of the data, with this explained in more detail in the following “Annual Survey of Hours and Earnings Data Cleaning” section

Annual Survey of Hours and Earnings Data Cleaning

Alongside the existence of extreme pension contribution values in the ASHE data, further investigation highlighted that there may be an underestimate of some individuals’ pension contributions. Consequently, DWP analysts have implemented a robust cleaning process on the raw data supplied by ONS.

Analysis identified that, for a significant number of records, employee contribution values are missing tax relief, with this particularly the case for individuals within pension schemes known to utilise the Relief at Source (RAS) tax relief method. For individuals in such schemes, tax relief is claimed separately by their pension provider, with the omission of tax relief when completing the survey therefore understandable. Consequently, an algorithm has been utilised to identify the records from which tax relief is omitted, with the employee contribution values for these records increased accordingly.

Alongside the correction of omitted of tax relief, adjustments have been made in instances where:

  • missing data prevents the calculation and assessment of contribution rates, including where employee earnings values are omitted
  • contribution values are slightly below AE minimums, with contribution values consequently rounded to meet these legal AE minimums
  • contributions are either extremely low or extremely high, suggesting survey error, with an algorithm implemented to reassign the contribution values in such instances

This cleaning process has been implemented from 2019 onwards, with this being the first year in which the current AE minimum contribution rates (8% of qualifying earnings) were required. Consequently, a consistent cleaning approach has been able to be applied across the 2019 to 2024 period. This may make comparisons with data before 2019 more challenging; however, this does not impact the trend nor narrative with any impact likely to be less than 1ppt on average contribution levels.

The approach has been Quality Assured and compared to other known sources. For example, Nest Insight’s Employer Survey[footnote 1] found around 4-in-10 employees were saving at default AE levels (this analysis finds around 3-in-10); comparisons to PPI’s DC Future Book[footnote 2], where pension contribution rates show a similar trend in average contribution rates; and the most common average contribution rate received within the ASHE data now being 8% (whereas previously it was 7%, strongly suggesting 1% of tax relief was not being captured within the data).

Key assumptions and limitations of the statistics

ONS Annual Survey of Hours and Earnings (ASHE)

ASHE notes:

  • ASHE is based on a 1% sample of employee jobs taken from HMRC PAYE records. Information is obtained from employers and treated confidentially. ASHE does not cover the self-employed nor does it cover employees not paid during the reference period
  • the 2024 ASHE data has a reference date of the week containing 17 April 2024
  • ASHE collects information on employee membership of the current employer’s workplace pension scheme. This does not include preserved rights in any former employer’s pension scheme or pensions paid by former employers
  • ASHE collects information from employers on employee jobs, although they are referred to in this Official Statistic as ‘employees’
  • the overall level of uncertainty arising from the sample size of ASHE is low, however uncertainty may be higher for particular subgroups
  • prior to the coronavirus (COVID-19) pandemic, the achieved sample size for ASHE was approximately 180,000 each year. However, given the challenges to data collection during the coronavirus pandemic, the final achieved sample size was 144,000 for 2020,142,000 for 2021 and 148,000 for 2022. As such, ASHE estimates for 2020, 2021 and 2022 are subject to more uncertainty than previous years. There has been an improvement in response rates in 2023 and 2024, with sample sizes returning closer to the levels seen before the COVID-19 pandemic, with 164,000 responses in 2023 and 173,000 responses in 2024
  • for the data from 2023 onwards, ONS have introduced improvements to the methods for processing returns to the survey. This has ensured that more high earners in each profession have been counted in the final data, thus increasing the achieved sample size and improving the quality of ASHE estimates. Therefore, ONS caution against comparing 2023 and 2024 with earlier years

For further information on ASHE please see the background notes section on the ONS website.

DWP Family Resource Survey (FRS)

FRS notes:

  • the latest FRS data was collected throughout the 2023/24 financial year and is not collected with reference to a specific time period like the ASHE data. Therefore, the two sources are not directly comparable
  • the impairment types used to define disability status were changed in the 2012 to 2013 survey to reflect new harmonised standards and therefore caution is needed where making comparisons over time
  • participation rates calculated using the FRS include participation in personal as well as workplace pensions

HMRC Real Time Information

RTI notes:

  • it is possible that some employments may be counted more than once as stopping saving. This occurs if an employment stops saving in one month, made no employee pension contributions for at least five tax months, restarted employee contributions after this time period and then stopped making employee pension contributions once again
  • the volumes of employments stopping saving and other figures estimated using HMRC’s RTI data may differ from other estimates, such as surveys and sampled data, due to differences in counting methodologies etc. RTI only includes information on when pension contributions were made, not when a request is made to stop saving
  • months refer to tax months. For example, the tax month of February 2023 was 6th February 2023 to 5th March 2023
  • as only employee pension contributions and tax relief under RAS schemes are reported through RTI, this analysis is based purely on employee pension contribution data. The analysis cannot identify employees saving into a workplace pension with employer-only contributions or via salary sacrifice. An employee with employer-only contributions, or saving via salary sacrifice, will be indistinguishable from an employee who is not saving into a pension, as both will have no employee pension contributions on RTI
  • self-employed individuals are not required to submit RTI, although there are some cases where these individuals are paid through PAYE and thus will appear on RTI
  • payments relating to company directors for NICs purposes, as defined via data item 84A in the RTI data items guide, are removed from the analysis. This impacts employments which had payments as both a company director and not
  • if the employment moved in and out of the population of interest, due to changing from employment to pension or from director to non-director, it would appear that the employment had ended. The classification of stoppage will vary depending on the length of time the employment is not in the population of interest and the length of time before the employment returns to the population of interest
  • if an individual has multiple employments, the individual could be represented multiple times in RTI
  • some information submitted by employers for RTI is Late, Missing or Incorrect. This is common for all administrative data
  • the population of identified pension savers may include employments that started pension saving before AE was introduced as mandatory. These cases usually relate to employers that provided the option of pension saving before AE was rolled out

Data from Private Pension Providers

Private Pension Provider notes:

  • data has been collected from 12 providers. However, for some providers data does not cover all months in the time series (for example, where a provider began sharing data after January 2020). When producing index statistics on contributions and active members, providers have only been included if data is available for all months in the time series, meaning the number of providers is consistent across all months. For the measure of stopping saving, all providers are included but some providers’ data does not cover all months in the time series
  • these providers account for the majority of the private pension market, though coverage is not comprehensive. Consequently, caution should be taken when applying any trends to the whole pension saving population

Status of the statistics

National, Official and Experimental statistics

These are Official Statistics.

The DWP FRS data is designated by the UK Statistics Authority as National Statistics.

Statistics sourced from HMRC Real Time Information and pension providers in this publication are badged as experimental statistics. This is due to the ongoing development of the data systems and statistics used to support. The methodologies used to produce these statistics are constantly monitored and are subject to revision as improved data sources and methodologies become available.

Quality Statement

All data sources used in this publication have undergone detailed quality assurance processes, including investigations of outliers.

ONS ASHE and DWP FRS have their own measures in place. In addition, we pay attention to the pension variables within these surveys.
HMRC RTI data and analysis used in this publication has undergone a detailed quality assurance process by both HMRC and DWP. Data from private pension providers has also been through a detailed quality assurance process within DWP.

Revisions

New ONS ASHE data is marked as provisional and it is possible that revisions will be made in later releases of the data, historically these have been minimal.

Feedback

We welcome feedback.

We would welcome any views you have using the following contact information:

Latest release

Workplace pensions participation and saving trends: 2009 to 2024

Previous releases

Workplace pensions participation and saving trends: 2009 to 2023

Workplace pensions participation and saving trends: 2009 to 2022

Workplace pensions participation and saving trends: 2009 to 2021

Workplace pensions participation and saving trends: 2009 to 2020

Workplace pensions participation and saving trends: 2009 to 2019

Workplace pensions participation and saving trends: 2008 to 2018

Automatic enrolment evaluation reports were published annually between 2013 and 2019. Read the latest Automatic Enrolment Evaluation Report 2019 publication.

The Office for National Statistics has previously published statistics on the participation in 2020 of all employees, using the same data source used in this publication for eligible employees. Read the latest Employee workplace pensions in the UK: 2021 provisional and 2020 final results publication.

DWP published analysis on saving and contribution levels in July 2025.

More information about AE

Automatic Enrolment - workplace pension duties

Workplace Pensions

More Information about HMRC RTI

Guidance on RTI Data Items from April 2021

Other National and Official statistics

Details of other National and Official Statistics produced by the Department for Work and Pensions can be found on the DWP website and at the following links.

Stat-Xplore – the DWP benefit statistics dissemination tool

Read a schedule of statistical releases over the next 12 months and a list of the most recent releases

In accordance with the Code of Practice for Statistics, all DWP official statistics are announced in the GOV.UK release calendar.