Official Statistics

Background Quality Report: Coronavirus Job Retention Scheme Statistics

Updated 16 December 2021

Contact

• Organisation unit - Knowledge, Analysis and Intelligence (KAI)
• Name - C Breeze, R Waddell
• Team - Statisticians, COVID-19 Analysis, Coordination and Evaluation team
• Email - CJRS.Statistics.Enquiries@hmrc.gov.uk

Statistical presentation

This section gives an outline of government policy surrounding the Coronavirus Job Retention Scheme (CJRS) and the statistics published. This includes details on the government’s contributions and the criteria for which people were eligible to claim, some background on the statistics produced, key definitions and the timeline and areas covered by the scheme.

Policy Background and Data description

The government announced the Coronavirus Job Retention Scheme on 20 March 2020. It was introduced to support employers through the COVID-19 period by helping keeping millions of people in employment; it has also been known as the furlough scheme.

The scheme is based around HMRC’s Pay As You Earn (PAYE) system. It works by providing grants to employers of up to a maximum 80% of salary to a maximum value of £2,500 per employee. Up to the end of July 2020, the scheme also met some of the cost of employer pension contributions and employer National Insurance Contributions.

From 1 July 2020 onwards, employees could be put on furlough under the CJRS for part of their hours and work the remainder. This was called flexible furlough. For staff on flexible furlough, the £2,500 monthly limit for claims was reduced by multiplying the limit by the proportion of the employee’s usual hours for which they had been put on furlough.

From 1 September 2020, the Government contribution supported 70% of salaries for hours not worked, reducing to 60% from 1 October. The employee continued to receive 80% of their wages with the remainder being made up by employer contributions.

The initial scheme was subsequently extended from 1 November 2020 for claim periods to 30 April 2021.

It was announced on 3 March 2021 that the CJRS would be extended, beyond 30 April to 30 September 2021.

From 1 July 2021, the Government contribution supported 70% of salaries for hours not worked, reducing to 60% from 1 August. The employee continued to receive 80% of their wages with the remainder being made up of employer contributions.

Up to date information about the Coronavirus Job Retention Scheme can be found on the GOV.UK website. You can read previous versions of the CJRS guidance on the national archives.

The data for these statistics comes from HM Revenue and Customs’ (HMRC) CJRS claims.

The statistics include figures from the start of the scheme in March 2020 and have been published approximately monthly since June 2020. The statistics are Experimental Statistics and subject to revisions. The publications can be found on the COVID-19 collections page.

HMRC separately publishes figures for the value of claims paid to employers, in its monthly publication on tax receipts. Those figures are on a cash basis and reflect the time that the payments were made rather than the date that claims were received, so are slightly different to the figures included in these statistics.

Contents of the CJRS Statistics

This section describes the content of the CJRS statistics and gives some details about the data used.

This publication includes daily time series of employments on furlough by sector, employer size, country and region, gender, age, and furlough type. There are also more detailed monthly snapshots which include breakdowns by:

• employer size
• broad industry sector and detailed industry sector, classified by Standard Industrial Classification (SIC) 2007 codes
• country and region and local authority (also broken down by broad industry sector) and UK Parliamentary constituency
• employee gender
• employee age
• furlough type (full or flexible)
• estimated annual pay
• how long jobs have been on furlough continuously

Cumulative statistics for total number of employments on furlough, total number of employers that have used the scheme and total value of claims made are also published. These statistics are also broken down by local authority and Parliamentary constituency. In the 16 December 2021 publication the total number of employments ever on furlough is additionally provided broken down by sector and employer size.

Also in the 16 December 2021 release, the total number of individuals ever on furlough since the start of the scheme is included broken down by age and gender. If an individual has been put on furlough in more than one job during the CJRS they will only appear in this data once. Individuals have only been counted where they can be reliably identified from their National Insurance number.

Information on the number of employers and employees on furlough, furlough dates, claim values and furlough type is based on data submitted by employers as part of their CJRS claim. Information on employer sector, size, employee age, gender, estimated annual pay and geography are based on data provided by employers to HMRC’s PAYE RTI system.

A unique claim reference number is assigned to each claim. The employer reference and employee NINO are used in aggregating the data.

Sector coverage

Employers operating in all industries were eligible to apply for support from the CJRS provided they were operating within the UK and had a UK bank account.

Statistical concepts and definitions

Employments

The CJRS allowed employers to claim support in paying the wages of employees they put on furlough through their PAYE scheme. For these statistics a person employed through a PAYE scheme counts as an employment. In the statistics the term job is also used to mean an employment.

An individual employed by more than one employer (PAYE scheme) is counted once for each employment from which they were put on furlough. Two individuals who were employed in the same role with an employer at different times are counted as two separate employments.

Employers

Employers are described as eligible to claim the CJRS if they employed staff for which the criteria to be eligible for furlough were met. The details of the eligibility for employers are detailed further in the statistical population section.

An employer is assumed within the CJRS statistics to have in effect the same scope as a Pay As You Earn (PAYE) Scheme. While most employers operate a single PAYE scheme, for some employers a single PAYE scheme does not relate exactly to what is commonly understood to be an employer.

For example, some organisations operate multiple payrolls, and in other situations, a group of companies may pool their payrolls together under one PAYE scheme. However, in our view PAYE schemes provides a reasonable proxy for employers for the purposes of this release.

Age of employee

The claimant’s age is calculated at March 2020 for employments on furlough up to 31 October, at 1 November 2020 for employments on furlough from 1 November to 30 April 2021 and at 1 May 2021 for employments on furlough from 1 May 2021 onwards.

Claim value

The claim value is the total amount claimed for by an employer in a CJRS claim to support the wages of employees put on furlough. The figures reported on the value of claims do not take account of voluntary repayments nor remove claims not paid for compliance reasons. Claims cancelled by employers within 72 hours of making the claim are excluded.

Employer size

Where it has been possible to match CJRS data to PAYE RTI, we have estimated the size of each employer in terms of number of employees estimated to be eligible for CJRS support.

Furlough Type

From July 2020, it was possible to put employees on furlough for some of their usual hours under the scheme as well as for the whole of their job. Employees put on furlough under the scheme for a part of their usual hours only are described as being on flexible furlough. On the other hand, the term full furlough describes situations where an employee was on furlough for all their usual working hours.

An employee on furlough could take part in training, volunteer for another employer or organisation, or work for another employer (if contractually allowed).

Sector

The industrial sector of employers is classified under the UK Standard Industrial Classification (SIC) 2007. The information used comes from the Interdepartmental Business Register (IDBR) produced by the Office for National Statistics (ONS). Where PAYE schemes are absent from the IDBR, we have used sector information from Companies House, linking on employer name where possible.

This provides UK Standard Industrial Classification (SIC) 2007 for employers that have made a claim. Where we have been unable to determine SIC codes, we have reported the sector as ‘unknown’.

Geographical breakdowns

There are geographical breakdowns in the data covering regions, local authorities and Parliamentary constituencies. These are based on employees’ residential addresses. The geographical breakdowns are derived using postcode lookup files from the ONS to link UK postcodes to geographic areas. More detailed information on the geographical data can be found using the Open Geography Portal.

Estimated Annual Pay

The estimated annual pay is based on the gross pay over the period April 2019 to March 2020. The gross pay figure is estimated as pay before any tax, National Insurance or pension contributions are deducted from employee’s pay. If a job does not span the full 12 months the pay from the other months is annualised, or if annual pay data is not available for that period then data from April 2020 to March 2021 is used.

These figures should be interpreted as the estimated gross pay rate for employments, as employees may have more than one job these figures may not represent the pay received by individuals. Additionally, as jobs may not last for a year or more, the figures should not be interpreted as total annual amounts of pay for jobs on furlough.

How long jobs have been on furlough continuously

Employments are defined as on furlough continuously if they have been on either full or flexible furlough without a break in furlough lasting more than 3 days.

Statistical unit

The units in the statistics are employers and employments.

Statistical population

All employers eligible to claim for the Coronavirus Job Retention Scheme.

The population of employers and jobs eligible to use the CJRS has been brought up to date to cover new employers and jobs each time the scheme has been extended. The criteria for eligibility for each phase of the scheme is explained below.

Coronavirus Job Retention Scheme Main Period

For furlough from the start of the scheme in March 2020 to 30 June 2020, employees were eligible if they were employed on 19 March 2020 and included in PAYE Full Payment Submissions for 2019 to 2020.

Employees employed on 28 February 2020 but who left their job before 19 March and were later re-employed by the same employer were also eligible for CJRS.

Introduction to flexible furlough

From 1 July 2020 employers could put their employees on flexible furlough. This introduced the possibility to put employees on furlough for some of their usual hours only as an alternative to putting them on furlough for all their usual working hours.

Coronavirus Job Retention Scheme Extension Period 1

For furlough from 1 November 2020 to 30 April 2021, employees were eligible if they were employed on 30 October 2020 and included in PAYE Full Payment Submissions between 20 March 2020 and 30 October 2020. Only employments in RTI submissions received by HMRC by 30 October are counted.

Employees employed on 23 September 2020 but who left their job before 30 October and were later re-employed by the same employer were also eligible for CJRS.

Coronavirus Job Retention Scheme Extension Period 2

For furlough from 1 May 2021 onwards, employees were eligible if they were employed on 2 March 2021 and included in PAYE Full Payment Submissions between 20 March 2020 and 2 March 2021.

Reference area

The geographic region covered by the Coronavirus Job Retention Scheme is the United Kingdom (UK).

Time coverage

The statistics cover the time period from March 2020 until the latest month for which a complete set of CJRS data is available.

Statistical processing

Source data

The data for CJRS claims comes from employer submissions to HMRC.

For larger claims employers had to send HMRC a separate file attachment containing the details of the jobs claimed for. Before 27 May 2021 this applied to all claims for 100 or more jobs on furlough. From this date employers claiming for between 16 and 99 jobs could also use this method to submit their claim.

The specification for the file attachment to claims covers job level information including NINO, full name, furlough start date and end dates, amount claimed per employee and three hours fields detailing the number of hours normally worked, the number of hours actually worked, and the number of hours on furlough.

The data in these statistics comes from combining details about furlough from the job level data provided by employers and the total number of staff and claim amount submitted for each claim.

Employers making larger claims may submit start and end dates for the periods that individual employees were on furlough in the attachment when those dates were different to the overall period covered by the claim. But for smaller claims HMRC does not collect the furlough start and end dates for each job separately.

For these jobs, we use the claim period to estimate when the job was on furlough. For example, if a job in one of these claims was on furlough from the 1 May 2021 to 16 May 2021, but the full claim period was from 1 May to 31 May, the job would appear to be on furlough for the full month. This leads to an overestimate in the number of jobs on furlough at the end of the month during periods where levels of furlough are decreasing.

RTI Data

The population of employments eligible for CJRS is linked to HMRC’s PAYE RTI data. The RTI data contains the NINO and name as well as information such as employee date of birth, gender and postcode. This allows for employer and employee level demographic breakdowns.

Frequency of data collection

The CJRS claims are updated into databases for analysis in real time.

The RTI data is updated when there is a change in the population. This is usually when there is a change in the eligibility, a change with the ONS postcode tables or if there has been an improvement to the methodology.

Data collection

Data on CJRS claims is sourced from the CJRS portal where employer claims are submitted and processed.

Data validation

Checks carried out on the data include:

• eligibility figures are compared across time and with labour market statistics for sense checking the size of the eligible population
• when running the code, the number of rows output after each step is quality assured and compared with previous runs. Unusual differences would alert us to errors within the code or data. Run times of the code are also checked in a similar manner
• outputs are checked against external publications (ONS BICS) as well as previous HMRC publications to sense check the scale of the numbers on furlough
• outputs are checked for consistency across the snapshots and time series produced, ensuring totals add up across the publication
• disclosure and dominance checking is carried out on the outputs to ensure HMRC does not breach its legal duty to maintain confidentiality of taxpayer information
• statistics production – a quality assurance checklist with the most common problems that we go through every publication to check for errors in the publication tables, graphs and commentary

Data compilation

In some circumstances HMRC holds incomplete information about employments, for example where a leaving date had not been submitted by employer. In this situation, an estimate of the probability that an individual was employed on the qualifying dates has been used.

The assessment of whether a person was employed on the qualifying dates is based on the methodology used for the joint HMRC/ONS statistics release Earnings and employment from PAYE RTI.

In producing the time series statistics some challenges had to be tackled. These included dealing with data on amendments to claims and claims for overlapping periods. In addition, claims where 16 or more staff are being put on furlough may include staff on furlough for varying periods. These factors combined with some incomplete data make counting the number of employees on furlough over time complex.

The method employed is designed to prevent over-counting employments so far as possible, however may in certain circumstances very slightly under or over count.

The demographic breakdowns of employments include all employees that employers have put on furlough where it has been possible to link CJRS data to classifying information (such as the employee’s gender and address).

This linking has been performed using employees’ National Insurance numbers to data held within HMRC’s PAYE RTI system. When the employer-submitted National Insurance numbers are not of sufficient quality to be matched with other HMRC data, the employments have been categorised as ‘unknown’.

Quality Management

Quality Assurance

All official statistics produced by the Knowledge, Analysis and Intelligence (KAI) team in HMRC must meet the standards in the Code of Practice for Statistics produced by the UK Statistics Authority and all analysts are required to adhere to best practice as set out in the ‘Quality’ pillar.

These are experimental statistics and improvements have been made over time to the code and processes used to produce the statistics, for example, the methodology used to define the eligible population. There is a checklist that summarises the key quality assurance tasks and is used as a starting point to consider the quality assurance actions to undertake.

Staff producing the statistics follow a documented and regularly reviewed plan of quality assurance tasks. These analysts carry out quality assurance tasks, update the checklist, and pass the outputs onto the Senior Responsible Officer for review and eventual sign off.

The quality assurance task list was created for this publication and contains checks specific to the data and structure of each publication, including the analytical code, the data tables, the charts and the commentary.

Quality assessment

The quality assurance for these statistics adhered to the framework described above and the specific procedures undertaken were as follows:

Stage 1 – Specifying the question

The content of the releases has been developed over the course of the scheme. We welcome comments on the content from users and where possible have developed the release taking these into account. This is an ongoing process for each publication.

Stage 2 – Developing the methodology

Methodology was agreed and developed in collaboration with data experts and others with relevant expertise, ensuring it was fit for purpose and would deliver the required outputs.

Stage 3 – Building and populating a model/piece of code

When writing the code:

• analysis was produced using the most appropriate software and in line with good practice guidance
• data inputs were checked to ensure they were fit-for-purpose by reviewing available documentation and, where possible, through direct contact with data suppliers
• quality assurance of the input data was carried out
• the analysis was audited by someone other than the lead analyst – checking code and methodology

Stage 4 – Running and testing the model/code

When running the code:

• results were compared with those produced in previous months and differences understood and determined to be genuine
• results were compared with comparable independent estimates, and differences understood
• results were determined to be explainable and in line with expectations

Stage 5 – Drafting the final output

When drafting the final output:

• checks were completed to ensure internal consistency (e.g. totals equal the sum of the components)
• the final outputs were independently proofread and checked

Relevance

User needs

The statistics published on the CJRS have been of wide interest to users including in the following categories:

• national government – policy makers and MPs
• regional and local governments
• academia and research bodies
• media
• business community
• general public

User satisfaction

Formal investigations into user satisfaction have not been undertaken, however a large amount of feedback from users following the releases has been received. The statistics have been developed considering user needs following engagement with a number of users including both government, non-governmental organisations and members of the public.

Completeness

These statistics cover all CJRS claims made by employers from the start of the scheme up to the latest claims’ deadline. During the scheme, the most recent data has been incomplete as under certain circumstances claims may still be filed or amended after the initial deadline.

Therefore the most recent figures should be considered provisional results and may be revised in future releases. Information on this has been included in the CJRS statistics releases.

Accuracy and reliability

Overall accuracy

This analysis is based on administrative data, and accuracy is addressed by eliminating non-sampling errors as much as possible through adherence to the quality assurance framework.

The potential sources of error include:

• employers entering incorrect data in their CJRS claim, or – in the case of the spreadsheet-type or comma separated value files submitted with larger claims (referred to as attachments below) – not submitting the information in the required format
• issues with attachments data submitted in a non-standard layout occasionally taking up to 10 days to be processed, meaning an employer could submit an attachment before the deadline but it hasn’t been processed until after the deadline
• the data from larger claims does not include a flag identifying whether the employee is on full or flexible furlough. Instead the information about hours worked, hours usually worked and hours on furlough that employers submit is used to derive in indicator of whether the employment has been put on flexible or full furlough
• postcode information used may be out of date and can lead to inaccurate geographical breakdowns
• mistakes in the code used to produce the statistics
• errors (for example transposing figures) when producing the data tables and commentary to be published

Sampling error

These statistics are based on data for all claims; samples are not used in this analysis. Sampling error is therefore not relevant.

Non-sampling error

Coverage error

All qualifying businesses had to apply to join the CJRS to receive financial support. Coverage error is therefore not relevant.

Measurement error

The statistics provide analysis of the claims that have been made to the CJRS. Any incorrectness within claims could affect the statistics produced. The department plans to release provisional estimates of the levels of error and fraud in CJRS claims in 2020-21 and other COVID-19 support scheme grants in HMRC’s Annual Report and Accounts during Autumn 2021.

Non-response error

In general, non-response error is not applicable to these statistics as they are not based on a survey.

Employers can amend their claim up to 14 days after the claims deadline. In some circumstances, late claims can be made with a reasonable excuse and in agreement with HMRC.

Some claims filed late or amended will be captured in the next publication and the relevant amendments to the published statistics made. But amendments or late filings submitted after the data for the publication is extracted will not be captured in the publication.

Processing error

It is possible that errors exist in the programming code used to analyse the data and produce the statistics. This risk is reduced through developing a good understanding of the complexities of CJRS data, and thoroughly reviewing and testing the programs that are used.

Data revision

Data revision – policy

The United Kingdom Statistics Authority (UKSA) Code of Practice for Official Statistics requires all producers of Official Statistics to publish transparent guidance on the policy for revisions. HMRC’s policy in handling revisions is to be transparent with users about the revision. To meet this commitment, we:

• explain the specific revisions policy for that where a revision is scheduled
• ensure that the need for major revisions for any series are pre-announced on the website
• detail how users will be informed of the need for revisions
• give explanations as to the nature and extent of the revisions that were necessary

The statistics have been produced as experimental statistics. This means that while they are of good quality the methods used are subject to improvements.

Data revision – practice

These statistics cover all CJRS claims made by employers from the start of the scheme up to the latest claims deadline. The data for the most recent month is incomplete as under certain circumstances claims may still be filed or amended after the main deadline; thus, the figures for the latest month should therefore be considered provisional results and will be revised in future releases.

In addition, some improvements have been made to the statistics during their lifespan. For example, as described in the releases, early in 2021 improvements were made to the way in which employments that had ever been on furlough are counted.

Seasonal adjustment

Seasonal adjustment is not used in these statistics.

Timeliness and punctuality

Timeliness

The analysis is published monthly covering all CJRS claims made by employers to the submission deadline for the previous month. The data for the latest month is incomplete as under certain circumstances claims may still be filed or amended after the main deadline; thus, the figures for the latest month should therefore be considered provisional results and will be revised in future releases.

Punctuality

In accordance with the Code of Practice for official statistics, the exact dates of publication of the statistics have been announced at least four weeks before publication whenever possible on the Research and statistics calendar of GOV.UK. Due to the need to produce statistics on the scheme as quickly as possible this was not the case for the first release in June 2020.

Any delays to the publication date will be announced on the HMRC National Statistics website. There have been no delays to scheduled publication dates.

Coherence and comparability

Geographical comparability

The geographic areas used for these statistics have been largely unchanged over the time that these statistics have been produced. Where there have been changes, they have generally involved local authority areas merging. Therefore, there are no substantial geographic comparability issues.

Comparability over time

Data for 100% of employers and employees have been used throughout these statistics so they are comparable in that regard.

During the scheme however the eligibility criteria has changed twice so the figures for the estimated population of jobs eligible changed between the initial scheme, and the CJRS extensions beginning in November 2020 and May 2021.

There have been some improvements made to the methodology used to produce the statistics. These are mentioned in the releases. This has the potential to cause some small differences which could affect the comparability.

Coherence – cross domain

Coherence – sub-annual and annual statistics

All statistics are presented as monthly outputs. No coherence issues between different time periods exist.

Coherence – internal

Rounding of numbers may cause minor internal coherence issues as the figures within a table may not sum to the total displayed. The totals between tables should be consistent.

Accessibility and clarity

News release

The Coronavirus Job Retention Scheme has been widely reported in the media.

Publication

The tables and associated commentary are published on the HMRC coronavirus (COVID-19) statistics webpage of GOV.UK.

The statistics and commentary are published as an accessible HTML document on GOV.UK. The supporting tables are laid out and formatted to be compatible with screen reading software.

Both documents comply with the accessibility regulations set out in the Public Sector Bodies (Websites and Mobile Applications) Accessibility Regulations 2018.

Further information can be found in HMRC’s accessible documents policy and GOV.UK guidance and tools for digital accessibility.

Online databases

This data is not available in any online databases.

Micro-data access

Micro-data access is currently not available.

Documentation on methodology

A description of the methodology used is contained in the HTML publication (see section titled ‘methodology’ in the latest release of these statistics).

Quality documentation

All official statistics produced by KAI, are produced under the principles set out in the Code of Practice for Statistics produced by the UK Statistics Authority and all analysts adhere to best practice as set out in the ‘Quality’ pillar.

Information about quality procedures for this analysis can be found in the quality management section of this document.

Cost and burden

All necessary data for the CJRS Statistics is obtained from an administrative data source, i.e. HMRC CJRS and HMRC PAYE RTI data. Data is also obtained from external data sources such as the ONS postcode lookup tables and SIC 2007 code. Beyond the requirements of the administrative systems, there is no additional burden on employers or HMRC staff to provide information.

Whilst there has been considerable investment in the system to produce these statistics, on top of that it is estimated to take about 15 days full time equivalent staff time to produce each monthly publication.

Confidentiality

Confidentiality – policy

HMRC has a legal duty to maintain the confidentiality of taxpayer information. This is set out by the Commissioners for Revenue and Customs Act 2005.

This analysis complies with this requirement.

Confidentiality – data treatment

The statistics in these tables are presented at an aggregate level so identification of individual employers is minimised.

Statistical disclosure control (SDC) is used to identify figures in the text and tables that pose potential disclosure risks. Where figures are identified as presenting a risk of disclosure action has been taken.

Some cells have been suppressed, shown as [c] in the tables. In other cases, we have combined categories to manage the risk of disclosure. SDC modifies data so that data subjects cannot be identified. Disclosure in this analysis is avoided by applying rules that prevent categories of data containing:

• small numbers of contributors, and
• employers that make a dominant contribution to a value Further information on anonymisation and data confidentiality best practice can be found on the Government Statistical Service’s website.