Official Statistics

Quality report: benefits in kind statistics

Published 30 September 2020

1. Introduction

The benefits in kind statistics publication contains a series of tables about the taxable benefits in kind paid by employers. The tables focus on two benefits: company cars and company car fuel. The tables show the number of recipients, the taxable value of the benefits and the associated Income Tax and National Insurance contributions (NIC) liabilities. Breakdowns are provided by type of benefit, by income level and there are breakdowns by CO2 emission level and fuel type.

Until 2016 to 2017, employers were required to provide HM Revenue & Customs with details of all taxable benefits in kind provided to directors and employees. This information was reported on form P11D at the end of each tax year. From 2016 to 2017, the requirement to make P11D returns has been removed if the employer has registered for voluntary payrolling. Data returned by employers on the P11D return are captured on the HM Revenue and Customs Employer Compliance System (ECS).

Most benefits in kind which are subject to voluntary payrolling do not have to be reported to HMRC. However there is a requirement to report company cars as part of PAYE submissions via Real-Time Information (RTI).

The combined P11D and RTI dataset for a tax year is matched to the Survey of Personal Incomes (SPI), a large sample drawn from administrative records of individuals in contact with HMRC over the course of the year. The income data in the matched dataset make it possible to produce statistics for tax and National Insurance liabilities on benefits in kind and breakdowns by income.

For the years 2017 to 2018 and 2018 to 2019 within this publication, the P11D data is supplemented with RTI data to account for car and car fuel benefits that were not reported on the P11D form due to payrolling.

The number of recipients and the amount of each car and car fuel benefit can be obtained from the combined P11D and RTI data, and are reported in Tables 4.1, 4.3, 4.4, and 4.5.

2. Relevance

This publication contains a series of tables about the taxable benefits in kind paid by employers in respect of company cars and company car fuel. The number of recipients and taxable values for Tables 4.1, 4.4, and 4.5 are based upon the P11D returns provided by employers for the relevant tax year, supplemented with RTI data related to car and car fuel benefits that were not reported on the P11D form due to payrolling The estimated tax and NICs liabilities also draw on data from HMRC’s annual Survey of Personal Incomes (SPI).

Other taxable benefits in kind and expenses are not included in this edition of the publication.

More detail on these exclusions is provided in the accompanying release document.

3. Accuracy and reliability

The following issues have a bearing on the accuracy of the figures provided. The tax, NICs and earnings breakdowns in Tables 4.1, 4.3 and Table 4.5 are based on the SPI sample survey and are therefore subject to sampling error. However, the sample size is large (the full matched sample contains between 90,000 and 100,000 individuals) so the estimates should be robust.

From April 2016, employers have been able to register for ‘voluntary payrolling’, removing any requirement to submit a P11D return when deducting tax through the payroll system. The number of schemes payrolling company cars has increased year on year since 2016 to 2017.

This has created incompleteness in HMRC’s data in the following ways:

• in the first year of voluntary payrolling 2016 to 2017, there was no system for reporting payrolled company cars

• in 2017 to 2018, payrolled company cars could be reported in PAYE submissions via Real-Time Information (RTI) but this was on a voluntary basis

• in 2018 to 2019 reporting of payrolled company cars became mandatory but analysis suggests there is significant under-reporting

It is difficult to produce accurate estimates of the number of unreported company cars. However initial analysis suggests that they account for a large part of the reduction in reported cars observed since 2015 to 2016.

More detail on accuracy is provided in the background documentation.

4. Timeliness and punctuality

Due to the time needed to receive and process tax returns and information provided by employers, the first publication of data for a new tax year normally occurs 15 months after the end of the tax year, in the form of projections in Table 4.5.

The SPI data for the new tax year (used to calculate tax and NIC liabilities and breakdowns by income) does not become available until several months later, and the P11D data are also subject to revision. Consequently full and final statistics are published in the following year’s release, 2 years and 3 months after the end of the tax year.

The punctuality record in recent years has been as follows:

• 2016: published on schedule in June 2016

• 2017: published on schedule in June 2017, revised in February 2018 to correct an error on Class 1A benefits in Table 4.1 and 4.5

• 2018: published on schedule in June 2018, revised in July 2018 to correct an error in the provisional tax liability 2016 to 2017 estimates that appeared in Table 4.5

• 2019: published on schedule in June 2019

• 2020: published in September 2020, three months late, as a result of demand on resources related to the COVID-19 pandemic.

5. Accessibility and clarity

Benefits in Kind statistics are available free to users via the National and Official Statistics section of the HMRC website.

Benefits in Kind statistics are released via a comprehensive HTML statistical bulletin which includes statistical commentary and a detailed description of data sources, methods, quality indicators and contextual information (e.g. describing the Benefits in Kind system).

The detailed tables are available separately in Excel format.

6. Coherence and comparability

There are no alternative sources for the information presented in the tables.

Since the statistics are based on P11D returns, the increasing take-up of voluntary payrolling of benefits in kind has implications for the coverage and completeness of the statistics. For 2017 to 2018 and 2018 to 2019 the P11D data has been supplemented with RTI data. However reporting of payrolled company cars was on a voluntary basis in 2017 to 2018 and analysis suggests there was significant underreporting in 2018 to 2019.

7. Trade-offs between output quality components

There is a trade-off in publishing the data in a timely fashion, and ensuring it is as accurate as possible, minimising revisions.

As a result, the figures for the last year published in Table 4.5 are partly based on projections and are subject to updating. Final figures for that year together with new projections for the subsequent year will be included in the next annual update.

8. Assessment of user needs and perceptions

Users can provide feedback by contacting the relevant statistician whose details are provided in the publication.

In addition, a user engagement exercise is being carried out between now and December 31st 2020 to better understand how the new statistics can meet user needs.

9. Performance, cost and respondent burden

Both the P11D, RTI and SPI data are taken from administrative records. The compilation of these statistics does not therefore impose any additional burden on employees or employers.

10. Confidentiality, transparency and security

The datasets we use contain confidential taxpayer information and are handled in accordance with strict security guidelines. The breakdowns provided in our table meet HMRC’s commitments to protect taxpayer confidentiality and personal information.

Information contained in the tables is not given out in final form until the day before publication and then only to nominated officials and ministers whose names are published on the HMRC website.