Quality report: Undisclosed foreign income in Self Assessment by UK residents, 2018 to 2019
Published 24 October 2024
1. Contact
Contact organisation unit: Knowledge, Analysis and Intelligence (KAI) - Customer Compliance Strategy
Contact name: H Whicker
Contact person function: Head of Tax Gaps Team
Contact email address: taxgap@hmrc.gov.uk
2. Official statistics in development
2.1 Overview
These statistics use the results of a random enquiry programme targeting individuals identified to HMRC through Automatic Exchange of Information (AEOI) data to estimate the scale of non-compliance relating to UK individuals’ foreign income and gains from financial assets.
The statistics are presented as ‘official statistics in development’. Official statistics in development are official statistics that are undergoing a development. They were previously called ‘experimental statistics’.
The findings provide new evidence on the scale of non-compliance relating to the declaration of foreign income for those individuals identified through the AEOI data. The analysis excludes any non-compliance relating to companies with any form of foreign income; and individuals, foreign income or jurisdictions not covered in the AEOI data, and therefore represents a subset of overall foreign income non-compliance. HMRC are evaluating the usefulness of this analysis for users. As part of this, HMRC welcome feedback which can be provided by emailing taxgap@hmrc.gov.uk. HMRC also welcomes ideas for further expansion and refinement.
2.2 Development plan
Following publication of these ‘official statistics in development’ HMRC will gather feedback from users. This will include engaging directly with external bodies, academics and other experts. Feedback can also be provided by emailing taxgap@hmrc.gov.uk.
3. Statistical presentation
3.1 Data description
This publication presents the findings of analysis into tax non-compliance by UK residents failing to declare foreign income which is notified to the UK through the Automatic Exchange of Information (AEOI) agreements. Provided is an estimate for the 2018 to 2019 tax year on the amount of under-declared tax liability from foreign income and gains by individuals with holdings of financial assets (such as bank accounts) outside the UK in the jurisdictions which are part of these data exchange agreements. Figures on the number of individuals in scope of the analysis and proportions estimated as under-declaring their tax liability are also provided.
3.2 Classification system
Estimates of under-declared liabilities are presented in absolute monetary terms. Population counts and related estimates are presented in absolute numerical terms and as proportions of the relevant population.
3.3 Sector coverage
These findings provide evidence on the scale of non-compliance relating to the declaration of foreign income for those individuals identified through the AEOI data only. This is therefore not a complete assessment of the tax gap for foreign income by UK individual taxpayers. The assessment excludes foreign income from companies.
3.4 Statistical concepts and definitions
Under-declared tax liability and tax at risk
The additional tax owed to HMRC as identified through this activity. The tax liability specifically relates to tax owed relating to foreign income and gains and is in scope of the Self Assessment tax regime as Income Tax and Capital Gains Tax.
Tax year
The estimate for the total under-declared tax liability is for the 2018 to 2019 tax year. This covers income and gains from period 6 April 2018 to 5 April 2019 relevant to the 2018 to 2019 Self Assessment tax return.
Compliance yield
Additional tax liabilities which arise through HMRC’s compliance activity.
The tax gap
The difference between the amount of tax that should, in theory, be paid to HMRC (‘theoretical tax liability’), and what is actually paid.
The ‘theoretical tax liability’ is the amount of tax that would be paid if all individuals, businesses and companies complied with the letter of the law and HMRC’s interpretation of Parliament’s intention in setting the law (referred to as the ‘spirit’ of the law). The total theoretical tax liability is the sum of the tax gap, plus the amount of tax received by HMRC.
3.5 Statistical unit
The units that are included in this publication are the values of under-declared tax liability, people and proportion of people.
3.6 Statistical population
The statistical population are individuals who are UK residents and have holdings of financial assets (such as bank accounts) outside the UK in the jurisdictions which are part of AEOI data exchange agreements.
3.7 Reference area
The geographic regions covered by the data are the UK and all jurisdictions participating in the AEOI data exchange agreements. All individuals under consideration are UK resident for tax purposes. See the OECD status of commitments for the Automatic Exchange of Information for a full list of jurisdictions undertaking exchanges through AEOI.
3.8 Time coverage
Estimates are presented for the 2018 to 2019 tax year. Data used to produce these estimates come from the 2017 to 2018 and 2018 to 2019 tax years.
4. Statistical processing
The scope of the analysis is limited to UK residents with foreign income notified to HMRC through 2017 and 2018 AEOI data. The analysis excludes any non-compliance relating to individuals, foreign income or jurisdictions not covered in the AEOI data, and therefore represents a subset of overall foreign income non-compliance. The analysis also excludes any non-compliance from companies with any form of foreign income.
Detailed information about the statistical processing methodology, including the source data, data collection and data compilation, can be found in the Methodology section of the ‘Undisclosed foreign income in Self Assessment by UK residents, 2018 to 2019’ publication.
4.1 Source data
Automatic Exchange of Information
Analysis is based on a random sample selected from the population of UK individuals identified to HMRC through Automatic Exchange of Information (AEOI). This covers information received through the Common Reporting Standard (CRS) and the United States Foreign Account Tax Compliance Act (FATCA).
AEOI relates to information collated by participating jurisdictions from their financial institutions and exchanged with other jurisdictions, see Standard for Automatic Exchange of Financial Account Information in Tax Matters.
These data sets provide HMRC with information on the foreign interest, dividends and gross proceeds from the sale or redemption of financial assets from UK residents in the jurisdictions which are part of these data exchange agreements.
A review of the effectiveness of jurisdictions’ implementation of the AEOI standard conducted by the Global Forum on Transparency and Exchange of Information for Tax Purposes identified that, whilst most jurisdictions are delivering as expected, 20% of jurisdictions are identified as being ‘non-compliant’ and have not completed operational frameworks to verify that Financial Institutions are effectively complying with all of the due diligence and reporting requirements. Risks of misreporting are therefore present, with implications on the accuracy of the outputs presented in this publication.
AEOI discrepancy profile
Information compiled from AEOI data is matched with Self Assessment tax return data to identify taxpayers who may have an under-declared tax liability on their foreign income. Taxpayers identified as having a discrepancy and being potentially non-compliant form the target population for the random enquiry programme activity.
Data is integrated into Connect - HMRC’s analytical tool which cross-references more than 22 billion lines of data including customers’ Self Assessment returns, property and financial data. Use of this system enables HMRC to identify known individuals from the AEOI data and detect possible non-compliance where data indicates a risk of tax being under-declared.
Uncertainty is introduced as a consequence of this process. Individuals remain untraced where HMRC have been unable to match the data compiled from AEOI data to a known individual. Assumptions are applied in the analysis to expand the analysis to cover these individuals. A further assumption is applied that individuals with no discrepancy are assumed to be compliant in this analysis.
There is a risk that additional non-compliance within the population is missed as a consequence of these assumptions. Non-compliance of those individuals identified as potentially being non-compliant from this data matching exercise is tested as part of the random sampling stage.
Random sample
A random sample of 400 cases across 2017 to 2018 and 2018 to 2019 is drawn from the population of individuals identified as having a discrepancy. A stratified sampling approach is used, based on the size of discrepancy between the information from AEOI data and the Self Assessment tax return data.
Analysis relating to non-compliance within the AEOI population is based on the results of this random sample and any under-declared foreign income identified.
Compliance yield
Under-declared foreign tax liability recovered by HMRC through its compliance activity. Compliance activity is recorded by caseworkers in internal administrative systems. The value used in this paper is based on unprompted voluntary disclosures, prompted disclosures following HMRC nudge letters to individuals identified through AEOI data, and HMRC compliance checks targeting individuals identified through AEOI data.
The compliance yield values are taken on a similar basis to those used in HMRC’s annual Measuring tax gaps publication. This technical note provides additional detail on how compliance yield relates to the tax gap.
4.2 Frequency of data collection
The random sampling activity was conducted for the 2017 to 2018 tax year and the 2018 to 2019 tax year. Compliance checks can then take many months or years to resolve.
4.3 Data collection
Cases selected for the random enquiry programme are first subject to a ‘one-to-many’ approach to encourage voluntary compliance. Taxpayers are notified of their potential discrepancy and given an opportunity to tell HMRC about any underdeclared foreign income by making a disclosure. Cases where there is still an outstanding or unexplained discrepancy following this stage proceed to a compliance check. The random enquiry programme data is recorded by HMRC compliance officers.
4.4 Data validation
The values of non-compliance recorded by HMRC compliance officers for the random sample activity have been checked by reviewing individual case notes. From this, the amounts specifically related to foreign income and gains for the relevant tax year were confirmed. In some instances taxpayers were identified as amending their tax return after being notified of their potential discrepancy at the ‘one-to-many’ stage. Associated values of these amendments were checked and confirmed by a HMRC analyst for this release.
The statistics were developed using R software by experienced analysts who have previously developed other analytical models for the Measuring tax gaps publication. R software was used to select the random sample and to then compile enquiry results and provide summary numbers. These processes have been reviewed by multiple analysts and independent cross-checks have been conducted.
4.5 Data compilation
Key data compilation steps are discussed in detail in the ‘Analytical steps and assumption’ section of the report.
Open cases
A minority of cases in the random sample were still ongoing at the point the analysis was produced. To include them in the analysis the values of non-compliance are forecasted by the responsible HMRC compliance officer.
Non-detection
It is unlikely that the random enquiry programme will identify all incorrect returns or the full scale of under-declaration of tax liabilities. For these ‘official statistics in development’ we have elected not to apply a non-detection multiplier to adjust for this potentially missed yield.
Combining samples
Sample sizes used in this analysis are relatively small and lead to high uncertainty in the results, particularly given the high variation in levels of non-compliance identified across the sample. To limit this uncertainty, we combine cases from 2 separate samples across 2 tax years.
Deselections
Cases in the random enquiry programme are not worked for several reasons and this is done in a non-random way. This means that the cases which are not worked are likely to be systematically different from the cases that are worked. For these ‘official statistics in development’ we have elected not to apply an adjustment to account for cases which have been deselected in the sample.
5. Quality management
5.1 Quality assurance
All official statistics produced by Knowledge, Analysis and Intelligence (KAI) at HMRC must meet the standards in the Code of Practice for Statistics produced by the United Kingdom Statistics Authority (UKSA) and all analysts adhere to best practice, as set out in the ‘Quality’ pillar.
Analytical quality assurance describes the arrangements and procedures put in place to ensure analytical outputs are error free and fit-for-purpose. It is an essential part of KAI’s way of working, as the complexity of our work and the speed at which we are asked to provide advice means there is a high risk of error, which can have serious consequences on KAI’s and HMRC’s reputation, decisions, and in turn, on peoples’ lives.
Every piece of analysis is unique, and as a result there is no single quality assurance (QA) checklist that contains all the QA tasks needed for every project. Nonetheless, analysts in KAI use a checklist that summarises the key QA tasks. This is used as a starting point for teams when they are considering what QA actions to undertake.
Teams amend and adapt it as they see fit, to take account of the level of risk associated with their analysis, and the different QA tasks that are relevant to the work.
Analysts and managers plan the QA tasks that need to be completed, documenting how each one is going to be carried out, then turn it into a QA checklist, which is specific to the estimate. Analysts and peers carry out the QA tasks, update the checklist, and pass it onto the Senior Responsible Officer for review and sign off.
5.2 Quality assessment
The QA for these statistics followed the framework described in the above ‘Quality assurance’ section to ensure that they meet the standards in the Code of Practice for Statistics.
The analysis has been through multiple reviews, where analysts, policy, compliance and risk stakeholders sense-checked the data, methodologies, and outputs and applied wider understanding and context to them. Any inconsistencies between the output data and the related datasets were identified and corrected during this process.
The quality assurance process ensures the models meet the standards set out in the code of practice. The specific procedures undertaken were as follows:
Stage 1 – Specifying the question
Checks that up-to-date documentation is agreed with stakeholders, setting out what outputs are needed and by when, how they will be used and all the parameters required for the analysis.
Stage 2 – Developing the methodology
Checks that any methodology is agreed and developed in collaboration with stakeholders and others with relevant expertise, ensuring it is fit for purpose and will deliver the required outputs.
Stage 3 – Building and populating a model/piece of code
Checks that analysis is produced using the most appropriate methodology, data and software, and is in line with good practice guidance.
Data inputs are checked to ensure they are fit–for–purpose, by reviewing available documentation and, where possible, through direct contact with data suppliers.
QA of the input data is carried out.
The analysis is audited by someone other than the lead analyst – checking the code and the methodology.
Stage 4 – Running and testing the model/code
The analysis was presented to policy, compliance and risk stakeholders to sense check the outputs. Results were determined to be explainable and in line with expectations.
Stage 5 – 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.
6. Relevance
6.1 User needs
This analysis is likely to be of interest to the following:
-
national government – policy makers and MPs
-
academics and research bodies
-
media
-
business community
-
general public
-
international revenue authorities and organisations
6.2 User satisfaction
The statistics are presented as ‘official statistics in development’.
HMRC are evaluating the usefulness of this analysis for users. As part of this, HMRC welcome feedback which can be provided by emailing taxgap@hmrc.gov.uk. HMRC also welcome ideas for further expansion and refinement.
HMRC consults with people via stakeholder forums and conferences, including the Chartered Institute of Taxation, the Association of Accounting Technicians, the Association of Taxation Technicians, the Association of Chartered Certified Accountants, the Institute for Fiscal Studies, Public Economics UK, the Commonwealth Association of Tax Administrators, and the Tax Administration Research Centre.
HMRC engages with academics, international institutions and foreign fiscal authorities, as well as the UK’s National Audit Office (NAO), the Office for Statistics Regulation (OSR) and the Office for National Statistics (ONS), to share methodologies and best practices to assist in the development of the tax gap.
In February 2023, HMRC presented at the Organisation for Economic Cooperation and Development (OECD) Forum on Tax Administration (FTA) Tax Gap Conference in Paris. Following the success of the OECD’s FTA Tax Gap Conference, a Community of Interest (COI) on Tax Gap was established, of which HMRC is now a member of.
The purpose of the Tax Gap COI is for tax administrations and stakeholders to share knowledge, good practice, and lessons learned on tax gap measurement to support all FTA members in their efforts of measuring the tax gap. HMRC is also a member of the Advisory Group to help with the running of the COI meetings. Meeting topics will be determined by the Advisory Group. The COI provides a forum to discuss tax gap development work such as those presented in this paper.
The OSR, the regulatory arm of the UKSA, completed a compliance check in 2019 of the extent to which HMRC’s tax gap statistics meet the standards for the Code of Practice for Statistics. The OSR praised the HMRC team, saying that it had “proven to be highly committed and engaged when working to enhance the trustworthiness, quality and value of these statistics”. They also stated that “HMRC is world-leading in measuring tax gaps and is setting the bar for others to follow”.
The International Monetary Fund (IMF) assessed the way in which the UK calculates its tax gap and concluded that HMRC produced one of the most comprehensive studies of the tax gap available internationally. They concluded that the models and methodologies used by HMRC to estimate the tax gap across taxes were sound and consistent with the general approaches used by other countries.
6.3 Completeness
These findings provide new evidence on the scale of non-compliance relating to the declaration of foreign income for those individuals identified through the AEOI data only. It therefore represents a subset of overall foreign income non-compliance.
The publication meets all requirements based on relevant legislation, regulation, and guidelines.
7. Accuracy and reliability
7.1 Overall accuracy
The analysis in this technical paper is presented as ‘official statistics in development’ - these are official statistics that are undergoing a development. They were previously called ‘experimental statistics’.
The statistics are published in line with the Code of Practice for Statistics. This code assures objectivity and integrity – providing the framework to ensure that statistics are trustworthy, good quality, valuable and give producers of official statistics the detailed practices they must commit to when producing and releasing official statistics.
We have assessed the estimate for total non-compliance using the ‘Measuring tax gaps’ publication’s uncertainty rating system. For this, we assess the uncertainty across 3 criteria: the model scope, the methodology used and the data underpinning the estimate. The estimate for non-compliance relating to foreign income from individuals notified to HMRC through the AEOI data is rated as having high uncertainty.
The potential sources of error include:
-
misreporting of an individual’s foreign income and gains through the AEOI data
-
misidentification of the known individuals owning the financial accounts notified to HMRC through AEOI
-
incorrectly calculating the discrepancy value when comparing information compiled from the AEOI data to what those individuals have declared to HMRC
-
sampling error working with the results of the random sample and the possibility these results may not reflect the characteristics of the wider population they have been sampled from
-
missing non-compliance within the random sample over the course of investigations, with no non-detection multiplier applied
-
mis-recording of the non-compliance identified
-
incorrectly forecasting the tax yield from random enquiry cases which are still ongoing
-
combining samples and assuming data from 2 different tax years are sufficiently similar
-
other key assumptions made in the analysis such as uplifting for those individuals untraced through the AEOI data and not making an adjustment for deselected cases
7.2 Sampling error
Where possible, HMRC has estimated the likely impact of sampling errors by calculating statistical confidence intervals. These give margins of error within which we would expect the true value to lie 95% of the time, if there were no systematic errors.
7.3 Non-sampling error
Coverage error
The statistics cover non-compliance relating to foreign income and gains from UK individuals identified to HMRC through the AEOI agreements. There is a risk that information reported to HMRC through AEOI may not capture all relevant income. HMRC then conducts data matching from this information to trace known individuals. Only traced individuals can be used as the target population for this activity. An uplift is applied in the analysis to account for untraced individuals in the outputs.
Information from the AEOI data is compared to taxpayer returns to identify those individuals who potentially may be non-compliant. Individuals with no discrepancy are assumed to be compliant. This process step may lead to non-compliance not being captured.
A random sample is selected from those individuals identified as having a discrepancy. A stratified sampling approach is used, based on the size of discrepancy between the information from AEOI data and the Self Assessment tax return data.
Measurement error
Cases selected for the random enquiry programme are first subject to a ‘one-to-many’ approach to encourage voluntary compliance. Cases where there is still an outstanding or unexplained discrepancy following this stage proceed to a compliance check.
Non-compliance can be difficult to identify. It is therefore unlikely that the random enquiry programme will identify all incorrect returns or the full scale of under-declaration of tax liabilities. No adjustment has been made to the results to account for this and the full scale of non-compliance could be higher than the estimates shown.
There is a risk that results for the random enquiries may be misreported. Values have been validated by reviewing individual case notes.
Non-response error
No adjustments have been made for cases selected for the random enquiry programme which could not be worked and were deselected. An assumption is made that deselected cases are not materially different to selected cases.
Processing error
It is possible that errors exist in the programming code used to analyse the data and produce statistics.
7.4 Data revision
Policy
The UKSA Code of Practice for Statistics requires all producers of official statistics to publish transparent guidance on the policy for revisions.
Practice
These estimates are likely to be subject to revision in any future iteration as the methodology is refined and through any improvements to the data available. A key contributor to the likelihood of revisions is current ongoing cases in the random enquiry programme closing for different amounts to what has been forecasted. Any data revisions will be detailed in future publications.
8. Timeliness and punctuality
8.1 Timeliness
The estimate for total under-declared liability is for the 2018 to 2019 tax year, estimated using the results of samples for the 2017 to 2018 tax year from the 2017 AEOI data and the 2018 to 2019 tax year from the 2018 AEOI data. This time lag is a result of the filing timeline for Self Assessment tax returns. It can then take several years for a sufficient number of cases in the random enquiry programme to be worked through to completion.
8.2 Punctuality
In ‘Measuring tax gaps’ 2022 and 2023, HMRC announced that it would be publishing an estimate of the tax gap arising from non-compliance by UK residents failing to declare their foreign income where this had been identified through data received from other fiscal authorities through Automatic Exchange of Information. In line with guidance for Civil Servants during an election period, HMRC’s Head of Profession for Statistics agreed that this additional breakdown of the tax gap should not be released within the election period alongside ‘Measuring tax gaps 2024 edition’. This paper is now being published.
9. Coherence and comparability
9.1 Geographical comparability
The geographic regions covered by the data are the UK and all jurisdictions participating in the AEOI agreements. All individuals under consideration are UK resident for tax purposes. See the Exchanges of information under the AEOI Standard for a full list of jurisdictions undertaking exchanges through AEOI.
9.2 Comparability over time
This is the first time HMRC has published statistics relating to foreign income non-compliance – consequently, it is not possible to compare estimates over time.
9.3 Cross domain coherence
There are no cross-domain coherence issues, since all terminology is standardised, is defined in the publication, and is consistent with publications such as ‘Measuring tax gaps 2024 edition’.
9.4 Internal coherence
The HMRC data used within the analytical model is coherent as it aligns to HMRC-wide definitions and terminology.
10. Accessibility and clarity
10.1 Publication
The ‘Undisclosed foreign income in Self Assessment by UK residents, 2018 to 2019’ paper is published in an HTML format on GOV.UK. GOV.UK writing style is followed to deliver the information clearly to the users.
10.2 Quality documentation
All official statistics produced by KAI must meet the standards in the Code of Practice for Statistics produced by the UKSA 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.
11. Cost and respondent burden
This activity has been designed to limit additional costs to HMRC and to function within existing processes. The direct HMRC staff cost is about £0.5 million per year. The enquiries also generate tax revenue and help HMRC to evaluate and improve the effectiveness of risk assessment processes. However, there is an opportunity cost for this allocation of resource.
Key assumptions, such as treating as compliant those individuals identified as having no discrepancy from the data matching stage, reduce the burden on taxpayers and HMRC compliance officers. The initial ‘one-to-many’ stage provides individuals with the opportunity to voluntarily inform HMRC of under-declared tax liability with limited involvement from HMRC compliance officers.
12. Confidentiality
12.1 Confidentiality policy
The statistics contained within the publication are presented at aggregate levels. As a result, issues of confidentiality relating to individuals’ or tax entities’ details are not applicable.
Production and dissemination follows the Code of Practice for Statistics and HMRC’s published policy on Confidentiality and Access.
A list of individuals who have pre-release access to these statistics is published on GOV.UK. The list is updated each year, maintaining the principles and rules set out in the pre-release access to official statistics.
12.2 Confidential data treatment
The statistics in these tables are presented at a highly aggregated level, to minimise the possibility of any individual person or organisation being identified.
If a potential risk did exist, statistical disclosure control (SDC) would be applied. SDC is the application of methods to ensure confidential data is not disclosed to parties who don’t have authority to access it.
SDC modifies data, so that the risk of data subjects being identified is within acceptable limits, while making the data as useful as possible.
Disclosure in this analysis is avoided by applying rules that prevent categories of data containing small numbers of contributors, and small numbers of contributors that are very dominant.
If a cell within a table is determined to disclose confidential data, its contents would be suppressed either by removing the data or combining categories.
Further information on anonymisation and data confidentiality best practice can be found on the Government Statistical Service’s website.