Corporate report

HMRC Statistics: Raising our performance

Published 15 September 2026

1. Introduction to the review

HM Revenue and Customs (HMRC) is one of the largest employers of professional analysts in government and has a strong reputation for delivering high-quality policy and operational analysis, research, and official statistics.

HMRC statistics are widely used by government departments, businesses, researchers, Parliament and the public. They affect important economic indicators and decision-making, and so confidence in the quality and transparency of HMRC statistics essential.

Due to the complexity of HMRC’s data landscape, there is an ongoing risk of errors which can affect official statistics produced both within HMRC and by other government departments (OGDs).

The purpose of this review is to identify and mitigate the risk of errors in HMRC’s official statistics, focusing on:

• HMRC official statistics that feed into market-sensitive official statistics produced by the Office for National Statistics (ONS) and HM Treasury (HMT)

• progress made on the recommendations from the Office for Statistics Regulation’s (OSR) Strengthening the quality of HMRC official statistics report

The review examines the root causes of recent errors, current processes and governance, and provides recommendations to reduce the likelihood of future issues. Its overarching aim is to strengthen trust, quality and confidence in HMRC’s official statistics by improving the systems, processes, governance and culture that underpin them.

1.1 Context

In 2020, the OSR conducted a review of HMRC’s official statistics (referred to in this report as the 2020 OSR review) and published a comprehensive report, Strengthening the quality of HMRC official statistics. HMRC has since improved its quality assurance processes and practices in line with the report’s recommendations.

However, in autumn 2025, there were 2 major errors in HMRC’s official statistics which affected downstream market-sensitive publications produced by ONS and HMT. These errors prompted a review of HMRC’s statistics affecting market-sensitive official statistics, building on the findings of the 2020 OSR review.

The 2 errors are summarised below:

VAT receipts error

In September 2025, there was a VAT receipts error leading to an upward revision of VAT cash receipts by £2.4 billion (3.1%) from April 2025 to August 2025, affecting the joint ONS and HMT Public Sector Finances publication.

Trade in goods statistics error

In October 2025, there was a trade in goods statistics error leading to an upward revision of UK exports by £5 billion (1.3%) from March to December 2024 and £6 billion (1.8%) from January to October 2025, affecting the ONS UK Trade publication and GDP estimate.

Annexes 1 and 2 respectively contain further details on these 2 errors.

1.2 Scope

This review is focused on HMRC statistics that feed directly into market-sensitive publications from OGDs, as these were considered the highest priority to strengthen assurance on. It is expected that lessons learned can be applied to other statistical products and analysis.

The 2 statistical products in scope are:

• HMRC tax receipts and National Insurance contributions for the UK, which is published monthly alongside the joint ONS and HMT Public Sector Finances publication

• UK trade in goods statistics, which feed into the ONS UK Trade, Balance of Payments, and GDP quarterly estimate publications

Whilst not in scope of this review, there have also been errors stemming from direct data feeds from HMRC systems which feed into ONS official statistics. Further reference to this issue is made in section 7.2 on error reporting.

1.3 Our approach

The review was led by HMRC’s Chief Statistician, Sean Whellams. We considered:

• the extent to which HMRC have met the recommendations set out in the 2020 OSR review

• opportunities to further reduce the risk of errors in HMRC’s statistical outputs

To explore these areas and develop recommendations, we:

• conducted interviews with a wide range of stakeholders including statistics producers and OGD stakeholders (Annex 3 lists the meetings undertaken during this review)

• held workshops with teams to reflect on the implementation of the recommendations set out in the 2020 OSR review

• reviewed previous errors and issues to identify root causes and cross-cutting themes

• considered with key stakeholders whether technical methodology changes would reduce the risk of errors in the tax receipts publication

1.4 Structure of report

Key findings and assessment of progress

This report begins by providing an executive summary and high-level overview of key findings and recommendations (sections 2 and 3). It then reviews the recommendations from the 2020 OSR review, assessing HMRC’s progress and identifying any remaining gaps (section 4).

Cross-cutting themes

Following this, it explores and provides recommendations for key cross-cutting themes that emerged during the review that are more broadly applicable to HMRC’s statistics: HMRC’s data environment (section 5), quality assurance and governance (section 6) and the error-reporting process (section 7).

Publication-specific analysis

Finally, it sets out publication-specific findings and recommendations for the ‘HMRC tax receipts and National Insurance contributions for the UK’ publication (section 8) and ‘UK trade in goods statistics’ (section 9).

1.5 Next steps

HMRC have agreed with the OSR to publicly report on progress made on the recommendations from this review. These updates will be published 2 to 3 times per year and set out actions taken to address the issues covered in this report. HMRC will continue to work closely with the OSR and welcome feedback or challenge to support our aim of reducing the risk of errors in our official statistics. Progress reporting will focus not only on the completion of actions, but on how these actions contribute to sustained improvements in the trustworthiness, quality and value of HMRC’s statistics.

2. Executive summary

In this report, HMRC have identified areas for improvement to mitigate risks and issues affecting its official statistics. These statistics inform major fiscal and economic decisions across government and are relied upon by a broad community of users, both internal and external to government.

The review focused on HMRC official statistics that feed into other market-sensitive official statistics and provides an assessment of progress following the 2020 OSR review.

The report finds that HMRC has made good progress since the 2020 OSR review.

HMRC adopted the OSR’s recommendations with the support of senior leaders, and HMRC’s Internal Audit (IA) function provided additional confidence in the progress on implementing recommendations.

However, some processes could be improved further to raise our performance.

The recommendations set out in this report form a programme of improvement designed to embed quality throughout the production process and strengthen confidence in HMRC’s statistical outputs. While many recommendations focus on process and systems, successful implementation will depend equally on leadership behaviours, effective challenge, analytical curiosity and clear accountability.

Table 1 provides a high-level summary of the key drivers of error identified by this review, the recommended actions to address them, and the intended outcomes.

Table 1: Addressing the key drivers of error in HMRC’s official statistics

Key drivers/findings	Recommendations	Intended outcomes
Understanding upstream data and system changes	Stronger engagement and change management	Earlier identification and management of risk
Inconsistent application of quality assurance (QA) processes and analytical challenge	Standardised QA processes and stronger analytical curiosity	Reduced likelihood of undetected errors
Governance and capability risks	Accountability and succession planning	Clear accountability and greater organisational resilience
Infrastructure constraints	Infrastructure risk management	More resilient statistical production processes
Communication challenges	Improved error reporting and engagement	Greater trust and transparency
Limited strategic profile of official statistics	Raise the profile of statistics within HMRC	Greater organisational recognition and prioritisation of official statistics
Methodological limitations	Methodological improvements	Improved quality and value of statistics

Collectively, these actions are intended to strengthen the trustworthiness, quality and value of HMRC’s official statistics, while increasing confidence among users and stakeholders. More detail on the key findings and recommendations can be found in section 3.

This report has 11 overarching findings and recommendations to strengthen the quality of HMRC’s official statistics (listed in Annex 4). 

These recommendations fall under the 7 themes listed below and are designed to be proportionate and practical, building on existing good practice. They have been agreed with the involvement of key stakeholders from OGDs. Delivery will be supported and driven by HMRC’s analytical senior leadership team, who will play a critical role in championing and embedding these improvements. The recommendations should also be applicable to other statistical products produced by HMRC and OGDs.

The 7 key themes for improvement that have been identified are:

1. Visibility and understanding of upstream data and processing changes

If changes to operational systems are not sufficiently understood by statistical producers, this can lead to misinterpretation or omission of data. Strengthening early and ongoing engagement between statistical producers, data suppliers, and transformation programmes is essential to ensure the impacts on statistics are fully understood and managed in advance.  

2. Quality assurance and analytical curiosity 

Quality assurance is strong in most areas but not consistently applied across all products. Valuable QA tools, process maps and checklists exist, but their use and documentation varies. Alongside this, strengthening analytical curiosity and learning systematically from near misses would reduce the likelihood of future issues.  

3. Governance and capability

Governance arrangements are generally good, however, risks increase where publications rely on multiple teams or a small number of key experts. Clearer accountability and succession planning for critical statistical roles would strengthen resilience, reduce dependency on individuals, and support continuity in statistical production.

4. Data environment and IT infrastructure 

HMRC’s data environment and IT infrastructure underpinning official statistics pose ongoing risks. The aging and complex nature of some of these systems increases vulnerability to disruption. While HMRC’s risk management processes provide a solid foundation, more explicit identification and monitoring of infrastructure dependencies at the statistical product level and greater visibility of IT transformation timelines are needed to help to further mitigate these risks.    

5. Communications 

Communication across HMRC and between OGDs has improved and in many areas is very effective. However, there would be benefit to establishing a clearer process for communicating errors that have an impact on OGD market-sensitive official statistics.

6. Raising the profile of official statistics within HMRC 

OGD stakeholders have suggested that HMRC’s official statistics are not sufficiently recognised as a strategic priority. Further work is required to raise the profile of these statistics within the department.

7. Methodological improvements

Some methodological changes are recommended for the tax receipts publication and UK trade in goods statistics.

3. Key findings and recommendations

Tables 2 to 8 below summarise the key findings and recommendations of this review. More detail about the findings and recommendations can be found in sections 5 to 9 of the report and Annex 4 lists all recommendations.

Table 2: Visibility and understanding of upstream data and processing changes 

Finding	Recommendation (summary)	Intended outcomes
1. Statistical producers are not always sufficiently engaged in, and therefore informed about or able to manage, planned changes to the source data or systems. Without a clear understanding of these upstream changes and effective management of any impact on statistics, there is a risk of misinterpreting new or altered data, leading to errors.	Frequent, ongoing engagement between analysts and data suppliers is vital to ensure upstream changes are more fully understood by statistics producers and that the statistical consequences of change are better managed (recommendation 5.1).	Upstream changes are identified, understood and managed earlier, reducing the risk of statistics being affected by data or system changes.

Table 3: Quality assurance and analytical curiosity

Finding	Recommendation (summary)	Intended outcomes
2. QA tools have been developed, and all statistical producer teams undertake a range of QA steps. However, we found that QA activities and the recording of near misses are not always applied or documented in a consistent way, and updates to process maps are inconsistent.	QA tools should be mandated and process maps reviewed, with auditing from Lead Statisticians and the Statistics Head of Profession. Lessons learnt from near misses should be shared to ensure more regular reporting (recommendations 6.1, 6.2 and 6.4).	More consistent and auditable quality assurance, with risks identified and addressed before they affect published statistics.
3. The implementation of Reproducible Analytical Pipelines (RAPs) in the production process has led to more automation of QA checks. While this reduces the risk of some errors, it can lead to less value being placed on analytical curiosity.	Analytical leaders need to champion analytical curiosity as part of the QA audit process. Evidence of analytical curiosity should be routinely demonstrated by the producer teams and those more senior undertaking the QA. (recommendations 6.2 and 6.3)	Cultural shift leading to increased analytical curiosity and therefore earlier identification and investigation of anomalies, reducing the likelihood of undetected errors.

Table 4: Governance and capability

Finding	Recommendation (summary)	Intended outcomes
4. There are clear owners for many statistical products in HMRC, especially where they focus on a single tax. However, in publications with contributions from many different teams, there is a greater risk of a lack of clarity in governance and ownership.	Statistical publications which rely on input from multiple teams should develop a responsibility assignment matrix to define responsibilities of various stakeholders (recommendation 6.5).	Clear accountability and ownership for statistical processes, data quality issues and decision-making.
5. Statistics are produced within teams where knowledge on subject matter and process can be limited to a very small number of people, meaning there is a risk if a few key individuals move on at the same time.	HMRC’s analytical senior leadership teams should identify key roles for the production of HMRC’s statistics and (i) identify the level of risk if the role is vacant and (ii) implement replacement strategies for these roles (recommendation 6.6).	Reduced dependency on individual experts and business continuity can be effectively maintained.

Table 5: Data environment and IT infrastructure 

Finding	Recommendation (summary)	Intended outcomes
6. Knowledge, Analysis and Intelligence (KAI) directorate’s analytical data environments are aging and at capacity, increasing the risk of data loss and delays to analytical outputs including statistical publications. Senior analysts are engaging with HMRC’s major IT change programmes as part of HMRC’s transformation roadmap and are escalating risks and issues as part of HMRC’s risk process1.	All official statistics should map their reliance on infrastructure as part of process mapping and identify mitigations. The timescales for delivery of improved analytical infrastructure by HMRC IT change programmes should be reported by these programmes as part of ongoing reports following this review (recommendations 5.2i and 5.2ii).	Improved management of infrastructure risks and greater resilience of statistical production processes.

Table 6: Communications  

Finding	Recommendation (summary)	Intended outcomes
7. External communications of errors have been produced on a consistent timetable and with agreed messaging between departments, however, informal communications between HMRC and OGDs could be improved.	A workshop should be held with key producers and Statistics Heads of Profession from HMRC, ONS, HMT and the OSR with a view to capturing an agreed process on communicating errors (recommendation 7.3).	More timely, consistent and transparent communication of errors across Departments.
8. Greater clarity is needed in differentiating between errors and routine revisions in quality reports and the reporting of changes in figures.	The quality reports for HMRC’s tax receipts publication and UK trade in goods statistics should be clear on how errors and revisions are defined and clarified where necessary (recommendation 7.4).	Improved user understanding of changes to statistics and greater transparency in reporting.

Table 7: Raising the profile of official statistics within HMRC 

Finding	Recommendation (summary)	Intended outcomes
9. While producing quality statistics falls under HMRC’s strategic objective of supporting wider government economic aims, more could be done to raise the profile of statistics within the department.	A project on how HMRC can develop a way to further raise the profile of its statistics within the department should be undertaken (recommendation 6.7).	Greater organisational recognition of the importance of official statistics, leading to stronger support, prioritisation and accountability.

Table 8: Methodological improvements 

Finding	Recommendation (summary)	Intended outcomes
10. For the tax receipts publication, some methodological improvements have been identified.	The tax receipts publication should align the reporting methodology of the main taxes to be consistent with accounting data on a monthly basis (recommendation 8.3).	Reduced risk of missing or misclassified data and improved consistency between statistical and accounting measures.
11. For the Trade Statistics publication, some methodological improvements have been identified.	Trade Statistics should expand QA processes to encompass more top-down QA and time series analysis. They should introduce subject matter experts and implement stronger governance practices around changes to Customs Procedures Codes (recommendations 9.3i, 9.3ii and 9.3iii).	Improved detection of data quality issues and greater confidence in the accuracy of trade statistics.

4. The 2020 OSR report ‘Strengthening the quality of HMRC official statistics’

In April 2020, the OSR conducted a review of HMRC’s official statistics (referred to in this report as the 2020 OSR review) and published its findings in a comprehensive report Strengthening the quality of HMRC official statistics. The report focused on the processes and principles surrounding HMRC’s production of official statistics, with the aim of further strengthening their quality through a set of recommendations.

4.1 Key findings and recommendations from the 2020 OSR review

There were 9 key findings and recommendations from the 2020 OSR review, which HMRC accepted in full. A summary of the OSR’s recommendations is set out below, alongside actions HMRC have taken to implement these recommendations.

While the scope of the 2020 OSR review was on the official statistics within HMRC’s KAI directorate, many of the recommendations were also applicable to statistics producer teams outside of KAI.

2020 OSR recommendation 1

Senior leaders in KAI should champion and support developments and innovations that can enhance the quality of official statistics.

Action taken

Senior leaders secured Spending Review funding to establish a new Deputy Director (DD)-led team to improve data quality and assurance in KAI.

The new team:

• produces and maintains QA guidance and training

• maintains the register of Business Critical Models, including model level assurance

• maintains a network of Quality Champions across HMRC who keep up to date with key QA developments and promote them across their areas

2020 OSR recommendation 2

Knowledge sharing should be prioritised at all levels of KAI, and more-effective mechanisms built to support this. As part of this, a cross-KAI quality network could be put in place, with senior support.

Action taken

The new DD-led team’s role includes championing and supporting knowledge-sharing and enhancements to the quality of data and statistics.

Mechanisms built to support this include: 

• a cross-HMRC Quality Champions network that shares best practice 

• creation and rollout of a Knowledge Management Framework, supported by Knowledge Management Champions — a data catalogue has been developed and shared which will support with knowledge management of its datasets  

2020 OSR recommendation 3

In some areas, there is the need for closer working with teams providing data to KAI. The responsibility for establishing and maintaining effective relationships should be shared between KAI and its partners. 

Action taken

Whilst improvements have been made, this has been an ongoing challenge which has led to some errors. Further recommendations to improve relationships are made in section 5.

2020 OSR recommendation 4

Each team should produce and maintain a process map, which shows how its statistics are produced, from data collection to publication. This map should highlight risk points and mitigating actions taken to ensure the quality of the statistics.

Action taken

Process maps were created for all HMRC’s statistical products and are reviewed and updated as required. These are used as a framework to undertake retrospective analysis of the process to produce the statistics.

When the process maps were updated in 2024, several improvements were put in place to support data quality. This included a new template for process maps that outlined key QA checks all statistical owners needed to put in place (or justify why they were not applicable) based on issues identified in previous analyses, this has improved the situation, but there are still minor inconsistencies.

2020 OSR recommendation 5

There should be a more consistent approach to, and clearer expectations for, QA across KAI. QA should be documented in each production round and available for audit.   

Action taken

The new DD-led team actively promotes use of an agreed QA framework and utilises the Quality Champion network to strengthen understanding and increase consistency of the quality assurance process.

Regular and tailored training has improved awareness of the QA framework, how to apply it, and how to support others, and many have reported that the proportionate QA framework is beneficial to better understand QA processes.

QA is undertaken in all areas though some challenges remain. These are covered further in section 6.

2020 OSR recommendation 6

Drawing on its relationship with the HMRC Chief Data Information Officer (CDIO) Data Exploitation team, KAI should move towards greater use of data science methods and tools.

Action taken

KAI undertook a large programme of work to use new technology and adopt RAPs for the vast majority of statistical products. These RAPs are mostly in the programming language R.

2020 OSR recommendation 7

HMRC should review the published information about the quality of its official statistics and update to ensure that it fully explains how HMRC producers assure themselves that their statistics are accurate, reliable, coherent and timely.

Action taken

HMRC now publishes background quality reports for all of its official statistics publications. These inform users about methodologies and the strengths and limitations of the statistics.

2020 OSR recommendation 8

HMRC should publish information in a transparent way that is easy to find.

Action taken

HMRC updated its GOV.UK pages for statistics, to improve the availability and accessibility of information on its statistics, drawing on best practice from other government departments. HMRC keeps users informed through a statistics announcements page.

2020 OSR recommendation 9

Each area of KAI should review its current publications, with a view to reducing the number or size of publications, to ensure analytical resource is being used effectively. 

Action taken

HMRC has conducted a consultation with users of statistics every 2 years, leading to reductions in number and frequency of publications, while still meeting user needs.

Implementation of recommendations

These recommendations were implemented and overseen by HMRC’s Audit and Risk Committee (ARC). An IA review in 2022 concluded that KAI were effectively monitoring the implementation of OSR recommendations, and there were no areas of particular concern. IA identified 2 areas where there was an opportunity to improve control:

• utilising process maps to assess the strength of controls in place around data quality

• clearly defining the role of KAI’s analytical data management function

KAI accepted and implemented IA’s recommendations.

4.2 Reflecting on the 2020 OSR review scope and remaining gaps

The ARC was satisfied that HMRC implemented all the recommendations as detailed from the 2020 OSR review. While there has been good progress on all the recommendations (as detailed in section 4.1 above), a small number of significant challenges remain, namely:

• analysts not being fully informed of, or engaged in, changes in the source data or HMRC processes which can impact the data (discussed further in section 5)

• analysts not consistently applying the QA processes which are available, nor being sufficiently curious (discussed further in section 6)

If the 2020 OSR review had gaps, then it seems that there were 2 of potential significance, though both were outside of the scope of the review at the time:

• while the review covered data science enablers such as RAPs in recommendation 6, the broad health of HMRC’s analytical IT environments was out of scope and continues to pose challenges, as detailed in section 5

• while Trade Statistics were not in scope of the OSR review, lessons learned and best practice was shared and adopted by the Borders and Trade team

5. HMRC’s data environment and IT infrastructure

HMRC has a large and complex data estate containing rich administrative information on the finances of individuals and businesses. These data underpin both operational work and a wide range of analytical outputs, including official statistics, policy analysis, and forecasting. 

To help manage this complexity and improve knowledge sharing, following a recommendation in the 2020 OSR review, KAI’s analytical data management function was created. This function has built up capacity to tackle the complexity and scale of change and ensure that change programmes fully address analytical needs.

The source data are typically populated through individuals, businesses or intermediaries filing data into “front-end” tax systems. Analytical users typically access copies of the source data through regular case-level extracts, usually extracted daily or monthly, rather than data being analysed directly on the systems. These data extracts are essential where operational systems do not permit analysis on their platforms or where analysis could compromise system performance impacting the Department’s main functions of assessing and collecting tax. However, this introduces 2 significant risks to the production of statistics:  

1. Limited visibility of upstream system and data changes – which has contributed to recent errors in HMRC’s official statistics.

2. Aging analytical infrastructure – which has not led to errors in HMRC’s official statistics to date but has been identified as a key risk going forward.

5.1 Limited visibility of upstream system and data changes

Both receipts and trade statistics have been affected by instances where changes in upstream data were not fully understood or not properly reflected in statistics production processes by analytical teams. Statistical producers have not always sufficiently engaged in, and therefore been informed about, planned changes to the source data or systems. Without a clear understanding of these upstream changes and effective management of any impact on statistics, there is a risk of misinterpreting new or altered data, leading to errors.

Strengthening engagement between data suppliers and statistical producers is essential to reducing this risk. Statistics producers need support from IT suppliers and data owners to ensure that they are not only informed of changes which impact their statistics, but are also more fully aware of the implication of upstream changes and their wider impact on the UK economy. This engagement needs to happen early enough for statistics producers to be able to shape change and ensure any consequences are well managed. This issue also links to section 6.5 on governance.

Recommendation 5.1

Frequent ongoing engagement between analysts and data suppliers will help ensure upstream changes are more fully understood by statistics producers and that the statistical consequences of change are better managed. This engagement should happen at different levels:

• there should be frequent engagement (for example, monthly) at working level between statistics producers and data suppliers

• there should be regular engagement (for example, biannual) at a senior level between senior analysts and data suppliers and where appropriate senior leaders should be joined up with an operational data expert “buddy” to support understanding

• the data management and IT support functions within analytical areas should proactively influence change programme plans to ensure that analysts are impacted by these changes and analytical needs are met

Some of this engagement could include OGD stakeholders where appropriate.

5.2 Aging analytical infrastructure

HMRC’s analytical data environments are aging and at capacity, increasing the risk of data loss and delays to analytical outputs including statistical publications.  Recent upgrades to the Trade Statistics environments have reduced the risk to their products.  However, the risk remains relevant for KAI’s statistical and analytical products.

While this has not caused an error in HMRC statistics to date, it was identified as part of this review as a risk for future errors. To mitigate this, KAI currently monitor and track an overarching risk for the data environment and report on this monthly through HMRC’s risk reporting process.

HMRC are investing in updating and securing these environments as part of HMRC’s IT transformation programme. This will create more robust and modern tools and systems for analysts to use for statistics production and other analysis. It will still be important to carry out an appropriate level of QA alongside the use of improved tools, especially during transition periods where unexpected issues might occur.

For some statistics, it is now possible to start transitioning to new tools and systems to reduce some of the risks associated with aging infrastructure. However, not all statistical processes can, or are, currently moving to new systems, and so there remains both a short-term risk for these statistics and also a medium-term risk that some elements of these IT transformation programmes are descoped or not fully delivered to required timescales. Clearer timeframes should be sought from and provided by transformation programmes, and where relevant to statistical production should be included in the reporting requirements following this review.

An example of a risk, and the work to mitigate it, is given below:

• there is a risk that the Real Time Information (RTI) server could reach or exceed its storage and processing capacity, delaying key economic data, leading to reputational impacts for HMRC and wider government

• to address this, a formal risk was raised on HMRC’s risk system, with a mitigation plan put in place

• the plan includes exploring moving the process onto new cloud platforms, with support from KAI’s technical expert teams

This approach is considered in HMRC as best practice for identifying, capturing and planning mitigations for a specific data infrastructure risk related to a statistical process. This approach of raising specific risks could be taken for other statistics, though the resource implications would need to be considered carefully, as moving processes can be resource intensive for analysts.

KAI have established links across IT transformation programmes, and the end-to-end change lifecycle within HMRC. This ensures that KAI is aware of changes and risks associated with changes to the data underlying statistics. However, changes can be complex, technical and carried out over a long timescale, so continued focus, prioritisation and engagement from all parties is needed to ensure risks are understood and controlled.

Recommendation 5.2i

All official statistics should map their reliance on infrastructure as part of process mapping. The dependencies should then be recorded as risks to the process and mitigations identified including short-term remedies (such as partial migration of the process where possible) and long-term remedies (such as transformation of the infrastructure through the IT transformation programme).

Recommendation 5.2ii

The timescales for delivery of improved analytical infrastructure by IT transformation programmes should be reported by these programmes as part of ongoing reports following this review. 

6. Quality assurance and governance

Robust and consistent QA is essential in maintaining the integrity of HMRC’s official statistics. Most of HMRC’s statistics are based on population data extracts and therefore do not require complex sampling methodology. Each statistical product has an accompanying published quality report to explain the methodology, strengths and weaknesses for users.

There are various QA tools which have been developed, and all statistics producers undertake a range of QA steps. However, the review found that these activities are not always used, applied or documented in a consistent way across statistical products. Strengthening governance, improving documentation and embedding analytical curiosity is important in reducing the risk of errors in HMRC’s official statistics.

6.1 Data process maps

Data process maps were identified in the 2020 OSR review as a powerful tool for visualising data flows and identifying structural weaknesses in the process or system. Comprehensive and updated process maps are essential for identifying risks and mitigating them.  

All HMRC statistical products have developed and produced process maps as a result of the 2020 OSR review; however, these are not consistent across all taxes, and updates need to be regularised.

Recommendation 6.1

Each key statistical product should review their data process maps and ensure they are up to date. This should then be reported back to the relevant Lead Statistician.

6.2 Standardising and documenting QA processes

Standard QA checks including investigating outliers should be undertaken for every publication, and many QA checks are now automated enabling previous issues to automatically be checked in future production processes. Evidence of QA checks should be recorded, reviewed and signed off for every publication, including:

• automated checks (which should be regularly reviewed for suitability)

• graphical plots

• any specific issues identified and action taken

A central QA checklist has been created and made available, but it is not consistently used, in part because there is no mandate or audit of its use. An outcome of this review is to make the QA checklist mandatory for all statistics producing teams, and regular auditing of a random sample of QA checklists from the Lead Statisticians and Statistics Head of Profession will help assure that they are more consistently used and recorded.

Recommendation 6.2

Consistently document QA checks including automated QA with regular auditing from Lead Statisticians and the Head of Profession for Statistics.

6.3 Analytical curiosity

Automation of QA is efficient and effective and has been observed to be widespread in the statistics production teams. However, it must complement, not replace, professional scrutiny. A strong culture of analytical curiosity is essential to identify anomalies and potential errors. Encouraging analysts to question unusual movements in data and share investigative approaches helps to normalise curiosity, strengthen capability, and support early identification of issues.

Recommendation 6.3

HMRC’s analytical Senior Leadership Team need to champion and strengthen analytical curiosity, drawing on the support of forums including the Quality Champions network and Learning network. Additionally, as part of the assurance/audit process which we are recommending, evidence of this activity should be captured.

6.4 Learning from near misses

Teams regularly experience near misses from a range of issues including manual entry errors and analytical errors. Though these are resolved before materialising as an error in a publication, they can still provide valuable insight into vulnerabilities in the systems and processes. A structured approach to logging near misses has been developed but is not consistently used. It is important to encourage more consistent use of this tool to support continuous improvement and help prevent future errors of a similar nature.

Recommendation 6.4

Develop a way to share and learn from near misses (for example, through a regular bulletin) to ensure they are more consistently used.

6.5 Governance

There are clear owners for many statistical products in HMRC, especially where they focus on a single tax. However, in publications with contributions from many different teams, such as the ‘HMRC tax receipts and National Insurance Contributions in the UK’ publication, there is a greater risk of a lack of clarity in governance and ownership.

For example, the tax receipts publication covers many different taxes. If there is an upstream issue with the quality of the data in the Corporation Tax part of the publication, responsibility for addressing the issue may be unclear. Potential owners could include:

• the co-ordination team in KAI who produce the statistics

• the Corporation Tax team in KAI as the analytical experts for that tax

• the accountants in Risk Control and Financial Accounting (RCFA) who own the method used

• the data owner or supplier

There would be merit in producing a responsibility assignment matrix to map the roles and responsibilities of statistical producers, input teams and data suppliers for publications that rely on the contributions of many different teams. A responsibility assignment matrix, also known as a RACI, defines who is Responsible, Accountable, Consulted, or Informed for each project deliverable or decision. This ensures clarity of roles and prevents accountability bottlenecks.

Recommendation 6.5

Statistical publications which rely on input from multiple teams should develop a responsibility assignment matrix to define responsibilities of various stakeholders.

6.6 Producer risk

Statistics are produced within teams where knowledge on subject matter and process can be limited to a very small number of people. This means that there is a risk that if a few key individuals move on at the same time, then the process(es) and quality could be compromised.

This risk could be reduced by:

• ensuring clearer documentation and process maps

• wider involvement of team members

• shared learning between colleagues such as talking through unusual data

• putting replacement strategies in place for key roles

Recommendation 6.6

HMRC’s analytical senior leadership teams should identify key roles for the production of HMRC’s statistics and (i) identify the level of risk if the role is vacant and (ii) implement replacement strategies for these roles.

6.7 Strengthening the profile of HMRC statistics

HMRC senior statistical producers and OGD stakeholders have identified that there is not an HMRC-level priority sufficiently articulated, or a mechanism which can be pointed to, as to why these statistics are important to get right.

One of HMRC’s strategic objectives is to support wider government economic aims, which includes providing data and analysis to OGDs and improving the use of HMRC data in official statistics. However, further work is required to raise the profile of statistics within the department to help foster a culture where statistical quality, challenge and accountability are recognised as a shared responsibility across HMRC.

Recommendation 6.7

HMRC should undertake a project driven by the support of analytical senior leaders to develop ways to raise the profile of its statistics, including clarifying their strategic importance and establishing mechanisms to support their prioritisation across the department.

7. Official statistics issues and error reporting

HMRC is a large producer of official statistics with around 130 publications per year. The vast majority of HMRC’s official statistics are produced regularly without significant issues or errors; there have been 6 major errors in around 170 tax receipts and trade in goods publications since 2020. Most of these errors were caused by changes to the underlying data source(s) not captured in the statistics or calculation errors. Further detail on the nature and scale of the errors within the scope of this review is provided in section 7.1. 

When errors occur in official statistics it is essential that communication is timely and transparent internally, to OGDs, and externally. The 2 recent errors in the tax receipts and trade statistics publications identified some areas for improvement in communication arrangements, including of the need for greater clarity around escalation and engagement with stakeholders. A consistent and agreed approach is required for reporting errors to ensure transparency and confidence in the handling of future issues.

This section explores the error handling process including reporting and communication of errors, focusing on the publications within the scope of this review to form recommendations.

7.1 Recent errors in the tax receipts and trade in goods publications

Since the OSR report was published in April 2020, HMRC have published around 800 publications, of which over 70 were the monthly HMRC tax receipts and National Insurance contributions (NICs) for the UK and over 100 were on UK trade in goods statistics. In that time, there have been 4 major errors in the ‘HMRC tax receipts and NICs’ publication and 2 major errors in the UK trade in goods statistics. The nature and scale of the errors within the scope of this review (those reported in 2025) are outlined below, with further details of all errors since 2020 in Annex 5.

HMRC tax receipts and NICs in the UK

In September 2025, there was an issue with the way VAT cash receipts had been recorded, impacting provisional cash receipts for April to August 2025. The error was the result of a change in financial processing leading to some VAT payments not being included in the tax receipts publication. This led to an upward revision of cash receipts by £2.4 billion from April to August (resulting in a 3% increase in VAT receipts).

UK trade in goods statistics

In October 2025, a new Customs Procedure Code (CPC) introduced in March 2024 had not been included in the Trade Statistics production system, and therefore trade movements associated with the CPC were wrongly excluded from the ‘UK overseas trade in goods statistics’ (OTS). The associated revision increased UK goods exports by around 1.3% (£5 billion) for the period January to December 2024, and 1.8% (£6 billion) for the period January to October 2025. 

7.2 Error reporting

Errors in HMRC’s official statistics are investigated and documented in line with the Code of Practice for Statistics, and the Head of Profession for Statistics (HoP) supports and ultimately approves the handling of errors. Errors are handled differently depending on whether they are classified as major or minor errors. As each case is unique, the HoP must exercise judgement in making this classification. Factors to consider when determining whether an error is major or minor include (but are not limited to): 

• the scale of the error

• the impact of the error and whether it affects headline figures

• whether the error relates to recent time points

• whether the statistics feed into other statistics or National Accounts

For minor errors, a light-touch form describing the error and agreed corrective action is completed by the producer and submitted to the HoP.

For major errors, a breach report is completed, signed off by the relevant DD or Lead Statistician, and submitted to the HoP. The HoP will provide oversight and support for investigating the error and will engage with the OSR as well as the National Statistician’s office where appropriate. Where an error in our statistics also impacts on OGD statistics, the HoP will oversee communications and engagement plans between HMRC and the OGD.

Direct data feeds

Some HMRC data feeds go directly to ONS to inform their official statistics and do not pass through HMRC analytical teams. There have been some errors in ONS official statistics stemming from these feeds but as they do not impact HMRC statistics they have not routinely been reported to the HMRC HoP. This highlights the importance of clear, understood arrangements to assure direct data feeds. These data feeds fall outside the scope of this review, but as erroneous data has impacted ONS market-sensitive statistics this matter will be considered separately.

Recommendation 7.2

HMRC and ONS should review assurance arrangements for direct data feeds that affect market‑sensitive statistics, through a separate process, and any issues should be reported back to the HMRC Head of Profession for Statistics.

7.3 Communicating errors

When errors occur, the Code of Practice for Statistics encourages producers to release revisions or corrections as soon as possible and in line with the organisation’s public policy. Delays in revisions and corrections being released can result in a perceived lack of transparency, though often producers need time to determine:

• whether there is an error at all

• what caused the error

• the approximate size and direction of the error

• when and how the error can be corrected

If error reporting is done too quickly, this can mean:

• there is pressure on the analysts to answer detailed questions before they have validated and understood the details of the error

• there are rushed communications increasing the risk of further issues

For the publications within the scope of this review, there is the added complexity of HMRC’s data being a major source of information for OGD statistics that are published on the same day as HMRC statistics. It is therefore imperative that OGD stakeholders such as the ONS and HMT are informed in a timely and consistent manner if an issue is discovered with HMRC data.

External communications on errors have been produced on a consistent timetable and with agreed messaging between departments. However, this review has identified areas where informal communications between HMRC and other government stakeholders could be improved.

Recommendation 7.3

A workshop should be held with key producers and Head of Profession for Statistics from HMRC, ONS, HMT and the OSR with a view to capturing an agreed process on communicating errors.

7.4 Errors vs revisions

We have also found that greater clarity is needed in differentiating between errors and routine revisions. For example, in the tax receipts process, for specific taxes where reported receipts are aligned monthly to finance (such as Corporation Tax), payments are initially placed into an unallocated ledger if they are not allocated to a specific tax head. In some cases, payments remain unallocated at the time of initial publication, meaning routine revisions are required once the payment can be allocated. This is different from identifying a genuine error in the data. The quality reports published alongside official statistics must clearly differentiate between these routine revisions and true errors.

Recommendation 7.4

The quality reports for HMRC’s tax receipts publication and UK trade in goods statistics should be clear on how errors and revisions are defined and clarified where necessary. 

8. HMRC tax receipts and National Insurance contributions for the UK

The monthly HMRC tax receipts and National Insurance contributions (NICs) for the UK publication is an important statistical output produced by the department. It provides a timely indicator of the UK’s fiscal position and feeds into the joint ONS and HMT Public sector finances publication.

These receipts statistics provisionally amounted to £938.8 billion in the 2025 to 2026 tax year, with the top 4 taxes contributing 86% of the total. Thus, the focus in this review has been on the top 4 taxes: Income Tax, NICs, VAT, and Corporation Tax.

The statistics are used by various OGD stakeholders with different priorities and needs:

• the ONS need to accurately measure outturn for their joint publication with HMT ‘Public sector finances’

• the Office for Budget Responsibility (OBR) use receipts data at a detailed level to help with economic forecasting

• HMT need to monitor the state of public finances and potential implications for the rest of the economy

The published monthly figures are labelled as provisional and are aligned to the HMRC accounts in the summer after the end of the financial year. The annual data are robust as they are aligned to the HMRC annual accounts which are audited by the National Audit Office. However, recent errors have led to significant corrections to the statistics, which have impacted ONS and HMT market-sensitive statistics.

8.1 Monthly process

The tax receipts publication has a robust monthly cycle with several opportunities for quality assurance and challenge. The process is summarised below:

1. On the first working day (WD1), KAI produces initial internal estimates on outturn for the prior month. These use operational data, such as the total amounts received in relevant bank accounts. These estimates undergo QA checks within KAI and are shared with HMT and OBR on WD2 who provide challenge and QA by comparing figures with amounts paid over to HMT by HMRC.

2. On WD4, Finance shares the relevant ledger amounts allocated to customer liabilities with KAI. Both teams review these against the latest profiles. There is a subsequent meeting on WD5 between KAI and finance to discuss early insights and identify anomalies.

3. Detailed data are shared with OGD stakeholders on WD7, and there is a subsequent meeting between HMRC, ONS, HMT and OBR on WD8 to review firmer monthly numbers and challenge any anomalies. They also consider the payover gap (the difference between reported and allocated tax receipts versus amounts paid over to HMT), which presents a key assurance metric.

4. Data are re-shared with OGD stakeholders with any subsequent revisions at WD11.

5. On WD15 HMRC publish the ‘HMRC tax receipts and NICs for the UK’ statistics at the same time as the joint ONS and HMT ‘Public sector finances’ publication.

The monthly process and working day meetings require rigorous conversations and clearly defined actions. It was encouraging that it was acknowledged by key OGD stakeholders that:

• KAI who chair the WD8 meeting have clear communications and action points from these meetings

• KAI has recently improved processes with an explicit focus on trying to better explain the payover gap, including ahead of the WD8 meeting, and increase understanding and use of finance data to give better assurance for QA purposes.

As these recent additional sources under scrutiny are produced by finance colleagues in the RCFA line of business, they are now included in WD8 meetings to increase engagement and aid explanations.

8.2 Recent errors

Since the OSR’s 2020 review, there have been over 70 publications of the tax receipts statistics. Over this period, there have been 4 major errors where figures have needed significant correction (see Annex 1 and 5 for further details).

It is challenging for analysts to keep on top of all changes to source data managed in different parts of the department, particularly given the scale of transformational change underway in HMRC. This was a key contributing factor for the recent VAT error, as well as the 2024 PAYE error.

The publication process relies on close collaboration between statistical and accounting experts. Strengthening this working relationship would improve outcomes and reduce levels of risk.

8.3 Methodology changes

Following consultation with key stakeholders from OGDs, HMRC have aligned the methodology used for the main 4 taxes on a monthly basis rather than undertaking the alignment as part of the end of the year reconciliation process (current practice).

A key benefit of more frequent in-year alignment is the single source of information and internal reporting consistency for HMRC’s largest taxes (Income Tax, NICs, Corporation Tax and VAT). Alignment with accounting data is expected to reduce the risk of future errors from missing data feeds which have caused recent errors in KAI’s published statistics (in October 2025 and January 2024). HMRC’s accounting data is also subject to a number of controls and assurances and can be reconciled to amounts paid over to HMT, providing additional confidence in accounting outturns.

The main issue with adopting this approach is that there can be large monthly differences between KAI and Finance figures. However, on balance, the anticipated benefits of adopting this approach outweigh the potential drawbacks. The impact of the main drawback is mitigated through additional analysis which can be reported to OGDs that explains much of the difference between KAI and Finance figures.

While more frequent in-year alignment carries benefit in reducing risk of error, it does not remove all risk and HMRC will continue to review controls and mitigations to ensure the risk of future errors is reduced.

Recommendation 8.3

Align the reporting methodology of the main taxes in the tax receipts publication to be consistent with accounting data on a monthly basis. This recommendation was implemented in August 2026, in changes to the publication HMRC Tax Receipts and National Insurance Contributions for the UK.

9. UK trade in goods statistics

HMRC collects and publishes data on the UK’s international trade in goods at a summary product and country level, and by UK regions and devolved administrations. These are split across 2 accredited official statistics series – the monthly UK overseas trade in goods statistics, and the quarterly UK regional trade in goods statistics. These statistics are important as they form key inputs in measuring trade between around 200 partner countries and details for over 9,000 commodity types, around 100 at a summary “chapter” level.

Data are passed to the ONS for their UK Trade statistics on a Balance of Payments basis (also known as economic ownership basis), published on the same day as OTS each month. The contents of the publications differ in several ways, for example HMRC only publish trade in goods whereas ONS publish statistics for goods and services. HMRC’s trade data feeds into UK Trade, Balance of Payments/National Accounts, and directly feeds into calculations of GDP.

9.1 Trade Statistics team

The trade statistics are produced by analysts in the Borders and Trade line of business in HMRC and at a professional level are led by 3 Grade 6 statisticians. Where significant issues occur, analysts call on support from the HoP who sits in KAI. The HoP team maintain regular contact with statistics producers in Borders and Trade, and the HoP role is HMRC-wide rather than focused on a single directorate. HMRC also has a network of Lead Statisticians with representation from each statistics producing team, including Borders and Trade. These roles are held by senior and experienced statisticians to provide support where issues arise and share best practice. There are regular meetings between the HoP and Lead Statisticians. Government Statistical Service (GSS) professional standards, best practice, and learning and development opportunities are consistently applied across HMRC.

There was a recent review of the organisational placement of the Trade Statistics team as the main official statistics producer outside KAI. Some trade statistics products have remained in Borders and Trade due to proximity to the operational and IT teams who work with traders and their data, largely on the Customs Declaration System (CDS). Retaining these functions close to the CDS delivers greater value than locating them in KAI. This proximity improves awareness of data and policy changes, strengthening trade analysis and enabling more agile engagement with international partners.

9.2 Recent errors

Since the 2020 OSR review, there have been over 100 publications, and 2 major errors in HMRC official statistics where the numbers needed significant correction. The 2 major errors are summarised below:

October 2025 (see Annex 2 for further details)

A new CPC, introduced in March 2024, was not included in the Trade Statistics production system, and therefore trade movements associated with the CPC were wrongly excluded from the OTS. The associated revision increased UK goods exports by around 1.3% (£5 billion) for the period January to December 2024, and 1.8% (£6 billion) for the period January to October 2025.

June 2021

Some export declarations for goods under temporary admission to be re-exported were erroneously included in the OTS, resulting in a downwards revision of UK exports by approximately £300 to £500 million per month between January 2020 and February 2021.

Direct data feeds

There have also been some errors in ONS official statistics caused by HMRC’s direct data feeds, which did not impact HMRC official statistics. These are addressed in section 7.2.

9.3 Quality management

Quality assurance

This review identified a range of strong QA checks already in place for trade statistics, including controls to ensure that amounts and average weights are within acceptable tolerance thresholds.

In the trade statistics publications, there is a lot of user interest at a detailed level, such as precious metals or oil and gas. QA is currently mostly approached from the bottom-up, checking declaration data as it flows through the system and giving assurance that declared data is credible, providing quality at lower levels of aggregation.

However, there is also user interest at an aggregate level, including from the ONS. While HMRC already examine aggregate level totals, as do ONS, recommendations set out in this report seek to strengthen aggregate level QA to more easily identify gaps in data which may indicate systemic data issues not apparent at line level detail. 2 potential areas for improvement that have been identified are:

• more top-down QA — the top 5 trade codes make up 50% to 60% of all imports and exports, meaning more top-down QA could help flag major errors

• more time series analysis – time series analysis could support in identifying where data might be missing through flagging unusual trends

Recommendation 9.3i

Expand QA processes for trade statistics to include:

• more top-down QA focused on the top 5 trade codes

• more time series analysis to identify abnormalities in trends including potential missing data

Subject matter experts

A recent ONS innovation is the introduction of a subject matter expert (SME) who understands certain asset classes and can use expert knowledge to verify and explain trends. It is worth investigating whether the Trade Statistics team in HMRC could benefit from this approach through either leveraging the knowledge of Customer Compliance Managers (CCMs) or formalising the roles of SMEs.

Recommendation 9.3ii

Introduce SMEs to support the explanation of trends either through closer engagement with HMRC CCMs or formalising SMEs within the Trade Statistics team.

Other recommendations 

The most recent error described above related to Customs Procedure Codes. These are codes deployed within the Customs system to distinguish different types of trade movements. Trade statisticians use these codes to apply the agreed methodology.  

While such CPC changes are relatively rare, it is clear that the impact of misapplying one of these codes is potentially high. Given that, strengthening governance around the end-to-end approach for adopting these changes is recommended.  

Recommendation 9.3iii

Implement stronger, auditable governance practice around changes to Customs Procedure Codes and their impact on Trade Statistics 

10. Acknowledgements

We would like to express our gratitude to everyone who gave their time to speak with us, for their cooperation and openness, to inform this review. 

11. Contact Information

Contact: S Whellams, S Delf, statisticsenquiries@hmrc.gov.uk

Publication date: 15 September 2026

Annex 1: Major error in HMRC tax receipts and National Insurance contributions (NICs) for the UK, September 2025 

The error  

In September 2025, an error was discovered relating to the introduction of new transitory accounts that VAT payments could flow through. VAT payments going through these accounts were not being correctly captured by the existing statistics methodology and so were excluded from the figures in the receipts publication. 

Impact  

The impact of this error on the HMRC tax receipts publication was: 

• an upward revision of VAT cash receipts by £2.4 billion from April 2025 to August 2025 (representing a 3% increase in VAT receipts) 

• an upward revision of total HMRC cash receipts by the same amount 

The error had downstream impacts on:  

• the ONS and HMT Public Sector Finances publication, including an overestimation of public sector net borrowing 

Corrective actions  

HMRC published an exceptional release with the corrected figures on 8th October 2025, and the joint ONS and HMT Public Sector Finance publication was corrected on the same day. 

Actions taken to prevent a similar error occurring include: 

• an After Action Review to understand what lead to the error and how similar instances can be prevented 

• checks to assure the completeness of data feeds for other taxes 

• additional checks implemented on VAT receipts 

• moving forward, the VAT receipts team will receive more information on new financial processing projects 

Annex 2: Major error in trade in goods statistics, October 2025 

The error  

In October 2025, an error was discovered where a new CPC was erroneously excluded from the Trade Statistics production system and therefore excluded from the OTS. CPCs on customs declarations enables HMRC Trade Statistics to determine whether the goods being moved should be included or excluded from the OTS. The new CPC was introduced in March 2024 as part of a switch from the legacy CHIEF Customs platform to the new CDS Customs platform. The CPC covers controlled goods permanently exported from an exercise warehouse and relates to commodities in the fuel sector.  

Impact  

The impact of this error for UK exports was:   

• January to December 2024 rose by around 1.3%, which was £5.1 billion more than the published annual total at the time of £405 billion  

• January to October 2025 rose by around 1.8% or £6.5 billion more than the published annual total at the time £355 billion  

The error had downstream impacts on:  

• the ONS UK Trade and GDP releases  

• the annual 2024 release of UK trade in goods by business characteristics  

• the UK regional trade in goods disaggregated by smaller geographical areas  

Corrective actions  

HMRC announced the corrections to the OTS in November 2025 and published the corrections in January 2026 together with the ONS.  

Actions taken to prevent a similar error occurring include: 

• a full review of all existing ‘out of trade’ CPCs to ensure no other CPCs were wrongly assigned 

• review of the governance process that decides which CPCs are included 

• additional QA checks, including additional aggregated data reviews to identify quality issues 

• a review of how Customs policy or process change translates into impact on the Trade Statistics 

Annex 3: List of meetings undertaken during this review  

Below is a full list of the people I have spoken to as part of the review. I would like to thank them for their time, thoughtfulness and openness, which helped inform the findings and recommendations in this report.   

HMRC individual meetings  

DD, KAI Direct Business Taxes (DBT)

DD, KAI Hub for Evidence, Assurance, Research and Technology (HEART)

DD, KAI HEART

DD, KAI Indirect Taxes (IT)

DD, KAI Personal Taxes (PT)

DD, RCFA Financial Accounting and Internal Tax

DD, BT Customs Service and Operations (CS&O)

G6 analyst, KAI DBT

G6 analyst, KAI HEART

G6 analyst, KAI IT

G6 analyst, KAI PT

G6 analyst, KAI PT

G6 analyst, BT CS&O

G6 accountant, RCFA

G7 analyst, KAI HEART

G7 analyst, KAI PT

G7 analyst, KAI PT

HMRC team meetings  

BT CS&O Trade Statistics team

KAI PT Income Tax and NICs receipts team

KAI DBT Corporation Tax receipts team

KAI IT VAT receipts team

External stakeholders  

ONS: DD, Public Sector Finances

ONS: DD, Trade Statistics

OBR: G6, Tax and Fiscal Forecasts

HMT: Head of Profession for Statistics

ONS: G6, Public Sector Finances

ONS: G6, Trade Statistics

ONS: G7, Public Sector Finances

HMT: G7, Public Sector Finances

Annex 4: Recommendations

Recommendation 5.1

Frequent and ongoing engagement between analysts and data suppliers will help ensure upstream changes are more fully understood by statistics producers and that the statistical consequences of change are better managed. This engagement should happen at different levels:

• there should be frequent engagement (for example, monthly) at working level between statistics producers and data suppliers

• there should be regular engagement (for example, biannual) at a senior level between senior analysts and data suppliers

• the data management and IT support functions within analytical areas should proactively influence change programme plans to ensure analytical needs are met

Some of this engagement could include OGD stakeholders where appropriate.

Recommendation 5.2i

All official statistics should map their reliance on infrastructure as part of process mapping. The dependencies should then be recorded as risks to the process and mitigations identified including short-term remedies (such as partial migration of the process where possible) and long-term remedies (such as transformation of the infrastructure through the IT Transformation programme).

Recommendation 5.2ii

The timescales for delivery of improved analytical infrastructure by IT Transformation programmes should be reported by these programmes as part of ongoing reports following this review. 

Recommendation 6.1

Each key statistical product should review their data process maps and ensure they are up to date. This should then be reported back to the relevant Lead Statistician.

Recommendation 6.2

Consistently document QA checks including automated QA with regular auditing from Lead Statisticians and the Head of Profession for Statistics.

Recommendation 6.3

HMRC’s analytical Senior Leadership Team need to champion and strengthen analytical curiosity, drawing on the support of forums including the Quality Champions network and Learning network. Additionally, as part of the assurance/audit process which we are recommending, evidence of this activity should be captured.

Recommendation 6.4

Develop a way to share and learn from near misses (for example, through a regular bulletin) to ensure they are more consistently used.

Recommendation 6.5

Statistical publications which rely on input from multiple teams should develop a responsibility assignment matrix to define responsibilities of various stakeholders.

Recommendation 6.6

HMRC’s analytical senior leadership teams should identify key roles for the production of HMRC’s statistics and (i) identify the level of risk if the role is vacant and (ii) implement replacement strategies for these roles.

Recommendation 6.7

HMRC should undertake a project driven by the support of analytical senior leaders to develop ways to raise the profile of its statistics, including clarifying their strategic importance and establishing mechanisms to support their prioritisation across the department.

Recommendation 7.2

HMRC and ONS should review assurance arrangements for direct data feeds that affect market‑sensitive statistics, through a separate process, and any issues should be reported back to the HMRC Head of Profession for Statistics.

Recommendation 7.3

A workshop should be held with key producers and Head of Profession for Statistics from HMRC, ONS, HMT and the OSR with a view to capturing an agreed process on communicating errors.

Recommendation 7.4

The quality reports for HMRC’s tax receipts publication and UK trade in goods statistics should be clear on how errors and revisions are defined and clarified where necessary. 

Recommendation 8.3

Align the reporting methodology of the main taxes in the tax receipts publication to be consistent with accounting data on a monthly basis. Implemented August 2026, in HMRC Tax Receipts and National Insurance Contributions for the UK.

Recommendation 9.3i

Expand QA processes for trade statistics to include:

• more top-down QA focused on the top 5 trade codes

• more time series analysis to identify abnormalities in trends including potential missing data

Recommendation 9.3ii

Introduce SMEs to support the explanation of trends either through closer engagement with HMRC CCMs or formalising SMEs within the trade statistics team.

Recommendation 9.3iii

Implement stronger, auditable governance practice around changes to Customs Procedure Codes and their impact on Trade Statistics 

Annex 5: Major errors in HMRC tax receipts and NICs in the UK and UK trade in goods statistics since April 2020

HMRC tax receipts and NICs in the UK

1. September 2025: there was an issue with the way VAT cash receipts had been recorded, impacting provisional cash receipts for April to August 2025. The error was the result of a change in financial processing leading to some VAT payments not being included in the tax receipts publication. This led to an upward revision of cash receipts by £2.4 billion from April to August (resulting in a 3% increase in VAT receipts).
2. January 2024: there was an issue with the way PAYE and NICs receipts had been recorded. The error was a result of a new payment mechanism causing payments to not be correctly captured as PAYE or NICs payments. The impact was an under-reporting of PAYE and NICs by £6.5 billion for the period April to December 2023 (2.2% of PAYE and NICs receipts), and an under-reporting of £1.8 billion for the period October 2022 to March 2023 (0.5% of PAYE and NICs receipts).
3. November 2022: there was a calculation error impacting VAT and Fines & Penalties for the 2021 to 2022 tax year. The impact was an under-reporting of VAT by approximately £750 million (0.5% of total VAT receipts in 2021 to 2022), and an under-reporting of Fines and Penalties by approximately £10 million (1.6% of total Fines & Penalties receipts in 2021 to 2022).
4. January 2021: there was a calculation error impacting several tax regimes in the 2018 to 2019 tax year. This resulted in receipts being underreported by £690 million.

UK trade in goods statistics

1. October 2025: a new CPC, introduced in March 2024, was not included in the Trade Statistics production system, and therefore trade movements associated with the CPC were wrongly excluded from the OTS. The associated revision increased UK goods exports by around 1.3% (£5 billion) for the period January to December 2024, and 1.8% (£6 billion) for the period January to October 2025.
2. June 2021: some export declarations for goods under temporary admission to be re-exported were erroneously included in the OTS, resulting in a downwards revision of UK exports by approximately £300 to £500 million per month between January 2020 and February 2021.