Official Statistics

5. Tax gaps: Corporation Tax

Published 18 December 2012

Introduction

This chapter explains Corporation Tax gaps, outlining the main components and how they are estimated. It describes the key methods, data sources, and analytical approaches used to estimate the tax gap, broken down by customer group. The chapter explains the use of random enquiry programme (REP) data and risk-based enquiries to measure the extent of non-compliance across all customer groups, outlined below.

Corporation Tax for small businesses

Overview

The Corporation Tax for small businesses model estimates the tax gap arising from incorrect returns filed by small, incorporated businesses. It draws on evidence from the random enquiry programme (REP), which provides a representative view of non‑compliance by subjecting selected returns to full compliance checks.

The small businesses population includes all live companies required to file a Corporation Tax return, excluding those that are dormant, dissolved, or outside the small business customer group.

Tax Gap Calculation (step-by-step)

1. Measure under‑declared tax in a random sample.

2. Calculate the average amount of under-declared tax in the random sample.

3. Apply a rolling average to the average amount of under-declared tax.   

4. Scale the sample to the whole population.

5. Apply the non-detection multiplier.

6. Add non‑payment (tax liability that will never be paid).

7. Subtract compliance yield.

8. Project incomplete years in-line with growth in tax liabilities.

Methodology

Data and Sampling

This section describes the data sources used for the Corporation Tax small businesses model and how the sample is selected, structured, and prepared for analysis.

Data sources

The Corporation Tax for small businesses model is based on data from the REP. Each selected return undergoes a full compliance check, during which caseworkers review the business’s records in detail and identify any under‑declared liabilities.

Population counts used to scale up the REP to the full population are drawn from HMRC administrative systems and reflect the total number of live small businesses required to file a Corporation Tax return. Cases from businesses that are dormant, dissolved, or otherwise outside the scope of the small business population are excluded.

Sampling approach

The REP for small businesses uses a stratified random sample of Corporation Tax returns. Since April 2013, stratification has been based on annual trading turnover. This allows the sample results to be weighted to reflect the structure of the underlying population and improves the accuracy of the tax gap estimate.

Before 2016 to 2017, samples were selected using the former SME customer group definition. From 2016 to 2017 onwards, the sample has been drawn solely from the small business customer group. Earlier years have been aligned to ensure consistency across the time series.

Sample sizes

Sample sizes for the REP vary by year, reflecting changes to customer group definitions, sampling design, and operational capacity. Annual sample size figures provide important context for the robustness of the estimates, as larger samples generally support more stable results.

A table of Corporation Tax small businesses REP sample sizes is published each year and shows how the number of cases selected has changed over time.

Table A5.1: Sample sizes for Corporation Tax REP

Accounting period ending in year	Sample size
2005 to 2006	419
2006 to 2007	460
2007 to 2008	492
2008 to 2009	491
2009 to 2010	480
2010 to 2011	490
2011 to 2012	448
2012 to 2013	576
2013 to 2014	419
2014 to 2015	458
2015 to 2016	266
2016 to 2017	451
2017 to 2018	332
2018 to 2019	354
2019 to 2020	324
2020 to 2021	327
2021 to 2022	359
2022 to 2023	337

Note for Table A5.1

1. Since the tax year 2016 to 2017 the Corporation Tax sample size given is for small businesses only. Before 2016 to 2017 HMRC’s former small and medium-sized enterprises (SME) customer group classification was used.

Deselections

Some sampled cases cannot be worked, for example if the business has ceased trading or has recently been subject to another compliance check. These cases are treated as deselections.

To avoid biasing the sample we include cases that are deselected from the sample but are still within the population of interest. If the business has undergone a recent compliance check, we substitute the outcome of this earlier compliance check into the case. If no such previous compliance check exists, we assign a value based on the average yield and probability of being non-compliant in the taxpayer’s stratum using data from results of compliance checks into businesses in the corresponding stratum.

Outliers

Outliers are individual cases with large yield which are disproportionately different from the yield of the other cases in the sample. Due to the nature of our samples, our estimates are particularly sensitive to outliers To ensure that these small number of cases do not have an undue influence on the tax gap calculation, their yield values are capped. This allows us to use all valid information while smoothing the year-on-year variability.

Yield data is modelled using a representative statistical distribution. The final value used for each tax year is calculated as a 3-year moving average of the 99.85th percentile from this distribution, calculated based only on the results of years where the sample was stratified. For years before stratification, and years where a full 3 years of stratified results are not available, a value based on the last 3 complete stratified years is used.

This approach allows all valid cases to be included while ensuring that outliers do not disproportionately influence the final estimate.

Estimating under-declared liabilities

This section explains how the results from the REP are transformed into an estimate of the total amount of under‑declared Corporation Tax for small businesses.

REP case outcomes

Each REP case provides a measured outcome for that business: either no adjustment or an identified amount of under‑declared Corporation Tax. These outcomes represent the observed levels of non‑compliance within the stratified sample.

Calculating average under‑declared liabilities

For each turnover stratum, the model calculates the average amount of under‑declared liability identified in the completed REP cases. This produces a set of stratum‑specific averages that reflect the differing compliance behaviour of businesses of different sizes.

Smoothing and stability

The current approach uses a smoothing technique to account for variations in under-declared liabilities due to yearly fluctuations. For each year, the estimate is based on that year and the 2 previous years. For each of these years, the rate of non-compliance is multiplied by the value of non-compliance. The most recent year is given a double weight. These figures are then averaged to produce the estimate.

Scaling to the population

The average under‑declared liability for each stratum is weighted by the number of live small businesses in that stratum. Summing these results across all strata produces an estimate of the total under‑declared liability for the entire small businesses Corporation Tax population.

Model Adjustments and Refinements

This section describes the adjustments applied to the estimated under‑declared liabilities to ensure the Corporation Tax for small businesses tax gap reflects the full extent of non‑compliance and produces a consistent time series.

Alignment of earlier years

For tax years prior to 2016 to 2017, REP data was collected under the former SME customer classification rather than the current small businesses customer group. To ensure consistency across the time series, an adjustment is applied to align earlier years with the modern population definition. This involves applying a conversion factor derived from historical data where both definitions overlap, allowing estimates based on SME samples to be made comparable with the small businesses population used in later years.

Non‑detection

Not all incorrect returns or under‑declared amounts will be identified through compliance checks, even within a random enquiry programme. To account for this, the model applies a non‑detection multiplier to the scaled under‑declared liabilities.

The multiplier is derived from analysis of REP results and expert judgement about the level of non‑detected non‑compliance in the small businesses population. It reflects the difference between the non‑compliance that is identified by caseworkers and the amount estimated to remain undetected.

The non‑detection multiplier varies by year. In earlier periods the model uses values originally based on analysis of international research, while more recent years reflect improvements in how Corporation Tax small businesses REP cases are worked. As caseworkers have become more effective at identifying non‑compliance from 2018 to 2019, the multiplier has been reduced to better represent the level of remaining undetected non‑compliance.

In addition to the central estimate, the model also applies lower and upper bounds to reflect uncertainty.

• The lower bound assumes that all non‑compliance has been detected (multiplier = 1)
• The upper bound applies a higher multiplier to reflect the plausible range of undetected non‑compliance

These ranges are used to describe the level of uncertainty around the central estimate and are updated when evidence about detection rates or case working effectiveness changes. The values are shown in Table A5.2.

Table A5.2: Comparison of adjustments for non-detection

Tax Years	Multiplier for central estimate	Multiplier for lower estimate	Multiplier for upper estimate
Before 2018 to 2019	1.457	1.000	1.914
2018 to 2019	1.403	1.000	1.807
2019 to 2020	1.332	1.000	1.664
2020 to 2021	1.261	1.000	1.521
2021 to 2022	1.225	1.000	1.450
2022 to 2023	1.189	1.000	1.379

Non‑payment

Some tax liabilities will not ultimately be paid. To reflect this, the model includes an estimate of non‑payment attributable to each tax year.

The method estimates eventual non-payment attributable to the year of tax debt creation. This does not extend beyond the 2018 to 2019 tax year. For years before 2018 to 2019 non-payment refers to tax debts that are written off or remitted in a tax year by HMRC and result in a permanent loss of tax.

Compliance yield

The REPs provide an estimate of the tax gap due to incorrect returns. However, HMRC carries out a wider programme of compliance activity to identify and correct erroneous returns. To calculate the net tax gap, it is necessary to subtract the compliance yield from this activity.

The figures for yield are taken from HMRC’s systems for recording the outcomes of compliance checks and relate to cases settled during each year rather than compliance checks into returns relating to a specific tax year. Yield by year of settlement is used as a proxy due to the extended timeline for completing all the compliance activity related to the liabilities for a year.

Compliance yield figures can be found in Table 5.2 of the online tables.

Projections for years with incomplete data

Due to the timing of returns and the duration of compliance checks, full REP data is not available for the most recent tax years. For the years after 2022 to 2023 the model applies projection factors based on the most recent complete information. The projections assume a stable gross tax gap and incorporate up‑to‑date information on liabilities, non‑payment and compliance yield. These projections are replaced with actual data in subsequent editions once sufficient REP cases have been settled.

Data features

In 2019 to 2020 and 2020 to 2021, compliance checks were only undertaken remotely (desk-based) due to the suspension of in-person compliance activity during the pandemic. From 2021 to 2022 onwards, cases have been handled either remotely or in person, depending on each situation and caseworker judgement.

Case-working methods (desk-based or face-to-face) can vary between years, which may influence how quickly or thoroughly compliance checks are completed. This variation does not affect the random sample’s representativeness but is important for understanding year-to-year trends in the data.

Timing

There are 2 timing factors that affect when Corporation Tax small businesses tax gap estimates can be produced. The first arises from delays inherent in the returns process. Companies have until a year after the end of their accounting period to submit their return. HMRC then has a further year in which a compliance check can be opened.

The second relates to the compliance checks themselves. Random enquiries can be complex and may take several years to complete.

To produce timely estimates despite these delays, forecasts are used for open or unsettled cases. Where possible, caseworker forecasts are used to estimate the likely outcome of ongoing compliance checks. Where these are not available, forecast values are generated based on the outcomes of settled cases with similar time-to-close durations. These forecasts allow the model to produce a best estimate for each tax year while recognising that results will be revised once more complete information becomes available.

Validation

HMRC undertakes a validation exercise each year to ensure that the underlying compliance check outcomes used in the Corporation Tax for small businesses model are accurate. As part of this programme, a sample of cases is reviewed to confirm that recorded compliance check outcomes, such as yield amounts, have been captured correctly. Any inaccuracies identified through this process are corrected before tax gap estimates are finalised.

Further, internal HMRC Quality Assurance (QA) processes are in-place to ensure quality estimates. As part of this, an independent analytical assurer reviews the model and estimates and completes QA documentation. Any issues found are communicated and rectified prior to publication.

Data Issues and Limitations

The Corporation Tax for small businesses model relies on REP case data and population counts from HMRC administrative systems. While these sources provide a strong foundation, several inherent limitations affect the precision of the tax gap estimates.

Compliance checks cannot identify all incorrect returns or the full scale of non‑compliance. As a result, the REP findings underestimate the tax gap. Although non-detection is addressed through a multiplier elsewhere in the methodology, the underlying limitation stems from the inherent constraints of compliance checks themselves.

Finally, the model depends on administrative population data that may be revised over time. Changes in customer group definitions, alignment exercises, or reclassification of cases can lead to updates to historical data. These revisions ensure consistency but highlight that population measures are not entirely static.

Sources of Error

There are 2 main sources of error associated with the results of REPs which could result in the true values of the tax gaps differing from the estimates produced. These are:

• sampling variation in the data: the whole population is not subject to a compliance check, so even though the sample is designed to be representative, its characteristics may differ from the population purely by chance
• systematic uncertainty where the sample results consistently tend to under-report the true values for the population, or where the sample does not include subgroups of the full population, for example those participating in avoidance

Uncertainty Rating

The uncertainty rating for the small businesses Corporation Tax gap estimate is ‘high’. This reflects that the model captures the majority of the tax base, uses a stratified REP methodology, and is supported by reasonable case data. However, uncertainty remains due to; the need to project results for years where REP data is not yet available; the use of forecasts for ongoing compliance checks; changes to ways of working; the use of rolling averages; and that not all non‑compliance can be identified through compliance checks alone. These factors contribute to variation in both the measured under‑declared liabilities and the adjustments applied for undetected non‑compliance. The rating has been increased from a ‘medium’ rating in the previous edition.

Corporation Tax for mid-sized businesses

Overview

The Corporation Tax for mid-sized businesses model estimates the tax gap using risk-based enquiries, focusing on the riskiest businesses rather than a random sample of businesses. To prevent a few very large under-declared tax cases from distorting the overall results, the model applies an extreme value methodology, ensuring a more typical picture of non-compliance.

Risk-based enquiry data is available from 2014 to 2015 onwards; earlier years use previous figures from ‘Measuring tax gaps 2020 edition’. As some risk-based enquiry cases remain open in recent years, projections are used to provide the best current estimates. A non-detection multiplier, based on expert judgement, accounts for undetected non-compliance.

Step-by-step tax gap calculation

1. Estimate under‑declared tax in observed risk-based enquiries.

2. Estimate the extreme value lower bound.

3. Estimate the upper bound.

4. Take the average of the lower bound and upper bound.

5. Apply the non‑detection multiplier.

6. Add non‑payment (tax liability that will never be paid).

7. Subtract compliance yield.

8. Project recent years in line with changes in tax liabilities.

Methodology

Data sources

The Corporation Tax for mid-sized businesses model mainly uses operational data from HMRC’s risk-based enquiries. These enquiries provide detailed information on identified under-declarations of tax within the mid-sized businesses population.

To scale the findings from these enquiries up to the whole mid‑sized businesses population, the model uses HMRC’s administrative data on how many businesses are in this group and their overall Corporation Tax liabilities. Operational data is available from 2014 to 2015 onward, so earlier years continue to use the figures published at the time.

Risk-based enquiry outcomes

HMRC carries out risk-based enquiries to investigate identified high-risk cases and determine whether the correct amount of Corporation Tax has been declared. The outcome of each enquiry might show no issues, or it may identify additional tax that should have been declared. These outcomes provide direct evidence of non-compliance that is used in the model.

Some enquiries identify very large amounts of under‑declared tax, while most identify much smaller amounts. To prevent a small number of unusually large cases from distorting the overall results, the model uses a statistical approach that reduces their influence while keeping them in the dataset. This makes the final estimates more stable and reflective of typical patterns of non‑compliance across the mid‑sized businesses population.

Forecasting open cases

Many risk-based enquiry cases for mid-sized businesses remain open at the point of estimation, so the model forecasts the expected compliance yield for these cases to complete the dataset used to estimate the tax gap. For open cases, the model pairs each case with a similar closed case and uses the observed yield from the matched case as the forecast. These forecast values are replaced by actual outcomes in subsequent editions once the underlying cases have been settled.

Extreme value methodology (lower bound estimate)

The extreme value methodology is used to estimate under‑declared tax when most of the value is concentrated in a small number of cases.

1. Use risk‑based enquiry results: the method begins with the outcomes of risk‑based enquiry cases, which show how much tax has been under‑declared in each case.

2. Identify extreme‑value behaviour: the results typically show that a small number of cases account for most of the under‑declared tax.

3. Apply a threshold cut‑off: cases that do not follow this extreme‑value pattern are removed so that only those consistent with the expected distribution are included.

4. Fit a power‑law model: the remaining above‑threshold cases are fitted to a statistical power‑law model to estimate under‑declared tax among high‑yield cases.

5. Estimate cases without risk-based enquiries: the model then estimates how many similar above‑threshold cases may exist among businesses that were not subject to risk‑based enquiries.

For ‘Measuring tax gaps 2026 edition, the extreme value method has been improved. The model now uses the observed yield for the above-threshold risk-based enquiry cases, where previously it used the under-declared tax estimated by the model for these cases.

The method does not assume the presence of additional high‑yield cases beyond those observed, so the resulting estimate is likely to underestimate the true level of under‑declared tax. For this reason, this method is used as a lower bound estimate, with further adjustments applied elsewhere in the methodology.

Upper bound estimate

The upper bound estimate is likely to overestimate the true level of under‑declared tax.

1. Assume average behaviour matches risk-based enquiry cases: the model assumes that businesses not subject to risk‑based enquiries have the same average tax gap percentage as those that were subject to risk-based enquiries.

2. Apply this average rate to the population not subject to risk-based enquiries: this percentage is applied across all businesses that were not selected for a risk‑based enquiry.

This method produces a deliberately higher estimate. As risk-based enquiry cases were selected based on expected high levels of non‑compliance, applying their average rate to the whole population overstates the likely level of non‑compliance.

Central estimate

As the lower bound method is likely to produce an underestimate and the upper bound method is likely to produce an overestimate, the average of these 2 results is used as a reasonable estimate of non-compliance for the tax gap.

Model Adjustments and Refinements

This section describes the adjustments applied to ensure the mid-sized businesses Corporation Tax gap reflects the full extent of non‑compliance and produces a consistent time series.

Non-detection

Not all incorrect returns or under‑declared tax will be identified through risk‑based enquiries. To account for this, the model applies a non‑detection multiplier, which adjusts the risk-based enquiry results to better reflect the true level of non‑compliance in the mid‑sized businesses population. This multiplier is based on HMRC expert opinion and is reviewed regularly to ensure it reflects the latest understanding of risk-based enquiry effectiveness.

Table A5.3: Non-detection multiplier in mid-sized businesses Corporation Tax gap

Tax Years	Multiplier
2014 to 2015 onwards	1.225

Non-payment

Some tax liabilities will not ultimately be paid. To reflect this, the model includes an estimate of non‑payment attributable to each tax year.

The method estimates eventual non-payment attributable to the year of tax debt creation. This does not extend back beyond the 2018 to 2019 tax year. For years before 2018 to 2019, non-payment refers to tax debts that are written off or remitted in a tax year by HMRC and result in a permanent loss of tax.

Compliance yield

To calculate the net tax gap, compliance yield is subtracted from the gross tax gap. Compliance yield for mid‑sized businesses differs from the small businesses approach because it is attributed to the accounting period (year of liability) rather than the year in which compliance activity is settled. This means it is different to the compliance yield published in HMRC’s Annual Report and Accounts.

In the mid‑sized businesses model, compliance yield is calculated as the total yield from closed cases plus the estimated yield from open cases, ensuring that all compliance activity is aligned to the correct liability year. This provides a more accurate reflection of the tax corrected within the period and helps maintain consistency with the risk‑based enquiry data underpinning the model.

Timing

Risk-based enquiries can be complex and may take several years to complete. This is partially accounted for by forecasting expected compliance yield for open cases.

Differences between the forecast yield and actual yield may lead to revised tax gap estimates in subsequent publications, but the use of forecasting reduces the chance that these revisions are significant. The tax gap for more recent years is likely to be subject to larger revisions because a higher proportion of the compliance yield is estimated.

Projections for recent years

There are more open cases in more recent accounting periods as there has been less time to complete these enquiries. The use of projected data for these years reduces the chance of large revisions to these years in future.

Both the compliance yield and gross tax gap figures for the tax years from 2022 to 2023 onward have been projected. This is done based on the percentage of compliance yield and gross tax gap to liabilities for 2021 to 2022.

The projections will lead to revised tax gap estimates in subsequent publications when these projections are replaced with actual estimates based on risk-based enquiry data.

Sources of Error

There are 3 main sources of error that may cause the true mid‑sized businesses Corporation Tax gap to differ from the model estimates.

First, systematic uncertainty arises when risk‑based enquiry results under‑report the true level of non‑compliance or when parts of the population are not fully captured.

Second, variations in risk‑based enquiry data occur because risking approaches change over time, which can affect the amount of tax identified and introduce differences between years.

Third, uncertainty in population numbers can affect results, as the definition of the mid‑sized businesses population can shift.

Some of this systematic uncertainty is addressed by the non‑detection multiplier.

Uncertainty rating

The uncertainty rating for the mid‑sized businesses Corporation Tax gap estimate is ‘medium’. The model captures most of the tax base and uses detailed operational data from risk-based enquiries, supported by an extreme value methodology that reduces sensitivity to unusually large cases. However, uncertainty remains because a proportion of cases are still open at the time of estimation and must be forecasted, and because the results can be affected by changes in how cases are identified and worked. These factors mean that estimates for the most recent years are more likely to be revised as additional risk-based enquiry outcomes become available. This uncertainty rating is unchanged from ‘Measuring tax gaps 2025 edition’.

Corporation Tax for large businesses

Overview

The Corporation Tax for large businesses model estimates the tax gap using risk-based enquiries, focusing on the riskiest businesses rather than a random sample of businesses. To prevent a few very large under-declared tax cases from distorting the overall results, the model applies an extreme value methodology, ensuring a more typical picture of non-compliance.

Risk-based enquiry data is available from 2014 to 2015 onwards; earlier years use previous figures from ‘Measuring tax gaps 2020 edition’. As some risk-based enquiry cases remain open in recent years, projections are used to provide the best current estimates. A non-detection multiplier, based on expert judgement, accounts for undetected non-compliance.

Step-by-step tax gap calculation

1. Estimate under‑declared tax in observed risk-based enquiries.

2. Estimate the extreme value lower bound.

3. Estimate the upper bound.

4. Take the average of the lower bound and upper bound.

5. Apply the non‑detection multiplier.

6. Add non‑payment (tax liability that will never be paid).

7. Subtract compliance yield.

8. Project recent years in line with changes in tax liabilities.

Methodology

Data sources

The Corporation Tax for large businesses model mainly uses operational data from HMRC’s risk‑based enquiries. These enquiries provide detailed information on identified under‑declarations of tax within the large businesses population.

To scale the findings from these enquiries up to the whole large businesses population, the model uses HMRC’s administrative data on how many businesses are in this group and their overall Corporation Tax liabilities. Operational data from the risk-based enquiries is available from 2014 to 2015 onward, so earlier years continue to use the figures published at the time.

Risk-based enquiry outcomes

HMRC carries out risk‑based enquiries to investigate identified high-risk cases and determine whether the correct amount of Corporation Tax has been declared. The outcome of each enquiry might show no issues, or it may identify additional tax that should have been declared. These outcomes provide direct evidence of non‑compliance that is used in the model.

Some enquiries identify very large amounts of under‑declared tax, while most identify much smaller amounts. To prevent a small number of unusually large cases from distorting the overall results, the model uses a statistical approach that reduces their influence while keeping them in the dataset. This makes the final estimates more stable and reflective of typical patterns of non‑compliance across the large businesses population.

Forecasting open cases

Many risk-based enquiry cases for large businesses remain open at the point of estimation, so the model forecasts the expected compliance yield for these cases to complete the dataset used to estimate the tax gap. For any open case expected to close within the next 6 months or currently going through litigation, the model uses caseworker assessments to forecast expected compliance yield.

For all remaining open cases, the model pairs each case with a similar closed case and uses the observed yield from the matched case as the forecast.

These forecast values are replaced by actual outcomes in subsequent editions once the underlying cases have been settled.

Extreme value methodology (lower bound estimate)

The extreme value methodology is used to estimate under‑declared tax when most of the value is concentrated in a small number of cases.

1. Use risk‑based enquiry results: the method begins with the outcomes of risk‑based enquiry cases, which show how much tax has been under‑declared in each case.

2. Identify extreme‑value behaviour: the results typically show that a small number of cases account for most of the under‑declared tax.

3. Apply a threshold cut‑off: cases that do not follow this extreme‑value pattern are removed so that only those consistent with the expected distribution are included.

4. Fit a power‑law model: the remaining above‑threshold cases are fitted to a statistical power‑law model to estimate under‑declared tax among high‑yield cases.

5. Estimate cases without risk-based enquiries: the model then estimates how many similar above threshold cases may exist among businesses that were not subject to risk-based enquiries.

For ‘Measuring tax gaps 2026 edition, the extreme value method has been improved. The model now uses the observed yield for the above-threshold risk-based enquiry cases, where previously it used the under-declared tax estimated by the model for these cases.

The method does not assume the presence of additional high‑yield cases beyond those observed, so the resulting estimate is likely to underestimate the true level of under‑declared tax. For this reason, this method is used as a lower bound estimate, with further adjustments applied elsewhere in the methodology.

Upper bound estimate

The upper bound estimate is likely to overestimate the true level of under‑declared tax.

1. Assume average behaviour matches risk-based enquiry cases: the model assumes that businesses not subject to risk‑based enquiries have the same average tax gap percentage as those that were subject to risk-based enquiries.

2. Apply this average rate to the population not subject to risk-based enquiries: this percentage is applied across all businesses that were not selected for a risk‑based enquiry.

This method produces a deliberately higher estimate. As risk-based enquiry cases were selected based on expected high levels of non‑compliance, applying their average rate to the whole population overstates the likely level of non‑compliance.

Central estimate

As the lower bound method is likely to produce an underestimate and the upper bound method is likely to produce an overestimate, the average of these 2 results is used as a reasonable estimate of non-compliance for the tax gap.

Model Adjustments and Refinements

This section describes the adjustments applied to ensure the large businesses Corporation Tax gap reflects the full extent of non‑compliance and produces a consistent time series.

Non-detection

Not all incorrect returns or under‑declared tax will be identified through risk‑based enquiries. To account for this, the model applies a non‑detection multiplier, which adjusts the risk-based enquiry results to better reflect the true level of non‑compliance in the large businesses population. This multiplier is based on HMRC expert opinion and is reviewed regularly to ensure it reflects the latest understanding of risk-based enquiry effectiveness.

Table A5.4: Non-detection multiplier in large businesses Corporation Tax gap

Tax Years	Multiplier
2014 to 2015 onwards	1.225

Non-payment

Some tax liabilities will not ultimately be paid. To reflect this, the model includes an estimate of non‑payment attributable to each tax year.

The method estimates eventual non-payment attributable to the year of tax debt creation. This does not extend back beyond the 2018 to 2019 tax year. For years before 2018 to 2019, non-payment refers to tax debts that are written off or remitted in a tax year by HMRC and result in a permanent loss of tax.

Compliance yield

To calculate the net tax gap, compliance yield is subtracted from the gross tax gap. Compliance yield for large businesses differs from the small businesses approach because it is attributed to the accounting period (year of liability) rather than the year in which compliance activity is settled. This means it is different to the compliance yield published in HMRC’s Annual Report and Accounts.

In the large businesses model, compliance yield is calculated as the total yield from closed cases plus the estimated yield from open cases, ensuring that all compliance activity is aligned to the correct liability year. This provides a more accurate reflection of the tax corrected within the period and helps maintain consistency with the risk‑based enquiry data underpinning the model.

Timing

Risk-based enquiries can be complex and may take several years to complete. This is partially accounted for by forecasting expected compliance yield for open cases.

Differences between the forecast yield and actual yield may lead to revised tax gap estimates in subsequent publications, but the use of forecasting reduces the chance that these revisions are significant. The tax gap for more recent years is likely to be subject to larger revisions because a higher proportion of the compliance yield is estimated.

Projections for recent years

There are more open cases in more recent accounting periods as there has been less time to complete these enquiries. The use of projected data for these years reduces the chance of large revisions to these years in future.

Both the compliance yield and gross tax gap figures for the tax years from 2023 to 2024 onward have been projected. This is done based on the percentage of compliance yield and gross tax gap to liabilities for 2022 to 2023.

The projections will lead to revised tax gap estimates in subsequent publications when these projections are replaced with actual estimates based on risk-based enquiry data.

Sources of Error

There are 2 main sources of error that may cause the true large businesses Corporation Tax gap to differ from the model estimates.

First, systematic uncertainty arises when risk‑based enquiry results under‑report the true level of non‑compliance or when parts of the population are not fully captured.

Second, variations in risk‑based enquiry data occur because risking approaches change over time, which can affect the amount of tax identified and introduce differences between years.

Some of this systematic uncertainty is addressed by the non‑detection multiplier.

Uncertainty rating

The uncertainty rating for the large businesses Corporation Tax gap estimate is ‘medium’. The model captures most of the tax base and uses detailed operational data from risk-based enquiries, supported by an extreme value methodology that reduces sensitivity to unusually large cases. However, uncertainty remains because a proportion of cases are still open at the time of estimation and must be forecasted, and because the results can be affected by changes in how cases are identified and worked. These factors mean that estimates for the most recent years are more likely to be revised as additional risk-based enquiry outcomes become available. This uncertainty rating is unchanged from ‘Measuring tax gaps 2025 edition’.