## _cr13314

## Source details

**Canonical URL:** [_cr13314](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13314.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13314.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13314.pdf.json)

---

### Preface and Executive Summary — objectives and program scope
- Assessment requested by Mr. Jonathan Athow (Director, Knowledge, Analysis and Intelligence Department (KAI)) from Her Majesty’s Revenue and Customs (HMRC).
- Assessment team led by Mr. Juan Toro (Assistant Director, FAD) with Messrs. Kentaro Ogata (Advisor, FAD), Selcuk Caner (Senior Economist, FAD), and Eric Hutton (Technical Assistance Advisor, FAD).
- Fieldwork: initial fact-finding visit April 23–25, 2013; follow-up visit held on July 1, 2012; continued work at IMF headquarters and incorporation of HMRC and FAD advisory committee comments.
- Main purpose: thorough review of HMRC’s tax gap analysis program, including models and methodologies, use in supporting HMRC operations, and dissemination approach.
- HMRC Vision Statement (Strategic Plan 2012–2015): “We will close the tax gap, our customers will feel that the tax system is simple for them and even-handed, and we will be seen as a highly professional and efficient organization.”
- Three critical objectives implied by the Vision Statement:
  - Measuring tax revenue losses to be closed.
  - Supporting efficiency in closing the tax gap and allocating resources to achieve high impact in reducing noncompliance.
  - Supporting perception of fairness through transparent and sufficiently detailed reporting of tax gap estimates.

### Main findings — program coverage, methods, reporting, and use
- Coverage and consistency:
  - HMRC’s program is comprehensive in tax coverage; tax gaps are estimated for most taxes administered by HMRC.
  - HMRC produces one of the most comprehensive studies of tax gap estimates internationally.
- Methodological approach:
  - HMRC employs “bottom-up” based estimates for direct tax gaps and “top-down” estimates for indirect tax gaps.
  - Approaches are consistent with international practices; HMRC has led application of some methodologies.
  - Sections II and III identify areas for methodological improvement.
- Reporting and dissemination:
  - Official publications are issued regularly; tax gap estimates are official statistics reported in a comprehensive and transparent manner.
  - Presentation, organization, and communication can be enhanced to reflect the tax gap program’s multiple objectives.
- Use in compliance management:
  - Tax gap estimates and related analyses provide useful information for compliance management.
  - HMRC’s decision to avoid mechanical links between marginal estimate changes and administrative action is sensible given complexity and data caveats.
  - HMRC is using the process to inform management and operational actions.
- Suggested medium-term research areas:
  - Assessing top-down models for income tax gaps.
  - Extending models to assess the size of the policy gap by tax type.
  - Comparing HMRC’s consumption-based VAT gap model with a value-added based VAT gap model.

### Key recommendations (selected)
- Defining the tax gap:
  - Distinguish between compliance and policy components of tax avoidance schemes in reporting the tax gap.
- Measuring the tax gap—direct taxes:
  - Segmentation of taxpayers in random enquiry programs and projection of results should be based on risk profiles.
  - Review the practice of excluding outliers from random audit samples.
  - Wage-levels assumed for ghosts and moonlighters estimates need a stronger basis.
  - Use targeted audit results to check tax gap estimates and assumptions.
  - Estimate domestically determined uplift factors.
- Measuring the tax gap—indirect taxes:
  - Better data needed for the amount of VAT collected on inputs into exempt supplies.
  - Use volume survey data for excise tax gap estimates.
- Reporting the tax gap—presentation and values:
  - Review aggregation of tax gap estimates—both calculation and reporting—for better representation by tax type.
  - Segment tax gap estimates by estimation method and by level of robustness and completeness.
  - Report the gross gap and the net gap in addition to the anticipated net gap.
  - Create a proper accruals report for VAT revenues.
- Using the tax gap:
  - Reporting both gross gap and net gap estimates would improve performance measurement.
  - Continue pursuing use of tax gap to support resource allocation to tackle noncompliance.

### Table 1 — Selected 2011 highlights (proportion of the 2011 Gap)
- Income Tax, NIC, Capital Gains Tax:
  - PAYE: small-medium enterprises (SMEs) — Bottom-up estimate based on random audit results. Proportion of the 2011 Gap: 2%
  - PAYE: large taxpayers — Constructed estimate based on the results for the SMEs. Proportion: 7%
  - Self-assessment: individuals and businesses — Bottom-up estimate based on random audit results. Proportion: 14%
  - Self-assessment: large partnerships — Constructed estimate based on error levels comparable to results for the SMEs. Proportion: 2%
  - Nondeclaration of income by individuals not in self-assessment — Bottom-up estimate based on cross-matching PAYE data with third party information. Proportion: 3%
  - “Moonlighters” — Estimate based on study results. Proportion: 6%
  - “Ghosts” — Estimate based on labor force survey and immigration data. Proportion: 4%
  - Avoidance — Estimate constructed using avoidance schemes being tracked in the "risk register." Proportion: 7%
- Corporation Tax:
  - Large business services (LBS) clients — Constructed estimate based on Tax under Consideration (TuC) data. Proportion: 4%
  - Large and complex businesses — Constructed estimate based on results for LBS clients. Proportion: 4%
  - Small-medium enterprises — Bottom-up estimate based on random-audit results. Proportion: 4%
- VAT:
  - Top down estimate based on consumption statistics (a bottom-up estimate is also performed to determine composition of the gap). Proportion: 30%
- Excises:
  - Alcoholic beverages, Tobacco — Top-down estimate based on consumption statistics. Proportion: 7%
  - Petroleum fuels — Top-down estimate based on travel distance statistics and fleet characteristics, and “cross-border shopping.” Proportion: 1%
- Notes:
  - “There are other components to the total estimate for some of these items, such as the addition of the value of nonpayment; this table only summarizes the main estimation methodology component.”
  - “Total adds to only 95 percent; the minor indirect and direct tax gaps estimates are left out.”

### Defining the tax gap — core concepts
- Common definition: tax gap = difference between current and potential collections (collections a revenue administration should collect given the current policy framework).
- IMF terminology:
  - “Compliance gap” — impact of compliance issues on revenue.
  - “Policy gap” — revenue loss attributable to provisions in tax laws that allow an exemption, a special credit, a preferential rate of tax, or a deferral of tax liability.
- Recommendation: link HMRC’s tax expenditure report estimates with compliance gap estimates to provide a comprehensive picture; recognize overlaps between tax expenditures when defining the policy gap.

---

### Compliance vs Policy Gap: purpose, HMRC approach, and recommendation
- Purpose:
  - Measure impact of both compliance issues and policy choices; policy gap analysis provides useful insights for revenue administration and policy discussions.
- HMRC approach:
  - HMRC defines the tax gap as “...the difference between tax collected and the tax that should be collected (the theoretical liability).”
  - Theoretical liability: “...the tax that would be paid if all individuals and companies complied with both the letter of the law and HMRC’s interpretation of the intention of Parliament in setting law (referred as the spirit of the law).”
  - HMRC explicitly includes revenue at risk associated with identified tax avoidance schemes until courts rule otherwise; where courts decide a scheme is legal HMRC does not count future uses within the tax gap measure; for VAT the theoretical liability is amended following court decisions.
- Accounting for tax avoidance:
  - Until legal ruling, including tax avoidance schemes in the compliance gap is appropriate using the administration’s interpretation.
  - Tax avoidance schemes deemed legal through litigation should be considered part of the policy gap, not the compliance gap.
  - Recommendation: identify tax avoidance schemes ruled legal through litigation as part of the policy gap and not report them as part of the compliance gap.

### Measuring direct tax gaps — framework overview
- Direct tax gap estimates are constructed by compiling results from multiple models/methodologies focused on distinct components.
- Major groups for direct taxation:
  - (i) income tax, NIC, and capital gains tax (noncorporation taxes);
  - (ii) corporation tax;
  - (iii) other direct taxes.
- The report focuses on (i) and (ii), which comprise 96 percent of the direct tax gap.
- For noncorporation taxes three basic estimation models/methodologies:
  1. Random-audit based estimation methodology—the random-enquiry program.
  2. Data-matching estimation methodology—cross-check of information.
  3. Ad-hoc specific taxpayer segment models—primarily to measure tax gap from “hidden” economy.
- Models are supplemented with direct operational data; mapping segments income into wages and salaries, business income, and other income, and layering taxpayer registration and filing coverage.

### Issues in noncorporation tax framework
- Four design criteria (Box 2):
  1. Captures the appropriate tax base.
  2. Covers all potential taxpayers.
  3. Accounts for all potential forms of noncompliance.
  4. No overlap between any two components of the framework.
- Table 2 summary:
  - Criterion 1. Captures the appropriate tax base. — Good. Comment: Missing foreign sources of other income.
  - Criterion 2. Covers all potential taxpayers. — Fair. Comment: Missing nonregistered employers.
  - Criterion 3. Accounts for all potential forms of noncompliance. — Excellent.
  - Criterion 4. No overlap between any two components of the framework. — Excellent.
  - Overall assessment: Good.
- Specific coverage gap: lack of coverage for withholdings of salaries and wages not being declared by nonregistered employers; may be implicitly included in “Moonlighters” and “Ghosts” estimates but not certain; recommendation to design explicit model coverage for unregistered employers.

---

### Random-enquiry based estimates — principles, HMRC assessment, and improvements
- Key principles (Box 3):
  1. Proper definition of the population: include current filers and taxpayers who should be filing.
  2. Risk-based taxpayer segments for sample selection: segment by similar risk profiles (size and sector).
  3. Proper sample selection: adequate sample sizes per segment.
  4. Comprehensive audit: scope cover all audit aspects relevant.
  5. Projection to the population: apply share of undeclared liability from sample to the total declared liability in segment.
  6. Projection to other populations: infer only forms of noncompliance designed to be captured.
  7. Accounting for undetected undeclared liability: apply an “uplift” factor at the taxpayer segment level.
- Assessment (Table 3 summary):
  - 1. Proper definition of the population. — Excellent / Excellent
  - 2. Risk-based taxpayer segments for sample selection. — Fair / Fair
  - 3. Proper sample selection. — Good / Good
  - 4. Comprehensive audit. — Good / Good
  - 5. Projection to the population. — Fair / Fair
  - 6. Projection to other populations. — Fair / Fair
  - 7. Accounting for undetected undeclared liability. — Fair / Fair
  - Overall assessment — Good / Good
- Identified shortcomings and recommendations:
  - Population segmentation: current strata (businesses: four strata by turnover; individuals: seven income classes) should be refined using revealed risk profiles; fact cited: 80 percent of the gap is from business taxpayers.
  - Sample selection and outliers:
    - Current practice excludes “enquiries with exceptionally high yield or tax at risk (outliers)”.
    - Observation: exclusions may be improper; recommendation: for legitimate outliers exclude from grossing-up but include undeclared liability values directly in the gap estimate; improve risk-based segmentation to distinguish true outliers.
  - Projection to other populations:
    - Problem: extrapolating small-medium taxpayer results to large taxpayers without proper adjustment; noted inconsistency where a 1 percent small-medium gap was being applied as a 1 to 2 percent value for large employers with a mid-range 1.5 percent used (implying a 50 percent increase over the base value).
    - Recommendation: model projections appropriately; use gross results for undeclared liabilities for projection.
  - Accounting for undetected undeclared liability (uplifts):
    - Current practice: most uplift factors cited as being from “U.S. Research”; uplifts not used in all cases.
    - Recommendation: adopt more consistent approach; in short term apply U.S. based estimates consistently; longer term develop domestic uplift values.

---

### Data-matching based estimates — principles, HMRC assessment, and improvements
- Key principles (Box 4):
  1. Availability of unique taxpayer identifiers.
  2. Unique identifiers in third party data and ability to map.
  3. Accounting for unmatched data.
  4. Comprehensive coverage of third-party information or complementary methods.
  5. Proper estimation of associated tax gap (micro-simulation preferred).
  6. Accounting for undetected undeclared liability (apply proportion estimate).
- Assessment (Table 4 summary):
  - 1. Availability of unique taxpayer identifiers. — Excellent
  - 2. Availability of unique identifiers in third party data. — Excellent
  - 3. Accounting for unmatched data. — Fair (Unmatched data is being ignored, possibly skewing the results.)
  - 4. Comprehensive scope of coverage. — Fair (Data matching must be supplemented to address lack of coverage, notably foreign-sourced other income.)
  - 5. Proper estimation of the associated tax gap. — Good
  - 6. Accounting for undetected undeclared liability. — Fair
  - Overall assessment — Fair
- Identified shortcomings and recommendations:
  - Unmatched data: current practice ignores unmatched data; recommendation to assess proportion and nature and adopt appropriate treatment (micro-simulation or calibrated averages).
  - Foreign-sourced other income: HMRC lacks access; recommendation to supplement via estimates and strengthened effective exchange of information (EOI) initiatives.

---

### Ad-hoc models for ghosts and moonlighters — scope and concerns
- Overlap risk: models for ghosts and moonlighters may overlap with nonregistered employers not submitting withholdings; attribution between PAYE and self-assessment matters for resource allocation.
- Methodological concerns: strong assumptions on number of participants and wage-levels; wage-level assumptions lack strong supporting evidence.
- Recommendation: extend models to explicitly capture unregistered employers; conduct research to ground wage-level assumptions; refine model design.

---

### Corporation tax gap framework — models, evaluation, and recommendations
- Framework models:
  - (1) random-enquiry based estimation methodology;
  - (2) direct program data (e.g., Tax under Consideration (TuC));
  - (3) ad-hoc estimates.
- Evaluation (Table 5 summary):
  - 1. Captures the appropriate tax base — Excellent
  - 2. Coverage of all potential taxpayers — Excellent
  - 3. Accounts for all potential forms of noncompliance — Fair (Undetected undeclared liabilities are not being accounted for all taxpayers.)
  - 4. No overlap between any two components of the framework — Excellent
  - Overall assessment: Good
  - Notable gap: undetected undeclared liabilities by large corporations not adequately covered.
- Random-enquiry program (Table 6 summary):
  - 1. Proper definition of the population — Excellent
  - 2. Risk-based taxpayer segments for sample selection — Fair
  - 3. Proper sample selection — Fair
  - 4. Comprehensive audit — Good
  - 5. Appropriate projection of the random audit data to the taxpayer population — Good
  - 6. Appropriate scope of projection of results — n/a
  - 7. Accounting for undetected undeclared liability — Fair
  - Overall assessment: Fair/Good
- Ad-hoc estimates and LBS projections:
  - Current method projects LBS technical risk and avoidance to large and complex businesses; produces stark contrasts in avoidance-to-technical-risk ratios (LBS roughly four to one; large and complex roughly one to four).
  - Comparative averages cited:
    - Measuring the Tax Gap 2012 averages: LBS (2006–07 to 2008–09): average GBP 1.4 billion in avoidance and average GBP 0.3 billion for technical risks subject to litigation.
    - Large and complex businesses (2009–10 to 2010–11): average avoidance GBP 0.25 billion and average derived technical risks GBP 1.0 billion.
- Recommendations (selected):
  - Estimate the gap from undeclared withholdings on employment income from nonregistered employers.
  - Segment businesses in random enquiry program by risk profiles.
  - Review exclusion of outliers and incorporate outlier results into the gap directly.
  - Establish “peer” segments between random enquiry and segments to which results are extended.
  - Change projection methods from small-medium employers to large employers so no additional error-bias assumptions are made and use results only to project undeclared liabilities.
  - Use data-matching across broader taxpayer populations.
  - Construct risk profile based segments of LBS businesses to improve peer group projections.
  - Compile targeted audit statistics to compare with gap estimates and use adjusted targeted audit results where random enquiry results are unavailable.
  - Determine domestic uplift factors for corporation tax estimates.

---

### Indirect taxes — VAT and excise models, evaluation, and improvements
- Indirect tax gap estimates mainly use top-down techniques and are designed to capture all possible sources of the tax gap.
- Indirect tax groupings: (1) VAT; (2) excise taxes; (3) other indirect taxes. Review focuses on VAT and excises, which comprise 92 percent of the indirect tax gap.
- VAT methodology (consumption-based top-down):
  - VAT Theoretical Total Liability (VTTL) components:
    - (i) final VAT payable by households on final consumption (consumer expenditure survey data);
    - (ii) VAT payable by government based on department accounts and supply-use tables;
    - (iii) expenditures by exempt sectors based on supply-use tables and surveys on proportion of input used by businesses that is not recoverable.
  - Internal bottom-up estimate supplements top-down to determine composition; bottom-up generally lower than top-down.
- Excise methodologies:
  - Alcoholic beverages, Tobacco — Top-down based on consumption surveys or data; noted conversion difficulties from value to volume for some commodities.
  - Petroleum fuels — Top-down using travel distance statistics, fleet characteristics, and “cross-border shopping”; Great Britain and Northern Ireland treated differently in method.
- Design criteria for top-down models (Box 5): independent source of statistics, accurate statistical data, consistency, sufficiently detailed statistical data.
- Evaluation (Table 8 summary):
  - 1. Independent source of statistics for the tax base. — VAT: Good; Excise: Excellent.
  - 2. Accurate statistical data. — VAT: Good; Excise: Good.
  - 3. Consistency in statistical data. — VAT: Good; Excise: Good.
  - 4. Sufficiently detailed statistical data. — VAT: Good; Excise: Excellent.
  - 5. Comprehensive statistical data. — VAT: Fair; Excise: Fair. Comment: VAT data on consumption by exempt suppliers could be improved; primary data for beer, spirits and wine is value based but tax is volume based.
  - 6. Accurate modeling of the tax structure. — VAT: Good; Excise: Excellent.
  - Overall assessment: Good (VAT) / Good (Excise)
- Recommended improvements — VAT:
  - Improve information on proportion of exempt to total supplies; use source-use tables where current or require taxpayers to report proportion of exempt to total supplies on VAT returns.
  - Improve method for determining proportion of VAT collected on inputs into exempt supplies based on better statistical data.
  - Investigate VAT return design to obtain more detailed taxpayer-reported data (taxable output by rate, exempt output, input tax credits split) to enhance model inputs.
- Recommended improvements — Excise:
  - Compare value data based excise models to volume-based survey data (e.g., General Lifestyle Survey) despite historical under-reporting in volume surveys.
  - Construct bottom-up excise estimates from audit and customs policing data to compare with top-down estimates.
  - Produce results for excise taxes using consumption volume survey data for comparison.

---

### Reporting the tax gap — objectives, presentation, aggregation, and values
- HMRC purposes for publishing tax gap:
  - Transparency, public interest, potential behavioral response, and internal use for operational decisions.
- HMRC practice:
  - Annual publication “Measuring Tax Gaps”; reports nominal terms and percent share of theoretical liabilities, broken down by tax heads and taxpayer behaviors, with error margins and notes on methodological changes; methodological annex provided.
- Aggregation issues:
  - Current aggregation hierarchy reported; recommendation to adopt consistent four-level aggregation:
    - Level One: Total Gap
    - Level Two: Direct Taxes / Indirect Taxes
    - Level Three (Direct Taxes): Income and capital gains tax; NIC; Corporation tax; Other direct taxes
    - Level Three (Indirect Taxes): VAT; Excises; Other indirect taxes
  - Recommendation: isolate estimates and report aggregates by major tax type; segment estimates by robustness/completeness.
- Values and dynamics:
  - Gap values change over time: bottom-up values decrease as debt stocks are reduced or new information is obtained; top-down gaps change with accrual and additional assessments.
  - HMRC hybrid approaches:
    - Direct taxes: older periods use actual compliance yield subtractions; recent periods use projections minus estimated compliance yield.
    - Indirect taxes: adjusted cash basis for collections used (shifting value by three months) as approximation of accrual collections.
  - Issue: netting anticipated future collections prior to reporting undermines transparency and ability to assess performance changes for a given period.
- Recommended reporting distinctions:
  - Report three measures:
    - Gross gap: gap measured at the due date for payment of tax liabilities.
    - Net gap: gap at the time of measurement.
    - Net gap with anticipated collections: net gap plus anticipated future collections (current HMRC reporting).
  - Rationale: presenting gross and net gap enables insight into voluntary compliance and administrative activity needed; explicitly identify portion attributable to assumptions about future collections.

---

### VAT accrual reporting and performance indicator guidance
- VAT gap improvements:
  - Use a true accrual value for VAT gap; current system should be able to track how payments are allocated against liabilities to generate accruals for proper interest calculations.
  - Report gross gap and current net gap in addition to anticipated net gap.
  - Investigate methods for generating a proper accruals report for VAT revenues.
- Tax gap as performance indicator:
  - A valid performance indicator must be measurable, verifiable, free from bias, tied to institution efforts, and based on clear transparent methodology.
  - Challenges:
    - Data limitations (dependence on third-party data and their changes).
    - Error margins large and difficult to quantify; publishing specific margins has risks.
    - Timeliness: tax gap estimates are backward looking and can lag; reliance on projections increases error.
    - Disincentive: bottom-up identification of new noncompliance can increase the measured gap and create perverse incentives.
  - Recommendations:
    - Do not use a single aggregate gap estimate as sole KPI.
    - Use both gross gap and net gap in measuring performance in encouraging voluntary compliance and enforcement.
    - Use tax gap estimates in combination with complementary indicators and intermediate indicators for systematic ‘health checks’.

---

### Tax gap as resource allocation tool
- Uses:
  - Inform risk analyses and distribution of revenue risks.
  - Help estimate marginal return on resources when combined with operational data.
- HMRC progress:
  - HMRC is linking tax gap with risk analyses and resource allocation; tax gap analysis informed Spending Review (SR) assessment.
- Recommendation:
  - Continue developing linkages between tax gap estimates and taxpayer risks; combine gap estimates with other compliance indicators for resource allocation decisions.

---

### Possible future research (medium term)
- Three research areas proposed:
  1. Assessing possible top-down models for the income tax gaps.
  2. Extending HMRC’s models to assess the size of the policy gap by tax type.
  3. Comparing the FAD Revenue Administration Gap Analysis Program (RA-GAP) value-added based VAT gap model to HMRC’s VAT gap model.
- Top-down direct tax model notes:
  - Top-down models can verify bottom-up estimates and estimate the policy gap; HMRC has documented data challenges.
  - Possible approach: use VAT-based proxies (value added identities Y = C + I + (X – M) = W + R and derivations) to approximate corporate income tax base (R) and NR = C – W + (I – D) + (X – M), with caveats on allowances and depreciation.
- Value-added based VAT model:
  - RA-GAP allows breakdown by sector; theoretically equivalent to consumption-based model if national accounts identities hold.
  - Issues: source-use tables lag; both models require projection to produce timely estimates; comparing time series of net VAT by both methods would be useful.
- Measuring the policy gap:
  - Top-down models can be extended by replacing current tax structure with normative version (example: a single-rate VAT with limited exemptions) to estimate policy gap.
  - For income taxes, micro-simulation models are suitable and often already used for tax expenditure analysis.
  - C-Efficiency ratio illustration for VAT: C-Efficiency = VAT Revenue / (Consumption * Standard Rate) as an approximation for combined policy and compliance effects.

---

### Illustrative components and treatment of direct taxes (summary)
- Noncorporation taxes (income tax, NIC, capital gains) use nine specific techniques across general noncompliance and avoidance for employers, self-assessors, and nondeclarants.
- Random enquiry program steps (gross sample results → grossing up → inflation/up-rate → uplift for undetected noncompliance → add nonpayment).
- Avoidance treatment: use stock of “tax at risk” from risk register and assume one third accrues in any given tax period (average length of scheme assumed three years).
- Corporation tax segmentation:
  - LBS (largest 800), large and complex (~9,000), SMEs in Corporate Tax Self Assessed program.
  - LBS derived gaps from TuC with about 50 percent of TuC for technical risks included; reported ratios and average GBP figures (see above).

*Italic: IMF Fiscal Affairs Department assessment of HMRC tax gap analysis program (content unit: _cr13314 as provided in the source).*

### Preface ................................................................................................................

### _cr13314 - Preface ................................................................................................................

### Preface and mission
- Assessment requested by Mr. Jonathan Athow (Director, Knowledge, Analysis and Intelligence Department (KAI)) from Her Majesty’s Revenue and Customs (HMRC).
- Assessment team led by Mr. Juan Toro (Assistant Director, FAD) with Messrs. Kentaro Ogata (Advisor, FAD), Selcuk Caner (Senior Economist, FAD), and Eric Hutton (Technical Assistance Advisor, FAD).
- Main purpose: thorough review of HMRC’s tax gap analysis program, including models and methodologies, use in supporting HMRC operations, and dissemination approach.
- Fieldwork: initial fact-finding visit April 23–25, 2013; follow-up visit held on July 1, 2012; continued work at IMF headquarters and incorporation of HMRC and FAD advisory committee comments.
- Report structure: Executive Summary; (I) Defining the Tax Gap; (II) Measuring the Direct Taxes Gaps—Framework and Methodologies; (III) Measuring the Indirect Taxes Gaps—Framework and Methodologies; (IV) Reporting the Tax Gap; (V) Using the Tax Gap; (VI) Possible Future Research on Tax Gap Analysis.

### Executive Summary — objectives and program scope
- The report assesses HMRC’s tax gap analysis program and provides advice on improving:
  - (1) models and methodologies employed;
  - (2) approach to disseminating results;
  - (3) use of results for compliance activities, revenue performance evaluation, and effectiveness assessment.
- HMRC Vision Statement (Strategic Plan 2012–2015): “We will close the tax gap, our customers will feel that the tax system is simple for them and even-handed, and we will be seen as a highly professional and efficient organization.”
- Three critical objectives implied by the Vision Statement:
  - Measuring tax revenue losses to be closed.
  - Supporting efficiency in closing the tax gap and allocating resources to achieve high impact in reducing noncompliance.
  - Supporting perception of fairness through transparent and sufficiently detailed reporting of tax gap estimates.

### Main findings
- Coverage and consistency:
  - HMRC’s program is comprehensive in tax coverage; tax gaps are estimated for most taxes administered by HMRC.
  - HMRC produces one of the most comprehensive studies of tax gap estimates internationally.
- Methodological approach:
  - HMRC employs “bottom-up” based estimates for direct tax gaps and “top-down” estimates for indirect tax gaps.
  - Approaches are consistent with international practices; HMRC has led application of some methodologies.
  - Sections II and III identify areas for methodological improvement.
- Reporting and dissemination:
  - Official publications are issued regularly; tax gap estimates are official statistics reported in a comprehensive and transparent manner.
  - Presentation, organization, and communication can be enhanced to reflect the tax gap program’s multiple objectives (see Section IV).
- Use in compliance management:
  - Tax gap estimates and related analyses provide useful information for compliance management.
  - HMRC’s decision to avoid mechanical links between marginal estimate changes and administrative action is sensible given complexity and data caveats.
  - HMRC is using the process to inform management and operational actions (see Section V).
- Suggested medium-term research areas (Section VI):
  - Assessing top-down models for income tax gaps.
  - Extending models to assess the size of the policy gap by tax type.
  - Comparing HMRC’s consumption-based VAT gap model with a value-added based VAT gap model.

### Key recommendations (Box 1)
- Defining the tax gap:
  - Distinguish between compliance and policy components of tax avoidance schemes in reporting the tax gap.
- Measuring the tax gap—direct taxes:
  - Segmentation of taxpayers in random enquiry programs and projection of results should be based on risk profiles.
  - Review the practice of excluding outliers from random audit samples.
  - Wage-levels assumed for ghosts and moonlighters estimates need a stronger basis.
  - Use targeted audit results to check tax gap estimates and assumptions.
  - Estimate domestically determined uplift factors.
- Measuring the tax gap—indirect taxes:
  - Better data needed for the amount of VAT collected on inputs into exempt supplies.
  - Use volume survey data for excise tax gap estimates.
- Reporting the tax gap—presentation:
  - Review aggregation of tax gap estimates—both calculation and reporting—for better representation by tax type.
  - Segment tax gap estimates by estimation method.
- Reporting the tax gap—values:
  - Report the gross gap and the net gap in addition to the anticipated net gap.
  - Create a proper accruals report for VAT revenues.
- Using the tax gap—performance measurement:
  - Reporting both gross gap and net gap estimates would improve performance measurement.
- Using the tax gap—resource allocation:
  - Continue pursuing use of tax gap to support resource allocation to tackle noncompliance.

### Table 1 — Summary of Tax Gap Estimates and Methodologies—2011 (selected highlights)
- Income Tax, National Insurance Contributions (NIC), Capital Gains Tax:
  - PAYE: small-medium enterprises (SMEs) — Bottom-up estimate based on random audit results. Proportion of the 2011 Gap: 2%
  - PAYE: large taxpayers — Constructed estimate based on the results for the SMEs. Proportion: 7%
  - Self-assessment: individuals and businesses — Bottom-up estimate based on random audit results. Proportion: 14%
  - Self-assessment: large partnerships — Constructed estimate based on error levels comparable to results for the SMEs. Proportion: 2%
  - Nondeclaration of income by individuals not in self-assessment — Bottom-up estimate based on cross-matching PAYE data with third party information. Proportion: 3%
  - “Moonlighters” — Estimate based on study results. Proportion: 6%
  - “Ghosts” — Estimate based on labor force survey and immigration data. Proportion: 4%
  - Avoidance — Estimate constructed using avoidance schemes being tracked in the "risk register." Proportion: 7%
- Corporation Tax:
  - Large business services (LBS) clients — Constructed estimate based on Tax under Consideration (TuC) data. Proportion: 4%
  - Large and complex businesses — Constructed estimate based on results for LBS clients. Proportion: 4%
  - Small-medium enterprises — Bottom-up estimate based on random-audit results. Proportion: 4%
- VAT:
  - Top down estimate based on consumption statistics (a bottom-up estimate is also performed to determine composition of the gap). Proportion: 30%
- Excises:
  - Alcoholic beverages, Tobacco — Top-down estimate based on consumption statistics. Proportion: 7%
  - Petroleum fuels — Top-down estimate based on travel distance statistics and fleet characteristics, and “cross-border shopping.” Proportion: 1%
- Notes:
  - “There are other components to the total estimate for some of these items, such as the addition of the value of nonpayment; this table only summarizes the main estimation methodology component.”
  - “Total adds to only 95 percent; the minor indirect and direct tax gaps estimates are left out.”

### I. Defining Tax Gaps — core concepts
- Purpose: clarify concepts and terminology; distinguish compliance and policy components.
- Common definition: tax gap = difference between current and potential collections (collections a revenue administration should collect given the current policy framework).
- IMF terminology:
  - “Compliance gap” — impact of compliance issues on revenue.
  - “Policy gap” — revenue loss attributable to provisions in tax laws that allow an exemption, a special credit, a preferential rate of tax, or a deferral of tax liability.
- Appendix I analyzes relationship between compliance issues and policy choices and how this informs potential avenues for improving revenue.

*Source: IMF Fiscal Affairs Department assessment of HMRC tax gap analysis program (Preface and Executive Summary as provided in the source content).*

### 3.      From the perspective of compliance management, it is recommendable to

### _cr13314 - 3.      From the perspective of compliance management, it is recommendable to

### Compliance vs Policy Gap: purpose and measurement
- It is recommendable to measure the impact of both compliance issues and policy choices.
- A more comprehensive analysis of the policy gap could provide useful insights for revenue administration and for policy discussions on tax regime designs.
- HMRC prepares a tax expenditure report that identifies the value of revenue foregone for select policy measures, but it does not, as is standard for a tax expenditure report, provide a value of the joint impact on potential revenue from the full set of tax expenditures working concurrently—this is what a policy gap measure would accommodate.
- To provide a comprehensive picture of revenue performance the HMRC’s work on estimating the value of individual tax expenditures and the estimates of the compliance gap should be linked.
- The policy gap is not simply the sum total of all tax expenditures, as there may be overlap between two individual tax expenditures (example given: zero-rating children’s clothing likely overlaps with zero-rating supplies to charities).

### Interaction of policy choices and compliance management
- The provision of exemptions, special credits, preferential rates of tax, or deferral of tax liabilities can present avenues for some taxpayers to engage in noncompliant activity.
- It is important for a tax administration to monitor revenue foregone through policy choices and compare these trends with estimated revenue losses due to compliance issues.
- Monitoring revenue foregone helps determine if revenue performance is impacted by unanticipated taxpayer take-up of tax instruments; otherwise changes could be misinterpreted as changes in taxpayer compliance.
- The administration should undertake special compliance programs to control the use of special treatments to deter and uncover potential abuses.
- Control activities provide a basis for estimating associated revenue cost and promote changes when abuses are detected.

### The HMRC approach to the tax gap
- HMRC’s tax gap analysis is focused on the compliance gap but is arguably a wider measure as it includes tax avoidance and legal interpretation of the tax laws.
- HMRC defines the tax gap as “...the difference between tax collected and the tax that should be collected (the theoretical liability).”
- HMRC defines the theoretical liability to be: “...the tax that would be paid if all individuals and companies complied with both the letter of the law and HMRC’s interpretation of the intention of Parliament in setting law (referred as the spirit of the law).”
  - This implies HMRC explicitly includes in its tax gap measures revenue at risk associated with identified tax avoidance schemes, whether or not they have ultimately been determined to be legal through litigation.
  - Where the courts decide that a scheme is legal HMRC does not count future uses within the tax gap measure.
  - For VAT the calculation of the theoretical liability is amended following court decisions.
- Given HMRC’s remit to “improve fairness and reduce the scope for evasion or avoidance,” inclusion of tax avoidance in tax gap coverage is appropriate.
- To help properly measure and monitor HMRC performance, tax gap analysis should cover tax avoidance in addition to tax evasion and other noncompliances.

### Accounting for tax avoidance in gap measurement
- Until a legal ruling is available, it is theoretically and practically appropriate to include tax avoidance schemes in the compliance gap.
  - The administration’s interpretation of tax law should be the basis for determining taxable activity or compliant activity until interpretation is overruled through appropriate litigation.
  - Top-down statistical data used in tax gap estimates would not capture activities recharacterized for tax avoidance; thus top-down estimates may de facto include tax avoidance.
  - Explicitly including tax avoidance schemes in bottom-up gap estimates provides consistency across estimate classes.
- Tax avoidance schemes deemed legal through litigation should be considered part of the policy gap, not the compliance gap. Rationale:
  - If a scheme is ruled legal, the only way to recover lost collectable revenue is through a change in the legislation—this is a policy choice.
  - Avoidance schemes deemed illegal fall under the compliance gap.
- Distinguishing legal avoidance (policy gap) from noncompliance would improve clarity and aid allocation of resources:
  - Actions to reduce leakage due to legal tax avoidance schemes require proposing changes to the legislation.
  - Actions for undetermined or illegal schemes require compliance management efforts.
- Recommendation:
  - Tax avoidance schemes ruled legal through litigation should be identified as being part of the policy gap and not reported as part of the compliance gap.

### Measuring the direct tax gaps—framework overview
- This section discusses estimation frameworks and methodologies for direct taxes; Section III analyzes indirect taxes (note: source structure).
- In many cases multiple models or methodologies are combined under a general estimation framework to arrive at the full estimate for a particular tax type.
- For direct taxes the tax gap estimates are constructed by compiling results from a number of models/methodologies focused on distinct components of the gap.
- Tax gaps in direct taxation are estimated mainly using bottom-up techniques across income categories.
- The tax gap estimates can be broken down into three major groups:
  - (i) the gap for income tax, National Insurance Contributions (NIC), and capital gains tax (collectively referred to herein as the noncorporation taxes);
  - (ii) the gap for corporation tax; and
  - (iii) the gap for other direct taxes.
- The report focuses on the first two categories, which comprise 96 percent of the direct tax gap.

### Estimation framework and models for noncorporation taxes
- The tax gap estimation framework for the first grouping relies on three basic estimation models/methodologies:
  1. A random-audit based estimation methodology—the random-enquiry program.
  2. A data-matching estimation methodology—through cross-check of information.
  3. Ad-hoc specific taxpayer segment models—primarily to measure tax gap from “hidden” economy.
- Results from these models are supplemented with direct operational data to complete coverage of potential sources of noncompliance.
- Models estimate tax gaps for several taxpayer types and segments, breaking down “general noncompliance” by potential sources of revenue leakage: employers, self-assessors, and nondeclarants.
- Mapping of models across the tax base/population (described in Figure 1) segments the tax base into three income types:
  - wages and salaries (red),
  - business income (blue),
  - other income (green: capital gains, interest, lettings, etc).
- Potential taxpayer coverage layered over income types includes registration and filing status distinctions (registered or not; filed or not; declared income or not).
- Color-coding in the mapping indicates methodology coverage:
  - purple = random-enquiry program,
  - light green = data-matching,
  - orange = ad-hoc modeling.
- Data on costs of tax avoidance and nonpayment complete the framework:
  - Tax avoidance and nonpayment estimates across all taxpayers use direct program data.

### Issues in the estimation framework for noncorporation taxes
- A framework combining independent models must be comprehensive: components should neither overlap nor have gaps, or overlap must be estimated and subtracted.
- Four design criteria for an effective framework (Box 2):
  1. Captures the appropriate tax base.
  2. Covers all potential taxpayers.
  3. Accounts for all potential forms of noncompliance.
  4. No overlap between any two components of the framework.
- HMRC’s estimation framework is generally sound but has issues with scope of coverage of the tax base and potential taxpayers.
- Table 2 evaluation (summary):
  - Criterion 1. Captures the appropriate tax base. — Good. Comment: Missing foreign sources of other income.
  - Criterion 2. Covers all potential taxpayers. — Fair. Comment: Missing nonregistered employers.
  - Criterion 3. Accounts for all potential forms of noncompliance. — Excellent.
  - Criterion 4. No overlap between any two components of the framework. — Excellent.
  - Overall assessment: Good.
- Specific coverage gap identified:
  - Lack of coverage for withholdings of salaries and wages not being declared by nonregistered employers—Criterion 2.
  - This gap may be implicitly included in estimates for “Moonlighters” (undeclared self-assessment earnings by taxpayers in the PAYE system) and “Ghosts” (undeclared self-assessment earnings by taxpayers not registered for self-assessment and not in the PAYE system); however this is not certain.
  - Explicit model design for these groups is oriented around own earnings.
  - Mapping (Figure 2 description) suggests faded coloration indicating possible gaps in coverage for undeclared liabilities from nonfiling.
  - Undeclared liability from nonfiling registered taxpayers can be estimated through random-enquiry methodology, but only if random audits capture a sufficient sample of registrants who ought to file but have not.
- Note: Box 3 (not reproduced here) is referenced for discussion on features of effective random-enquiry programs.

*Italic: IMF staff report content as provided in the source unit.*

### 18.      For nonfilers, it is not certain whether the random-enquiry program will capture

### _cr13314 - 18.      For nonfilers, it is not certain whether the random-enquiry program will capture

### Random-enquiry based tax gap estimates for noncorporation income taxes — key principles (Box 3)
- Criteria for an effective random-audit based gap estimation methodology:
  1. Proper definition of the population: include current filers and taxpayers who should be filing; excluding nonfilers will exclude undeclared liabilities from nonfilers.
  2. Risk-based taxpayer segments for sample selection: segment population by similar risk profiles (key characteristics: taxpayer size and main sector of activity).
  3. Proper sample selection: adequate sample sizes per segment; trade-off between accuracy and cost.
  4. Comprehensive audit: scope should cover all audit aspects relevant under other audit selection processes.
  5. Projection to the population: apply share of undeclared liability from sample to total declared liability in that segment; group taxpayers by risk profiles for un-sampled segments.
  6. Projection to other populations: only infer forms of noncompliance that the random-audit program is designed to capture (under-declaration of liabilities).
  7. Accounting for undetected undeclared liability: apply an “uplift” factor at the taxpayer segment level to account for undeclared liability not detected by audits.

### Assessment of HMRC random-enquiry programs (Table 3) — summary evaluation
- General statement: HMRC’s random-enquiry programs to assess undeclared liabilities from employers and self-assessors are generally good but present shortcomings.
- Table 3 evaluations (levels preserved):
  - 1. Proper definition of the population. — Excellent / Excellent
  - 2. Risk-based taxpayer segments for sample selection. — Fair / Fair
    - Comment: Business taxpayers’ stratification can be enhanced to improve accuracy of estimates.
  - 3. Proper sample selection. — Good / Good
  - 4. Comprehensive audit. — Good / Good
  - 5. Projection to the population. — Fair / Fair
    - Comment: Better segmentation of the population could enhance the projection.
  - 6. Projection to other populations. — Fair / Fair
    - Comment: Better segmentation of the population could enhance the projection.
  - 7. Accounting for undetected undeclared liability. — Fair / Fair
    - Comment: Not being accounted for in all cases, values being used could be improved.
  - Overall assessment — Good / Good

### Identified shortcomings and recommended improvements for the random-enquiry program
- General: Improvements needed in population segmentation, treatment of outliers, how inferences are applied to other taxpayer segments, and estimating undetected undeclared liability (uplifts).
- Criterion 2 — Population segmentation:
  - Current changes indicated: businesses being selected from four strata based on turnover; individuals selected from seven income classes.
  - Fact: Results to date suggest that 80 percent of the gap is from business taxpayers.
  - Recommendation: Analyze all random enquiry data to identify groups with similar revealed risk profiles and stratify taxpayers by a combination of size and type of activity.
- Criterion 3 — Sample selection and outliers:
  - Current practice: self-assessment documentation states “enquiries with exceptionally high yield or tax at risk (outliers) are not representative of the population and distort the results and so have been excluded from all analysis in this report.”15
  - Observation: Data on number and values associated with excluded outliers suggest improper identification of outliers may be occurring; exclusions often within expected error margins.16
  - Recommendation: For legitimate outliers, exclude from grossing-up factor but include their undeclared liability values directly in the gap estimate; risk-based segmentation should ease true outlier identification.
- Criterion 5 — Projection to the population:
  - Key assumption criticized: using results from one population to infer another assumes similar compliance behavior; problematic when extrapolating from small-medium taxpayers to large taxpayers.
  - Recommendation: Stratify results by risk profiles to improve projection to un-sampled segments.
- Criterion 6 — Projection to other populations:
  - Current issue: results from small-medium employers being improperly extended to large employers; total gap from small used as estimate for large, discounting known information on large-employer noncompliance.
  - Specific numerical inconsistency noted: the estimate of the gap value for the small-medium employers is 1 percent, but in applying the value to the large employers this 1 percent figure is being interpreted as being a gap value of 1 to 2 percent, and so a mid-range value of 1.5 percent being applied, which implies a 50 percent increase over the base value.17
  - Recommendation: Use random-enquiry results from small-medium employers to produce an appropriately modeled estimate for large employers; prefer applying gross results for undeclared liabilities.
- Criterion 7 — Accounting for undetected undeclared liability (uplifts):
  - Current practice: most uplift factors cited as being from “U.S. Research”; uplifts not used in all cases.
  - Recommendation: Adopt a more consistent, coherent approach to estimating and applying uplifts. If using U.S. based estimates, apply consistently across taxpayers in the short term; develop domestically determined uplift values in the longer term.

### Data-matching based tax gap estimates — key principles (Box 4)
- Criteria for an effective data-matching based gap estimation methodology:
  1. Availability of unique taxpayer identifiers.
  2. Availability of unique identifiers in third party data and ability to map to tax authority identifiers.
  3. Accounting for unmatched data: unmatched data may overstate or understate gaps; treatment depends on proportion of unmatched data.
  4. Comprehensive coverage: third-party information ideally universal; if limited, need relative coverage info or complementary methods.
  5. Proper estimation of the associated tax gap: micro-simulation preferred; if unavailable, use average effective rate for taxpayers with similar income/type.
  6. Accounting for undetected undeclared liability: apply estimate of proportion of undetected undeclared liability to detected undeclared liability.

### Assessment of HMRC data-matching program (Table 4) — summary evaluation
- General statement: overall principles appear sound, but shortcomings detected; overall assessment: Fair.
- Table 4 evaluations (levels preserved):
  - 1. Availability of unique taxpayer identifiers. — Excellent
  - 2. Availability of unique identifiers in third party data. — Excellent
  - 3. Accounting for unmatched data. — Fair
    - Comment: Unmatched data is being ignored, possibly skewing the results.
  - 4. Comprehensive scope of coverage. — Fair
    - Comment: The data matching program has to be supplemented to address the lack of coverage.
  - 5. Proper estimation of the associated tax gap. — Good
  - 6. Accounting for undetected undeclared liability. — Fair
    - Comment: Further information on this component is necessary to complete the evaluation.
  - Overall assessment — Fair

### Identified shortcomings and recommended improvements for the data-matching program
- Criterion 3 — Accounting for unmatched data:
  - Current practice: unmatched data is being ignored.
  - Risk: ignoring unmatched data may skew results; treating all unmatched as undeclared tax could overstate gap; ignoring all unmatched could understate gap.
  - Recommendation: Assess proportion and nature of unmatched data and adopt an appropriate treatment (micro-simulation or calibrated averages as per Box 4).
- Criterion 4 — Scope of coverage:
  - Issue: HMRC lacks access to foreign sources of “other income”; domestic third-party data cover a broad range but not nondomestic sources.
  - Recommendation: Supplement data-matching with estimates for undeclared foreign sourced other income, including through strengthened effective exchange of information (EOI) initiatives; current EOI initiatives encouraging but more work needed to obtain sufficiently complete foreign-sourced income data.

### Ad-hoc model-based estimates for noncorporation taxes — ghosts and moonlighters
- Scope and coverage:
  - Observation: models for ghosts and moonlighters may overlap with population of nonregistered employers not submitting income tax withholdings; possible overlap with PAYE tax gap if undeclared wages of employed individuals are being misattributed.
  - Implication: overall gap value may not change materially, but distinction matters for resource allocation (targeting self-employed vs. identifying employers).
  - Recommendation: Extend models to explicitly capture unregistered employers and clarify attribution between self-assessment and PAYE gaps.
- Methodological improvements:
  - Issue: strong assumptions on number of participants and assumed wage levels; number of participants has some foundation, but wage-level assumptions lack supporting evidence or analysis.
  - Importance: these estimates form a very significant portion of the income tax gap and the tax gap as a whole.
  - Recommendation: conduct more research to ground assumptions on wage levels and refine model design.

*Source: IMF staff assessment as presented in the supplied text.*

### 26.      The tax gap estimation framework for the second income tax grouping

### 26.      The tax gap estimation framework for the second income tax grouping

### Framework and models for corporation tax gap estimation
- The corporation tax gap estimation framework relies on three basic models:
  - (1) a random-audit based estimation methodology—the random-enquiry program;
  - (2) direct program data; and
  - (3) ad-hoc estimates.
- Results from these models and methodologies are supplemented with direct operational data to complete coverage of potential sources of noncompliance.
- A mapping of models and methodologies across the general tax base and tax population was prepared for corporation tax. The mapping:
  - Segments coverage by form of noncompliance (bottom pie chart): undeclared liabilities (light blue), avoidance and “technical risk” (red), and other noncompliance (green).
  - Segments coverage by taxpayer population (upper pie chart) with colors identifying model/methodology: purple = random enquiry based estimate; orange = ad hoc model; dark blue = direct program data; transparent = no estimate.

### Key differences vs. noncorporation tax framework
- The corporation tax framework includes a component supplementing direct program data on tax avoidance and technical risk for all taxpayers; for income tax, NIC, and capital gains tax this pertains only to certain taxpayer and income type segments.
- The corporation tax framework covers only one form of income, so no income-type dimension is required (unlike income tax, NIC, and capital gains tax frameworks).

### Evaluation of the corporation tax estimation framework (summary of Table 5)
- Criteria and evaluation:
  - 1. Captures the appropriate tax base — Excellent
  - 2. Coverage of all potential taxpayers — Excellent
  - 3. Accounts for all potential forms of noncompliance — Fair (Undetected undeclared liabilities are not being accounted for all taxpayers.)
  - 4. No overlap between any two components of the framework — Excellent
- Overall assessment: Good
- Notable gap: undetected undeclared liabilities by large corporations are not adequately covered.

### Issues and suggested data checks
- The assumption that tax gaps for large taxpayers arise exclusively from tax avoidance and technical risk should be tested (Criterion 3). A proper estimation should not assume sizes of particular components of the tax gap.
- Data from targeted audits (audits performed during regular operations) could be used to:
  - check the validity of the assumption that undeclared liabilities are negligible in large businesses, or
  - establish an estimate of the possible size of undeclared liabilities.

### The random-enquiry based estimates — evaluation (summary of Table 6)
- Evaluation of the Random Audit Program for Small-Medium Corporation Taxpayers:
  - 1. Proper definition of the population — Excellent
  - 2. Risk-based taxpayer segments for sample selection — Fair (Segmentation is not based on risk profiles)
  - 3. Proper sample selection — Fair (There may be issues with the treatment of outliers and the sample size)
  - 4. Comprehensive audit — Good
  - 5. Appropriate projection of the random audit data to the taxpayer population — Good
  - 6. Appropriate scope of projection of results — n/a
  - 7. Accounting for undetected undeclared liability — Fair (Values could be improved)
- Overall assessment: Fair/Good

### Identified shortcomings and improvements for the random-enquiry program
- Criterion 2: Population segmentation
  - Segmentation should be based on risk profiles (combination of size and type of activity informed by risk-based profiles) to improve sampling efficiency and accuracy.
- Criterion 3: Sample selection
  - Treatment of outliers needs review; sample size should be reassessed.
  - Documentation notes outliers are excluded, but “high tax at risk combined with small sample sizes for the program can result in one large settlement substantially inflating that overall population for that year.”
  - For small sample sizes it is difficult to determine whether an anomalous value is an outlier or representative; when sample size is small, sample distribution may deviate from the population distribution.
  - A review indicates including excluded outliers yields a fairly stable trend, more stable than results excluding outliers.
- Criterion 7: Accounting for undetected undeclared liability
  - A more consistent and coherent approach to estimating and applying uplifts is required.
  - Uplifts can significantly affect final tax gap estimates; research into the general range of domestic values is needed.
  - Preferably uplifts should be determined for individual taxpayer segments (uplift size likely correlated to risk factors).

### The ad-hoc model based estimates — issues and improvements
- Purpose: extend data on technical risk and tax avoidance from the LBS segment to the large and complex business segment to extend coverage.
- Methodological issues:
  - The projection method used could be improved by following guidelines for projecting data from one taxpayer segment to another (as outlined for random audit programs).
  - Segmentation of LBS businesses and large and complex businesses based on risk profiles should be conducted to improve projections.
  - Current approach applies a general level of technical risk and tax avoidance from LBS businesses to large and complex businesses, producing stark contrasts in avoidance-to-technical-risk ratios:
    - LBS businesses: ratio roughly four to one (avoidance to technical risk).
    - Large and complex businesses: ratio roughly one to four (avoidance to technical risk).
  - Suggest identifying more targeted source LBS business segments and target large and complex business segments to improve projection accuracy.
- Comparative values cited:
  - Measuring the Tax Gap 2012 averages:
    - LBS (2006–07 to 2008–09): average GBP 1.4 billion in avoidance and average GBP 0.3 billion for technical risks subject to litigation.
    - Large and complex businesses (2009–10 to 2010–11): average avoidance GBP 0.25 billion and average derived technical risks GBP 1.0 billion.

### Recommendations (as listed)
- An estimate of the gap from undeclared withholdings on employment income from nonregistered employers is needed.
- Segmentation of businesses in the random enquiry program should be based on risk profiles.
- The practice of excluding outliers from the random audit samples should be reviewed, and outlier results should be incorporated into the gap directly.
- Establish “peer” segments between businesses covered under the random enquiry program and those to which the results are being extended.
- Change the manner in which results from small-medium employers are projected to large employers so that:
  - the projection makes no additional assumption as to error bias, and
  - results are used only to project undeclared liabilities.
- The data-matching exercise should be used across a broader segment of the taxpayer population, not just for taxpayers not registered for self-assessment.
- Conduct research to better establish a basis for wage-level assumptions used in estimates for ghosts and moonlighters.
- Construct risk profile based segments of LBS businesses to allow better establishment of “peer” groups between LBS and large and complex businesses.
- Compile statistics on targeted audit results to use as a basis for comparison with tax gap estimates (targeted audit data should establish a floor level for any estimate).
- Use targeted audit results where random enquiry results are not available, once adjusted for selection bias associated with targeted audit results.
- Test and prove the assumption that there is no tax gap other than tax avoidance or technical risk for LBS taxpayers.
- Determine domestic uplift factors to be used in tax gap estimates for corporation taxation.

### Transition to indirect taxation (overview)
- The report transitions to indirect tax gaps, noting the framework and model are essentially one piece for indirect taxes; model designed to capture all possible sources of the tax gap.
- Indirect tax gap estimates mainly use top-down techniques.
- Indirect tax gap estimates are broken into three major groupings:
  - (1) the gap for VAT;
  - (2) the gap for excise taxes; and
  - (3) the gap for other indirect taxes.
- The review focuses on VAT and excise taxes, which comprise 92 percent of the indirect tax gap.

### Methodologies for VAT and excise taxes (summary of Table 7)
- VAT:
  - A top-down approach based on consumption statistics estimates the VAT Theoretical Total Liability (VTTL).
  - VTTL components:
    - (i) estimate of final VAT payable by households on final consumption, based on consumer expenditure survey data;
    - (ii) VAT payable by government based on department accounts and supply-use tables;
    - (iii) expenditures by exempt sectors based on supply-use tables and surveys on proportion of input used by businesses that is not recoverable.
  - Adjustments for special treatment of housing and charities and reimbursements under special relief programs or schemes.
  - For internal purposes a bottom-up estimate supplements the top-down estimate and allows disaggregation by aspects of noncompliance; the bottom-up estimate is generally lower than the top-down estimate.
- Excises: Alcoholic beverages, tobacco
  - Top-down estimates based on surveys or consumption data published by the Office of National Statistics determine total consumption.
  - Tax gap base: difference between total consumption and tax-paid level of consumption.
  - Conversion difficulties (e.g., wine expenditure value to consumption volumes) have required development of a new model with a slightly more complex estimation methodology, but still following the top-down approach.
- Excises: Petroleum fuels
  - Top-down estimates based on travel distance statistics, fleet characteristics, and “cross-border shopping.”
  - Two methods by location:
    - Great Britain: composite estimate of domestically purchased fuel versus cross-border shopping; domestic consumption estimated using fleet composition, distance travelled, and fuel efficiency; tax gap = estimated total consumption + estimated cross-border shopping − duty paid consumption − licit cross-border shopping.
    - Northern Ireland: estimate based on Great Britain estimates and estimates of relative market shares.

### Design criteria for effective top-down gap estimation methodology (Box 5)
1. Independent source of statistics for the tax base:
   - Rely on good statistics on the size of the tax base derived from sources other than taxpayer records; check nominally independent data sources to ensure they do not rely on tax records to impute critical missing values or control totals.
2. Accurate statistical data:
   - Statistical data needs to be relatively accurate, with detailed documentation on compilation methods and, ideally, an indication of the estimation error.
3. Consistency in statistical data:
   - When combining statistical data from different sources, ensure consistency in definitions and scale; verify survey consumption levels are consistent with national accounts and that no definitional changes occurred over time.
4. Sufficiently detailed statistical data:
   - Statistical data must be sufficiently detailed to model the policy framework; commodities and sectors should be disaggregated to match the most detailed definitions in tax rate schedules.

*Source: IMF team based on HMRC publications (content unit: _cr13314 - 26.      The tax gap estimation framework for the second income tax grouping).*

### 5. Comprehensive statistical data: The statistical data used needs to cover the full tax

### _cr13314 - 5. Comprehensive statistical data: The statistical data used needs to cover the full tax base for the tax type.

### Data and model quality for indirect taxes
- Summary finding: Overall the models for the VAT and the excise taxes meet the criteria for effective design, but there are areas for improvement. Table 8 summarizes the appraisal of how the indirect taxation framework meets the criteria outlined in Box 5.
- Table 8. Evaluation of the Indirect Tax Top-down Based Estimates (criteria, evaluation for the VAT Model, evaluation for the Excise Tax Models, comments)
  - 1. Independent source of statistics for the tax base. — VAT: Good; Excise: Excellent. Comment: Some HMRC based data is used to supplement third-part data.
  - 2. Accurate statistical data. — VAT: Good; Excise: Good.
  - 3. Consistency in statistical data. — VAT: Good; Excise: Good.
  - 4. Sufficiently detailed statistical data. — VAT: Good; Excise: Excellent.
  - 5. Comprehensive statistical data. — VAT: Fair; Excise: Fair. Comment: The VAT data on consumption by exempt suppliers could be improved; the primary data used for beer, spirits and wine is value based but the tax is volume based.
  - 6. Accurate modeling of the tax structure. — VAT: Good; Excise: Excellent. Comment: The modeling of how tax accrues on inputs to exempt supplies could be improved for the VAT.
  - Overall assessment: Good (VAT) / Good (Excise)
  - Note: Evaluation levels = excellent, good, fair, poor, missing; n/a = not available.

### Improvements suggested — VAT gap model
- Criterion 5: Comprehensive statistical data
  - Improve information on the proportion of exempt to total supplies. Current input: HMRC survey.
  - Ideal source: calculate proportion based on data from source-use statistical tables to improve independence of data sources and likely quality.
  - Alternative when source-use tables are perceived outdated: require taxpayers to record the proportion of exempt to total supplies on their VAT return.
  - Rationale: Data that taxpayers are required to report on a return would probably be more reliable than information obtained through HMRC conducted surveys.
  - Additional point: Data reported on returns may better capture actual apportioning methodologies used by taxpayers compared to source-use table based modeling, which is largely restricted to apportioning potential input tax credits based on the proportion of exempt output.
  - Footnote: For this purpose current source-use tables are not required, they only need to be reasonably current such that they reflect current business practices. Tables produced within the last three to four years would likely have proportional values for a sector that are reasonably representative of current business activity (barring any significant price shocks). (Footnote retained as in source.)

### Improvements suggested — Excise gap models
- Criterion 5: Comprehensive statistical data
  - Compare results from value data based excise models to volume-based survey data.
  - Recognize HMRC historical issues with volume reported data, including under-reporting; nevertheless illustrative calculations using volume based survey data (such as provided in the General Lifestyle Survey from the Office of National Statistics) should be produced and compared against value based calculations.
  - Construct bottom-up estimates for excise taxes to compare with top-down estimates:
    - Bottom-up should be compiled based on audit data for domestic producers on under-declaration or nondeclaration.
    - Include audit and policing data from customs on under-declaration, mis-declaration, or nondeclaration.
  - Recommendation (explicit): Bottom-up estimates of the excise taxes should be constructed in order to compare and contrast with the results from the top-down estimates.

### Additional recommendations on data and modeling
- The method for determining the proportion of VAT collected on inputs into the production of exempt supplies should be based on better statistical data.
- Results for excise tax based on consumption volume survey data should be produced for comparison and contrast to the expenditure value based survey data.
- Comment on VAT return design: The VAT return in use by the HMRC has been over-simplified. In most countries with a VAT taxpayers are required to provide information on the level of their taxable output for the period by tax rate, and their level of exempt output, detail on input tax credit claims is also generally required to be reported such as the amount of input tax credit being claimed that was paid on imports versus that paid to domestic suppliers. As this is all data that a taxpayer would need to have compiled in order to compute their liability, having the taxpayer report it on the tax form would have a negligible impact on the taxpayer’s compliance costs while providing substantial benefit to the HMRC. This would also provide for better data than a survey, as the reporting requirements for tax return data are typically more stringent than for a survey, and the sample size would be larger (all filing taxpayers).

### Reporting the tax gap — objectives and HMRC approach
- Purposes HMRC identifies for publicizing tax gap estimates:
  - Transparency: HMRC believes information used in high level operational decision making should be transparent.
  - Public interest: The Information Commissioner has ruled that the tax gap estimates are a matter of public interest and thus should be published because disclosure will facilitate public debate and enable the public to assess HMRC's performance. As the estimates are published as Official Statistics, HMRC has to abide by the Code of Practice for Official Statistics.
  - Taxpayer behavioral response: HMRC expects that showing most tax liabilities are being collected and most peer taxpayers pay the tax due could positively affect taxpayers’ compliance behavior.
  - Internal use for operational decisions: Tax gap analyses provide information to help make operational decisions; analyses should be properly documented and shared transparently with those affected by operational decisions.
- Communication strategy: Because gap analyses serve multiple purposes, a clear communication strategy is important. Operational gap analyses may not always be relevant for affecting taxpayer behavior; conversely, considerations to affect behavior should not discourage candid assessments of tax gaps.

### Presentation of tax gap results — HMRC practice and critiques
- HMRC reporting practice:
  - Reports results annually in the “Measuring Tax Gaps” publication.
  - Reports tax gap estimates in nominal terms and in percent share of theoretical liabilities, broken down by tax heads and taxpayer behaviors, with error margins and known biases, historical series, and notes on methodological changes.
  - Publication includes a methodological annex sufficiently detailed to assess validity; on-line information is also available.
- Limitations and suggestions:
  - HMRC does not publish how tax gap estimates are to be used in making operational decisions; this is reasonable as operational changes should not be mechanically linked to changes in the gap.
  - The method by which tax gap estimates are being aggregated should be reviewed due to general inconsistencies in aggregation across the report.
  - Recommendation: Use a consistent manner to report the tax gap to improve transparency and to provide a better breakdown of results for informing resource allocations.

### Aggregation — current practice and suggested hierarchy
- Table 9. Current Apparent Aggregation Hierarchy (as reported)
  - Level One: Total Gap
  - Level Two: Direct Taxes / Indirect Taxes
  - Level Three (examples): Income, NIC, and capital gains tax; Corporation tax; VAT; Excises and other indirect Taxes
  - Level Four / Five: Further breakdowns including breakdown by individual estimation components, and specific excise duties such as Beer duty, Spirits duty, Cigarette duty, Hand-rolled tobacco duty, Great Britain diesel duty, etc.
- Suggested aggregation (Table 10) — four basic levels of aggregation to be applied consistently:
  - Level One: Total Gap
  - Level Two: Direct Taxes / Indirect Taxes
  - Level Three (Direct Taxes): Income and capital gains tax; NIC; Corporation tax; Other direct taxes
  - Level Three (Indirect Taxes): VAT; Excises; Other indirect taxes
- Recommendation bullets:
  - The method by which the gap estimates are being aggregated, both in their calculation and reporting, should be reviewed to have isolated estimates and reported aggregates by major tax type.
  - The tax gap estimates should be segmented by the level of robustness and completeness (e.g., grouping gap estimates with similar level of margins of error).

### Values for the tax gap — dynamics, measurement approaches, and issues
- Dynamics of gap values:
  - The value of the gap can change over time; for bottom-up estimates the value for a given period will tend to decrease on subsequent measurements as debt stocks are reduced or new information is obtained.
  - For top-down estimates the gap will change over time as revenue accrues through collection of arrears or additional assessments.
  - The dynamic nature affects relative measures (percent of potential revenue or percent of GDP), which are also dynamic.
- HMRC’s hybrid approaches:
  - For direct taxes: use differing values for collections depending on tax period. For older periods actual compliance yield (actual collections against identified liabilities) are subtracted from the estimated tax gap. For more recent periods gap projections are used, with estimated compliance yield subtracted.
  - For indirect taxes: an adjusted cash basis for collections is used. The adjusted cash collection—shifting the value by three months—is used as an approximation of accrual collections.
- Issues identified:
  - Assumptions of future collections performance undermine using the gap as a means to assess performance changes over time and reduce the gap estimates’ ability to show how revenue performance has changed for a given period.
  - While pragmatic for dealing with data issues (e.g., lack of accruals data for the VAT) and producing comparable measures across periods, netting anticipated compliance yields prior to reporting may do more harm than good to transparency.
- Recommended reporting distinctions:
  - Report the tax gap in three manners to provide nuanced breakdowns:
    - Gross gap: the gap as measured at the due date for payment of tax liabilities.
    - Net gap: the gap at the time of measurement.
    - Net gap with anticipated collections: the net gap plus anticipated future collections (how HMRC is currently reporting).
  - Rationale: Presenting gross and net gap enables insight into taxpayer compliance behavior and the nature of administrative activity needed; explicitly identifying the portion of the gap attributable to assumptions about future potential collections would increase transparency and allow tracking of current and projected collection performance.
  - Note in source: While the definition for the gross gap is somewhat narrow in definition—a broader definition would be the gap in the absence of any active or passive intervention by the administration—it has the advantage of being practical; measuring this gap is a relatively simple matter of identifying those payments made on-time.

*Source: _cr13314 - 5. Comprehensive statistical data: The statistical data used needs to cover the full tax base for the tax type.*

### 56.      The estimates for the VAT gap could be improved, and made more consistent

### 56.      The estimates for the VAT gap could be improved, and made more consistent

### Improving VAT gap estimation and accrual reporting
- The VAT gap estimates could be improved and made more consistent with gap estimates for other major tax types by using a true accrual value.
- The current system does not provide a report on how VAT payments are accruing, but it should be capable of producing such a report—the system must be able to track how payments are being allocated against liabilities in order to do proper interest calculations for payments in arrears.
- Recommendations:
  - Values for the gross gap and the current net gap should be reported, in addition to the anticipated net gap figure currently used.
  - Methods for generating a proper accruals report for VAT revenues should be investigated.

### A. Tax Gap as Performance Indicator for Revenue Collection
- A tax gap estimate has characteristics suitable for a performance indicator linked to pre-determined consequences: it must be measurable, verifiable and free from bias, tied to the institution’s efforts in meeting objectives and goals, and based on a clear and transparent methodology acceptable to those who monitor performance.
- A tax gap is directly linked to the tax administration’s primary objective—to collect taxes—and is a quantified figure (absolute amount or scaled to relevant tax base or overall economic size).
- Challenges in using a tax gap estimate as a performance indicator:
  - Data limitations:
    - The usefulness of a tax gap comes from comparing actual tax collection with an objective estimate of tax potential using (preferably) independent third-party data.
    - Reliance on third-party data means the gap estimate inherits problems and limitations of those data sources; changes in third-party data can change the gap estimate independently of institutional performance.
    - Wherever feasible, cross-check the gap estimate against alternative estimates that in theory should produce the same gap number.
  - Error margins:
    - Any tax gap estimate—even the most developed and sophisticated model—has a potentially large margin of error, difficult to precisely quantify; standard statistical methods are generally of limited use.
    - One must assess carefully whether changes (or differences) in the estimate are due to spurious factors or real ones.
    - Improving accuracy through model expansion may not be worthwhile if it does not materially affect the estimate relative to the margin of error (examples given: improving the estimate from 20.0 ± 3 to 19.5 ± 3 may not be worth the effort; or from 20 ± 2 to 19 ± 4 where error margin change outweighs apparent precision improvement).
    - Footnote discussion: margins around levels can be systematically biased; error in year-on-year changes can be much less than error in estimated levels. Publication of specific margins of error has risks; broad indication of margins could still be useful.
  - Timeliness:
    - Tax gap estimates are generally backward looking and can have significant time lags.
    - Lag varies by data source; measures relying on detailed statistical data have greatest lag.
    - Even tax record based measures can lag because taxpayers might have up to twelve months after the end of a period to file returns, and then data must be captured, processed, and analyzed.
    - Current estimates often rely on forecasting and thus are an estimate of an estimate, substantially increasing error margin.
  - Disincentive to identify new noncompliance (bottom-up approach):
    - Bottom-up approaches rely on current knowledge of noncompliance behaviors; identifying more sources of noncompliance increases the estimated gap.
    - If an increased gap is interpreted as indicating lower administration performance, this creates a perverse incentive not to identify new noncompliance behaviors.
    - It is possible to split the gap into previously known noncompliance and newly identified noncompliance, but this complicates the framework.
- Using a single gap estimate as sole KPI:
  - A single gap estimate as the sole Key Performance Indicator on compliance and/or administration efficiency could be misleading due to potentially large margins of error.
  - Small changes in the number (within a country) and/or small differences across countries should not lead to strong conclusions.
  - Assessing compliance and administration efficiency requires comprehensive analysis of several related indicators (sub-indicators).

### Issues observed and recommendations for using tax gap estimates
- HMRC practice:
  - HMRC has stopped using the aggregate tax gap estimate as a KPI; it is used to help make strategic decisions and business planning, including assessing scale and direction of new investments.
  - Given data limitations and margins of error, the aggregate estimate alone should not be the primary KPI for administrative performance; used in combination with complementary indicators it can serve as a performance indicator.
  - HMRC uses the aggregate tax gap and sub-components to assess whether operational strategies are broadly generating expected results (a ‘health check’), but could perform this health check more systematically by using intermediate indicators and related gap analyses more explicitly.
- Distinguishing voluntary compliance and enforcement:
  - HMRC’s aims to encourage voluntary compliance and to crack down on deliberate noncompliance suggest both components should be identified in tax gap measurement.
  - Gross gap captures degree to which administration encourages voluntary compliance (filing and paying on time).
  - Net gap captures additional collection efforts needed by the administration.
- Bottom-up measurement limits:
  - Changes in tax gap estimates based on bottom-up techniques should not be the sole basis for conclusions on compliance levels; bottom-up measures only capture risk factors identifiable by the administration.
  - A small estimated gap does not necessarily mean the true gap is small (example: an administration incapable of detecting any noncompliance would estimate the gap at zero).
- Recommendations:
  - In measuring HMRC’s performance in encouraging voluntary compliance and enforcing collection, both gross gap and net gap estimates should be employed.
  - Tax gap estimates, particularly bottom-up methods, should not be the sole basis for drawing conclusions on changes in taxpayer compliance.

### B. Tax Gap as a Resource Allocation Tool
- A performance measurement indicator should be integrated into a broader performance management framework with consequences for over- and under-performance; management can use performance measurement to reallocate resources, refocus activities and assess staff performance.
- Tax gap measurement (and its analyses) can inform:
  - Risk analyses:
    - Tax gap analyses help understand underlying causes of the gap and distribution of revenue risks, which can inform resource reallocation.
    - Information obtained in the process (e.g., random enquiry program) can be useful more generally within the tax administration.
  - Marginal return on resource (marginal effectiveness of additional investment):
    - A tax gap estimate by itself cannot inform effectiveness of individual activities, but combined with operational information (resource allocations, operational strategies) it might be possible to establish link between operational input and compliance outcome.
- HMRC progress:
  - HMRC is linking tax gap with risk analyses and resource allocation; this process has informed operational management such as the preparation of the Spending Review (SR)—tax gap analysis allowed HMRC to “assess whether the overall balance of the SR package was proportionate and whether it was sufficiently ambitious.”
  - HMRC should continue pursuing this development.
- Recommendation:
  - HMRC should continue pursuing the use of tax gap to support resource allocation to tackle noncompliance, specifically:
    - Further development of linkages between tax gap estimates and taxpayer risks.
    - In assessing taxpayers’ compliance, combine tax gap estimates with other compliance indicators.

### VI. Possible future research on tax gap analysis
- Three possible research areas to improve HMRC’s tax gap analysis in the medium term:
  1. Assessing possible top-down models for the income tax gaps.
  2. Extending HMRC’s models to assess the size of the policy gap by tax type.
  3. Comparing and contrasting the output of the FAD Revenue Administration Gap Analysis Program (RA-GAP) value-added based VAT gap model to results from HMRC’s VAT gap model.

### A. Top Down Direct Tax Gap Models
- A top-down model could verify broad validity of bottom-up estimates and estimate the policy gap.
- HMRC has documented serious modeling challenges and significant data issues for top-down models.
- Benefits:
  - Results from a top-down model can serve as a point of comparison to improve overall analysis of size and trends in direct tax gaps.
  - Costs of doing the exercise must be assessed to determine whether benefits outweigh costs.
- Possible approach:
  - Start with top-down VAT gap analysis and expand it as a proxy for corporate income taxation.
  - VAT is tax on value added (Y) which comprises labor income (W; wage and salaries, etc.) and gross capital income (R; corporate profit, etc.), therefore VAT base less wage (Y – W = R) could be used as a proxy of corporate income tax.
  - Identities presented:
    - Y = C + I + (X – M) = W + R
    - ∴ R = C – W + I + (X – M)
    - Net tax base (NR) net of deductible depreciation expenses (D): NR = C – W + (I – D) + (X – M)
  - Investment is added back (treatment under VAT—fully creditable—is equivalent to full expensing under corporate income taxation), but adding full investment overestimates the tax base because depreciation (D) must be taken into account.
  - The last term—the net export—is added back because export is taxable output and import is deductible costs under corporate income tax, whereas export is zero-rated and import is taxable under VAT.
- Key challenge:
  - Accounting for allowances specific to individual taxpayer circumstances requires reliable third-party data that are often unavailable; hence top-down models may not be practical as a primary gap indicator for direct taxation, though the route merits closer examination.

### B. Value-Added Based VAT Gap Model
- Using the RA-GAP VAT gap model allows breakdown of the VAT gap by sector of economic activity, useful for compliance management.
- Advantage over traditional consumption-based VAT gap model: results compiled by sector of economic activity permit decomposition of the gap by sector.
- Theoretically, the value-added based approach used in RA-GAP should produce the same results as a consumption statistics based model used by HMRC if national accounting identity holds between consumption in the consumption model and outputs and inputs in the value-added model.
  - Relationship noted: C [+G] = Y - I - X + M [- G]
  - Consumption-based approach corresponds to left-hand side; value-added based approach corresponds to right-hand side.
- Comparing models:
  - Analyzing differences between the two model outputs could be extremely useful for improving models.
  - Issue with RA-GAP: primary data source (source-use statistical tables) are usually produced with a few years lag.
  - To produce timelier estimates, RA-GAP inflates estimated net VAT by sector using time series data on GDP by economic activity.
  - Consumption statistics used by HMRC are not fully current either; both models require projection to produce up-to-date potential collections estimates.
  - It would be beneficial to compare the time series of net VAT estimated by both methods and the impact on estimated potential collections over time.

### C. Measuring the Policy Gap
- (Section heading present; content beyond this point not included in supplied text.)

*IMF staff assessment excerpt from _cr13314 - 56.      The estimates for the VAT gap could be improved, and made more consistent*

### 74.      Producing estimates of the policy gap, in addition to the compliance gap, would

### _cr13314 - 74.      Producing estimates of the policy gap, in addition to the compliance gap, would

### Estimating the policy gap: purpose and distinction
- Producing estimates of the policy gap, in addition to the compliance gap, would be useful for analyzing revenue performance as a whole and for providing context to assessing the size of the compliance gap.
- The tax gap is comprised of both a compliance gap component and a policy gap component.
- HMRC is currently only including a portion of the policy gap in the estimates (tax avoidance); some of the models in use could be extended to estimate the full policy gap.
- Tax expenditure estimates identify the revenue foregone due to particular elements of the tax system, while the policy gap measures the net effect on potential revenue of the interaction of all the elements in the tax structure.

### Methodological approaches to estimating the policy gap
- Top-down models:
  - In general top-down models can be easily extended to estimate the policy gap.
  - Top-down models generally involve creating an estimate of potential revenue by modeling how the current tax applies to the tax base.
  - Modeling the policy gap requires replacing the current tax structure in the model with a normative version of the tax structure.
  - Example normative VAT: a single rate tax structure with exemptions limited to financial services, and zero-rating limited to exports.
- Taxes without existing top-down models:
  - New models might be required to estimate the policy gap.
  - For income taxes, a micro-simulation model would be a suitable tool for estimating the policy gap.
  - Such models are likely to already exist and be used for tax expenditure estimates and policy analysis.

### Relationship to tax-expenditure estimation and revenue indicators
- Estimation of the policy gap is a related but separate exercise from the production of tax expenditure estimates.
- Execution and results will differ despite both being based on comparing the current tax structure to a normative structure.
- Example indicator: C-Efficiency ratio for VAT defined as:
  - C-Efficiency = VAT Revenue / (Consumption * Standard Rate)
  - (Consumption * Standard Rate) is an approximation for full potential revenue for the tax—i.e., covering both policy and compliance—so the c-efficiency measure is a general approximation of the impact of both on VAT receipts.

### Illustrative framework (components of the tax gap)
- The report illustrates the relationship between compliance issues and policy choices on tax revenues, allowing comparison of the relative extent receipts are influenced by compliance issues or tax policy choices.
- Definitions used in the illustration:
  - Actual Collections
  - Compliance Gap: Uncollected revenue due to taxpayer noncompliance
  - Policy Gap: Foregone revenue due to policy design
  - Current Potential Revenue
  - Full Potential Revenue

### Estimation models and methodologies for direct taxes (income tax, NIC, capital gains)
- Coverage and composition:
  - Two income tax groupings described: income tax, NIC, and capital gains; and corporation tax.
  - These two groups represent 96 percent of the direct tax gap.
- Income tax / NIC / capital gains: high-level methods
  - There are nine separate specific techniques in place for estimating the tax gap for this group, broken down into two major categories (general noncompliance or avoidance) across three major taxpayer types (employers, self-assessors, nondeclarants).
  - General noncompliance methodologies are based on:
    - Results from direct random audit
    - Results from random audits for another taxpayer segment
    - Data matching
    - Statistical estimates of the size of the potential tax base for the segment
  - Random enquiry program approach:
    - Gross results for the sample population up to the full population (based on the relative amount of tax liability)
    - “Up-rate” the value to take into account the lapse in time (using trends in the national accounts statistics for Gross Operating Surplus)
    - Apply an “uplift” factor to account for undetected noncompliance to arrive at estimated under-declared liabilities
    - Add data on nonpayment by the taxpayer segment for the tax period to arrive at the total amount of general noncompliance
  - Larger businesses and partnerships not included in random enquiry program:
    - Assumed to have roughly the same level of noncompliance as the average results for the taxpayers of the same taxpayer type under the random enquiry program; add nonpayment to this estimate
  - Nondeclarants (“Hidden Economy”):
    - Third party sources of information form the basis of the estimates (statistical data or using data obtained from other parties involved in the transactions)
- Avoidance estimation:
  - Same data source and method for all taxpayers: the stock of tax at risk recorded in the “Risk Register”
  - The register includes data on taxpayers involved in known tax avoidance schemes and the impact on their tax liability
  - It is assumed that only one third of the registered “tax at risk” accrues in any given tax period (i.e., it is assumed the average length of the tax avoidance scheme is three years)

### Summary table highlights for noncorporation taxes (methodology mapping)
- General Noncompliance:
  - Employers (up to 250 employees): PAYE, NIC — Random enquiry, grossed up, inflation, uplift, plus nonpayment
  - Employers dealt with by Large Business Service or “Large and Complex”: PAYE, NIC — Assumed same risk as average employers in random enquiry, plus nonpayment
  - Self-assessors (‘Business taxpayers’ up to 4 partners): Income Tax, Capital Gains Tax, NIC — Random enquiry, grossed up, inflation, uplift, plus nonpayment
  - Self-assessors (‘Nonbusiness taxpayers’ individuals and trusts): Income Tax, Capital Gains Tax — Random enquiry, grossed up, inflation, uplift, plus nonpayment
  - Partnerships with 5 or more partners: Income Tax, Capital Gains Tax, NIC — Assumed same risk as self-assessors in random enquiry, plus nonpayment
  - Nondeclarants (employees and pensioners with undeclared nonemployment income): Income Tax, Capital Gains Tax — Data matching between bank records and income tax returns, statistical data on “lettings”
  - Moonlighters: Income Tax — Survey results from other countries
  - Ghosts: Income Tax — Statistical data on labor force survey and immigration data
- Avoidance:
  - All taxpayers: Income Tax — Data from the risk register on identified avoidance schemes, annualized

### Corporation tax: segmentation and methods
- Corporate tax gap segmented into three sub-groups:
  - The largest 800 corporations monitored and audited by the Large Business Service (LBS)
  - Around 9,000 smaller businesses classified as large and complex
  - SMEs covered by the Corporate Tax Self Assessed program
- Methodologies by sub-group:
  - General NonCompliance:
    - Large Business Services clients: Assumed to be zero
    - Large and complex business: Nonpayment
    - Small-medium sized enterprises: Random enquiry for undeclared liability, grossed up for the population and inflation, times an “uplift,” plus nonpayment
  - Avoidance and “technical risks”:
    - Businesses handled by LBS: Data from the LBS case management system on the amount of “Tax under Consideration” (TuC)
    - Large and complex businesses: Assumption that the risk is the same as the average for businesses in the LBS, plus nonpayment
    - Small-medium sized enterprises: Data from the risk register on identified avoidance schemes, annualized
- Treatment and assumptions:
  - For LBS businesses, the tax gap is derived from the value of TuC for a given period, after removing any value of TuC associated with technical risks that will not be subject to litigation.
  - For large and complex businesses, the gap is estimated based on the relative size of the gap (as compared to total tax liability) for the LBS businesses.
  - For the LBS the ratio of avoidance to technical risk is roughly five to one, while for the large and complex businesses it ends up being one to four.
  - Note: Generally about 50 percent of the value of TuC from technical risks appears to be included.
- Reported illustrative values (average figures cited):
  - LBS average values for 2006–07 to 2008–09: BRP 1.4 billion in avoidance to an average of BRP 0.3 billion for technical risks subject to litigation.
  - Large and complex businesses average values for 2009–10 to 2010–11: BRP 0.25 billion for avoidance, derived value BRP 1.0 billion for technical risks.

*Prepared by the IMF team based on HMRC publications.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13314.pdf_