## tarea2025009-print-pdf

## Source details

**Canonical URL:** [tarea2025009-print-pdf](https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025009-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tar/2025/english/tarea2025009-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/tar/2025/english/tarea2025009-print-pdf.pdf.json)

---

### Executive Summary — Core conclusion, methods, and next steps
- Core conclusion:
  - SRC has sufficient quality data available from operational audits to assess the PIT/SSC gaps.
- Methods and focus:
  - Two main methods for estimating tax gaps: top-down and bottom-up; each has strengths and weaknesses and are complementary.
  - Top-down gap estimates have already been conducted by SRC and MoF; these were outside the scope of the mission review.
  - Mission focused on data availability for more granular gap assessment using the bottom-up approach.
  - Bottom-up approach for PIT and SSC relies on detailed records from SRC and operational audits of taxpayers.
- Recommendation on approach:
  - A bottom-up approach using data from operational audits will likely be particularly effective for estimating the PIT and SSC gaps due to SRC’s extensive data on audit results and taxpayer characteristics.
  - Enhanced gap analysis can compare employment statistics from the Labor Force Survey (LFS) with SRC data using NACE codes; if ARMSTAT shows higher employment in certain sectors than SRC, it may indicate underreporting.
- Next steps proposed:
  - IMF can assist SRC to estimate the PIT/SSC gap for the income years 2020 to 2022, contingent on SRC making the data available to IMF.
  - Assistance could be provided remotely in November 2024 or, if more time is required, in February 2025.

### Key Recommendations (PIT and SSC Gap)
- That the SRC have the data available by November 1, 2024 in order for IMF to be able to help in measuring the PIT/SSC gaps.
  - Target date: November 1, 2024
- That the PIT/SSC gap be estimated using the bottom-up method based on comprehensive audits conducted by the SRC.
  - Target date: December 30, 2024
- That the SRC uses the labor input method and compares employment statistics from the LFS with SRC employment statistics.
  - Target date: June 30, 2025

### Background — request, system context, and past CD
- Request and scope:
  - The Chairman of the Armenian State Revenue Committee (SRC) requested IMF assistance to measure the PIT and SSC gaps; request confirmed at the 2024 Spring Meeting.
  - Purpose: obtain a detailed understanding of data availability so that an estimation of the PIT and SSC Gap can be made.
- System context:
  - Armenian PIT and SSC systems operate within a framework including a well-established withholding tax system, an employee register, and mandatory electronic submission of withholding returns.
  - PIT and SSC are collected by SRC at the same time; SRC is responsible for ensuring correct withholding of PIT and SSC from employers.
- Past CD and gap estimates:
  - VAT compliance gap in 2019 and 2020 estimated at 0.8 percent of GDP and 8.9 percent of potential VAT on average for 2019 and 2020.
  - Armenia CIT compliance gap estimated as a three-year average at 35.2 percent of potential CIT liability for income years 2020, 2021, and 2022.
  - Excluding audit adjustments with no immediate impact on revenue, CIT gap estimated at 26.4 percent of potential CIT liability.
  - Measured in terms of GDP the CIT gap was 1.4 percent or 1 percent excluding audit adjustments with no immediate impact on revenue.
  - CIT gap for 2023 predicted at 34.1 percent of potential CIT liability; excluding audit adjustments with no immediate impact on revenue predicted at 25.5 percent of potential CIT liability in 2023.
  - MLM predicted auditing the top 10 percent of corporations with the highest risk is likely to raise revenue by three times compared to SRC’s existing risk model.
- SRC and MoF PIT gap work:
  - SRC estimates the PIT gap at 0.94 percent of GDP in 2022; MoF estimates it at 1.35 percent of GDP in 2022.
  - For 2020 and 2021 SRC estimates the PIT gap at 0.67 percent and 1.09 percent of GDP; MoF estimates at 1.05 and 1.13 in 2020 and 2021 respectively.
  - Main reason for difference: SRC calculates at sectoral level while MoF uses aggregated data; SRC considers its more granular estimates more precise.

### Findings — methodologies and data suitability
- PIT gap measurement methodologies:
  - Top-down approaches:
    - Use independent statistical data (typically national accounts) to model a potential tax base and compare to the actual tax base observed in the tax administration.
    - Capture overall size of non-compliance, including informal activities, but lack granular insights across taxpayer types.
  - Bottom-up approaches:
    - Use tax administration’s detailed taxpayer records (random audits, operational audits targeted by risk criteria, other interventions).
    - Allow direct, detailed assessment and breakdown across taxpayer segments, but typically miss undetectable noncompliance.
- Operational audits (bottom-up) and selection bias correction:
  - Must account for non-random selection of audited taxpayers to avoid Sample Selection Bias.
  - The Heckman method is used to correct for Sample Selection Bias.
    - Two-stage procedure: first stage estimates probability of audit selection using a probit model and risk scores; second stage models PIT/SSC uncovered using taxpayer characteristics and a regressor accounting for selection derived from stage one.
  - Suitability considerations:
    - If audits focus narrowly on specific business parts, undisclosed taxes may be missed, leading to underestimated tax gaps.
    - Most SRC audits are comprehensive, which helps minimize missing discrepancies.
    - If audits target specific companies (sector/type), increased extrapolation is required, complicating reliable estimation.
    - A fraction of noncompliance could be undetectable even with best auditing efforts, potentially underestimating the tax gap.
- Consumption survey method (top-down):
  - Uses household consumption expenditure surveys to estimate underreporting among self-employed by comparing relationship between food expenditure and income for self-employed and employees (Pissarides and Weber 1989 methodology).
  - Assumes employees report income accurately due to third-party reporting/withholding; infers underreporting when self-employed food expenditure exceeds expected level given reported income and household characteristics.
  - Limitations: does not provide granularity on drivers of non-compliance; relies on stronger assumptions than audit-based approaches; sensitive to non-response and measurement error in surveys.
- Data availability for PIT gap measurement:
  - SRC maintains detailed tax registry data and audit result records (comprehensive audits, thematic audits, thematic inspections).
  - SRC monitors risk scores for selection into comprehensive audits but does not track selection criteria for thematic audits/inspections.
  - Operational audits' data richness supports bottom-up modeling, subject to correction for selection bias as described.

### Data inventory and key statistics (audit and survey data)
- Audit program and volume:
  - SRC conducts around 1,000 comprehensive audits each year.
  - Comprehensive audits cover all aspects of the taxpayer including PIT; taxpayers selected for audit each year will have their unaudited tax returns audited for up to three years earlier.
- Audit outcomes (2020–2022):
  - Total assessed PIT from comprehensive audits and thematic inspections:
    - 2020: 1 billion AMD
    - 2021: 1.3 billion AMD
    - 2022: 900 million AMD
  - Total assessed SSC:
    - 2020: 6 million AMD
    - 2021: 21 million AMD
    - 2022: 12 million AMD
  - Most of the assessed PIT and SSC relate to audits where taxpayers agreed to the findings and adjusted their tax returns accordingly.
- Thematic audits (2020–2022):
  - Purpose: uncover unregistered workers; unannounced; impose penalties of 250,000 AMD for each unregistered worker.
  - Associated penalties covering income years 2020-2022 amount to a total of 1.4 billion AMD.
  - Number of thematic audits ranged from 700-1,300 between 2020-2022.
- Tax registry data availability:
  - SRC collects and stores comprehensive tax registry data and background information on all taxpayers, and records audit results from comprehensive audits, thematic audits, and thematic inspections.
  - SRC tracks risk scores used to select taxpayers for comprehensive audits.
  - SRC does not track the criteria for selecting taxpayers for thematic audits and thematic inspections, as these criteria change regularly.
  - Implication: data facilitates PIT/SSC gap estimation based on comprehensive audits, with thematic audits/inspections serving as a supplementary measure.
- Survey data:
  - Labor Force Survey (LFS) by ARMSTAT:
    - Data: 7.788 completed interviews among households from 2020 and onwards.
    - Coverage: about 1 percent of the total population of households in Armenia.
    - Contains information on approximately 23,000 individuals’ labor force participation.
    - Anonymized data is publicly available.
    - Sectoral information: LFS contains information about employment at a sectoral level; possible to estimate employment at a sectoral level using NACE codes with reasonable degree of certainty with information on 23,000 individuals.
  - Household Integrated Living Conditions Survey (HILC) by ARMSTAT:
    - Covers 5,184 households from 2020 onward.
    - Reports monthly food expenditures and income broken down by sources such as salary and self-employment, and household characteristics.
    - Contains information about approximately 17,500 household members.
    - Anonymized data is publicly available.
  - Suitability note:
    - According to Armenian tax legislation, most self-employed taxpayers, including sole proprietors, are subject to corporate income tax (CIT).
    - Therefore the methodology developed by Pissarides and Weber (1989) building on Household Consumption survey data is not suitable for estimating the PIT/SSC gap.

### PIT Gap methodology to be applied — rationale and sequencing
- Rationale for using operational audits:
  - Data from operational audits are well-suited for estimating the PIT/SSC gap because of SRC’s extensive data on audit outcomes, audit selection, and taxpayer characteristics.
  - This supports methodologies that account for audit selection bias, such as the Heckman two-step estimator.
- Recommended primary approach:
  - Estimate the PIT/SSC gap based on comprehensive audits conducted by the SRC.
  - Yearly data on roughly 1,000 comprehensive audits should serve as the primary tool.
  - PIT/SSC gap estimation should be conducted using the Heckman two-step estimator to correct for selection bias.
- Role of thematic audits:
  - Data on thematic audits could be utilized to estimate the PIT/SSC gap due to unregistered workers.
  - Because thematic audits do not record under-reported PIT/SSC and the risk criteria used to select them are not available, thematic audit data should be used primarily as a supplementary measure.
- Recommended sequencing in practice:
  - Use bottom-up approach based on operational audits initially.
  - After initial tax gap estimates using the bottom-up method, enhance results by applying the labor input method.

### Labor-input method — approach, limitations, and future application
- Method:
  - Compare employment and reported hours in labor force surveys (LFS) with registered employment and hours in tax registry; difference in reported hours interpreted as underdeclared hours.
  - Multiply calculated differences in the number of full-time equivalents by the per capita values of output and value added to measure the prevalence and value of underreported hours.
  - Both the LFS by ARMSTAT and SRC record employment and hours worked by sector, allowing breakdown of uncovered employment by industry sectors.
- Limitations:
  - Questionnaire surveys like LFS can be distorted by non-response and measurement errors.
  - If respondents differ systematically from non-respondents or misreport hours worked, the PIT/SSC gap estimate will be inaccurate.
- Future application and sequencing:
  - Recommended that SRC compare employment statistics from the LFS with employment statistics in SRC data to give greater certainty/coverage of the PIT/SSC gaps.
  - Both ARMSTAT and SRC can measure employment at a sectoral level following NACE code, making sectoral comparison feasible.
  - If higher employment in certain sectors is found according to ARMSTAT compared to SRC data, this indicates underreporting.
  - Suggested deadline for SRC to implement comparison: June 2025.
  - Follow-up: investigate high-underreporting sectors and conduct targeted audits; after sufficient targeted audits are collected, re-measure the PIT gap using the Heckman two-step estimator.

### Next Steps — IMF assistance and SRC preparatory requirements
- IMF assistance proposed:
  - IMF will assist the SRC to estimate the PIT/SSC gap for the income years 2020 to 2022 using the bottom-up method.
  - Timing: remotely in November 2024 or if more time is required February 2025.
- SRC preparatory requirement:
  - That the SRC have the data available by November 1, 2024 in order for IMF to be able to help in measuring the PIT/SSC gaps.
- Future work:
  - SRC to use the labor input method and compare LFS employment statistics with SRC employment statistics by NACE codes; suggested deadline June 2025.
  - Targeted audits and subsequent re-estimation using Heckman two-step estimator if labor-input identifies sectors with high underreporting.

*IMF Technical Assistance Report*

### Preface ___________________________________________________________________________ 6

### Preface

### Mission overview
- A capacity development (CD) mission team comprising Mr. Graham Whyte, Fiscal Affairs Department (mission lead), Mr. Soren Pedersen (FAD) and Mr. Tobias Gabel Christiansen (FAD short-term expert) undertook a scoping mission (providing in country and remote support) during the period September 2 – 10, 2024.
- Purpose: undertake a comprehensive review of data availability for estimating the Personal Income Tax (PIT) and Social Security Contribution (SSC) gaps.
- Mission follows prior CD provided to the SRC to measure the Value Added Tax (VAT) and Corporate Income Tax (CIT) gaps.
- Workshops and meetings held with Mr. Arsen Sarikyan, Head of Development and Administration Strategy Programs Department (SRC) and his staff, representatives from the Ministry of Finance (MoF), the State Social Security Service (SSSS), and the Armenian Statistical Committee (ARMSTAT).
- Report findings are based on data and information provided by the SRC.
- Final draft submitted to Mr. Ashot H. Muradyan, Deputy Chairman and Mr. Arsen Sarikyan on September 10, 2024.
- Report consists of: Executive Summary; (I) Background; (II) Findings; and (III) Next Steps.

### Acknowledgements
- The IMF team expresses appreciation to the SRC, MoF, ARMSTAT and the SSSS for cooperation.
- Special acknowledgement to Mr. Arsen Sarikyan and his team, and interpreters Ms. Lilit Simonyan and Mr. Khachatur Adumyan.

### Key institutional acronyms (selected)
- AMD Armenian Dram - the currency of Armenia
- ARMSTAT Statistical Committee - Republic of Armenia
- CD Capacity Development
- CIT Corporate Income Tax
- FAD Fiscal Affairs Department
- GDP Gross Domestic Product
- ILO International Labor Organization
- IMF International Monetary Fund
- LFS Labor Force Survey
- MoF Armenian Ministry of Finance
- NACE Nomenclature of Economic Activities
- PIT Personal Income Tax
- SRC State Revenue Committee of Armenia
- SSC Social Security Contributions
- SSSS State Social Security Service
- STX Short Term Expert
- TA Technical Assistance
- TIN Taxpayer Identification Number
- VAT Value Added Tax

---

### Executive Summary

### Core conclusion
- SRC has sufficient quality data available from operational audits to assess the PIT/SSC gaps.

### Methods and focus
- Two main methods for estimating tax gaps: top-down and bottom-up; each has strengths and weaknesses and are complementary.
- Top-down gap estimates have already been conducted by SRC and MoF; these were outside the scope of the mission review.
- Mission focused on data availability for more granular gap assessment using the bottom-up approach.
- Bottom-up approach for PIT and SSC relies on detailed records from SRC and operational audits of taxpayers.

### Data strengths and limitations
- SRC’s detailed tax registry data and background information on all taxpayers is very important for a tax gap model to produce reliable estimates.
- SRC keeps records of audit results from comprehensive audits, thematic audits, and thematic inspections.
- SRC monitors risk scores used to select taxpayers for comprehensive audits, but does not track criteria for selecting taxpayers for thematic audits and inspections, making thematic audits/inspections unsuitable for tax gaps estimation.

### Recommendation on approach
- A bottom-up approach using data from operational audits will likely be particularly effective for estimating the PIT and SSC gaps due to SRC’s extensive data on audit results and taxpayer characteristics.
- Enhanced gap analysis can compare employment statistics from the Labor Force Survey (LFS) with SRC data using NACE codes; if ARMSTAT shows higher employment in certain sectors than SRC, it may indicate underreporting.

### Next steps proposed
- IMF can assist SRC to estimate the PIT/SSC gap for the income years 2020 to 2022, contingent on SRC making the data available to IMF.
- Assistance could be provided remotely in November 2024 or, if more time is required, in February 2025.

---

### Recommendations (PIT and SSC Gap)

- 1 The SRC have the data available by November 1, 2024 in order for IMF to be able to help in measuring the PIT/SSC gaps.
  - Target date: November 1, 2024
- 2 The PIT/SSC gap be estimated using the bottom-up method based on comprehensive audits conducted by the SRC.
  - Target date: December 30, 2024
- 3 The SRC uses the labor input method and compares employment statistics from the LFS with SRC employment statistics.
  - Target date: June 30, 2025

---

### I. Background

### Request and scope
- The Chairman of the Armenian State Revenue Committee (SRC) requested IMF assistance to measure the PIT and SSC gaps; request confirmed at the 2024 Spring Meeting.
- Purpose of the mission: obtain a detailed understanding of data availability so that an estimation of the PIT and SSC Gap can be made.
- If suitable data is available, follow-up assistance can be provided on request to help calculate the PIT and SSC gaps.

### System context
- Armenian PIT and SSC systems operate within a framework including a well-established withholding tax system, an employee register, and mandatory electronic submission of withholding returns.
- PIT and SSC are collected by SRC at the same time; SRC is responsible for ensuring correct withholding of PIT and SSC from employers.

### Past CD and gap estimates
- Previous CD: VAT gap measured in 2022 using IMF’s Revenue Administration VAT gap model; CIT gap measured in 2024 by a bottom-up approach based on operational audits.
- A machine learning model (MLM) was developed to select cases for future audits; subsequent CD delivered prediction of the CIT gap for 2023 based on non-audited CIT declarations.
- VAT compliance gap in 2019 and 2020 estimated at 0.8 percent of GDP and 8.9 percent of potential VAT on average for 2019 and 2020.
  - Note: VAT gap calculation limited to 2019 and 2020 due to significant changes in SRC accounting systems.
- Armenia CIT compliance gap estimated as a three-year average at 35.2 percent of potential CIT liability for income years 2020, 2021, and 2022.
  - Excluding audit adjustments with no immediate impact on revenue, CIT gap estimated at 26.4 percent of potential CIT liability.
  - Measured in terms of GDP the CIT gap was 1.4 percent or 1 percent excluding audit adjustments with no immediate impact on revenue.
- MLM predicted auditing the top 10 percent of corporations with the highest risk is likely to raise revenue by three times compared to SRC’s existing risk model; MLM results in higher average CIT corrections compared to current SRC strategy when auditing taxpayers other than those in the top 10 percent.
- CIT gap for 2023 predicted at 34.1 percent of potential CIT liability; excluding audit adjustments with no immediate impact on revenue predicted at 25.5 percent of potential CIT liability in 2023.

### SRC and MoF PIT gap work
- SRC and MoF have both measured the PIT gap based on discrepancies between national accounts statistics and SRC data; wages and salaries derived from output and value added in national accounts, then potential tax to be withheld from employers is calculated from estimated wages and salaries.
- PIT gap calculated as the difference between potential tax derived from national accounts and reported taxes to the SRC.
- SRC estimates the PIT gap at 0.94 percent of GDP in 2022; MoF estimates it at 1.35 percent of GDP in 2022.
- For 2020 and 2021 SRC estimates the PIT gap at 0.67 percent and 1.09 percent of GDP; MoF estimates at 1.05 and 1.13 in 2020 and 2021 respectively.
- Main reason for difference: SRC calculates at sectoral level while MoF uses aggregated data; SRC considers its more granular estimates more precise because sectoral estimation can account for different legislative rules across sectors.

---

### II. Findings

### PIT gap measurement methodologies
- Two approaches: top-down and bottom-up; complementary with distinct strengths and limitations.

Top-down approaches
- Use independent statistical data (typically national accounts) to model a potential tax base and compare to the actual tax base observed in the tax administration.
- Capture overall size of non-compliance, including informal activities, but lack granular insights across taxpayer types.

Bottom-up approaches
- Use tax administration’s detailed taxpayer records (random audits, operational audits targeted by risk criteria, other interventions).
- Allow direct, detailed assessment and breakdown across taxpayer segments, but typically miss undetectable noncompliance.

Operational audits (bottom-up)
- Must account for non-random selection of audited taxpayers to avoid Sample Selection Bias.
- The Heckman method is used to correct for Sample Selection Bias.
  - Two-stage procedure: first stage estimates probability of audit selection using a probit model and risk scores; second stage models PIT/SSC uncovered using taxpayer characteristics and a regressor accounting for selection derived from stage one.
  - Once fitted, tax gap can be split into subcomponents based on taxpayer information for granular insights.
- Suitability considerations:
  - If audits focus narrowly on specific business parts, undisclosed taxes may be missed, leading to underestimated tax gaps.
  - Most SRC audits are comprehensive, which helps minimize missing discrepancies.
  - If audits target specific companies (sector/type), increased extrapolation is required, complicating reliable estimation.
  - A fraction of noncompliance could be undetectable even with best auditing efforts, potentially underestimating the tax gap.

Consumption survey method (top-down)
- Uses household consumption expenditure surveys to estimate underreporting among self-employed by comparing relationship between food expenditure and income for self-employed and employees (Pissarides and Weber 1989 methodology).
- Assumes employees report income accurately due to third-party reporting/withholding; infers underreporting when self-employed food expenditure exceeds expected level given reported income and household characteristics.
- Captures most aspects of informal economy because surveys cover registered, non-filers, and non-registered taxpayers.
- Limitations: does not provide granularity on drivers of non-compliance; relies on stronger assumptions than audit-based approaches; sensitive to non-response and measurement error in surveys.

### Data availability for PIT gap measurement
- SRC maintains detailed tax registry data and audit result records (comprehensive audits, thematic audits, thematic inspections).
- SRC monitors risk scores for selection into comprehensive audits but does not track selection criteria for thematic audits/inspections.
- Operational audits' data richness supports bottom-up modeling, subject to correction for selection bias as described.

---

### III. Next Steps

### Application of the methodology – Additional CD
- IMF assistance to estimate the PIT/SSC gap for income years 2020 to 2022 conditional on SRC making the data available.
- Assistance could be provided remotely in November 2024 or in February 2025 if more time is required.

### Future application of labor input method
- SRC can undertake enhanced gap analysis comparing LFS employment statistics with SRC employment statistics using NACE codes to identify sectors with potential underreporting.

*IMF Technical Assistance Report — Preface and Executive Summary (September 2–10, 2024 mission).*

### 22.      The labor-input method compares employment from labor force surveys (LFS) with

### 22.      The labor-input method compares employment from labor force surveys (LFS) with registered employment.

### Labor-input method: approach and limitations
- Method: Compare employment and reported hours in labor force surveys (LFS) with registered employment and hours in tax registry; difference in reported hours interpreted as underdeclared hours.
- Estimation of tax gap: Multiply calculated differences in the number of full-time equivalents by the per capita values of output and value added to measure the prevalence and value of underreported hours.
- Sectoral breakdown: Both the LFS by ARMSTAT and SRC record employment and hours worked by sector, allowing breakdown of uncovered employment by industry sectors.
- Limitations:
  - Questionnaire surveys like LFS can be distorted by non-response and measurement errors.
  - If respondents differ systematically from non-respondents or misreport hours worked, the PIT/SSC gap estimate will be inaccurate.

### Use in practice (recommendation within combined methodology)
- Recommended sequencing:
  - Use bottom-up approach based on operational audits initially.
  - After initial tax gap estimates using the bottom-up method, enhance results by applying the labor input method.

### Explicit recommendation
- That the SRC uses the labor input method and compares employment statistics from the LFS with employment statistics in SRC data.

---

### B. Data Available for PIT Gap Measurement — Audit Data
- SRC audit program:
  - Prepares and implements an annual audit program including comprehensive audits, thematic audits, and thematic inspections.
  - Annual program targets taxpayers assessed as representing the highest risk to revenue by SRC's internally developed risk assessment system.
  - Comprehensive audits: selected using risk scores; thematic audits: selected using several binary criteria; thematic inspections introduced in 2022 and are desk-type audits selected on same criteria as thematic audits.
- Volume and scope:
  - SRC conducts around 1,000 comprehensive audits each year.
  - Comprehensive audits cover all aspects of the taxpayer including PIT; taxpayers selected for audit each year will have their unaudited tax returns audited for up to three years earlier.
- Audit outcomes (2020–2022):
  - Total assessed PIT from comprehensive audits and thematic inspections:
    - 2020: 1 billion AMD
    - 2021: 1.3 billion AMD
    - 2022: 900 million AMD
  - Total assessed SSC:
    - 2020: 6 million AMD
    - 2021: 21 million AMD
    - 2022: 12 million AMD
  - Most of the assessed PIT and SSC relate to audits where taxpayers agreed to the findings and adjusted their tax returns accordingly.
- Thematic audits (2020–2022):
  - Purpose: uncover unregistered workers; unannounced; impose penalties of 250,000 AMD for each unregistered worker.
  - Associated penalties covering income years 2020-2022 amount to a total of 1.4 billion AMD.
  - Number of thematic audits ranged from 700-1,300 between 2020-2022.
- Tax registry data availability:
  - SRC collects and stores comprehensive tax registry data and background information on all taxpayers, and records audit results from comprehensive audits, thematic audits, and thematic inspections.
  - SRC tracks risk scores used to select taxpayers for comprehensive audits.
  - SRC does not track the criteria for selecting taxpayers for thematic audits and thematic inspections, as these criteria change regularly.
  - Implication: data facilitates PIT/SSC gap estimation based on comprehensive audits, with thematic audits/inspections serving as a supplementary measure.

### B. Data Available for PIT Gap Measurement — Survey Data
- Labor Force Survey (LFS) by ARMSTAT:
  - Follows ILO recommendations on how to collect LFS data.
  - Data: 7.788 completed interviews among households from 2020 and onwards.
  - Coverage: about 1 percent of the total population of households in Armenia.
  - Contains information on approximately 23,000 individuals’ labor force participation.
  - Anonymized data is publicly available.
  - Sectoral information: LFS contains information about employment at a sectoral level; possible to estimate employment at a sectoral level using NACE codes with reasonable degree of certainty with information on 23,000 individuals.
- SRC tax return data:
  - SRC records the number of employees and hours worked in PIT returns, facilitating comparison between tax registry records and LFS data for PIT/SSC gap estimation.
- Household Integrated Living Conditions Survey (HILC) by ARMSTAT:
  - Covers 5,184 households from 2020 onward.
  - Carried out monthly throughout the year, covering different households each month.
  - Reports monthly food expenditures and income broken down by sources such as salary and self-employment, and household characteristics including number of household members, children, age, gender, educational level, etc.
  - Contains information about approximately 17,500 household members.
  - Anonymized data is publicly available.
- Suitability notes:
  - According to Armenian tax legislation, most self-employed taxpayers, including sole proprietors, are subject to corporate income tax (CIT).
  - Therefore the methodology developed by Pissarides and Weber (1989) building on Household Consumption survey data is not suitable for estimating the PIT/SSC gap.

### Conclusion on data availability
- There is sufficient quality data available from operational audits to assess the PIT/SSC gaps.
- There is ample data to also implement the labor input method.

---

### C. PIT Gap Methodology to be Applied
- Rationale for using operational audits:
  - Data from operational audits are well-suited for estimating the PIT/SSC gap because of SRC’s extensive data on audit outcomes, audit selection, and taxpayer characteristics.
  - This supports methodologies that account for audit selection bias, such as the Heckman two-step estimator.
- Advantages:
  - Operational audits provide granular gap estimates for different segments of taxpayers, enabling identification of different causes of non-compliance (types of mistakes, demographic breakdowns, sector-specific issues).
  - Granular insight enables tax authorities to tailor interventions and improve overall compliance.
- Recommended primary approach:
  - Estimate the PIT/SSC gap based on comprehensive audits conducted by the SRC.
  - Yearly data on roughly 1,000 comprehensive audits should serve as the primary tool.
  - PIT/SSC gap estimation should be conducted using the Heckman two-step estimator to correct for selection bias.
- Role of thematic audits:
  - Data on thematic audits could be utilized to estimate the PIT/SSC gap due to unregistered workers.
  - Because thematic audits do not record under-reported PIT/SSC and the risk criteria used to select them are not available, thematic audit data should be used primarily as a supplementary measure.

### Recommendation
- That the PIT/SSC gap be estimated using the bottom-up method based on comprehensive audits conducted by the SRC.

---

### III.   Next Steps

### A. Application of the Methodology – Additional CD
- Proposed IMF assistance:
  - IMF will assist the SRC to estimate the PIT/SSC gap for the income years 2020 to 2022 using the bottom-up method.
  - Timing: IMF is proposing that it can be done remotely in November 2024 or if more time is required February 2025.
- SRC preparatory requirement:
  - That the SRC have the data available by November 1, 2024 in order for IMF to be able to help in measuring the PIT/SSC gaps. (Recommendation)

### B. Future Application of Labor Input Method
- Recommended process:
  - SRC should compare employment statistics from the LFS with employment statistics in SRC data to give greater certainty/coverage of the PIT/SSC gaps.
  - Both ARMSTAT and SRC can measure employment at a sectoral level following NACE code, making sectoral comparison feasible.
  - If higher employment in certain sectors is found according to ARMSTAT compared to SRC data, this indicates underreporting.
  - Suggested deadline: June 2025 to give SRC enough time to gather data from ARMSTAT and compare with SRC’s own data.
- Follow-up actions if labor-input identifies high-underreporting sectors:
  - SRC should investigate these sectors in more detail, which might result in targeted audits in these sectors.
  - After sufficient targeted audits are collected in these sectors, the PIT gap should be measured again using the Heckman two-step estimator.
- Recommendation
  - That the SRC uses the labor input method and compares employment statistics from the LFS with employment statistics in SRC data.

*IMF Technical Assistance Report*

---


_Source: https://www.imf.org/-/media/files/publications/tar/2025/english/tarea2025009-print-pdf.pdf_