## wpiea2025209 — Appendix I–V (selected material)

## Source details

**Canonical URL:** [wpiea2025209 — Appendix I–V (selected material)](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025209.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025209.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025209.pdf.json)

---

### Overview
- The paper examines the relationship between tax administration performance and VAT compliance gaps using panel data for 111 countries during the period 2010–2023.
- Tax administration performance is measured primarily with the Tax Administration Diagnostic Assessment Tool (TADAT), with scores ranging from ‘A’ (best) to ‘D’ (worst).
- VAT compliance gaps are estimated using the “reverse RA-GAP” method based on the Revenue Administration Gap Analysis Program (RA-GAP).

### Theoretical foundations
- Core framework: Allingham and Sandmo (A-S) tax compliance risk theory.
  - Representative taxpayer chooses evasion share Γ based on income I, tax rate τ, perceived detection probability p, and fine rate X:
    - Γ* = Γ*(I, τ, p, X)
  - First-order predictions:
    - Fine rate X and detection probability p have a negative effect on the compliance gap.
    - Effects of income I and tax rate τ are indeterminate without further assumptions.
- Extensions and complements:
  - Addition of a tax morale variable (social acceptability a).
  - Alignment with Compliance Risk Management (CRM) approaches.
- Key behavioral determinants and equations:
  - p = p( e_TA , η_TADAT )     (2)
  - X = X( e_TA , η_TADAT )     (3)
  - a = a(value of government services, trust in tax administration, social norms, ..)     (4)
- Empirical specification (log-linear):
  - Log(훤∗) = 훼 + 훽 Log( e_TA ) + 훾 Log( η_TADAT ) + 휆 Log( a ) + 휈 Log( I ) + 휌 Log( τ ) + 휀           (5)

### Data and empirical approach
- Core datasets: TADAT, RA-GAP (reverse RA-GAP), ISORA (tax administration budget and human resources), supplemented by IMF and World Bank macro and index series.
- Panel composition: 111 countries—mainly Emerging Market Economies (EMEs) and Low Income Developing Countries (LIDCs)—for 2010–2023.
- TADAT scoring and use:
  - 55 performance dimensions grouped into 9 Performance Outcome Areas (POAs).
  - Scoring conversion: ‘D’ = 1, ‘C’ = 2, ‘B’ = 3, ‘A’ = 4.
  - Overall η_TADAT = unweighted average of 55 dimension scores.
  - η_TADAT calculated for the year of a country’s first TADAT assessment.
- Key tax-administration resource measures:
  - Revenue Administration operating expenditure as a share of GDP (푒_TA) from ISORA 2018–2022.
  - Tax administration staff per million inhabitants (퐿_TA) as FTE per million from ISORA 2018–2022.
- Social acceptability proxy: World Bank Rule of Law Annual Index (annual levels for 2010–2023, range -2.50 to 2.50).

### Reverse RA-GAP (VAT compliance gap measurement)
- Identity and estimation:
  - (1 − ) ≡ Ec (1 − TEG/PV2) (1 − NTG/PV3)                      (1)
  - Ec = C-efficiency from IMF FAD public datasets.
  - TEG approximated using GTED; NTG approximated from GDP of public administration, education, health.
- Sample coverage: 111 countries, 2010–2023.
- Calibration: 208 VAT gap observations from 43 countries (IMF RA-GAP direct top-down estimates) for 2010–2023 used to calibrate proxies.
- Advantages: enables multi-year, cross-country VAT compliance gap estimates even in data-scarce settings.
- Caveats: relies on indirect estimates and proxies; calibration smooths differences and cannot capture all variations.

### Main empirical findings (preserve numeric precision)
- Estimated elasticity of VAT compliance gap with respect to tax administration performance: -0.7 (reported more precisely as -0.71 in Table 2, Column 1; range across specifications between -0.466 and -0.784).
- Simulation: improvement in TADAT score from 1.85 (close to a ‘D+’) to 2.32 (close to a ‘C+’) is associated with:
  - 6.6 percentage point reduction in the VAT compliance gap — from 42.8 to 36.2 percent.
  - Increase in VAT revenue equivalent to 0.6 percent of GDP.
- VAT-induced CIT effects and combined revenue impact:
  - Under a “consistent evader” assumption, a 0.6 percent of GDP reduction in the VAT compliance gap is associated with a 0.7 percent of GDP reduction in the CIT compliance gap (based on a sample of 12 EME).
  - Total combined effect: approximately 1.3 percent of GDP in additional revenue from both VAT and CIT.
- Timeframe for TADAT improvement:
  - Based on evidence from about 30 countries with repeat TADATs, improving from 1.85 to 2.32 could take an average of 5.8 years (reported also as 5.81 years in aggregate language).
- Tax morale (Rule of Law semi-elasticity and revenue effect):
  - Estimated semi-elasticity of the tax morale indicator: -0.13.
  - Increasing tax morale index from -0.62 to -0.57 could raise revenues on the order of 0.04 percent of GDP, on average, by reducing both VAT and CIT compliance gaps.
- Additional empirical associations (selected, exact values preserved where reported):
  - Log (휂_TADAT): Column 1 -0.710**; Column 2 -0.690**; Column 3 -0.466*; Column 4 -0.784**.
  - Lagged Log (I): positive and significant in some specifications (e.g., 0.143*** in Column 1).
  - Lagged (a) (Rule of Law): negative and significant in some specifications (e.g., -0.127*** in Column 1).
  - Lagged Log (Imp): negative and significant in some specifications (e.g., -0.195*** in Column 1).
  - Lagged Log (Agr): positive and significant in some specifications (e.g., 0.119*** in Column 1).
  - In VAT-to-GDP robustness (Table AV.1), Log(휂_TADAT) estimates include 0.756*** (5.1), 0.957** (5.2), 0.462* (5.3), 0.722*** (5.4).
- Non-significant or nuanced findings:
  - Revenue Administration operating expenditure to GDP ratio (푒_TA) and staff per million (퐿_TA) are not consistently statistically significant predictors of VAT compliance gap in the main analysis; explanations include measurement error, reporting differences, off-budget financing, or TADAT capturing effective use of resources better.

### Empirical strategy, identification, and robustness
- Econometric approach:
  - Mundlak-Krishnakumar framework augmented with Hausman-Taylor estimators to handle time-invariant variables and address endogeneity.
  - Expanded estimating equation: 훤*_{it} = 훼1 + 훼2(X_i) + 훼3(Z_{it−1}) + 훼4(N_i) + T_t + v_i + θ_{it}
    - X_i: time-invariant variables including η_TADAT, e_TA or L_TA.
    - Z_{it−1}: lagged time-varying macro and institutional variables (I, Imp, Agr, τ, number of VAT rates, a).
    - N_i: exogenous time-invariant instruments (e.g., mean Rule of Law a̅ and log mean agriculture-to-GDP).
- Main robustness checks:
  - Alternative dependent variable: VAT-to-GDP (using accounting identity between C-efficiency, compliance gap, and policy gap).
    - Result: estimated elasticity 0.76 (Column 5.1), and counterfactual TADAT improvement from 1.85 to 2.32 implied an increase of 0.7 percentage points of GDP in VAT revenue — similar to the VAT compliance gap specification (0.6 percentage points of GDP).
  - Subsample restricted to countries with RA-GAP mission estimates: same sign for TADAT but statistical insignificance due to reduced variation and selection.
  - Recognized empirical challenges: limited time variation in TADAT scores, measurement imprecision in ISORA and proxies, and potential endogeneity/reverse causality.

### Policy-relevant implications and recommendations
- Strengthen tax administration performance comprehensively:
  - Improvements across registration, filing, payment, compliance risk management, audit, dispute resolution, accountability and transparency are needed to reduce noncompliance.
  - Targeted one-off reforms are unlikely to yield sustained compliance effects when capacity is weak.
- Time horizons and sequencing:
  - Gains accrue over multi-year horizons (e.g., average observed improvement from 1.85 to 2.32 in repeat TADATs took 5.8 years).
- Whole-of-government and social acceptability:
  - Promoting tax morale (social acceptability of the tax system) is an additional lever to reduce compliance gaps; tax administrations can lead a whole-of-government approach.
- Tailor compliance strategies to economic structure:
  - Where VAT is mainly collected at the border, customs can support VAT compliance via import controls and third-party data sharing.
  - In economies dominated by agriculture with many small cash-based enterprises, tax administrations need tailored compliance risk strategies.
- Data and measurement improvements:
  - Continued expansion of standardized RA-GAP VAT assessments and improved time-series data on tax administration performance would reduce measurement imprecision and strengthen causal inference.

### VAT-induced CIT compliance gap (key relations and counterfactual)
- Formal relation:
  - VAT-induced CIT Compliance Gap Amount = (τ_CIT / τ) x VAT Compliance Gap Amount.
  - Total CIT Compliance Gap Amount = K x (τ_CIT / τ) x VAT Compliance Gap Amount, where K > 1.
- Empirical log-log fit (ECLAC, 12 EMEs):
  - Log(CIT Compliance Gap Amount) = 0.00482799 + 0.575622 Log(τ_CIT / τ) + 0.986122 Log(VAT Compliance Gap Amount); R2 Adj. = 0.64; N = 12.
- Counterfactual example:
  - Using world average τ_CIT = 20 percent and τ = 15 percent, a 0.6 percent of GDP variation in the VAT compliance gap implies an associated 0.7 percent of GDP variation in the CIT compliance gap (robust to alternative average rates: CIT = 22.5 percent, VAT = 13.5 percent).

### Summary conclusions (exact reported magnitudes)
- Empirical evidence suggests tax administration effectiveness (η_TADAT) is a key mechanism shaping VAT compliance.
- Estimated elasticity: -0.71 (Table 2, Column 1) with a range across specifications.
- Policy-relevant simulation:
  - Raising a tax administration’s TADAT score from 1.85 (approximately a ‘D+’) to 2.32 (approximately a ‘C+’) is associated with:
    - Additional VAT revenue equivalent to 0.6 percent of GDP from improved compliance.
    - Additional CIT revenue equivalent to 0.7 percent of GDP under the consistent-evader assumption.
    - Approximately 1.3 percent of GDP in additional revenue from both taxes in total.
- Caveats: measurement imprecision of VAT compliance gaps, limited time variation in TADAT scores, and endogeneity risks necessitate continued data improvements and further analysis.

*Source: IMF Working Paper — Closing the gap: How tax administration performance shapes compliance (selected Appendices I–V and Box 2).*

### Appendix I. Literature .................................................................................................

### Appendix I. Literature

### Overview
- The paper examines the relationship between tax administration performance and VAT compliance gaps, grounded in a theoretical framework and empirical analysis using panel data for 111 countries during the period 2010–2023.
- Tax administration performance in this analysis is measured primarily with the Tax Administration Diagnostic Assessment Tool (TADAT), with scores ranging from ‘A’ (best) to ‘D’ (worst).
- VAT compliance gaps are estimated using a method referred to as the “reverse RA-GAP” based on the Revenue Administration Gap Analysis Program (RA-GAP).

### Key prior studies and literature synthesis
- Foundational cross-country studies include Silvani and Brondolo (1993) and Agha and Haughton (1996), which correlates VAT design and administration resources with compliance gaps; subsequent relevant studies include Christie and Holzner (2006), Reckon (2009), Keen (2015), Das-Gupta et al (2016), Crivelli (2018), CASE (2020), and Butu (2021).
- Appendix I (the literature appendix) synthesizes these and other adaptations of the Allingham and Sandmo (A-S) model, including extensions that incorporate third-party information, behavioral factors (trust, fairness, social norms), game-theoretic approaches, and firm- and macro-level adaptations.

### Theoretical foundations used in the paper
- The empirical analysis is based on the Allingham and Sandmo (A-S) tax compliance risk theory (Allingham and Sandmo (1972); Sandmo (2005)), with a representative taxpayer choosing an evasion share Γ based on income I, tax rate τ, perceived detection probability p, and fine rate X:
  - Γ* = Γ*(I, τ, p, X)
- First-order predictions:
  - The fine rate and the probability of detection have a negative effect on the compliance gap.
  - The effects of income and the tax rate are indeterminate.
- The A-S framework is complemented by adding a tax morale variable (the degree of the taxpayer’s social acceptability of the tax system) and is aligned with modern Compliance Risk Management (CRM) approaches.

### Data and empirical approach
- Core datasets: TADAT, RA-GAP (reverse RA-GAP method for VAT compliance gaps), and ISORA (for tax administration budget and human resources), supplemented by macroeconomic variables, VAT rates, and international indexes from IMF and World Bank datasets and, where needed, country MoF websites.
- Panel composition: 111 countries—mainly Emerging Market Economies (EMEs) and Low Income Developing Countries (LIDCs)—for 2010–2023.

### Main empirical findings (preserve numeric precision)
- Estimated elasticity of VAT compliance gap with respect to tax administration performance: -0.7.
  - If a tax administration raises its TADAT score from 1.85 (close to a ‘D+’) to 2.32 (close to a ‘C+’), the expectation is an increase of 0.6 percent to GDP in additional VAT revenue.
  - Incorporating the fall in the CIT compliance gap induced by the decrease in the VAT compliance gap yields an additional 0.7 percent of GDP in CIT revenue.
  - The total combined effect is approximately 1.3 percent of GDP in additional revenue from both VAT and CIT.
- Timeframe for TADAT improvement:
  - Based on evidence from about 30 countries with repeat TADATs, improving from 1.85 to 2.32 could take an average of 5.8 years.
- Tax morale (semi-elasticity):
  - Estimated semi-elasticity of the tax morale indicator: -0.13.
  - An increase in the tax morale indicator from -0.62 to -0.57 could raise revenues on the order of 0.04 percent of GDP, on average, by reducing both VAT and CIT compliance gaps.

### Policy-relevant implications distilled from the analysis
- Strengthening tax administration performance can substantially reduce VAT noncompliance and mobilize additional domestic revenue, although benefits may accrue over multi-year horizons (e.g., improvements observed over an average of 5.8 years in repeat TADATs).
- Efforts to improve the social acceptability of the tax system (tax morale), supported by tax administrations and other public agencies, provide an additional channel to reduce compliance gaps and increase revenue.
- Compliance strategies should be tailored to economic structure:
  - In economies where VAT is primarily collected at the border, customs administrations can support VAT compliance by strengthening import controls and providing revenue-related information as third-party data.
  - In economies dominated by agriculture with many small, often informal cash-based enterprises, tax administrations need tailored and robust compliance risk strategies to detect and address noncompliance.

*Source: IMF Working Paper — Closing the gap: How tax administration performance shapes compliance (Appendix I. Literature).*

### Appendix II). The business’s income corresponds to the true value added it generates, which also defines the

### wpiea2025209 - Appendix II). The business’s income corresponds to the true value added it generates, which also defines the 

### Theoretical framework: tax compliance risk model
- The VAT-able base is defined as the true value added subject to VAT for a representative taxpayer (referred to as the VAT-able base).
- The model follows tax compliance risk theory (also called "portfolio theory," "expected utility model," and "deterrence model") and mirrors Compliance Risk Management (CRM) approaches used by tax administrations.
- Key behavioral determinants:
  - Probability of detection, p, influenced by tax administration expenditure and effectiveness.
  - Fine rate, X, determined by statutory law but effectively contingent on administration capacity to apply penalties.
  - Non-financial determinants captured by social acceptability, a (value of government services, trust in tax administration, social norms, political legitimacy, cost of compliance).
- Equations linking administration resources and effectiveness to enforcement outcomes:
  - p = p( e_TA , η_TADAT )     (2)
  - X = X( e_TA , η_TADAT )     (3)
  - a = a(value of government services, trust in tax administration, social norms, ..)     (4)

### Empirical model specification
- Log-linear empirical equation proposed to relate the VAT compliance gap (휞∗, percent of potential VAT liability) to enforcement and non-financial factors:
  - Log(훤∗) = 훼 + 훽 Log( e_TA ) + 훾 Log( η_TADAT ) + 휆 Log( a ) + 휈 Log( I ) + 휌 Log( τ ) + 휀           (5)
- Hypotheses on expected signs:
  - 훽 and 훾 expected to be negative (greater expenditure and greater effectiveness lower the VAT compliance gap).
  - 휆 expected to be negative (higher social acceptability reduces noncompliance).
  - Controls: income level I and tax rate τ included per theory.

### Enforcement facilitators and hinderers
- Facilitators (increase probability of detection and fine application):
  - Application of VAT on imports.
  - Legal powers to control and sanction noncompliance.
  - Degree of simplicity of tax legislation.
  - Existence of third-party data (customs, financial, etc.).
- Hinderers:
  - Large agriculture sector.
  - Fragile and Conflict-Affected States (FCS) and Small Island conditions associated with weak institutional capacity.
  - Prevalence of small enterprises transacting in cash.
- Practical note: effective penalty enforcement may require years-long efforts to pursue fines through courts; administration expenditure proxies both detection and effective fine application.

### Dependent variable and measurement challenges
- VAT compliance gap expressed as percentage points of the potential VAT liability (revenue that would be obtained under full legal compliance at the standard rate).
- Traditional proxy VAT revenue-to-GDP is problematic because it conflates compliance gap with tax policy gap, and is affected by the standard VAT rate and final consumption-to-GDP ratio.
- Need to control for tax policy gap; lack of global datasets for tax policy gap introduces imprecision in prior empirical work.

### Data and novel measurement: the Reverse Method (RA-GAP Reverse Method)
- Purpose: construct a consistent time series of annual VAT compliance gap estimates for a large cross-section of countries.
- Sample coverage using the Reverse Method: 111 countries over the period 2010–2023.
- Methodology overview:
  - Based on C-efficiency (Ec) accounting identity linking actual VAT revenue to potential revenue under full enforcement at the standard rate.
  - Compliance gap () is estimated as a residual using Ec, the tax expenditure gap (TEG/PV2), and the non-taxable goods and services gap (NTG/PV3):
    - (1 − ) ≡ Ec (1 − TEG/PV2) (1 − NTG/PV3)                      (1)
  - Tax policy gap decomposed into:
    - Tax expenditure gap to potential (TEG/PV2).
    - Nontaxable goods and services gap to potential (NTG/PV3).
  - Normative tax policy: all goods and services taxable at the standard rate except public administration, public education, and public health.
  - Comprehensive tax policy: all goods and services taxable at the standard rate, no exemptions.
- Calibration and proxies:
  - Only 25 countries have precise RA-GAP mission-based variables; to expand sample, proxies are calibrated against RA-GAP direct estimates.
  - Ec figures from IMF FAD public datasets.
  - TEG approximated using GTED (Global Tax Expenditure Datasets).
  - NTG approximated from GDP of sectors public administration, education, and health.
  - Calibration example: 208 VAT gap observations from 43 countries (IMF RA-GAP direct top-down estimates) for 2010–2023 used to calibrate proxies.
- Advantages and caveats:
  - Advantage: top-down method captures all forms of noncompliance in principle and enables multi-year, cross-country estimates even in data-scarce settings.
  - Caveat: relies on indirect estimates and proxies; proxies have methodological limitations and calibration smooths but cannot capture all differences.

### Tax policy and tax-base proxies used
- VAT-able base proxy: per capita final consumption in constant US dollars based on per capita PPP (IMF datasets) for 2010–2023.
  - Rationale: true value added subject to VAT approximated by final consumption; per-taxpayer proxy uses per capita final consumption assuming VAT payers are fixed proportion of population.
- Standard VAT rate: IMF database series for 2010–2023 (standard rate defined as default rate applied to most goods and services, excluding reduced, zero, or special rates).

### Independent variable: Tax administration strength (TADAT)
- Effectiveness measure: η_TADAT derived from IMF’s Tax Administration Diagnostic Assessment Tool (TADAT) datasets.
- TADAT assesses 55 areas of tax administration performance grouped into nine dimensions; applied in more than 100 countries.
- Scoring conversion:
  - ‘D’ = 1 (worst), ‘C’ = 2, ‘B’ = 3, ‘A’ = 4 (best).
  - The unweighted average of the 55 performance dimension scores yields overall TA effectiveness score η_TADAT.
- Timing: η_TADAT calculated for the year in which the first TADAT assessment was conducted in a country.
- Rationale for using TADAT:
  - Standardized assessment applied across countries.
  - Conducted by independent external assessors based on evidence (not self-reported).
  - Results reviewed with the tax administration before final assessment.

### Expected empirical findings and robustness checks (as outlined)
- Primary question: are 훽 and 훾 negative and statistically significant (i.e., do higher e_TA and higher η_TADAT reduce the VAT compliance gap)?
- Expected negative sign for 휆 (social acceptability reduces noncompliance).
- Robustness strategy: use accounting identity between C-efficiency, compliance gap, and tax policy gap to estimate tax policy gap consistently and conduct robustness checks (see Appendix V referenced).

*Source: IMF Working Paper — Closing the gap: How tax administration performance shapes compliance (content unit: wpiea2025209 - Appendix II).*

### Box 2: Understanding TADAT and Its Value for Assessing Tax Administration Performance

### Box 2: Understanding TADAT and Its Value for Assessing Tax Administration Performance

### What is TADAT?
- TADAT (Tax Administration Diagnostic Assessment Tool) is a standardized, evidence-based framework developed by the IMF and international partners to objectively assess the performance of tax administrations worldwide.
- Launched in 2014, TADAT enables countries to benchmark their tax administration systems against international good practices, identify strengths and weaknesses, and monitor progress over time.
- Uses: inform reforms, help mobilize domestic revenues, and guide capacity development providers.

### Performance Outcome Areas and Scoring
- TADAT evaluates tax administration performance across 9 Performance Outcome Areas (POAs) and 55 dimensions:
  - (POA1) Integrity of the Registered Taxpayer Base
  - (POA2) Effective Risk Management
  - (POA3) Supporting Voluntary Compliance
  - (POA4) Timely Filing of Tax Declarations
  - (POA5) Timely Payment of Taxes
  - (POA6) Accurate Reporting in Declarations
  - (POA7) Effective Tax Dispute Resolution
  - (POA8) Efficient Revenue Management
  - (POA9) Accountability and Transparency
- Scoring: quantitative and qualitative dimensions scored from D or 1 (lowest) to A or 4 (highest).

### Advantages of TADAT
- Standardization & Comparability: first globally harmonized tool enabling objective, cross-country, and over-time comparisons using the same methodology and scoring system.
- Outcome-Focused: emphasizes outcomes rather than inputs or processes.
- Evidence-Based: assessments require documented evidence.
- Comprehensive Coverage: covers 9 key areas and 55 dimensions for a holistic view.
- Supports Reform & Capacity Development: helps identify reform priorities and coordinate international support.
- Peer Learning & Transparency: publication of Performance Assessment Reports (PARs) fosters peer-to-peer learning and accountability.
- Global Adoption: used in over 100 countries, with over 4,000 officials trained in its methodology.

### Data Sources and Key Variables
- Tax administration resources:
  - Revenue Administration operating expenditure as a share of GDP (푒_TA) from ISORA 2018–2022 operational expenditure datasets; GDP data from IMF’s WEO database.
  - Tax administration staff per million inhabitants (퐿_TA) as FTE per million inhabitants from ISORA 2018–2022.
- Social acceptability of the tax system (풂): World Bank’s Rule of Law Annual Index, annual levels for 2010–2023 (ranges from -2.50 to 2.50).
- Other country-level controls: imports-to-GDP (Imp), agriculture-to-GDP (Agr), number of VAT rates, dummies for fragile and conflict-affected states.
- Summary statistics (Table 1 highlights):
  - VAT compliance gap (Reverse RAGAP method): Obs 1,461; Mean 34.46683; Std. dev. 12.87028; Min 0.277445; Max 68.95093
  - Average TADAT score: Obs 1,554; Mean 2.184229; Std. dev. 0.611021; Min 1.085106; Max 3.717283
  - Final consumption per capita (USD PPP): Obs 1,504; Mean 11085.67; Std. dev. 9836.082; Min 491.7177; Max 69644.1
  - VAT general rate (percent): Obs 1,487; Mean 15.87872; Std. dev. 4.105571; Min 3; Max 25
  - Rule of Law Index: Obs 1,430; Mean -0.25939; Std. dev. 0.726583; Min -1.83763; Max 2.124782
  - Imports to GDP: Obs 1,489; Mean 46.63105; Std. dev. 20.70977; Min 10.79023; Max 191.4582
  - Agriculture to GDP: Obs 1,537; Mean 12.33571; Std. dev. 10.25236; Min 0.456543; Max 64.35418
  - FTEs per million in inhabitants: Obs 1,400; Mean 339.6279; Std. dev. 347.2633; Min 20.61503; Max 2028.169
  - Revenue Administration operating expenditure as a share of GDP: Obs 1,428; Mean 0.611655; Std. dev. 4.509766; Min 0.007683; Max 45.89782

### Empirical Strategy and Identification
- Initial inspection: inverse relationship between overall TADAT score (휂_TADAT) and VAT compliance gap (휞*); LIDC cluster toward lower TADAT, higher VAT compliance gaps.
- Alternative composite and sectoral TADAT scores were tested but did not significantly improve model fit relative to overall score.
- Main econometric challenges:
  - Limited time variation in key variables: TADAT scores often available only once; resource variables limited years.
  - Potential endogeneity and reverse causality between compliance and tax administration variables.
- Estimation approach:
  - Mundlak-Krishnakumar framework augmented with Hausman-Taylor estimators to address time-invariant variables and endogeneity.
  - Expanded estimating equation (6): 훤*_{it} = 훼1 + 훼2(X_i) + 훼3(Z_{it−1}) + 훼4(N_i) + T_t + v_i + θ_{it}
    - X_i: time-invariant variables including TADAT score (휂_TADAT), expenditure-to-GDP (푒_TA) or staff per million (퐿_TA)
    - Z_{it−1}: lagged time-varying macro and institutional variables (final consumption per capita I, Imp, Agr, standard VAT rate τ, number of VAT rates, Rule of Law Index a)
    - N_i: exogenous time-invariant instruments (e.g., mean Rule of Law a̅ and log mean agriculture-to-GDP)
    - T_t: time fixed effects; v_i: country random effects; θ_{it}: error term
- Dataset notes: 25 countries in sample have more than one TADAT score (repeat TADAT).

### Main Results (Hausman-Taylor Estimates, Table 2)
- Inverse relationship between VAT compliance gap and TADAT scores robust across specifications (Columns 1–4).
- Estimated elasticities (interpretation in text):
  - Improving tax administration effectiveness by one percent is correlated with a reduction in the VAT compliance gap of between 0.47 and 0.78 percent, all else equal.
- Selected coefficient results (Table 2 highlights):
  - Log (휂_TADAT): Column 1 -0.710**; Column 2 -0.690**; Column 3 -0.466*; Column 4 -0.784**
  - Lagged Log (I): positive and significant (e.g., 0.143*** in Column 1)
  - Lagged (a) (Rule of Law): negative and significant (e.g., -0.127*** in Column 1)
  - Lagged Log (Imp): negative and significant (e.g., -0.195*** in Column 1)
  - Lagged Log (Agr): positive and significant (e.g., 0.119*** in Column 1)
  - Log (No. rates): included in Column 4 with coefficient -0.0645 (not reported as significant)
- Observations and sample:
  - Observations range across columns: 1,315; 1,224; 1,199; 1,265
  - Number of countries: 109; 101; 99; 109
  - Year fixed effects: YES
- Statistical significance: *** p<0.01, ** p<0.05, * p<0.1
- Most variables log-transformed except Rule of Law Index (a).

### Simulations and Fiscal Impact
- Using estimated elasticity -0.71 (Table 2, Column 1) and the specification maximizing observations:
  - Scenario: improvement in a hypothetical tax administration’s TADAT score from 1.85 (approximately a D+) to 2.32 (approximately a C+).
  - Result:
    - Associated with a 6.6 percentage point reduction in the VAT compliance gap — from 42.8 to 36.2 percent.
    - Leads to an increase in VAT revenue equivalent to 0.6 percent of GDP.
  - Under the “consistent evader” assumption, associated reduction in the CIT compliance gap:
    - Using a sample of 12 EME, a reduction of 0.6 percent to GDP in the VAT compliance gap is associated with a reduction of 0.7 percent to GDP in the CIT compliance gap.
    - On this basis, one could expect approximately 1.3 percent of GDP in additional revenue from both taxes.
- Timing: for the eight tax administrations with the highest percent increase in TADAT scores following a second evaluation, improvement from 1.85 to 2.32 took on average 5.8 years (noting model does not account for intertemporal dynamics).

### Additional Findings and Interpretations
- Tax administration operating expenditure to GDP ratio (푒_TA) and staff per million inhabitants (퐿_TA) are not statistically significant predictors of VAT compliance gap in this analysis; possible explanations include measurement error in ISORA data, cross-country institutional differences in reporting, off-budget financing, and that TADAT may better capture the effective use of resources.
- Rule of Law Index:
  - Semi-elasticity estimated at -0.13: improving the index from -0.62 to -0.57 (reflecting average variation among top-improving administrations) could raise revenues by 0.04 percent of GDP on average via reduced VAT and CIT noncompliance.
- Per capita final consumption is positively associated with the VAT compliance gap in the estimates; the result is noted as requiring further research.
- Higher imports-to-GDP ratio is correlated with lower VAT compliance gap; higher agriculture-to-GDP ratio is correlated with higher VAT compliance gap.
- Number of VAT rates (policy complexity proxy) generally not significant; most countries in sample (91 out of 111) remained at two rates during the analysis period.
- Controls for Fragile and Conflict-affected States and Small Island States were not statistically significant.

### Robustness Checks
- Alternative dependent variable: VAT-to-GDP ratio using accounting identity linking C-efficiency, compliance gap, and policy gap (Appendix V).
  - Result: estimated elasticity 0.76; improvement in TADAT score from 1.85 to 2.32 associated with an increase of 0.7 percentage points to GDP in VAT revenue — similar to the VAT compliance gap specification (0.6 percentage points to GDP).
- Subsample robustness: restricting to countries with VAT compliance gap estimated by IMF revenue administration missions yields same sign for TADAT but statistical insignificance due to selection and reduced variation.

### Comparison with Prior Analysis
- Adan et al. (2023) results (total revenue dependent variable):
  - Increase in tax administration strength from 40th to 60th percentile associated with increase in total tax revenues of 1.4 percent of GDP (TADAT strength score) and 1.8 percent (ISORA composite).
- Current study comparable mapping:
  - Equivalent variation in TADAT score corresponds to a decrease in the compliance gap equivalent to additional 0.4 percent of GDP in VAT collection and 0.5 percent of GDP in CIT collection, totaling 0.9 percent of GDP in that comparison.

### Conclusions
- Empirical evidence suggests tax administration effectiveness, as measured by a composite TADAT score, is a key mechanism shaping tax compliance.
- Estimated elasticity: -0.71 (Table 2, Column 1).
- Policy-relevant simulation:
  - Raising a tax administration’s TADAT score from 1.85 (approximately a ‘D+’) to 2.32 (approximately a ‘C+’) is associated with:
    - Additional VAT revenue equivalent to 0.6 percent of GDP from improved compliance.
    - Additional CIT revenue equivalent to 0.7 percent of GDP under a consistent-evader assumption.
    - On this basis, one could expect approximately 1.3 percent of GDP in additional revenue from both taxes.
- Implication: improvements in tax administration performance, measured and benchmarked through TADAT, can yield measurable gains in tax compliance and revenue mobilization, although measurement and causality challenges warrant continued data improvements and further analysis.

*Source: Box 2 from the IMF working paper "Closing the gap: How tax administration performance shapes compliance" (authors’ calculations and referenced datasets).*

### 1.3 percent of GDP in additional revenue from both taxes. The answer to our question thus appears to be: tax

### wpiea2025209 - 1.3 percent of GDP in additional revenue from both taxes. The answer to our question thus appears to be: tax

### Key empirical finding
- Tax administration performance (as a whole, reflecting the interaction of all its key aspects) does shape compliance and appears to matter significantly for compliance and revenue outcomes.
- Strengthening the tax administration system in a comprehensive way — which some countries have achieved over an average period of 5.81 years — could lead to a significant reduction in the compliance gap, resulting in higher tax collection.
- An example aggregate magnitude reported in the study: 1.3 percent of GDP in additional revenue from both taxes.

### Measurement, data, and methodological caveats
- TADAT score data availability: data is only available for the year in which the assessment was conducted; lack of a time series for tax administration performance (effectiveness) variables is a key limitation.
- The study applies a methodology that combines time-invariant variables within a panel data framework as a first step to address the time-series limitation.
- VAT compliance gap measurement imprecision: the VAT compliance gap is an unobservable variable and cannot be measured with complete accuracy based on currently available methodologies.
- Estimates are based on an indirect approach that uses data on C-efficiency, tax expenditures, and national accounts.
- Improvements possible as underlying data are updated and as new individual country (standardized) RA-GAP assessments for the VAT are added to the sample.
- Endogeneity risk: compliance levels may influence tax administration effectiveness (reverse causality). The authors employed recently developed econometric tools to mitigate potential endogeneity effects and note alternative approaches (e.g., event studies) could be used.

### Policy implications and recommended strategy
- Comprehensive strengthening: reducing noncompliance visibly and consistently requires strengthening many aspects of tax administration (registration, filing, payment, compliance risk management, audit, dispute resolution, accountability and transparency — all performance aspects that TADAT measures).
- Targeted one-off reforms are unlikely to yield sustained compliance effects when tax administration capacity is weak.
- Time horizon: the process of improving tax administration to reduce noncompliance may take years (not months).
- Whole-of-government and social acceptability:
  - If the tax administration, along with the whole government, assumes an active role in fostering an environment that promotes the social acceptability of the tax system, it will open another channel for reducing the compliance gap.
  - Tax administrations can be instrumental in leading a whole-of-government approach to tax compliance and in promoting the tax system’s legitimacy.
- Economy structure matters:
  - In countries where the VAT is primarily collected at the border, customs administrations can help improve compliance by strengthening import controls and systematically providing this information to the tax administration as third-party information.
  - Tax administrations face greater challenges in economies dominated by the agricultural sector and must apply appropriately designed compliance risk strategies.

### Literature synthesis (selected points from Appendix I and AI.1)
- Earlier cross-country studies recognized the importance of tax administration effectiveness but were often limited by data:
  - Silvani and Brondolo (1993): correlated VAT design characteristics with compliance gap; noted omitted variable of administration effectiveness.
  - Agha and Haughton (1996): used budgetary resources as percentage of VAT collection to proxy administrative effectiveness.
  - Christie and Holzner (2006), Reckon (2009): used judicial/legal effectiveness and corruption perception index as proxies for audit capacity/institutional robustness.
  - Keen (2015), CASE (2020): incorporated administrative inputs (staff, costs, IT expenditure, audits, e-filing) but noted remaining limitations in capturing performance fully.
  - Butu (2021): expanded models with broader institutional indicators (fiscal freedom, government effectiveness, Human Development Index, Corruption Perceptions Index, share at risk of poverty).
  - Das-Gupta et al. (2016), Crivelli (2018): constructed effectiveness/strength indices but, due to data, used tax collection rather than compliance gaps as the dependent variable.
- This study (Baer and others (2025)):
  - Dependent variable: VAT compliance gap (IMF RA-GAP Reverse Method).
  - Estimation: Panel.
  - Sample: 111 countries, 1,315 observations, 2010-2023.
  - Key tax administration performance findings: TADAT effectiveness (negative sign), ISORA number of resources (staff) (uncertain sign), contribution to promoting greater social acceptability of the tax system (negative sign).

### Theoretical framework and extensions (Appendix II summary)
- Allingham and Sandmo (A-S) compliance risk model:
  - Taxpayer chooses evaded portion Γ to maximize expected utility: Max E[(1 - p)U(I1) + pU(I2)] with Γ as decision variable.
  - Higher probability of detection p → lower optimal Γ*. Higher penalty rate X → lower Γ*.
  - Income I and tax rate τ effects ambiguous without additional assumptions; under decreasing relative risk aversion, Γ* decreases when income declines.
- Indirect tax (sales/VAT) firm model (Sandmo (2005)):
  - Firm chooses production and evasion Γ to maximize expected profits; output decision independent of detection probability and fine, but detection probability and fine affect the sales tax compliance gap.
- Behavioral and other extensions:
  - Include withholding, labor supply interactions, non-monetary fines, moral costs, complexity and compliance costs, perceived benefits of government services, social group influences.
  - Central element preserved: deterrent effects of probability of detection and sanctions.

### VAT-induced CIT compliance gap (Appendix IV)
- Conceptual assumptions:
  - (i) consistent compliance behavior by the representative VAT taxpayer;
  - (ii) the representative VAT taxpayer is also subject to Corporate Income Tax (CIT).
- Illustrative numeric example:
  - Standard VAT rate of 10 percent; $100 sale evaded → VAT evaded $10.
  - Assuming 15 percent CIT rate and consistent evader: VAT-induced CIT evasion = $15.
- Formal relationship (A4.1 and A4.2):
  - VAT-induced CIT Compliance Gap Amount = (τ_CIT / τ) x VAT Compliance Gap Amount.
  - Total CIT Compliance Gap Amount = K x (τ_CIT / τ) x VAT Compliance Gap Amount, where K > 1 represents the additional non-VAT-induced CIT noncompliance proportion.
- Empirical estimation using ECLAC data for 12 Latin American EMEs (top-down RA-GAP methodology):
  - Log-log estimation result (A4.3):
    - Log(CIT Compliance Gap Amount) = 0.00482799 + 0.575622 Log(τ_CIT / τ) + 0.986122 Log(VAT Compliance Gap Amount)
    - R2 Adj. = 0.64
    - N = 12 observations
  - Model provides a fairly good fit for the sample.
- Counterfactual exercise for an LIDC with an average TADAT score:
  - Input values: world average CIT standard rate = 20 percent; world average VAT standard rate = 15 percent; 0.6 percent of GDP variation in the VAT compliance gap from the authors’ exercise.
  - Resulting associated impact: 0.7 percent of GDP in the CIT compliance gap due to VAT-induced variation.
  - Result robust to alternative average rates used for LIDCs (CIT = 22.5 percent, VAT = 13.5 percent).

*Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025209.pdf (IMF Working Paper: Closing the gap: How tax administration performance shapes compliance)*

### Appendix V. Robustness Check: VAT Revenue to

### Appendix V. Robustness Check: VAT Revenue to GDP as the Dependent Variable

### Arithmetic analysis
- Two C-Efficiency identities (following Keen (2013)):
  - 퐶−퐸푓푓 = (100−Γ)/(100−푃)    (5.1)
  - 퐶−퐸푓푓 = 푉퐴푇 / (퐹퐶′ x 휏)   (5.2)
- Combining (5.1) and (5.2) and dividing by GDP yields:
  - 푉퐴푇/퐺퐷푃 = (퐹퐶′ x 휏 x (100−Γ)/(100−푃)) / 퐺퐷푃    (5.3)
- Log-linear empirical specification (from (5.3)):
  - Log(푉퐴푇/퐺퐷푃) = ạ + b Log(퐹퐶′/퐺퐷푃) + c Log(휏) + d Log(100−Γ) + e Log(100−푃) + f    (5.4)
- Alternative formulation expressing Log(100−Γ*) as a function of tax administration performance and other factors:
  - Log(100− Γ*) = 훼′ + 훽′ Log(푒푇퐴) + 훾′ Log(휂푇퐴퐷퐴푇) + 휆′ Log(푎) + 휈′ Log(퐼) + 휌′ Log(휏) + 휀′    (5.5)
- Substituting (5.5) into (5.4) produces an expanded log-linear model (equations (5.6) and (5.7)) where coefficients on Log(휏) and other covariates combine (e.g., (c + d x ρ′) on Log(휏)).

### Empirical setup and controls
- Tax-administration performance measure used: TADAT score (Log(휂푇퐴퐷퐴푇)).
- Additional tax-administration measures considered in alternative specifications:
  - Log(푒푇퐴) (time invariant endogenous in some columns)
  - Log(퐿푇퐴) (staff per million inhabitants, included in Column (5.3))
- Additional regressors and invariants:
  - Rule of Law Index mean, 푎̅ (treated as time-invariant exogenous)
  - Log(Aggr) mean, Log(퐴푔푟)̅̅̅̅̅̅̅̅̅̅̅̅̅ (treated as time-invariant exogenous)
  - Lagged Log(퐼), Lagged(푎), Lagged Log(100−푃), Lagged Log(퐹퐶′/퐺퐷푃), Lagged Log(휏), Lagged Log(Imp), Lagged Log(Agr)
  - Number of VAT rates (Log (No. rates) included in Column (5.4))
- Estimation framework: Mundlak-Krishnakumar to account for time-invariant exogenous variables.
- Transformation: logarithmic transformation for most variables except the rule of law perception index (푎).

### Results (Table AV.1 — Dependent variable: VAT to GDP)
- Coefficient estimates (selected, by column):
  - Log(휂푇퐴퐷퐴푇):
    - (5.1): 0.756***
    - (5.2): 0.957**
    - (5.3): 0.462*
    - (5.4): 0.722***
  - Log(푒푇퐴):
    - (5.2): 0.125
  - Log(퐿푇퐴):
    - (5.3): 0.344***
  - 푎̅:
    - (5.1): 0.0960
    - (5.2): 0.0856
    - (5.3): 0.136
    - (5.4): 0.0555
  - Log(Aggr)̅:
    - (5.2): -0.0212
    - (5.4): 0.00966
  - Lagged Log(퐼):
    - (5.1): -0.0417
    - (5.2): -0.104**
    - (5.3): -0.0805**
    - (5.4): -0.0235
  - Lagged(푎):
    - (5.1): 0.0436
    - (5.2): 0.0367
    - (5.3): 0.0313
    - (5.4): 0.0533*
  - Lagged Log(100−푃):
    - (5.1): 0.647***
    - (5.2): 0.457***
    - (5.3): 0.668***
    - (5.4): 0.632***
  - Lagged Log(퐹퐶′/퐺퐷푃):
    - (5.1): -0.134***
    - (5.2): -0.148***
    - (5.3): -0.132***
    - (5.4): -0.145***
  - Lagged Log(휏):
    - (5.1): 0.357***
    - (5.2): 0.286***
    - (5.3): 0.310***
    - (5.4): 0.478***
  - Lagged Log(Imp):
    - (5.1): 0.142***
    - (5.2): 0.143***
    - (5.3): 0.146***
    - (5.4): 0.133***
  - Lagged Log(Agr):
    - (5.1): 0.0120
    - (5.2): -0.0374
    - (5.3): 0.0186
    - (5.4): 0.00803
  - Log (No. rates):
    - (5.4): -0.0234
- Sample and fixed effects:
  - Observations:
    - (5.1): 1,315
    - (5.2): 1,224
    - (5.3): 1,199
    - (5.4): 1,265
  - Number of Countries:
    - (5.1): 109
    - (5.2): 101
    - (5.3): 99
    - (5.4): 109
  - Year FE: YES in all columns
- Statistical significance legend:
  - *** p<0.01, ** p<0.05, * p<0.1
- Note: Standard errors are not shown for simplicity. P = Policy Gap to Potential. FC’/GDP = Final Consumption to GDP. Source for table: Authors’ own calculations.

### Key interpretation and quantitative finding
- Estimated effectiveness elasticity (Log(휂푇퐴퐷퐴푇)) in Column (5.1): 0.756***.
- Counterfactual magnitude: improving a tax administration’s TADAT score from 1.85 (close to a D+) to 2.32 (close to a C+) implies:
  - An increase of 0.7 percentage points of GDP in VAT revenue.
- Robustness: this VAT-to-GDP specification yields results quite similar to the model using VAT compliance gap as the dependent variable.

*Source: Authors’ own calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025209.pdf_