## Section III.C. — Disentangling Economic and Design Drivers of PIT Revenue Performance

## Source details

**Canonical URL:** [Section III.C. — Disentangling Economic and Design Drivers of PIT Revenue Performance](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022020-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022020-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022020-print-pdf.pdf.json)

---

### Data, Objective, and Estimation Approach
- Objective: Statistically disentangle whether improvements in PIT revenue performance (focus on the 2006–2018 period) are associated with economic characteristics or policy/design variables.
- Dependent variable: PIT revenue as a share of GDP (PIT-to-GDP ratio), using specification in Eq.1.
- Key explanatory variable groups:
  - Economic variables: natural logarithm of GDP per capita; share of agriculture in total GDP; share of self-employment; public wage bill as a share of GDP; inflation; natural resource producing country dummy (robustness check).
  - Policy/design variables: PIT liability threshold (multiples of GDP per capita); PIT minimum non-zero marginal rate; PIT maximum marginal rate; income required to pay the maximum rate (multiples of GDP per capita); dummy whether SSCs can be deducted from the PIT base.
  - Controls: CIT and VAT collection levels (percent of GDP); government effectiveness (robustness).
- Estimation:
  - Lagged instrumental variable (IV) two stage least squares (2SLS) using first lags of covariates as instruments.
  - Robust errors to correct for heteroskedasticity and serial autocorrelation.
  - Interaction of economic and policy vectors with country-group dummies for EME and LIDC; AEs are the base group.
- Data sources and coverage: panel dataset covering 2006–2018 constructed from WoRLD, WEO, WDI, Penn World Tables 9.1, SWIID, EY Personal Tax Guides, and IBFD tax guides.

### Main Regression Results (summary)
- Specification: Main regression reported in Table 1 (Column 1). Robustness checks in Columns 2–6; results broadly stable. Year dummies jointly not statistically different from zero.
- GDP per capita (natural logarithm):
  - Base coefficient (AEs) positive and statistically significant.
  - Marginal effects: a one percent increase in GDP per capita is associated with an increase of the PIT revenue ratio of 0.001 percentage points of GDP in EMEs and 0.002 percentage points of GDP in LIDCs.
- Share of agriculture:
  - Base (AEs) negative but not statistically significant; interactions produce small net effects that are generally statistically insignificant.
- Share of self-employment:
  - Base coefficient (AEs) positive and significant.
  - Interactions for EMEs and LIDCs offset base effect, producing expected negative and statistically significant net effects:
    - LIDCs: a decrease of one percentage point in the share of self-employment is associated with a PIT revenue ratio increase of approximately 0.07 percentage points of GDP.
    - EMEs: a decrease of one percentage point in the share of self-employment is associated with a PIT revenue ratio increase of 0.02 percentage points of GDP.
- Public sector wage bill (share of GDP):
  - Positively and statistically significantly associated with PIT in the base (AEs); effect mitigated in LIDCs and EMEs.
  - Marginal effects of a one percentage point of GDP increase in public wage bill:
    - LIDCs: 0.15 increase in the PIT revenue ratio.
    - EMEs: 0.3 percentage points of GDP increase.
- Inflation:
  - Base (AEs) marginal effect negative but statistically insignificant.
  - Interactions for EMEs and LIDCs offset base effect, resulting in a positive net effect for EMEs and LIDCs (evidence of bracket creep in these groups).
- Tax liability threshold:
  - Base (AEs) coefficient negative; interactions with LIDC and EME dummies positive and larger, yielding a net positive and statistically significant effect on PIT revenue for LIDCs and EMEs:
    - An increase in the threshold equal to one time GDP per capita increases the PIT revenue ratio by:
      - 0.02 percentage points of GDP in LIDCs.
      - 0.04 percentage points of GDP in EMEs.
  - Possible interpretation: increasing exemption thresholds may be accompanied by reductions in other deductions and exemptions, yielding net increases in PIT revenue.
- Lowest non-zero PIT rate:
  - Marginal effect positive and significant.
  - A one percentage point increase in this rate increases the PIT revenue ratio by:
    - 0.05 percentage points of GDP in LIDCs.
    - 0.09 percentage points of GDP in EMEs.
- Top marginal PIT rate:
  - Coefficient is negative and statistically significant at the 1 percent level in the base estimates (AEs); for EMEs the coefficient on the top rate is positive and significant:
    - EMEs: a 1 percentage point increase in the top PIT rate is associated with an increase in the PIT revenue ratio of 0.08 percentage points.
    - LIDCs: net impact (base plus interaction) suggests a small positive net effect of 0.01.
  - Discussion: negative base coefficient may be consistent with top rates being beyond the top of the Laffer curve in AEs.
- SSC deductibility:
  - Coefficient negative and statistically significant overall (base); effect smaller for EMEs; net effect for LIDCs appears positive and statistically significant.
- CIT and VAT controls:
  - CIT revenue coefficient:
    - Base (AEs): negative and statistically significant (substitution between CIT and PIT in AEs).
    - EMEs: interaction positive but small.
    - LIDCs: net effect positive, significant and relatively large — a 1 percentage point rise in CIT revenue is associated with an increase by 0.14 percentage points of GDP in PIT revenue.
  - VAT revenue coefficient:
    - Positive and significant for AEs and LIDCs (rising VAT associated with rising PIT, possibly reflecting administrative improvements).
    - Negative for EMEs (potential substitution effects).
- Robustness and instrumentation:
  - IV (2SLS) regressions using first lags as instruments.
  - Durbin and Wu-Hausman p-value 0.3192 (fail to reject instrument exogeneity).
  - Results stable across robustness checks (Columns 1–6).

### Selected key coefficients (Table 1, Column 1 excerpts)
- Natural logarithm of GDP per capita (constant 2010 US$): 5.722*** (base); EMEs#c. −5.626***; LIDCs#c. −5.506***.
- Self-employed, total (% of total employment): 0.124*** (base); EMEs#c. −0.144***; LIDCs#c. −0.197***.
- Public sector wage bill (% of GDP): 0.891*** (base); EMEs#c. −0.591***; LIDCs#c. −0.742***.
- Inflation, consumer prices (annual %): −0.339 (base); EMEs#c. 0.446*; LIDCs#c. 0.445*.
- Tax liability threshold (multiples of GDP per capita): −6.062*** (base); EMEs#c. 6.106***; LIDCs#c. 6.083***.
- Lowest non-zero marginal PIT rate: 0.126*** (base); EMEs#c. −0.0354*; LIDCs#c. −0.0772***.
- Maximum marginal PIT rate: −0.143*** (base); EMEs#c. 0.223***; LIDCs#c. 0.156***.
- Corporate Income Tax (CIT) Revenue as a % of GDP: −0.638*** (base); EMEs#c. 0.670***; LIDCs#c. 0.777***.
- VAT Revenue as a % of GDP: 0.419*** (base); EMEs#c. −0.435***; LIDCs#c. −0.317**.
- SSC deduction: −1.958*** (base); EMEs#c. 1.086***; LIDCs#c. 2.084***.
- Observations: 839 (Columns vary 839–846); R-squared: 0.870–0.881; Fixed effects: YES; Time effects: NO (except one column with YES).

### Decomposition of PIT-to-GDP Growth (2006–2018) — Aggregate and Group Results (Table 2)
- PIT-to-GDP ratio increases during 2006–2018:
  - AEs: 3.6 percent (equivalent to 0.30 percentage points of GDP).
  - EMEs: 24.3 percent (equivalent to 0.55 percentage points of GDP).
  - LIDCs: 46.5 percent (equivalent to 0.67 percentage points of GDP).
- Contributions (percent change for the period and associated PIT-to-GDP effects in percentage points of GDP):
  - GDP per capita (constant 2010 US$): percent change AEs 12.1, EMEs 21.2, LIDCs 31.2 → associated PIT changes 0.12 (AEs), 0.11 (EMEs), 0.11 (LIDCs).
  - Agriculture, forestry, and fishing (% of GDP): percent change AEs −6.3, EMEs −6.4, LIDCs −16.2 → associated PIT changes 0.00 (AEs), 0.06 (EMEs), −0.06 (LIDCs).
  - Self-employed, total (% of total employment): percent change AEs −7.9, EMEs −10.2, LIDCs −7.5 → associated PIT changes −0.03 (AEs), 1.06 (EMEs), 0.72 (LIDCs).
  - Wage Bill as a Percentage Of GDP: percent change AEs 5.1, EMEs 17.4, LIDCs 29.7 → associated PIT changes 0.08 (AEs), 1.97 (EMEs), 0.35 (LIDCs).
  - Inflation, consumer prices (annual %): percent change AEs −41.9, EMEs −46.5, LIDCs −32.2 → associated PIT changes 0.06 (AEs), −1.26 (EMEs), −0.34 (LIDCs).
  - Tax liability threshold (multiples of GDP per capita): percent change AEs −0.2, EMEs −24.1, LIDCs −1.5 → associated PIT changes 0.00 (AEs), −0.06 (EMEs), 0.00 (LIDCs).
  - Lowest non-zero marginal PIT rate (percent): percent change AEs 1.9, EMEs 4.7, LIDCs −17.3 → associated PIT changes 0.01 (AEs), 0.25 (EMEs), −0.19 (LIDCs).
  - Maximum marginal PIT rate (percent): percent change AEs 1.0, EMEs −13.1, LIDCs −15.6 → associated PIT changes −0.01 (AEs), −1.50 (EMEs), −0.12 (LIDCs).
  - CIT revenue as a % of GDP: percent change AEs −10.7, EMEs 9.2, LIDCs 30.5 → associated PIT changes 0.04 (AEs), 0.04 (EMEs), 0.11 (LIDCs).
  - VAT Revenue as a % of GDP: percent change AEs 4.7, EMEs 7.9, LIDCs 21.8 → associated PIT changes 0.02 (AEs), −0.04 (EMEs), 0.11 (LIDCs).
  - PIT threshold to pay highest marginal rate (multiples of GDP per capita): percent change AEs 44.7, EMEs −24.2, LIDCs −14.8 → associated PIT changes 0.00 (AEs), 0.01 (EMEs), 0.00 (LIDCs).
  - SSC deduction: percent change AEs −2.9, EMEs 4.2, LIDCs −10.6 → associated PIT changes 0.01 (AEs), −0.11 (EMEs), −0.02 (LIDCs).
- Total associated PIT-to-GDP change from covariates: 0.30 (AEs), 0.55 (EMEs), 0.67 (LIDCs) percentage points of GDP.
- Interpretation highlights:
  - GDP per capita growth and public wage bill increases account for important positive contributions to PIT revenue across groups (e.g., wage bill contribution: 1.97 percentage points in EMEs; 0.35 in LIDCs).
  - Inflation and some policy/design changes have negative associated effects (e.g., inflation associated reductions: EMEs −1.26; LIDCs −0.34).
  - In LIDCs many observed changes in PIT design variables are associated with negative changes in PIT-to-GDP ratio (e.g., fall in lowest non-zero marginal PIT rate associated with −0.19 percentage points).
  - In EMEs policy changes (notably reduction in the top marginal rate) are associated with a net mitigation of PIT growth: top marginal rate reduction associated with −1.50 percentage points of GDP in PIT revenue.

### PIT Share within the Tax Structure and Total Tax Revenue (Appendix 2 / Table 3)
- Empirical setup: total tax revenue as a share of GDP regressed on PIT_share and interactions with development group dummies, instrumented with first lags.
- Key coefficients (Table 3 / Appendix 2 specifications):
  - PIT_share (AE base): 0.288*** (Column 1); 0.295*** (Column 2); 0.284*** (Column 3).
  - LIDC#c.PIT_share interaction: −0.0629** (Column 1); −0.0904*** (Column 2); −0.0964*** (Column 3).
  - EME#c.PIT_share interaction: −0.191*** (Column 1); −0.222*** (Column 2); −0.229*** (Column 3).
  - VAT_share baseline: 0.0675*** (Appendix 2); VAT interactions not strongly significant.
  - TradeTaxes_share baseline: −0.308*** with positive interactions for LIDCs and EMEs in Appendix 2 specifications.
- Interpretation:
  - In AEs a higher PIT_share is strongly associated with higher total tax revenue.
  - In LIDCs the positive association of PIT_share with total tax revenue is weaker (interaction negative and significant).
  - EMEs are intermediate between LIDCs and AEs.
  - Trade taxes, excises, property taxes and CIT shares can be relatively more conducive to total revenue in LIDCs than in AEs, per interaction patterns.

### PIT Progressivity, Redistributive Capacity, and Caveats (Box 2)
- Measures referenced:
  - Reynolds–Smolensky (difference between before- and after-tax Gini).
  - Kakwani (difference between Gini of before-tax income and Gini of the income tax), decomposes redistribution into progressivity and size components.
  - Dardanoni and Lambert (2002) transplant approach to make redistribution indices comparable across countries by using a common pre-tax distribution.
  - Benítez and Vellutini (2021) implement Dardanoni and Lambert methodology to estimate progressive and redistributive capacity indices (Figure 12 referenced).
- Comparative findings (2006–2017 summarized):
  - Progressive capacity often higher in LIDCs than in AEs.
  - Aggregate PIT tax rate much lower in LIDCs than in AEs.
  - Redistributive capacity (difference in Gini) markedly lower in LIDCs than in AEs.
    - Median redistributive capacity: 1.7 in LIDCs and 3.7 Gini points in AEs (2018).
  - Redistributive capacity of LIDCs comparable to EMEs, but:
    - Progressive capacity generally lower in EMEs (many EMEs have flat-rate PIT regimes).
    - Aggregate PIT tax rate higher in EMEs than in LIDCs.
- Caveats:
  - Capital income taxation is typically lower in LIDCs (Appendix 3).
  - Widespread exemptions are commonplace and likely limit effectiveness of PIT in LIDCs more than in other groups.
  - Policy factors not or imperfectly captured would typically erode measured redistributive impact in LIDCs.

### Appendix 3 — Stylized Facts on Treatment of Capital Income and Small Business Regimes
- Interest:
  - Average rates on returns to savings (interest income): AEs 18 percent; EMEs 18 percent; LIDCs 15.
- Dividends (average nominal dividend rates in 2019):
  - AEs averaged 22 percent.
  - EMEs averaged 13 percent.
  - LIDCs averaged 13 percent.
- Capital gains taxation approaches:
  - Share of countries using a special rate for capital gains: AEs 57 percent; EMEs 41 percent; LIDCs 60 percent.
  - Share taxing capital gains as part of ordinary income: AEs 37 percent; EMEs 31 percent; LIDCs 26 percent.
- Presumptive/special regimes for small business:
  - Prevalence: LIDCs almost 90 percent; EMEs 59 percent; AEs 35 percent.
  - Optional regimes can create arbitrage opportunities.
- Relative importance of CIT in total tax revenues:
  - EMEs: CIT represents 19 percent of total tax revenues.
  - LIDCs: CIT represents 16 percent of total tax revenues.
  - AEs: CIT represents 10 percent of total tax revenues.

### Key policy-relevant implications from the reported analysis
- Economic structure and administrative capacity matter strongly:
  - GDP per capita growth, expansion of the public wage bill, and declines in share of self-employment are major drivers of PIT revenue increases in EMEs and LIDCs.
  - Digitalization and improved information systems can help monitor self-employed and capital income and boost PIT performance (noted in Section II).
- PIT design changes during 2006–2018 often contributed negatively to PIT revenue in LIDCs and EMEs (reductions in lowest non-zero and top marginal rates).
- As countries develop a larger fraction of additional tax revenue tends to come from the PIT; this association is much weaker in LIDCs, implying reliance on other taxes (indirect taxes, trade taxes) for revenue gains in low-income contexts.
- Redistributive potential:
  - PIT can be relatively progressive in LIDCs but its redistributive impact is constrained by the small aggregate PIT rate and by exemptions and weak capital income taxation.
  - There is scope in EMEs to raise revenue and improve progressivity through better PIT design (e.g., addressing flat-rate regimes, capital income taxation, and tax expenditures).
- Interpretation caveats:
  - Regression results may be affected by omitted variables; further research is suggested, including explicit information on PIT tax expenditures to understand counterintuitive threshold results.

*Source: IMF Working Paper — Progress of the Personal Income Tax in Emerging Markets and Developing Countries (Section III.C., Appendices 2–3, Box 2).*

### References .............................................................................................................

### References

### I. Introduction and Scope
- Focus: Progress of the Personal Income Tax (PIT) in Emerging Markets and Developing Countries.
- Dataset: Assembled PIT characteristics of 157 countries over the 2006–2018 period, covering revenue data and tax design data (rates, thresholds, deductions, tax credits).
- Core analytical objectives:
  - Examine main drivers of PIT growth (structural economic factors vs. policy changes) focusing on 2006–2018.
  - Explore association between PIT share in total revenue and overall revenue performance.
  - Assess contribution of PIT growth to inequality reduction using a novel index of redistributive capacity based on simulated microdata.

### II. Key Findings (summary)
- PIT levels by development group (2019):
  - Advanced Economies (AEs): PIT raised revenue averaging 8.6 percent of GDP.
  - Low-Income Developing Countries (LIDCs): PIT averaged 2.1 percent of GDP.
  - Emerging Market Economies (EMEs): PIT averaged 3.1 percent of GDP.
- Historical growth (1990–2019):
  - LIDCs: PIT revenue grew from 1 percent of GDP in 1990 to 2.1 percent in 2019.
  - EMEs: PIT revenue grew from 1.9 percent of GDP in 1990 to 3.1 percent in 2019.
  - AEs: PIT revenue fell from 9.6 percent of GDP in 1990 to 8.6 percent in 2019.
- Drivers of PIT growth (2006–2018):
  - Primary positive contributors: increase in GDP per capita, growth in public sector wage bill, decrease in share of agriculture and self-employment (both associated with informal sector size).
  - Policy changes (liability threshold, lowest non-zero marginal rate, top PIT rate): statistically significant but net negative effect on PIT revenue in LIDCs and EMEs, consistent with observed reductions in lowest non-zero and top marginal rates.
- PIT share vs. total tax revenue:
  - Positive association between the share of PIT within the tax structure and total tax revenue collection—implying revenue increases are more than proportional due to PIT improvements.
  - This positive association is weaker in LIDCs than in EMEs and AEs, indicating LIDCs have relied relatively more on other taxes (e.g., indirect taxes, trade taxes) for revenue gains.
- Redistributive capacity:
  - PIT redistributive power is non-negligible but lower than in AEs, mainly because PIT revenue as a share of GDP remains smaller.
  - Counter to common beliefs, PIT is relatively progressive in LIDCs; many EMEs, by contrast, do not show comparable progressivity.

### III. PIT Design Considerations for LIDCs and EMEs
- Tax base composition:
  - Primary taxable income forms: earnings from employment, income from self-employment, profits from unincorporated businesses, returns to capital (interest, dividends, rents, royalties, capital gains).
- Institutional and economic constraints:
  - LIDCs and EMEs have larger shares of self-employment and agriculture and smaller shares of labor income earned by employees.
  - LIDCs feature lower literacy and less developed institutional arrangements to measure income and enforce taxes.
  - Informal economy prevalence constrains PIT coverage and enforcement; many LIDCs exempt income from agriculture.
- Administrative capacity implications:
  - ISORA evidence: higher-income countries have lower cost of collection (0.89 currency units to collect 100 currency units) versus 1.10 for lower-income countries.
  - Digitalization and better information systems can improve monitoring and compliance for self-employed and capital income.

### IV. Revenue Mix and Policy Lessons
- Average tax revenue mix (2019, share of total tax and social security contribution revenue):
  - Indirect taxes dominance in developing countries: indirect taxes raised 47.7 percent (LIDCs) and 40.1 percent (EMEs).
  - PIT share of tax+SSCs: 13.4 percent (LIDCs) and 12.7 percent (EMEs).
- Developmental sequence:
  - Historically and empirically, countries tend to rely first on indirect taxation, then gradually expand PIT as administrative and economic conditions improve.
  - Examples: some EMEs retain single flat-rate PITs; some resource-rich countries have yet to levy PIT; several LIDCs adopted progressive PIT designs from the outset.
- Efficiency vs. equity trade-offs:
  - Income taxes can have higher efficiency costs in contexts with limited institutional capacity and may be more easily evaded/avoided.
  - Empirical literature suggests consumption taxes tend to be more growth-friendly; however, a well-designed and enforced PIT can raise additional revenue and improve progressivity while managing efficiency costs.

### V. Historical Context (Box 1)
- PIT adoption in AEs:
  - Mid-nineteenth century onward; UK consolidated PIT in 1842; Switzerland adopted PIT in 1939.
  - Early PITs were low-rate, often flat; strong increase in progressivity occurred in early twentieth century (e.g., France top marginal rate to 50 percent in 1920 from 2 percent in 1915; U.S. top marginal rate to 67 percent in 1917 from 7 percent in 1913).
  - By 1939 PIT was dominant in many AEs (44 percent of tax revenue in the US; 47 percent in Germany).
  - Development of welfare state and social spending reinforced PIT compliance incentives.
- PIT rollout in EMEs and LIDCs:
  - By 1950, two thirds of EMEs had introduced PIT; equivalent share for LIDCs was reached only by 1990.

### VI. Empirical Strategy and Sections Overview
- Analytical structure in paper:
  - Section II: Reviews PIT design issues in context of LIDCs and EMEs.
  - Section III: Analyzes economic and policy determinants of PIT revenue performance across country groups; correlation of tax structures and tax revenue performance.
  - Section IV: Examines redistributive effect of PIT using novel redistributive capacity index based on simulated microdata.
  - Section V: Concluding remarks.
- Empirical components referenced:
  - Figures: Self-employment, Agriculture and Urbanization (2019); Average Tax Revenue Mix by Development Group (2019); PIT Revenue (1990–2019); PIT Revenue as a Share of Tax Revenue, Excluding SSCs; PIT Revenue Growth Index 1990–2019 (1990=100); Evolution of PIT liability threshold (multiples of GDP per capita); Average non-zero lowest marginal PIT rate; Top marginal PIT rate; Share of PIT revenue growth explained by policy/economic variables (2006–2018); Share of PIT in revenue and revenue performance by development group; Evolution of disposable income inequality; Progressive and redistributive capacities of the PIT by country groups (2006–2017).
  - Tables: PIT Revenue and its Economic and Policy Determinants (2006–2018); Accounting for the Increase in PIT revenue in 2006–2018; Tax Revenue and Tax Shares (2006–2017).
  - Appendices: Summary Statistics; Tax Revenue and Tax Shares – Additional Regressions; Stylized Facts of the PIT’s Treatment of Capital Income.

*Source: IMF Working Paper — Progress of the Personal Income Tax in Emerging Markets and Developing Countries*

### Section III.C.

### Section III.C.

### B. Key Changes in Measurable PIT Policy Variables
- The PIT is levied on taxable income (gross income adjusted for allowances).  
- The standard/basic allowance (PIT liability threshold) exempts a certain level of income for all individuals and:
  - Allows a basic level of consumption to be income tax-free.
  - Excludes low-income individuals from the PIT.
  - Lends basic progressivity even when a single (“flat”) rate is used (a “flat” rate existed in 15 percent of sampled EMEs in 2020).
- Allowances can be implemented as deductions from taxable income or as credits against PIT liability (Zee 2005).
- Liability threshold differences (2019 averages):
  - LIDCs: an individual needs to earn at least 1.3 times GDP per capita to be liable for PIT.
  - EMEs: 0.8 times GDP per capita.
  - AEs: 0.2 times GDP per capita.
- Liability threshold trends (2006-2019, LIDCs):
  - 2006: 1.4 times GDP per capita.
  - 2014: maximum 1.9 times GDP per capita.
  - 2019: decreased to 1.3 times GDP per capita.
- Lowest non-zero marginal PIT rate (averages):
  - Declined in LIDCs from approximately 12.8 percent in 2006 to 10.4 percent in 2019.
  - Relatively stable in EMEs and AEs (Figure 7).
- Top marginal PIT rate trends (2006–2019):
  - Slight increase in AEs by 0.5 percentage points.
  - Declined for LIDCs by 3.4 percentage points.
  - Declined for EMEs by one percentage point.
- Levels of top marginal rates (most recent reported):
  - AEs: 39 percent.
  - LIDCs: 29 percent.
  - EMEs: 24 percent (with an uptick in recent years).

### C. Disentangling the Economic from the Design Drivers of the PIT Revenue Performance — Specification and Estimation
- Objective: Statistically disentangle whether improvements in PIT revenue performance (focus on the 2006-2018 period) are associated with economic characteristics or policy/design variables.
- Dependent variable: PIT revenue as a share of GDP (PIT-to-GDP ratio), using specification in Eq.1.
- Key explanatory variable groups:
  - Economic variables (휀): natural logarithm of GDP per capita, share of agriculture in total GDP, share of self-employment, public wage bill as a share of GDP, inflation, and a natural resource producing country dummy (robustness check).
    - Logs used for GDP per capita to capture potential non-linearity.
  - Policy/design variables (휏): PIT liability threshold (multiples of GDP per capita), PIT minimum non-zero marginal rate, PIT maximum marginal rate, income required to pay the maximum rate (multiples of GDP per capita), and a dummy indicating whether SSCs can be deducted from the PIT base.
    - Expected signs: liability threshold, top threshold, and SSC deduction dummy expected to have negative signs; minimum non-zero rate and maximum rate expected to be positively related to PIT revenue.
- Control variables: CIT and VAT collection levels (percent of GDP) to capture substitution or complementarity effects; government effectiveness estimate included as robustness.
- Estimation approach:
  - Lagged instrumental variable (IV) two stage least squares (2SLS) to address simultaneity, using lagged covariates as instruments.
  - Robust errors to correct for heteroskedasticity and serial autocorrelation.
  - Interaction of economic and policy vectors with country-group dummies for EME and LIDC to estimate group-specific marginal effects; AEs are the base group.
- Data sources and coverage:
  - Panel dataset covering 2006-2018.
  - Constructed using the IMF World Revenue Longitudinal Database (WoRLD); the World Economic Outlook dataset (WEO); the World Development Indicators (WDI); the Penn World Tables 9.1; the Standardized World Income Inequality Database; and PIT rates and thresholds from EY Personal Tax Guides and IBFD tax guides.

### C. Disentangling the Economic from the Design Drivers of the PIT Revenue Performance — Results (summary of main findings)
- Specification overview:
  - Main regression reported in Table 1 (Column 1). Columns 2–6 present robustness checks: dropping share of agriculture (Column 2); dropping share of self-employment (Column 3); adding natural resource dummy (Column 4); adding government effectiveness (Column 5); adding year dummies (Column 6).
  - Year dummies in Column 6 are jointly not statistically different from zero.
- GDP per capita (natural logarithm):
  - Base coefficient (AEs) is positive and statistically significant.
  - For EMEs and LIDCs the coefficient is smaller (interaction with dummies).
  - Marginal effects stated: a one percent increase in GDP per capita is associated with an increase of the PIT revenue ratio of 0.001 percentage points of GDP in EMEs and 0.002 percentage points of GDP in LIDCs.
- Share of agriculture:
  - Base coefficient (AEs) negative but not statistically significant.
  - Interaction with EME dummy more than offsets base negative effect, but net impact remains statistically insignificant.
  - For LIDCs, interaction offsets negative base estimate, yielding a small positive net coefficient, statistically insignificant.
  - Exception: Column 3 (dropping self-employment) where all coefficients show expected negative signs but are not statistically significant.
- Share of self-employment:
  - Base coefficient (AEs) positive and significant (unexpected).
  - For EMEs and LIDCs, interactions more than offset base effect, producing expected negative and statistically significant net effects:
    - LIDCs: a decrease of one percentage point in the share of self-employment is associated with a PIT revenue ratio increase of approximately 0.07 percentage points of GDP.
    - EMEs: a decrease of one percentage point in the share of self-employment is associated with a PIT revenue ratio increase of 0.02 percentage points of GDP.
- Public sector wage bill (share of GDP):
  - Positively and statistically significantly associated with PIT in the base (AEs).
  - Effect is mitigated in LIDCs and EMEs relative to AEs.
  - Marginal effects reported: an increase in one percentage point of GDP of the public wage bill yields:
    - 0.15 increase in the PIT revenue ratio in LIDCs.
    - 0.3 percentage points of GDP increase in EMEs.
- Inflation:
  - Base coefficient (AEs) marginal effect negative but statistically insignificant (contrary to expectations).
  - Interactions for EMEs and LIDCs offset this, resulting in a positive net effect for EMEs and LIDCs, pointing to the existence of bracket creep in these groups.
- Tax liability threshold (policy variable):
  - Base regression (AEs) coefficient negative.
  - Interactions with LIDC and EME dummies are positive and larger, producing a counterintuitive net positive and statistically significant effect on PIT revenue for LIDCs and EMEs:
    - An increase in the threshold equal to one time GDP per capita increases the PIT revenue ratio, all else equal, by:
      - 0.02 percentage points of GDP in LIDCs.
      - 0.04 percentage points of GDP in EMEs.
  - Possible explanation: increasing exemption thresholds may be accompanied by reductions in other deductions and exemptions, yielding net increases in PIT revenue; further research suggested to incorporate information about PIT tax expenditures.
- Lowest non-zero PIT rate:
  - Marginal effect positive and significant.
  - A one percentage point increase in this rate increases the PIT revenue ratio by:
    - 0.05 percentage points of GDP in LIDCs.
    - 0.09 percentage points of GDP in EMEs.
- Top marginal PIT rate:
  - Coefficient is negative and statistically significant at the 1 percent level in the base estimates.
  - May be consistent with top rates being beyond the top of the Laffer curve in AEs (discussion truncated in source).
- Robustness:
  - Results across Columns 1–6 are broadly stable; inclusion of year dummies does not materially change main findings.

*IMF Working Papers — Progress of the Personal Income Tax in Emerging Markets and Developing Countries (Section III.C.)*

### conclusion might also be premature, as the regression might suffer from omitted variable that determine the tax

### wpiea2022020-print-pdf - conclusion might also be premature, as the regression might suffer from omitted variable that determine the tax

### Regression findings on PIT determinants
- For EMEs the coefficient of the top marginal PIT rate is positive and significant: a 1 percentage point increase in the top PIT rate comes along with an increase in the PIT revenue ratio of 0.08 percentage points.
- For LIDCs the net impact of base coefficient and the interaction with the dummy variable suggest a statistically significant small positive net effect of 0.01.
- The coefficient for the deductibility of SSCs from the PIT base is negative and statistically significant overall; this effect is smaller for EMEs, while the net effect for LIDCs appears to be positive and statistically significant.
- The coefficient for the threshold of the top PIT rate is not statistically different from zero across AEs, EMEs and LIDCs and across specifications.
- The coefficient for CIT revenue is negative and statistically significant for the base regression (AEs), indicating substitution between CIT and PIT in AEs.
  - In EMEs the interaction term is positive but small (mitigating the negative base effect).
  - In LIDCs the net effect of CIT is positive, significant and relatively large: a 1 percentage point rise in CIT revenue is associated with an increase by 0.14 percentage points of GDP in PIT revenue.
- The coefficient for VAT revenue:
  - Positive and significant for AEs and LIDCs, indicating rising VAT is associated with rising PIT (possibly reflecting improvements in revenue administration).
  - Negative for EMEs, indicating potential substitution effects.
- Natural resource producer dummy coefficients are not statistically different from zero (positive for AEs and LIDCs, negative for EMEs, but insignificant).
- Government effectiveness (robustness test, Column 5) is positive and statistically significant for the base (AEs). Interactions for EMEs and LIDCs yield smaller positive and statistically significant coefficients. Most coefficients remain robust in sign and size with inclusion of government effectiveness, except natural logarithm of GDP per capita which is smaller in magnitude in this specification.

### Key coefficients and statistical signals (selected from Table 1)
- Natural logarithm of GDP per capita (constant 2010 US$): 5.722*** (base), with EMEs#c. −5.626*** and LIDCs#c. −5.506*** (Column 1).
- Self-employed, total (% of total employment): 0.124*** (base); EMEs#c. −0.144***; LIDCs#c. −0.197***.
- Public sector wage bill (% of GDP): 0.891*** (base); EMEs#c. −0.591***; LIDCs#c. −0.742***.
- Inflation, consumer prices (annual %): −0.339 (base) with EMEs#c. 0.446* and LIDCs#c. 0.445* (Column 1).
- Tax liability threshold (multiples of GDP per capita): −6.062*** (base); EMEs#c. 6.106***; LIDCs#c. 6.083***.
- Lowest non-zero marginal PIT rate: 0.126*** (base); EMEs#c. −0.0354*; LIDCs#c. −0.0772***.
- Maximum marginal PIT rate: −0.143*** (base); EMEs#c. 0.223***; LIDCs#c. 0.156***.
- Corporate Income Tax (CIT) Revenue as a % of GDP: −0.638*** (base); EMEs#c. 0.670***; LIDCs#c. 0.777***.
- VAT Revenue as a % of GDP: 0.419*** (base); EMEs#c. −0.435***; LIDCs#c. −0.317**.
- PIT threshold to pay highest marginal rate (multiples of GDP per capita): −0.0151 (base); EMEs#c. 0.0140; LIDCs#c. 0.0145.
- SSC deduction: −1.958*** (base); EMEs#c. 1.086***; LIDCs#c. 2.084***.
- Observations: 839 (Columns vary 839–846); R-squared: 0.870–0.881 (specifications); Fixed effects: YES; Time effects: NO (except one column with YES).
- Endogeneity / instrumentation: IV (2SLS) regressions used; instruments are first lags of economic variables; Durbin and Wu-Hausman p-value is 0.3192 (fail to reject instrument exogeneity).

### Decomposition of PIT-to-GDP growth (2006–2018) — aggregate and group results (from Table 2)
- PIT-to-GDP ratio increases during 2006–2018:
  - AEs: 3.6 percent (equivalent to 0.30 percentage points of GDP).
  - EMEs: 24.3 percent (equivalent to 0.55 percentage points of GDP).
  - LIDCs: 46.5 percent (equivalent to 0.67 percentage points of GDP).
- Contributions (percent change for the period and associated PIT-to-GDP effects in percentage points of GDP):
  - GDP per capita (constant 2010 US$): percent change AEs 12.1, EMEs 21.2, LIDCs 31.2 → associated PIT changes 0.12 (AEs), 0.11 (EMEs), 0.11 (LIDCs).
  - Agriculture, forestry, and fishing, value added (% of GDP): percent change AEs −6.3, EMEs −6.4, LIDCs −16.2 → associated PIT changes 0.00 (AEs), 0.06 (EMEs), −0.06 (LIDCs).
  - Self-employed, total (% of total employment): percent change AEs −7.9, EMEs −10.2, LIDCs −7.5 → associated PIT changes −0.03 (AEs), 1.06 (EMEs), 0.72 (LIDCs).
  - Wage Bill as a Percentage Of GDP: percent change AEs 5.1, EMEs 17.4, LIDCs 29.7 → associated PIT changes 0.08 (AEs), 1.97 (EMEs), 0.35 (LIDCs).
  - Inflation, consumer prices (annual %): percent change AEs −41.9, EMEs −46.5, LIDCs −32.2 → associated PIT changes 0.06 (AEs), −1.26 (EMEs), −0.34 (LIDCs).
  - Tax liability threshold (multiples of GDP per capita): percent change AEs −0.2, EMEs −24.1, LIDCs −1.5 → associated PIT changes 0.00 (AEs), −0.06 (EMEs), 0.00 (LIDCs).
  - Lowest non-zero marginal PIT rate (percent): percent change AEs 1.9, EMEs 4.7, LIDCs −17.3 → associated PIT changes 0.01 (AEs), 0.25 (EMEs), −0.19 (LIDCs).
  - Maximum marginal PIT rate (percent): percent change AEs 1.0, EMEs −13.1, LIDCs −15.6 → associated PIT changes −0.01 (AEs), −1.50 (EMEs), −0.12 (LIDCs).
  - CIT revenue as a % of GDP: percent change AEs −10.7, EMEs 9.2, LIDCs 30.5 → associated PIT changes 0.04 (AEs), 0.04 (EMEs), 0.11 (LIDCs).
  - VAT Revenue as a % of GDP: percent change AEs 4.7, EMEs 7.9, LIDCs 21.8 → associated PIT changes 0.02 (AEs), −0.04 (EMEs), 0.11 (LIDCs).
  - PIT threshold to pay highest marginal rate (multiples of GDP per capita): percent change AEs 44.7, EMEs −24.2, LIDCs −14.8 → associated PIT changes 0.00 (AEs), 0.01 (EMEs), 0.00 (LIDCs).
  - SSC deduction: percent change AEs −2.9, EMEs 4.2, LIDCs −10.6 → associated PIT changes 0.01 (AEs), −0.11 (EMEs), −0.02 (LIDCs).
- Total associated PIT-to-GDP change from covariates (Table 2, Column “Total”): 0.30 (AEs), 0.55 (EMEs), 0.67 (LIDCs) percentage points of GDP.
- Interpretation highlights:
  - GDP per capita and public wage bill increases account for important positive contributions to PIT revenue across groups (e.g., wage bill contribution: 1.97 percentage points in EMEs; 0.35 in LIDCs).
  - Inflation and some policy/design changes have negative associated effects (e.g., inflation associated reductions for EMEs −1.26 and LIDCs −0.34).
  - In LIDCs many observed changes in PIT design variables are associated with negative changes in PIT-to-GDP ratio (e.g., lowest non-zero marginal PIT rate in LIDCs fell from 12 in 2006 to 10 in 2018 and is associated with a −0.19 percentage points of GDP reduction in PIT revenue).
  - In EMEs policy changes (notably reduction in the top marginal rate) are associated with a net mitigation of PIT growth: top marginal rate reduction associated with −1.5 percentage points of GDP in PIT revenue.

### PIT in the broader tax structure (association of PIT share with total tax revenue)
- Empirical setup: total tax revenue as a share of GDP regressed on PIT_share and interactions with development group dummies, instrumented (2SLS) with first lags.
- Key regression results (Table 3):
  - PIT_share coefficient (AE base): 0.288*** (Column 1); 0.295*** (Column 2); 0.284*** (Column 3).
  - LIDC#c.PIT_share interaction: −0.0629** (Column 1); −0.0904*** (Column 2); −0.0964*** (Column 3).
  - EME#c.PIT_share interaction: −0.191*** (Column 1); −0.222*** (Column 2); −0.229*** (Column 3).
  - VAT_share coefficient (added in Column 2): 0.0920***; VAT interactions show smaller cross-group variation.
  - TradeTaxes_share coefficient (added in Column 3): −0.100***; interactions with development groups not strongly significant.
  - Observations: 5,636; R-squared: 0.326–0.362.
- Interpretation:
  - In AEs a higher PIT_share is strongly associated with higher total tax revenue.
  - In LIDCs the association of PIT_share with total tax revenue is positive but markedly weaker (interaction negative and significant), implying that increases in total tax revenue in LIDCs are less associated with PIT than in AEs.
  - EMEs are intermediate between LIDCs and AEs.
  - In contrast to PIT, shares of trade taxes, excise taxes, property taxes and CIT are relatively more conducive to total revenue in LIDCs than in AEs (when tested in isolation).
- Takeaway: As countries develop, a larger fraction of additional tax revenue comes from the PIT, but this relationship is much weaker within LIDCs, suggesting a development threshold beyond which PIT becomes a more significant revenue contributor.

### PIT progressivity and inequality context
- Market income is unequally distributed; PIT can provide redistribution via progressive design.
- Inequality remains a pressing issue in LIDCs and EMEs; disposable income inequality (Gini) on average remains significantly higher in LIDCs and EMEs than in AEs.
- The Musgrave-Thin index is noted as a common measure of the redistributive effect of PIT (difference between before- and after-tax Gini). It decomposes into progressivity and size (aggregate tax rate) components.

*Source: IMF Working Paper — Progress of the Personal Income Tax in Emerging Markets and Developing Countries (sections and tables excerpted from the provided PDF).*

### Box 2. Measures of Progressivity and Redistribution

### Box 2. Measures of Progressivity and Redistribution

### Measures and indices of progressivity and redistribution
- Traditional (Pigou 1928) measure: ratio of the change in the average tax rate to the change in taxable income for a given level of taxable income.
- Peter, Buttrick, and Duncan (2010) measure: captures progressivity over the full distribution by calculating the average tax rate progression over 100 data points in the before-tax income distribution (estimated as the slope of a regression of the average tax rate on income).
- Reynolds and Smolensky (1977) approach: measures redistributive effects of income tax as the difference between the Gini coefficients of before-tax income and after-tax income (Reynolds–Smolensky index). Captures redistribution over the entire income range.
- Kakwani (1977) progressivity index: defined as the difference between the Gini coefficients of before-tax income and the income tax. Kakwani decomposed total redistributive effects into:
  - Progressivity, and
  - The “size” of taxation in the aggregate, as measured by the aggregate tax rate (the average tax rate in the economy).
- Limitation: both Reynolds–Smolensky and Kakwani indexes are functions of the before-tax income distribution and therefore do not fully reflect the intrinsic progressivity of the tax system.
- Dardanoni and Lambert (2002) methodological contribution: “transplant” tax (or transfer) regimes into a common base with an identical pre-tax distribution to make redistribution indices comparable across countries and years.
- Benítez and Vellutini (2021) implement the Dardanoni and Lambert methodology to estimate progressive and redistributive capacity indices reported in Figure 12.

### Comparative findings by country group
- Progressivity and redistributive outcomes (2006–2017 observations summarized):
  - Progressive capacity (how fast average tax rates change with taxable income):
    - Often higher in LIDCs than in AEs (upper left panel of Figure 12).
  - Aggregate tax rate of the PIT:
    - Generally much lower in LIDCs than in AEs (upper right panel of Figure 12).
  - Redistributive capacity (difference in Gini coefficients before and after tax):
    - Markedly lower in LIDCs than in AEs (lower left panel of Figure 12).
    - Median redistributive capacity: 1.7 in LIDCs and 3.7 Gini points in AEs (2018).
    - Redistributive capacity of LIDCs is comparable to that of EMEs, but:
      - Progressive capacity is generally lower in EMEs,
      - Aggregate tax rate is higher in EMEs than in LIDCs.
    - The relatively low progressivity observed in EMEs may be explained by the important number of countries with a flat-rate PIT regime within this group.
- Interpretation:
  - A median capacity of PIT in LIDCs to reduce income inequality by about 1.7 Gini points is characterized as far from negligible, despite being lower than in AEs.
  - There is scope for improving progressivity and reducing inequality in EMEs through better PIT design.

### Caveats affecting measured redistributive impact
- Policy factors not or imperfectly captured here would typically erode the redistributive impact of the PIT in LIDCs.
- Important caveats:
  - Capital income taxation is typically lower in LIDCs (Appendix 3).
  - Widespread exemptions are commonplace in all development groups but are likely to limit the effectiveness of PIT in LIDCs more than in other country groups (Coady, Gupta, and Bastagli 2015).

### Key conclusions and policy-relevant findings (from concluding remarks)
- The PIT serves as an effective revenue and redistribution instrument in AEs.
- In LIDCs and EMEs:
  - PIT revenue and redistributive effects remain relatively modest compared with AEs.
  - Over the last fifteen years, PIT revenue rose markedly faster in LIDCs and EMEs than in AEs.
  - The improvement in PIT revenue has been mostly associated with changes in economic variables rather than policy changes. Key drivers include:
    - GDP per capita growth,
    - Growth in the size of the public sector wage bill,
    - Decline in the share of self-employment, which further supported PIT revenue.
  - Policy changes during the 2006–2018 period contributed negatively to PIT revenue in LIDCs and EMEs.
  - Improvements in CIT revenue go together with rising PIT in both EMEs and LIDCs, possibly reflecting improvements in tax administration capacity.
- Development and revenue pattern:
  - As economies grow and raise more tax revenue, a larger fraction comes from the PIT.
  - This relationship is much weaker within LIDCs, suggesting additional revenue there comes from all revenue sources jointly rather than predominantly from the PIT.
  - This pattern is different in EMEs and especially in AEs, consistent with the existence of a development threshold beyond which the PIT becomes an increasingly significant contributor of tax revenue.
- Overall assessment:
  - Redistributive capacity of the PIT in LIDCs is significant due to progressive PIT design but constrained by the small size of the aggregate PIT rate.
  - Redistribution through the PIT in LIDCs is considerably smaller than in AEs but similar to EMEs.
  - Policy implication: there is scope to raise revenue, improve progressivity, and reduce inequality in EMEs through better PIT design.

*Box 2. Measures of Progressivity and Redistribution — IMF Working Paper content unit*

### Appendix 2. Tax Revenue and Tax Shares –

### Appendix 2. Tax Revenue and Tax Shares – Additional Regressions

### Regression findings (summary of reported coefficients)
- Sample: Observations = 5,636 for all specifications.
- R-squared by specification:
  - (1) 0.326
  - (2) 0.273
  - (3) 0.309
  - (4) 0.271
  - (5) 0.265
  - (6) 0.271
- Constant (AE) by specification (standard errors in parentheses):
  - (1) 15.40*** (0.554)
  - (2) 23.24*** (0.408)
  - (3) 25.37*** (0.201)
  - (4) 27.75*** (0.509)
  - (5) 26.33*** (0.388)
  - (6) 27.75*** (0.440)
- Country-group indicators (standard errors in parentheses):
  - LIDC:
    - (1) -0.763 (0.603)
    - (2) -7.563*** (0.457)
    - (3) -6.717*** (0.288)
    - (4) -10.87*** (0.559)
    - (5) -9.544*** (0.433)
    - (6) -10.99*** (0.493)
  - EME:
    - (1) -3.834*** (0.611)
    - (2) -11.52*** (0.469)
    - (3) -10.52*** (0.395)
    - (4) -16.23*** (0.560)
    - (5) -14.14*** (0.438)
    - (6) -15.75*** (0.522)

- Reported tax-share coefficients (each column corresponds to a specification where that tax share is the focus; standard errors in parentheses):
  - PIT_share:
    - Coefficient 0.288*** (0.0178)
    - Interaction: LIDCs#c.PIT_share -0.0629** (0.0253)
    - Interaction: EMEs#c.PIT_share -0.191*** (0.0300)
  - VAT_share:
    - Coefficient 0.0675*** (0.0155)
    - Interaction: LIDCs#c.VAT_share 0.00153 (0.0172)
    - Interaction: EMEs#c.VAT_share -0.00744 (0.0177)
  - TradeTaxes_share:
    - Coefficient -0.308*** (0.0308)
    - Interaction: LIDCs#c.TradeTaxes_share 0.208*** (0.0317)
    - Interaction: EMEs#c.TradeTaxes_share 0.198*** (0.0331)
  - Excises_share:
    - Coefficient -0.277*** (0.0388)
    - Interaction: LIDCs#c.Excises_share 0.251*** (0.0421)
    - Interaction: EMEs#c Excises_share 0.399*** (0.0463)
  - Propr_share:
    - Coefficient -0.227*** (0.0393)
    - Interaction: LIDCs#c.Propr_share 0.167** (0.0687)
    - Interaction: EMEs#c.Propr_share 0.612*** (0.125)
  - CIT_share:
    - Coefficient -0.254*** (0.0294)
    - Interaction: LIDCs#c.CIT_share 0.249*** (0.0316)
    - Interaction: EMEs#c.CIT_share 0.290*** (0.0325)

- Dependent variable across specifications: "Tax Revenue as a % of GDP" (reported as column heading).

### Interpretation points evident from coefficients
- A positive baseline association between PIT_share and tax revenue as a % of GDP: PIT_share coefficient 0.288***.
- Interaction terms indicate differential associations for LIDCs and EMEs:
  - PIT_share association weaker in LIDCs (interaction -0.0629**) and substantially weaker in EMEs (interaction -0.191***).
- VAT_share has a positive baseline association (0.0675***) with no significant interactions for LIDCs or EMEs.
- TradeTaxes_share and Excises_share show negative baseline associations, with positive and significant interactions for LIDCs and EMEs (indicating less negative or positive net association in those groups).
- Property (Propr_share) and CIT_share show negative baseline associations with positive and significant interactions for LIDCs and EMEs (implying different net effects by country group).

### Appendix 3. Stylized Facts of the PIT’s Treatment of Capital Income

### Capital income definition and context
- Capital income arises from investment in the form of interest, dividends, capital gains, and others such as rents, annuities, royalties.
- Recent attention to taxation of capital income is motivated by evidence documenting:
  - Decline in the share of labor income in advanced and emerging economies (IMF 2017; Bengtsson and Waldenström 2018; Karabarbounis and Neiman 2014).
  - Reduction in progressivity of tax systems (Gerber et al. 2018).

### Interest
- Nominal PIT rates applied to interest are, on average, higher in more developed countries.
- Where income is taxed on a schedular basis, interest rates are lower than the top marginal PIT rates on employment income.
- Average rates levied on returns to savings (interest income):
  - AEs: 18 percent (above lowest marginal rates on labor income of 15 percent, below top marginal rate of 39 percent).
  - EMEs: 18 percent (below top marginal rate on labor income of 24 percent).
  - LIDCs: 15 (below top marginal rate on labor income of 29 percent).
- Reference: Figure 12 (reported averages).

### Dividends
- Special dividend PIT rates commonly applied when corporate profits are distributed as dividends.
- Nominal dividend rates rise with level of development.
- Average nominal dividend rates in 2019:
  - AEs averaged 22 percent.
  - EMEs averaged 13 percent.
  - LIDCs averaged 13 percent.

### Capital gains
- Capital gains taxation approaches:
  - Taxed at the regular rate on overall income.
  - Taxed separately at a special rate for capital gains.
- Predominant form: special rates for capital gains.
- Share of countries using a special rate to tax capital gains:
  - AEs: 57 percent.
  - EMEs: 41 percent.
  - LIDCs: 60 percent.
- Share of countries taxing capital gains as part of ordinary income:
  - AEs: 37 percent.
  - EMEs: 31 percent.
  - LIDCs: 26 percent.

### Business income and special regimes
- Business income comprises part labor and part return on investment; self-employed and unincorporated business taxpayers pose enforcement challenges, especially in LIDCs and EMEs.
- Countries implement special tax regimes for small business taxpayers, including presumptive tax regimes that may replace income taxes.
- Prevalence of presumptive/special schemes:
  - LIDCs: almost 90 percent of countries have some scheme to tax on a presumptive basis.
  - EMEs: 59 percent.
  - AEs: 35 percent.
- Note: Optional regimes can create arbitrage opportunities.
- Relative importance of corporate income tax (CIT) in total tax revenues:
  - EMEs: CIT represents 19 percent of total tax revenues.
  - LIDCs: CIT represents 16 percent of total tax revenues.
  - AEs: CIT represents 10 percent of total tax revenues.

*IMF Working Paper: Progress of the Personal Income Tax in Emerging Markets and Developing Countries — Appendix 2 and Appendix 3 excerpts.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022020-print-pdf.pdf_