## 2. Sensitivity to Optimal Income Tax Schedules to Key Model Parameters

## Source details

**Canonical URL:** [2. Sensitivity to Optimal Income Tax Schedules to Key Model Parameters](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020105-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020105-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020105-print-pdf.pdf.json)

---

### Introduction and framing
- Income inequality rose in many countries between 1985 and 2015, with increases in the share of income accruing to top income groups.
- Fiscal redistribution (taxes and transfers) is central to addressing income inequality; differences in fiscal redistribution account for over three-quarters of the difference in average disposable income inequality between advanced and Latin American countries.
- Chapter focus: design choices for income taxes and cash transfers, and trade-offs between targeting precision, incentive (efficiency) costs, and administrative, social, and political costs.

### Targeting approaches: definitions and comparative benefits
- Universal transfers:
  - Available to everyone without eligibility conditions (example: Universal Basic Income).
- Quasi-universal (categorical) transfers:
  - Eligibility defined by observable categories highly correlated with low income (example: households with children or elderly).
- Means-tested transfers:
  - Eligibility based on measured income (or “means”), e.g., General Minimum Income (GMI) programs where benefit = gap between means and needs.
- In practice, safety net systems use combinations of targeting methods; outside of universal transfers, targeting is typically the norm.

### Benefits and inefficiencies of means-tested targeting
- Perfect means-tested targeting with a fixed budget maximizes poverty reduction by:
  - Perfect beneficiary targeting: only the poor receive transfers and all the poor receive transfers.
  - Perfect transfer targeting: entire transfer budget goes to the poor, each receiving exactly their poverty gap.
- Inefficiencies of uniform (non-targeted) transfers:
  - Leakage: non-poor households receive transfers.
  - Overpayment: some poor households receive transfers greater than their poverty gaps.
- These inefficiencies reduce poverty impact for a given budget (illustrated by Figure 1 in source).

### Incentive costs and optimal tax-schedule shape
- Perfect means-tested transfers create strong work-disincentive effects near eligibility thresholds (very high effective marginal income tax rates at those points).
- Mirrlees optimal income-tax framework used: universal lump-sum transfer plus non-linear income tax schedule to manage equity-efficiency trade-offs.
- Under standard assumptions (same labor supply elasticity across individuals, declining social value of income, clustered income distribution), the optimal marginal tax schedule is an incomplete U-shape:
  - Relatively high marginal tax rates for low-income individuals (especially very low income) to allow faster withdrawal of transfers and reduce fiscal cost.
  - Relatively low and flat marginal tax rates for middle-income individuals where aggregate efficiency cost is high.
  - Medium marginal tax rates for high-income individuals reflecting lower aggregate efficiency cost and low social value of extra income.
- Model parameter example (Figure 2 note):
  - compensated earnings supply elasticity = 0.25
  - inequality aversion = 0.5
  - Pareto distribution parameter (PI) = 2.5
  - Under PI = 2.5, the top 5 percent have approximately 16.6 percent share of total income.
  - Optimal grant = 10,718 — or 36 percent of median income of 30,000.
  - As inequality aversion increases to 1.0, the optimal grant increases to 54 percent of median income.

### Negative Income Tax (NIT)
- NIT specification: uniform grant g to all plus proportional tax rate t on earned income y0 so final income y1 = g + (1-t) y0.
- Equivalence: NIT is equivalent to a means-tested scheme with threshold g/t: those with y0 below g/t receive net transfers; those above pay net taxes.
- Restricting withdrawal rate (constant t) implies departure from Mirrlees optimal schedule and a lower efficient amount of redistribution.
- In high-informality contexts, NIT may be simplified (high exemption threshold) and rely more on consumption taxes, reducing optimal redistribution.

### Categorical targeting: empirical illustrations and trade-offs
- Eligibility based on exogenous characteristics can be highly but imperfectly correlated with poverty.
- Empirical (Indian 2011 NSS household) illustrations:
  - 33 percent of households have children aged 0–5 years.
  - 50 percent have children aged 0–10 years.
  - 62 percent have either children aged 0–10 years or elderly 65 years or above.
  - Uniform benefits for children up to 5 years: coverage of bottom quintile ~50 percent, falling to ~15 percent for top quintile.
  - Expanding eligibility to children up to 10 years increases overall household coverage from 33 percent to 50 percent and coverage of the bottom quintile to around 72 percent.
  - Including the elderly increases overall coverage to 62 percent and bottom quintile coverage to nearly 80 percent.
  - If benefit levels are held constant, the total transfer budget for expanded child benefit (0–10yrs) would be 57 percent higher than the (0–5yrs) program, and 87 percent higher when the elderly are included; under a fixed budget, individual transfer levels would need to fall by 36 percent and 46 percent, respectively.

### Proxy-means testing (PMT) and tiered approaches
- PMT assigns continuous scores based on observable household characteristics correlated with welfare.
- PMT trade-offs and performance:
  - PMT can outperform simple categorical targeting in coverage and benefit incidence for a given budget.
  - Expanding PMT coverage to 40 percent of the population can reach ~80–90 percent coverage for the bottom quintile.
  - To ensure near-universal coverage (above 80 percent) of each of the bottom three deciles, program coverage may need to expand to 50 percent of the population.
  - PMT suffers from undercoverage (particularly on the margin) and requires regular updating of scoring systems.
- Tiered universal benefit using PMT scores (universal coverage but differentiated benefit levels) can:
  - Increase share of benefits to the bottom 30 percent to over 40 percent (example with 4:2:1 benefit ratios across tiers).
  - Reduce share accruing to the richest 30 percent to just above 15 percent.
  - Eliminate eligibility undercoverage while retaining benefit differentiation.

### Administrative, social, and political costs of targeting
- Implementation constraints: low administrative capacity, large informal sector, multiple/volatile income sources, poor bookkeeping—making income verification difficult.
- Additional costs of means-testing:
  - Administrative costs of income verification.
  - Private costs to beneficiaries for take-up.
  - Social costs such as beneficiary stigma.
  - Political costs related to building broad support for programs.
- These costs can make precise means-testing infeasible; countries may prefer simpler schemes (NIT, categorical targeting, PMT) despite higher efficiency costs and lower optimal redistribution.

### Strengthening fiscal redistribution — guidance and country-group priorities
- Short-term objective: use feasible tax and transfer instruments to approximate the Mirrlees optimal schedule.
- Medium-term objective: strengthen administrative capacity to implement more sophisticated redistributive instruments.

- Advanced countries — key policy areas:
  - Personal income taxation (PIT):
    - Room exists to increase progressivity; top marginal PIT rates have declined since 1980.
    - Proliferation of exemptions reduces effective progressivity; progressivity could be enhanced without harming growth.
    - More progressive PIT can strengthen insurance amid rising income volatility.
  - Capital income taxation (CIT):
    - Capital income often taxed more lightly than labor income; integration of PIT and CIT needed to reduce income reclassification and tax avoidance.
  - Wealth taxation:
    - Wealth taxes can strengthen progressivity but have been reduced in some OECD countries on revenue/administrative grounds.
    - Property taxes (real estate/land) are underused; Norregaard (2015) estimates additional revenue potential around 2 percent of GDP where property taxes are modest.
  - Consumption taxation:
    - Enhancing C-efficiency could increase revenues by over 1 percentage point of GDP in many advanced economies.
    - Efficient taxation of fossil fuels would increase tax revenues by nearly 3 percentage points of GDP on average in advanced countries and requires revenue recycling to mitigate welfare impacts (e.g., via universal transfers).
  - Means-tested transfers:
    - High marginal tax rates (MTRs) from benefit withdrawal create strong disincentives; reducing these MTRs via in-work benefits and wage subsidies can improve efficiency and equity if financed by higher top MTRs and efficient consumption taxes.
    - Active labor market programs and conditionality (job search, training) can enhance efficiency.

- Emerging and developing countries — key policy areas:
  - Constraints: weaker administrative capacity, large informal sectors, competing development spending needs.
  - Consumption taxation:
    - VAT is a primary short-term revenue source, but many countries have sizeable exemptions; raising C-efficiency requires reducing exemptions, increasing compliance (risk-based auditing), and possibly raising registration thresholds to protect small businesses.
    - Higher excises on luxury goods and negative-externality items (fossil fuels, alcohol, tobacco) can raise revenue and improve welfare.
    - Coady and others (2019) estimate efficient pricing of fossil fuels would generate additional revenues of nearly 4 percentage points of GDP in sub-Saharan Africa and 7 percentage points of GDP in Middle East and North Africa.
    - Revenues should be reallocated to progressive spending and social safety nets to protect vulnerable households.
  - Income and wealth taxes:
    - Gradually expand PIT coverage, starting with high tax thresholds and lowering them as capacity grows.
    - Strengthen CIT to reduce profit shifting and recover declining CIT rates.
    - Remove unnecessary exemptions/deductions and avoid harmful international tax competition; international cooperation may be needed.
    - Wealth taxation and property taxes have revenue potential but require significant administrative capacity; Norregaard (2015) estimates property tax potential at 0.5– (numeric range truncated in source excerpt).

### Social safety nets: coverage, design, and trade-offs
- Key challenge in developing countries: expand coverage (particularly low in low-income economies) to reduce poverty and protect poor and vulnerable households from increases in consumption and other taxes and from economic reforms.
- Expansion approaches:
  - Demographic transfers (e.g., child benefits and social pensions) can be used to quickly increase coverage.
  - PMT transfers and public works programs based on self-selection via a work requirement and relatively low wages can complement demographic transfers to better reach lower-income groups.
  - Consider expanding PMT coverage combined with decreasing transfer levels as administrative capacity strengthens.
- Conditional cash transfers (CCTs):
  - Attractive due to complementarity with in-kind transfers (education and health) but require access to quality service delivery and investments in administrative capacity for monitoring.
- Distributional interaction with tax systems:
  - Relatively large leakage of benefits to higher-income groups increases the importance of strengthening progressive income tax systems to claw back these benefits from high-income groups.

### Expanding tax capacities and empirical revenue milestones
- Fiscal thresholds and evidence:
  - Recent research finds countries move to a higher growth path once tax revenue reaches about 15 percent of GDP (Gaspar, Jaramillo, and Wingender 2016).
  - About half of low-income countries—and a third of emerging market economies—have tax ratios below this 15 percent threshold.
- Historical country revenue increases (reported increases):
  - Georgia increased tax revenue by 12.9 percent of GDP during 2004–08.
  - Maldives raised revenue by 11 percent of GDP during 2011–15.
  - Dominica: 7.5 percent (2002–06).
  - Ghana: 7.3 percent (2000–04).
  - Mauritania: 6.1 percent (2010–14).
  - Mozambique: 6.1 percent (2007–11).
  - Guinea: 5.8 percent (2008–12).
  - Malawi: 5.7 percent (2003–07).
  - Cambodia: 5.0 percent (2012–16).
- Revenue loss from profit shifting:
  - Revenue losses from profit shifting have been estimated at nearly 2 percentage points of GDP on average in developing countries.
- Policy implication:
  - Expanding tax capacities is crucial for enhancing fiscal redistribution while avoiding crowding out of public spending critical for inclusive growth and poverty reduction.
  - Financing universal or quasi-universal transfers places a premium on ambitious progressive and efficient taxation reforms, including progressive income and consumption taxation, and efficient taxation of fossil fuel consumption.
- Additional note in source header: "1.0 percent of GDP over 5–10 years with sufficient investment in information and" (text fragment present in source).

### Concluding remarks: optimal taxation, targeting costs, and policy choices
- Optimal income taxation theory:
  - Perfect targeting (all poor receive transfers equal to their poverty gap) maximizes poverty reduction for a given transfer budget, but costs to targeting (incentive, administrative, social, political) make perfect targeting often suboptimal or infeasible.
  - Efficiency requires benefits be withdrawn at a rate below 100 percent to preserve labor supply incentives.
  - Mirrlees framework involves a universal income transfer but assumes capacity for a sophisticated non-linear income tax system; in practice this often transforms the universal transfer into a heavily means-tested transfer.
- Trade-offs and feasibility:
  - Administrative constraints, especially in economies with large informal sectors, imply means-tested transfers may not be feasible.
  - Universal transfers involve substantial leakage to higher-income groups under budget constraints, reducing poverty impact unless financed by progressive taxation or higher budgets that may have efficiency costs.
  - Quasi-universal (categorical) transfers introduce trade-offs between coverage, poverty impact, fiscal cost, and horizontal equity.
  - Proxy-means testing can enhance poverty impact but retains many challenges; using scores to differentiate transfer levels rather than strict inclusion/exclusion can reduce challenges.
- Country specificity and political economy:
  - Appropriate choice among universal and targeted transfers depends on social and political preferences and administrative and fiscal constraints.
  - Greater reliance on universal transfers raises the importance of progressive income taxation and efficient consumption taxes to claw back leakage.
  - Even advanced economies with capacity for sophisticated means-testing still use quasi-universal benefits, indicating social and political factors shape policy design.

### Annex summaries
- Annex 1 — Targeting approaches and costs:
  - Common targeting approaches: demographic, geographic, proxy-means, self-selection.
  - Targeting can vary benefit levels across eligible beneficiaries to reduce leakage.
  - Costs of targeting: incentive, administrative, private, social, political.
  - Relative importance of costs by method: administrative costs higher with individual assessment; incentive costs lower with categorical targeting; private costs higher with self-selection; social costs vary.
- Annex 2 — Sensitivity of optimal income tax schedules to key model parameters:
  - Sensitivity dimensions include inequality aversion, β; compensated earnings-supply elasticity, ε; Pareto parameter, α.
  - Graphical results illustrate impacts on marginal and average tax rates and implied distributional outcomes for varied parameter values.

*Italicized source: wpiea2020105-print-pdf - 1.0 percent of GDP over 5–10 years with sufficient investment in information and administrative systems.*

### References  _______________________________________________________________________________________  24

### wpiea2020105-print-pdf - References

### Figures
- 1. Benefits of Perfect Means-tested Targeting _____________________________________________________7
- 2. Optimal Marginal and Average Income Tax Schedules __________________________________________9
- 3. Negative Income Tax Scheme _________________________________________________________________ 11
- 4. Coverage Under Alternative Categorical and Proxy-means Targeting _________________________  13
- 5. Benefit Share and Program Coverage  _________________________________________________________ 15

### Annexes
- 1. Targeting Approaches and Costs ______________________________________________________________ 21

*https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020105-print-pdf.pdf*

### 2. Sensitivity to Optimal Income Tax Schedules to Key Model Parameters _______________________ 23

### 2. Sensitivity to Optimal Income Tax Schedules to Key Model Parameters

### Introduction
- Income inequality rose in many countries between 1985 and 2015, with increases in the share of income accruing to top income groups.
- Fiscal redistribution (taxes and transfers) is central to addressing income inequality; differences in fiscal redistribution account for over three-quarters of the difference in average disposable income inequality between advanced and Latin American countries.
- The chapter focuses on design choices for income taxes and cash transfers, and the trade-offs between targeting precision, incentive (efficiency) costs, and administrative, social, and political costs.

### Targeted and Universal Transfers
- Definitions:
  - Universal transfers: available to everyone without eligibility conditions (e.g., a Universal Basic Income).
  - Quasi-universal (categorical) transfers: eligibility defined by observable categories highly correlated with low income (e.g., households with children or elderly).
  - Means-tested transfers: eligibility based on measured income (or “means”), e.g., General Minimum Income (GMI) programs where benefit = gap between means and needs.
- In practice, safety net systems use combinations of targeting methods; outside of universal transfers, targeting is typically the norm.

### Benefits of Means-tested Targeting
- With a fixed budget, perfect means-tested targeting maximizes poverty reduction by:
  - Perfect beneficiary targeting: only the poor receive transfers and all the poor receive transfers.
  - Perfect transfer targeting: the entire transfer budget goes to the poor, each receiving exactly their poverty gap.
- Uniform (non-targeted) transfers exhibit two inefficiencies:
  - Leakage: non-poor households receive transfers.
  - Overpayment: some poor households receive transfers greater than their poverty gaps.
- These inefficiencies reduce poverty impact for a given budget by the area represented in the Figure 1 construction.

### Incentive Costs
- Perfect means-tested transfers create strong work-disincentive effects near eligibility thresholds (effectively very high marginal income tax rates at those points).
- The Mirrlees optimal income-tax framework considers a universal lump-sum transfer plus a non-linear income tax schedule to manage equity-efficiency trade-offs.
- Under standard assumptions (same labor supply elasticity across individuals, social value of income declining with income, clustered income distribution), the optimal marginal tax schedule is an incomplete U-shape:
  - Relatively high marginal tax rates for low-income individuals (especially low-income) to allow faster withdrawal of transfers and reduce fiscal cost.
  - Relatively low and flat marginal tax rates for middle-income individuals due to high aggregate efficiency cost where most taxpayers are located.
  - Medium marginal tax rates for high-income individuals reflecting lower aggregate efficiency cost and low social value of extra income.
- Model parameter example (Figure 2 note):
  - compensated earnings supply elasticity = 0.25
  - inequality aversion = 0.5
  - Pareto distribution parameter (PI) = 2.5
  - Under PI = 2.5, the top 5 percent have approximately 16.6 percent share of total income.
  - Optimal grant = 10,718 — or 36 percent of median income of 30,000.
  - As inequality aversion increases to 1.0, the optimal grant increases to 54 percent of median income.
- Practical implementation: ability to implement non-linear tax schedules often transforms policy into sophisticated means-tested transfers (e.g., GMI with high withdrawal rates), with advanced countries also using quasi-universal categorical transfers.

### Administrative, Social, and Political Costs
- Implementation constraints: low administrative capacity, large informal sector, multiple/volatile income sources, poor bookkeeping—making income verification difficult.
- Additional costs of means-testing include:
  - Administrative costs of income verification.
  - Private costs to beneficiaries for take-up.
  - Social costs such as beneficiary stigma.
  - Political costs related to building broad support for programs.
- These costs can make precise means-testing infeasible; countries may prefer simpler schemes (NIT, categorical targeting, PMT) despite higher efficiency costs and lower optimal redistribution.

### Negative Income Tax (NIT)
- NIT: a uniform grant g to all plus a proportional tax rate t on earned income y0 so final income y1 = g + (1-t) y0.
- Equivalent to a means-tested scheme with threshold g/t: those with y0 below g/t receive net transfers; those above pay net taxes.
- Restricting withdrawal rate (constant t) implies higher departure from the Mirrlees optimal schedule and thus a lower efficient amount of redistribution.
- In high-informality contexts, NIT may be further simplified (high exemption threshold) and rely more on consumption taxes, reducing optimal redistribution.

### Categorical Targeting
- Eligibility based on exogenous characteristics (children, elderly, location, disability) can be highly but imperfectly correlated with poverty.
- Trade-offs:
  - Better poverty targeting for included categories but exclusion of poor without the category.
  - Coverage vs. leakage vs. fiscal cost trade-offs.
- Empirical illustrations (Indian 2011 NSS based):
  - 33 percent of households have children aged 0–5 years.
  - 50 percent have children aged 0–10 years.
  - 62 percent have either children aged 0–10 years or elderly 65 years or above.
  - Uniform benefits for children up to 5 years: coverage of bottom quintile ~50 percent, falling to ~15 percent for the top quintile.
  - Expanding eligibility to children up to 10 years increases overall household coverage from 33 percent to 50 percent and coverage of the bottom quintile to around 72 percent.
  - Including the elderly increases overall coverage to 62 percent and bottom quintile coverage to nearly 80 percent.
  - If benefit levels are held constant, the total transfer budget for expanded child benefit (0–10yrs) would be 57 percent higher than the (0–5yrs) program, and 87 percent higher when the elderly are included; under a fixed budget, individual transfer levels would need to fall by 36 percent and 46 percent, respectively.

### Proxy-means Testing (PMT)
- PMT assigns continuous scores based on observable household characteristics correlated with welfare.
- PMT trade-offs:
  - PMT can outperform simple categorical targeting in coverage and benefit incidence for a given budget.
  - PMT tends to have higher coverage of lower-income groups; e.g., expanding PMT coverage to 40 percent of the population can reach ~80–90 percent coverage for the bottom quintile.
  - To ensure near-universal coverage (above 80 percent) of each of the bottom three deciles, program coverage may need to expand to 50 percent of the population.
  - PMT suffers from undercoverage (particularly on the margin) and requires regular updating of scoring systems.
- Tiered universal benefit using PMT scores (universal coverage but differentiated benefit levels) can:
  - Increase the share of benefits to the bottom 30 percent to over 40 percent (example with 4:2:1 benefit ratios across tiers).
  - Reduce the share accruing to the richest 30 percent to just above 15 percent.
  - Eliminate eligibility undercoverage while retaining benefit differentiation.

### Strengthening Fiscal Redistribution — General Guidance
- Short-term objective: use feasible tax and transfer instruments to approximate the Mirrlees optimal schedule.
- Medium-term objective: strengthen administrative capacity to implement more sophisticated redistributive instruments.

### Advanced Countries — Key Policy Areas
- Personal income taxation (PIT):
  - Room exists to increase progressivity; top marginal PIT rates have declined since 1980.
  - Proliferation of exemptions reduces effective progressivity; empirical studies suggest progressivity could be enhanced without harming growth.
  - More progressive PIT can strengthen insurance role amid rising income volatility.
- Capital income taxation (CIT):
  - Capital income is often taxed more lightly than labor income; integration of PIT and CIT needed to reduce income reclassification and tax avoidance.
- Wealth taxation:
  - Wealth taxes can strengthen progressivity but have been reduced in some OECD countries on revenue/administrative grounds.
  - Property taxes (real estate/land) are underused; Norregaard (2015) estimates additional revenue potential around 2 percent of GDP where property taxes are modest.
- Consumption taxation:
  - Enhancing C-efficiency could increase revenues by over 1 percentage point of GDP in many advanced economies.
  - Efficient taxation of fossil fuels would increase tax revenues by nearly 3 percentage points of GDP on average in advanced countries and requires revenue recycling to mitigate welfare impacts (e.g., via universal transfers).
- Means-tested transfers:
  - High marginal tax rates (MTRs) from benefit withdrawal create strong disincentives; reducing these MTRs via in-work benefits and wage subsidies can improve efficiency and equity if financed by higher top MTRs and efficient consumption taxes.
  - Active labor market programs and conditionality (job search, training) can enhance efficiency.

### Emerging and Developing Countries — Key Policy Areas
- Constraints: weaker administrative capacity, large informal sectors, competing development spending needs on physical and human capital.
- Consumption taxation:
  - VAT is a primary short-term revenue source, but many countries have sizeable exemptions; raising C-efficiency requires reducing exemptions, increasing compliance (risk-based auditing), and possibly raising registration thresholds to protect small businesses.
  - Higher excises on luxury goods and negative-externality items (fossil fuels, alcohol, tobacco) can raise revenue and improve welfare.
  - Coady and others (2019) estimate efficient pricing of fossil fuels would generate additional revenues of nearly 4 percentage points of GDP in sub-Saharan Africa and 7 percentage points of GDP in Middle East and North Africa.
  - Revenues should be reallocated to progressive spending and social safety nets to protect vulnerable households.
- Income and wealth taxes:
  - Gradually expand PIT coverage, starting with high tax thresholds and lowering them as capacity grows.
  - Strengthen CIT to reduce profit shifting and recover declining CIT rates.
  - Remove unnecessary exemptions/deductions and avoid harmful international tax competition; international cooperation may be needed.
  - Wealth taxation and property taxes have revenue potential but require significant administrative capacity; Norregaard (2015) estimates property tax potential at 0.5– (note: full numeric range cut off in source excerpt).

*Source: IMF working paper — chapter "2. Sensitivity to Optimal Income Tax Schedules to Key Model Parameters."*

### 1.0 percent of GDP over 5–10 years with sufficient investment in information and

### wpiea2020105-print-pdf - 1.0 percent of GDP over 5–10 years with sufficient investment in information and

### Social safety nets: coverage, design, and trade-offs
- Key challenge in developing countries: expand coverage (particularly low in low-income economies) to reduce poverty and protect poor and vulnerable households from increases in consumption and other taxes and from economic reforms more generally.
- Expansion approaches:
  - Demographic transfers (e.g., child benefits and social pensions) can be used to quickly increase coverage.
  - Proxy-means targeted (PMT) transfers and public works programs based on self-selection via a work requirement and relatively low wages can complement demographic transfers to better reach lower-income groups.
  - Consideration could be given to expanding coverage in PMT programs combined with decreasing transfer levels as administrative capacity strengthens.
- Targeting complementarities and requirements:
  - Conditional cash transfers (CCTs) that condition eligibility on school attendance, health and nutrition clinic use are attractive due to complementarity with in-kind transfers (education and health), but must be complemented by access to quality service delivery.
  - These programs can require significant investments in administrative capacity to coordinate line ministries and monitor beneficiary compliance.
- Distributional interaction with tax systems:
  - Relatively large leakage of benefits to higher-income groups increases the importance of strengthening progressive income tax systems to claw back these benefits from high-income groups.

### Expanding tax capacities and fiscal redistribution
- Fiscal thresholds and evidence:
  - Recent research finds countries move to a higher growth path once tax revenue reaches about 15 percent of GDP (Gaspar, Jaramillo, and Wingender 2016).
  - About half of low-income countries—and a third of emerging market economies—have tax ratios below this 15 percent threshold.
- Historical country experiences increasing tax revenue (exact increases reported):
  - Georgia increased tax revenue by 12.9 percent of GDP during 2004–08.
  - Maldives raised revenue by 11 percent of GDP during 2011–15.
  - Dominica: 7.5 percent (2002–06).
  - Ghana: 7.3 percent (2000–04).
  - Mauritania: 6.1 percent (2010–14).
  - Mozambique: 6.1 percent (2007–11).
  - Guinea: 5.8 percent (2008–12).
  - Malawi: 5.7 percent (2003–07).
  - Cambodia: 5.0 percent (2012–16).
- Revenue loss from profit shifting:
  - Revenue losses from profit shifting have been estimated at nearly 2 percentage points of GDP on average in developing countries (Cobham and Janský, 2018; Crivelli, de Mooij, and Keen, 2016).
- Policy implications:
  - Expanding tax capacities is crucial for enhancing fiscal redistribution while avoiding crowding out of public spending crucial for inclusive growth and poverty reduction.
  - Financing universal or quasi-universal transfers places a premium on ambitious progressive and efficient taxation reforms, including progressive income and consumption taxation, and efficient taxation of fossil fuel consumption.

### Concluding remarks: optimal taxation, targeting costs, and policy choices
- Optimal income taxation theory:
  - Perfect targeting (all poor receive transfers equal to their poverty gap) maximizes poverty reduction for a given transfer budget, but costs to targeting (incentive, administrative, social, political) make perfect targeting often suboptimal or infeasible.
  - Efficiency requires that benefits are withdrawn at a rate below 100 percent to preserve labor supply incentives.
  - The optimal income taxation framework involves a universal income transfer but assumes governments can implement a sophisticated non-linear income tax system; in practice this effectively transforms the universal transfer into a heavily means-tested transfer.
- Trade-offs and feasibility:
  - Administrative constraints, especially in economies with large informal sectors, imply means-tested transfers may not be feasible.
  - Universal transfers involve substantial leakage to higher-income groups under budget constraints, reducing poverty impact unless financed by progressive taxation or higher budgets that may have efficiency costs.
  - Quasi-universal (categorical) transfers (age, gender, location) introduce trade-offs between coverage, poverty impact, fiscal cost, and horizontal equity; can generate social and political costs.
  - Proxy-means testing can enhance poverty impact but retains many challenges of quasi-universal targeting; using scores to differentiate transfer levels rather than strict inclusion/exclusion can reduce challenges.
- Country specificity and political economy:
  - The appropriate choice among universal and targeted transfers is country specific, depending on social and political preferences and administrative and fiscal constraints.
  - Greater reliance on universal transfers raises the importance of progressive income taxation and efficient consumption taxes to claw back leakage.
  - Even advanced economies with capacity for sophisticated means-testing still use quasi-universal benefits (e.g., child benefits, social pensions), indicating social and political factors shape policy design.

### Annex 1 — Targeting approaches and costs (summary)
- Common targeting approaches:
  - Demographic targeting: eligibility based on household characteristics (presence of children, elderly, disabled) correlated with low income.
  - Geographic targeting: eligibility based on living in poor communities using national poverty maps.
  - Proxy-means targeting: statistical scoring of socio-economic characteristics to predict income/consumption/poverty and determine eligibility.
  - Self-selection targeting: eligibility based on willingness to participate in work programs at relatively low wages to induce high-wage individuals to opt out.
- Targeting can also vary benefit levels across eligible beneficiaries to reduce leakage.
- Costs of targeting:
  - Incentive costs: households may change behavior (e.g., reduce labor supply) or hide income to become eligible.
  - Administrative costs: collecting and verifying household income data is costly and verification is imperfect, especially in informal settings.
  - Private costs: opportunity costs (e.g., workfare participation), queuing, certification fees, transportation.
  - Social costs: stigma from public identification as poor; targeting errors (excluding some poor, including some non-poor) can create community tensions.
  - Political costs: excluding middle classes may reduce broad-based support; efficient targeting may increase political support via perceived indirect benefits.
- Relative importance of costs by method:
  - Administrative costs more important with individual/household assessment.
  - Incentive costs less important with categorical targeting (but shifted to tax side).
  - Private costs more important with self-selection.
  - Social costs vary with program design.

### Annex 2 — Sensitivity of optimal income tax schedules to key model parameters (noted themes)
- Sensitivity dimensions include:
  - Inequality aversion, β.
  - Compensated earnings-supply elasticity, ε.
  - Pareto parameter, α.
- Graphical results illustrate impacts on marginal and average tax rates and implied distributional outcomes for varied parameter values (figures present distributions and tax schedules for multiple parameter settings and median income scenarios).

*Italicized source: wpiea2020105-print-pdf - 1.0 percent of GDP over 5–10 years with sufficient investment in information and administrative systems.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020105-print-pdf.pdf_