## wpiea2020037-print-pdf

## Source details

**Canonical URL:** [wpiea2020037-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020037-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020037-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020037-print-pdf.pdf.json)

---

### I. Introduction — key findings and fiscal context
- Italy’s tax system: considerable complexity, multiple deductions and credits, substitute taxes and special regimes, extensive use of reduced rates and exemptions.
- Largest identified PIT tax expenditures: estimated revenue loss of about 5 percent of GDP (authors’ EUROMOD estimate for 2018: total €89.1 billion, equal to 5.1 percent of GDP).
- VAT policy gap: about 22 percent of the total tax liability (CASE, 2018) versus 16.5 percent EU average → revenue loss estimated at about 3½ percent of GDP.
- VAT compliance gap: estimated about 26 percent of the total tax liability, compared with about 12.3 percent EU average (CASE, 2018).
- Personal income tax gap for the self-employed: estimated about 67 percent.
- Revenue loss from tax evasion: estimated about €109 billion or close to 6 percent of GDP (MEF, 2019a).
- Average tax wedge on labor income in Italy: 47.9 percent; EU-15 average: 41.8 percent; OECD average: 35.9 percent.
- Italy’s corporate income tax rate: 24 percent; standard VAT rate: 22 percent.
- Outdated cadastral valuation for property taxes erodes the tax base and imposes a disproportionately high tax burden on the poor.
- Paper objective: assess comprehensive reform options to reduce the tax burden on labor toward the EU average; simulate revenue and distributional implications for PIT, VAT, and recurrent property taxes (together account for 65 percent of tax revenue; 19 percent of GDP).
- Simulated PIT reforms target a reduction in the tax wedge on labor income by about 4.5 percent—equivalent to 2 percent of GDP in budget revenue.
- Reform principles: reduce complexity, broaden tax base, lower marginal tax rates; preserve progressivity unless explicitly testing a flatter alternative.

### II. Data, models, and simulation methodology
- Microsimulation model: EUROMOD (version H1.0+ with software version V3.1.4), reproducing the 2018 Italian tax-benefit system; policy rules at 2018 setting.
- Underlying microdata: EU-SILC (using EU-SILC data of 2016). EU-SILC lacks disaggregated consumption information; VAT simulations require imputing consumption from Italian Household Budget Survey (HBS) 2017.
- Matching and imputation approach (following Sanchez and others, 2016):
  - Estimate consumption function using SHIW 2016 (log-consumption on log-disposable income and covariates); impute aggregated household consumption into EU-SILC.
  - Aggregate HBS 2017 consumption on 276 goods/services into 12 COICOP categories; compute consumption shares at household level.
  - Propensity score matching between HBS and EU-SILC on aggregated consumption and covariates; impute disaggregated consumption shares to matched EU-SILC households.
  - Multiply matched consumption shares by total imputed consumption in EU-SILC to obtain household consumption levels for 12 categories.
- Weighted average VAT rates by COICOP computed by applying current Italian VAT rate structure to the 276 HBS goods/services.
- EUROMOD is static: simulations show first-round budgetary and distributive effects only; do not capture behavioral effects on labor supply or residential investment.
- VAT consumer behavior modeled via two simplified approaches:
  - Fixed quantities: consumed quantities unchanged; consumption expenditure and savings adjust with price changes.
  - Fixed expenditure shares: total consumption expenditure constant; expenditure on each good adjusts to keep each good’s expenditure share constant.
- Matching quality: mean bias reduced from 7.5 before matching to 2.4 after matching; post-matching tests indicate group means do not differ for matched variables.

### IV. Reform simulation objectives, scenarios, and estimated effects
- Reform objective: shift taxation away from labor income toward consumption and property while keeping revenue-neutrality relative to a reduction in the tax wedge on labor toward EU-15 average.
- Reform principles: reduce complexity, broaden tax base, lower marginal tax rates.
- Overall calibration: target reduction in tax wedge on labor income by 4.5 percentage points, equivalent to 2 percent of GDP in budget revenue.
- Revenue-neutral design: additional receipts from VAT and property tax reforms used to finance reductions in labor income taxation.
- Simulations quantify changes in pre- and post-reform tax liability and changes in household disposable income across the income distribution.

A. Personal Income Tax (PIT) reform simulations
- Two PIT reform options evaluated; both target average reduction in the tax wedge on labor of 4.5 percent.
- Net cost of reform proposals to the budget: about €36 billion (2 percent of GDP).
- Reform 1 (low- and middle-income focus):
  - Simplify PIT to four tax brackets.
  - Reduce tax rate on first taxable income bracket to 9 percent (from 23 percent).
  - Preserve second bracket (income between 15,000–28,000) at 27 percent.
  - Increase tax rate on third bracket (income between 28,000–55,000) to 41 percent (from 38 percent).
  - Merge last two brackets into one for income above 55,000 at 44 percent.
  - Eliminate the National Income Tax Bonus (€80 bonus) and the tax credit for building refurbishment and construction.
  - Distributional outcomes (summary):
    - All households pay lower taxes; middle-income groups get highest reduction in PIT liability.
    - Households with incomes below 20,000 virtually pay no income tax due to lower rate and interactions with credits/deductions.
    - In absolute terms, wealthier families benefit more because they pay a lower tax rate on the first part of their income.
    - Annual disposable income increases:
      - about €820 for the lowest five deciles
      - about €1,600 for the highest three deciles
- Reform 2 (flatter two-bracket alternative):
  - Structure: taxable income below 30,000 taxed at 15 percent; taxable income above 30,000 taxed at 44 percent.
  - Cost offset: elimination of the National Income Tax Bonus and tax credit for building refurbishment and construction.
  - Distributional outcomes:
    - In absolute terms wealthier families benefit significantly more.
    - Average annual disposable income increases:
      - by about €530 for households in the first five deciles
      - by about €2,700 for the highest three deciles
    - Relative to disposable income:
      - middle-income households increase by about 3.5 percent
      - highest three deciles increase by about 5 percent
  - Note: even flatter PIT reform (e.g., 15 and 20 percent rates) would be exponentially costlier and highly regressive (IMF, 2019).
- Fiscal cost and offsets:
  - Proposed changes to brackets and rates estimated cost about €50 billion.
  - Elimination of National Income Tax Bonus saves €8.0 billion.
  - Elimination of tax credit on refurbishment and construction saves €5.7 billion.

B. VAT reform simulations — revenue-neutral and revenue-increasing options
- Policy aim: generate additional revenue to compensate half- or full cost of income tax reform (1–2 percent of GDP) while minimizing impact on poorer households.
- Revenue-neutral VAT alternatives:
  - Option A: eliminate 4 percent reduced rate and reduce standard rate to 17.5 percent (calibrated revenue neutral).
  - Option B: eliminate 10 percent reduced rate and reduce standard rate to 18.5 percent (calibrated revenue neutral).
- Distributional outcomes for revenue-neutral options:
  - Eliminating 10 percent reduced rate (lowering standard to 18.5 percent): largely neutral distributionally; disposable income mostly unchanged across income levels.
  - Eliminating 4 percent reduced rate (lowering standard to 17.5 percent): regressive; redistributes income from poorer to wealthier households by an estimated 0.4 percent of disposable income; reduces disposable income of two lowest deciles by about €50 and increases disposable income of highest decile by about €250.
- Revenue-increasing VAT reforms:
  - Eliminate 10 percent reduced rate and set standard rate to 21 percent → additional revenue about 1 percent of GDP.
  - Eliminate 10 percent reduced rate and set standard rate to 24 percent → additional revenue about 2 percent of GDP.
- Distributional impacts of revenue-increasing VAT:
  - Reforms are regressive; negative impact increases with revenue target.
  - 1 percent of GDP VAT reform reduces disposable income of first two deciles by about 1.5 percent; for highest two deciles reduction slightly below 1 percent.
  - VAT liability translates to close to 14 percent of household disposable income among poorer households (for 1 percent reform) and about 9 percent among wealthier households.
  - Absolute additional VAT payments:
    - poorer households: about €150 (1 percent of GDP reform) and €300 (2 percent reform)
    - wealthier households: about €700 (1 percent reform) and €1,500 (2 percent reform)
- Simulation calibration note: revenue implications differ by 0.1–0.3 percent of GDP depending on behavioral assumption; simulations calibrated at mid-point between fixed quantities and fixed expenditure shares.

C. Property tax reform simulations — market valuation and base broadening
- Valuation methodology:
  - Current taxable values in EU-SILC multiplied by ratio of market to taxable value by geographical region.
  - Updated taxable value formula: Taxable Value_i = Cadastral Value_i × 1.05 × 1.60 × Ratio_i, where i are the 20 geographical regions; multiplicative factors 1.05 and 1.6 applied as in current system; Ratio_i reflects remaining gap between market and taxable values by region.
- Revenue-neutral property tax reform (market valuation, broadening to include all immovable property including main residences):
  - Current property tax revenue on immovable property (about 1 percent of GDP in 2018) can be obtained with a tax rate of 0.3 percent under market valuation.
  - Comparison: current maximum allowed tax rate of 1.06 percent for other residences.
- Distributional effects of a revenue-neutral market-valued property tax:
  - Increases disposable income of poorest households by 3 percent.
  - Decreases disposable income of wealthier households by 0.5 percent.
  - Property tax liability (first decile): before reform 4.8 percent of disposable income → after reform about 2 percent → annual savings about €275.
  - Property tax liability (two highest deciles): before reform about 2 percent → after reform 2.6 percent → higher annual property taxes about €460.
  - Equity/liquidity mitigation: suggested deferral scheme where tax liability payable upon property transfer (sale or inheritance).
- Base-broadening property tax variants (market valuation):
  - Tax rate 0.55 percent on all properties → additional revenue about 1 percent of GDP.
  - Tax rate 0.8 percent on all properties → additional revenue about 2 percent of GDP.
- Distributional effects for 1 percent of GDP property reform (0.55 percent rate):
  - Disposable income of first decile increases by 0.8 percent.
  - Disposable income decreases from 1 percent in second decile to about 3 percent in highest decile.
  - Resulting property tax liability increases with disposable income from 4 percent to above 5 percent.
- Distributional effects for 2 percent of GDP property reform (0.8 percent rate): impacts roughly twice those of 1 percent reform.
  - Absolute additional property tax:
    - Poorer households (first two deciles): about €30 (1 percent reform) and €120 (2 percent reform).
    - Wealthier households (highest two deciles): about €1,700 (1 percent reform) and €2,900 (2 percent reform).

D. Distributive effects of revenue-neutral shifts from PIT to VAT and/or property tax (offsetting a PIT reform lowering labor tax burden by 2 percent of GDP)
- Financing options:
  1. Fully by VAT (2 percent of GDP).
  2. Split equally VAT and property tax (1 percent of GDP each).
  3. Fully by property tax (2 percent of GDP).
- PIT → VAT (fully VAT-financed, 2 percent of GDP):
  - Disposable income of middle-income households (third to seventh deciles) increases by 2.5 percent on average → average annual tax relief €620 per household.
  - Disposable income of highest three deciles increases by 0.8 percent on average → €250.
  - Disposable income of low-income households (first two deciles) decreases by 1 percent on average → €100.
- PIT → VAT and property tax (1 percent VAT + 1 percent property):
  - Disposable income of middle-income households increases by 2.0 percent on average → €500.
  - Disposable income of low-income households increases by 0.6 percent on average → €60.
  - Disposable income of high-income households decreases by 0.7 percent on average → €600.
  - Adding property tax progressivity reduces gains for middle-income households and shifts costs to high-income households.
- PIT → property tax (fully property-financed, 2 percent):
  - Disposable income of low-income households increases by 1.4 percent on average.
  - Disposable income of middle-income households increases by 1.7 percent on average.
  - Disposable income of high-income households decreases by 1.5 percent on average.
  - Percentage change in disposable income becomes negative from the 8th decile onwards.

### Conclusions and policy implications
- Broad principle: broader tax base combined with low tax rates desirable to limit deadweight loss; narrow bases plus high marginal rates increase efficiency costs.
- Italian tax-code inefficiencies: tax expenditures have eroded the base, increasing reliance on higher marginal rates (especially on labor income); complexity and inefficiency contributed to a significant compliance gap.
- Summary implications of simulated reforms:
  - PIT reform that lowers tax burden on labor while preserving progressivity benefits low- and middle-income households the most; simplification and streamlining poorly targeted credits/deductions can offset costs.
  - Base-broadening VAT reform should be carefully designed to limit regressive effects; eliminating 10 percent reduced rate while lowering the standard rate can be almost neutral distributionally.
  - Updating property valuation from cadastral to market values is essential to address equity concerns and increase revenue; cadastral valuations erode base and disproportionately burden the poor.
  - A revenue-neutral shift from labor income to consumption and property taxation can be designed to benefit low- and middle-income households, lowering their overall tax liability while raising progressivity.
- Implementation note: tax administration strengthening and holistic strategy required to realize compliance gains; detailed organizational assessment beyond paper’s scope.

### Appendix — matching and imputation methodology (consumption imputation) and key data summaries
- Consumption function estimation (SHIW 2016 OLS highlights):
  - Income (logs): -0.789*** (standard error 0.037)
  - Income Squared (logs): 0.069*** (0.002)
  - R2: 0.599; Observations: 16,333; F-statistic: 997.89
- Imputation comparison (selected):
  - SHIW Income (logs): -0.811*** (0.019)
  - EU-SILC Income (logs): -0.825*** (0.006)
  - R2: 0.587 (SHIW) and 0.981 (EU-SILC)
  - Observations: 16,333 (SHIW) and 21,704 (EU-SILC)
- Propensity score matching diagnostics (Appendix Table 3):
  - Probit regression selected coefficients (examples): Consumption -0.001*** (0.000); Male Household Head -0.105* (0.015); Self Employed 0.073** (0.018); Pseudo R2: 0.036; Observations: 38,650.
  - PS test: U Mean Bias 7.5; M Mean Bias 2.4.
- Consumption shares (Appendix Table 4; means in percent of total consumption expenditures and household consumption monthly euros):
  - Food and non-alcoholic beverages: HBS Mean 27.2; Household Consumption 466.4. EU-SILC Mean 26.1; Household Consumption 452.8.
  - Home fuels, electricity, rents: HBS Mean 18.9; Household Consumption 309.1. EU-SILC Mean 18.2; Household Consumption 314.5.
  - Hospitality and catering: HBS Mean 5.0; Household Consumption 120.8. EU-SILC Mean 5.6; Household Consumption 120.2.
  - Other (aggregate): HBS Mean 8.4; Household Consumption 180.6. EU-SILC Mean 8.8; Household Consumption 171.7.
  - Observations: HBS 16,946; EU-SILC 21,704.
- Key micro-simulation findings (selected):
  - Using household consumption instead of disposable income yields a somewhat more progressive tax system in Italy: baseline PIT liability increases from 1.7 to 42 percent; VAT liability increases from 12.3 to 13 percent; property tax liability decreases from 4.6 to 2.7 percent.
  - Revenue-neutral shift away from income tax toward VAT and property tax: net impact for middle-income households (third to seventh deciles) an increase in household consumption of about 3.6 percent (€700); impact on low- and high-income households almost unchanged.

*Source: wpiea2020037-print-pdf - REFERENCES and accompanying sections from the supplied IMF PDF.*

### REFERENCES_______________________________________________________________________________________________ 35

### wpiea2020037-print-pdf - REFERENCES_______________________________________________________________________________________________ 35

### I. Introduction — key findings and fiscal context
- Italy’s tax system is characterized by considerable complexity, multiple deductions and credits, substitute taxes and special regimes, and extensive use of reduced rates and exemptions.
- Largest identified PIT tax expenditures cause an estimated revenue loss of about 5 percent of GDP.
- The VAT policy gap in Italy is estimated at about 22 percent of the total tax liability (CASE, 2018), versus 16.5 percent for the EU average, resulting in a revenue loss estimated at about 3½ percent of GDP.
- Average tax wedge on labor income in Italy is 47.9 percent, compared with the EU-15 average of 41.8 percent and the OECD average of 35.9 percent.
- Italy’s corporate income tax rate is 24 percent (above the EU average of 21.3 percent); the standard VAT rate is 22 percent.
- Outdated cadastral valuation for property taxes erodes the tax base and imposes a disproportionately high tax burden on the poor.
- VAT compliance gap estimated at about 26 percent of the total tax liability, compared with about 12.3 percent for the EU average (CASE, 2018).
- Personal income tax gap for the self-employed estimated at about 67 percent.
- Revenue loss from tax evasion estimated at about €109 billion or close to 6 percent of GDP (MEF, 2019a).
- The paper assesses comprehensive reform options aimed at reducing the tax burden on labor toward the EU average and simulates revenue and distributional implications for PIT, VAT, and recurrent property taxes, which together account for 65 percent of tax revenue (19 percent of GDP).
- Simulated PIT reforms aim to reduce the tax wedge on labor income by about 4.5 percent—equivalent to 2 percent of GDP in budget revenue.
- Reform principles: reduce complexity, broaden the tax base, lower marginal tax rates; preserve progressivity unless explicitly testing a flatter alternative.

### II. Data, models, and simulation methodology
- Microsimulation for PIT and property taxes: EUROMOD (version H1.0+ with software version V3.1.4), reproducing the 2018 Italian tax-benefit system; policy rules run at their 2018 setting.
- Underlying microdata: EU-SILC (European Union Statistics on Income and Living Conditions), using EU-SILC data of 2016.
- EU-SILC lacks disaggregated consumption information; VAT simulations require imputing consumption from Italian Household Budget Survey (HBS) 2017.
- Matching and imputation approach (following Sanchez and others, 2016):
  - Estimate a consumption function using SHIW (Survey on Household Income and Wealth) 2016 to regress aggregated household consumption (in logs) on disposable income (in logs) and household covariates common to EU-SILC.
  - Impute aggregated household consumption into EU-SILC using estimated parameters and EU-SILC disposable income and covariates.
  - Aggregate HBS 2017 consumption data on 276 goods/services into 12 COICOP categories: 1) food and non-alcoholic beverages; 2) alcoholic beverages and tobacco; 3) clothing and footwear; 4) home fuels, electricity, rents (excluding imputed rents); 5) furniture and household services; 6) health; 7) transport; 8) communication; 9) recreation and culture; 10) education; 11) hospitality and catering; 12) other goods and services.
  - Compute consumption shares in HBS at household level.
  - Use propensity score matching to match HBS and EU-SILC households on aggregated consumption (observed in HBS and predicted in EU-SILC) and household covariates; impute disaggregated consumption shares to matched EU-SILC households.
  - Multiply matched consumption shares by total imputed consumption in EU-SILC to obtain household consumption levels for 12 categories.
- Weighted average VAT rates for each COICOP category computed by applying current Italian VAT rate structure to the 276 HBS goods/services.
- EUROMOD is a static model; simulations show first-round budgetary and distributive effects and do not capture second-round or behavioral effects on labor supply or residential investment.
- VAT simulations attempt to incorporate consumer behavior using two simplified approaches (as in Capeau and others, 2014):
  - Fixed quantities: consumed quantities unchanged; consumption expenditure and savings adjust with price changes.
  - Fixed expenditure shares: total consumption expenditure constant; expenditure on each good adjusts to keep each good’s expenditure share constant.

### III. Overview of the current tax system — design details and major tax provisions
- Personal Income Tax (IRPEF) structure (progressive on five brackets):
  - 23 percent for annual income between zero and €15,000.
  - 27 percent for €15,001–€28,000.
  - 38 percent for €28,001–€55,000.
  - 41 percent for €55,001–€75,000.
  - 43 percent on income exceeding €75,000.
- Additional regional taxes: 1.23 percent to 3.33 percent; local taxes: 0 to 0.9 percent.
- Italy offers about 140 tax deductions and credits on labor income (MEF, 2019b).
- Largest identified PIT tax expenditures (authors’ EUROMOD estimate for 2018): total €89.1 billion, equal to 5.1 percent of GDP. Largest items:
  - Tax credit for wage income from employment, pensions, self-employment, and similar income: €33.7 billion, 1.9 percent of GDP. This regime substitutes for the absence of a lower personal income tax threshold.
  - Tax exemption for contributions to welfare and pension schemes for employees and self-employed: €27.2 billion, 1.5 percent of GDP. Pension contributions are excluded; pension income is taxed—avoids double taxation.
  - Tax credit for dependent relatives: €10.3 billion, 0.6 percent of GDP.
  - National income tax bonus (€80 bonus): €8.0 billion, 0.5 percent of GDP. Granted to employees with annual taxable income above €8,000 and phased down for annual salaries between €24,600–€26,600.
  - Tax credit for building refurbishment and construction: €5.7 billion, 0.3 percent of GDP.
  - Tax credit for medical expenses and health assistance services: €3.0 billion, 0.2 percent of GDP.
  - Tax credit on mortgage interest payments: €1.2 billion, 0.1 percent of GDP.
- Sum of largest PIT tax expenditures: €89.1 billion, 5.1 percent of GDP.
- Notes on definitions and government reporting:
  - Authors use a comprehensive definition of tax expenditures following Heady and Mansour (2019).
  - Government report (MEF, 2018) using a legal benchmark approach estimates tax expenditures at about €35 billion; using the same definition authors find about €30 billion. The 2020 government update estimates the cost at €43 billion (MEF, 2019).

### IV. Reform simulation objectives and scope
- Objective: reduce tax burden on labor toward the EU average through comprehensive reform combining PIT simplification, VAT base-broadening, and property tax reform.
- Simulate PIT reforms that reduce number of taxable brackets, streamline credits and deductions, reduce marginal rates while preserving progressivity (and assess alternative flatter reform).
- Simulate VAT reforms that streamline reduced rates and exemptions and adjust the standard rate.
- Simulate property tax reforms redefining the tax base using market value rather than cadastral value to allow substantial rate reductions for comparable revenue.
- Revenue-neutral reform simulations: additional receipts from VAT and property tax reforms used to finance reductions in labor income taxation.
- Simulations quantify changes in pre- and post-reform tax liability and changes in household disposable income across the income distribution.

*Source: wpiea2020037-print-pdf - REFERENCES_______________________________________________________________________________________________ 35*

### Box 1. PIT Tax Expenditures in the European Union

### Box 1. PIT Tax Expenditures in the European Union

### Overview
- Tax expenditures are government revenues foregone as a result of differential or preferential treatment relative to some benchmark system and include allowances, exemptions, rate relief, credits, and tax deferrals.
- Cross-country comparisons are difficult due to definitional, classification and benchmark differences and limited comparable data.
- OECD (2010) cross-country analysis: budgetary cost of tax expenditures ranges from about ½ percent of GDP in the Netherlands to close to 5 percent of GDP in Italy.
- European Commission (2016) study (housing, pension, education, health): revenue loss from these categories estimated for Italy at about 3½ percent of GDP; average revenue loss among selected EU countries estimated at about 2½ percent of GDP.

### A. Personal Income Tax (PIT) in Italy — Incidence and Major Tax Expenditures
- Average incidence of the PIT (net of tax credits and deductions) for all Italian households: about 20 percent of gross annual disposable income.
- PIT liability by disposable income decile:
  - first decile: about 2 percent of disposable income (about €100)
  - highest decile: about 29 percent (about €26,000)
- Interaction of multiple tax credits and deductions:
  - Tax credits on income source and family tax credits increase average household disposable income by about 10 percent for the first four deciles; their incidence on the three highest deciles falls below 4 percent.
  - National Income Tax Bonus (the €80 bonus) increases disposable income of the second to seventh deciles by about 1.5 percent.
  - Deductions for social security contributions (SSC Deductions) reduce progressivity: contribution to average household disposable income for the three highest income deciles (about 3.6 percent) doubles that of the first three 1.9
  - Other tax credits and deductions (building refurbishment, mortgage interest payments, medical expenses) increase disposable income from about 0.5 percent (first decile) to about 1.5 percent (highest deciles) and tend to benefit wealthier households more.

### B. Value-Added Tax (VAT)
- Italy VAT structure:
  - standard rate: 22 percent
  - reduced rate: 10 percent (selected goods including some food items, restaurants and hotels, pharmaceutical and medical products, energy products and recreational services)
  - reduced rate: 4 percent (food items, educational textbooks and newspapers, school and company canteens)
  - exemptions: actual rentals paid by tenants, dental, hospital and other medical services, education, insurance, passenger transport (by sea, taxi, underground or tram), and postal services
- Effective average VAT rate in Italy: about 10 percent (below EU-15 average of 12 percent); standard rate above EU-15 average of 21.6 percent.
- Estimated revenue losses from reduced rates/exemptions using microsimulation:
  - policy gap for reduced-rate taxation: about €22 billion (1.3 percent of GDP)
  - revenue loss associated with the reduced rate of 4 percent: €27 billion (1.5 percent of GDP)
  - revenue loss for exempted goods: about €18 billion (1 percent of GDP)
- Incidence of VAT by decile:
  - VAT liability decreases from about 12 percent of disposable income in the first two deciles to below 9 percent for the highest deciles.
  - Application of reduced rates and exemptions increases disposable income by about 7 percent in the lowest deciles and about 4 percent in the highest deciles.
  - Reduced rate of 4 percent increases disposable income of poorer families by about 3 percent, and by about 1 percent for the wealthiest.
- Targeting of reduced rates/exemptions:
  - Basic foods (average VAT rate of 6.5 percent): poorer households consume about 30 percent of their income on these goods (Decile 1).
  - Energy products and drinking water (reduced 10 percent) and rental fee exemptions: poorer households consume about 20 percent of income on these goods and services.
  - Reduced rate of 10 percent on restaurants and hotel stays clearly benefits wealthy households the most.

- Representative household consumption shares by decile (selected items — share of household disposable income):
  - Decile 1: Food and non-alcoholic beverages 30.7; Home fuels, electricity, rents 23.3; Hospitality and catering 3.6; Other 7.4
  - Decile 10: Food and non-alcoholic beverages 11.3; Home fuels, electricity, rents 7.5; Hospitality and catering 4.6; Other 6.1
  - VAT rates by category: Food and non-alcoholic beverages 6.5; Alcoholic beverages and tobacco 22.0; Home fuels, electricity, rents 16.5; Hospitality and catering 11.4; Other 20.6

### C. Immovable Property Taxes (IMU and TASI)
- Two taxes on owned real estate:
  - IMU: municipal tax on ownership
  - TASI: tribute for indivisible service (lighting, road maintenance)
- Tax base and exemptions:
  - main residence (owner-occupied) exempt from IMU and TASI unless luxury home (stately homes, villas, castles and buildings with high artistic and historic value)
  - second homes and commercial property subject to both taxes
  - rural land subject to IMU except land owned by small independent farmers
- Valuation system:
  - based on estimates of market rental values from 1988–89; cadastral value increased by 5 percent (multiplied by 1.05) and then further increased by property-specific coefficients (1.6 for houses; between 1.055–1.4 for other property types)
  - national average ratio of market to taxable value: 2.1 for primary residences and other dwellings (large regional variation)
- Tax rates and municipal flexibility:
  - IMU basic tax rate: 0.76 percent on updated cadastral value; municipalities can set rate in range between 0.46–1.06 percent
  - TASI national standard rate: 0.1 percent; municipalities can set rate in range between 0–0.33 percent
  - sum of IMU and TASI rates cannot exceed the maximum IMU rate
- Distributional incidence:
  - Property tax liability decreases with income: about 5 percent of household disposable income in the first decile to about 2 percent in the highest deciles.
  - Exemption on main residences equivalent to about 0.2 percent of GDP; increases disposable income by about 0.9 percent in the lowest decile and about 0.3 percent in the highest decile.
  - Without the exemption, incidence would increase to 5.7 percent (poorest) and 2.4 percent (wealthiest).
- Assessment disparities and regressivity:
  - National average ratio of market to taxable value by region (selected):
    - All Italy: Primary Dwelling average market value €190,550; Ratio of Market Value to Taxable Value 2.08
    - All Italy: Other Housing average market value €134,764; Ratio 2.14
  - Strong negative relationship between assessment ratios and average market value across regions leads to declining effective average tax rate as market values increase.
  - Positive relationship between household disposable income and average market value across regions; combined with assessment disparities, the effective average tax rate declines with income, producing regressive outcomes.
- Regional examples (primary dwelling average market value and ratio to taxable value):
  - Trentino Alto Adige: Average Market Value €335,738; Ratio 3.22
  - Molise: Average Market Value €112,270; Ratio 1.53

### IV. Reform Scenarios and Simulation Results — Overview
- Reform objective: shift taxation away from labor income toward consumption and property while keeping revenue-neutrality relative to a reduction in the tax wedge on labor toward other EU-15 countries’ average.
- Reform principles: reduce complexity, broaden tax base, lower marginal tax rates.
- Reform scenarios:
  - (1) PIT design options: simplify structure (fewer brackets), streamline credits/deductions, reduce marginal tax rates while preserving progressivity; alternative flatter tax assessed. Target reduction in tax wedge on labor income by 4.5 percentage points, equivalent to 2 percent of GDP in budget revenue.
  - (2) VAT and property tax reform: streamline reduced rates and exemptions with possible reduction in standard rate; revalue property tax base to market values allowing substantial rate reductions for comparable revenue.
  - (3) Distributive effects: simulate revenue-neutral shifts away from labor income for alternative VAT and property tax reform options.

### A. Simulation of a Personal Income Tax Reform — Details and Estimated Effects
- Two PIT reform options evaluated both targeting an average reduction in the tax wedge on labor of 4.5 percent.
- Net cost of the reform proposals to the budget: about €36 billion (2 percent of GDP).
- Reform 1 (low- and middle-income focus):
  - Simplify PIT to four tax brackets.
  - Reduce tax rate on the first taxable income bracket to 9 percent (from 23 percent).
  - Preserve second bracket (income between 15,000–28,000) at 27 percent.
  - Increase tax rate on third bracket (income between 28,000–55,000) to 41 percent (from 38 percent).
  - Merge last two brackets into one for income above 55,000 at 44 percent.
  - Eliminate the National Income Tax Bonus (the €80 bonus) and the tax credit for building refurbishment and construction.
  - Distributional simulation result: increases disposable income of households between the third- and seventh-income deciles (incomes between 20,000–40,000) by about [text truncated in source at this point].

*Source: Authors’ calculations and cited studies as reported in the source document.*

### 5.5 percent, twice as high as the benefit for the wealthiest households (Figure 5). While all

### wpiea2020037-print-pdf - 5.5 percent, twice as high as the benefit for the wealthiest households (Figure 5). While all

### Reform 1 and Reform 2: Personal Income Tax (PIT) reforms — distributional impacts and fiscal costs
- Reform 1 (proposed reform described):
  - All households along the income distribution would pay lower taxes; middle-income groups get the highest reduction in PIT liability.
  - Households with incomes below 20,000 virtually pay no income tax due to the lower tax rate and the interaction of tax credits and deductions.
  - In absolute (euro) terms, the proposed reform benefits wealthier families more because they pay a lower tax rate on the first part of their income.
  - Annual disposable income increases:
    - about €820 for the lowest five deciles of the income distribution
    - about €1,600 for the highest three deciles
- Alternative base-broadening options considered:
  - Elimination of the National Income Tax Bonus may address labor-supply distortions created by large variability of effective marginal income tax rates (Impact Assessment Office, 2017).
  - Elimination of the tax credit for building refurbishment and construction motivated by equity considerations because it is a relatively costly tax expenditure that benefits wealthy households relatively more.
- Reform 2: A flatter PIT reform (two brackets simulated):
  - Structure: taxable income below 30,000 taxed at 15 percent; taxable income above 30,000 taxed at 44 percent.
  - Cost of this reform compensated with elimination of the National Income Tax Bonus and the tax credit for building refurbishment and construction (as in Reform 1).
  - Distributional outcomes:
    - In absolute (euro) terms, the flatter reform benefits wealthier families significantly more.
    - It increases average annual disposable income:
      - by about €530 for households in the first five deciles
      - by about €2,700 for the highest three deciles
    - When assessed relative to disposable income:
      - disposable income of middle-income households is estimated to increase by about 3.5 percent
      - disposable income increases by about 5 percent among the highest three deciles
  - An even flatter PIT reform (for instance, with rates at 15 and 20 percent) would be exponentially costlier to the budget, and highly regressive (IMF, 2019).
- Fiscal cost and offsets:
  - Proposed changes to the tax structure (tax brackets and rates only) have a cost to the budget estimated at about €50 billion.
  - Elimination of the National Income Tax Bonus produces savings of €8.0 billion.
  - Elimination of the tax credit on refurbishment and construction produces savings of €5.7 billion.

### VAT reform simulations — revenue-neutral and revenue-increasing base-broadening options
- Policy aim: generate additional revenue to compensate half- or the full cost of the income tax reform (1–2 percent of GDP), while minimizing impact on poorer households.
- Revenue-neutral base-broadening VAT reform alternatives:
  - Option A: eliminate reduced rate of 4 percent and reduce standard rate to 17.5 percent (calibrated to be revenue neutral).
  - Option B: eliminate reduced rate of 10 percent and reduce standard rate to 18.5 percent (calibrated to be revenue neutral).
- Distributional outcomes for revenue-neutral options:
  - Eliminating the reduced rate of 10 percent (lowering standard to 18.5 percent) is largely neutral for income distribution; disposable income mostly unchanged across income levels.
  - Eliminating the reduced rate of 4 percent (lowering standard to 17.5 percent) is regressive:
    - Redistributes income from poorer to wealthier households by an estimated 0.4 percent of disposable income.
    - In absolute terms, reduces disposable income of the two lowest deciles by about €50 and increases disposable income of the highest decile by about €250.
  - Explanation: the reduced rate of 4 percent is targeted to goods largely consumed by poorer households; the reduced rate of 10 percent is not always targeted to poorer-consumed categories.
- Base-broadening VAT reforms for higher revenue:
  - Two simulated reforms:
    - Eliminate reduced rate of 10 percent and set standard rate to 21 percent to generate additional revenue of about 1 percent of GDP.
    - Eliminate reduced rate of 10 percent and set standard rate to 24 percent to generate additional revenue of about 2 percent of GDP.
  - Distributional impacts:
    - Revenue-increasing VAT reforms are regressive; negative impact on income distribution increases proportionally with revenue targeted.
    - A 1 percent of GDP VAT reform reduces disposable income of the first two deciles by about 1.5 percent; for the highest two deciles, the reduction is slightly below 1 percent.
    - Resulting VAT liability translates to:
      - close to 14 percent of household disposable income among poorer households (for the 1 percent of GDP reform)
      - about 9 percent among wealthier households (for the 1 percent of GDP reform)
    - The 2 percent of GDP reform has roughly twice the impact of the 1 percent reform.
    - In absolute (euro) terms:
      - poorer households pay additional VAT estimated at €150 (for the 1 percent of GDP reform) and €300 (for the 2 percent of GDP reform)
      - wealthier households pay additional VAT estimated at about €700 (for the 1 percent reform) and €1,500 (for the 2 percent reform)
- Note on simulation calibration:
  - Revenue implications of VAT tax reforms differ somewhat (0.1–0.3 percent of GDP) depending on whether households modify consumed quantities; simulations calibrated at mid-point between fixed quantities and fixed expenditure shares.

### Property tax reform simulations — valuation and base broadening
- Objectives: generate additional revenue to compensate half- or full cost of the income tax reform (1–2 percent of GDP), while minimizing impact on poorer households.
- Valuation system linked to market value — simulation methodology:
  - Current taxable values reported in EU-SILC are multiplied by the ratio of market to taxable value of properties by geographical region.
  - Updated taxable value formula used in simulations:
    - Taxable Value_i = Cadastral Value_i × 1.05 × 1.60 × Ratio_i
    - Where i are the 20 geographical regions in Italy; multiplicative factors (1.05; 1.6) applied to all properties as in the current system; and the ratio reflecting the remaining gap between market and taxable values is applied according to geographical region.
- Revenue-neutral property tax reform based on market valuation:
  - With broadening to include all immovable property (including owner-occupied main residences) and improvement of the valuation system, current property tax revenue can be obtained with a significantly lower tax rate.
  - Current revenue from property tax on immovable property (about 1 percent of GDP in 2018) can be obtained with a tax rate of 0.3 percent.
  - Comparison: current maximum allowed tax rate of 1.06 percent for other residences.

*Source: wpiea2020037-print-pdf*

### 0.4 percent on main residences.

### wpiea2020037-print-pdf - 0.4 percent on main residences.

### Property valuation and a revenue-neutral market-valued property tax
- A revenue-neutral property tax based on market valuation:
  - Increases disposable income of poorest households by 3 percent.
  - Decreases disposable income of wealthier households by 0.5 percent.
- Impact on property tax liability (first decile):
  - Before reform: 4.8 percent of disposable income.
  - After reform: about 2 percent of disposable income.
  - Annual savings: about €275.
- Impact on property tax liability (two highest deciles):
  - Before reform: about 2 percent of disposable income.
  - After reform: 2.6 percent of disposable income.
  - Higher annual property taxes: about €460.
- Equity and liquidity considerations:
  - Property taxes can burden highly indebted or liquidity-constrained households.
  - Suggested mitigation: a deferral scheme where tax liability becomes payable only when the property is transferred (for instance through sale or inheritance).
- Cross-country context:
  - Tax rates on immovable property for countries in the European Union range between 0.4-1 percent, on average.

### Base-broadening property tax reforms for higher revenue (market valuation)
- Reform variants assuming all immovable property (including main residence) is taxed on market value:
  - Tax rate of 0.55 percent on all properties → additional revenue of about 1 percent of GDP.
  - Tax rate of 0.8 percent on all properties → additional revenue of about 2 percent of GDP.
- Distributional effects for the 1 percent of GDP revenue increase (tax rate 0.55 percent):
  - Disposable income of households in the first decile increases by 0.8 percent.
  - Disposable income decreases from 1 percent in the second decile to about 3 percent in the highest decile.
  - Resulting property tax liability increases with disposable income from 4 percent to above 5 percent.
- Distributional effects for the 2 percent of GDP revenue increase (tax rate 0.8 percent):
  - Impacts roughly twice as large as the 1 percent of GDP reform.
  - Disposable income decreases from 0.5 percent in the first decile to 5 percent in the highest decile.
  - In absolute euro terms:
    - Poorer households (first two deciles) pay additional property tax estimated at €30 (for the 1 percent of GDP reform) and €120 (for the 2 percent of GDP reform).
    - Wealthier households (highest two deciles) pay additional property tax estimated at about €1,700 (for the 1 percent of GDP reform) and €2,900 (for the 2 percent of GDP reform).

### Distributive effects of revenue-neutral shifts from PIT to VAT and/or property tax
- Three financing options to offset a PIT reform lowering labor tax burden by 2 percent of GDP:
  1. Fully financed through the proposed VAT reform.
  2. Equally financed through the proposed VAT and property tax reforms (1 percent of GDP each).
  3. Fully financed through the proposed property tax reform.
- PIT → VAT (fully financed by VAT, revenue-neutral shift of 2 percent of GDP):
  - Disposable income of middle-income households (third to seventh deciles) increases by 2.5 percent on average → average annual tax relief of €620 per household.
  - Disposable income of highest three deciles increases by 0.8 percent on average → €250.
  - Disposable income of low-income households (first two deciles) decreases by 1 percent on average → €100.
- PIT → VAT and property tax (1 percent of GDP VAT + 1 percent of GDP property tax):
  - Disposable income of middle-income households increases by 2.0 percent on average → €500.
  - Disposable income of low-income households increases by 0.6 percent on average → €60.
  - Disposable income of high-income households decreases by 0.7 percent on average → €600.
  - Adding property tax progressivity reduces gains for middle-income households and shifts costs to high-income households.
- PIT → property tax (fully financed by property tax, revenue-neutral shift of 2 percent of GDP):
  - Disposable income of low-income households increases by 1.4 percent on average.
  - Disposable income of middle-income households increases by 1.7 percent on average.
  - Disposable income of high-income households decreases by 1.5 percent on average.
  - Percentage change in disposable income becomes negative from the 8th decile onwards.

### Conclusions and policy implications
- Principle: A broader tax base combined with low tax rates is desirable to limit deadweight loss; narrow bases plus high marginal rates increase efficiency costs.
- Italian tax-code inefficiencies:
  - Tax expenditures have eroded the tax base, increasing reliance on higher marginal rates (especially on labor income).
  - Complexity and inefficiency have contributed to a significant compliance gap.
  - Broad-based revenue needs require a comprehensive base-broadening reform.
- Summary implications of reforms analyzed:
  - Personal income tax (PIT) reform that lowers the tax burden on labor while preserving progressivity benefits low- and middle-income households the most; simplification can lower bracket count and streamlining poorly targeted tax credits/deductions can offset costs.
  - Base-broadening VAT reform should be carefully designed to limit regressive effects; eliminating the reduced rate of 10 percent while lowering the standard rate can be almost neutral distributionally.
  - Updating property valuation from cadastral to market values is essential to address equity concerns and increase revenue; current cadastral valuations erode the tax base and disproportionately burden the poor.
  - A revenue-neutral shift from labor income to consumption and property taxation can be designed to benefit low- and middle-income households, lowering their overall tax liability while raising progressivity.
- Implementation note:
  - Tax administration strengthening and a holistic strategy are required to realize compliance gains; detailed organizational assessment is beyond the paper’s scope.

### Appendix — Matching and imputation methodology (consumption imputation)
- Matching approach overview:
  1. Estimate a consumption function using SHIW 2016 to predict aggregated household consumption (log-consumption on log-disposable income and covariates) using OLS.
     - Selected coefficient highlights (Appendix Table 1):
       - Income (logs): -0.789*** (standard error 0.037)
       - Income Squared (logs): 0.069*** (0.002)
       - R2: 0.599
       - Observations: 16,333
       - F-statistic: 997.89
  2. Impute aggregated household consumption in EU-SILC using estimated parameters.
     - Comparison (Appendix Table 2):
       - SHIW Income (logs): -0.811*** (0.019)
       - EU-SILC Income (logs): -0.825*** (0.006)
       - Income Squared (logs): 0.071*** (SHIW) and 0.073*** (EU-SILC)
       - R2: 0.587 (SHIW) and 0.981 (EU-SILC)
       - Observations: 16,333 (SHIW) and 21,704 (EU-SILC)
  3. Match individual households in HBS and EU-SILC using propensity score matching on predicted consumption and covariates.
     - Matching quality: mean bias reduced from 7.5 before matching to 2.4 after matching; post-matching null that group means do not differ cannot be rejected for any variable.
  4. Compute disaggregated consumption shares from HBS across 12 COICOP categories and impute them to matched EU-SILC households to derive category-level consumption.
     - 12 COICOP categories used: 1) food and non-alcoholic beverages; 2) alcoholic beverages and tobacco; 3) clothing and footwear; 4) home fuels, electricity, rents (excluding imputed rents); 5) furniture and household services; 6) health; 7) transport; 8) communication; 9) recreation and culture; 10) education; 11) hospitality and catering; 12) other goods and services.
     - Imputed consumption in EU-SILC closely matches HBS; 95 percent of consumption is matched well across datasets.

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020037-print-pdf.pdf*

### Appendix Table 3. Matching Households in HBS and EU-SILC

### Appendix Table 3. Matching Households in HBS and EU-SILC

### Probit regression (psmatch2) — coefficients and significance
- Consumption: coefficient -0.001*** (0.000)
- Male Household Head: coefficient -0.105* (0.015)
- Married Household Head: coefficient -0.044* (0.015)
- Age 30-64: coefficient -0.843 (0.021)
- Age 65+: coefficient -0.707 (0.023)
- Lower secondary Education: coefficient 0.051*** (0.020)
- Upper secondary Education: coefficient 0.142* (0.020)
- Tertiary Education: coefficient 0.127* *0.024)
- Self Employed: coefficient 0.073** (0.018)
- Constant: 0.834 (0.028)
- Pseudo R2: 0.036
- LR chi2 prob. (Unmatched): 0.000
- LR chi2 prob. (Matched): 0.002
- Observations: 38,650

### Propensity score (PS) test — unmatched (U) and matched (M) means and % bias
- Consumption
  - U: Treated 22,675; Control 22,675; % bias -0.0
  - M: Treated 22,675; Control 23,270; % bias -4.1***
- Male Household Head
  - U: Treated 0.613; Control 0.655; % bias -8.7
  - M: Treated 0.613; Control 0.604; % bias 1.9**
- Married Household Head
  - U: Treated 0.494; Control 0.527; % bias -6.7
  - M: Treated 0.494; Control 0.499; % bias -1.0***
- Age 30-64
  - U: Treated 0.429; Control 0.541; % bias -22.7
  - M: Treated 0.429; Control 0.422; % bias 1.3***
- Age 65+
  - U: Treated 0.365; Control 0.394; % bias -6.1
  - M: Treated 0.365; Control 0.371; % bias -1.3***
- Lower secondary Education
  - U: Treated 0.259; Control 0.283; % bias -5.6
  - M: Treated 0.259; Control 0.268; % bias -2.1*
- Upper secondary Education
  - U: Treated 0.384; Control 0.342; % bias 8.8
  - M: Treated 0.384; Control 0.390; % bias -1.3***
- Tertiary Education
  - U: Treated 0.163; Control 0.148; % bias 4.1
  - M: Treated 0.163; Control 0.153; % bias 2.7*
- Self Employed
  - U: Treated 0.162; Control 0.144; % bias 4.9
  - M: Treated 0.162; Control 0.140; % bias 6.2*

### Balance and summary statistics
- U Mean Bias: 7.5
- M Mean Bias: 2.4
- Notes: Standard errors in parentheses. ***(**,*) indicate significance at 1(5,10) percent in the probit regression. In the propensity score test (PS test), a t-test on the null hypothesis that the mean value of each variable is the same in the treatment group and then non-treatment group is done before and after matching. ***(**,*) indicate that the null hypothesis cannot be rejected at 1(5,10) percent.

*Italic source: Appendix Table 3 from the provided IMF PDF.*

### Appendix Table 4. Consumption Shares in HBS and EU-SILC

- Food and non-alcoholic beverages
  - HBS Mean 27.2; Std. Dev. 13.5; Household Consumption 466.4
  - EU-SILC Mean 26.1; Std. Dev. 12.4; Household Consumption 452.8
- Alcoholic beverages and Tobacco
  - HBS Mean 2.4; Std. Dev. 4.1; Household Consumption 44.3
  - EU-SILC Mean 2.5; Std. Dev. 3.9; Household Consumption 47.3
- Clothing and footwear
  - HBS Mean 5.5; Std. Dev. 7.4; Household Consumption 122.6
  - EU-SILC Mean 5.7; Std. Dev. 7.3; Household Consumption 115.0
- Home fuels, electricity, rents
  - HBS Mean 18.9; Std. Dev. 14.3; Household Consumption 309.1
  - EU-SILC Mean 18.2; Std. Dev. 13.5; Household Consumption 314.5
- Furniture and household services
  - HBS Mean 4.9; Std. Dev. 7.4; Household Consumption 111.2
  - EU-SILC Mean 5.2; Std. Dev. 7.8; Household Consumption 99.5
- Health
  - HBS Mean 6.4; Std. Dev. 8.5; Household Consumption 123.8
  - EU-SILC Mean 6.1; Std. Dev. 8.1; Household Consumption 110.9
- Transport
  - HBS Mean 11.8; Std. Dev. 12.3; Household Consumption 287.4
  - EU-SILC Mean 11.9; Std. Dev. 11.9; Household Consumption 256.8
- Communication
  - HBS Mean 3.6; Std. Dev. 3.1; Household Consumption 64.0
  - EU-SILC Mean 3.5; Std. Dev. 2.8; Household Consumption 63.0
- Recreation and culture
  - HBS Mean 5.3; Std. Dev. 6.5; Household Consumption 131.1
  - EU-SILC Mean 5.8; Std. Dev. 6.7; Household Consumption 124.1
- Education
  - HBS Mean 0.6; Std. Dev. 2.1; Household Consumption 15.7
  - EU-SILC Mean 0.6; Std. Dev. 2.1; Household Consumption 13.8
- Hospitality and catering
  - HBS Mean 5.0; Std. Dev. 7.0; Household Consumption 120.8
  - EU-SILC Mean 5.6; Std. Dev. 7.2; Household Consumption 120.2
- Other
  - HBS Mean 8.4; Std. Dev. 7.3; Household Consumption 180.6
  - EU-SILC Mean 8.8; Std. Dev. 7.3; Household Consumption 171.7
- Observations: HBS 16946; EU-SILC 21704
- Notes: The shares are computed as average of household level ratios, in percent of total consumption expenditures. Household consumption is for monthly expenditures in euro.

### Key findings from figures and micro-simulation results
- Density of Consumption: Appendix Figure 1 presents density of consumption in HBS and EU-SILC (authors’ own calculations).
- Using household consumption instead of disposable income yields a somewhat more progressive tax system in Italy (Appendix Figure 2):
  - Baseline PIT liability increases from 1.7 to 42 percent
  - VAT liability increases from 12.3 to 13 percent
  - Property tax liability decreases from 4.6 to 2.7 percent
- Revenue-neutral shift away from income tax toward VAT and property tax (Appendix Figure 3):
  - Net impact for middle-income households (third to seventh deciles): increase in household consumption of about 3.6 percent (€700)
  - Impact on low- and high-income households: almost unchanged

*Italic source: Extracted content from the supplied IMF PDF (Appendix Tables 3 and 4 and accompanying text).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020037-print-pdf.pdf_