## _wp14124 - References

## Source details

**Canonical URL:** [_wp14124 - References](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14124.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14124.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14124.pdf.json)

---

### I. Introduction and study purpose
- Latin America (LA) experienced strong economic growth and improved social indicators in the last decade:
  - The region’s real GDP has grown by an annual average of over 4 percent.
  - Unemployment declined to multi-year lows; public debt and inflation declined significantly.
- Study objectives:
  - Document developments in social indicators (income inequality, access to education and basic services, poverty and polarization) via historical trends and cross-regional comparisons; investigate convergence of income levels across population segments.
  - Explore possible drivers behind the decline in income inequality in Latin America as a whole using correlation and econometric techniques (including panel regressions), extending the sample beyond LA and using country-fixed effects to control for country-specific differences.

### Main methodological notes
- Multiple methodologies deployed: correlation analysis, econometric panel regressions, growth incidence curves, and fixed-effects panel analysis for robustness.
- Sample definitions:
  - "Latin America" refers to Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela (unless otherwise noted).
  - Middle class defined as people with per capita income between $10-$50 per day (2005 PPP) following Milanovic and Yitzhaki (2002).

### Major high-level findings (summary)
- Regional trends:
  - Latin America and Sub-Sahara Africa are the only regions that experienced declines in income inequality in the last decade.
  - Latin America saw a decline in its Gini coefficient of around 3 Gini points over the last decade; SEDLAC suggests around 4 Gini points, World Bank data suggest around 3 Gini points.
  - Poverty and polarization rates have declined since the 1990s; the middle class surged from 20 percent of the population a decade ago to almost half of the population.
- Attribution and drivers:
  - Well-designed policies explain more than half of the decline in income inequality in Latin America (as estimated in the paper).
  - Econometric analysis suggests the largest contributors to the decline:
    - Higher education spending: explains almost one-fourth of the total decline in income inequality.
    - Higher FDI (partly reflecting strong economic fundamentals).
    - Higher tax revenue.
  - Appreciating exchange rates have a small but dampening effect on equality.
  - The Kuznets curve is confirmed: economic growth has been conducive to declining income inequality, though its impact is limited.
- Correlation results:
  - Tax revenues (including direct and property taxation) and spending on education are negatively correlated with inequality changes in the LA sample.

### II. Social indicators: stylized facts
- Key statistics and distributional facts:
  - Gini coefficient: now hovering at around 50 Gini points (out of 100, World Bank, 2014).
  - Gini decline: unweighted average decline around 3-4 Gini points in the last decade.
  - Middle class: almost half of Latin America’s population is now regarded as middle class (up from 20 percent a decade ago).
  - Poverty: almost 85 million people living below $2.5 per day (World Bank data).
  - Richest 10 percent of households in Latin America possess on average 37 percent of the total per capita income.
  - Poorest 10 percent possess on average 1.5 percent of total per capita income.
  - The richest households earn 25 times more than the poorest households on average across LA.
  - Cross-country extremes: income gap between highest and lowest earners ranges from 55 times more (Honduras) to 15 times more (Uruguay and El Salvador).
- Skill premia and education:
  - Declining skill premia: in 2012 high-skilled workers earned on average 2.7 times the wages of low-skilled workers (compared to 4 times more in 2000).
  - Increase in the relative supply of workers with completed secondary and tertiary education is posited as a main factor behind declining returns to education.
  - Educational inequality (Gini for years of education) is high in many countries, notably Panama and Ecuador; gaps in years of schooling between high- and low-income households vary (from 5 years in Argentina to 8½ years in Panama).
  - Gender differences in years of education: in half of LA countries considered, men have more years of schooling than women (positive difference).
- Infrastructure, basic services and health:
  - Access to sewage varies widely: from 10 percent coverage in Paraguay to almost universal coverage in Chile; within-country quintile disparities are large (example: in Peru 20 percent of lower-income households have access to sewage versus almost 90 percent of higher-income households).
  - Skilled physicians often do not attend births in the least wealthy households in LA, in contrast to emerging Europe.
  - Rural–urban disparities in under-five mortality are much higher in LA than in emerging Europe.
  - Stunting among children and other health outcomes are less favorable in LA than in emerging Europe despite similar health spending levels.

### III. Cross-country patterns and exceptions
- Countries with large declines in income inequality (sample of over 170 countries, 1990–2012; large = at least 3 Gini points decline since the 1990s):
  - Out of 29 countries with large declines, almost half are in Latin America.
  - Examples and magnitudes (selected from Table 1): Mali -17.5; Peru -13.1; Bolivia -11.6; Kyrgyz Republic -11.5; El Salvador -9.8; Ecuador -9.7; Brazil -7.4; Mexico -4.9; Argentina -4.8; Paraguay -4.7; Colombia -3.1.
- Countries with large increases in income inequality (Table 2 examples):
  - Indonesia 8.4; China, P.R.: Mainland 6.3; Costa Rica 3.4; Honduras 3.1.
- Cross-country convergence:
  - Conditional convergence detected: countries with higher initial inequality tend to experience larger declines in Gini coefficient; similar conditional convergence observed for poverty rates.
- Notable country exceptions and drivers:
  - Honduras and Costa Rica experienced rising Gini coefficients over the last decade; Honduras also experienced a rising poverty rate.
  - Honduras’ rise in inequality attributed to increased dispersion of labor incomes in rural areas between tradable and non-tradable sectors, overvalued currency, poor agricultural exports, segmented labor markets, poor educational progress.
  - Costa Rica’s rising inequality attributed to a large informal sector, widening wage gaps, and a large unskilled labor force stemming from 1980s crisis effects on schooling.

### IV. What determines income inequality in Latin America? (analysis summary)
- Simple correlations (LA sample):
  - Negative correlation between changes in total tax revenue (percent of GDP) and changes in the Gini coefficient.
  - Negative correlation between changes in direct tax revenue (percent of GDP) and changes in the Gini coefficient.
  - Negative correlation between changes in property tax revenue (percent of GDP) and changes in the Gini coefficient.
  - Negative correlation between changes in education spending (percent of GDP) and changes in the Gini coefficient.
  - Interpretation caveats: correlations do not indicate causation; variables chosen reflect data availability and policy relevance.
- Econometric panel results (44 emerging and developing countries, 1990–2010, country and time fixed effects):
  - Policies explain more than half of the decline in income inequality in Latin America (per the paper’s econometric estimates).
  - Higher education spending is estimated to explain almost one-fourth of the total decline.
  - Higher FDI and higher tax revenue are also important contributors.
  - Appreciating exchange rates have a small dampening (inequality-increasing) effect.
  - The Kuznets curve holds in the sample: growth has been conducive to declining income inequality, but the quantitative impact is limited.

### V. Policy-relevant implications highlighted in the study
- Role of well-designed policies:
  - Targeted social transfers and increases in education spending can substantively contribute to inequality reductions.
  - Strengthening tax revenue capacity—especially through progressive direct taxes and property taxes—can be associated with reduced inequality and can finance redistributive spending.
- Structural impediments and priorities:
  - Continue educational upgrading and expand access to reduce skill premia and support inclusive growth.
  - Improve infrastructure and basic services access to reduce within-country disparities.
  - Monitor exchange rate policies and external factors that can have distributional consequences.
- Specific empirical policy notes:
  - Based on the panel estimates (2001–10), education spending, taxation, depreciating exchange rates (expected negative sign), and FDI are associated with lower inequality.
  - From Table 4 results:
    - Real effective exchange rate: 0.02 (0.06)** 
    - Tax revenue (in percent of GDP): -0.08 (-0.04)** 
    - Education spending (in percent of GDP): -0.72 (-0.23)** 
    - FDI (in percent of GDP): -0.12 (0.06)** 
    - Observations: 274
    - Countries: 38
    - Time period: 2001-10
    - Adjusted R-square: 0.98
    - Time and country dummies: √
  - Higher education spending explains almost a fourth of the overall decline alone (i.e., 0.8 out of the 3 Gini points).
  - Higher FDI and higher tax revenue jointly explain more than a forth of the total decline in the Gini coefficient over the last decade.
  - Exchange rate policies have been associated with higher inequality in the region because currencies have appreciated on average in LA over the last decade.
  - Common external factors (proxied by time dummies) explain the remaining change in LA’s Gini coefficient over the last decade.
- Fiscal context:
  - Tax revenue was 20 percent in LA versus 34 percent in OECD countries in 2012.
  - Income and profit taxes account for one-fourth of total tax revenue in LA versus 35 percent in OECD countries.
  - Only Colombia and Uruguay collect more than 1 percent of GDP through recurrent property taxes.

### VI. Kuznets hypothesis, empirical specification and results
- Kuznets hypothesis:
  - Inverted U-shaped relationship between GDP per capita and the Gini coefficient: rising inequality at early stages of growth, declining inequality after a turning point.
  - Contested in the literature; some studies support it, others reject it when controlling for country-specific effects.
- Estimated Kuznets specification (natural logarithms; country and time fixed effects):
  - Ln(Gini_it) = β1 ln(GDPPC_it) + β2 [ln(GDPPC_it)]^2 + T_t + C_i + ɛ_it
  - Table 3 reported estimates:
    - GDP per capita: 0.45 (0.06)** 
    - Squared GDP per capita: -0.03 (0.00)** 
    - Observations: 910
    - Countries: 44
    - Time period: 1990-2010
    - Adjusted R-square: 0.94
    - Time and country dummies: √
  - Interpretation:
    - Coefficients satisfy β1 > 0 and β2 < 0 and are statistically significant, supporting the presence of a Kuznets curve in this specification.
    - The Kuznets curve does not explain the bulk of variations in inequality across countries or over time.
    - Analysis suggests that only one eight of the decline in Latin America’s income inequality in the last decade can be associated with recent strong growth momentum.

### VII. Caveats, limitations, and suggested extensions
- Limitations:
  - Inherent limitations of the data set and cross-country regression analysis, including inability to establish causality with full confidence.
  - The list of policies is not exhaustive; excluded variables due to data limitations include social/cash transfers, health spending, quality/types of education, infrastructure spending, and access to basic services.
  - The analysis uses the aggregate Gini coefficient; other dependent variables could be informative (e.g., disaggregated income sources, 90/10 income ratio, wage dispersion, labor income share).
  - Time dummies are a crude proxy for non-policy external factors.
- Heterogeneity:
  - Country- and time-specific heterogeneity noted: different policies may have different effects in different countries and times.
- Suggested extensions:
  - Incorporate social transfers, health spending, infrastructure measures, and disaggregated inequality measures where data permit.
  - Explore country-specific case studies to complement cross-country panel findings.

### VIII. Conclusions
- Despite recent improvements, Latin America remains the most unequal region in the world.
- Empirical panel analysis indicates well-designed policies can explain more than half of the decline in income inequality (averaging around 3 Gini points over the last decade), with education spending explaining almost one Gini point.
- Stronger FDI and tax revenues were also important contributors to the decline; growth dynamics (Kuznets channel) were less pivotal.
- Policy recommendations emphasized:
  - Improving access of low-income families to education to boost equality of opportunity and lower income inequality over the long run; strengthen access to quality education given relatively high spending but poor outcomes in LA.
  - Attracting stronger FDI by maintaining strong economic fundamentals.
  - Raising tax revenue and improving tax composition and administration (progressivity, address evasion, broaden tax base, exploit property taxation where feasible) to finance redistributive policies.
- Final caution: cross-country regression results should be interpreted carefully; causality is difficult to establish and policy effects vary by country and time.

*Italic source attribution: IMF working paper content as provided in the supplied document.*

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

### _wp14124 - References

### I. Introduction and study purpose
- Latin America (LA) experienced strong economic growth and improved social indicators in the last decade:
  - The region’s real GDP has grown by an annual average of over 4 percent.
  - Unemployment declined to multi-year lows; public debt and inflation declined significantly.
- Study objectives:
  - Document developments in social indicators (income inequality, access to education and basic services, poverty and polarization) via historical trends and cross-regional comparisons; investigate convergence of income levels across population segments.
  - Explore possible drivers behind the decline in income inequality in Latin America as a whole using correlation and econometric techniques (including panel regressions), extending the sample beyond LA and using country-fixed effects to control for country-specific differences.

### Main methodological notes
- Multiple methodologies deployed: correlation analysis, econometric panel regressions, growth incidence curves, and fixed-effects panel analysis for robustness.
- Sample definitions:
  - "Latin America" refers to Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela (unless otherwise noted).
  - Middle class defined as people with per capita income between $10-$50 per day (2005 PPP) following Milanovic and Yitzhaki (2002).

### Major high-level findings (summary)
- Regional trends:
  - Latin America and Sub-Sahara Africa are the only regions that experienced declines in income inequality in the last decade.
  - Latin America saw a decline in its Gini coefficient of around 3 Gini points over the last decade; SEDLAC suggests around 4 Gini points, World Bank data suggest around 3 Gini points.
  - Poverty and polarization rates have declined since the 1990s; the middle class surged from 20 percent of the population a decade ago to almost half of the population.
- Attribution and drivers:
  - Well-designed policies explain more than half of the decline in income inequality in Latin America (as estimated in the paper).
  - Econometric analysis suggests the largest contributors to the decline:
    - Higher education spending: explains almost one-fourth of the total decline in income inequality.
    - Higher FDI (partly reflecting strong economic fundamentals).
    - Higher tax revenue.
  - Appreciating exchange rates have a small but dampening effect on equality.
  - The Kuznets curve is confirmed: economic growth has been conducive to declining income inequality, though its impact is limited.
- Correlation results:
  - Tax revenues (including direct and property taxation) and spending on education are negatively correlated with inequality changes in the LA sample.

*Sources for methodology and findings: the study’s approaches include correlation plots, growth incidence curves, SEDLAC data, World Bank data, and panel econometric analysis (44 emerging and developing countries, period 1990–2010, with country and time fixed effects).*

### II. Social indicators: stylized facts

- Inequality, poverty, polarization, and middle class:
  - Gini coefficient: now hovering at around 50 Gini points (out of 100, World Bank, 2014).
  - Gini decline: unweighted average decline around 3-4 Gini points in the last decade.
  - Middle class: almost half of Latin America’s population is now regarded as middle class (up from 20 percent a decade ago).
  - Poverty: almost 85 million people living below $2.5 per day (World Bank data).
- Distributional statistics:
  - Richest 10 percent of households in Latin America possess on average 37 percent of the total per capita income.
  - Poorest 10 percent possess on average 1.5 percent of total per capita income.
  - The richest households earn 25 times more than the poorest households on average across LA.
  - Cross-country extremes: income gap between highest and lowest earners ranges from 55 times more (Honduras) to 15 times more (Uruguay and El Salvador).
- Skill premia and education:
  - Declining skill premia: in 2012 high-skilled workers earned on average 2.7 times the wages of low-skilled workers (compared to 4 times more in 2000).
  - Increase in the relative supply of workers with completed secondary and tertiary education is posited as a main factor behind declining returns to education.
  - Educational inequality (Gini for years of education) is high in many countries, notably Panama and Ecuador; gaps in years of schooling between high- and low-income households vary (from 5 years in Argentina to 8½ years in Panama).
  - Gender differences in years of education: in half of LA countries considered, men have more years of schooling than women (positive difference).
- Infrastructure and basic services:
  - Access to sewage varies widely: from 10 percent coverage in Paraguay to almost universal coverage in Chile (by income quintile disparities are large within countries; example: in Peru 20 percent of lower-income households have access to sewage versus almost 90 percent of higher-income households).
- Health outcomes and access:
  - Skilled physicians often do not attend births in the least wealthy households in LA, in contrast to emerging Europe.
  - Rural–urban disparities in under-five mortality are much higher in LA than in emerging Europe.
  - Stunting among children and other health outcomes are less favorable in LA than in emerging Europe despite similar health spending levels.

### III. Cross-country patterns and exceptions
- Countries with large declines in income inequality (sample of over 170 countries, 1990–2012; large = at least 3 Gini points decline since the 1990s):
  - Out of 29 countries with large declines, almost half are in Latin America.
  - Examples and magnitudes (selected from Table 1): Mali -17.5; Peru -13.1; Bolivia -11.6; Kyrgyz Republic -11.5; El Salvador -9.8; Ecuador -9.7; Brazil -7.4; Mexico -4.9; Argentina -4.8; Paraguay -4.7; Colombia -3.1.
- Countries with large increases in income inequality (Table 2 examples):
  - Indonesia 8.4; China, P.R.: Mainland 6.3; Costa Rica 3.4; Honduras 3.1.
- Cross-country convergence:
  - Conditional convergence detected: countries with higher initial inequality tend to experience larger declines in Gini coefficient; similar conditional convergence observed for poverty rates.
- Notable country exceptions and drivers:
  - Honduras and Costa Rica experienced rising Gini coefficients over the last decade; Honduras also experienced a rising poverty rate.
  - Honduras’ rise in inequality attributed to increased dispersion of labor incomes in rural areas between tradable and non-tradable sectors, overvalued currency, poor agricultural exports, segmented labor markets, poor educational progress.
  - Costa Rica’s rising inequality attributed to a large informal sector, widening wage gaps, and a large unskilled labor force stemming from 1980s crisis effects on schooling.

### IV. What determines income inequality in Latin America? (analysis summary)

- Simple correlations (LA sample):
  - Negative correlation between changes in total tax revenue (percent of GDP) and changes in the Gini coefficient: increases in tax revenue associated with decreases in income inequality.
  - Negative correlation between changes in direct tax revenue (percent of GDP) and changes in the Gini coefficient.
  - Negative correlation between changes in property tax revenue (percent of GDP) and changes in the Gini coefficient.
  - Negative correlation between changes in education spending (percent of GDP) and changes in the Gini coefficient.
  - Interpretation caveats: correlations do not indicate causation; variables chosen reflect data availability and policy relevance.
- Econometric panel results (44 emerging and developing countries, 1990–2010, country and time fixed effects):
  - Policies explain more than half of the decline in income inequality in Latin America (per the paper’s econometric estimates).
  - Higher education spending is estimated to explain almost one-fourth of the total decline.
  - Higher FDI and higher tax revenue are also important contributors.
  - Appreciating exchange rates have a small dampening (inequality-increasing) effect.
  - The Kuznets curve holds in the sample: growth has been conducive to declining income inequality, but the quantitative impact is limited.

### V. Policy-relevant implications highlighted in the study
- Emphasize the role of well-designed policies in reducing inequality:
  - Targeted social transfers and increases in education spending can substantively contribute to inequality reductions.
  - Strengthening tax revenue capacity—especially through progressive direct taxes and property taxes—can be associated with reduced inequality and can finance redistributive spending.
- Address structural impediments:
  - Continue educational upgrading and expand access to reduce skill premia and support inclusive growth.
  - Improve infrastructure and basic services access to reduce within-country disparities.
  - Monitor exchange rate policies and external factors that can have distributional consequences.

*Italic source attribution: IMF working paper content as provided in the supplied document.*

### Appendix for the country list and data description). Formulated by Simon Kuznets in the

### _wp14124 - Appendix for the country list and data description). Formulated by Simon Kuznets in the

### Kuznets hypothesis and background
- The Kuznet’s hypothesis postulates an inverted U-shaped relationship between GDP per capita and the Gini coefficient: economic growth is associated with rising income inequality up to a turning point, after which further growth is associated with declining inequality.
- Mechanism described: labor moves from lower productive agricultural sectors to higher productive industrial sectors where average income is higher and wages are less uniform, initially raising inequality; later social-welfare policies and transfer payments reduce the urban-rural income gap.
- The hypothesis is controversial: some studies support a Kuznets curve (e.g., Barro (2000, 2008); Acemoglu and Robinson (2002)); others reject it when controlling for country-specific effects (e.g., Deininger and Squire, 1998; Higgins and Williamson, 1999; Savvides and Stengos, 2000). Criticisms of negative findings often point to inconsistent income inequality data.

### Empirical specification and estimation approach
- Variables are demeaned using country-specific means (country fixed effects) to focus on within-country changes rather than cross-country levels.
- Time dummies are included to capture common global shocks (business cycles, external conditions, growth spurts).
- Estimated model (all variables in natural logarithm):
  - Ln(Gini_it) = β1 ln(GDPPC_it) + β2 [ln(GDPPC_it)]^2 + T_t + C_i + ɛ_it
  - Gini_it = Gini coefficient of country i at time t.
  - GDPPC = real GDP per capita (PPP).
  - Estimation: ordinary least squares with heteroskedasticity-consistent standard errors.
- Rationale: focusing on within-country variation reduces omitted variable bias and captures changes over time within countries.

### Kuznets specification results (Table 3)
- Estimated coefficients and statistics (dependent variable: natural logarithm of Gini; all explanatory variables in natural logarithm):
  - GDP per capita: 0.45 (0.06)** 
  - Squared GDP per capita: -0.03 (0.00)** 
  - Observations: 910
  - Countries: 44
  - Time period: 1990-2010
  - Adjusted R-square: 0.94
  - Time and country dummies: √
- Interpretation:
  - Signs satisfy β1 > 0 and β2 < 0 and are statistically significant, supporting the presence of a Kuznets curve in this specification.
  - However, the Kuznets curve does not explain the bulk of variations in inequality across countries or over time.
  - The analysis suggests that only one eight of the decline in Latin America’s income inequality in the last decade can be associated with recent strong growth momentum.

### Policies to the rescue — variables considered
- Panel econometric analysis with time and fixed effects for a sample of 38 emerging and developing economies over 2001-2010.
- Policies considered (selected for data availability and recent large changes):
  - Government spending on education (expect negative association with Gini; spending as share of GDP).
  - Tax revenue (higher tax revenue expected to be associated with lower inequality if progressive and if it funds targeted transfers/education).
  - Foreign Direct Investment (FDI) (sign ambiguous apriori: may lower inequality when abundant unskilled labor exists; may raise inequality via skill-biased technological change).
  - Exchange rate policies (depreciations expected to lower inequality by shifting production toward tradable, unskilled-labor-intensive sectors).
- Estimated model (levels used for easier interpretation):
  - Gini_it = β1 FDI_it + β2 educ + β3 tax + β4 reer + T_t + C_i + ɛ_it
  - FDI = foreign direct investment (share of GDP)
  - educ = education spending (share of GDP)
  - tax = tax revenue (share of GDP)
  - reer = real effective exchange rate
  - Time and country dummies included; country dummies capture institutional characteristics.

### Income Inequality Panel Regression results (Table 4)
- Estimated coefficients and statistics (dependent variable: Gini coefficient; standard errors in parentheses; ** denotes significance at the 1 percent level):
  - Real effective exchange rate: 0.02 (0.06)** 
  - Tax revenue (in percent of GDP): -0.08 (-0.04)** 
  - Education spending (in percent of GDP): -0.72 (-0.23)** 
  - FDI (in percent of GDP): -0.12 (0.06)** 
  - Observations: 274
  - Countries: 38
  - Time period: 2001-10
  - Adjusted R-square: 0.98
  - Time and country dummies: √
- Main empirical findings from the panel:
  - Education spending, taxation, depreciating exchange rates (expected negative sign), and FDI are associated with lower inequality; FDI is found associated with lower inequality in this sample.
  - Based on the estimated model and average total changes over the last decade, the four policies explain more than half of the recent decline in income inequality in Latin America.
  - Higher education spending explains almost a fourth of the overall decline alone (i.e., 0.8 out of the 3 Gini points).
  - Higher FDI and higher tax revenue jointly explain more than a forth of the total decline in the Gini coefficient over the last decade.
  - Exchange rate policies have been associated with higher inequality in the region because currencies have appreciated on average in LA over the last decade.
  - Common external factors (proxied by time dummies) explain the remaining change in LA’s Gini coefficient over the last decade.

### Caveats, limitations, and suggested extensions
- Limitations emphasized:
  - Inherent limitations of the data set and cross-country regression analysis, including inability to establish causality with full confidence.
  - The list of policies is not exhaustive; excluded variables due to data limitations include social/cash transfers, health spending, quality/types of education, infrastructure spending, and access to basic services.
  - The analysis uses the aggregate Gini coefficient; other dependent variables could be informative (e.g., disaggregated income sources, 90/10 income ratio, wage dispersion, labor income share).
  - Time dummies are a crude proxy for non-policy external factors.
- The authors note country- and time-specific heterogeneity: different policies may have different effects in different countries and times.

### Conclusions and policy implications
- Despite recent improvements, Latin America remains the most unequal region in the world.
- Empirical panel analysis indicates well-designed policies can explain more than half of the decline in income inequality (averaging around 3 Gini points over the last decade), with education spending explaining almost one Gini point.
- Stronger FDI and tax revenues were also important contributors to the decline; growth dynamics (Kuznets channel) were less pivotal.
- Specific policy implications highlighted:
  - Improving access of low-income families to education is an efficient tool to boost equality of opportunity and lower income inequality over the long run; strengthening access to quality education is pivotal given relatively high spending but poor outcomes in LA.
  - Stronger FDI can help lower inequality; attracting FDI depends on strong economic fundamentals.
  - Raising tax revenue (noting tax revenue was 20 percent in LA versus 34 percent in OECD countries in 2012) could be associated with declining inequality by financing redistributive policies; composition matters (income and profit taxes account for one-fourth of total tax revenue in LA versus 35 percent in OECD countries).
  - Policy suggestions include making income tax systems more progressive, addressing tax evasion risks, bringing more informal operators into personal income tax, and more fully utilizing property taxes (only Colombia and Uruguay collect more than 1 percent of GDP through recurrent property taxes).
- Final caution: cross-country regression results should be interpreted carefully; causality is difficult to establish and policy effects vary by country and time.

*Source: Authors' calculations.*

### REFERENCES

### _wp14124 - REFERENCES

### References (selected citations)
- Acemoglu, D., and J.A. Robinson, 2002, “The Political Economy of the Kuznets Curve,” Review of Development Economics, Vol. 6(2), pp. 183–203.  
- Azevedo, J. P., M. E. Dávalos, C. Diaz-Bonilla, B. Atuesta, and R. A. Castañeda, 2013, “Fifteen Years of Inequality in Latin America: How Have Labor Markets Helped?” Policy Research Working Paper 6384 (Washington: The World Bank).  
- Barro, R. J., 2008, “Inequality and Growth Revisited,” ADB Working Paper on Regional Economic Integration, No. 11 (Manila: Asian Development Bank).  
- Barro, R. J., 2000, “Inequality and Growth in a Panel of Countries,” Journal of Economic Growth, Vol. 5, pp. 5–32.  
- Berg, A., and J. Ostry, 2011, “Inequality and Unsustainable Growth: Two Sides of the Same Coin,” IMF Staff Position Note 11/08 (Washington: International Monetary Fund).  
- Bucheli, M., N. Lustig, M. Rossi, and F. Amábile, 2012, “Social Spending, Taxes and Income Redistribution in Uruguay,” ECINEQ Working Papers 263 (Verona: Society for the Study of Economic Inequality).  
- CEDLAS (Universidad Nacional de La Plata) and The World Bank, SEDLAC: Socio-Economic Database for Latin America and the Caribbean, available at: http://sedlac.econo.unlp.edu.ar/eng/  
- Coble, D., and N. Magud, 2010, “A Note on Terms of Trade Shocks and the Wage Gap,” IMF Working Paper 10/241 (Washington: International Monetary Fund).  
- Cornia, G., 2012, “Inequality Trends and their Determinants,” UNU-WIDER Working Paper 2012/09 (Helsinki: United Nations University).  
- Cruces, G., C. Garcia-Domenech, and L. Gasparini, 2011, “Inequality in Education. Evidence for Latin America,” UNU-WIDER Working Paper No. 2011/93 09 (Helsinki: United Nations University).  
- Deininger, K., and L. Squire, 1998, “New Ways of Looking at Old Issues: Inequality and Growth,” Journal of Development Economics, Elsevier, Vol. 57(2), pp. 259–87.  
- Deaton, A., 2010, “Price Indexes, Inequality, and the Measurement of World Poverty,” American Economic Review, Vol. 100 (1), pp. 5-34.  
- Deaton, A. and A. Heston, 2010, “Understanding PPPs and PPP-based National Accounts,” American Economic Journal: Macroeconomics, Vol. 2(4), pp. 1-35.  
- Deaton, A., 2013, Reshaping the World, in Measuring the Real Size of the World Economy: The Framework, Methodology, and Results of the International Comparison Program (ICP), (World Bank: May).  
- Fields, G., 2001, Distribution and Development. A New Look at the Developing World (Cambridge: MIT Press).  
- Gasparini, L., and G. Cruces, 2010, “A Distribution in Motion: The Case of Argentina,” in Declining Inequality in Latin America: A Decade of Progress? ed. by López Calva and N. Lustig, Chapter 5 (Brookings Institution and UNDP).  
- Gigliarano, C., and P. Muliere, 2012, “Measuring Income Polarization Using an Enlarged Middle Class,” ECINEQ 2012 – 27.  
- Goñi, E., H. López, and L. Servén, 2008, “Fiscal Redistribution and Income Inequality in Latin America,” Policy Research Working Paper 4487 (Washington: The World Bank).  
- Hidalgo, J.C., 2014, “Growth without Poverty Reduction: The Case of Costa Rica,” Economic Development Bulletin, No. 18 (Washington: The Cato Institute).  
- Higgins, M., J. Williamson, 1999, “Explaining Inequality the World Round: Cohort Size, Kuznets Curves, and Openness,” NBER Working Paper No. 7224 (Cambridge: National Bureau of Economic Research).  
- International Monetary Fund, 2014a, “Fiscal Policy and Income Inequality,” Policy Paper, (Washington: March).  
- International Monetary Fund, 2014b, Western Hemisphere—Regional Economic Outlook, (Washington: April).  
- Jaumotte, F., S. Lall, and C. Papageorgiou, 2008, “Rising Income Inequality: Technology, or Trade and Financial Globalization,” IMF Working Paper 08/185 (Washington: International Monetary Fund).  
- Klasen, S., T. Otter, and C. Villalobos, 2012, “The Dynamics of Inequality Change in a Highly Dualistic Economy: Honduras, 1991–2007,” Discussion paper No 215s (Göttingen: Ibero America Institute for Economic Research).  
- Lopez-Calva, L. F., and N. Lustig, 2010, Declining Inequality in Latin America: A Decade of Progress? (Washington: Brookings Institution and UNDP).  
- Milanovic, B., and S. Yitzhaki, 2002, “Decomposing World Income Distribution: Does The World Have A Middle Class?” The Review of Income and Wealth, Vol. 48(2), pp. 155–78.  
- Organisation for Economic Co-operation and Development, 2014, Revenue Statistics in Latin America 1990–2012, available at http://www.oecd.org/ctp/latin-america-tax-revenues-continue-to-rise-but-are-low-and-varied-among-countries-according-to-new-oecd-eclac-ciat-report.htm  
- Organisation for Economic Co-operation and Development, 2011, “An Overview of Growing Income Inequalities in OECD Countries: Main Findings,” in Divided We Stand: Why Inequality Keeps Rising, pp. 21–45 (Paris: OECD Publishing).  
- Ostry, J.D., A. Berg, and C. G. Tsangarides, 2014, “Redistribution, Inequality, and Growth,” Staff Discussion Note 14/2, (Washington: International Monetary Fund).  
- Peñalosa, Garcia Cecilia and Elsa Orgiazzi, 2013, “Factor Components of Inequality: A Cross-Country Study,” Review of Income and Wealth, International Association for Research in Income and Wealth, Vol. 59(4), pp. 689-727, December.  
- Ravallion, M., and Chen, S., 2003, “Measuring Pro-poor Growth,” Economic Letters, Vol. 78 (1), pp. 93–99.  
- Reynolds, L., 1996, “Some Sources of Income Inequality in Latin America,” Journal of Interamerican Studies and World Affairs, Vol. 38, No. 2/3, pp. 39–46.  
- Savvides, A., and T. Stengos, 2000, “Income Inequality and Economic Development: Evidence from the Threshold Regression Model,” Economics Letters, Vol. 69, pp. 207–12.  
- Soares, S., R. Guerreiro Osorio, F. Veras Soares, M. Medeiros, and E. Zepeda, 2009, “Conditional Cash Transfers in Brazil, Chile and Mexico: Impacts Upon Inequality,” Estudios Economicos, Vol. 0 (Special I), pp. 207–24. (El Colegio de México, Centro de Estudios Económicos).  
- Stampini, M., and L. Tornarolli, 2012, “The Growth of Conditional Cash Transfers in Latin America and the Caribbean: Did They Go Too Far?” Policy Note IDB-DB-185, (Washington, Inter-American Development Bank).  
- Stern, D.I., 2004, “The Rise and Fall of the Environmental Kuznets Curve,” World Development, Vol. 32, No. 8, pp. 1419–39.  
- Trujillo, L. and S. Villafañe (2011), “Factores asociados a la dinámica distributiva: una aproximación desde la descomposición por fuentes de ingresos en la Argentina reciente 2002-2010” (Factors Related to Distributive Dynamics: An Approach from Disaggregation by Sources of Income in Recent Argentina 2022-2010), in Novick, M. and S. Villafañe (eds.), Distribución del Ingreso. Enfoques y políticas públicas desde el Sur, (Distribution of Income. Approaches and Public Policies from the South); Ministry of Labor, Employment and Social Security, and UNDP, Buenos Aires.  
- Willem te Velde, D., 2003, “Foreign Direct Investment and Income Inequality in Latin America. Experiences and Policy Implications,” (United Kingdom: Overseas Development Institute).  
- World Bank, 2011, A Break with History: Fifteen Years of Inequality Reduction in Latin America (Washington: April).  
- World Bank, 2014, World Development Indicators, Washington.  
- World Health Organization, 2014, Health Equity Monitor, available at http://www.who.int/gho/health_equity/en/

### Annex — Figures and Captions (figure captions and source notes preserved)
- Figure A1. LA6 - Distribution of Per Capita Household Income by Highest and Lowest Decile (percent of total)  
  - Countries shown: Brazil, Chile, Colombia, Mexico, Peru, Uruguay.  
  - Sources: Centro de Estudios Distributivos Laborales y Sociales (CEDLAS); Socio-Economic Database for Latin America and the Caribbean (SEDLAC); The World Bank; and authors' calculations.  
- Figure A2. OTHER LA - Distribution of Per Capita Household Income by Highest and Lowest Decile (percent of total)  
  - Countries shown: Argentina, Bolivia, Ecuador, Paraguay, Venezuela.  
  - Sources: Centro de Estudios Distributivos Laborales y Sociales (CEDLAS); Socio-Economic Database for Latin America and the Caribbean (SEDLAC); The World Bank; and authors' calculations.  
- Figure A3. CAPDR - Distribution of Per Capita Household Income by Highest and Lowest Decile (percent of total)  
  - Countries shown: Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, Panama (data for Nicaragua and Panama noted elsewhere).  
  - Sources: Centro de Estudios Distributivos Laborales y Sociales (CEDLAS); Socio-Economic Database for Latin America and the Caribbean (SEDLAC); The World Bank; and authors' calculations.  
- Figure A4. LA6 Growth Incidence Curves During the Last Decade¹ (Rate of annual growth of household per capita income, in percent)  
  - Note: Growth incidence curves of household per capita income for Brazil, Chile, Colombia, Mexico, Peru and Uruguay (deciles). The changes are 2004-12 in Brazil, 2000-11 in Chile, 2001-12 in Colombia, 2000-12 in Mexico, 2003-12 in Peru, and 2000-12 in Uruguay.  
  - Sources: CEDLAS; SEDLAC; The World Bank; and authors' calculations.  
- Figure A5. Other LA - Growth Incidence Curves During the Last Decade¹ (Rate of annual growth of household per capita income, in percent)  
  - Note: Growth incidence curves for Argentina, Bolivia, Ecuador, Paraguay, and Venezuela (deciles). The changes are 2003-13 in Argentina, 2000-12 Bolivia, 2003-12 Ecuador, 2001-11 Paraguay, and 2000-06 Venezuela.  
  - Sources: CEDLAS; SEDLAC; The World Bank; and authors' calculations.  
- Figure A6. CAPDR - Growth Incidence Curves During the Last Decade¹ (Rate of annual growth, percent)  
  - Note: Growth incidence curves for Costa Rica, Dominican Republic, El Salvador, Guatemala, and Honduras. Data for Nicaragua and Panama were not available. The changes are 2001-12 in Costa Rica, 2000-11 in Dominican Republic, 2004-12 in El Salvador, 2000-11 in Guatemala, and 2001-11 in Honduras.  
  - Sources: CEDLAS; SEDLAC; The World Bank; and authors' calculations.  
- Figure A7. LA6: Distribution of Per Capita Household Income (by decile)  
  - Countries shown: Brazil, Chile, Colombia, Mexico, Peru, Uruguay.  
  - Sources: Socio-Economic Database for Latin America and the Caribbean; and authors' calculations.  
- Figure A8. OTHER LA: Distribution of Per Capita Household Income (by decile)  
  - Countries shown: Argentina, Bolivia, Ecuador, Paraguay, Venezuela, Panama.  
  - Sources: Socio-Economic Database for Latin America and the Caribbean; and authors' calculations.  
- Figure A9. CAPDR: Distribution of Per Capita Household Income (by decile)  
  - Countries shown: Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, Panama.  
  - Sources: Socio-Economic Database for Latin America and the Caribbean; and authors' calculations.

### Appendix: Data sources and sample countries
- Data Sources used in the analysis:
  - Gini index data: World Bank’s Povcal database.  
  - Tax revenue and education spending: World Development Indicators.  
  - Real effective exchange rate: IMF’s Information Notice System.  
  - Real GDP per capita (PPP) and FDI: IMF’s World Economic Outlook.  
- Sample of countries for Kuznets analysis:  
  Argentina, Bangladesh, Belarus, Bolivia, Brazil, Bulgaria, Chile, China, P.R., Colombia, Costa Rica, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ghana, Guatemala, Honduras, Hungary, Indonesia, Jamaica, Kazakhstan, Latvia, Lithuania, Malaysia, Mexico, Nicaragua, Nigeria, Panama, Paraguay, Peru, Philippines, Poland, Romania, South Africa, Sri Lanka, Thailand, Trinidad and Tobago, Turkey, Uganda, Ukraine, Uruguay, Venezuela, and Vietnam.  
- Sample of countries for regression analysis on policies:  
  Argentina, Bangladesh, Belarus, Bolivia, Brazil, Bulgaria, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ghana, Guatemala, Hungary, Indonesia, Jamaica, Kazakhstan, Latvia, Lithuania, Malaysia, Mexico, Nicaragua, Panama, Paraguay, Peru, Philippines, Poland, Romania, South Africa, Sri Lanka, Thailand, Turkey, Uganda, Ukraine, and Vietnam.

*Source: _wp14124 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14124.pdf_