## _wp08185

## Source details

**Canonical URL:** [_wp08185](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08185.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08185.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08185.pdf.json)

---

### I. Introduction and objectives
- Aim: examine the role of globalization (trade and financial) in affecting within-country distribution of income over the past two decades and to identify/estimate different channels through which globalization affects income distribution.
- Contributions:
  - Jointly analyze trade and financial globalization (not just trade).
  - Disaggregate subcomponents (e.g., exports of manufacturing vs. agriculture; portfolio debt and equity flows vs. FDI).
  - Use a new, more methodologically consistent cross-country inequality dataset (Povcal supplemented with LIS).
- Main empirical focus: changes in inequality (within-country variation) rather than levels, to avoid biases from mixing income and expenditure surveys and cross-country comparability issues.

### II. Data and cross-country trends
- Sample: 51 countries (20 advanced, 31 developing/emerging), period 1981–2003; unbalanced panel (observations only for years with actual survey data).
- Inequality:
  - Gini coefficients: inequality has risen in most regions over the past two decades, except for declines in some sub-Saharan African countries and in France among large advanced economies; emerging market trends are heterogeneous (e.g., sharply rising in China, little change in India, falling in Brazil).
  - Quintile shares: rising Gini largely driven by increasing shares of richer quintiles at the expense of middle quintiles; income share of poorest quintile changes little.
  - Absolute incomes: real per capita incomes have risen across virtually all regions and all quintiles in the recent globalization period.
- Globalization and technology measures:
  - World trade (imports + exports)/GDP rose from 36 percent to 55 percent since 1980; world trade in real terms grew five times since 1980.
  - Total cross-border financial assets rose from 58 percent of global GDP in 1990 to 131 percent in 2004.
  - Share of FDI in emerging market total liabilities rose from 17 percent in 1990 to 38 percent in 2004; portfolio equity liabilities rose from 2 percent to 11 percent over the same period.
  - ICT capital (share of ICT capital in total capital stock) has risen rapidly across all income levels over the past 20 years and is used as proxy for technological progress.

### III. Empirical specification (variables, controls, identification)
- Dependent variable: natural logarithm of the Gini coefficient (within-country demeaned changes).
- Trade globalization measures:
  - De jure: average tariff rate (TARIFF).
  - De facto: non-oil exports-to-GDP (X) and non-oil imports-to-GDP (M).
  - Sectoral export splits: agriculture, manufacturing, services.
- Financial globalization measures:
  - De jure: Chinn-Ito capital account openness index (KA_OPEN).
  - De facto: ratios of FDI, portfolio equity, and debt liabilities to GDP; inward FDI stock-to-GDP; foreign assets and liabilities A and L.
- Controls:
  - Technological progress: ICT capital share (K_ICT).
  - Education: average years of education (H), share with secondary/higher education (POP_SH).
  - Sectoral employment shares: E_AGR, E_IND.
  - Domestic financial development: private credit/GDP (CREDIT).
  - Real per capita GDP (Y) included in robustness checks but excluded from main reported specifications.
- Estimation: OLS with heteroskedasticity-consistent standard errors; country demeaned variables and time dummies; robustness checks include IV (lagged exports/FDI, export-weighted partner GDP, distance-weighted industrial countries’ FDI assets) and sensitivity tests (leave-one-out, dropping variables, including GDP per capita).

### IV. Key empirical findings
- Opposite effects of trade vs. financial globalization:
  - Trade liberalization and export growth are associated with lower income inequality.
  - Increased financial openness—particularly inward FDI stock (FDI/GDP)—is associated with higher inequality.
- Statistical significance (selected reported results):
  - Export-to-GDP ratio coefficients: -0.066, -0.057, -0.048, -0.056, -0.055 (t-statistics reported: (2.18)**, (2.56)**, (2.15)**, (2.41)**, (2.16)**).
  - 100 minus tariff rate coefficients: -0.002, -0.002, -0.003, -0.002, -0.003, -0.003 (t-statistics reported: (2.27)**, (2.52)**, (2.71)***, (2.61)***, (2.50)**, (2.98)***).
  - Ratio of inward FDI stock to GDP coefficients: 0.039, 0.040, 0.038, 0.035, 0.039, 0.029 (t-statistics reported: (2.92)***, (3.01)***, (3.06)***, (2.57)**, (2.96)***, (2.03)**).
  - Share of ICT in total capital stock coefficients: 0.032, 0.031, 0.027, 0.030, 0.033, 0.047 (t-statistics reported: (1.87)*, (1.98)**, (1.62), (2.03)**, (2.01)**, (3.16)***).
  - Credit to private sector (percent of GDP) coefficients: 0.049, 0.051, 0.049, 0.050, 0.042, 0.041 (t-statistics reported: (3.38)***, (3.49)***, (3.81)***, (3.54)***, (3.06)***, (2.58)***).
- Magnitudes (benchmark, one standard deviation changes from sample mean):
  - One standard deviation increase in export-to-GDP → reduces Gini by 3.4 percent.
  - One standard deviation decrease in tariffs → reduces Gini by 1.7 percent.
  - One standard deviation increase in inward FDI (stock/GDP) → increases Gini by 2.7 percent.
  - One standard deviation increase in technological progress (ICT capital share) → increases inequality by 1.7 percent.
  - One standard deviation increase in domestic financial deepening (private credit/GDP) → increases inequality by 2.6 percent.
  - One standard deviation reduction in agriculture employment share → reduces inequality by 3.3 percent.
  - One standard deviation increase in industry employment share → reduces inequality by 2.3 percent.
- Sectoral nuances:
  - Agricultural exports: coefficient -0.03 (t-statistic (2.49)**) — largest and statistically significant inequality-reducing coefficient among export sectors.
  - Manufacturing exports: coefficient -0.002 (t-statistic (0.10)) — not significant.
  - Service exports: coefficient -0.006 (t-statistic (0.38)) — not significant.
- Education:
  - Average years of education: coefficients reported negative (e.g., -0.219, -0.216, -0.182, -0.328, -0.359, -0.16) with some t-statistics indicating weak to moderate significance (e.g., (1.84)*, (1.91)* in some columns).
  - Population share with at least a secondary education: small positive coefficients in some specifications (e.g., 0.003, 0.004) with mixed significance.
- Sectoral employment and productivity:
  - Agriculture employment share: positive coefficients (e.g., 0.05, 0.05, 0.052, 0.058) with significance (t-statistics ~ (2.00)** to (2.49)**).
  - Industry employment share: negative coefficients (e.g., -0.102, -0.095, -0.098, -0.096) with significance (t-statistics ~ (2.50)** to (2.90)***).
  - Relative labor productivity of industry: 0.128 (t-statistic (3.03)***).
  - Relative labor productivity of agriculture: -0.037 (t-statistic (1.67)*).

### V. Quintile-level and heterogeneity results
- Quintile-level effects (selected):
  - Export-to-GDP ratio by quintile: Quintile 1: 0.439 (2.47)**; Quintile 2: 0.631 (3.52)***; Quintile 3: 0.690 (3.68)***; Quintile 4: 0.492 (2.58)**; Quintile 5: -2.220 (3.57)***. Test all coefficients equal to zero [p-value] = [0.02]**.
  - 100 minus tariff rate by quintile: Quintile 1: 0.021 (2.16)**; Quintile 2: 0.020 (2.04)**; Quintile 3: 0.017 (1.67)*; Quintile 4: 0.013 (1.32); Quintile 5: -0.070 (2.12)**. Test all coefficients equal to zero [p-value] = [0.28].
  - Ratio of inward FDI stock to GDP by quintile: Quintile 1: -0.400 (3.91)***; Quintile 2: -0.385 (3.74)***; Quintile 3: -0.326 (3.02)***; Quintile 4: -0.163 (1.48); Quintile 5: 1.241 (3.47)***. Test all coefficients equal to zero [p-value] = [0.00]***.
  - Credit to private sector (percent of GDP) by quintile: Quintile 1: -0.373 (3.30)***; Quintile 2: -0.625 (5.47)***; Quintile 3: -0.709 (5.94)***; Quintile 4: -0.437 (3.59)***; Quintile 5: 2.136 (5.39)***. Test all coefficients equal to zero [p-value] = [0.00]***.
- Regional heterogeneity (selected):
  - Different regional technology effect: Share of ICT in total capital stock * dummy for developing Asia: 0.033 (1.99)**; Share of ICT in total capital stock * dummy for Latin America and the Caribbean: -0.028 (1.91)*.
  - Advanced-economy interactions: Share of imports from developing economies * dummy for advanced economies: -0.104 (2.20)**; Ratio of inward debt stock to GDP * dummy for advanced economies: -0.083 (2.65)***; Ratio of outward FDI stock to GDP * dummy for advanced economies: 0.069 (2.68)***.
- IV diagnostics and model fit:
  - Anderson Test: 188.1, p-value [0.00].
  - Hansen Test: 2.01, p-value [0.37].
  - Observations across specifications: 279–288 (varies by column); Adjusted R-squared (within) reported around 0.27–0.32.

### VI. Decomposition of observed change in inequality (contributions)
- Average annual change in Gini (early 1980s onward): 0.45 percent increase per year (average).
- Contributions to the 0.45 percent annual average increase in Gini:
  - Technological progress alone explains nearly 0.35 percent per year of the 0.45 percent annual increase.
  - Globalization and financial deepening together contributed another 0.1 percent per year.
  - Increased access to education and a shift in employment away from agriculture together contributed 0.1 percent per year toward a reduction in the Gini.
- Net effect summary:
  - Technology is the dominant disequalizing force.
  - Globalization has a small net adverse impact because trade globalization tends to reduce inequality while financial globalization (FDI in particular) tends to increase it.

### VII. Mechanisms and interpretation
- Channels identified:
  - Technological progress: increases the premium on skills, substitutes away low-skill inputs, and raises returns to capital—exacerbating income dispersion.
  - FDI and financial openness: tend to increase returns to skills and capital; FDI often locates in higher-skill, technology-intensive sectors, increasing inequality; outward FDI in developed countries can reduce lower-skill employment.
  - Trade liberalization: agricultural export growth and tariff reduction can raise incomes for lower-income groups and reduce inequality; imports from developing countries can lower prices for poorer consumers in advanced economies.
  - Domestic financial deepening: can increase inequality if access is uneven and benefits accrue to those with collateral.
  - Education expansion mitigates disequalizing effects by increasing the share of the population able to benefit from skill-biased change.

### VIII. Policy implications and recommendations
- Address technological-driven inequality:
  - Expand access to education and raise average years of schooling to allow broader participation in skill-biased technological gains.
  - Policies to increase skill acquisition and reduce skills gaps can counteract disequalizing effects of technology and FDI.
- Manage financial globalization:
  - Strengthen domestic institutions and financial inclusion to ensure financial deepening benefits broader population rather than disproportionately favoring the rich.
  - Monitor sectoral composition of FDI and promote complementarities with local skill development.
- Leverage trade:
  - Support exports in sectors that benefit lower-income groups (e.g., agriculture in many developing countries) while designing complementary policies to help displaced workers transition to higher-productivity sectors.
- Complementary redistributive and social policies:
  - Use gains from globalization and growth to finance education and policies that expand economic opportunity and limit backlash against liberalization.

### IX. Data classification and sample (appendix highlights)
- Income group definitions:
  - Low income: $875 or less
  - Lower-middle income: $876–$3,465
  - Upper-middle income: $3,466–$10,725
  - High income: $10,726 or more
  - Taiwan Province of China is included in the high-income group.
- Countries used in estimation (sample of 51 countries):
  - Advanced economies (20): Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Israel, Italy, Japan, Korea, the Netherlands, Norway, Singapore, Spain, Sweden, the United Kingdom, and the United States.
  - Developing economies (31): Argentina, Bangladesh, Bolivia, Brazil, Chile, China, Costa Rica, Ecuador, Egypt, El Salvador, Ghana, Guatemala, Honduras, India, Indonesia, the Islamic Republic of Iran, Kenya, Malaysia, Mexico, Pakistan, Panama, Paraguay, Peru, the Philippines, Sri Lanka, Thailand, Turkey, Uganda, Uruguay, Venezuela, and Zambia.

*Source: IMF Working Paper section 7 — "Decomposition of the Change in Income Inequality" (content as provided).*

### Appendix I. Variable Definitions and Data Sources................................................................17

### _wp08185 - Appendix I. Variable Definitions and Data Sources................................................................17

### Appendix I
- Appendix I. Variable Definitions and Data Sources................................................................17

### Appendix II
- Appendix II. Income Country Groups and Estimation Sample ...............................................20

### References
- References................................................................................................................................33

### Tables
- 1.   Income Inequality Panel Regressions..............................................................21
- 2.   Quintile Income Shares Regressions..............................................................22
- 3.   Income Inequality Panel Regressions (Regional Heterogeneity).............................23

### Figures
- 1.   Income Inequality Within Income Country Groups and Selected Countries................24
- 2.   Income Shares Within Income Country Groups..................................................26
- 3.   Income by Quintile in Selected Regions.........................................................27
- 4.   Trade Liberalization Within Income Country Groups..........................................28
- 5.   Financial Liberalization Within Income Country Groups......................................29
- 6.   Technological Development Within Income Country Groups................................30

*Source: _wp08185 - Appendix I. Variable Definitions and Data Sources................................................................17*

### 7.   Decomposition of the Change in Income Inequality...........................................31

### 7.   Decomposition of the Change in Income Inequality

### I. Introduction and objectives
- Aim: examine the role of globalization (trade and financial) in affecting within-country distribution of income over the past two decades and to identify/estimate different channels through which globalization affects income distribution.
- Contributions:
  - Jointly analyze trade and financial globalization (not just trade).
  - Disaggregate subcomponents (e.g., exports of manufacturing vs. agriculture; portfolio debt and equity flows vs. FDI).
  - Use a new, more methodologically consistent cross-country inequality dataset (Povcal supplemented with LIS).
- Main empirical focus: changes in inequality (within-country variation) rather than levels, to avoid biases from mixing income and expenditure surveys and cross-country comparability issues.

### II. Data and cross-country trends
- Inequality:
  - Gini coefficients: inequality has risen in most regions over the past two decades, except for declines in some sub-Saharan African countries and in France among large advanced economies; emerging market trends are heterogeneous (e.g., sharply rising in China, little change in India, falling in Brazil).
  - Quintile shares: rising Gini largely driven by increasing shares of richer quintiles at the expense of middle quintiles; income share of poorest quintile changes little.
  - Absolute incomes: real per capita incomes have risen across virtually all regions and all quintiles in the recent globalization period.
- Globalization and technology measures:
  - World trade (imports + exports)/GDP rose from 36 percent to 55 percent since 1980; world trade in real terms grew five times since 1980.
  - Total cross-border financial assets rose from 58 percent of global GDP in 1990 to 131 percent in 2004.
  - Share of FDI in emerging market total liabilities rose from 17 percent in 1990 to 38 percent in 2004; portfolio equity liabilities rose from 2 percent to 11 percent over the same period.
  - ICT capital (share of ICT capital in total capital stock) has risen rapidly across all income levels over the past 20 years and is used as proxy for technological progress.

### III. Empirical specification (sample, variables, identification)
- Sample: 51 countries (20 advanced, 31 developing/emerging), period 1981–2003; unbalanced panel (observations only for years with actual survey data).
- Dependent variable: logarithm of the Gini coefficient (within-country demeaned changes).
- Trade globalization measures:
  - De jure: average tariff rate (TARIFF).
  - De facto: non-oil exports-to-GDP (X) and non-oil imports-to-GDP (M).
  - Sectoral export splits: agriculture, manufacturing, services.
- Financial globalization measures:
  - De jure: Chinn-Ito capital account openness index (KA_OPEN).
  - De facto: ratios of FDI, portfolio equity, and debt liabilities to GDP; inward FDI stock-to-GDP; foreign assets and liabilities A and L.
- Controls:
  - Technological progress: ICT capital share (K_ICT).
  - Education: average years of education (H), share with secondary/higher education (POP_SH).
  - Sectoral employment shares: E_AGR, E_IND (agriculture, industry).
  - Domestic financial development: private credit/GDP (CREDIT).
  - Real per capita GDP (Y) included in robustness checks but excluded from main reported specifications to capture full effects.
- Estimation: OLS with heteroskedasticity-consistent standard errors; country demeaned variables and time dummies; robustness checks include IV (lagged exports/FDI, export-weighted partner GDP, distance-weighted industrial countries’ FDI assets) and sensitivity tests (leave-one-out, dropping variables, including GDP per capita).

### IV. Key empirical findings
- Opposite effects of trade vs. financial globalization:
  - Trade liberalization and export growth are associated with lower income inequality.
  - Increased financial openness—particularly inward FDI stock (FDI/GDP)—is associated with higher inequality.
- Statistical significance (benchmark results; significance levels noted in text):
  - Exports-to-GDP reduction in inequality, tariff reductions reduce inequality, inward FDI increases inequality (significant at the 5 percent level).
- Magnitudes (benchmark model, one standard deviation changes from sample mean):
  - One standard deviation increase in export-to-GDP → reduces Gini by 3.4 percent.
  - One standard deviation decrease in tariffs → reduces Gini by 1.7 percent.
  - One standard deviation increase in inward FDI (stock/GDP) → increases Gini by 2.7 percent.
  - One standard deviation increase in technological progress (ICT capital share) → increases inequality by 1.7 percent.
  - One standard deviation increase in domestic financial deepening (private credit/GDP) → increases inequality by 2.6 percent.
  - One standard deviation reduction in agriculture employment share → reduces inequality by 3.3 percent.
  - One standard deviation increase in industry employment share → reduces inequality by 2.3 percent.
- Sectoral nuances:
  - Agricultural exports have the largest (and statistically significant) inequality-reducing coefficient among export sectors; manufacturing and services coefficients not statistically different from agriculture but smaller.
  - FDI tends to flow to higher-skill, technology-intensive sectors, increasing the premium on skills and raising inequality.
- Education:
  - Higher average years of education reduce inequality (facilitates broader benefit from technology and FDI).
  - For a given average education level, a larger share with secondary/higher education (greater dispersion) tends to increase inequality (sometimes imprecisely estimated).
- Sectoral employment and productivity:
  - Shift away from agriculture to industry improves distribution by raising incomes of low-earning groups and increasing agricultural productivity; confirmed when replacing employment shares with sectoral productivity measures.

### V. Robustness and heterogeneity
- Endogeneity addressed via instrumental variables (lagged export/FDI, export-weighted partner GDP, distance-weighted FDI supply); instruments validated by Anderson and Hansen tests; results robust to IV (see column 6 of Table 1).
- Other robustness checks: dropping countries one at a time, dropping single variables, including GDP per capita; results broadly robust.
- Quintile-level estimations (income shares) confirm main results:
  - Export growth raises income shares of bottom four quintiles and reduces share of richest quintile.
  - Tariff reduction benefits mostly bottom three quintiles; decreases top quintile share.
  - Financial globalization, technological progress, and financial deepening mainly benefit the richest 20 percent.
- Heterogeneous effects:
  - Advanced vs. developing countries: differentiated effects—FDI assets increase inequality in advanced countries; debt liabilities-to-GDP and share of imports from developing countries contribute to reducing inequality in advanced countries.
  - Regional differences among developing countries:
    - Technological progress has a stronger disequalizing effect in developing Asia and weaker (insignificant) effect in Latin America—consistent with Asia’s larger share of technology-intensive manufacturing.

### VI. Decomposition of observed change in inequality (contributions)
- Average annual change in Gini (early 1980s onward): 0.45 percent increase per year (average).
- Contributions to the 0.45 percent annual average increase in Gini:
  - Technological progress alone explains nearly 0.35 percent per year of the 0.45 percent annual increase.
  - Globalization and financial deepening together contributed another 0.1 percent per year.
  - Increased access to education and a shift in employment away from agriculture together contributed 0.1 percent per year toward a reduction in the Gini.
- Net effect: small adverse net impact of globalization on inequality because trade globalization tends to reduce inequality while financial globalization (FDI in particular) tends to increase it; technological change is the dominant disequalizing force.

### VII. Mechanisms and interpretation
- Technological progress and financial openness mainly operate by increasing the premium on higher skills, raising returns to capital, and thereby exacerbating income dispersion.
- Trade liberalization and export growth (particularly agricultural exports) can reduce inequality by increasing incomes of those employed in lower-income sectors (agriculture) and benefiting poorer consumers via tariff reductions.
- Domestic financial deepening can increase inequality if institutional quality is weak and benefits accrue disproportionately to those with collateral/income.
- Education expansion (average years) mitigates the disequalizing effects by enabling broader participation in skill-intensive activities.

### VIII. Policy implications (implied by findings)
- Address technological-driven inequality:
  - Expand access to education and raise average years of schooling to allow broader participation in skill-biased technological gains.
  - Policies to increase skill acquisition and reduce skills gaps can counteract disequalizing effects of technology and FDI.
- Manage financial globalization:
  - Strengthen domestic institutions and financial inclusion to ensure financial deepening benefits broader population rather than disproportionately favoring the rich.
  - Monitor sectoral composition of FDI and promote complementarities with local skill development.
- Leverage trade:
  - Support exports in sectors that benefit lower-income groups (e.g., agriculture in many developing countries) while designing complementary policies to help displaced workers transition to higher-productivity sectors.
- Complementary redistributive and social policies:
  - Use gains from globalization and growth to finance education and policies that expand economic opportunity and limit backlash against liberalization.

*Source: IMF Working Paper section 7 — "Decomposition of the Change in Income Inequality" (content as provided).*

### 0.3 percent a year on average, technology has been the main driving force while globalization

### _wp08185 - 0.3 percent a year on average, technology has been the main driving force while globalization

### Overview finding
- "0.3 percent a year on average, technology has been the main driving force while globalization has in fact provided a small counterweight by tending to reduce inequality."

### Channels through which globalization and technology affect income distribution
- Trade (goods and services)
  - Opening trade in agriculture raises incomes of those dependent on agriculture in developing countries by increasing agricultural income and shifting underemployed agricultural workers to manufacturing or services.
  - Reduction in tariffs is associated with a reduction in inequality by allowing cheaper imports to substitute for more expensive domestically produced consumption and intermediate inputs (consistent with Stolper-Samuelson).
  - In developed countries, imports from developing countries reduce inequality by increasing effective income of poorer segments through lower-priced noncompeting imports and by substituting lower-paying low-end manufacturing jobs with higher-paying service sector jobs (retail and consumer finance).
  - Imports from other advanced economies do not have the same beneficial impact on inequality because higher-end imports affect higher-paying domestic employment and form a small component of poorer households’ consumption.
- Foreign Direct Investment (FDI) and financial globalization
  - Inward FDI is associated with increases in income inequality in both developed and developing countries.
  - Outward FDI in developed countries is additionally associated with rising inequality by reducing employment opportunities in relatively lower skill sectors.
  - FDI tends to locate in higher skill and higher technology sectors in host countries, increasing returns to those with higher skills and education.
  - In developing countries, a large share of FDI goes into low-end manufacturing and natural resource sectors, increasing employment for those with higher skills than agricultural workers, thereby raising relative demand for higher-skilled labor.
- Technology
  - Technological progress increases the premium on skills and substitutes away low-skill inputs, increasing the relative demand for higher skills and exacerbating income inequality (Birdsall, 2007).
  - The adverse effect of technology on income inequality is more evident in Asia than in Latin America because manufacturing is a greater share of the economy in many Asian countries and new technology affects a larger share of the population there.
- Financial deepening
  - Overall financial deepening is associated with higher growth, but a disproportionately larger share of increased finance goes to those who already have higher incomes and assets usable as collateral.
  - Broader access to finance, not just aggregate financial deepening, matters for sharing benefits more widely.

### Empirical synthesis and contrasts
- Both globalization and technology tend to increase returns to skills and education in developed and developing countries.
- While incomes have increased across all segments of the population in virtually all regions and countries, incomes of the already well off have increased disproportionately more during the recent era of globalization.
- The contribution of increased globalization to inequality has in general been relatively minor because two offsetting effects operate:
  - Increased trade tends to reduce income inequality.
  - Foreign direct investment tends to exacerbate income inequality.
- The analysis supports Bernanke's notion that technological change played a major role in increasing inequality, while globalization played a smaller role (see Bernanke (2008)), and runs counter to arguments attributing inequality increases primarily to international trade.

### Policy implications and recommendations
- Broader access to education and training
  - Greater access to education and training can increase the share of the population able to take advantage of opportunities from globalization and technology.
  - Returns to investment in education for all countries have risen in the recent era of globalization.
- Broader access to finance
  - Broader access to finance would benefit poorer segments of the population; it is not just aggregate financial deepening but how broadly it is available that matters.
- Areas for further empirical research
  - Examine the impact of government policies, and fiscal policies in particular, on income distribution—limited currently by absence of a comprehensive cross-country database of government redistribution policies.
  - Examine the impact of FDI by sector, given distributional consequences are likely to vary by sector.
  - Examine the impact of technological progress by sector and type of technology—limited by availability of comprehensive data across countries and over time, so likely feasible only for single countries or small groups of countries.

### Conclusion (V. CONCLUSION)
- Estimates using a new and more reliable dataset on inequality and detailed measures of globalization suggest the observed rise in inequality across both developed and developing countries over the past two decades is largely attributable to the impact of technological change.
- Both globalization and technological progress increase the relative demand for skills and education.
- Broader access to education would allow a greater segment of the population to benefit from globalization and technological change and could accelerate poverty reduction.

*Source: _wp08185 - 0.3 percent a year on average, technology has been the main driving force while globalization (IMF PDF chapter).*

### APPENDIX II. INCOME COUNTRY GROUPS AND ESTIMATION SAMPLE

### APPENDIX II. INCOME COUNTRY GROUPS AND ESTIMATION SAMPLE

### Income country groups
- Income group definitions:
  - Low income: $875 or less
  - Lower-middle income: $876–$3,465
  - Upper-middle income: $3,466–$10,725
  - High income: $10,726 or more
- Taiwan Province of China is included in the high-income group.

### Countries used in estimation (sample of 51 countries)
- Advanced economies (20): Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Israel, Italy, Japan, Korea, the Netherlands, Norway, Singapore, Spain, Sweden, the United Kingdom, and the United States.
- Developing economies (31): Argentina, Bangladesh, Bolivia, Brazil, Chile, China, Costa Rica, Ecuador, Egypt, El Salvador, Ghana, Guatemala, Honduras, India, Indonesia, the Islamic Republic of Iran, Kenya, Malaysia, Mexico, Pakistan, Panama, Paraguay, Peru, the Philippines, Sri Lanka, Thailand, Turkey, Uganda, Uruguay, Venezuela, and Zambia.

### Table 1: Income inequality panel regressions (dependent variable: natural logarithm of Gini) — key reported coefficients and statistics
- Trade globalization
  - Export-to-GDP ratio: -0.066 -0.057 -0.048 -0.056 -0.055 (2.18)** (2.56)** (2.15)** (2.41)** (2.16)**
  - Agricultural exports: -0.03 (2.49)**
  - Manufacturing exports: -0.002 (0.10)
  - Service exports: -0.006 (0.38)
  - Import-to-GDP ratio: 0.011 (0.38)
  - 100 minus tariff rate: -0.002 -0.002 -0.003 -0.002 -0.003 -0.003 (2.27)** (2.52)** (2.71)*** (2.61)*** (2.50)** (2.98)***
- Financial globalization
  - Ratio of inward FDI stock to GDP: 0.039 0.040 0.038 0.035 0.039 0.029 (2.92)*** (3.01)*** (3.06)*** (2.57)** (2.96)*** (2.03)**
  - Ratio of inward portfolio equity stock to GDP: -0.0001 (0.15)
  - Ratio of inward debt stock to GDP: 0.002 (0.12)
  - Ratio of outward FDI stock to GDP: 0.0005 (0.39)
  - Capital account openness index: -0.001 (0.17)
- Control variables (selected)
  - Share of ICT in total capital stock: 0.032 0.031 0.027 0.030 0.033 0.047 (1.87)* (1.98)** (1.62) (2.03)** (2.01)** (3.16)***
  - Credit to private sector (percent of GDP): 0.049 0.051 0.049 0.050 0.042 0.041 (3.38)*** (3.49)*** (3.81)*** (3.54)*** (3.06)*** (2.58)***
  - Population share with at least a secondary education: 0.003 0.003 0.002 0.004 0.004 0.002 (1.42) (1.47) (0.77) (1.82)* (2.08)** (0.84)
  - Average years of education: -0.219 -0.216 -0.182 -0.328 -0.359 -0.16 (1.17) (1.20) (1.00) (1.84)* (1.91)* (0.87)
  - Agriculture employment share: 0.05 0.05 0.052 0.058 (2.00)** (2.05)** (2.21)** (2.49)**
  - Industry employment share: -0.102 -0.095 -0.098 -0.096 (2.50)** (2.78)*** (2.26)** (2.90)***
  - Relative labor productivity of agriculture: -0.037 (1.67)*
  - Relative labor productivity of industry: 0.128 (3.03)***
- Sample and goodness-of-fit
  - Observations: 288 288 284 279 288 285
  - Adjusted R-squared (within): 0.29 0.30 0.31 0.32 0.27
- IV estimation diagnostics
  - Anderson Test: 188.1
    - p-value [0.00]
  - Hansen Test: 2.01
    - p-value [0.37]
- Notes: Heteroskedasticity-robust t-statistics in parentheses; * denotes significance at the 10 percent level, ** denotes significance at the 5 percent level, and *** denotes significance at the 1 percent level. All explanatory variables are in natural logarithm, except the tariff measure, the capital account openness index, and the population share with at least a secondary education. Left- and right-hand-side variables are demeaned using country-specific means and the equations include time dummies. FDI = foreign direct investment; ICT = information and communications technology.

### Table 2: Quintile income shares regressions (dependent variable: quintile income share) — selected coefficients and tests
- Export-to-GDP ratio by quintile:
  - Quintile 1: 0.439 (2.47)** 
  - Quintile 2: 0.631 (3.52)*** 
  - Quintile 3: 0.690 (3.68)*** 
  - Quintile 4: 0.492 (2.58)** 
  - Quintile 5: -2.220 (3.57)*** 
  - Test all coefficients equal to zero [p-value] = [0.02]**
- 100 minus tariff rate by quintile:
  - Quintile 1: 0.021 (2.16)** 
  - Quintile 2: 0.020 (2.04)** 
  - Quintile 3: 0.017 (1.67)* 
  - Quintile 4: 0.013 (1.32) 
  - Quintile 5: -0.070 (2.12)** 
  - Test all coefficients equal to zero [p-value] = [0.28]
- Ratio of inward FDI stock to GDP by quintile:
  - Quintile 1: -0.400 (3.91)*** 
  - Quintile 2: -0.385 (3.74)*** 
  - Quintile 3: -0.326 (3.02)*** 
  - Quintile 4: -0.163 (1.48) 
  - Quintile 5: 1.241 (3.47)*** 
  - Test all coefficients equal to zero [p-value] = [0.00]***
- Share of ICT in total capital stock by quintile:
  - Quintile 1: -0.177 (1.32)
  - Quintile 2: -0.223 (1.65)*
  - Quintile 3: -0.218 (1.54)
  - Quintile 4: -0.207 (1.44)
  - Quintile 5: 0.830 (1.77)*
  - Test all coefficients equal to zero [p-value] = [0.59]
- Credit to private sector (percent of GDP) by quintile:
  - Quintile 1: -0.373 (3.30)***
  - Quintile 2: -0.625 (5.47)***
  - Quintile 3: -0.709 (5.94)***
  - Quintile 4: -0.437 (3.59)***
  - Quintile 5: 2.136 (5.39)***
  - Test all coefficients equal to zero [p-value] = [0.00]***
- Education and employment shares (selected)
  - Population share with at least a secondary education: coefficients range -0.035 to 0.094, t-statistics reported; [p-value] = [0.14] for joint test
  - Average years of education: coefficients range 1.844 to -3.99; [p-value] = [0.80] for joint test
  - Agriculture employment share: coefficients -0.460 -0.789 -0.981 -0.568 2.777 with t-statistics (1.76)* (2.98)*** (3.55)*** (2.02)** (3.02)***; joint [p-value] = [0.00]***
  - Industry employment share: coefficients 1.081 0.866 0.603 0.084 -2.623 with t-statistics (3.07)*** (2.43)** (1.62) (0.22) (2.12)**; joint [p-value] = [0.09]*
- Sample and fit
  - Observations: 271 (for each quintile)
  - R-squared: 0.34, 0.36, 0.33, 0.18, 0.35 (by quintile)
- Notes: T-statistics in parentheses; * denotes significance at the 10 percent level, ** denotes significance at the 5 percent level, and *** denotes significance at the 1 percent level. All explanatory variables in natural logarithm, except the tariff measure and the population share with at least a secondary education. Left- and right-hand-side variables are de-meaned using country-specific means; equations include time dummies and are estimated jointly using the seemingly unrelated regressions estimator. FDI = foreign direct investment; ICT = information and communications technology.

### Table 3: Income inequality panel regressions (regional heterogeneity) — selected coefficients
- Common model coefficients (reported)
  - Export-to-GDP ratio: -0.063 -0.071 (2.23)** (3.17)***
  - 100 minus tariff rate: -0.002 -0.004 (2.24)** (3.53)***
  - Ratio of inward FDI stock to GDP: 0.031 0.041 (2.28)** (3.03)***
  - Share of ICT in total capital stock: 0.035 0.037 (2.12)** (2.11)**
  - Credit to private sector (percent of GDP): 0.058 0.041 (3.94)*** (3.29)***
  - Population share with at least a secondary education: 0.001 0.002 (0.35) (0.82)
  - Average years of education: -0.100 -0.124 (0.54) (0.65)
  - Agriculture employment share: 0.074 0.052 (2.59)** (2.31)**
  - Industry employment share: -0.090 -0.139 (2.23)** (3.96)***
- Additional variables for advanced economies
  - Share of imports from developing economies: 0.018 (0.57)
  - Share of imports from developing economies * dummy for advanced economies: -0.104 (2.20)**
  - Ratio of inward debt stock to GDP: 0.014 (0.78)
  - Ratio of inward debt stock to GDP * dummy for advanced economies: -0.083 (2.65)***
  - Ratio of outward FDI stock to GDP: 0.000 (0.31)
  - Ratio of outward FDI stock to GDP * dummy for advanced economies: 0.069 (2.68)***
- Different regional technology effect
  - Share of ICT in total capital stock * dummy for developing Asia: 0.033 (1.99)**
  - Share of ICT in total capital stock * dummy for Latin America and the Caribbean: -0.028 (1.91)*
- Observations: 282 282
- Adjusted R-squared: 0.32 0.35
- Notes: T-statistics in parentheses; * denotes significance at the 10 percent level, ** denotes significance at the 5 percent level, and *** denotes significance at the 1 percent level. All explanatory variables in natural logarithm, except the tariff measure and the population share with at least a secondary education. Left- and right-hand-side variables are de-meaned using country-specific means; equations include time dummies and are estimated jointly using the seemingly unrelated regressions estimator. FDI = foreign direct investment; ICT = information and communications technology.

### Figures and notes (selected)
- Figure 1: Shows averages of country Gini coefficients by income group and Gini coefficients in selected advanced and emerging market economies. Notes: Income country groups defined in the appendix. Trends after 2000 based on earnings data for full-time, year-round workers. Trends for pre-1992 Germany based on West Germany data.
- Figure 2: Income shares by quintile across income country groups. Notes: Income country groups are defined in the appendix.
- Figure 3: Income by quintile in selected regions; income/consumption share data applied to per capita real GDP levels from PWT 6.2 to calculate per capita income by quintile.
- Figure 4: Trade liberalization within income groups — "De Facto" trade openness (ratio of imports and exports to GDP) and "De Jure" trade openness (100-Tariff rate). Notes: Tariff rates are average of the effective rate and average un-weighted tariff rates.
- Figure 5: Financial liberalization — "De Facto" financial openness (ratio of cross-border assets and liabilities to GDP) and "De Jure" financial openness (capital account openness index). Notes: De jure index based on principal components from disaggregated capital and current account restriction measures.
- Figure 6: ICT capital (percent of total capital stock). ICT data from Jorgenson and Vu (2005).
- Figure 7: Decomposition of the change in income inequality into contribution of globalization, technology, and other factors. Notes: Contribution of each variable computed as annual change in the variable times the relevant regression coefficient.

*Source: APPENDIX II. INCOME COUNTRY GROUPS AND ESTIMATION SAMPLE (extracted from the supplied content).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08185.pdf_