## wp1854

## Source details

**Canonical URL:** [wp1854](https://www.imf.org/-/media/files/publications/wp/2018/wp1854.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1854.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1854.pdf.json)

---

### Correlations, collinearity, and index construction
- Pairwise correlations reported:
  - Between the two de facto flow variables (trade and FDI): 0.59.
  - Between the two de jure restriction variables (tariffs and capital account restrictions): 0.57.
  - Other reported pairwise values include 0.30 and 0.59 in descriptive comparisons.
- Country fixed effects analysis shows strong positive associations persist after netting out unobserved country-specific, time-invariant heterogeneity.
- Composite index construction and properties:
  - Composite index aggregates eight de facto and de jure measures via principal component analysis.
  - Index range: 0 (no globalization) to 100 (maximum globalization).
  - Median (mean) change from one period to the next: 2.1 (2.9) points.
  - Squared term of globalization is included to allow nonlinearity.
- Motivation: high correlations among sub-components motivate use of an overall composite index (no presumption about dominant underlying components).

### Data, measures, and identification
- Composite index components:
  - De facto: trade, FDI, portfolio investment, income payments to nonresidents.
  - De jure: import barriers, tariffs, taxes on trade, capital account restrictions.
- Dependent variables and sources:
  - Average per capita GDP growth (World Development Indicators; robustness with Penn World Tables).
  - Gini index of income inequality (baseline: SWIID; robustness: PovcalNet and All the Ginis (ATG)); correlation between SWIID Ginis and PovcalNet/ATG in the sample: p = .89.
  - Income growth by income deciles (Global Income and Consumption Project, GCIP).
- Controls and estimation setup:
  - Lagged ln GDP per capita (previous 5-year period); inequality regressions include squared logged GDP per capita.
  - Population growth rate, average life expectancy, average years of schooling, Polity index; all averaged over previous 5-year period.
  - Period fixed effects and country fixed effects used.
  - Baseline dynamic panel: y_it = β g_{i,t-1} + γ X_{i,t-1} + country FE + period FE + ε_it (variables averaged in 5-year periods).
- IV strategy:
  - Instrument = one-period-lagged, inverse-distance-weighted globalization scores of all other countries j ≠ i at time t-1 (spatial lag with temporal lag).
  - Distance measure: population-weighted distance between agglomerations (Mayer and Zignago 2011).
  - First-stage and exclusion restriction assumptions discussed; limitations noted (LATE, sensitivity to outliers).

### Main empirical findings — GDP/capita growth
- OLS fixed effects (controlling for country and year FE and lagged GDP per capita):
  - A one-point increase in globalization is associated with an increase in the 5-year growth rate by 0.3 percentage points (average annual effect 0.06 percentage points).
- Selected Table 2 coefficients (Growth – Main Results):
  - Column (1) OLS: 0.0033*** (0.0009)
  - Column (2) OLS: 0.0031*** (0.0009)
  - Column (3) IV: 0.0016 (0.0052)
  - Column (4) IV: 0.0018 (0.0048)
  - Column (5) OLS: 0.0101*** (0.0025)
  - Column (6) IV: 0.0470*** (0.0133)
  - Column (7) OLS: 0.0041*** (0.0012)
  - Column (8) IV: 0.0492** (0.0215)
- Nonlinear effects:
  - Economic Globalization² in Table 2:
    - Column (6): -0.0001*** (0.0000)
    - Column (8): -0.0002*** (0.0001)
  - Evidence of positive but diminishing marginal effects of globalization on growth (OLS and IV).
  - Marginal effect becomes statistically insignificant at the five percent level at a globalization score of about 77.
  - Approximately 14 percent of country-period observations in the sample surpass a globalization score of 77.
- Magnitude examples:
  - An average low income country with globalization score 41 would be expected to increase its total 5-year-period growth rate by about 2.2 percentage points when increasing globalization by one point.
  - Expected average annual growth effects reported:
    - Average LIC: 0.40 percentage points.
    - Average MIC: 0.36 percentage points.
  - Mean (median) period-to-period increase in globalization index: about three (two) points.
- Channels and Table 5 (Growth – Channels; coefficients multiplied by 100):
  - Economic Liberalization: OLS 0.447*** (0.170); IV 1.472*** (0.541)
  - Economic Liberalization²: -0.003** (0.001) in OLS
  - Economic Flows: OLS 0.883*** (0.233); IV 3.410*** (1.154)
  - Economic Flows²: -0.007*** (0.002); IV -0.024** (0.010)
  - Tariff Reductions: OLS 0.748** (0.322); IV 1.894** (0.853)
  - Capital Account Liberalization: OLS 0.640** (0.290); IV 2.109** (0.827)
  - Trade (% GDP): OLS 0.148*** (0.043); IV 0.190*** (0.069)
  - FDI stock (% GDP): OLS 0.042** (0.016); IV -0.066 (0.044)
- IV diagnostics:
  - First-stage coefficient on IV (α) = 0.66; t = 4.81; p < 0.001.
  - Kleibergen-Paap statistics reported; examples: K-P weak id. F-statistics reported across specifications (e.g., 23.171, 23.755, 11.858, 5.219; critical values cited: 16.38 for one endogenous regressor; 7.03 for two endogenous regressors).
  - In small-sample IV, weak instrument bias cannot be ruled out (example F = 5.2).

### Main empirical findings — inequality (Gini and distribution)
- Baseline Gini (Table 3) main coefficients:
  - Column (1) OLS: 0.104*** (0.039)
  - Column (2) OLS: 0.098*** (0.038)
  - Column (3) IV: 0.368** (0.160)
  - Column (4) IV: 0.359** (0.148)
  - Column (5) OLS: 0.043 (0.141)
  - Column (6) IV: -0.254 (0.336)
  - Column (7) OLS: 0.104* (0.061)
  - Column (8) IV: 0.343 (0.359)
- Magnitude interpretation:
  - A one-point increase in economic globalization leads to a rise in the Gini index of about one third of a point.
  - Given average change in globalization of about three points per period, this implies an economically substantial effect.
  - Blackburn (1989) equivalence: a one-point change in Gini ≈ lump-sum transfer of 2 percent of mean income from bottom half to upper half.
- Nonlinearity and heterogeneity:
  - OLS shows no strong evidence for significant nonlinearity; IV marginal effects plotted show consistently positive effects, statistically significant only for values larger than 60.
  - Inequality-increasing effect particularly strong in highly globalized, advanced economies; positive but weaker association for developing countries cannot be ruled out.
  - IV in developing-country subsample yields positive but statistically insignificant coefficient; IV may be weak in smaller sample.
- Decile-level distributional effects (Table 4):
  - Full Sample (Panel A) Economic Globalization coefficients:
    - Decile 1: 0.005 (0.010)
    - Decile 2: 0.008 (0.008)
    - Decile 3: 0.009 (0.008)
    - Decile 4: 0.010 (0.007)
    - Decile 5: 0.011 (0.007)
    - Decile 6: 0.011 (0.007)
    - Decile 7: 0.012* (0.007)
    - Decile 8: 0.012* (0.006)
    - Decile 9: 0.013** (0.006)
    - Decile 10: 0.016** (0.006)
    - Top 5%: 0.017** (0.008)
    - Top 1%: 0.017** (0.008)
  - Developing Countries (Panel B) Economic Globalization coefficients:
    - Decile 1: 0.056* (0.033)
    - Decile 2: 0.051* (0.027)
    - Decile 3: 0.049** (0.024)
    - Decile 4: 0.046** (0.023)
    - Decile 5: 0.044** (0.021)
    - Decile 6: 0.042** (0.020)
    - Decile 7: 0.040** (0.019)
    - Decile 8: 0.038** (0.017)
    - Decile 9: 0.036** (0.016)
    - Decile 10: 0.030** (0.013)
    - Top 5%: 0.039* (0.021)
    - Top 1%: 0.039* (0.021)
  - No evidence of absolute income losses for any decile in the analyses.
- Relative shares and top incomes (Table 10):
  - Using decile income shares, a one-point increase in globalization increases top 10% income share by about 0.33 percentage points and top 1% share by 0.24 percentage points.
  - Example 2SLS coefficients on income shares:
    - Decile 1: -0.034** (0.017)
    - Decile 10: 0.330*** (0.101)
    - Top 5%: 0.363*** (0.113)
    - Top 1%: 0.240** (0.097)

### Channels: de jure vs de facto and indicator-level evidence
- Decomposition findings:
  - Both de jure liberalization and de facto flows contribute to the positive growth effect; nonlinearity visible for both.
  - Both de jure and de facto measures contribute to increasing inequality (Table 6, columns 1–4).
- Individual indicator correlational evidence (OLS fixed effects):
  - Trade, FDI, capital account liberalization, and tariffs reduction each positively associated with growth.
  - Nonlinearity (diminishing marginal returns) significant mainly for capital account liberalization.
  - Trade-related indicators show positive association with growth at all levels.
- Distributional channels for individual indicators (Table 6, columns 5–8):
  - Increases in FDI are significantly associated with rising inequality (FDI stock (% GDP): OLS 0.032** (0.015)).
  - Trade indicators show no significant association with rising inequality in correlational regressions.
  - Interpretation: capital flows (FDI) more likely to drive inequality-increasing effects than trade flows.

### Robustness and sensitivity
- Outcome substitutions:
  - Growth regressions replicated with Penn World Tables; coefficients and significance barely affected (IV linear effect now significant at 10% in one specification).
  - Inequality regressions replicated with PovcalNet and ATG (smaller sample); coefficient on globalization positive, similar size, often significant at 5%.
  - Correlation between SWIID and PovcalNet/ATG Ginis in sample: p = .89.
- Econometric variations and additional checks:
  - Adding investment, debt, government expenditure (shares of GDP) as controls: coefficients barely affected.
  - Running regressions without controls: coefficients barely affected.
  - Contemporaneous (non-lagged) globalization regressions: growth point estimates stay significant and slightly smaller; inequality coefficients negligibly affected.
  - Dropping the last decade (2005–2014): average growth effect remains positive and significant; inequality coefficients lose significance.
- Outlier and IV sensitivity:
  - Winsorizing variables at 1st and 99th percentiles yields similar results.
  - Leave-one-country-out sensitivity: point estimates and confidence intervals not sensitive to exclusion of any single country.
  - Alternative distance measure for IV (largest agglomerations vs population-weighted agglomeration distances) slightly reduces instrument relevance but IV still passes tests and second-stage estimates remain similar.
- Representative robustness coefficients (selected tables):
  - Table 8 (PWT GDP figures): Column (1) OLS: 0.0043*** (0.0010); Column (3) IV: 0.0073* (0.0043).
  - Table 9 (alternative Gini sources): Column (1) OLS: 0.104** (0.047); Column (3) IV: 0.457** (0.221).
  - Table 11 column (1) OLS: Economic Globalization (t-1) 0.007*** (0.003); IV 0.042*** (0.014).
  - Table 12 column (1) OLS: Economic Globalization (t-1) 0.093** (0.036); IV 0.304** (0.123).

### Role of domestic policies and mitigation
- Redistribution and market vs net inequality:
  - More globalized economies are more unequal in market income (before taxes and transfers); net-income Gini rises less due to redistribution.
  - Regressions using market-income Gini show globalization’s effect on market-income inequality is slightly larger than on net-income inequality.
- Education and policy interactions (Table 7):
  - Average years of schooling: an increase by one year associated with a drop in net Gini of one point (Table 3, col 2).
  - Interaction of globalization with government education expenditure (% GDP):
    - Full sample: interaction not statistically significant.
    - Developing countries subsample: interaction negative and statistically significant — globalization’s positive association with inequality becomes smaller the more countries spend on education.
    - Suggestive threshold: developing countries spending less than 3 percent of GDP on education show a significant link between globalization and rising inequality; those spending more than 3 percent do not show a significant association.
  - Example Table 7 coefficients:
    - Economic Globalization main effects: Column (1) OLS 0.176*** (0.046); Column (2) IV 0.495** (0.199)
    - Economic Globalization x Education Exp. (%GDP): Column (2) -0.003 (0.013); Column (4) -0.053*** (0.017)
    - Education Exp. (%GDP): Column (3) -0.170 (0.728); Column (4) 1.888** (0.850)
- Interpretation: taxes/transfers and investments in education can mitigate distributional costs of globalization, but current levels of redistribution in many countries are insufficient to fully offset inequality rises.

### Conclusion — synthesis and policy implications
- Joint summary:
  - Economic globalization increases average incomes while distributing gains unevenly.
  - Across countries: positive yet diminishing marginal returns to globalization on growth; substantial growth benefits for early- and medium-stage globalizers; little additional growth for the most globalized economies.
  - Within countries: globalization increases income inequality; gains concentrate at the top of national income distributions; poor in many countries do not see significant absolute income gains on average.
- Policy implications:
  - Domestic policies matter: taxes and transfers partially mitigate inequality increases; education spending appears effective in reducing globalization’s inequality effect in developing countries.
  - Assessment of globalization’s consequences should account for nonlinearity and initial levels of globalization.
- Research implications:
  - Future work should analyze the interplay between globalization, growth, and inequality jointly and study policy-specific and country-specific mechanisms.

*Source: IMF Working Paper (wp1854) — content as provided.*

### 0.30 and 0.59; the pairwise correlations between the two de facto flow variables (trade and FDI)

### wp1854 - 0.30 and 0.59; the pairwise correlations between the two de facto flow variables (trade and FDI)

### Correlations, collinearity, and index construction
- Pairwise correlations reported:
  - Between the two de facto flow variables (trade and FDI): 0.59.
  - Between the two de jure restriction variables (tariffs and capital account restrictions): 0.57.
  - Other reported pairwise values include 0.30 and 0.59 in descriptive comparisons.
- Country fixed effects analysis (Figure 2) shows these strong positive associations persist after netting out unobserved country-specific, time-invariant heterogeneity.
- Conclusion: individual sub-components of economic globalization are highly correlated, motivating:
  - Use of an overall composite index aggregated from individual de jure and de facto measures (no presumption about dominant underlying components).
- KOF-like composite index properties:
  - Index range: 0 (no globalization) to 100 (maximum globalization).
  - Median (mean) change from one period to the next: 2.1 (2.9) points.
  - Squared term of globalization is included to allow nonlinearity.

### Globalization, growth, and inequality — descriptive stylized facts
- Cross-country correlations (Figure 3):
  - Economic globalization is positively correlated with GDP per capita levels.
  - Economic globalization is negatively correlated with net income inequality when high-income countries (HICs) are included; the association turns positive when HICs are excluded.
- Within-country (1990–2014) changes (Figure 4):
  - Countries that globalized more between 1990 and 2014 tended to grow more (change in ln GDP per capita).
  - Countries that globalized more also saw, on average, stronger increases in net inequality over the same period (true for full sample and for developing countries only).
- Caveats emphasized: collinearity, unobserved country-specific heterogeneity, and endogeneity can bias simple correlations.

### Data and method — measures and variables
- Measures of economic globalization:
  - Composite index combines eight de facto and de jure measures via principal component analysis (de facto: trade, FDI, portfolio investment, income payments to nonresidents; de jure: import barriers, tariffs, taxes on trade, capital account restrictions).
  - Index slow to change; median (mean) period-to-period change: 2.1 (2.9) points.
- Dependent variables:
  - Average per capita GDP growth (World Development Indicators; robustness with Penn World Tables).
  - Gini index of income inequality (baseline: SWIID; robustness: PovcalNet and All the Ginis (ATG)). Correlation between SWIID Ginis and PovcalNet/ATG in the sample: p = .89.
  - Income growth by income deciles (Global Income and Consumption Project, GCIP).
- Controls:
  - Lagged ln GDP per capita (previous 5-year period); in inequality regressions, squared term of logged GDP per capita.
  - Population growth rate, average life expectancy, average years of schooling, Polity index; all averaged over previous 5-year period.
  - Period fixed effects and country fixed effects used.
  - Parsimonious lagged controls used in baseline; extensive controls used in robustness.
- Identification (dynamic panel / IV strategy):
  - Baseline dynamic panel: y_it = β g_{i,t-1} + γ X_{i,t-1} + country FE + period FE + ε_it (variables averaged in 5-year periods).
  - IV exploits geographically diffusive character of globalization:
    - Instrument = one-period-lagged, inverse-distance-weighted globalization scores of all other countries j ≠ i at time t-1 (spatial lag with temporal lag).
    - Distance measure: population-weighted distance between agglomerations (Mayer and Zignago 2011).
  - First-stage and exclusion restriction assumptions discussed; limitations noted (LATE, sensitivity to outliers).

### Empirical results — growth (main findings)
- OLS fixed effects (controlling for country and year FE and lagged GDP per capita):
  - A one-point increase in globalization is associated with an increase in the 5-year growth rate by 0.3 percentage points (translating into an average annual growth effect of 0.06 percentage points).
- IV diagnostics and first stage:
  - First-stage coefficient on IV (α) = 0.66; t = 4.81; p < 0.001.
  - Kleibergen-Paap statistics pass Stock and Yogo thresholds (critical values cited: 16.38 for one endogenous regressor; 7.03 for two endogenous regressors).
- Nonlinear effects (allowing squared globalization term):
  - Evidence of positive but diminishing marginal effects of globalization on growth (both OLS and IV).
  - Marginal effect becomes statistically insignificant at the five percent level at a globalization score of about 77.
  - Approximately 14 percent of country-period observations in the sample surpass a globalization score of 77.
  - Example: An average low income country with globalization score 41 (e.g., Burkina Faso in the most recent period) would be expected to increase its total 5-year-period growth rate by about 2.2 percentage points when increasing globalization by one point.
  - Expected average annual growth effects:
    - Average LIC: 0.40 percentage points.
    - Average MIC: 0.36 percentage points.
  - Mean (median) period-to-period increase in globalization index: about three (two) points.
- Subsample results:
  - Regressions restricted to low- and middle-income countries support stronger growth effects there.
  - In small-sample IV, weak instrument bias cannot be ruled out (F = 5.2 reported for one smaller sample).

### Empirical results — inequality (main findings)
- Baseline results (Table 3):
  - Robustly positive and statistically significant effect of economic globalization on the Gini coefficient of net incomes.
  - IV and OLS both indicate a causal positive effect; adding controls does not change inference.
  - Magnitude: A one-point increase in economic globalization leads to a rise in the Gini index of about one third of a point.
    - Given average change in globalization of about three points per period, this implies an economically substantial effect.
    - Blackburn (1989) equivalence: a one-point change in Gini ≈ lump-sum transfer of 2 percent of mean income from bottom half to upper half.
- Nonlinearity and heterogeneity:
  - No strong evidence for significant nonlinearity in OLS; IV marginal effects plotted (Figure 6) show consistently positive effects, statistically significant only for values larger than 60 (approximate level that average MIC reaches).
  - Inequality-increasing effect particularly strong in highly globalized, advanced economies; a positive but weaker association for developing countries cannot be ruled out.
  - IV in the developing country subsample yields positive but statistically insignificant coefficient; IV may be weak in this smaller sample.
- Decile-level results (Table 4, Figure 7):
  - Income growth concentrated at top deciles:
    - Point estimates largest for top decile; statistically significant at the 10 percent level (top four deciles at 10%; top two deciles at 5%).
    - For poorest 60 percent, effect not statistically significant in full sample.
  - Developing-country subsample:
    - Evidence for growth-enhancing effect of globalization for all income deciles; point estimates are larger than in full sample.
    - Possible poverty-reducing effect in developing countries, but confidence intervals overlap—cautious interpretation needed.
  - No evidence of absolute income losses for any decile in the analyses.

### Channels (de jure vs de facto; indicators)
- Decomposition into de jure liberalization and de facto flows:
  - Both de jure and de facto dimensions contribute to the positive growth effect; nonlinearity visible for both.
  - Reassuring: similar results whether using composite index or decomposed dimensions.
- Individual indicator correlational evidence (OLS fixed effects):
  - Each of the four major indicators (trade, FDI, capital account liberalization, tariffs reduction) is positively associated with growth rates.
  - Nonlinearity (diminishing marginal returns) significant mainly for capital account liberalization.
  - Trade-related indicators show positive association with growth at all levels.
- Distributional channels:
  - Both de jure and de facto measures contribute to increasing inequality (columns 1–4, Table 6).
  - For individual indicators (columns 5–8, Table 6):
    - Increases in foreign direct investment (FDI) are significantly associated with rising inequality.
    - Trade indicators show no significant association with rising inequality in these correlational regressions.
  - Interpretation: capital flows (FDI) more likely to drive inequality-increasing effects than trade flows, consistent with theory that FDI benefits high-skilled workers.

### Robustness checks
- Outcome data substitutions:
  - Growth regressions replicated with Penn World Tables; coefficients and significance barely affected (IV linear effect now significant at 10% in one spec).
  - Inequality regressions replicated with PovcalNet and ATG (smaller sample); coefficient on globalization positive, similar size, often significant at 5%.
  - Correlation between SWIID and PovcalNet/ATG Ginis in sample: p = .89.
- Relative shares and top incomes (Table 10):
  - When using decile income shares instead of absolute growth, results indicate:
    - Bottom share falls in relative terms while top 10% and top 1% gain.
    - A one-point increase in globalization increases top 10% income share by about 0.33 percentage points and top 1% share by 0.24 percentage points.
- Econometric specification variations:
  - Adding investment, debt, government expenditure (shares of GDP) as controls: coefficients barely affected.
  - Running regressions without controls: coefficients barely affected.
  - Contemporaneous (non-lagged) globalization regressions: growth point estimates stay significant and slightly smaller; inequality coefficients negligibly affected.
  - Dropping the last decade (2005–2014): average growth effect remains positive and significant; inequality coefficients lose significance (limited data in shorter panel).
- Outlier and sensitivity analyses:
  - Winsorizing variables at 1st and 99th percentiles yields similar results.
  - Leave-one-country-out sensitivity (Young 2017 approach): point estimates and confidence intervals not sensitive to exclusion of any single country.
  - Alternative distance measure for IV (distance between largest agglomerations vs population-weighted agglomeration distances) slightly reduces instrument relevance but IV still passes tests and second-stage estimates remain similar.

### Role of domestic policies
- Redistribution and market vs net inequality:
  - More globalized economies are more unequal in market income (before taxes and transfers).
  - However, more globalized countries tend to redistribute more; net inequality patterns differ from market-inequality patterns (Figure 8).
  - Redistribution reduces but does not fully offset globalization-induced rises in inequality:
    - Regressions using market-income Gini as dependent variable show globalization’s effect on market-income inequality is slightly larger than on net-income inequality (Table 7, cols 1–4).
- Education and policy interaction:
  - Average years of schooling: an increase by one year is associated with a drop in net Gini of one point (Table 3, col 2).
  - Interaction of globalization with government expenditure on education (% of GDP) in market-inequality regressions:
    - Full sample: interaction not statistically significant.
    - Developing countries subsample: interaction negative and statistically significant — the positive association between globalization and inequality becomes smaller the more countries spend on education.
    - Suggestive threshold: developing countries spending less than 3 percent of GDP on education show a significant link between globalization and rising inequality; those spending more than 3 percent do not show a significant association.
  - Interpretation: taxes/transfers and investments in education can help mitigate distributional costs of globalization, but current levels of redistribution in many countries are insufficient to fully offset inequality rises.

### Conclusion — synthesis of findings
- Economic globalization increases average incomes but distributes gains unevenly:
  - Across countries: positive yet diminishing marginal returns to globalization on growth; substantial growth benefits for early- and medium-stage globalizers; little additional growth for the most globalized economies.
  - Within countries: globalization increases income inequality; gains concentrate at the top of national income distributions; poor in many countries do not see significant absolute income gains on average.
- Policy implications:
  - Domestic policies matter: taxes and transfers partially mitigate inequality increases; education spending appears effective in reducing globalization’s inequality effect in developing countries.
  - Assessment of globalization’s consequences should account for nonlinearity and initial levels of globalization.
- Research implications:
  - Future work should analyze the interplay between globalization, growth, and inequality jointly and study policy-specific and country-specific mechanisms.

*Source: IMF Working Paper (wp1854) — content as provided.*

### 2010. Journal of Development Economics, 104, 184–198.

### wp1854 - 2010. Journal of Development Economics, 104, 184–198.

### Major thematic clusters in the referenced literature
- Globalization, trade, and growth
  - Bloom, N., Draca, M., & Van Reenen, J. (2016). Trade induced technical change? The impact of chinese imports on innovation, IT and productivity. Review of Economic Studies, 83(1), 87–117.
  - Dorn, F., Fuest, C., & Potrafke, N. (2018). Globalization and income inequality revisited. CESifo Working Paper, forthcoming.
  - Dreher, A. (2006). Does globalization affect growth? Evidence from a new index of globalization. Applied Economics, 38(10), 1091–1110.
  - Dreher, A., & Gaston, N. (2008). Has Globalization Increased Inequality? Review of International Economics, 16(3), 516–536.
  - Felbermayr, G., & Gröschl, J. (2013). Natural disasters and the effect of trade on income: A new panel IV approach. European Economic Review, 58, 18–30.
  - Frankel, B. J. A., & Romer, D. (1999). Does Trade Cause Growth? American Economic Review, 89(3), 379–399.
  - Grossman, G. M., & Helpman, E. (1991). Trade, knowledge spillovers, and growth. European Economic Review, 35(2–3), 517–526.
  - Grossman, G. M., & Helpman, E. (2015). Globalization and Growth. American Economic Review, 105(5), 100–104.
  - Harrison, A., McLaren, J., & McMillan, M. S. (2010). Recent Findings on Trade and Inequality. NBER Working Paper Series, 16425.
  - Helpman, E. (2016). Globalization and Wage Inequality. NBER Working Paper, 22944.
  - Helpman, E., Itshoki, O., & Redding, S. (2010). Inequality and unemployment in a global economy, 78(4), 1239–1283.
  - Fajgelbaum, P. D., & Khandelwal, A. K. (2016). Measuring the Unequal Gains from Trade. Quarterly Journal of Economics, 1113–1180.
  - Ebenstein, A., Harrison, A., McMillan, M., & Philipps, S. (2014). Estimating the Impact of Trade and Offshoring on American Workers using the Current Population Surveys. The Review of Economics and Statistics, 96(4), 581–595.

- Income inequality, education, and technology
  - De Gregorio, J., & Lee, J.-W. (2002). Education and income inequality: new evidence from cross-country data. Review of Income and Wealth, 48(3), 395–416.
  - Galor, O., & Moav, O. (2004). From Physical to Human Capital Accumulation: Inequality and the Process of Development. Review of Economic Studies, 71, 1001–1026.
  - Galor, O., & Zeira, J. (1993). Income Distribution and Macroeconomics. The Review of Economic Studies, 60(1), 35.
  - Goldin, C., & Katz, L. F. (2010). The Race between Education and Technology. Belknap Press.
  - Gruber, L., & Kosack, S. (2014). The tertiary tilt: Education and inequality in the developing world. World Development, 54, 253–272.
  - Checchi, D., & van de Werfhorst, H. G. (2014). Educational Policies and Income Inequality. IZA Discussion Paper, 8222.
  - Behar, A. (2016). The endogenous skill bias of technical change and wage inequality in developing countries. Journal of International Trade and Economic Development, 25(8), 1101–1121.

- Financial globalization, capital flows, and macroeconomic outcomes
  - Borensztein, E., De Gregorio, J., & Lee, J.-W. (1998). How does foreign direct investment affect economic growth? Journal of International Economics, 45(1), 115–135.
  - Biglaiser, G., & DeRouen, K. (2010). The effects of IMF programs on U.S. foreign direct investment in the developing world. Review of International Organizations, 5(1), 73–95.
  - Furceri, D., & Loungani, P. (2018). The distributional effects of capital account liberalization. Journal of Development Economics, 130, 127–144.
  - Furceri, D., Loungani, P., & Ostry, J. D. (2017). The Aggregate and Distributional Effects of Financial Globalization: Evidence from Macro and Sectoral Data. Paper presented at the Eighteenth Jacques Polak Annual Research Conference.
  - Ghosh, A. R., Ostry, J. D., & Qureshi, M. S. (2016). When Do Capital Inflow Surges End in Tears? American Economic Review, 106(5), 581–585.
  - Cordella, T., & Ospino, A. (2017). Financial Globalization and Market Volatility. World Bank Policy Research Working Paper, 8091.
  - IMF. (2016). Global Trade: What’s behind the Slowdown. In IMF World Economic Outlook

- Measurement, identification, and empirical methods
  - Imbens, G. W., & Angrist, J. D. (1994). Identification and estimation of local average treatment effects. Econometrica, 62(2), 467–475.
  - Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143.
  - Deaton, A. (2009). Instruments of development: Randomization in the tropics, and the search for the elusive keys to economic development. NBER Working Paper, 14960.
  - Ferreira, F. H., Lustig, N., & Teles, D. (2015). Appraising cross-national income inequality databases: An introduction. Journal of Economic Inequality, 13(4), 1–30.
  - Felbermayr, G., & Gröschl, J. (2013). Natural disasters and the effect of trade on income: A new panel IV approach. European Economic Review, 58, 18–30.
  - Bazzi, S., & Clemens, M. A. (2013). Blunt instruments: Avoiding common pitfalls in identifying the causes of economic growth. American Economic Journal: Macroeconomics, 5(2), 152–186.
  - Implied use of standard datasets and measures: Feenstra, R. C., Inklaar, R., & Timmer, M. P. (2015). The next generation of the penn world table. American Economic Review, 105(10), 3150–3182.

### Representative methodological and disciplinary emphases (by selected citations)
- Causal identification and instruments: Imbens & Angrist (1994); Bazzi & Clemens (2013); Deaton (2009).
- Dynamic panel estimation techniques: Blundell & Bond (1998).
- Construction and appraisal of cross-national inequality and macro datasets: Ferreira, Lustig, & Teles (2015); Feenstra, Inklaar, & Timmer (2015).
- Use of natural experiments and IV strategies in trade and disaster studies: Felbermayr & Gröschl (2013).

### Key publication and bibliographic signals present in the unit
- Multiple references to Journal of Development Economics, Review of Economic Studies, American Economic Review, Quarterly Journal of Economics, World Development, NBER Working Papers, CESifo and IZA discussion papers.
- Recurring topics: trade-induced technical change; globalization and inequality; FDI and growth; education and income distribution; financial globalization and volatility; measurement and identification strategies.

*Source: wp1854 - 2010. Journal of Development Economics, 104, 184–198.*

### 2016. Washington DC.

### wp1854 - 2016. Washington DC.

### Main empirical findings on GDP/capita growth
- Table 2 (Growth – Main Results): Economic Globalization coefficients:
  - Column (1) OLS: 0.0033*** (standard error (0.0009))
  - Column (2) OLS: 0.0031*** (0.0009)
  - Column (3) IV: 0.0016 (0.0052)
  - Column (4) IV: 0.0018 (0.0048)
  - Column (5) OLS: 0.0101*** (0.0025)
  - Column (6) IV: 0.0470*** (0.0133)
  - Column (7) OLS: 0.0041*** (0.0012)
  - Column (8) IV: 0.0492** (0.0215)
- Table 2: Economic Globalization² in columns (6) and (8):
  - Column (6): -0.0001*** (0.0000)
  - Column (8): -0.0002*** (0.0001)
- Table 2: Other controls (selected coefficients, standard errors):
  - GDP/capita (ln): ranges from -0.1958*** (0.0422) to -0.5192*** (0.1579) across columns
  - Population Growth (%): selected estimates include -0.0199** (0.0081), -0.0188** (0.0075), -0.0151** (0.0075)
  - Life Expectancy: 0.0048*** (0.0017), 0.0068*** (0.0021), 0.0030* (0.0017)
- Table 5 (Growth – Channels): coefficients (coefficients multiplied by 100 for readability; standard errors in parentheses):
  - Economic Liberalization: OLS 0.447*** (0.170); IV 1.472*** (0.541)
  - Economic Liberalization²: -0.003** (0.001) in OLS; -0.003 (0.005) in IV
  - Economic Flows: OLS 0.883*** (0.233); IV 3.410*** (1.154)
  - Economic Flows²: -0.007*** (0.002); IV -0.024** (0.010)
  - Tariff Reductions: OLS 0.748** (0.322); IV 1.894** (0.853)
  - Capital Account Liberalization: OLS 0.640** (0.290); IV 2.109** (0.827)
  - Capital Account Liberalization²: -0.165** (0.076)
  - Trade (% GDP): OLS 0.148*** (0.043); IV 0.190*** (0.069)
  - FDI stock (% GDP): OLS 0.042** (0.016); IV -0.066 (0.044)
- Notes on growth regressions:
  - Dependent variable: GDP/capita growth (averages of 5 year periods).
  - Estimators: OLS and 2SLS fixed effects regressions. Standard errors clustered at country-level.
  - Significance notation: * p<.10, ** p<.05, *** p<.01.

### Main empirical findings on income inequality (Gini)
- Table 3 (Inequality – Main Results): Economic Globalization coefficients:
  - Column (1) OLS: 0.104*** (0.039)
  - Column (2) OLS: 0.098*** (0.038)
  - Column (3) IV: 0.368** (0.160)
  - Column (4) IV: 0.359** (0.148)
  - Column (5) OLS: 0.043 (0.141)
  - Column (6) IV: -0.254 (0.336)
  - Column (7) OLS: 0.104* (0.061)
  - Column (8) IV: 0.343 (0.359)
- Table 3: Economic Globalization² reported in some specifications:
  - Example coefficients: 0.001 (0.001) and 0.004 (0.003) where included
- Table 6 (Inequality – Channels): selected coefficients (coefficients multiplied by 100 for readability; standard errors in parentheses):
  - Economic Liberalization: OLS 0.062** (0.028); IV 0.304** (0.138)
  - Economic Flows: OLS 0.044 (0.027); IV 0.156* (0.081)
  - Tariff Reductions: OLS reported 0.190 (0.141)
  - FDI stock (% GDP): OLS 0.032** (0.015)
- Table 7 (Market Gini and Education Spending): Economic Globalization and interaction with education expenditure:
  - Economic Globalization main effects: e.g., Column (1) OLS 0.176*** (0.046); Column (2) IV 0.495** (0.199)
  - Economic Globalization x Education Exp. (%GDP): Column (2) -0.003 (0.013); Column (4) -0.053*** (0.017)
  - Education Exp. (%GDP): Column (3) -0.170 (0.728); Column (4) 1.888** (0.850)
- Notes on inequality regressions:
  - Dependent variable: Gini index of net income (unless otherwise noted).
  - Averages of 5 year periods. OLS and 2SLS fixed effects regressions. Standard errors clustered at country-level. Significance: * p<.10, ** p<.05, *** p<.01.

### Distributional effects across income deciles
- Table 4 (Income Growth by Decile), Panel A: Full Sample, Economic Globalization coefficients (with standard errors):
  - Decile 1: 0.005 (0.010)
  - Decile 2: 0.008 (0.008)
  - Decile 3: 0.009 (0.008)
  - Decile 4: 0.010 (0.007)
  - Decile 5: 0.011 (0.007)
  - Decile 6: 0.011 (0.007)
  - Decile 7: 0.012* (0.007)
  - Decile 8: 0.012* (0.006)
  - Decile 9: 0.013** (0.006)
  - Decile 10: 0.016** (0.006)
  - Top 5%: 0.017** (0.008)
  - Top 1%: 0.017** (0.008)
- Table 4, Panel B: Developing Countries, Economic Globalization coefficients (with standard errors):
  - Decile 1: 0.056* (0.033)
  - Decile 2: 0.051* (0.027)
  - Decile 3: 0.049** (0.024)
  - Decile 4: 0.046** (0.023)
  - Decile 5: 0.044** (0.021)
  - Decile 6: 0.042** (0.020)
  - Decile 7: 0.040** (0.019)
  - Decile 8: 0.038** (0.017)
  - Decile 9: 0.036** (0.016)
  - Decile 10: 0.030** (0.013)
  - Top 5%: 0.039* (0.021)
  - Top 1%: 0.039* (0.021)
- Notes on decile regressions:
  - 2SLS fixed effects regressions. Averages of 5-year periods. All explanatory variables lagged by one period. Standard errors clustered at country-level.

### Robustness and sensitivity evidence
- Table 8 (Robustness – PWT GDP figures): Economic Globalization coefficients using PWT GDP:
  - Examples: Column (1) OLS: 0.0043*** (0.0010); Column (2) OLS: 0.0039*** (0.0010)
  - Column (3) IV: 0.0073* (0.0043); Column (4) IV: 0.0072* (0.0041)
  - Column (6) IV (developing sample): 0.0410*** (0.0124)
  - Column (8) IV (developing sample): 0.0452*** (0.0165)
- Table 9 (Robustness – alternative Gini sources): Economic Globalization coefficients when using PovcalNet and All the Ginis:
  - Column (1) OLS: 0.104** (0.047)
  - Column (3) IV: 0.457** (0.221)
  - Column (7) OLS: 0.149*** (0.051)
  - Column (8) IV: 0.635 (0.468)
- Table 10 (Robustness – Relative Income Shares): Economic Globalization coefficients (2SLS) on income shares (dependent variable: income share of deciles and top percentiles):
  - Decile 1: -0.034** (0.017)
  - Decile 2: -0.032* (0.018)
  - Decile 3: -0.035* (0.019)
  - Decile 4: -0.041** (0.019)
  - Decile 5: -0.046** (0.019)
  - Decile 6: -0.049*** (0.019)
  - Decile 7: -0.049*** (0.018)
  - Decile 8: -0.040** (0.019)
  - Decile 9: -0.007 (0.026)
  - Decile 10: 0.330*** (0.101)
  - Top 5%: 0.363*** (0.113)
  - Top 1%: 0.240** (0.097)
- Additional robustness: Tables 11 and 12 present multiple sensitivity checks (e.g., alternative controls, no lag, excluding last decade, winsorized distance instrument), reporting persistent positive associations of economic globalization with growth and with higher Gini in many specifications. Example summary coefficients:
  - Table 11 column (1) OLS: Economic Globalization (t-1) 0.007*** (0.003); IV 0.042*** (0.014)
  - Table 12 column (1) OLS: Economic Globalization (t-1) 0.093** (0.036); IV 0.304** (0.123)

### Data, sample, and descriptive statistics
- Appendix 1: Sample includes 138 countries listed (selection excerpted in source; full list provided in appendix).
- Appendix 2: Selected descriptive statistics and data sources (mean, S. D., min, max, source):
  - Economic Globalization: Mean 51.77, S. D. 19.10, Min 9.31, Max 97.24 (KOF (2016))
  - Economic Restrictions: Mean 51.38, S. D. 23.09, Min 4.42, Max 95.98 (KOF (2016))
  - Economic Flows: Mean 52.69, S. D. 20.94, Min 3.40, Max 99.35 (KOF (2016))
  - Trade (% GDP): Mean 75.82, S. D. 45.19, Min 0.67, Max 410.25 (KOF (2016), based on World Bank data)
  - FDI Stock (% GDP): Mean 16.84, S. D. 21.14, Min 0.00, Max 209.11 (KOF (2016), based on UNCTAD data)
  - Tariff Reduction: Mean 7.08, S. D. 2.34, Min 0.00, Max 10.00 (KOF (2016))
  - Capital Account Liberalization: Mean 3.80, S. D. 2.95, Min 0.00, Max 10.00 (KOF (2016))
  - IV (instrument): Mean 42.16, S. D. 10.22, Min 21.38, Max 65.17 (own calculations based on Mayer and Zignago (2011) and KOF (2016))
  - GDP/Capita Growth (WDI): Mean 0.09, S. D. 0.15, Min -0.82, Max 1.08 (World Bank (2017a))
  - Gini (Net Income): Mean 38.17, S. D. 8.94, Min 18.15, Max 63.90 (Solt (2016))
  - Gini (Market Income): Mean 46.39, S. D. 7.23, Min 23.00, Max 73.42 (Solt (2016))
  - Population Growth: Mean 1.75, S. D. 1.44, Min -3.77, Max 15.53 (World Bank (2017a))
  - Education (Barro and Lee): Mean 6.59, S. D. 3.07, Min 0.45, Max 13.18 (Barro and Lee (2013))
  - Life Expectancy: Mean 65.39, S. D. 10.81, Min 29.27, Max 83.09 (World Bank (2017a))
  - Investment (% GDP): Mean 23.12, S. D. 7.37, Min 3.58, Max 58.97 (World Bank (2017a))
  - Debt (% GDP): Mean 74.18, S. D. 94.22, Min 1.38, Max 1042.64 (World Bank (2017a))
  - Government Expenditure (% GDP): Mean 15.65, S. D. 6.46, Min 0.00, Max 88.43 (World Bank (2017a))
  - Education Expenditure (% GDP): Mean 4.22, S. D. 2.28, Min 0.00, Max 36.39 (World Bank (2017a) and IMF (2017))
  - Mean income and income shares by decile (Lahoti et al. (2016)) reported (e.g., Mean Income of Decile 1 (ln) Mean 5.80, S. D. 1.60; Income Share of Decile 10 Mean 39.38, S. D. 10.76)
- Notes on instrumentation and tests:
  - Kleibergen-Paap (K-P) underidentification test p-values and K-P weak identification test F-statistics are reported across tables (examples: Table 2 K-P underid. p in some IV columns 0.001; K-P weak id. F 23.171, 23.755, 11.858, 5.219 depending on specification).
  - Several IV specifications report K-P underidentification test p-values of 0.000 or 0.001 and K-P weak id. F-statistics ranging from below 10 to above 30 in robustness checks.

*Source: wp1854 - 2016. Washington DC. (tables, appendices, and figures as provided in the source PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1854.pdf_