## _wp0528

## Source details

**Canonical URL:** [_wp0528](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0528.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0528.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0528.pdf.json)

---

### Introduction: purpose, data, and measurement
- Purpose and scope
  - Analyze the impact of income inequality on growth and test the validity of the Kuznets curve.
  - Examine relationships among economic growth, income distribution, government spending, and poverty reduction.
  - Specific questions: Is inequality harmful for growth? Is inequality related to per capita income (Kuznets curve)? How responsive is poverty to growth and changes in inequality? Would increases in government expenditures reduce poverty and improve income distribution? What minimum annual per capita real GDP growth is needed for sub‑Saharan African countries to reach poverty targets under the Millennium Development Goals by 2015?
- Data and sample
  - New dataset assembled as a panel for 82 countries for the period 1965–2003.
  - Data averaged over periods of three to seven years depending on survey availability; minimum observations per country = three; maximum = seven.
  - Entire sample: 380 observations and 290 intervals.
- Measurement
  - Poverty: percent of population living on less than $1 a day at 1993 prices, PPP adjusted.
  - Income distribution: Gini coefficient (range 0 to 1).

### Theory and evidence: channels and the Unified Model
- Literature context
  - No consensus on inequality–growth relationship.
  - Classical channel: higher marginal propensity to save of the rich → higher aggregate saving → higher physical capital accumulation → higher growth.
  - Modern channels where inequality lowers growth:
    - Increased rent-seeking and reduced security of property rights.
    - Greater social/political instability and volatile/inefficient redistribution policies.
    - Median-voter taxation effects reducing incentives.
    - With imperfect credit markets, poor cannot invest in human/physical capital.
- The Unified Model (Galor; Galor and Moav)
  - Early development: physical capital scarce; inequality raises aggregate saving and physical capital accumulation → positive effect on growth.
  - Later development: return to human capital rises; credit constraints relax; negative effect of inequality on human capital accumulation subsides; net effect may become insignificant or negative depending on stage.
  - Empirical implication: inequality may correlate positively with growth in short-to-medium term in low-financial-development contexts, but negatively in the long run.

### Empirical strategy and methodology
- Econometric approach
  - Panel regression techniques addressing simultaneity, omitted variables, and unobserved country-specific effects.
  - Averaged-period panel (3–7 year intervals) to match household survey timing.
  - Estimators discussed include OLS, Fixed Effects, and GMM dynamic panel estimators.
- Paper structure (sections and appendices)
  - The paper covers analytical review, data issues, panel regression analysis, MDG feasibility for poverty reduction, summary of empirical results, Appendix I (methodology), Appendix II (dataset).

### Key empirical findings — growth, inequality, and financial depth
- Main empirical conclusions
  - Results challenge the belief that income inequality has a universally negative effect on growth.
  - Results confirm the Kuznets curve (inverted-U relationship between inequality and per capita income).
  - Credit market imperfections in low- and medium-income countries identified as a likely reason for a positive short-to-medium term link between inequality and growth.
  - Higher government spending has a statistically significant impact on reducing inequality and poverty.
- Growth and inequality (panel regression GR_it specification)
  - Model variables: GR (per capita GDP growth), GINI_it-1, LogY_it-1, INV_it (investment/GDP), INF_it (CPI inflation), HFI (dummy for high financial intermediation), interaction GINI_it-1 * HFI, country and time effects.
- Selected coefficient estimates from Table 2 (columns (1)–(6); t-statistics in parentheses)
  - Inequality (Gini index): 0.07, 0.04, 0.07, 0.02, -0.05, -0.04 (t-statistics: (3.2), (1.9), (2.7), (0.4), (1.9), (1.8))
  - Log (per capita income): -1.36, -0.86, -1.63, -0.71, 0.98, -0.98 (t-statistics: (4.9), (4.1), (3.7), (3.6), (2.6), (2.6))
  - Investment/GDP: 0.19, 0.21, 0.21, 0.23, 0.17, 0.17 (t-statistics: (6.5), (7.0), (5.8), (7.5), (5.3), (5.3))
  - Inflation rate: -0.03, -0.03, -0.03, -0.05, -0.02, -0.01 (t-statistics: (5.8), (7.2), (5.0), (2.5), (1.3), (0.8))
  - Inequality * HFI: -0.07, 0.08 (t-statistics: (1.9), (1.6))
  - HFI Dummy: 3.67, -2.44 (t-statistics: (2.4), (1.5))
  - Countries: 82, 82, 56, 26, 6, 64
  - Observations: 269, 269, 164, 105, 81, 181
- Interpretation
  - Positive estimated β1 (inequality) in short- to medium-term columns suggests inequality can be associated with higher growth where financial intermediation is low.
  - Interaction GINI*HFI strongly negative in some specifications; inequality loses explanatory power in countries with developed financial markets.
  - Long-term (10–20 year averages) estimated inequality coefficients are negative and statistically significant at the 10 percent level, consistent with adverse long-term effects.

### Determinants of income inequality (Kuznets and other drivers)
- Regression specification: Log GINI_it on LogY_it, Log²Y_it, Log EXP_it (government expenditure % GDP), Log EDUC (secondary enrollment), POPGR, regional and measurement dummies, country and time effects.
- Key empirical estimates from Table 3 (t-statistics in parentheses)
  - Log (per capita GDP): 1.12, 1.25, 0.98 (t-statistics: (5.4), (7.1), (6.3))
  - Log² (per capita GDP): -0.16, -0.19, -0.15 (t-statistics: (6.1), (7.3), (6.7))
  - Log (government expenditure as % of GDP): -0.15, -0.17 (t-statistics: (3.8), (4.13))
  - Population growth: 0.02, 0.02 (t-statistics: (2.5), (2.4))
  - Log (secondary school enrollment): -0.18, -0.15 (t-statistics: (4.6), (3.7))
  - Dummy for sub-Saharan Africa: 0.08 (t-statistic: (4.9))
  - Dummy for Latin America: 0.14 (t-statistic: (12.1))
  - Dummy for transition economies: -0.07 (t-statistic: (6.6))
  - Dummy for income-based inequality: 0.04 (t-statistic: (3.2))
  - Adjusted R-squared: 0.19, 0.31, 0.61
  - Number of countries: 90; Observations: 378
- Findings
  - Strong evidence of non-monotonic (inverted-U) relationship consistent with Kuznets; estimated turning point ≈ 4,000 (PPP dollars, 1993 prices).
  - Per capita income explains about 20 percent of variation in inequality in baseline specification.
  - Higher government expenditures (% of GDP) and higher secondary school enrollment reduce inequality.
  - Higher population growth increases inequality.
  - Regional effects: Latin America and sub‑Saharan Africa are respectively 0.14 and 0.08 (in log‑GINI terms corresponding to 14 and 8 points) more unequal than average; transition economies are less unequal.

### Poverty, growth, inequality, and government expenditures
- Poverty regression specification: ∆P_it on GR_it (per capita real GDP growth), ∆GINI_it (change in inequality), ∆EXP_it (change in government expenditures as % of GDP), and initial GINI_it-1.
- Selected coefficient estimates from Table 4 (t-statistics in parentheses)
  - Per capita real GDP growth: -0.30, -0.29, -0.30, -0.32, -0.28 (t-statistics: (11.4), (10.5), (11.2), (8.5), (8.2))
  - Change in inequality: 0.57, 0.30, 0.83, 0.45 (t-statistics: (6.7), (2.9), (5.7), (3.2))
  - Change in government expenditure: -1.01, -0.77, -1.01 (t-statistics: (8.3), (5.7), (5.7))
  - Initial level of inequality: 0.05, 0.04 (t-statistics: (3.2), (3.3))
  - All coefficients in Table 4 reported significant at the 1% level.
- Magnitude interpretation reported in text
  - One percentage point reduction in government expenditures to GDP ratio would increase poverty by 0.7 percentage points.
- Implications
  - Growth reduces poverty, but the effectiveness depends on changes in income distribution and initial inequality.
  - For a given growth rate, poverty reduction is smaller in more unequal societies.
- Regional elasticities from Table 5 (selected)
  - Whole sample: Growth elasticity of poverty = -1.08; Inequality elasticity = 1.40; Poverty = 28; Inequality (Gini index) = 42; Gov't spending (% of GDP) = 3; Investment (% of GDP) = 22; Human capital (secondary enrollment) = 25.
  - Latin America: Growth elasticity = -1.31; Inequality elasticity = 2.02; Poverty = 16; Inequality = 52; Gov't spending = 22; Investment = 20; Human capital = 52.
  - Sub‑Saharan Africa: Growth elasticity = -0.79; Inequality elasticity = 1.20; Poverty = 49; Inequality = 44; Gov't spending = 22; Investment = 19; Human capital = 28.
  - Middle East and North Africa: Growth elasticity = -1.15; Inequality elasticity = 1.44; Poverty = 17; Inequality = 40; Gov't spending = 27; Investment = 23; Human capital = 47.
  - East and South Asia: Growth elasticity = -0.79; Inequality elasticity = 1.35; Poverty = 30; Inequality = 38; Gov't spending = 19; Investment = 26; Human capital = 53.
  - Transition countries: Growth elasticity = -1.41; Inequality elasticity = 1.30; Poverty = 31; Inequality = 31; Gov't spending = 30; Investment = 19; Human capital = 71.
  - Interpretation: Elasticities vary by region; transition countries show the largest poverty responsiveness to growth.

### Projections for Sub‑Saharan Africa to meet MDG poverty target by 2015
- Baseline facts and assumptions
  - Poverty rate in Sub‑Saharan Africa in 2001 = 47 percent of the population.
  - MDG poverty target by 2015 = 22 percent of the population.
  - Estimated growth elasticity of poverty (this paper, whole sample) = -1.08.
  - Average population growth assumed = at least 2 percent.
- Required growth scenarios (as reported)
  - To reduce poverty from 47 percent in 2001 to 22 percent by 2015 with constant inequality: per capita real GDP needs to grow by at least 4.5 percent a year.
  - Given population growth of at least 2 percent, the minimum needed economic growth (GDP growth) should be about 6.5 percent a year.
  - If inequality (Gini index) improves by 0.5 percentage point every year (declining from about 0.48 in 2001 to 0.40 in 2015), then required real GDP per capita growth = 5.5 percent a year.
- Interpretation
  - Achieving MDG poverty reduction in Sub‑Saharan Africa requires sustained high growth and/or substantial improvements in income distribution.

### Dataset and measurement notes (M2 as % of GDP; Appendix II)
- "M2 as % of GDP" defined as broad money/GDP, lines 34 plus 35 of the IFS.
- Appendix II presents a cross‑country table with country name, household survey year, per capita annual growth, GINI, secondary enrollment (%), population growth rate, government expenditure (% GDP), investment (% GDP), inflation rate (%), GNP per capita PPP (US$), credit as % of GDP, M2 as % of GDP, and poverty (% of population).
- Table coverage: low, middle, and high‑income countries across regions; some rows include negative growth rates (e.g., transition economies in the 1990s).
- Intended use: provide country-level controls and a financial depth indicator (M2 as % of GDP) for empirical analysis.

### Policy-relevant implications and recommendations
- Address credit market imperfections to prevent short-run trade-offs where inequality raises savings and capital accumulation but undermines long-term growth via poor human capital accumulation.
- Strengthen financial intermediation to reduce the positive short- to medium-term link between inequality and growth and enable broader human capital investment.
- Prioritize targeted government spending on social sectors (education, health, housing) and infrastructure to reduce inequality and enhance poverty reduction; composition matters more than aggregate size.
- Protect pro‑poor programs during fiscal consolidation (example cited: Chile’s protection of basic health and child nutrition programs in the 1980s–1990s).
- Invest in education (increase secondary enrollment) to reduce inequality; address high population growth that is associated with higher inequality.
- Recognize Kuznets dynamics: modest increases in inequality may accompany early-stage development, but long-run policies should mitigate adverse distributional effects to sustain pro‑poor growth.

*Source: _wp0528 (PDF chapter/section).*

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

### _wp0528 - References..... ........................................................................................................

### Introduction
- Purpose and scope
  - Analyze the impact of income inequality on growth and test the validity of the Kuznets curve.
  - Examine relationships among economic growth, income distribution, government spending, and poverty reduction.
  - Specific questions addressed:
    - Is inequality harmful for growth?
    - Is inequality related to the level of per capita income (Kuznets curve)?
    - How responsive is poverty to economic growth and changes in inequality?
    - Would an increase in government expenditures reduce the incidence of poverty and improve the income distribution?
    - What is the minimum annual per capita real GDP growth needed for sub-Saharan African countries to reach their respective poverty targets, under the Millennium Development Goals, by 2015?
- Data and sample
  - New dataset on inequality and poverty assembled as a panel for 82 countries for the period 1965–2003.
  - Data averaged over periods of three to seven years, depending on availability of inequality and poverty data.
  - Minimum number of observations per country is three and maximum is seven.
  - Two household surveys for one country define an interval of three to seven years in length.
  - Entire sample includes 380 observations and 290 intervals.
- Measurement
  - Poverty measured using the World Bank’s definition: percentage of the population living on less than $1 a day at 1993 prices, adjusted for purchasing power parity.
  - Income distribution measured by the Gini coefficient (range 0 to 1).

### Theory and Evidence
- Literature context
  - No consensus on the relationship between income inequality and growth.
  - Early/classical thinking: greater inequality might promote growth via higher aggregate savings (rich have higher marginal propensity to save).
  - Modern approaches emphasize channels where inequality lowers growth:
    - Encourages rent-seeking that reduces security of property rights.
    - Greater propensity for social/political instability and volatile or populist redistribution policies.
    - Median-voter effects leading to higher and more inefficient taxation.
    - When inequality coexists with imperfect credit markets, poorer people cannot invest in human and physical capital.
- The Unified Model (Galor, 2000)
  - Provides intertemporal reconciliation between classical and modern approaches.
  - Early stage of development:
    - Physical capital scarce; inequality raises aggregate savings → higher physical capital accumulation → higher economic growth.
  - Later stage of development:
    - Return to human capital rises due to capital-skill complementarity.
    - Human capital becomes main engine of growth.
    - Credit constraints become less binding as wages increase; adverse effect of inequality on human capital accumulation subsides; net effect of inequality on growth becomes insignificant.

### Channels Through Which Inequality Can Affect Growth (summary of Figure 1)
- Classical approach (Keynes 1920; Kaldor 1956; Bourguignon 1981)
  - Higher marginal propensity to save of the rich → higher aggregate saving → higher physical capital accumulation → higher economic growth.
- Modern approaches (Persson and Tabellini 1991; Alesina and Perotti 1993; Alesina and Rodrik 1994; Keefer and Knack 2000)
  - Inequality → higher rent-seeking, social tension, political instability → reduced security of property rights → lower investment.
  - Inequality → greater demand for redistribution, median-voter taxation effects → lower incentives.
  - Inequality with imperfect credit markets → poor unable to invest in human capital → lower human capital accumulation → lower long-run growth.
- Unified approach (Galor 2000)
  - Low-income stage: inequality fosters saving and physical capital accumulation → positive effect on growth.
  - Later stage: inequality hinders human capital accumulation when credit constraints bind → negative effect on growth; as constraints relax, effect diminishes.

### Empirical Strategy and Methodology (as described)
- Econometric approach
  - Employ panel regression techniques addressing simultaneity, omitted variables, and unobserved country-specific effects.
  - Use of averaged-period panel (3–7 year intervals) to match household survey timing.
- Structure of the paper
  - Section II: review of analytical arguments and literature.
  - Section III: data issues and recommendation to use more consistent data to reduce measurement error.
  - Section IV: panel regression analysis and evaluation.
  - Section V: discussion of Millennium Development Goals feasibility for poverty reduction given estimated growth elasticity of poverty.
  - Section VI: summary of empirical results.
  - Appendix I: empirical methodology.
  - Appendix II: complete dataset used.

### Key Empirical Findings (as reported)
- Main results
  - The empirical results challenge the belief that income inequality has a negative effect on growth.
  - The results confirm the validity of the Kuznets curve.
  - Credit market imperfections in low- and medium-income countries are identified as a likely reason for the positive link between inequality and growth over the short-to-medium term.
  - Evidence that higher government spending has a statistically significant impact on reducing inequality and poverty.
- Interpretation caveats
  - Data quality, period length, and estimation technique may explain differences from previous studies.

_Italic: Source: _wp0528 - References..... ........................................................................................................ (PDF chapter/section)._

### Box 1. The Unified Model

### Box 1. The Unified Model

### Overview
- The unified approach complements the research of Galor and Weil (1999, 2000) by encompassing transitions between the Malthusian Regime, the Post-Malthusian Regime, and the Modern Growth Regime, focusing on historical evolution of the relationship between population growth, technological change, and economic growth.
- Galor and Moav (1999) argue that inequality has a positive effect on capital accumulation but a negative effect on human capital accumulation in the presence of credit constraints. In early development physical capital is scarce, returns on physical capital exceed returns on human capital, and development is driven mainly by capital accumulation.
- In early stages, the positive effect of inequality on aggregate saving (because marginal propensity to save increases with wealth) can more than offset negative effects on human capital investment; in later stages the negative effect on human capital dominates.

### Empirical evidence on growth determinants and inequality
- Empirical evidence indicates growth depends on human capital, economic policies (openness to international trade, sound monetary and fiscal policies—small budget deficits and absence of high inflation), and a well-developed financial system; other factors include geography, initial incomes, and level of corruption.
- Strong evidence suggests growth is higher in countries with lower initial per capita income and in countries that have experienced a sharp fall in output (e.g., transition economies in the early 1990s).
- Cross-country growth regressions including inequality have produced mixed results:
  - Negative-impact findings: Alesina and Rodrik (1994), Clarke (1995), Perotti (1996), Panizza (2002).
  - Critiques of negative association robustness: Deininger and Squire (1998).
  - Positive-impact findings: Forbes (2000) finds positive short- and medium-term correlation between inequality and growth using fixed-effect estimations and attributes prior negative bias to country-specific effects and omitted variables.
  - Smith (2001) finds evidence that especially at low per capita income levels, income inequality may be associated with higher aggregate savings.

### B. Kuznets’s Law
- Kuznets (1955) proposed an inverted-U relationship between income inequality and per capita income: inequality rises during industrialization (urbanization and concentrated urban savings) and later declines due to (i) slower growth in population of wealthier classes, (ii) exploitation of new wealth-creation opportunities, (iii) shift of workers to higher-income industries.
- Earlier literature (1960s–1970s) largely supported an income-inequality relationship (Ahluwalia 1976). Later studies challenged it (Anand and Kanbur (1992)); Li, Squire, and Zou (1998) argue the Kuznets curve works better for cross-sections at a point in time than for within-country evolution.
- Empirical patterns over the past three decades: more countries experienced some worsening in inequality; South and East Asian economies grew at high per capita rates since the early 1970s while maintaining moderate inequality (increasing over time, notably China); Latin American countries grew slower while maintaining high inequality.

### C. Growth, Inequality, Government Spending, and Poverty
- Economic growth generally reduces poverty, but the amount of poverty reduction at a given growth rate depends on changes in income distribution and initial inequalities.
- If income inequality increases, growth can coexist with stagnant incomes for the poorest—a distributional offset.
- Fiscal policy matters: composition of government spending is critical. Increasing total government expenditures is not necessarily the answer; targeted social spending can protect the poor even during fiscal consolidation (example cited: Chile in the 1980s and 1990s).
- Larger public spending on social sectors (education, health, housing) and infrastructure is necessary to alleviate poverty and promote human development; markets for education and health are imperfect.
- However, a larger government (public expenditure to GDP) may harm growth prospects if it sustains ineffective programs and bloated bureaucracy; retrenchment can cut programs benefiting the poor and, where public employment acts as safety net, can increase inequality.

### Data issues (summary)
- Data quality and measurement errors are major concerns for cross-country inequality and poverty studies. The dataset used is an unbalanced panel of 82 countries observed from 1965 to 2003.
- Effort prioritized comparability within countries over time rather than across countries. Observations used in regressions have at least three survey observations; intervals are constructed from nationally representative surveys of length three or more years, based on expenditures or income per person.
- World Bank poverty/inequality dataset covers about 60 developing and transition countries but many have only one or two observations several years apart; author expanded data with IMF staff reports, PRSPs, and OECD inequality data.
- Survey methodological differences: some surveys measure income, others consumption/expenditures; for developing/transition countries slightly more than half of observations are based on expenditures, the remainder on income. Expenditure-based surveys typically yield lower inequality estimates than income-based surveys.
- The most widely used poverty indicator for developing and transition economies is the percent of the population living below $1 a day of consumption or income at 1993 prices, adjusted for PPP.

### IV. Econometric results

A. Growth and Inequality
- Panel regression framework and specification (equation reproduced as in source):
  GR_it = α_1t + β_1 GINI_it-1 + β_2 LogY_it-1 + β_3 INV_it + β_4 INF_it + β_5 GINI_it-1 * HFI + β_6 HFI + μ_i + ν_t + ε_it
  - GR: average growth rate of per capita GDP at 1993 prices and PPP adjusted.
  - GINI_it-1: Gini index in the previous period.
  - LogY_it-1: natural log of beginning-period per capita GDP (1993 prices, PPP).
  - INV_it: share of gross capital formation in GDP.
  - INF_it: average CPI inflation rate.
  - HFI: dummy = 1 for countries with high financial intermediation (above sample median measured by share of M2 and credit to private sector in GDP).
  - μ_i: country-specific unobservable; ν_t: time-specific factor.
- Main findings excluding inequality/financial intermediation:
  - Negative and significant correlation between growth and initial income per capita (poorer countries tend to grow faster).
  - Strong association between investment shares and GDP growth.
  - Macroeconomic instability (inflation) is negatively correlated with growth.
- Inequality–growth relationship:
  - Estimated coefficients on inequality (GINI_it-1) are positive in columns (1) to (4) (short- to medium-term effects).
  - Columns (3) and (4) split sample by low and high financial intermediation: effect of inequality on growth differs across financial development.
  - Expected signs: β_1 > 0, β_5 < 0, and β_6 > 0 (positive effect of inequality on growth weaker in countries with high financial intermediation).
  - Interaction term GINI_it-1 * HFI is strongly negative in column (1); F-test indicates GINI_it-1 * HFI and HFI are jointly highly significant. Coefficient for HFI (β_6) is positive and highly significant.
  - Inequality coefficient insignificance in column (4) consistent with inequality having no explanatory power in countries with developed financial markets.
  - Long-term (columns (5) and (6), data as 10–20 year averages): estimated inequality coefficients are negative and statistically significant at the 10 percent level.
- Conclusion: Credit market imperfections may explain a positive short- to medium-term link between inequality and growth in countries with low financial development; over the long term inequality may adversely affect growth.

- Selected numerical estimates from Table 2 (as presented):
  - Inequality (Gini index): 0.07, 0.04, 0.07, 0.02, -0.05, -0.04 (corresponding to columns (1)–(6); t-statistics in parentheses: (3.2), (1.9), (2.7), (0.4), (1.9), (1.8))
  - Log (per capita income): -1.36, -0.86, -1.63, -0.71, 0.98, -0.98 (t-statistics: (4.9), (4.1), (3.7), (3.6), (2.6), (2.6))
  - Investment/GDP: 0.19, 0.21, 0.21, 0.23, 0.17, 0.17 (t-statistics: (6.5), (7.0), (5.8), (7.5), (5.3), (5.3))
  - Inflation rate: -0.03, -0.03, -0.03, -0.05, -0.02, -0.01 (t-statistics: (5.8), (7.2), (5.0), (2.5), (1.3), (0.8))
  - Inequality * HFI: -0.07, 0.08 (t-statistics: (1.9), (1.6))
  - HFI Dummy: 3.67, -2.44 (t-statistics: (2.4), (1.5))
  - Countries: 82, 82, 56, 26, 6, 64 (presentation corresponds to table layout)
  - Number of observations: 269, 269, 164, 105, 81, 181

B. Determinants of Income Inequality
- Regression specification (equation reproduced as in source):
  Log GINI_it = β_1 LogY_it + β_2 Log²Y_it + β_3 Log EXP_it + β_4 Log EDUC + β_5 POPGR + Regional Dummies + Dummy for Income based Inequality + μ_i + ν_t + ε_it
  - Log GINI_it: natural log of Gini index.
  - LogY_it and Log²Y_it: test nonlinear (quadratic) relationship (Kuznets’ curve).
  - Log (EXP): natural log of government expenditures as % of GDP.
  - EDUC: secondary school enrollment rate (percent of secondary school-aged population).
  - POPGR: percent change in population.
  - Dummies: Sub-Saharan Africa (D1), Latin America (D2), East Europe/former Soviet Union prior to 1995 (D3), income-based Gini (D4).
- Findings:
  - Clear evidence of a non-monotonic (inverted-U) relationship between inequality and per capita income (Kuznets’ curve). Estimated coefficients: about 1 on the linear LogY term and -0.15 on the squared term; estimated coefficients stable over time.
  - Estimated Kuznets turning point at about 4,000 in PPP dollars of 1993 prices.
  - Per capita income explains about 20 percent of variation in inequality (first column of Table 3).
  - Government size (government expenditures as % of GDP) has a negative and highly significant coefficient: higher government spending tends to reduce inequality.
  - Secondary school enrollment (proxy for human capital) has a negative and highly significant coefficient: improvement in education reduces inequality.
  - Population growth has a positive coefficient: high population growth increases inequality.
  - Regional dummies: Latin America dummy = 0.14 (highly significant), Sub-Saharan Africa dummy = 0.08 (highly significant); Latin America and sub-Saharan Africa are 14 and 8 points more unequal than the average, respectively. Transition economies dummy is negative and significant. Dummy for income-based Gini (D4) is significant and implies income-based inequality is on average about four percentage points higher than expenditure-based inequality.

- Selected numerical estimates from Table 3 (as presented):
  - Log (per capita GDP): 1.12, 1.25, 0.98 (t-statistics: (5.4), (7.1), (6.3))
  - Log (per capita GDP) squared: -0.16, -0.19, -0.15 (t-statistics: (6.1), (7.3), (6.7))
  - Log (government expenditure as % of GDP): -0.15, -0.17 (t-statistics: (3.8), (4.13))
  - Population growth: 0.02, 0.02 (t-statistics: (2.5), (2.4))
  - Log (secondary school enrollment): -0.18, -0.15 (t-statistics: (4.6), (3.7))
  - Dummy for sub-Saharan Africa: 0.08 (t-statistic: (4.9))
  - Dummy for Latin America: 0.14 (t-statistic: (12.1))
  - Dummy for transition economies: -0.07 (t-statistic: (6.6))
  - Dummy for income based inequality: 0.04 (t-statistic: (3.2))
  - Adjusted R-squared: 0.19, 0.31, 0.61
  - Number of countries: 90; Number of observations: 378 (columns as presented)

C. Role of Growth, Equity, and Government Expenditures in Reducing Poverty
- Poverty measurement: headcount change in poverty is used with $1 a day (1993 PPP) as benchmark for low-income countries.
- Regression specification (equation reproduced as in source):
  ∆P_it = α_i + β_1 GR_it + β_2 ∆GINI_it + β_3 ∆EXP_it + β_4 GINI_it-1 + μ_i + ν_t + ε_it
  - ∆P_it: headcount change in poverty (percentage points) between two household survey years.
  - GR_it: per capita real GDP growth rate between survey years.
  - ∆GINI_it: change in inequality (percentage points).
  - ∆EXP_it: change in government expenditures as percent of GDP (proxy for social spending/infrastructure).
  - GINI_it-1: initial inequality level.
- Findings from Table 4:
  - Per capita real GDP growth coefficient: -0.30, -0.29, -0.30, -0.32, -0.28 (t-statistics: (11.4), (10.5), (11.2), (8.5), (8.2)) — higher growth reduces poverty (in percentage points).
  - Change in inequality coefficient: 0.57, 0.30, 0.83, 0.45 (t-statistics: (6.7), (2.9), (5.7), (3.2)) — increases in inequality raise poverty.
  - Change in government expenditure coefficient: -1.01, -0.77, -1.01 (t-statistics: (8.3), (5.7), (5.7)) — increases in government expenditure reduce poverty.
  - Initial level of inequality coefficient: 0.05, 0.04 (t-statistics: (3.2), (3.3)) — higher initial inequality reduces the poverty-reducing impact of growth.
  - All estimated coefficients in Table 4 are reported significant at the 1% level.
  - Magnitude interpretation: One percentage point reduction in government expenditures to GDP ratio would increase poverty by 0.7 percentage points (as reported in text).
- Implication: Growth reduces poverty more when accompanied by decreased inequality; for the same growth, poverty reduction is smaller in more unequal societies.

- Regional elasticity estimates (Table 5, selected values as presented):
  - Whole sample: Growth elasticity of poverty = -1.08; Inequality elasticity = 1.40; Poverty (% of pop.) = 28; Inequality (Gini index) = 42; Gov't spending (% of GDP) = 3; Investment (% of GDP) = 22; Human capital (secondary enrollment) = 25.
  - Latin America: Growth elasticity = -1.31; Inequality elasticity = 2.02; Poverty = 16; Inequality = 52; Gov't spending = 22; Investment = 20; Human capital = 52.
  - Sub-Saharan Africa: Growth elasticity = -0.79; Inequality elasticity = 1.20; Poverty = 49; Inequality = 44; Gov't spending = 22; Investment = 19; Human capital = 28.
  - Middle East and North Africa: Growth elasticity = -1.15; Inequality elasticity = 1.44; Poverty = 17; Inequality = 40; Gov't spending = 27; Investment = 23; Human capital = 47.
  - East and South Asia: Growth elasticity = -0.79; Inequality elasticity = 1.35; Poverty = 30; Inequality = 38; Gov't spending = 19; Investment = 26; Human capital = 53.
  - Transition countries: Growth elasticity = -1.41; Inequality elasticity = 1.30; Poverty = 31; Inequality = 31; Gov't spending = 30; Investment = 19; Human capital = 71.
- Interpretation: Elasticities vary substantially by region; transition countries show the highest poverty elasticity to growth (a 10 percent decline in real per capita growth would lead to a 14 percent increase in poverty incidence).

### Policy-relevant implications and recommendations (derived from findings)
- Address credit market imperfections to avoid short-run trade-offs where inequality raises savings/capital accumulation but undermines long-term growth through poor human capital accumulation.
- Strengthen financial intermediation to reduce the positive short- to medium-term link between inequality and growth and to enable broader human capital investment.
- Prioritize targeted government spending on social sectors (education, health) and infrastructure to reduce inequality and enhance poverty reduction; composition of spending matters more than aggregate increases.
- Protect pro-poor programs during fiscal consolidation to maintain social outcomes (example: Chile’s protection of basic health and child nutrition programs during 1980s–1990s fiscal adjustment).
- Invest in education (secondary enrollment) as a means to reduce inequality; address high population growth which is associated with higher inequality.
- Recognize that in the early stages of development, modest increases in inequality may accompany rising per capita income (Kuznets dynamics); however, long-run policies should mitigate adverse distributional effects to sustain pro-poor growth.

*Source: _wp0528 - Box 1. The Unified Model*

### 2015. The poverty line in this case is based on one dollar per day in purchasing power parity.

### _wp0528 - 2015. The poverty line in this case is based on one dollar per day in purchasing power parity.

### Global and regional poverty trends (1981–2001)
- Global poverty has declined significantly over the past two decades; most of the improvement was due to sharp reductions in China and India.
- East Asia and the Pacific: poverty rate fell from about 58 percent of the population in 1981 to 15 percent in 2001, mainly because of dramatic poverty reduction in China.
- Sub-Saharan Africa: poverty rates increased from 42 percent of the population in 1981 to 47 percent in 2001, mainly explained by stagnant annual per capita growth in sub-Saharan Africa over the past two decades.
- Recent declines in poverty in several Asian economies were due to both improved income distribution and sustained rapid growth.

### Estimated elasticities of poverty
- Estimated growth elasticity of poverty (this paper, whole sample): -1.08.
- Estimated inequality elasticity of poverty (this paper, whole sample): 1.40.
- World Bank studies (including Ravallion, 1997): growth elasticity of poverty -1.68 and inequality elasticity of poverty 1.90.

### Projections and required growth for Sub‑Saharan Africa (to reach 22 percent poverty by 2015)
- Based on the paper's estimated growth elasticity of poverty, per capita real GDP in sub‑Saharan African countries needs to grow by at least 4.5 percent a year to reduce poverty from 47 percent in 2001 to 22 percent by 2015 (Table 7).
- Given an average population growth of at least 2 percent, this implies that the minimum needed economic growth should be about 6.5 percent a year.
- If inequality, as measured by the Gini index, improves by half a percentage point every year (average inequality declining from about 0.48 in 2001 to 0.40 in 2015), then real GDP needs to grow by 5.5 percent a year.

### Key empirical conclusions and policy-relevant findings
- Short-to-medium term relationship: panel regression results suggest that an increase in a country’s level of income inequality may have a positive relationship with subsequent economic growth; credit market imperfections may be a source of this positive link between inequality and growth.
- Long-term relationship: in the long term, inequality would have an adverse impact on growth.
- Kuznets curve: results confirm the Kuznets hypothesis—income inequality first increases and later decreases during economic development; inequality rises with per capita income to a certain level, estimated at about 4,000 in PPP dollars (1993 prices), and declines thereafter.
- Per capita income explains only about 20 percent of variations in inequality across countries or over time.
- Higher per capita growth is associated with higher rates of poverty reduction; the variation in poverty reduction for similar growth rates reflects the degree of income inequality.
- For the same growth in per capita income, poverty is reduced more in countries with low initial inequality than in countries with high initial inequality.
- Other things equal, growth leads to less poverty reduction in unequal societies than in egalitarian ones.
- Specific restatement: Sub‑Saharan African countries will need to grow by at least 6.5 percent a year to reduce poverty from 47 percent of the population in 2001 to 22 percent by 2015 (assuming that the level of inequality remains constant).

### Data and methodological notes (selected)
- Poverty line: percentage of population living on less than $1 a day at 1993 prices and adjusted for purchasing power parity.
- Poverty figures source: World Bank (poverty figures extracted from http://www.worldbank.org/research/povmonitor on August 4, 2004) except Middle East and North Africa (derived from Appendix II data).
- Per capita annual growth figures are derived from the IMF WEO database.
- Appendix highlights: panel estimation techniques discussed include Ordinary Least Squares (OLS), Fixed Effects, and Generalized Method-of-Moments (GMM) dynamic panel estimators; concerns addressed include unobserved time- and country-specific effects and likely endogeneity of some regressors.

*Canonical source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0528.pdf*

### 10. M2 as % of GDP:  Broad money/GDP, lines 34 plus 35 of the IFS.

### _wp0528 - 10. M2 as % of GDP:  Broad money/GDP, lines 34 plus 35 of the IFS.

### Dataset description
- Appendix II table presenting cross‑country data covering many countries and multiple survey years.
- The table reports a broad set of socioeconomic and macroeconomic variables for each country–year observation, with one column labelled "M2 as % of GDP" (broad money/GDP, lines 34 plus 35 of the IFS).
- The dataset is presented as a panel-like listing; entries include country name and survey year alongside multiple indicators in fixed columns.

### Key variables included (as labelled in the table)
- Household survey year
- Per capita annual growth (growth)
- Inequality index (GINI)
- Secondary school enrollment (percent)
- Population growth rate
- Government expenditure as % of GDP
- Investment as % of GDP
- Inflation rate (percent)
- GNP per capita, PPP (US$)
- Credit as % of GDP
- M2 as % of GDP
- Poverty (percent of population)

### Coverage and format notes
- The table spans low, middle, and high‑income countries (examples across Africa, Asia, Europe, Latin America, and OECD countries).
- Data are presented row by row; some rows include negative growth rates and negative entries where reported (e.g., transition economies in the 1990s).
- Numeric entries are reported in the original precision and formatting as in the source table.

### Intended use and content focus
- Purpose: to provide country‑level control variables and financial depth indicator (M2 as % of GDP) for empirical analysis of inequality, growth, and related macroeconomic relationships.
- The "M2 as % of GDP" column is the primary focus of this content unit, intended to measure broad money relative to economic output for comparative cross‑country analysis.

*Source: _wp0528 - 10. M2 as % of GDP:  Broad money/GDP, lines 34 plus 35 of the IFS.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0528.pdf_