## _wp04214 - 1. Summary Statistics and Correlations

## Source details

**Canonical URL:** [_wp04214 - 1. Summary Statistics and Correlations](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04214.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04214.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04214.pdf.json)

---

### Introduction and motivation
- Universal public provision is often modeled as delivering uniform services (examples cited: Meltzer and Richard (1981), Epple and Romano (1996), Doepke and de la Croix (2004a, 2004b)).
- Empirical evidence indicates uniform provision frequently fails redistributive goals where institutions are weak due to political capture and bureaucratic inefficiencies.
- Public interventions (education, health, housing, transportation) can disproportionately benefit middle and upper income groups (Le Grand (1982); Birdsall (1997); World Bank 2001, 2004).
- In many developing countries, public education incidence is skewed toward richer households; spending on primary education is small relative to tertiary education and incidence to the top quintile can be three to four times that received by the bottom quintile.

### Theoretical framework and key mechanisms
- Dynamic model of public education spending with rent-seeking appropriation.
- Key mechanism:
  - Richer individuals are less credit constrained and can spend more on rent seeking, capturing a larger share of public spending.
  - Governance quality Q_t (0 < Q_t < 1) moderates appropriation: higher Q_t reduces the role of rent seeking; Q_t = 1 implies equal access, Q_t = 0 implies access depends strongly on rent seeking.
- Production and preferences (selected equations preserved):
  - Budget constraint: y_it = c_it + T y_it + x_it+1.
  - Allocation rule (Tullock-contest extension): s_it+1 = x_it+1^{1-Q_t} / ∫_0^1 x_jt+1^{1-Q_t} dj.
  - Production: y_it+1 = A a_it+1^γ (s_it+1 T Y_t)^α, with 0 < γ < 1 and 0 < α < 1.
  - Preferences: U_it = ln(c_it) + ln(y_it+1).
- Equilibrium tax support: all individuals support T = α/(1+α).

### Analytical results and propositions
- Individual optimal choices (from maximization subject to model constraints):
  - x_it+1 = α (1-Q_t) y_it (1- T)/(1 + α(1-Q_t))
  - c_it = y_it (1- T)/(1 + α(1-Q_t))
  - s_it+1 = y_it^{1-Q_t} / ∫_0^1 y_jt^{1-Q_t} dj
- Incidence bias measure (variance of ln share):
  - Bias = (1-Q_t)^2 σ_t^2.
  - Interpretation: incidence bias decreases with governance quality Q_t and increases with income inequality σ_t^2.
- Proposition 1: Stronger governance and lower current income inequality imply a smaller incidence bias.
- Intergenerational dynamics:
  - μ_{t+1} = ln A + α(μ_t + σ_t^2/2) + α ln(T) - α(1-Q_t)^2 σ_t^2/2
  - σ_{t+1}^2 = γ^2 + (α(1-Q_t))^2 σ_t^2 = γ^2 + α^2 Bias
  - Steady-state variance for constant Q: σ^2* = γ^2 / [1- (α(1-Q))^2]
- Mobility and growth:
  - Intergenerational mobility (complement of intertemporal correlation): 1 – Cov(ln(y_it+1), ln(y_it))/(σ_{t+1} σ_t) = 1 - {1/[(γ /α(1−Q_t)σ_t)^2 + 1] }^{1/2}; higher Q_t ⇒ higher mobility.
  - Average income growth (log): g_{t+1} = ln A + (α-1)(μ_t + σ_t^2/2) + α ln(T) + γ^2 /2 + α(α −1)(1-Q_t)^2 σ_t^2/2.
- Proposition 2: Better governance quality in period t reduces income inequality in next period and in steady state, increases intergenerational mobility, and enhances growth.

### Comparison: public versus private education regimes
- Private regime allocations (no public provision):
  - y_it = c_it + e_it+1
  - e_it+1 = α y_it /(1+α)
  - c_it = y_it /(1+α)
  - y_it+1 = A a_it+1^γ [ α y_it /(1+α) ]^α
- Utility differential between public and private regimes:
  - U_it - R_it = ln[1/(1 + α(1-Q_t))] - α Q_t ln(y_it) + α Q_t μ_t + α Q_t(2-Q_t) σ_t^2/2.
- Implications:
  - Richer households prefer private education (term - α Q_t ln(y_it)).
  - Majority support for public education depends on governance quality Q_t; expression inside utility differential is negative at Q_t = 0 and positive at Q_t = 1.

### Empirical strategy, data, and measures
- Data source: Davoodi et al. (2003) benefit incidence (BIA) dataset for 36 developing and transition countries in the 1990s.
- Two dependent measures of pro-rich bias:
  - Q5/Q1 = total education spending accruing to wealthiest quintile relative to poorest quintile.
  - Q5 = share of public education benefits accruing to wealthiest quintile (percent).
- Governance indicators (rescaled 0–1, higher = better):
  - Gastil (average of civil liberties and political rights, rescaled).
  - Regulatory quality (Kaufmann et al. (1999)).
  - Rule of law (Kaufmann et al. (1999)).
- Controls (entered one at a time in regressions): Gini (gross income inequality), IGDP (real per capita GDP PPP averaged over 1980s), Credit (private sector credit/GDP), Educ (average years schooling age 25+ in 1990).

### Descriptive statistics (Table 1 — exact figures)
- Means:
  - Q5/Q1 = 1.76
  - Q5 = 23.64
  - Regulatory quality = 0.53
  - Rule of Law = 0.46
  - Gastil = 0.59
  - Gini = 44.11
  - IGDP = 2611.63
  - Educ = 4.85
  - Credit = 24.25
- Standard deviations:
  - Q5/Q1 = 1.28
  - Q5 = 9.49
  - Regulatory quality = 0.09
  - Rule of Law = 0.10
  - Gastil = 0.23
  - Gini = 9.94
  - IGDP = 1675.14
  - Educ = 2.32
  - Credit = 17.90
- Maximum values:
  - Q5/Q1 = 5.02
  - Q5 = 46.00
  - Regulatory quality = 0.75
  - Rule of Law = 0.74
  - Gastil = 1
  - Gini = 63.00
  - IGDP = 6614.40
  - Educ = 9.26
  - Credit = 79.04
- Minimum values:
  - Q5/Q1 = 0.11
  - Q5 = 4.60
  - Regulatory quality = 0.38
  - Rule of Law = 0.28
  - Gastil = 0
  - Gini = 24.53
  - IGDP = 508.70
  - Educ = 0.84
  - Credit = 3.18
- Examples (Table A1 benefit incidence extremes):
  - Chile (CHL): Q1 = 42, Q5 = 4.6, Q5/Q1 = 0.1 (lowest bias).
  - Nepal (NPL): Q1 = 11, Q5 = 46, Q5/Q1 = 4.4 (highest bias).
  - Nicaragua (NIC): Q1 = 9, Q5 = 40, Q5/Q1 = 4.4 (high bias).

### Empirical estimation results (highlights)
- OLS results (Tables 2 and 3):
  - Coefficient on governance indicators is consistently negative and statistically significant at the 1 percent level across governance indicators and across both bias measures.
  - Gini often positive but not statistically significant in OLS.
  - Adding controls (IGDP, Credit, Educ) does not overturn qualitative results; Educ is significant in some specifications when dependent variable is Q5.
- IV strategy (Table 4):
  - Instruments: distance from the equator (Lat) and fraction of population speaking English and European languages.
  - Hansen J-Statistic p-values reported (examples): 0.17, 0.49, 0.37, 0.11, 0.28, 0.14 — instruments not rejected in these specifications.
  - IV results: governance measures remain negative and statistically significant; coefficients on governance generally significant at 5 and 10 percent levels in some specifications.
  - After instrumenting, coefficients on other controls are not consistently significant.

### Empirical conclusions and policy implications
- Empirical tests support Proposition 1: lower governance quality is associated with higher pro-rich bias in public education spending.
- Robustness: results hold in both OLS and IV specifications controlling for Gini, IGDP, Credit, and Educ.
- Policy-relevant implications from model and evidence:
  - Improving governance quality reduces pro-rich incidence bias in public education spending.
  - Better governance reduces future inequality, increases intergenerational mobility, and raises average growth rates.
  - Majority support for public education depends on governance quality; poor governance can shift political support toward regimes that perpetuate inequality.
- Suggested directions for further research:
  - Empirical study of the inequality and growth implications of incidence bias requires panel data; current cross-country coverage limits such analysis.
  - Extend framework to incidence bias in other public spending areas (health, welfare, infrastructure) to assess whether governance quality has similar effects.

*Source: _wp04214 - 1. Summary Statistics and Correlations (IMF Working Paper).*

### 1. Summary Statistics and Correlations .................................................................................

### _wp04214 - 1. Summary Statistics and Correlations .................................................................................

### Introduction
- Universal public provision is commonly assumed to supply a uniform level of services; models cited include Meltzer and Richard (1981), Epple and Romano (1996), Doepke and de la Croix (2004a, 2004b).
- Empirical evidence and policy awareness indicate that uniform provision often fails to achieve redistributive goals, especially in countries with weak institutions due to political capture and bureaucratic inefficiencies.
- Le Grand (1982) documents that public interventions (education, health, housing, transportation) often primarily benefit the middle class and the rich.
- In many developing countries, public education incidence is skewed in favor of richer households: spending on primary education is small relative to tertiary education (Birdsall (1997)); in some countries the incidence accruing to the top quintile is three to four times larger than that received by the bottom quintile (World Bank 2001, 2004).

### Theoretical framework and mechanisms
- A simple dynamic model of public education spending is constructed to study incidence bias when spending is subject to appropriation through rent seeking.
- Key mechanism:
  - Richer individuals are less credit constrained and can spend more on rent seeking, allowing appropriation of a larger share of public spending.
  - Prevailing governance quality moderates the ability to appropriate: higher governance quality reduces room for appropriation.
- Model implications:
  - Weak governance results in a higher level of rent seeking and a skewed incidence of public spending toward the rich.
  - This skew adversely affects future income distribution, intertemporal mobility, and impedes economic growth.
  - Support for public education relative to private education also depends on governance quality.

### Empirical evidence and summary statistics
- Cross-section evidence presented for 36 developing countries shows a negative bivariate relationship between the share of public education spending accruing to the wealthiest quintile relative to the poorest and an indicator of governance.
- Illustrative country observations:
  - Malaysia (MYS): one of the lowest bias in public education spending and the best quality of governance in the sample.
  - Mozambique (MOZ): very high pro-rich bias in education spending and poor enforcement of rule of law.
- Additional summarized empirical findings:
  - The relationship between governance quality and pro-rich bias is robust after controlling for an economy's level of development and additional characteristics.
  - The moderating effect of governance on incidence bias remains across various measures of governance quality and incidence bias.

### Implications and comparisons
- Differential access to public funds (non-uniform incidence) matters for growth, inequality, and political sustainability; this differentiates the paper from prior work that assumes uniform access.
- Comparison with related mechanisms:
  - Unlike models where leakages ensure political support (e.g., Gelbach and Pritchett (2002), Ravallion (1999)), here leakages arise directly from rent seeking.
  - The framework highlights the role of governance factors as significant in shaping the distribution of public spending.
- The paper connects to literature on growth and inequality effects of public education (Gradstein and Justman (1997), Doepke and de la Croix (2004a)), and empirical studies such as Barro (1991) and Gylfason and Zoega (2003).

### Organization of the paper (as presented)
- Model presentation followed by equilibrium analysis and comparison with a private education system.
- Empirical testing using a recently compiled cross-country data set confirming model predictions.
- Figures and appendices included:
  - Figures: Bias in Public Education Spending and Good Governance; Convergence to a Steady State.
  - Appendices: Preferences Over Governance Quality; Data.
  - Appendix Table: Benefit Incidence of Public Spending on Education.

*Source: _wp04214 - 1. Summary Statistics and Correlations .................................................................................*

### Section III. Empirical implications are then studied in Section IV, and Section V concludes

### _wp04214 - Section III. Empirical implications are then studied in Section IV, and Section V concludes

### Basic framework
- Population: infinite number of successive generations of households, measure normalized to one; each household consists of a parent and a child.
- Initial income: y_i0; period t income: y_it.
- Public education financed by proportional income tax with rate T; disposable income y_it(1- T); balanced budget implies aggregate public education spending TY_t.
- Households allocate disposable income between consumption c_it and individual rent-seeking investment x_it+1 intended to increase share of public education resources for offspring (no productive value).
- Budget constraint (normalized prices): y_it = c_it + T y_it + x_it+1 (equation (1)).
- Allocation rule for shares (Tullock-contest extension):
  s_it+1 = x_it+1^{1-Q_t} / ∫_0^1 x_jt+1^{1-Q_t} dj  (equation (2)), where 0 < Q_t < 1 is governance quality.
  - Q_t = 1 implies equal access to public education; Q_t = 0 implies access depends heavily on rent seeking.
- Production input for child: s_it+1 T Y_t.
- Individual production capability: A_it = A a_it^γ (equation (4)), with 0 < γ < 1.
- Production technology (Cobb-Douglas variant): y_it+1 = A_it+1 (s_it+1 T Y_t)^α, 0 < α < 1 (equation (3)).
- Preferences (logarithmic, no discounting): U_it = ln(c_it) + ln(y_it+1) (equation (5)).
- Under these assumptions, all individuals unanimously support tax rate T = α/(1+α).

### Equilibrium allocations — key analytic results
- Individual optimal choices (from maximization of (5) subject to (1)–(4)):
  - x_it+1 = α (1-Q_t) y_it (1- T)/(1 + α(1-Q_t))
  - c_it = y_it (1- T)/(1 + α(1-Q_t))
  - s_it+1 = y_it^{1-Q_t} / ∫_0^1 y_jt^{1-Q_t} dj  (equation (6))
  - Individual rent-seeking efforts decrease with Q_t.
- Next-period income (substituting into production):
  - y_it+1 = A a_it+1^γ [ y_it^{1-Q_t} T Y_t / ∫_0^1 y_jt^{1-Q_t} dj ]^α  (equation (7))
- Utility expression (after substitution):
  - U_it = ln[y_it(1-T)/(1 + α(1-Q_t))] + ln{A a_it+1^γ [ y_it^{1-Q_t} T Y_t / ∫_0^1 y_jt^{1-Q_t} dj ]^α }  (equation (8)

### Incidence bias, inequality, and propositions
- Log-share expression:
  - ln(s_it+1) = (1-Q_t) ln(y_it) - ln[ ∫_0^1 y_jt^{1-Q_t} dj ]  (equation (9))
- Variance of ln(s_it+1) — incidence bias measure:
  - Bias = (1-Q_t)^2 σ_t^2  (equation (10))
  - Interpretation: incidence bias decreases with governance quality Q_t and increases with income inequality σ_t^2.
- Proposition 1: Stronger governance and lower level of current income inequality imply a smaller incidence bias.
- Intergenerational inequality dynamics:
  - ln(y_it+1) = ln(A) + γ ln(a_it+1) + α(1-Q_t) ln(y_it) + α ln(TY_t) - α ln[ ∫_0^1 y_jt^{1-Q_t} dj ]  (equation (11))
- Mean and variance dynamics:
  - μ_{t+1} = ln A + α(μ_t + σ_t^2/2) + α ln(T) - α(1-Q_t)^2 σ_t^2/2  (equation (12a))
  - σ_{t+1}^2 = γ^2 + (α(1-Q_t))^2 σ_t^2 = γ^2 + α^2 Bias  (equation (12b))
  - Steady-state variance (for constant Q): σ^2* = γ^2 / [1- (α(1-Q))^2]
- Intergenerational mobility measure (complement of intertemporal correlation):
  - 1 – Cov(ln(y_it+1), ln(y_it))/(σ_{t+1} σ_t) = 1 - {1/[(γ /α(1−Q_t)σ_t)^2 + 1] }^{1/2}  (equation (13))
  - Higher Q_t ⇒ higher mobility.
- Average income growth (log):
  - g_{t+1} = ln A + (α-1)(μ_t + σ_t^2/2) + α ln(T) + γ^2 /2 + α(α −1)(1-Q_t)^2 σ_t^2/2  (equation (14))
  - Differentiation: growth increases with governance quality Q_t and decreases with incidence bias.
- Proposition 2: Better governance quality in period t reduces income inequality in next period and in steady state, increases intergenerational mobility, and enhances growth.
  - Mechanism: better governance reduces marginal benefit from rent seeking, yielding more egalitarian distribution of public education and, because of diminishing returns to schooling, higher aggregate growth.

### Comparison with private education regime
- Private regime budget: y_it = c_it + e_it+1  (equation (15))
- Private education production: y_it+1 = A_it+1 (e_it+1)^α  (equation (16))
- Private-optimal allocations:
  - e_it+1 = α y_it /(1+α)
  - c_it = y_it /(1+α)  (equation (17))
  - y_it+1 = A a_it+1^γ [ α y_it /(1+α) ]^α  (equation (18))
- Utility in private regime (R_it) and in public regime (U_it) lead to utility differential:
  - U_it - R_it = ln[1/(1 + α(1-Q_t))] - α Q_t ln(y_it) + α Q_t μ_t + α Q_t(2-Q_t) σ_t^2/2  (equation (21))
- Implications:
  - Richer households prefer private education (term - α Q_t ln(y_it)).
  - For median voter, sign of ln[1/(1 + α(1-Q_t))] + α Q_t(2-Q_t) σ_t^2/2 increases in Q_t; it is negative at Q_t = 0 and positive at Q_t = 1.
  - Majority support for public education depends on governance quality; governance quality is an additional determinant of preference for public vs private education.

### Empirical evidence — data and estimation
- Data: cross-country dataset from Davoodi et al. (2003) for 36 developing and transition countries in the 1990s; sample restricted to 36 countries with BIA in the 1990s (Table A1).
- Two dependent measures of pro-rich bias:
  - Ratio Q5/Q1 = total education spending accruing to wealthiest quintile relative to poorest quintile.
  - Q5 = share of public education benefits accruing to wealthiest quintile (percent).
- Governance measures (rescaled 0–1, higher = better):
  - Gastil: average of civil liberties and political rights (rescaled).
  - Regulatory quality: Kaufmann et al. (1999).
  - Rule of law: Kaufmann et al. (1999).
- Controls considered (added one at a time): Gini (gross income inequality), IGDP (real per capita GDP PPP averaged over 1980s), Credit (private sector credit/GDP), Educ (average years schooling age 25+ in 1990).
- Descriptive statistics (Table 1 — exact figures):
  - Mean Q5/Q1 = 1.76; Q5 = 23.64; Regulatory quality = 0.53; Rule of Law = 0.46; Gastil = 0.59; Gini = 44.11; IGDP = 2611.63; Educ = 4.85; Credit = 24.25.
  - Standard deviation Q5/Q1 = 1.28; Q5 = 9.49; Regulatory quality = 0.09; Rule of Law = 0.10; Gastil = 0.23; Gini = 9.94; IGDP = 1675.14; Educ = 2.32; Credit = 17.90.
  - Maximum Q5/Q1 = 5.02; Q5 = 46.00; Regulatory quality = 0.75; Rule of Law = 0.74; Gastil = 1; Gini = 63.00; IGDP = 6614.40; Educ = 9.26; Credit = 79.04.
  - Minimum Q5/Q1 = 0.11; Q5 = 4.60; Regulatory quality = 0.38; Rule of Law = 0.28; Gastil = 0; Gini = 24.53; IGDP = 508.70; Educ = 0.84; Credit = 3.18.
- Examples from Table A1 (benefit incidence extremes):
  - Chile (CHL): Q1 = 42, Q5 = 4.6, Q5/Q1 = 0.1 (lowest bias).
  - Nepal (NPL): Q1 = 11, Q5 = 46, Q5/Q1 = 4.4 (highest bias).
  - Nicaragua (NIC): Q1 = 9, Q5 = 40, Q5/Q1 = 4.4 (high bias listed).
- OLS results (Tables 2 and 3 — highlights):
  - Coefficient on governance indicators is consistently negative and statistically significant at the 1 percent level across governance indicators and across both bias measures.
  - Gini often positive but not statistically significant in OLS specifications.
  - Additional controls (IGDP, Credit, Educ) do not overturn qualitative results; some are significant in certain specifications (e.g., Educ significant when dependent variable is Q5).
- IV strategy and results (Table 4):
  - Endogeneity concern: used IV estimation, instruments: distance from the equator (Lat) and fraction of population speaking English and European languages.
  - Hansen J-Statistic p-values reported (exact values shown in Table 4): e.g., 0.17, 0.49, 0.37, 0.11, 0.28, 0.14 — unable to reject instrument validity in these specifications.
  - IV results: governance measures remain negative and statistically significant; coefficients on governance now generally significant at 5 and 10 percent levels (some specifications).
  - After instrumenting for governance, coefficients on other controls are not consistently significant.

### Empirical conclusions and implications
- Empirical tests support Proposition 1: lower governance quality is associated with higher pro-rich bias in public education spending.
- Robustness: result holds in both OLS and IV specifications, accounting for controls such as Gini, IGDP, Credit, and Educ.
- Policy-relevant implications derived from analytical model and empirical evidence:
  - Improving governance quality reduces pro-rich incidence bias in public education spending.
  - Better governance reduces future inequality, increases intergenerational mobility, and raises average growth rates.
  - Majority support for public education depends on governance quality; poor governance can shift political support toward regimes that perpetuate inequality.
- Issues for further research:
  - Empirical study of inequality and growth implications of incidence bias requires panel data; current coverage limits such analysis.
  - Extending framework to incidence bias in other public spending areas (health, welfare, infrastructure) to test whether governance quality plays a similar role.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04214.pdf*

### REFERENCES

### _wp04214 - REFERENCES

### References

- Banerjee, A., and A. Newman, 1993, “Occupational Choice and the Process of Development,” Journal of Political Economy, Vol. 101 (April) pp. 274–98.
- Barro, R., 1991, “Economic Growth in a Cross Section of Countries,” Quarterly Journal of Economics, Vol. 106 (May), pp. 407–43.
- ———, and J. Lee, 2001, “International Data on Educational Attainment: Updates and Implications,” Oxford Economic Papers, Vol. 53, No. 3, pp. 541–63.
- Beck, T., A. Demirgüç-Kunt, and R. Levine, 2000, “A New Database on Financial Development and Structure,” World Bank Economic Review, Vol. 14 (September), pp. 597-605.
- Bertocchi, D., and M. Spagat, 2004, “The Evolution of Modern Educational Systems: Technical vs. General Education, Distributional Conflict, and Growth,” Journal of Development Economics, Vol. 73, No. 2, pp. 559–82.
- Birdsall, N., 1997, “Public Spending on Higher Education in Developing Countries: Too Much or Too Little?” Economics of Education Review, Vol. 15, No. 4, pp. 407–19.
- Bourguignon, F., and T. Verdier, 2000, “Oligarchy, Democracy, Inequality and Growth,” Journal of Development Economics, Vol. 62 (August), pp. 285–313.
- Castello, A., and R. Domenech, 2002, “Human Capital Inequality and Economic Growth: Some New Evidence,” Economic Journal, Vol. 112 (March), pp. C187–200.
- Chong, A., and M. Gradstein, 2004, “Inequality and Institutions” (unpublished; Washington: Inter-American Development Bank).
- Davoodi, H.R., E.R.Tiongson, and S.S. Asawanuchit, 2003, “How Useful Are Benefit Incidence Analyses of Public Education and Health Spending,” IMF Working Paper 03/227 (Washington: International Monetary Fund).
- Deininger, K., and L. Squire, 1996, “A New Data Set Measuring Income Inequality,” World Bank Economic Review, Vol. 10 (September), pp. 565–91.
- Doepke, M., and D. de la Croix, 2004a, “Public Versus Private Education when Differential Fertility Matters,” Journal of Development Economics, Vol. 73 (February), pp. 607-29.
- ———, 2004b, “To Segregate or to Integrate: Education Politics and Democracy,” CEPR Discussion Paper No.390 (London: Centre for Economic Policy Research).
- Epple, D., and R. Romano, 1996, “Public Provision of Private Goods,” Journal of Political Economy, Vol. 104 (February), pp. 57–84.
- Esping-Anderson, G., 1990, The Three Worlds of Welfare Capitalism (Cambridge, UK: Polity Press).
- Ferreira, F., 2001, “Education for the Masses?: The Interaction Between Wealth, Educational and Political Inequalities,” Economics of Transition, Vol. 9, No. 2, pp. 533–52.
- Galor, O., and J. Zeira, 1993, “Income Distribution and Macroeconomics,” Review of Economic Studies, Vol. 60, No. 1, pp. 35–52.
- Gelbach, J.B, and L. Pritchett, 2002, “Is More for the Poor Less for the Poor? The Politics of Means-Tested Targeting,” The B.E. Journals in Economic Analysis and Policy, Vol. 2, No. 1, pp. 1–26.
- Gradstein, M., 2004, “Voting on Meritocracy,” European Economic Review, Vol. 48, No. 4, pp. 797–803.
- ———, and M. Justman, 1997, “Democratic Choice of an Education System: Implications for Growth and Income Distribution,” Journal of Economic Growth, Vol. 2, pp. 169-83.
- Gylfason, T., and G. Zoega, 2003, “Education, Social Equality and Economic Growth: A View of the Landscape,” CESifo Economic Studies, Vol. 49, No. 4, pp. 557–79.
- Hall, R., and C. Jones, 1999, “Why Do Some Countries Produce So Much Output Per Worker than Others,” Quarterly Journal of Economics, Vol. 114 (February), pp. 83-116.
- Hwang, J., 2004, “Asset Distribution and Tertiary Education Expenditures in Developing Countries,” Economics of Education Review (forthcoming).
- Knack, S., and P. Keefer, 1995, “Institutions and Economic Performance: Cross-country Tests Using Alternative Institutional Measures,” Economics and Politics, Vol. 7 (November), pp. 207–27.
- La Porta, R., F. Lopes-de-Silanes, A. Shliefer, and R. Vishny, 1998, “Law and Finance,” Journal of Political Economy, Vol. 106 (December), pp. 1113–55.
- Le Grand, J., 1982, The Strategy of Equality: Redistribution and the Social Services (London: George Allen and Unwin).
- Mankiw, G.N., D. Romer, and D.N. Weil, 1992. “A Contribution to the Empirics of Economic Growth,” Quarterly Journal of Economics, Vol. 107 (May), pp. 407–37.
- Meltzer, A.H., and S.F. Richard, 1981, “A Rational Theory of the Size of Government,” Journal of Political Economy, Vol. 79 (October), pp. 914–27.
- Milanovic, B., 2002, Worlds Apart: International and World Inequality 1950–2000 (Washington: World Bank).
- Nitzan, S., 1994, “Modeling Rent-Seeking Contests,” European Journal of Political Economy, Vol. 10 (November), pp. 41–60.
- Ravallion, M., 1999, “Are Poorer States Worse at Targeting their Poor?” Economics Letters, Vol. 65, No. 3, pp. 373–76.
- Rodrik, D., A. Subramanian, and F. Trebbi, 2002, “Institutions Rule: The Primacy of Institutions Over Geography and Integration in Economic Development” (unpublished; Washington: International Monetary Fund).
- Su, X., 2004, “The Allocation of Public Funds in a Hierarchical Education System,” Journal of Economic Dynamics and Control, Vol. 28 (December), pp. 2485–2510.
- World Bank, 2001, World Bank Development Report: Attacking Poverty (New York: Oxford University Press; Washington: World Bank).
- World Bank, 2004, World Bank Development Report: Making Services Work for Poor People (New York: Oxford University Press; Washington: World Bank).

*Source: _wp04214 - REFERENCES.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04214.pdf_