## _wp0721 — Government Effectiveness (Chapter 12) and Appendix II: The Efficiency of Public Investment in Latin America

## Source details

**Canonical URL:** [_wp0721 — Government Effectiveness (Chapter 12) and Appendix II: The Efficiency of Public Investment in Latin America](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp0721.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp0721.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp0721.pdf.json)

---

### Overview / Introduction
- JEL Classification Numbers: E6, H5, O54
- Keywords: Public Expenditure, Fiscal Policy, and Latin America
- Author(s) E-Mail Address: bclements@imf.org; cfaircloth@imf.org; mverhoeven@imf.org
- Scope: assesses trends in public expenditures in Latin America for 17 countries and trends in spending from the mid-1990s to 2006; examines: (i) cyclicality of government spending; (ii) public investment; (iii) public employment; and (iv) social expenditures.
- Sample: 17 countries (Argentina, Bolivia, Brazil, Chile, Colombia, Mexico, Peru, Paraguay, Uruguay, Venezuela, Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama). 2006 data refer to IMF staff projections as of end-August 2006.
- Contextual facts:
  - Public debt ratios remain high at over 50 percent of GDP.
  - Prudent maximum debt for a typical emerging market may be as low as 25 percent of GDP (cited estimate).

### Trends in Public Spending
- Aggregate fiscal trends
  - Fiscal balances weakened in the late 1990s with primary balances deteriorating by over 1 percentage point of GDP.
  - Revenues have risen by an average of about 3½ percentage points of GDP since 2002.
  - Oil producers experienced a boost in receipts of over 7 percent of GDP.
  - Noncommodity revenues rose by about 1 percent of GDP on average since 2002.
  - Public debt (weighted average) projected to remain above 50 percent of GDP by end-2006, little changed from mid-1990s.
- Trends by periods and regions
  - First phase (1995–2001): spending rose by 3 percentage points of GDP (median increase 1½ percent).
  - Downturn and early recovery (2002–04): spending ratios declined; 13 of 17 countries trimmed spending-to-GDP ratios between 2001 and 2004.
  - Second phase (2004–06): outlays rose by about 1½ percentage points of GDP; outlays now surpass 2001 peak ratio to GDP.
  - Country highlights:
    - Brazil: spending rose by 8½ percentage points of GDP in first phase (driven by higher nonwage current outlays).
    - Honduras and Uruguay: outlays rose by over 5½ percent of GDP.
    - Bolivia and Chile: spending rose by about 4½ percent of GDP.
    - Second phase extremes: Venezuela increased over 9 percent of GDP; Bolivia, Brazil, and Nicaragua each 2½ percent of GDP or greater.
  - No convergence: countries with higher initial 1995 spending tended to experience sharper increases (Uruguay exception).
- Real spending and volatility
  - Real outlays accelerated over the past two years, with outlays climbing by an average of 8 percent per annum.
  - 15 of 17 countries increased real outlays by 5 percent or more per annum in 2005–06.
  - Real spending has been volatile and procyclical, following real revenues and the economic cycle.
- Composition of spending
  - Primary current outlays drove the rise in spending-to-GDP ratios.
  - Wage bills rose by about ½ percentage point of GDP over the period (mostly late 1990s).
  - Nonwage current outlays increased by about 3 percentage points of GDP.
  - Capital expenditures:
    - Hovered near 6 percent of GDP in late 1990s, fell nearly 1 percentage point of GDP through 2004.
    - 2006 projected to rise close to ½ percentage point to about 5½ percent of GDP.
    - Share of public investment remains below end-of-1990s levels.
  - Social spending:
    - Broad measure surged by about 2 percentage points of GDP through 2002, then reversed somewhat.
    - Higher spending for education and social insurance/pensions accounted for most of the increase.
    - In five countries (Chile, Colombia, Mexico, Peru, Uruguay) where recent data are available, social spending fell by about ¼ percentage point of GDP on average (because growth outpaced real outlays).

### Key Expenditure Policy Issues
- Major issues examined: the cyclicality of spending; public investment; public employment; social expenditures.

H3 Cyclicality of Government Spending
- Observations and empirical points:
  - Fluctuations in real spending have continued to follow a procyclical pattern.
  - Akitoby and others (2006) find that in about two-thirds of Latin American countries there is a statistically significant short-term procyclical relationship between real output shocks and real primary expenditure.
  - Expenditure especially procyclical in Costa Rica, Guatemala, and Venezuela.
  - By expenditure category:
    - Capital outlays appear most procyclical; excluding an extreme estimate for Bolivia, average coefficient for statistically significant observations is about 4.1 (i.e., a one percent shock to output boosts capital spending by about 4 percent).
    - Spending on goods and services responds less markedly but coefficients for significant observations are generally higher than one.
  - Asymmetry and econometric summary:
    - Regression of primary expenditures-to-GDP on lagged spending, output gap (positive and negative), log terms of trade, and lagged public debt.
    - System GMM preferred estimator; output gap has a statistically significant impact on spending.
    - Differences between positive and negative output gap coefficients are not statistically significant.
    - System GMM estimated coefficient on output gap ~ 0.08.
    - Given observed average changes in output gaps (~4 percentage points for the 17 countries), the model predicts spending ratios would rise by about 0.5 percentage point of GDP between 2002 and 2006; observed increase was 0.8 percentage point of GDP.

H3 Public Investment
- Levels and regional comparison:
  - Public investment averaged about 4½–5 percent of GDP over 1990-2006.
  - Even accounting for private sector participation in infrastructure (~2 percentage points of GDP in available data), total investment lags other regions.
- Drivers and inefficiencies:
  - Decline partly reflected 1990s privatizations not fully offset by private investment.
  - Infrastructure quality varies: Chile, El Salvador, Panama outperform predicted levels by PPP-adjusted per capita income; Paraguay and Argentina have sizable lags.
  - Nonparametric production-function analysis for seven countries finds wide performance range: spending appears most efficient in Chile and Mexico; returns much less favorable in Bolivia and Colombia (with caveats about data in Colombia).
  - Countries relying more heavily on the public sector for provision of infrastructure tend to be less efficient; higher aggregate public infrastructure spending also linked to lower efficiency.
- Contributing institutional factors:
  - Weak public institutions, project selection and appraisal shortcomings, volatility of capital spending causing erratic contractor cash flows and disrupted maintenance.

H3 Public Sector Employment and Government Effectiveness
- Wage bills and employment:
  - General government wages comparable to other regions; Latin American average wages about 7.0 percent of GDP and 32.8 percent of general government expenses.
  - 2005 wages ranged from 12½ percent of GDP in Honduras to 4½ percent of GDP in Nicaragua; Latin American average 7 percent of GDP in 2004–05.
  - Public to manufacturing sector wage ratios in line with other regions.
  - 1990s data suggest public sector employment levels not particularly high relative to other regions.
- Quality of government services:
  - Bureaucratic quality index and World Bank government effectiveness measures show Latin America roughly in line with income-predicted levels but with notable within-region dispersion.
  - Chile and Mexico: bureaucratic quality ~35–40 percent higher than predicted by income.
  - Paraguay and Venezuela: more than one standard deviation below fitted values.
  - Bureaucratic quality increased 1990–1998 but broadly flat since; World Bank measure shows decline in Latin America between 1998 and 2005.
- Empirical relationship:
  - Average wages have no statistically significant effect on government quality after controlling for per-capita income.
  - No significant correlation within Latin America between increases in wage bill (share of GDP) and improvements in bureaucratic quality (1996–2005).
- Policy implication: reforms should focus on institutional weaknesses (patronage, lack of merit-based hiring/promotions, absence of performance evaluation, internal inequities in remuneration).

H3 Social Spending
- Scale and composition:
  - Social spending about 13 percent of GDP, accounting for half of primary government spending.
  - Social spending higher than Emerging Asia but lower than OECD and Eastern Europe & Central Asia.
- Outcomes and efficiency:
  - Education and health indicators broadly in line with development level but lag fast-growing regions; net secondary enrollment around 60 percent (compared with 70 percent in a comparator sample).
  - Repetition rates high; poor performance on international examinations indicates quality weaknesses.
  - Social spending has been procyclical and more volatile than aggregate spending.
- Poverty and targeting:
  - Poverty estimated at 41 percent of the population in 2005, exceeding level predicted by development due to high inequality; regional Gini coefficients average over 0.50.
  - Much social spending has been poorly targeted; on average regressive—poorest 20 percent receiving less than a fifth of benefits of total social outlays in many cases.
  - Incidence by category (poorest quintile ; richest quintile; unweighted averages where available):
    - Education: 20.2 ; 20.4
    - Primary education: 29.0 ; 7.9
    - Secondary education: 13.2 ; 18.3
    - Tertiary education: 1.95 ; 52.1
    - Health: 20.6 ; 17.6
    - Social security: 5.6 ; 51.2
    - Total social spending: 15.0 ; 30.4
  - Distribution varies by country: Chile, Costa Rica, Uruguay more progressive; Bolivia, Peru, Nicaragua more regressive.
- Policy-relevant developments:
  - Targeted conditional cash transfer programs expanded (examples listed in source).
  - These programs have been effective and well-targeted; spending on them remains modest (about 1–1½ percent of GDP) and a small share of total social spending.
- Efficiency opportunity: reorient spending toward programs with higher incidence for the poor (primary education, social assistance), reduce regressivity of social insurance and tertiary subsidies.

### Main Findings and Efficiency Opportunities
- Primary expenditures have risen as a share of GDP over the past ten years, driven mainly by current spending and social expenditures.
- Real spending remains procyclical.
- Capital expenditures generally low; despite 2006 uptick, public investment remains low relative to most developing regions.
- There is substantial scope to improve the efficiency of:
  - Public investment
  - Public employment
  - Social spending
- Potential efficiency gains from adopting best practices in project appraisal, implementation, and institutional strengthening.

### Policy Recommendations (enumerated)
- Limit procyclicality of spending:
  - Strengthen political commitment to restrict spending growth during good times.
  - Consider explicit ceilings for expenditure growth while leaving automatic stabilizers on revenues free to operate.
  - Use well-designed fiscal rules (for expenditure growth, deficits, debt) with enforceable sanctions and political commitment.
- Reduce public debt and improve debt structure:
  - Reduce public debt toward prudent levels to lower crisis probability and the need for contractionary fiscal adjustments.
  - Lengthen debt maturities and increase reliance on domestic-currency debt where feasible.
- Improve efficiency of public investment:
  - Strengthen project selection, appraisal, and management.
  - Increase reliance on private sector provision where efficiency gains are likely, complemented by stronger institutional frameworks and management of public-private partnership fiscal risks.
  - Stabilize capital spending to avoid erratic flows that impair implementation and maintenance.
- Reform public employment and bureaucratic quality:
  - Focus reforms on institutional weak spots: reduce patronage, establish merit-based hiring and promotions, implement performance evaluation, and address pay inequities.
  - Higher average wages alone unlikely to improve service quality without institutional reforms.
- Reorient social spending toward pro-poor outcomes:
  - Reform public social insurance to reduce excessive generosity and move schemes toward actuarial soundness.
  - Consider increased user fees for higher education combined with subsidies for low-income families to preserve access.
  - Improve secondary education quality to reduce repetition rates.
  - Expand well-targeted conditional cash-transfer programs and increase their share where fiscal space allows.
- Reduce expenditure volatility:
  - Pursue policies to reduce stop-and-go capital spending and smooth social sector outlays to improve program effectiveness over the long term.

### Appendix II — The Efficiency of Public Investment in Latin America

H3 Methodology
- Objective: assess efficiency of public spending by relating public investment to infrastructure outputs (rails, roads, electricity, water, telecommunications).
- Technique: Free Disposable Hull Analysis (FDH), a nonparametric efficiency frontier method.
  - Efficient countries receive efficiency score = 1.
  - For less efficient country A relative to efficient country E, input efficiency score = (Espending)/(Aspending). This score < 1.
  - Interpretation: input efficiency score = minimum level of spending (as a share of actual investment spending) that other countries have needed to achieve same or higher infrastructure outputs.
  - Maximum potential savings from efficiency enhancement (narrow interpretation) = 1 − input efficiency score.
- Advantages: not an econometric exercise; no functional form imposed on frontier.
- Caveats and limitations:
  - FDH assumes inputs and outputs related through a production process and common production technology.
  - Measures technical efficiency narrowly; results sensitive to choice of inputs/outputs, sample selection, and measurement error.
  - Differences in production technologies, policy objectives, unidentified inputs, and exogenous factors (for example, geography) can affect rankings.
  - Exercise largely based on data through 2001; interpret rankings cautiously.
- Private sector participation:
  - Primary input is total spending (public + private) to reflect sizeable private participation.
  - Adjusted efficiency scores correct for private sector participation via a two-step truncated regression and score adjustment.

H3 Data
- Coverage:
  - Period: 1990s and early 2000s (largely through 2001).
  - Countries: Argentina, Bolivia, Brazil, Chile, Colombia, Mexico, Peru.
  - Sectors: rails, roads, electricity, water, telecommunications.
- Appendix Table 1 — Infrastructure Spending in Latin America, 1991–2001 (Average, in percent of GDP):
  - Argentina: Total 1.5; Public 0.4; Private 1.2
  - Bolivia: Total 7.4; Public 3.2; Private 4.2
  - Brazil: Total 2.4; Public 1.3; Private 1.1
  - Chile: Total 4.6; Public 1.6; Private 3.0
  - Colombia: Total 5.0; Public 3.5; Private 1.5
  - Mexico: Total 1.7; Public 0.6; Private 1.1
  - Peru: Total 1.7; Public 0.6; Private 1.1
- Appendix Table 2 — Infrastructure outcome average annual changes (country highlights; in percent):
  - Argentina: Rail lines -0.76; Roads -1.18; Paved roads -0.77; Electricity losses -0.81; Electric power cons. (kWh per capita) 5.77; Main telephone lines (per 1000 workers) 10.55
  - Bolivia: Rail lines -1.92; Roads -0.10; Paved roads 7.33; Electricity losses -0.66; Electricity access by households 2.77; Electric power cons. (kWh per capita) 4.83; Water access 1.50; Main telephone lines (per 1000 workers) 11.64
  - Brazil: Rail lines -0.48; Roads -0.92; Paved roads -5.58; Electricity losses 0.79; Electricity access by households 6.29; Electric power cons. (kWh per capita) 2.61; Water access 0.60; Main telephone lines (per 1000 workers) 20.63
  - Chile: Rail lines -6.28; Roads -1.53; Paved roads 3.18; Electricity losses -4.26; Electric power cons. (kWh per capita) 10.21; Water access 0.46; Main telephone lines (per 1000 workers) 16.30
  - Colombia: Rail lines 0.01; Roads -1.08; Paved roads 1.69; Electricity losses -0.71; Electricity access by households 0.59; Electric power cons. (kWh per capita) 0.04; Water access 0.00; Main telephone lines (per 1000 workers) 12.88
  - Mexico: Rail lines 0.68; Roads 2.98; Paved roads 1.66; Electricity losses 1.32; Electric power cons. (kWh per capita) 4.85; Water access 1.15; Main telephone lines (per 1000 workers) 9.91
  - Peru: Rail lines 0.59; Roads 0.39; Paved roads 4.29; Electricity losses -0.47; Electricity access by households -0.14; Electric power cons. (kWh per capita) 4.19; Water access 0.79; Main telephone lines (per 1000 workers) 14.03
- Observations:
  - Chile achieved the most noteworthy gains in road, electricity, and telephones.
  - Countries with relatively high spending (Bolivia and Colombia) did not necessarily achieve the most rapid progress in outcomes.

H3 Results
- Aggregation: FDH efficiency scores computed for each sector and aggregated using each sector’s share in public sector investment spending; both unadjusted and adjusted (for private participation) scores reported.
- Main findings:
  - Large inefficiencies exist; many countries display significant gaps from the efficiency frontier (score = 1.0).
  - Inefficiencies especially noteworthy in Bolivia and Colombia.
  - Chile and Mexico rank among the most efficient countries.
    - Mexico: high Public Efficiency 0.974; Rank 1; Level of Public Efficiency: High; Adjusted Public Efficiency 0.824; Adjusted Rank 1; Adjusted Level of Public Efficiency: High.
    - Argentina: Public Efficiency 0.804; Rank 2; Level of Public Efficiency: High; Adjusted Public Efficiency 0.530; Adjusted Rank 4; Adjusted Level: Medium.
    - Chile: Public Efficiency 0.802; Rank 3; Level of Public Efficiency: High; Adjusted Public Efficiency 0.732; Adjusted Rank 2; Adjusted Level: High.
    - Peru: Public Efficiency 0.655; Rank 4; Level: Medium; Adjusted 0.598; Adjusted Rank 3; Level: Medium.
    - Brazil: Public Efficiency 0.410; Rank 5; Level: Medium; Adjusted 0.432; Adjusted Rank 5; Level: Medium.
    - Bolivia: Public Efficiency 0.218; Rank 6; Level: Low; Adjusted Public Efficiency 0.036; Adjusted Rank 7; Level: Low.
    - Colombia: Public Efficiency 0.202; Rank 7; Level: Low; Adjusted Public Efficiency 0.253; Adjusted Rank 6; Level: Low.
  - Adjustment for differing levels of private sector participation has little effect on country rankings.
  - Caution: results for Colombia should be interpreted with caution.

### Summary and Policy Implications
- Main findings recap:
  - Primary expenditures trended upward since mid-1990s, driven by increases in current spending (notably nonwage outlays and social insurance).
  - Capital expenditures generally low; public investment remains low relative to most developing regions despite 2006 uptick.
  - Real spending growth has been rapid and procyclical, constraining faster declines in public debt.
  - Efficiency of government expenditure varies widely; large potential gains from better project appraisal, implementation, and institutional strengthening.
  - Social spending is large but often regressive and volatile; targeted conditional transfers show promise.
- Recommended priorities:
  - Limit procyclicality of spending through political commitment, fiscal rules, and expenditure ceilings.
  - Reduce public debt and improve debt structure (lengthen maturities, increase domestic-currency debt where feasible).
  - Improve public investment efficiency via stronger project appraisal, management, private participation where appropriate, and stabilizing capital spending.
  - Reform public employment by addressing patronage, introducing merit-based systems, performance evaluation, and fixing pay inequities.
  - Reorient social spending toward pro-poor programs (primary education, social assistance), reform social insurance, and expand well-targeted conditional cash-transfer programs.
  - Reduce expenditure volatility to improve long-term program effectiveness.

*Source: _wp0721 — Chapter 12 “Government Effectiveness” and Appendix II “The Efficiency of Public Investment in Latin America” (IMF staff analysis provided).*

### 2006. It also examines key policy issues, including the cyclicality of spending, public

### Public Expenditure, Fiscal Policy, and Latin America

### Overview / Introduction
- JEL Classification Numbers: E6, H5, O54
- Keywords: Public Expenditure, Fiscal Policy, and Latin America
- Author(s) E-Mail Address: bclements@imf.org; cfaircloth@imf.org; mverhoeven@imf.org

### Trends in Public Spending
- Primary expenditures have trended upward for the past ten years as a share of GDP, driven by increases in current spending, in particular for social expenditures.
- Figures and tables referenced in the source document covering trends include:
  - Figure 1: Fiscal Trends
  - Figure 2: Primary Public Spending by Region
  - Figure 3: Initial Primary Spending Levels (1995) and Increases (1995–2006)
  - Figure 4: Real GDP, Revenues, and Primary Spending
  - Figure 5: Public Sector Expenditures
  - Figure 6: Average Public Sector Capital Spending
  - Figure 7: Trends in Social Spending
  - Table 3: Public Investment in Latin America and Other Regions, 1990–2006
  - Table 6: Government Effectiveness in Latin America and Other Regions
  - Table 7: General Government Social Spending in Latin America and Other Regions, 2004
  - Table 8: Distribution of Benefits from Social Spending to the Top and Bottom Quintiles

Key empirical points (as presented):
- Time spans and comparisons explicitly used: 1990–2006; 1995; 1995–2006; 2004.
- Aggregate fiscal trend material is organized under:
  - II.A Overview of Aggregate Fiscal Trends
  - II.B Trends in Government Expenditure

### Key Expenditure Policy Issues
- The paper examines key policy issues, including:
  - The cyclicality of spending
  - Public investment
  - Public employment
  - Social expenditures

A. The Cyclicality of Government Spending
- Fluctuations in real spending have continued to follow a procyclical pattern.
- Empirical assessments and summaries include:
  - Table 1: Estimates of the Short-Run Response of Spending to Output Shocks
  - Table 2: Response of Expenditure to Output Gaps
  - Appendix I: Econometric Methodology Used to Assess the Cyclicality of Spending

B. Public Investment
- The analysis addresses the level, composition, and efficiency of public investment.
- Supporting exhibits and appendices:
  - Table 3: Public Investment in Latin America and Other Regions, 1990–2006
  - Figure 8: Institutional and Infrastructural Performance—Deviation of Actual Indices from Values Predicted by PPP-Adjusted Per Capita Income
  - Figure 9: Public Share of Infrastructure Spending and Efficiency
  - Appendix II: The Efficiency of Public Investment in Latin America

C. Public Sector Employment
- Public sector employment and wage dynamics are examined with comparison across regions.
- Relevant tables and figures:
  - Table 4: General Government Wages and Salaries in Latin America and Other Regions, 2004
  - Table 5: Government Employment in Latin America and Other Regions, 1990s
  - Figure 10: Relationship Between GDP Per Capita and Bureaucratic Quality
  - Figure 11: The Quality of the Bureaucracy

D. Social Spending
- Social expenditures are identified as a major driver of the increase in current spending.
- Key material:
  - Figure 7: Trends in Social Spending
  - Table 7: General Government Social Spending in Latin America and Other Regions, 2004
  - Table 8: Distribution of Benefits from Social Spending to the Top and Bottom Quintiles

### Main Findings and Efficiency Opportunities
- Primary expenditures have risen as a share of GDP over the past ten years, driven mainly by current spending and social expenditures.
- Real spending remains procyclical.
- There is substantial scope to improve the efficiency of:
  - Public investment
  - Public employment
  - Social spending

### Structure of the Report
- Main sections:
  - I. Introduction
  - II. Trends in Public Spending
    - A. Overview of Aggregate Fiscal Trends
    - B. Trends in Government Expenditure
  - III. Key Expenditure Policy Issues
    - A. The Cyclicality of Government Spending
    - B. Public Investment
    - C. Public Sector Employment
    - D. Social Spending
  - IV. Summary and Policy Implications
- Appendices:
  - Appendix I: Econometric Methodology Used to Assess the Cyclicality of Spending
  - Appendix II: The Efficiency of Public Investment in Latin America

*Source: _wp0721 - 2006 (IMF working paper content provided).*

### 12.    Government    Effectiveness .....................................................................................

### _wp0721 - 12.    Government    Effectiveness .....................................................................................

### Introduction
- Scope: assesses trends in public expenditures in Latin America for 17 countries and trends in spending from the mid-1990s to 2006, and examines: (i) cyclicality of government spending; (ii) public investment; (iii) public employment; and (iv) social expenditures.
- Sample: 17 countries (Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Mexico, Peru, Paraguay, Uruguay, Venezuela, Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama). 2006 data refer to IMF staff projections as of end-August 2006.
- Contextual facts:
  - Public debt ratios remain high at over 50 percent of GDP.
  - Prudent maximum debt for a typical emerging market may be as low as 25 percent of GDP (cited estimate).

### Trends in Public Spending
- Aggregate fiscal trends
  - Fiscal balances weakened in the late 1990s with primary balances deteriorating by over 1 percentage point of GDP.
  - Revenues have risen by an average of about 3½ percentage points of GDP since 2002.
  - Oil producers experienced a boost in receipts of over 7 percent of GDP.
  - Noncommodity revenues rose by about 1 percent of GDP on average since 2002.
  - Higher primary surpluses and macroeconomic improvements underpinned the region’s expansion.
  - Public debt (weighted average) projected to remain above 50 percent of GDP by end-2006, little changed from mid-1990s.

- Trends by periods and regions
  - First phase (1995–2001): spending rose by 3 percentage points of GDP (median increase 1½ percent).
  - Downturn and early recovery (2002–04): spending ratios declined; 13 of 17 countries trimmed spending-to-GDP ratios between 2001 and 2004.
  - Second phase (2004–06): outlays rose by about 1½ percentage points of GDP; outlays now surpass 2001 peak ratio to GDP.
  - Country highlights:
    - Brazil: spending rose by 8½ percentage points of GDP in first phase (driven by higher nonwage current outlays).
    - Honduras and Uruguay: outlays rose by over 5½ percent of GDP.
    - Bolivia and Chile: spending rose by about 4½ percent of GDP.
    - Second phase extremes: Venezuela increased over 9 percent of GDP; Bolivia, Brazil, and Nicaragua each 2½ percent of GDP or greater.
  - No convergence: countries with higher initial 1995 spending tended to experience sharper increases (Uruguay exception).

- Real spending and volatility
  - Real outlays accelerated over the past two years, with outlays climbing by an average of 8 percent per annum.
  - 15 of 17 countries increased real outlays by 5 percent or more per annum in 2005–06.
  - Real spending has been volatile and procyclical, following real revenues and the economic cycle.

- Composition of spending
  - Primary current outlays drove the rise in spending-to-GDP ratios.
  - Wage bills rose by about ½ percentage point of GDP over the period (mostly late 1990s).
  - Nonwage current outlays increased by about 3 percentage points of GDP.
  - Capital expenditures:
    - Hovered near 6 percent of GDP in late 1990s, fell nearly 1 percentage point of GDP through 2004.
    - 2006 projected to rise close to ½ percentage point to about 5½ percent of GDP.
    - Share of public investment remains below end-of-1990s levels.
  - Social spending:
    - Broad measure surged by about 2 percentage points of GDP through 2002, then reversed somewhat.
    - Higher spending for education and social insurance/pensions accounted for most of the increase.
    - In five countries (Chile, Colombia, Mexico, Peru, Uruguay) where recent data are available, social spending fell by about ¼ percentage point of GDP on average (because growth outpaced real outlays).

### Key Expenditure Policy Issues
- A. Cyclicality of Government Spending
  - Procyclical fiscal policy has limited faster public debt reduction during recoveries.
  - Country evidence:
    - Akitoby and others (2006) find that in about two-thirds of Latin American countries there is a statistically significant short-term procyclical relationship between real output shocks and real primary expenditure.
    - Expenditure especially procyclical in Costa Rica, Guatemala, and Venezuela.
  - By expenditure category:
    - Capital outlays appear most procyclical. Excluding an extremely high estimate for Bolivia, the average coefficient for statistically significant observations is about 4.1 (i.e., a one percent shock to output boosts capital spending by about 4 percent).
    - Spending on goods and services responds less markedly but coefficients for significant observations are generally higher than one, indicating more-than-proportional response to shocks.
  - Regional comparison:
    - Share of countries with procyclical spending is higher in Latin America than for developing countries overall; average coefficient values are somewhat higher.
  - Asymmetry test (Table 2 results summary):
    - Regression of primary expenditures-to-GDP on lagged spending, output gap (positive and negative), log terms of trade, and lagged public debt.
    - System GMM preferred estimator; output gap has a statistically significant impact on spending.
    - Differences between positive and negative output gap coefficients are not statistically significant.
    - System GMM estimated coefficient on output gap ~ 0.08.
    - Given observed average changes in output gaps (~4 percentage points for the 17 countries), the model predicts spending ratios would rise by about 0.5 percentage point of GDP between 2002 and 2006; observed increase was 0.8 percentage point of GDP.

- B. Public Investment
  - Public investment low relative to other regions:
    - Public investment averaged about 4½–5 percent of GDP over 1990-2006.
    - Even accounting for private sector participation in infrastructure (~2 percentage points of GDP in available data), total investment lags other regions.
  - Drivers and inefficiencies:
    - Decline partly reflected 1990s privatizations not fully offset by private investment.
    - Infrastructure quality varies by country; Chile, El Salvador, Panama outperform predicted levels by PPP-adjusted per capita income; Paraguay and Argentina have sizable lags.
    - Nonparametric production function analysis for seven countries finds wide performance range: spending appears most efficient in Chile and Mexico; returns much less favorable in Bolivia and Colombia (with caveats about data in Colombia).
    - Countries relying more heavily on the public sector for provision of infrastructure tend to be less efficient; higher aggregate public infrastructure spending also linked to lower efficiency.
  - Contributing factors to inefficiency:
    - Weak public institutions, project selection and appraisal shortcomings, volatility of capital spending leading to erratic contractor cash flows and disrupted maintenance.

- C. Public Sector Employment and Government Effectiveness
  - Public sector wage bills:
    - General government wages comparable to other regions; Latin American average wages about 7.0 percent of GDP and 32.8 percent of general government expenses (regional table summaries).
    - Wide variance: 2005 wages ranged from 12½ percent of GDP in Honduras to 4½ percent of GDP in Nicaragua; Latin American average 7 percent of GDP in 2004–05.
    - Public to manufacturing sector wage ratios in line with other regions.
  - Employment levels:
    - 1990s data suggest public sector employment levels not particularly high relative to other regions.
  - Quality of government services:
    - Bureaucratic quality index and World Bank government effectiveness measures show Latin America roughly in line with income-predicted levels but with notable within-region dispersion.
    - Chile and Mexico: bureaucratic quality ~35–40 percent higher than predicted by income; Paraguay and Venezuela: more than one standard deviation below fitted values.
    - Bureaucratic quality increased 1990–1998 but has been broadly flat since; the World Bank measure shows decline in Latin America between 1998 and 2005.
  - Implication: average wages have no statistically significant effect on government quality after controlling for per-capita income; no significant correlation within Latin America between increases in wage bill (share of GDP) and improvements in bureaucratic quality (1996–2005).
  - Recommended focus: tackle institutional weaknesses (patronage, lack of merit-based hiring/promotions, absence of performance evaluation, internal inequities in remuneration).

- D. Social Spending
  - Scale and composition:
    - Social spending about 13 percent of GDP, accounting for half of primary government spending.
    - Social spending higher than Emerging Asia but lower than OECD and Eastern Europe & Central Asia.
  - Outcomes and efficiency:
    - Education and health indicators broadly in line with development level but lag fast-growing regions; net secondary enrollment around 60 percent (compared with 70 percent in a comparator sample).
    - Repetition rates high; poor performance on international examinations indicates quality weaknesses.
    - Social spending has been procyclical and more volatile than aggregate spending, hindering long-term gains that require sustained effort.
  - Poverty and targeting:
    - Poverty estimated at 41 percent of the population in 2005, exceeding level predicted by development due to high inequality; regional Gini coefficients average over 0.50.
    - Much social spending has been poorly targeted; on average regressive—poorest 20 percent receiving less than a fifth of benefits of total social outlays in many cases.
    - Incidence by category (poorest quintile vs richest quintile; unweighted averages where available):
      - Education: 20.2 ; 20.4
      - Primary education: 29.0 ; 7.9
      - Secondary education: 13.2 ; 18.3
      - Tertiary education: 1.95 ; 52.1
      - Health: 20.6 ; 17.6
      - Social security: 5.6 ; 51.2
      - Total social spending: 15.0 ; 30.4
    - Distribution varies by country: Chile, Costa Rica, Uruguay more progressive; Bolivia, Peru, Nicaragua more regressive.
  - Policy-relevant developments:
    - Targeted conditional cash transfer programs expanded (e.g., Jefes y Jefas de Hogar, Familias, Bolsa Familia, Chile Solidario, Familias en Acción, Oportunidades).
    - These programs have been effective and well-targeted; spending on them remains modest (about 1–1½ percent of GDP) and a small share of total social spending.
  - Efficiency opportunities: reorient spending toward programs with higher incidence for the poor (primary education, social assistance), reduce regressivity of social insurance and tertiary subsidies.

### Summary and Policy Implications
- Main findings
  - Primary expenditures have trended upward since mid-1990s, driven by increases in current spending (notably nonwage outlays and social insurance).
  - Capital expenditures generally low; despite 2006 uptick, public investment remains low relative to most developing regions.
  - Real spending growth has been rapid and procyclical, constraining faster declines in public debt.
  - Efficiency of government expenditure varies widely: large potential gains from adopting best practices in project appraisal, implementation, and institutional strengthening.
  - Social spending is large but often regressive and volatile; targeted conditional transfers show promise.

- Policy recommendations (enumerated)
  - Limit procyclicality of spending:
    - Strengthen political commitment to restrict spending growth during good times.
    - Consider explicit ceilings for expenditure growth while leaving automatic stabilizers on revenues free to operate.
    - Use well-designed fiscal rules (for expenditure growth, deficits, debt) with enforceable sanctions and political commitment.
  - Reduce public debt and improve debt structure:
    - Reduce public debt toward prudent levels to lower crisis probability and the need for contractionary fiscal adjustments.
    - Lengthen debt maturities and increase reliance on domestic-currency debt where feasible.
  - Improve efficiency of public investment:
    - Strengthen project selection, appraisal, and management.
    - Increase reliance on private sector provision where efficiency gains are likely, complemented by stronger institutional frameworks and management of public-private partnership fiscal risks.
    - Stabilize capital spending to avoid erratic flows that impair implementation and maintenance.
  - Reform public employment and bureaucratic quality:
    - Focus reforms on institutional weak spots: reduce patronage, establish merit-based hiring and promotions, implement performance evaluation, and address pay inequities.
    - Higher average wages alone unlikely to improve service quality without institutional reforms.
  - Reorient social spending toward pro-poor outcomes:
    - Reform public social insurance to reduce excessive generosity and move schemes toward actuarial soundness.
    - Consider increased user fees for higher education combined with subsidies for low-income families to preserve access.
    - Improve secondary education quality to reduce repetition rates.
    - Expand well-targeted conditional cash-transfer programs and increase their share where fiscal space allows.
  - Reduce expenditure volatility:
    - Pursue policies to reduce stop-and-go capital spending and smooth social sector outlays to improve program effectiveness over the long term.

### Appendix I — Econometric methodology (brief)
- Short-run elasticity estimates (Akitoby and others, 2006): error-correction model estimating short-run elasticity of government spending with respect to output (coefficient β0 in model).
- Asymmetry and cyclicality tests: regression specification follows modified IMF (2007) and Balassone and Francese (2004) frameworks, estimating primary expenditures-to-GDP on lagged spending, log terms of trade, lagged public debt, and output gap; positive and negative output gap coefficients tested separately; system GMM used to address endogeneity.

*Source: IMF staff analysis in chapter "12. Government Effectiveness" from the provided PDF content.*

### APPENDIX II.  THE EFFICIENCY OF PUBLIC INVESTMENT IN LATIN AMERICA

### APPENDIX II.  THE EFFICIENCY OF PUBLIC INVESTMENT IN LATIN AMERICA

### A. Methodology
- Objective: assess efficiency of public spending by relating public investment to infrastructure outputs (rails, roads, electricity, water, telecommunications).
- Technique: Free Disposable Hull Analysis (FDH), a nonparametric efficiency frontier method.
  - Efficient countries (produce more outputs for equal or less spending) receive efficiency score = 1.
  - For less efficient country A relative to efficient country E, input efficiency score = (Espending)/(Aspending). This score < 1.
  - Interpretation: input efficiency score = minimum level of spending (as a share of actual investment spending) that other countries have needed to achieve same or higher infrastructure outputs.
  - Maximum potential savings from efficiency enhancement (narrow interpretation) = 1 − input efficiency score.
- Advantages of FDH:
  - Not an econometric exercise; no functional form imposed on frontier.
- Key caveats and limitations:
  - FDH assumes chosen inputs and outputs are related through a production process and that countries share the same production technology.
  - Measures technical efficiency narrowly; results sensitive to choice of inputs/outputs, sample selection, and measurement error.
  - Differences in production technologies, policy objectives, unidentified inputs, and exogenous factors (for example, geography) can affect rankings.
  - Exercise largely based on data through 2001; interpret rankings cautiously.
- Private sector participation:
  - Private sector undertakes a sizeable share of infrastructure investment in Latin America; thus the study uses total spending (public + private) as primary input.
  - An adjusted measure corrects for private sector participation’s impact on efficiency rankings by:
    - Step 1: running a truncated regression of efficiency on private sector share across countries.
    - Step 2: estimating corrected efficiency score for each sector using regression and country private share, implying downward adjustment for countries with high private participation.

### B. Data
- Coverage:
  - Period: 1990s and early 2000s (largely through 2001).
  - Countries: Argentina, Bolivia, Brazil, Chile, Colombia, Mexico, Peru.
  - Sectors: rails, roads, electricity, water, telecommunications.
- Spending data source: Calderón and Servén (2004), with updates by Luis Andres (World Bank).
- Appendix Table 1 — Infrastructure Spending in Latin America, 1991–2001 (Average, in percent of GDP) 1/
  - Note: Includes spending on rail, roads, electricity, water, and telecommunications.
  - Argentina: Total 1.5; Public 0.4; Private 1.2
  - Bolivia: Total 7.4; Public 3.2; Private 4.2
  - Brazil: Total 2.4; Public 1.3; Private 1.1
  - Chile: Total 4.6; Public 1.6; Private 3.0
  - Colombia: Total 5.0; Public 3.5; Private 1.5
  - Mexico: Total 1.7; Public 0.6; Private 1.1
  - Peru: Total 1.7; Public 0.6; Private 1.1
- Infrastructure outcome changes (Appendix Table 2 — Average annual change, in percent):
  - Variables reported: Rail lines (km/1000 capita), Roads (meters/capita), Paved roads (meters/capita), Electricity losses (% of total output), Electricity access by households (% of total pop), Electric power cons. (kWh per capita), Water access (% of total pop), Main telephone lines (per 1000 workers).
  - Country-specific average annual changes (periods vary by indicator as in source):
    - Argentina:
      - Rail lines: -0.76
      - Roads: -1.18
      - Paved roads: -0.77
      - Electricity losses: -0.81
      - Electric power cons. (kWh per capita): 5.77
      - Main telephone lines (per 1000 workers): 10.55
    - Bolivia:
      - Rail lines: -1.92
      - Roads: -0.10
      - Paved roads: 7.33
      - Electricity losses: -0.66
      - Electricity access by households: 2.77
      - Electric power cons. (kWh per capita): 4.83
      - Water access: 1.50
      - Main telephone lines (per 1000 workers): 11.64
    - Brazil:
      - Rail lines: -0.48
      - Roads: -0.92
      - Paved roads: -5.58
      - Electricity losses: 0.79
      - Electricity access by households: 6.29
      - Electric power cons. (kWh per capita): 2.61
      - Water access: 0.60
      - Main telephone lines (per 1000 workers): 20.63
    - Chile:
      - Rail lines: -6.28
      - Roads: -1.53
      - Paved roads: 3.18
      - Electricity losses: -4.26
      - Electric power cons. (kWh per capita): 10.21
      - Water access: 0.46
      - Main telephone lines (per 1000 workers): 16.30
    - Colombia:
      - Rail lines: 0.01
      - Roads: -1.08
      - Paved roads: 1.69
      - Electricity losses: -0.71
      - Electricity access by households: 0.59
      - Electric power cons. (kWh per capita): 0.04
      - Water access: 0.00
      - Main telephone lines (per 1000 workers): 12.88
    - Mexico:
      - Rail lines: 0.68
      - Roads: 2.98
      - Paved roads: 1.66
      - Electricity losses: 1.32
      - Electric power cons. (kWh per capita): 4.85
      - Water access: 1.15
      - Main telephone lines (per 1000 workers): 9.91
    - Peru:
      - Rail lines: 0.59
      - Roads: 0.39
      - Paved roads: 4.29
      - Electricity losses: -0.47
      - Electricity access by households: -0.14
      - Electric power cons. (kWh per capita): 4.19
      - Water access: 0.79
      - Main telephone lines (per 1000 workers): 14.03
  - Observations from data:
    - Chile achieved the most noteworthy gains in road, electricity, and telephones.
    - Countries with relatively high spending (Bolivia and Colombia) did not necessarily achieve the most rapid progress in outcomes.

### C. Results
- Aggregation method:
  - FDH efficiency scores computed for each sector (rails, roads, electricity, water, telecommunications) and aggregated using each sector’s share in public sector investment spending.
  - Both unadjusted (public efficiency) and adjusted (corrected for private sector participation) scores reported.
- Main findings:
  - Overall, large inefficiencies exist; many countries display significant gaps from the efficiency frontier (score = 1.0).
  - Inefficiencies are especially noteworthy in Bolivia and Colombia.
  - Chile and Mexico rank among the most efficient countries.
    - Chile’s high efficiency attributed to substantial progress in improving infrastructure indicators.
    - Mexico’s high efficiency attributed to continued progress in output indicators combined with low levels of expenditures.
  - Adjustment for differing levels of private sector participation has little effect on country rankings.
  - Caution: results for Colombia should be interpreted with caution (see main text for further discussion).
- Appendix Table 3 — Aggregated Public Efficiency Scores
  - Mexico:
    - Public Efficiency: 0.974
    - Rank: 1
    - Level of Public Efficiency: High
    - Adjusted Public Efficiency: 0.824
    - Adjusted Rank: 1
    - Adjusted Level of Public Efficiency: High
  - Argentina:
    - Public Efficiency: 0.804
    - Rank: 2
    - Level of Public Efficiency: High
    - Adjusted Public Efficiency: 0.530
    - Adjusted Rank: 4
    - Adjusted Level of Public Efficiency: Medium
  - Chile:
    - Public Efficiency: 0.802
    - Rank: 3
    - Level of Public Efficiency: High
    - Adjusted Public Efficiency: 0.732
    - Adjusted Rank: 2
    - Adjusted Level of Public Efficiency: High
  - Peru:
    - Public Efficiency: 0.655
    - Rank: 4
    - Level of Public Efficiency: Medium
    - Adjusted Public Efficiency: 0.598
    - Adjusted Rank: 3
    - Adjusted Level of Public Efficiency: Medium
  - Brazil:
    - Public Efficiency: 0.410
    - Rank: 5
    - Level of Public Efficiency: Medium
    - Adjusted Public Efficiency: 0.432
    - Adjusted Rank: 5
    - Adjusted Level of Public Efficiency: Medium
  - Bolivia:
    - Public Efficiency: 0.218
    - Rank: 6
    - Level of Public Efficiency: Low
    - Adjusted Public Efficiency: 0.036
    - Adjusted Rank: 7
    - Adjusted Level of Public Efficiency: Low
  - Colombia:
    - Public Efficiency: 0.202
    - Rank: 7
    - Level of Public Efficiency: Low
    - Adjusted Public Efficiency: 0.253
    - Adjusted Rank: 6
    - Adjusted Level of Public Efficiency: Low

_Italic: Source — APPENDIX II. THE EFFICIENCY OF PUBLIC INVESTMENT IN LATIN AMERICA (content unit: _wp0721 - APPENDIX II. THE EFFICIENCY OF PUBLIC INVESTMENT IN LATIN AMERICA)_

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp0721.pdf_