## wpiea2022048-print-pdf

## Source details

**Canonical URL:** [wpiea2022048-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022048-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022048-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022048-print-pdf.pdf.json)

---

### Key findings
- During the COVID-19 response, selected advanced economies (AEs) increased health expenditures by an average of 1.5 percent of GDP in 2020, while the increase averaged 0.9 percent of GDP in emerging markets (EMs) and in low-income developing countries (LIDCs).
- Previous IMF work estimates that LIDCs would need to increase, on average, their annual health spending by 4.6 percent of GDP by 2030 to meet the Sustainable Development Goal on health (SDG3).
- AEs have higher life expectancy and a higher composite index of health outcomes than EMs and LIDCs.
- Using bias-corrected Data Envelopment Analysis (DEA), the study finds significant cross-country variation in health spending efficiency:
  - LIDCs lose on average 8.2 life-years due to inefficient use of health resources; Sub-Saharan Africa loses 10.5 life-years on average.
  - EMs waste the greatest share of health resources by income group at 1.4 percent of GDP; the Caucasus and Central Asia region wastes 1.9 percent of GDP.
- Over 2007–17, the vast majority of countries’ health systems became more efficient; LIDCs made notable gains in achieving improved health outcomes for their spending levels.
- Both efficiency and technological progress rose over 2007–17 as reflected in gains in the Malmquist productivity index.

### Methodology
- Efficiency scores are derived using bias-corrected Data Envelopment Analysis (DEA) with bootstrap correction (200 bootstrap samples) to address measurement bias (Simar and Wilson, 2007).
- Outputs used include life expectancy, health-adjusted life expectancy (HALE), and a composite health index constructed from eight health outcome indicators (maternal mortality rate, neonatal mortality rate, under-5 mortality rate, incidence of tuberculosis, HIV incidence, cardiovascular/cancer incidence, adolescent fertility rate, and HALE).
- Both input- and output-oriented efficiency scores are estimated; output-oriented uses maximization over δ with inverse 1/δ as technical output-efficiency score (ranges 0 to 1); input-oriented minimizes δ as input-efficiency score (ranges 0 to 1).
- Bootstrapping applied to correct DEA’s upward bias and enable inferential methodology (Simar and Wilson, 2007; Bogetoft and Otto, 2011).

### Trends and quantitative patterns
- The relationship between health spending and life expectancy exhibits concavity: additional spending yields smaller gains in life-years in AEs versus regions like Sub-Saharan Africa.
- From 2007–17:
  - 146 countries became more output-efficient; 136 countries became more input-efficient.
  - 27 countries saw a deterioration in output-efficiency; 37 countries saw a deterioration in input-efficiency.
  - Many countries show both efficiency improvements and technological progress (Malmquist index); several LIDCs and EMs experienced technological regress while showing relative efficiency gains.
- Correlations of efficiency measures across outputs:
  - Correlation of output-oriented efficiency scores (life expectancy vs HALE): 0.94.
  - Correlation of input-oriented efficiency scores (life expectancy vs HALE): 0.88.
  - Correlation of output-oriented scores (life expectancy vs composite health index): 0.92.
  - Correlation of input-oriented scores (life expectancy vs composite health index): 0.87.

### Cross-sectional patterns and magnitudes
- Sample: 173 countries total: 34 AEs, 84 EMs, and 55 LIDCs.
- Group averages and dispersion (2017):
  - AEs are on average more efficient than EMs, which are more output-efficient than LIDCs.
  - LIDCs are modestly more input-efficient than EMs.
  - Considerable within-group variation, especially among EMs and LIDCs; less variation among AEs.
- Years of life lost due to inefficiency:
  - LIDCs: 8.2 life-years lost on average.
  - Sub-Saharan Africa: 10.5 life-years lost on average.
- Resources wasted due to inefficiency (input-oriented, percent of GDP):
  - EMs: 1.4 percent of GDP could be saved on average.
  - AEs: 1.2 percent of GDP could be saved on average.
  - LIDCs: 1.0 percent of GDP lost due to inefficiency.
  - Regional results:
    - Emerging and Developing Asia: 0.8 percent of GDP.
    - Caucasus and Central Asia: 1.9 percent of GDP.
    - Emerging and Developing Europe: 1.6 percent of GDP.
  - When using HALE as output, AEs’ resources lost are 1.8 percent of GDP (larger than EMs).

### Major drivers examined
- Policy and institutional variables included:
  - Universal Health Coverage (UHC) index (WHO): geometric mean of 14 tracer indicators, normalized 0–100.
  - Gini coefficient: income inequality (0 to 1).
  - Control of Corruption index: from Worldwide Governance Indicators (ranges -2.5 to 2.5).
  - Controls: total health spending per capita (HXPC), ln(GDP per capita) averaged 2013–17, ln(primary school enrollment rate), ln(share of public health expenditure in total health spending).
- Estimation approach:
  - Cross-sectional two-stage-least-squares (right-hand-side variables instrumented with their five-year lags); control-of-corruption averaged over 2013–17.

### Key regression results (selected coefficients; robust standard errors in parentheses)
- Output-oriented efficiency (All countries)
  - Universal Health Care (UHC): 0.005** (0.002)
  - ln Gini (income): -0.097*** (0.031)
  - Control of Corruption: 0.014* (0.008)
  - ln (Health Expenditure per capita PPP): -0.053** (0.025)
  - ln (GDP per capita): 0.013 (0.022)
- Output-oriented efficiency (EMDEs)
  - Universal Health Care (UHC): 0.004** (0.002)
  - ln Gini (income): -0.122*** (0.038)
  - Control of Corruption: 0.028** (0.013)
  - ln (Health Expenditure per capita PPP): -0.047* (0.025)
- Output-oriented efficiency (AEs)
  - Universal Health Care (UHC): -0.001 (0.006)
  - ln Gini (income): -0.006 (0.058)
  - Control of Corruption: -0.003 (0.012)
- Input-oriented efficiency (All countries)
  - Universal Health Care (UHC): 0.009* (0.005)
  - ln Gini (income): -0.213*** (0.072)
  - Control of Corruption: 0.067*** (0.020)
  - ln (Health Expenditure per capita PPP): -0.088 (0.058)
- Input-oriented efficiency (EMDEs)
  - Universal Health Care (UHC): 0.010** (0.004)
  - ln Gini (income): -0.184** (0.083)
  - Control of Corruption: 0.077** (0.034)
  - ln (Health Expenditure per capita PPP): -0.112* (0.059)
- Input-oriented efficiency (AEs)
  - Coefficients statistically insignificant for AEs (small sample).

Note: Significance levels reported in source: *** p<0.01, ** p<0.05, * p<0.1.

### Interpretation of empirical findings
- UHC allocation:
  - Higher UHC index is associated with greater input- and output-efficiency for the full sample and for EMDEs.
  - For AEs, the UHC–efficiency relationship is statistically insignificant.
  - Coverage expansion has fiscal implications; improving UHC above certain thresholds in EMDEs is feasible only with adequate total health spending.
- Income inequality:
  - Higher Gini is associated with lower efficiency and worse health outcomes; negative relationship stronger for EMDEs and for input-oriented efficiency.
- Corruption / governance:
  - Worse control of corruption is associated with more inefficient health spending, especially for EMDEs; link not statistically significant for AEs in these regressions.
- Other variables:
  - Negative coefficients on ln(Health Expenditure per capita PPP) reflect nonlinearity/concavity in the health production function.

### Drivers of inefficiency and quantified policy scenarios
- Reallocating health funds toward UHC of essential services substantially raises overall efficiency, particularly in EMs and LIDCs.
  - Elevating countries’ UHC to their respective income group’s 75th percentile would:
    - Increase LIDCs’ average life expectancy by 3.4 years.
    - Increase EMs’ average life expectancy by 2.2 years.
    - Enable LIDCs to save 0.37 percent of GDP in health spending while maintaining life expectancy.
    - Enable EMs to save 0.39 percent of GDP in health spending while maintaining life expectancy.
- Addressing broader economic and institutional factors:
  - Improving LIDCs’ Gini coefficient to the 75th percentile would increase life expectancy by 2.1 years.
  - Increasing EMs’ control-of-corruption index to the group’s 75th percentile would enable EMs to save 0.27 percent of GDP in health spending while maintaining the same health outcomes.
  - Gains from reducing inequality to the income-group 75th percentile:
    - Life expectancy increases: EMs: 1.7 years; LIDCs: 2.1 years.
    - Savings in health spending: EMs: 0.17 percent of GDP; LIDCs: 0.12 percent of GDP.
  - Improving control of corruption to the income-group 75th percentile:
    - EMs: additional 1.6 years of life expectancy; avoid waste of 0.27 percent of GDP.
    - LIDCs: smaller but positive gains.

### Robustness and alternative outputs
- Using HALE as the output yields highly consistent results, except for a weaker relationship of corruption to efficiency.
- A composite health-output index (maternal mortality rate, neonatal mortality rate, under-5 mortality rate, incidence of tuberculosis, HIV, cardiovascular/cancer disease, adolescent fertility rate, and HALE) produces results that do not alter the study’s main findings.

### Policy implications and conclusions
- Policymakers can improve health spending efficiency through:
  - Increasing allocation toward essential health coverage (UHC) while recognizing fiscal implications and need for adequate total health spending.
  - Reducing income inequality to improve the conversion of health spending into life expectancy, with particularly large potential gains in EMDEs.
  - Fighting corruption to reduce waste and improve efficiency, especially in EMDEs.
- Limits to efficiency gains:
  - Efficiency improvements have limits under constrained budgets; adequacy of total health spending is crucial to achieve higher coverage and outcomes.
- Suggested next steps:
  - Complement macro-efficiency analysis with micro-efficiency studies focusing on hospitals, pharmaceutical procurement, public financial management, medical human resources, and other system-level details.

### Data sources and measurement specifics (selected)
- Output: life expectancy at birth, measured in years, for the latest year available, 2017.
- Input: total (public plus private) health expenditures per capita, measured in purchasing-power parity (PPP) dollars, averaged over 2013–17.
- Sample: 173 countries: 34 AEs, 84 EMs, 55 LIDCs.
- Key variable medians:
  - UHC medians: LIDCs: 44.5; EMs: 69.8; AEs: 81.5.
  - Gini medians: AEs: 0.30; EMs and LIDCs: 0.39.
  - Control of Corruption medians: LIDCs: -0.81; EMs: 0.30; AEs: 1.55.
- Appendix III output index composition (averages over indicated year ranges):
  - (i) maternal mortality rate (2014–17)
  - (ii) neonatal mortality rate (2015–18)
  - (iii) under-5 mortality rate (2015–18)
  - (iv) incidence of tuberculosis (2015–18)
  - (v) HIV incidence (2015–18)
  - (vi) cardiovascular/cancer incidence (2010–16)
  - (vii) adolescent fertility rate (2014–17)
  - (viii) HALE (2016)

*Source: IMF staff calculations and analysis from the IMF Working Paper "Patterns and Drivers of Health Spending Efficiency" (excerpts from the PDF content provided).*

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

### wpiea2022048-print-pdf - References .............................................................................................................

### Key findings
- During the COVID-19 response, selected advanced economies (AEs) increased health expenditures by an average of 1.5 percent of GDP in 2020, while the increase averaged 0.9 percent of GDP in emerging markets (EMs) and in low-income developing countries (LIDCs).
- Previous IMF work estimates that LIDCs would need to increase, on average, their annual health spending by 4.6 percent of GDP by 2030 to meet the Sustainable Development Goal on health (SDG3).
- AEs have higher life expectancy and a higher composite index of health outcomes than EMs and LIDCs.
- Using bias-corrected Data Envelopment Analysis (DEA), the study finds significant cross-country variation in health spending efficiency:
  - LIDCs lose on average 8.2 life-years due to inefficient use of health resources; Sub-Saharan Africa loses 10.5 life-years on average.
  - EMs waste the greatest share of health resources by income group at 1.4 percent of GDP; the Caucasus and Central Asia region wastes 1.9 percent of GDP.
- Over 2007–17, the vast majority of countries’ health systems became more efficient; LIDCs made notable gains in achieving improved health outcomes for their spending levels.
- Both efficiency and technological progress (a favorable shift of the production frontier) rose over 2007–17 as reflected in gains in the Malmquist productivity index.

### Methodology
- Efficiency scores are derived using bias-corrected Data Envelopment Analysis (DEA) with bootstrap correction to address measurement bias (Simar and Wilson, 2007).
- Outputs used include life expectancy, health-adjusted life expectancy (HALE), and a composite health index constructed from eight health outcome indicators (maternal mortality rate, neonatal mortality rate, under-5 mortality rate, incidence of tuberculosis, HIV incidence, cardiovascular/cancer incidence, adolescent fertility rate, and HALE).
- Both input- and output-oriented efficiency scores are estimated; results are robust to alternative output variables.

### Trends and quantitative patterns
- The relationship between health spending and life expectancy exhibits concavity: additional spending yields smaller gains in life-years in AEs versus regions like Sub-Saharan Africa.
- From 2007–17:
  - Most countries improved spending efficiency.
  - Technological progress contributed to overall productivity gains in health.
- Regional and income-group disparities persist: AEs are the most efficient group on average; LIDCs and Sub-Saharan Africa experience the largest losses in life expectancy from inefficiency.

### Drivers of inefficiency and quantified policy scenarios
- Reallocating health funds toward universal health coverage (UHC) of essential services substantially raises overall efficiency, particularly in EMs and LIDCs.
  - Elevating countries’ UHC to their respective income group’s 75th percentile would:
    - Increase LIDCs’ average life expectancy by 3.4 years, or
    - Enable LIDCs to save 0.37 percent of GDP in health spending without compromising life expectancy.
- Addressing broader economic and institutional factors also reduces inefficiency, though effects are generally more modest than UHC-focused reallocations:
  - Improving LIDCs’ Gini coefficient to the 75th percentile would increase life expectancy by 2.1 years (the 100th percentile is defined as the most equal country).
  - Increasing EMs’ control-of-corruption index to the group’s 75th percentile would enable EMs to save 0.27 percent of GDP in health spending while maintaining the same health outcomes.

### Literature context and robustness
- Prior global estimates report mean efficiency scores around 0.80 (Greene, 2004) and OECD averages near 0.83 (Afonso and St. Aubyn, 2005), indicating potential to increase outcomes by about 20 percent with current resources.
- Hospital-level and regional studies corroborate substantial inefficiency with possible savings in staff inputs and expanded services if inefficiencies are eliminated.
- Results in this paper are consistent with previous empirical research and robust to alternative output measures, including HALE and the composite health index.

*Source: IMF Working Papers — Patterns and Drivers of Health Spending Efficiency (excerpts from the PDF content provided).*

### 2.3 percent, or 1.8 years (1.2 years at age 65), when moving from current positions to the efficiency frontier.

### 2.3 percent, or 1.8 years (1.2 years at age 65), when moving from current positions to the efficiency frontier.

### Evolution and trends in health spending efficiency
- Relatively few studies assess evolution over time; findings cited:
  - Gupta and Verhoeven (2001): little change in EMDEs from second half of the 1980s to first half of the 1990s, with improvements in the Western Hemisphere region.
  - Samut and Cafri (2016): OECD countries—health care efficiency increased mildly up to mid-decade in the 2000s, then steadily fell, with a lower level by 2010 than at the start of the decade; decline accompanied by a regress in technology and an overall decline in the Malmquist Productivity Index.
- Between 2007 and 2017:
  - 146 (136) countries became more output (input) efficient over time.
  - 27 (37) countries saw a deterioration in output (input) efficiency.

### Factors associated with efficiency (sector-specific and cross-sectoral)
- Health service provision and system design:
  - Immunization of children enhances health outcomes for given resources (Pérez-Cárceles et al., 2018; Alexander et al. 2003).
  - Integrated care (multiple conditions addressed simultaneously) can improve allocative efficiency (Hou et al. 2021 — Zimbabwe study).
  - Market concentration in hospitals (e.g., de facto monopolies) can erode efficiency (Puig-Junoy, 2000).
  - Balance of public vs private provision: Samut and Cafri (2016) find higher prevalence of private hospital care associated with greater efficiency, while public care is associated with lower efficiency.
  - Hospital-centric systems can be inefficient relative to primary health care provision (Vietnam qualitative World Bank analysis — Teo and Huong, 2020).
- Broader institutional and socioeconomic factors:
  - Institutional quality and corruption: worse values of a synthetic measure of corruption in public contracts associated with poorer execution efficiency of health infrastructure projects (Cavalieri et al., 2017 — Italy).
  - Income and inequality:
    - Higher overall income leads to a more productive process of transforming health resources into improved outcomes (Greene, 2004).
    - Inequality measured by the Gini coefficient is detrimental to efficient health provision (Greene, 2004).
- Complementary sector services:
  - Higher levels of education associated with better health outcomes given existing health spending (Samut and Cafri, 2016; Alexander et al., 2003).
  - Sanitation and nutrition services reduce disease and undernourishment, enhancing effectiveness of health resources (Kathuria and Sankar, 2005; Alexander et al. 2003).

### Data, methodology, and measurement specifics
- Output and input definitions:
  - Output: life expectancy at birth, measured in years, for the latest year available, 2017.
  - Input: total (public plus private) health expenditures per capita, measured in purchasing-power parity (PPP) dollars, averaged over 2013–17.
- Sample:
  - 173 countries total: 34 AEs, 84 EMs, and 55 LIDCs.
- Efficiency measurement approach:
  - Nonparametric Data Envelopment Analysis (DEA) used to derive piecewise-linear frontier and input- and output-oriented efficiency scores.
  - Both input- and output-oriented scores derived; output-oriented uses maximization over δ with inverse 1/δ as technical output-efficiency score (ranges 0 to 1); input-oriented minimizes δ as input-efficiency score (ranges 0 to 1).
  - Bootstrapping applied (200 bootstrap samples) to correct upward bias of DEA and enable inferential methodology (Simar and Wilson, 2007; Bogetoft and Otto, 2011).
- Robustness checks:
  - Primary findings also presented using health-adjusted life expectancy (HALE) and a composite health index; correlations between life expectancy and alternative outputs:
    - Correlation of output-oriented efficiency scores (life expectancy vs HALE): 0.94.
    - Correlation of input-oriented efficiency scores (life expectancy vs HALE): 0.88.
    - Correlation of output-oriented scores (life expectancy vs composite health index): 0.92.
    - Correlation of input-oriented scores (life expectancy vs composite health index): 0.87.

### Cross-sectional patterns and magnitudes
- Group averages and dispersion (2017):
  - Advanced Economies (AEs) are on average more efficient than Emerging Markets (EMs), which are more output-efficient than Low-Income Developing Countries (LIDCs).
  - LIDCs are modestly more input-efficient than EMs.
  - Considerable within-group variation, especially among EMs and LIDCs; less variation among AEs.
- Years of life lost due to inefficiency:
  - Figures show distributions by income group and by region (visuals not reproduced here).
- Resources wasted due to inefficiency (input-oriented, percent of GDP):
  - EMs could save 1.4 percent of GDP on average by closing input-efficiency gaps.
  - AEs could prevent waste of 1.2 percent of GDP.
  - LIDCs’ resources lost due to inefficiency: 1.0 percent of GDP.
  - Regional results (percent of GDP):
    - Emerging and Developing Asia: 0.8 percent of GDP.
    - AEs: 1.2 percent of GDP.
    - Caucasus and Central Asia: 1.9 percent of GDP.
    - Emerging and Developing Europe: 1.6 percent of GDP.
  - When using HALE as output, results similar except AEs’ resources lost are more sizeable at 1.8 percent of GDP and larger than those of EMs (Appendix III, Table AIII.1).

### Evolution over time and decomposition (2007–17)
- Share of countries improving efficiency:
  - More countries showed improvements than declines over the decade (details by income group in figures).
  - Output-efficiency improvements: higher share for EMs than AEs, and for LIDCs than EMs.
  - Input-efficiency improvements: lower the income status, the lower the share of countries showing improvement.
- Decomposition using Malmquist Productivity Index:
  - Efficiency Change (EC) captures how much closer to the frontier a unit gets between two periods.
  - Technical Change (TC) captures frontier shifts (technological progress or regress) between periods at each country's input mix.
  - Malmquist index M_h = EC_h * TC_h.
  - Between 2007 and 2017 many countries show both efficiency improvements and technological progress (upper-right quadrant in Figure 12).
  - Several LIDCs and EMs experienced technological regress while showing relative efficiency gains; implies frontier shifted proportionally less for these countries than for peers, limiting overall productivity gains.

*Source: IMF staff calculations, from IMF Working Paper "Patterns and Drivers of Health Spending Efficiency" (figures and tables referenced are part of the source chapter).*

### Section IV had revealed interesting patterns in efficiency across income groups, regions, and time, and the

### wpiea2022048-print-pdf - Section IV had revealed interesting patterns in efficiency across income groups, regions, and time, and the

### Major drivers examined
- Universal Health Coverage (UHC) index (WHO): based on 14 tracer indicators across four essential service areas (infectious diseases; reproductive, maternal, newborn and child health; non-communicable diseases; and service capacity and access).
- Income distribution: Gini coefficient included to capture the role of income inequality.
- Corruption / governance: Control of Corruption index from the Worldwide Governance Indicators (averaged over 2013–17).
- Additional controls: total health spending per capita (HXPC), ln(GDP per capita) averaged 2013–17, ln(primary school enrollment rate), and ln(share of public health expenditure in total health spending).
- Estimation approach: cross-sectional two-stage-least-squares (right-hand-side variables instrumented with their five-year lags); control-of-corruption averaged over 2013–17; where data missing for one or more years the average is taken over available years.

### Empirical model (specification)
- Outcome: Effi — life expectancy efficiency score in 2017 for country i.
- Main regressors: UHC, ln(Gini), Corr (control of corruption).
- Controls: ln(GDP per capita), ln(Health Expenditure per capita PPP), ln(Primary School Enrollment), ln(Share of Public Health Expenditure).
- Estimation: two-stage-least-squares to address simultaneity bias; Tobit not used due to fractional outcome nature.

### Sample distributions (medians and dispersion)
- UHC medians:
  - LIDCs: 44.5
  - EMs: 69.8
  - AEs: 81.5
- Gini medians:
  - AEs: 0.30
  - EMs and LIDCs: 0.39
- Control of Corruption medians:
  - LIDCs: -0.81
  - EMs: 0.30
  - AEs: 1.55

### Key regression results (Table 1: selected coefficients; robust standard errors in parentheses)
- Output-oriented efficiency (All countries)
  - Universal Health Care (UHC): 0.005** (0.002)
  - ln Gini (income): -0.097*** (0.031)
  - Control of Corruption: 0.014* (0.008)
  - ln (Health Expenditure per capita PPP): -0.053** (0.025)
  - ln (GDP per capita): 0.013 (0.022)
- Output-oriented efficiency (EMDEs)
  - Universal Health Care (UHC): 0.004** (0.002)
  - ln Gini (income): -0.122*** (0.038)
  - Control of Corruption: 0.028** (0.013)
  - ln (Health Expenditure per capita PPP): -0.047* (0.025)
- Output-oriented efficiency (AEs)
  - Universal Health Care (UHC): -0.001 (0.006)
  - ln Gini (income): -0.006 (0.058)
  - Control of Corruption: -0.003 (0.012)
- Input-oriented efficiency (All countries)
  - Universal Health Care (UHC): 0.009* (0.005)
  - ln Gini (income): -0.213*** (0.072)
  - Control of Corruption: 0.067*** (0.020)
  - ln (Health Expenditure per capita PPP): -0.088 (0.058)
- Input-oriented efficiency (EMDEs)
  - Universal Health Care (UHC): 0.010** (0.004)
  - ln Gini (income): -0.184** (0.083)
  - Control of Corruption: 0.077** (0.034)
  - ln (Health Expenditure per capita PPP): -0.112* (0.059)
- Input-oriented efficiency (AEs)
  - All above coefficients statistically insignificant for AEs (small sample).

Note: Significance levels reported as in source: *** p<0.01, ** p<0.05, * p<0.1.

### Interpretation of findings
- UHC allocation
  - Higher allocation toward essential health services (UHC index) is associated with greater input- and output-efficiency for the full sample and for EMDEs.
  - For AEs, the UHC–efficiency relationship is statistically insignificant, consistent with many AEs already having high UHC coverage.
  - Because coverage expansion has fiscal implications, improving UHC above certain thresholds in EMDEs is only feasible with greater total health spending adequacy.
- Income inequality
  - Greater income inequality (higher Gini) is associated with lower efficiency and worse overall health outcomes.
  - This negative relationship is more pronounced for EMDEs and particularly affects input-oriented efficiency (waste of health sector resources).
- Corruption / governance
  - Worse control of corruption (lower index value) is associated with more inefficient health spending, especially for EMDEs.
  - For AEs the corruption–efficiency link is not statistically significant in these regressions.
- Other variables
  - Higher ln(Health Expenditure per capita PPP) shows a negative coefficient in several specifications, reflecting nonlinearity/concavity in the health production function and the role of spending levels in the efficiency context.
  - ln(GDP per capita), ln(primary school enrollment), and ln(share of public health expenditure) are generally not robustly significant across all specifications.

### Quantitative relevance: gains from policy actions (moving bottom 75% to income-group 75th percentile)
- If bottom 75 percent of LIDCs reach LIDCs’ 75th percentile in UHC:
  - Average increase in life expectancy: 3.4 years (LIDCs).
- If bottom 75 percent of EMs reach EMs’ 75th percentile in UHC:
  - Average increase in life expectancy: 2.2 years (EMs).
- Savings (while maintaining life expectancy) from reaching UHC 75th percentile:
  - EMs: 0.39 percent of GDP
  - LIDCs: 0.37 percent of GDP
- Gains from reducing income inequality to the income-group 75th percentile:
  - Life expectancy increases:
    - EMs: 1.7 years
    - LIDCs: 2.1 years
  - Savings in health spending:
    - EMs: 0.17 percent of GDP
    - LIDCs: 0.12 percent of GDP
- Gains from improving control of corruption to the income-group 75th percentile:
  - EMs: additional 1.6 years of life expectancy; avoid waste of 0.27 percent of GDP in health spending.
  - LIDCs: smaller but still positive gains (figures shown visually in source).

### Robustness and alternative outputs
- Using HALE as the output yields highly consistent results, except for a weaker relationship of corruption to efficiency (see Appendix II, Table AII.7).
- A third alternative health-output index (maternal mortality rate, neonatal mortality rate, under-5 mortality rate, incidence of tuberculosis, HIV, cardiovascular or cancer disease, adolescent fertility rate, and health-adjusted life expectancy) produces results that do not alter the study’s main findings (Appendix III).

### Policy implications and conclusions
- Policymakers can improve health spending efficiency through:
  - Increasing allocation toward essential health coverage (UHC) while recognizing fiscal implications and the need for adequate total health spending.
  - Reducing income inequality to improve the conversion of health spending into life expectancy, with particularly large potential gains in EMDEs.
  - Fighting corruption to reduce waste and improve efficiency, especially in EMDEs.
- Limits to efficiency gains:
  - Efficiency improvements have limits under constrained budgets; adequacy of total health spending is crucial to achieve higher coverage and outcomes.
- Suggested next steps:
  - Complement macro-efficiency analysis with micro-efficiency studies focusing on hospitals, pharmaceutical procurement, public financial management, medical human resources, and other system-level details to derive more actionable interventions.

*Italic: Source — IMF staff calculations and analysis from the IMF Working Paper (Patterns and Drivers of Health Spending Efficiency) excerpt.*

### Appendix Table AI.1 Sources for and Descriptions of Data and Variables

### Appendix Table AI.1 Sources for and Descriptions of Data and Variables

### Variables, Uses, Sources, and Descriptions
- Life expectancy  
  - Use: Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators, at: https://data.worldbank. Org/indicator/ SP.DYN.LE00.IN (accessed August 2020)  
  - Description: Life expectancy at birth, in years. Indicates the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life.

- Per capita health spending  
  - Use: Input in efficiency analysis  
  - Source: WHO Global Health Expenditure database, at: https://apps.who.int/nha/database/Select/Indicators/en  
  - Description: Total (public and private) per capita health expenditure, in purchasing-power parity dollars, calculated as domestic general government health expenditure per capita, PPP (SH.XPD.GHED.PP.CD), plus domestic private health expenditure per capita, PPP (SH.XPD.PVTD.PP.CD).

- Universal Health Coverage (UHC) index  
  - Use: Explanatory variable in second-stage regression  
  - Source: WHO, “UHC Service Coverage Index”, at: https://www.who.int/ data/gho/indicator- metadata- registry/imr-details/4834 (accessed August 2020)  
  - Description: Coverage of the population with essential health services, computed as the geometric mean of 14 tracer indicators and normalized to a 0–100 scale. The 14 indicators relate to family planning, pregnancy and delivery, child immunization, child treatment, tuberculosis, HIV/AIDS, malaria, water & sanitation, hypertension, diabetes, tobacco, hospital access, health workforce, and health security. Metadata can be found at https://www.who.int/healthinfo/universal_health_ coverage/UHC_ Tracer_I ndicators_Metadata.pdf.

- Share of public in total health spending  
  - Use: Explanatory variable in second-stage regression  
  - Source: WHO Global Health Expenditure database  
  - Description: Share of public health expenditure in total health expenditures (public / (public + private)).

- School enrollment, primary (% gross)  
  - Use: Explanatory variable in second-stage regression  
  - Source: World Bank, “World Development Indicators, at: https://data.worldbank.org/indicator/SE.P RM.ENRR (accessed August 2020)  
  - Description: Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown. Primary education provides children with basic reading, writing, and mathematics skills along with an elementary understanding of such subjects as history, geography, natural science, social science, art, and music (SE.PRM.ENRR).

- GDP per capita  
  - Use: Control variable in second-stage regression  
  - Source: World Bank, “World Development Indicators, at: https://data.worldbank.org/indicator/NY.G DP.PCAP.PP.CD  
  - Description: Gross domestic product in PPP dollars, divided by the population size (NY.GDP.PCAP.PP.CD).

- Gini coefficient  
  - Use: Explanatory variable in second-stage regression  
  - Source: IMF staff calculations, using data from the Luxembourg Income Study; the Organisation for Economic Co-operation and Development’s Income Distribution Database; Eurostat Income Inequality Statistics; the Socio-Economic Database for Latin America and the Caribbean; and PovcalNet.  
  - Description: Measure of the degree of income inequality in a country, ranging from 0 (all individuals have the same income) to 1 (one individual has all income).

- Control of Corruption  
  - Use: Explanatory variable in second-stage regression  
  - Source: World Bank, “Worldwide Governance Indicators” (Kauffman et al., 2010), at: https://info. Worldbank.org/ governance/wgi/  
  - Description: Perception of extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests. The index ranges from -2.5 to 2.5 (worst and best possible outcome, respectively).

- Maternal mortality rate (per 100,000 live births)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators”, at: http://data.worldbank.org/indicator/SH.ST A.MMRT  
  - Description: The estimated number of women, between the age of 15–49, who die from pregnancy-related causes while pregnant or within 42 days of termination of pregnancy, per 100,000 live births.

- Neonatal mortality rate (per 1,000 live births)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators”, at: http://data.worldbank.org/indicator/SH.DY N.NMRT  
  - Description: The number of newborn infants (neonates) who die before reaching 28 days of age, per 1,000 live births.

- Under-5 mortality rate (per 1,000 live births)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators”, at: http://data.worldbank.org/indicator/SH.DY N.MORT  
  - Description: The probability that a newborn baby will die before reaching age five, if subject to age-specific mortality rates of the specified year, per 1,000 live births.

- Tuberculosis incidence (per 100,000 population)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators”, at: http://data.worldbank.org/indicator/SH.TB S.INCD  
  - Description: The estimated rate of new and relapse cases of tuberculosis in a given year, expressed per 100,000 people. All forms of tuberculosis are included, including cases of people living with HIV.

- HIV incidence (per 1,000 uninfected population)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: UNAIDS (2020), at: http://aidsinfo.unaids.org?did=55da49cd6 4e925b94e70b0ce&r=world&t=2018&tb=d&bt=dnli&ts=0,0&tr=world&aid=5970eccef7341ed11f26de5d&sav=Population:%20All%20ages&tl=2  
  - Description: Number of people newly infected with HIV per 1,000 uninfected population.

- Cardiovascular or cancer incidence  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: WHO (2018), at: https://unstats.un.org/sdgs/indicators/data base/?indicator=3.4.1  
  - Description: The probability of dying between the ages of 30 and 70 years from cardiovascular diseases, cancer, diabetes or chronic respiratory diseases, defined as the percent of 30-year-old-people who would die before their 70th birthday from these diseases, assuming current mortality rates at every age and that individuals would not die from any other cause of death (e.g. injuries or HIV/AIDS).

- Adolescent fertility rate (births per 1,000 adolescent females aged 15 to 19)  
  - Use: Component of Own Index; Output in efficiency analysis  
  - Source: World Bank, “World Development Indicators”, at: http://data.worldbank.org/indicator/SP.AD O.TFRT  
  - Description: The number of births per 1,000 women between the age of 15 to 19.

- Healthy-adjusted life expectancy (HALE)  
  - Use: Output in efficiency analysis  
  - Source: World Health Organization, “Global Health Observatory”, at: https://www.who.int/data/gho/indicator- metadata-registry/imr-details/66  
  - Description: Average number of years that a person can expect to live in "full health" by taking into account years lived in less than full health due to disease and/or injury.

### Appendix III: Output Index Composition (for robustness check)
- The output index uses the following variables, averaged over the indicated year ranges:  
  - (i) maternal mortality rate (2014–17)  
  - (ii) neonatal mortality rate (2015–18)  
  - (iii) under-5 mortality rate (2015–18)  
  - (iv) incidence of tuberculosis (2015–18)  
  - (v) HIV incidence (2015–18)  
  - (vi) cardiovascular/cancer incidence (2010–16)  
  - (vii) adolescent fertility rate (2014–17)  
  - (viii) HALE (2016)

*Source: Appendix Table AI.1, wpiea2022048-print-pdf — Appendix Table AI.1 Sources for and Descriptions of Data and Variables*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022048-print-pdf.pdf_