## sstssaea

## Source details

**Canonical URL:** [sstssaea](https://www.imf.org/-/media/files/publications/dp/2025/english/sstssaea.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/dp/2025/english/sstssaea.pdf.md)
- [Structured JSON version](/-/media/files/publications/dp/2025/english/sstssaea.pdf.json)

---

### Executive Summary — Key Findings and Quantities
- Scope: Analysis of social spending targets in IMF-supported programs across sub-Saharan Africa from 2002 to 2024.
- Country coverage: Since 2002, 37 out of 45 sub-Saharan African countries have engaged in IMF-supported programs.
- Program sample: Database covers 152 programs approved between January 2002 and September 2024, with 98 including social spending targets.
- PRGT 2009 requirement: Introduction of a requirement in 2009 that PRGT-supported programs include social spending targets led to increased use of social spending floors; since 2010, 89 percent of programs included such targets.
- Definition and measurement: Social spending defined as on-budget government spending on health and education, and social protection (social insurance and social assistance) (IMF 2019a, 2024).
- Financing gap: Closing health and education gaps to achieve Sustainable Development Goals would require additional spending of almost 8 percentage points of GDP on average by 2030.
- Evolution and form of targets:
  - Initially focused on education and health; expanded to include social protection.
  - Trend toward more specific minimum thresholds for parts of social spending.
  - Predominantly formulated as indicative targets rather than quantitative performance criteria.
- Ambition and completion:
  - Median social spending target in sub-Saharan Africa = 3.7 percent of GDP.
  - Median social spending target as a share of tax revenue (medians, 2002–24): Sub-Saharan Africa = 35 percent; Europe = 50 percent; Asia and Pacific = 32 percent; Middle East and Central Asia = 10 percent; Western Hemisphere = 9 percent.
  - Average completion rate for social spending targets: Sub-Saharan Africa = 69 percent; Europe = 62 percent; Middle East and Central Asia = 62 percent; Asia and Pacific = 66 percent; Western Hemisphere = 74 percent.
  - Completion rate by target specificity: Nonspecific targets = 69 percent; Specific targets = 67 percent.
  - In sub-Saharan Africa, 11 percent of social spending targets were subsequently lowered; where targets were missed, median shortfall ≈ 15 percent of the original target.
- Effectiveness and outcomes:
  - Health and education spending tend to rise during programs (excluding pandemic years 2020–21).
  - Increases are larger, on average, during programs that include a social spending target versus programs without such a target.
  - Tentative evidence suggests school enrollment rates increase and infant mortality decreases in the two years after social spending targets are implemented; evidence limited by data frequency and coverage.

### Definitions, Context, and Policy Role
- Definition used in paper: Social spending = on-budget government spending on health and education, plus social protection (social insurance and social assistance) (IMF 2019a, 2024).
- Policy role: Social spending promotes human capital accumulation, inclusive growth, reduced inequality, protection during structural changes, and consumption stabilization during shocks.
- Observed shortfall: Social spending generally below desirable levels in sub-Saharan Africa; substantial financing constraints and reliance on limited external funding.
- Program conditionality intent: “Target” refers to program conditionality designed to prevent decline in social spending under adverse macroeconomic conditions or help improve adequacy; not intended as an SDG-level benchmark.
- Literature: Mixed evidence on program effects; effectiveness depends on design, implementation, institutional capacity, and country commitment.

### Coverage, Design, and Specificity of Social Spending Targets
- Coverage and trends:
  - Since 2002, of 45 sub-Saharan African countries, 37 engaged in IMF-supported programs; database includes 152 programs (Jan 2002–Sept 2024), 98 with social spending targets.
  - Since 2010 average share of sub-Saharan African programs with social targets: 89 percent, compared with other regions: Middle East and Central Asia = 67 percent; Western Hemisphere = 63 percent; Asia and Pacific = 40 percent; Europe = 13 percent.
  - Most targets cover education and health; almost 90 percent of sub-Saharan African social spending targets reference education or health.
  - About half of sub-Saharan African targets reference social protection spending; share covering social protection has increased over time and in the last five years most targets covered education, health, and social protection.
- Form of conditionality:
  - Out of 152 new programs (2002–2024): 93 programs included social spending ITs; 5 programs included social spending QPCs (Rwanda 2002 and 2006; Chad 2014; Senegal 2020; Guinea-Bissau 2023).
  - QPCs on social spending are rare in other regions: 4 QPCs in 59 programs with quantitative social spending conditionality outside sub-Saharan Africa.
  - ITs are used to promote adaptability, country ownership, and accommodate limited capacity/data constraints.
- Specificity:
  - Social spending targets in sub-Saharan Africa have tended to be nonspecific over 2002–24, compared with most other regions.
  - For sub-Saharan African programs: Targets linked to both education and health spending are not prescriptive about the type of spending or project in 73 percent of cases; for social protection spending, the nonspecific share ≈ 60 percent.
  - Over 2020–24, half of the social spending targets in sub-Saharan Africa set goals in terms of a specific type of spending or project.
  - Exclusions: Wages explicitly excluded in 54 percent of specific target definitions versus 19 percent of nonspecific targets; specific targets explicitly exclude externally financed spending in 29 percent of cases versus 26 percent for nonspecific targets.
  - The share of targets explicitly excluding wages rose to 37 percent over 2020–24 compared with 14 percent over 2006–10.

### Ambition, Magnitudes, and Regional Comparisons
- Median social spending target in sub-Saharan African programs = 3.7 percent of GDP.
- Examples of range:
  - Targets < 1 percent of GDP: Mozambique in 2022; Democratic Republic of the Congo in 2021; Guinea in 2017.
  - Target = 9 percent of GDP: Senegal in 2021.
- Specific vs nonspecific magnitudes (averages):
  - Specific targets: 22 percent of tax revenues on average; 14 percent of government expenditures.
  - Nonspecific targets: 42 percent of tax revenues; 25 percent of government expenditures.
- Trend: Overall, social spending targets in sub-Saharan African programs have declined as a share of both tax revenue and government expenditures over time; decline amplified by growing share of specific targets and increased exclusion of wages.

### Target Completion, Revisions, and Shortfalls
- Completion rates:
  - Average completion rate for social spending targets: Sub-Saharan Africa = 69 percent.
  - Completion rates for indicative targets (ITs) in sub-Saharan African programs: Social spending ITs = 68 percent; ITs linked to fiscal revenues = 59 percent.
  - Completion by specificity: Nonspecific targets = 69 percent; Specific targets = 67 percent.
- Revisions and shortfalls:
  - In sub-Saharan Africa, 11 percent of social spending targets were subsequently lowered.
  - Where targets were missed in sub-Saharan Africa, the median shortfall was about 15 percent of the original target.
  - Share of completed targets completed only because of downward revision: Social spending targets = 5 percent; Revenue targets = 12 percent.

### Spending Behavior During IMF-Supported Programs and Fiscal Consolidation
- Descriptive and regression evidence:
  - Before-and-after comparisons indicate health and education spending tend to rise during IMF-supported programs (analysis excluding pandemic years 2020–21 where noted).
  - Median increases in both real health and real education spending over the course of IMF-supported programs are higher in sub-Saharan Africa than in most other regions; interquartile range for education spending changes in sub-Saharan Africa lies above zero.
  - Mean and median percentage changes in real spending are higher in programs with social spending targets than in programs without such targets.
  - Panel fixed effects regression shows both real health and education spending increase on average during program years; results are not significant when looking at health or education spending as a share of government expenditure or GDP.
- Interaction with fiscal consolidation:
  - Programs with social spending ITs: Fiscal deficits were reduced by 0.6 percentage points of GDP on average during IMF-supported programs with social spending ITs.
  - Education spending as a share of total government spending rose by 0.31 percentage points on average.
  - Health spending as a share of total government spending rose by 0.75 percentage points on average.
  - For larger fiscal consolidations, the increase in the median share of social spending is mostly because of a decline in total spending, while social spending was less affected.
  - Regression analysis of programs with social spending targets showed a nonsignificant small negative effect of fiscal balance improvements on education and health spending as a share of GDP, and a positive but not statistically significant effect on education and health spending as a share of total government spending.
  - Interpretation: Social spending targets appear to have limited the effect of spending cuts associated with fiscal consolidation on education and health spending.

### Evidence on Outcomes — Education and Health Indicators
- Education outcomes:
  - Across 36 program cases in sub-Saharan Africa with available data: Median share of young people not in education falls from >23 percent in the year before the social spending target was implemented to 21 percent two years after the start of the program.
  - Event study patterns: Share of young people not in education typically above average in year before targets implemented and falls in the two years after implementation.
  - Literacy rates: Median literacy rates are found to stay constant during IMF-supported programs with education spending targets; event study results for literacy are not significant.
  - Caveat: Structural variables like literacy are slow moving and data are updated infrequently.
- Health outcomes:
  - According to 39 program cases: Share of births attended by skilled health staff increased from 62.5 percent to over 67 percent over the first two years of IMF-supported programs with health spending targets.
  - Median infant mortality rate falls from 55.6 per 1,000 live births to 49.3 during the first two years of the programs.
  - Event study: Share of births attended by skilled health workers increases somewhat during programs with health spending targets, though differences are not statistically significant; some decline in infant mortality after implementation, but rates remain well above average.
  - Caveat: Health outcome data updated infrequently; results should be treated with caution.

### Policy Considerations and Recommendations
- Design and targeting:
  - Social spending targets should be tailored to the country context, including the adequacy and efficiency of existing social safety nets.
  - Design should consider distributional effects of macroeconomic adjustment (IMF 2019b).
  - Decisions on inclusion or exclusion of public teachers’ and health workers’ wages from social spending targets have important implications and require country-specific judgment.
  - Recent trend toward excluding wages suggests emphasis on protecting nonwage social spending.
- Efficiency, complementarity, and governance:
  - Opportunities may exist to improve the efficiency of social spending in some cases.
  - Social spending targets should avoid crowding out private provision of health, education, or other social services.
  - Development partners can help better define social spending targets and aid effective implementation when domestic capacity is limited.
  - Weak governance, corruption, and lack of implementation capacity may explain poor social outcomes despite high target completion rates.
  - Good public financial management practices, transparent public procurement, strong anti-corruption frameworks, and robust oversight are key to ensuring social spending floors yield durable development outcomes.

### Data, Measures, and Econometric Specification
- Data sources:
  - Program data sourced from MONA (Monitoring of Fund Arrangements) database; covers all IMF Executive Board–approved programs with ex ante or ex post conditionality, including precautionary facilities and nonfinancing instruments like PCI; excludes SMPs and urgent financing via RCF/RFI.
  - Education expenditure data from UNESCO Institute for Statistics (aligned with ISCED 2011).
  - Public health spending data from WHO’s Global Health Expenditure Database (GHED), focusing on current health spending per SHA 2011; GHED includes external transfers routed through government.
  - Total government spending, GDP, inflation, and revenue data from IMF World Economic Outlook database.
  - No comprehensive database for social protection spending; many governments do not use functional budget classifications and social protection spending can be done by several ministries.
- Four measures of social spending used in econometric analysis:
  - Social spending as a share of GDP.
  - Real social spending.
  - Real per capita social spending.
  - Social spending in percent of total government spending.
- Econometric specification (fixed effects regression):
  - Let Yit denote social spending in country i at time t, with Yit0 the social spending variable in the year prior to the program approval.
  - Specification: ∆Yit,t0 = α + ∑5s=2 βs Dit,s + γi + εit, where ∆Yit,t0 = Yit − Yit0; Dit,s is a dummy for each year s of the program; γi country fixed effects; εit error term.
- Limitations:
  - Data frequency and coverage constrain causal inference; particularly limited data on social outcomes and the generalization of social spending targets in PRGT-supported programs after 2010.

*Source: sstssaea - references. (sstssaea - references.pdf) — International Monetary Fund*

### references.

### sstssaea - references.

### Identifiers and Classification
- ISBN: 9798229014595 (paper)
- ISBN: 9798229015998 (ePub)
- ISBN: 9798229015943 (WebPDF)
- Classification: HJ923.L3 2025
- Subjects (LCSH): Government spending policy—Sub-Saharan Africa. | International Monetary Fund—Economic assistance—Sub-Saharan Africa. | Conditionality (Internation relations)—Sub-Saharan Africa.

### Acknowledgments and Publication Notes
- Authors acknowledge comments and suggestions from: Ali Abbas; Isabell Adenauer; Omer Akbal; Matthieu Bellon; Nicolo Bird; Fernanda Brollo; Felipe Camêlo; Wenjie Chen; Costas Christou; Mariana Colacelli; Pavis Devahasadin; Luc Eyraud; Claire Gicquel; Matt Grieger; Iacovos Ioannou; Pablo Lopez Murphy; Montfort Mlachila; Catherine Pattillo; Jesmin Rahman; Nagwa Riad; Jennifer Ribarsky; Lisbeth Rivas; Cemile Sancak; Axel Schimmelpfennig; Martin Schindler; Abebe Selassie; Andrew Tiffin; Petia Topalova; Olaf Unteroberdoerster; Delia Velculescu; Mercedes Vera Martin; Genevieve Verdier; Jaroslaw Wieczorek; Karim Youssef.
- Useful discussions with Emine Hanedar, Marialuz Moreno Badia, and Baoping Shang.
- Danielle Bieleu thanked for formatting assistance.
- Departmental Paper Series: views expressed are those of the author(s) and do not necessarily represent IMF, its Executive Board, or IMF management.
- Publication orders: International Monetary Fund, Publication Services, P.O. Box 92780, Washington, DC 20090, USA; T. +(1) 202.623.7430; publications@IMF.org; IMFbookstore.org; elibrary.IMF.org.

### Executive Summary — Key Findings and Quantities
- Scope: Analysis of social spending targets in IMF-supported programs across sub-Saharan Africa from 2002 to 2024.
- Country coverage: Since 2002, 37 out of 45 sub-Saharan African countries have engaged in IMF-supported programs.
- Program sample: Database covers 152 programs approved between January 2002 and September 2024, with 98 including social spending targets.
- PRGT 2009 requirement: Introduction of a requirement in 2009 that PRGT-supported programs include social spending targets led to increased use of social spending floors; since 2010, 89 percent of programs included such targets.
- Definition and measurement: No uniform definition of social spending across countries; the paper defines social spending as on-budget government spending on health and education, and social protection (social insurance and social assistance) per IMF 2019a, 2024.
- Financing gap: Closing health and education gaps to achieve Sustainable Development Goals would require additional spending of almost 8 percentage points of GDP on average by 2030.
- Evolution of targets:
  - Initially focused on education and health; expanded to include social protection.
  - Trend toward more specific minimum thresholds for parts of social spending.
  - Predominantly formulated as indicative targets rather than quantitative performance criteria.
- Ambition and completion:
  - Median social spending target in sub-Saharan Africa is higher than in all other regions except Europe (as a share of tax revenues).
  - Completion rate for social spending targets is high and compares favorably with other regions and other fiscal conditionality types.
  - Downward adjustments to targets after initial program agreement occur in a small number of cases.
- Effectiveness during fiscal consolidation:
  - Many IMF-supported programs prescribed fiscal consolidation, but social spending has generally been protected from wider spending cuts.
  - Health and education spending tend to rise during programs (excluding pandemic years 2020–21).
  - The rise in health and education spending during IMF-supported programs is more pronounced in sub-Saharan Africa than in most other regions.
  - Increases are larger, on average, during programs that include a social spending target versus programs without such a target.
  - No evidence that social spending cuts resulted from fiscal consolidations under IMF programs when social spending targets were present.
- Outcomes:
  - Tentative evidence suggests school enrollment rates increase and infant mortality decreases in the two years after social spending targets are implemented.
  - Findings limited by lack of regularly updated data on education and health outcomes.

### Acronyms and Abbreviations (as listed)
- CI = Confidence interval
- GRA = General Resources Account
- IT = Indicative target
- MONA = Monitoring of Fund Arrangements database
- PCI = Policy Coordination Instrument
- PRGT = Poverty Reduction and Growth Trust
- QPC = Quantitative performance criterion
- RCF = Rapid Credit Facility
- RFI = Rapid Financing Instrument
- SMP = Staff-Monitored Program

### Section 1 — Introduction: Definitions and Context
- Definition used in paper: Social spending = on-budget government spending on health and education, plus social protection (social insurance and social assistance) (IMF 2019a, 2024).
- Policy role: Social spending promotes human capital accumulation, inclusive growth, reduced inequality, protection during structural changes, and consumption stabilization during shocks.
- Observed shortfall: Social spending generally below desirable levels in sub-Saharan Africa; substantial financing constraints and reliance on limited external funding.
- Program conditionality intent: “Target” refers to program conditionality designed to prevent decline in social spending under adverse macroeconomic conditions or help improve adequacy; does not imply an ideal level to reach SDGs.
- Literature: Mixed evidence on program effects—some studies find lasting increases in social spending (Clements, Gupta, and Nozaki 2013; IMF 2019b; Gupta, Schena, and Yousefi 2020), others find program participation lowered social spending (Stubbs and others 2020; Kentikelenis and Stubbs 2024). Effectiveness depends on design, implementation, institutional capacity, and country commitment.

### Section 2 — Social Spending Targets in Sub-Saharan Africa (Selected Findings)
- A. More Programs with Social Spending Targets
  - Since 2002, of 45 sub-Saharan African countries, 37 engaged in IMF-supported programs.
  - Database: 152 programs (Jan 2002–Sept 2024), 98 with social spending targets.
  - 2009 PRGT reform required safeguarding social and priority spending where possible, increasing use of social spending floors.
  - Since 2010 average share of sub-Saharan African programs with social targets: 89 percent, compared with other regions:
    - Middle East and Central Asia: 67 percent
    - Western Hemisphere: 63 percent
    - Asia and Pacific: 40 percent
    - Europe: 13 percent
  - Some programs introduced social spending targets during reviews after approval in specified countries (Benin 2010; Burundi 2008; Comoros 2009; Congo 2008; Djibouti 2008; Ghana 2003 and 2009; Liberia 2008; Mali 2008; Mongolia 2017; Niger 2008; Senegal 2007; Solomon Islands 2011; Tajikistan 2002; Tanzania 2010; Zambia 2008).
  - Instances of two social spending targets in a program: Liberia 2008; Uganda 2021; Côte d’Ivoire 2023.
- B. Mostly Indicative Targets
  - Out of 152 new programs in sub-Saharan Africa (2002–2024):
    - 93 programs included social spending ITs.
    - 5 programs included social spending QPCs (Rwanda 2002 and 2006; Chad 2014; Senegal 2020; Guinea-Bissau 2023).
  - QPCs on social spending are rare in non-sub-Saharan regions: 4 QPCs in 59 programs with quantitative social spending conditionality.
  - Two programs began with ITs and later transformed into QPCs: Tunisia 2016; Pakistan 2019.
  - ITs are used to promote adaptability, country ownership, and accommodate limited capacity/data constraints.
- C. Growing Coverage of Social Protection
  - Most targets cover education and health; fewer reference social protection.
  - Almost 90 percent of sub-Saharan African social spending targets reference education or health.
  - About half of sub-Saharan African targets reference social protection spending.
  - Regional contrasts: Asia and Pacific and Europe targets tend to cover social protection more than sub-Saharan Africa; Middle East and Central Asia and the Americas show fairly similar coverage across education, health, and social protection.
  - Share of targets covering social protection in new sub-Saharan African programs has increased over time; in the last five years most targets covered education, health, and social protection.

### Data and Method Notes Referenced
- Program data sourced from MONA (Monitoring of Fund Arrangements) database; covers all IMF Executive Board–approved programs with ex ante or ex post conditionality, including precautionary facilities and nonfinancing instruments like PCI; excludes SMPs and urgent financing via RCF/RFI.
- Coverage and specificity analysis relies on technical memoranda of understanding describing definitions and data requirements for IMF programs.

*Source: sstssaea - references. (sstssaea - references.pdf) — International Monetary Fund*

### 1. Share of Targets Including Education, Health,

### 1. Share of Targets Including Education, Health, and Social Protection between 2002 and 2024, by Region

### D. Increasingly Specific Target Definitions — key findings
- Social spending targets in sub-Saharan Africa have tended to be nonspecific over 2002–24, compared with most other global regions.
- Sub-Saharan African programs typically set objectives for total social spending by social ministries rather than tying targets to a functional budget classification or specific projects.
- For sub-Saharan African programs:
  - Targets linked to both education and health spending are not prescriptive about the type of spending or project in 73 percent of cases.
  - For social protection spending, the equivalent share of nonspecific definitions is around 60 percent.
- There has been evolution toward more prescriptive targets over time:
  - Over 2020–24, half of the social spending targets in sub-Saharan Africa set goals in terms of a specific type of spending or project.
- The shift toward more specific targets may reflect:
  - Progress in public financial management, often with support from IMF technical assistance.
  - Greater government capacity to monitor social spending.

### Box 1 — classification methodology (summary)
- Identification: Word search selected targets related to education, health, social, pro-poor, anti-poverty, cash transfer spending, and spending aimed at supporting vulnerable households.
- Nonspecific definition: Measured by current or capital spending of sectoral ministries (e.g., Ministry of Education, Ministry of Health, Ministry of Social Affairs, and sometimes Ministries of Women, Youth, Water, Development, Agriculture).
  - Risks: Administrative overhead absorption, limited effects on effective service provision, omission of social programs run by other entities.
- Specific definition: Spending related to specific programs expected to have positive effects on defined social outcomes (examples: vocational training, cash transfers, programs for vulnerable families).
  - Requirements: Technical capacity to identify programs and track program-specific spending; functional budget classification facilitates monitoring.
- Trade-offs: Specific coverage enhances cost-effectiveness and granularity but may limit flexibility and reach; nonspecific coverage is broader and flexible but may reduce efficiency and focus.

### E. Relatively “Ambitious” Targets — key statistics and patterns
- Median social spending target in sub-Saharan African programs:
  - Equivalent to 3.7 percent of GDP.
- Examples of range across countries:
  - Targets less than 1 percent of GDP: Mozambique in 2022; Democratic Republic of the Congo in 2021; Guinea in 2017.
  - Target of 9 percent of GDP: Senegal in 2021.
- As a share of tax revenue (medians, 2002–24):
  - Sub-Saharan Africa median = 35 percent.
  - Europe median = 50 percent.
  - Asia and Pacific median = 32 percent.
  - Middle East and Central Asia median = 10 percent.
  - Western Hemisphere median = 9 percent.
- As a share of government expenditures, sub-Saharan African targets are higher than in the Western Hemisphere and in the Middle East and Central Asia.
- Specific versus nonspecific target magnitudes:
  - Specific targets: 22 percent of tax revenues on average; 14 percent of government expenditures.
  - Nonspecific targets: 42 percent of tax revenues; 25 percent of government expenditures.
- Exclusions and trends:
  - Wages excluded explicitly in 54 percent of specific target definitions versus 19 percent of nonspecific targets.
  - Specific targets explicitly exclude externally financed spending in 29 percent of cases versus 26 percent for nonspecific targets.
  - The share of targets explicitly excluding wages rose to 37 percent over 2020–24 compared with 14 percent over 2006–10.
  - Overall, social spending targets in sub-Saharan African programs have declined as a share of both tax revenue and government expenditures over time; this decline is amplified by the growing share of specific targets and increased exclusion of wages.

### Box 2 — Rwanda (high-level summary and figures)
- Growth and social progress:
  - Real GDP growth averaged 7.5 percent per year from 2000 to 2024.
  - Poverty rate declined; life expectancy increased by nearly 18 years since 1990; infant mortality reduced by two-thirds.
- IMF-supported program design and priority spending:
  - 2002 program defined recurrent priority expenditures as total budget outlays identified by government as essential.
  - 2006 program outlined an increase in education and health expenditures by 0.6 percent of GDP, raising priority spending from 5 percent to 6 percent between 2005 and 2006.
  - More recent programs have maintained priority spending at around 10 percent of GDP, with over half allocated to capital expenditures.
  - Priority spending monitored through the integrated financial management system at the program level.
- Interpretation:
  - Rwanda’s broad definition of priority spending spans social sectors and productivity-enhancing strategies and reflects program ownership and alignment with national strategies.

### Box 3 — Côte d’Ivoire (high-level summary and figures)
- Context and outturns:
  - GDP growth since 2011 exceeded the sub-Saharan African average by approximately 3 percentage points.
  - Social outturns lag some regional peers; poverty rate declining but a large share still below the poverty line.
  - Côte d’Ivoire has consistently met its social spending targets under IMF-supported programs.
- Definitions and shifts in social spending targets:
  - Since 2009, a broad “pro-poor” definition captured total spending aimed at the poorest, encompassing approximately 30 percent of total government expenditures.
  - The most recent program introduced a more targeted measure of social spending in addition to the pro-poor aggregate (the pro-poor aggregate is still reported as a memorandum item).
  - Targeted social spending accounts for approximately 28 percent of pro-poor spending.
- Purpose:
  - The move toward a targeted measure aims to address specific poverty-reduction priorities where Côte d’Ivoire lags peers while maintaining transparency through publication of the pro-poor memo item.

*Sources: MONA database; World Economic Outlook; Rwandan authorities; Ivoirien authorities; IMF Staff Reports; and authors’ calculations.*

### Box 3. Increasing Social Spending and Supporting Inclusive Growth in Côte d’Ivoire

### Box 3. Increasing Social Spending and Supporting Inclusive Growth in Côte d’Ivoire (Concluded)

### Authorities’ Reform Priorities and Targeted Social Spending Definition
- The targeted social spending measure reflects the authorities’ reform priorities to promote human capital development, especially among women and young people, including through the establishment of the Government Youth Program (PJGouv).
- Health insurance coverage has been expanded as part of wider efforts to improve social welfare.
- The targeted definition of social spending under the current program includes separate categories for spending to improve youth employment and social protection.

### Trends in Education and Health Spending
- Real education spending per capita rose from about $23,000 in 2011 to $45,000 in 2020.
- Real health spending per capita increased from less than $9,000 to close to $22,000 over the same period.
- Nominal amounts were deflated using the consumer price index with 2014 as the base year.
- Despite real increases, other areas of public spending also rose, so social spending on education has fallen when measured as a share of total spending.

### Target Completion Rates and Comparisons Across Regions
- The average completion rate for social spending targets:
  - Sub-Saharan Africa: 69 percent
  - Europe: 62 percent
  - Middle East and Central Asia: 62 percent
  - Asia and Pacific: 66 percent
  - Western Hemisphere: 74 percent
- Completion rate by target specificity:
  - Nonspecific targets: 69 percent
  - Specific targets: 67 percent
- Completion rates for indicative targets (ITs) in sub-Saharan African programs:
  - Social spending ITs: 68 percent
  - ITs linked to fiscal revenues: 59 percent
- Target revisions and shortfalls:
  - In sub-Saharan Africa, 11 percent of social spending targets were subsequently lowered.
  - Where targets were missed in sub-Saharan Africa, the median shortfall was about 15 percent of the original target.
  - Share of completed targets completed only because of downward revision:
    - Social spending targets: 5 percent
    - Revenue targets: 12 percent

### Changes in Health and Education Spending During IMF-Supported Programs
- Before-and-after comparisons indicate health and education spending tend to rise during IMF-supported programs.
- Real social spending on health and education increases on average during IMF-supported programs, but health or education spending as a share of government expenditure is little changed on average.
- Median increases in both real health and real education spending over the course of IMF-supported programs are higher in sub-Saharan Africa than in most other regions; the interquartile range for education spending changes in sub-Saharan Africa lies above zero.
- Simple descriptive comparisons:
  - Mean and median percentage changes in real spending are higher in programs with social spending targets than in programs without such targets.
- Data limitations:
  - Cross-country data on social spending are limited, especially for low-income countries; analysis focuses on health and education because they are covered by 90 percent of sub-Saharan African social spending targets since 2002.
  - Analysis excludes pandemic years 2020–21 where noted.

### Econometric and Panel Regression Evidence
- A panel regression analysis of sub-Saharan African programs, controlling for country fixed effects, shows that both real health and education spending increase on average during program years.
- Results are not significant when looking at health or education spending as a share of government expenditure or GDP.
- A similar analysis did not find a significant relationship between the existence of social spending targets and changes in health or education spending over the course of the program, but tests are affected by the very small number of sub-Saharan African programs without social spending targets after 2010.
- Structural limitations:
  - Lack of data on social outcomes in sub-Saharan African countries and the generalization of social spending targets in all PRGT-supported programs after 2010 prevent a systematic analysis of the actual effect of those targets on education, health, or poverty.

### Social Spending Targets and Fiscal Consolidation
- Median and average movements by program with social spending ITs:
  - Fiscal deficits were reduced by 0.6 percentage points of GDP on average during IMF-supported programs with social spending ITs.
  - Education spending as a share of total government spending rose by 0.31 percentage points on average.
  - Health spending as a share of total government spending rose by 0.75 percentage points on average.
- For larger fiscal consolidations, the increase in the median share of social spending is mostly because of a decline in total spending, while social spending was less affected.
- Regression analysis of programs with social spending targets:
  - Showed a nonsignificant small negative effect of fiscal balance improvements on education and health spending as a share of GDP.
  - Showed a positive, albeit still not statistically significant, effect on education and health spending as a share of total government spending.
- Interpretation: Social spending targets appear to have limited the effect of spending cuts associated with fiscal consolidation on education and health spending.

### Evidence on Outcomes: Education Indicators
- Across 36 program cases in sub-Saharan Africa with available data:
  - The median share of young people not in education falls from more than 23 percent in the year before the social spending target was implemented to 21 percent two years after the start of the program.
- Event study patterns:
  - The share of young people not in education is typically above average in the year before social spending targets are implemented and falls in the two years after implementation.
- Literacy and structural outcomes:
  - Median literacy rates are found to stay constant during IMF-supported programs with education spending targets.
  - Event study results for literacy are not significant.
  - Note: Structural variables like literacy rates are slow moving and statistics are updated infrequently, so improvements may take longer to appear in the data.

*Source: IMF DEPARTMENTAL PAPERS — Social Spending Targets in IMF-Supported Programs in Sub-Saharan Africa (Box 3, concluded). The analysis uses MONA; UNESCO Institute for Statistics; WHO; World Development Indicators; and IMF Staff Calculations.*

### 1. Median Share of Youth Not in Education,

### 1. Median Share of Youth Not in Education, Employment, or Training before and after IMF-Supported Programs with Education Targets

### Education and youth outcomes
- Sample and timing
  - Analysis covers the period from two years before the start of the program to two years after the program agreement year (horizontal axis: year relative to the year in which the program was agreed).
  - Where outcomes data are missing, they are assumed to stay constant at the same level as the previous year.
  - Event study regressions include year and country fixed effects.
  - Countries are removed from the sample where any observation falls below the 1st percentile or above the 99th percentile of the sample.
  - Lighter shaded areas correspond to 90 percent confidence intervals about the central estimates, while darker areas correspond to 95 percent confidence intervals, based on robust standard errors.

- Key empirical notes on education outcomes
  - Event study results compare outcome levels before and after programs with other years across the same sample of countries in sub-Saharan Africa.
  - Data on education outcomes are updated infrequently in some cases; results should be treated with caution.
  - The horizontal axis in figures uses the convention Year of program (first year = 0), covering years −2, −1, 0, 1, 2, etc.

### Health outcomes (comparison and key statistics)
- Sample and headline changes
  - According to data from 39 program cases in sub-Saharan Africa, the share of births attended by skilled health staff increased from 62.5 percent to over 67 percent over the first two years of IMF-supported programs with health spending targets.
  - Median infant mortality rate falls from 55.6 per 1,000 live births to 49.3 during the first two years of the programs.
  - For sub-Saharan African countries more broadly, the median infant mortality rate falls by 1.5 per year (contextual comparison).

- Event study findings and interpretation
  - Event study analysis shows the share of births attended by skilled health workers increases somewhat during programs with health spending targets, although the differences are not statistically significant.
  - Event study analysis shows some decline in infant mortality after the implementation of health spending targets, but mortality rates remain well above average levels across the other years in the sample.
  - Similar to education statistics, data on the share of births attended by skilled health workers are updated infrequently; results should be treated with caution.

- Figure conventions referenced
  - Figures illustrating health outcomes use percent scales for births attended (e.g., 56 to 70 percent axes) and per 1,000 live births scales for infant mortality (e.g., 44 to 58 per 1,000).
  - Event study figures use year labels relative to program start (e.g., −2 to 4).

### Program design, implementation, and outcomes
- Trends in social spending targets
  - Social spending targets have become an increasingly important part of IMF-supported program design since 2010.
  - In sub-Saharan African countries, social spending targets have generally been in the form of ITs, not QPCs.
  - Coverage of social spending targets has increased over time to include social protection in many programs.
  - Target definitions have become increasingly specific.

- Ambition and completion
  - Targets in sub-Saharan African programs are larger than in other regions when measured as a share of fiscal revenues.
  - Targets have been achieved in many cases and completion rates for social spending ITs are high relative to other types of fiscal ITs.

- Spending behavior during programs
  - Education and health spending seem to increase over the course of most IMF-supported programs.
  - Programs including social spending targets have been accompanied by higher real education or health spending.
  - There is no evidence that education and health spending as a ratio of total spending or as a ratio of GDP have declined during programs.

- Caveats on causal interpretation
  - It is not possible to definitively conclude about the role played by social spending targets in observed increases in spending or outcomes.
  - Further work is required to fully test the robustness of findings given infrequent updates to education and health outcome data in some cases.

### Policy considerations and recommendations
- Design and targeting
  - Social spending targets should be tailored to the country context, including the adequacy and efficiency of existing social safety nets.
  - The design of social spending measures should consider the distributional effects of any macroeconomic adjustment (IMF 2019b).
  - Decisions on inclusion or exclusion of public teachers’ and health workers’ wages from social spending targets have important implications and need country-specific judgment.
  - The more frequent exclusion of wages from social target calculations in recent years suggests targets have often been designed to focus on protecting nonwage social spending.

- Efficiency, complementarity, and governance
  - There may be opportunities to improve the efficiency of social spending in some cases.
  - Social spending targets should avoid crowding out private provision of health, education, or other social services, which can complement the public sector.
  - Development partners can help better define social spending targets and aid effective implementation when domestic capacity is limited.
  - Weak governance, corruption, and lack of implementation capacity may account for poor social outcomes despite high social spending target completion rates.
  - Good public financial management practices with transparent public procurement processes, strong anti-corruption frameworks, and robust oversight are key to ensuring social spending floors yield durable development outcomes.

### Data sources, measures, and econometric specification
- Data sources and measurement notes
  - Education expenditure data are sourced from the UNESCO Institute for Statistics, aligned with the International Standard Classification of Education (ISCED) adopted by the UNESCO General Conference in 2011.
  - Public health spending data are based on the WHO’s Global Health Expenditure Database (GHED) and focus on current health spending as per the System of Health Accounts 2011 (SHA 2011) framework; GHED includes external transfers routed through government.
  - Total government spending, GDP, inflation, and revenue data come from the IMF World Economic Outlook database.
  - There is no comprehensive database for social protection spending because many governments do not use functional budget classifications and social protection spending can be done by several ministries.

- Four measures of social spending used in econometric analysis
  - Social spending as a share of GDP.
  - Real social spending.
  - Real per capita social spending.
  - Social spending in percent of total government spending.

- Econometric specification (fixed effects regression) as presented
  - Let Yit denote social spending in country i at time t, with Yit0 the social spending variable in the year prior to the program approval.
  - The fixed effects regression model is formally expressed as:
    ∆Yit,t0 = α + ∑5s=2 βs Dit,s + γi + εit,
    where ∆Yit,t0 = Yit − Yit0 denotes the change in social spending during program years; Dit,s is a dummy variable for each year t of the program; γi accounts for country-specific fixed effects; εit is the error term.

*Sources: MONA database; World Development Indicators; and authors’ calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/dp/2025/english/sstssaea.pdf_