Back Matter
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 2 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 3 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 4 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 5 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 6 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 7 https://isni.org/isni/0000000404811396, International Monetary Fund

Annex 1. Defining and Measuring Social Spending

There are different ways to measure social spending. While some elements of public spending on the wage bill and subsidies may have a social component, it is difficult to isolate this component in a manner that accounts for cross-country differences. Similarly, some social spending is carried out not by the government but by households or NGOs and aid agencies.

For the purposes of this paper we use a traditional definition of “social spending.” In line with IMF (2019b) we define as “social” all public spending on social protection (social insurance and social assistance), education services, and health services (Figure 1). This may understate the amount of social spending individual countries engage in but allows for a better cross-country comparison.

Other forms of government spending or policies may affect outcomes in a way similar to social spending. For example, capital spending on sanitation and clean energy should have an impact on health outcomes, as should regulations concerning workplace safety or food and medicines. Requiring individuals to contribute to private pension schemes can be a partial substitute for a public pension system and therefore reduce poverty amongst the elderly.

This annex addresses the issues outlined above. The efficiency of spending as estimated in this paper is a relative concept: a country’s spending is deemed more (less) efficient to the extent that it has better (worse) outcomes than other countries for a given measured spending level.1 Insofar as there are factors other than measured budgetary social spending which affect the outcomes we consider, our estimates of spending efficiency could be biased. However, the portion of Middle East and Central Asia countries’ public and private spending with social aspects not included in budgetary social spending seems broadly to occupy the middle ground compared to other country groups or should matter little due to other considerations. Therefore, the (relative) efficiency we consider should be robust to such issues.

Budgetary spending with social aspects. Higher public employment and/ or average compensation of public employees than is justified by the extent of public service provision is a form of social protection benefiting public employees. Subsidies, such as energy subsidies, can also be a form of social protection as they amount to a universal transfer to households, albeit one that tends to be greater for better-off households. There is evidence of excess spending on the public wage bill and considerable spending on energy subsidies in a number of Middle East and Central Asia countries (Annex Figures 1 and 2), however this should have little or no effect on the outcome variables considered in this paper. Such spending is in effect a transfer which mainly benefits relatively well-off households, allowing them, for example, to spend more on (private) education and health care which should have little or no effect on overall education and health outcomes, especially considering that the funds could go to improving public education and health care for a greater number of people. In the case of excess wage bill spending, Middle East and Central Asia countries appear to be in line with other country groups on average.

Annex Figure 1.
Annex Figure 1.

Pre-Tax Energy Subsidies, 2017

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF, FAD Country-level Energy Subsidies by Energy Product and Externality Component; and IMF staff calculations.Note: Pre-tax energy subsidies are estimated as the amount by which the cost of supplying energy products exceeds the price paid by its users. They do not take into account foregone revenue from unduly low taxation and are not always explicitly included in government budget figures.
Annex Figure 2.
Annex Figure 2.

Excess Wage Bill

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF Public-Private Sector Wage Premium Dataset; IMF Government Compensation and Employment Dataset, 2016; Worldwide Bureaucracy Indicators, World Bank; and IMF staff calculations.Note: “Excess wage bill” is defined as the amount by which the government wage bill exceeds what it would be if the public wage premium over the private sector were zero. The public wage premium is the amount by which public-sector pay exceeds private-sector pay for comparable levels of education, experience, etc. This concept of the excess wage bill implicitly assumes no public-sector employment surplus or deficit. Negative estimates of the wage premium are set to zero. The excess wage bill (EWB) is calculated as EWB = WB × WP/(1 + WP ), where WB is the actual wage bill and WP is the public-sector wage premium over the private sector (percent of private-sector earnings). Data labels use International Organization for Standardization (ISO) country codes.

Household “social spending.” Household spending may have effects similar to that of budgetary social spending (at least for the households with sufficient resources to undertake such spending). This includes charitable spending, such as zakat contributions.2 However, charitable spending appears either to be small, or to be part of the government’s budget and therefore should be captured by budgetary social spending data; for example, zakat contributions in Saudi Arabia are large because it is mandatory and collected like a tax (Annex Figure 3). One estimate is that only a quarter of total zakat contributions are made through formal certified organizations.3 However, even this suggests that total (formal and informal) contributions remain small relative to other forms of social spending, especially if one assumes that countries with mandatory contributions have much lower informal contribution levels. Private spending on education and health care, which is substantial in many Middle East and Central Asian countries (especially for health care), can also be a partial substitute for government spending in those areas (Annex Figure 4, Figure 6).4 Our regression analysis therefore controls for private health spending. The scarcity of data on private education spending prevents us from similarly controlling for such spending. However, the limited data available suggest that private education spending is small relative to public education spending, while the ratio of the two is broadly in line with other country groups, and therefore any bias in our results should also be small.

Annex Figure 3.
Annex Figure 3.

Zakat Contributions

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: International Policy Centre for Inclusive Growth Working Paper #168; Jordan National Zakat Fund; Palestinian Zakat Fund; UAE Zakat Fund; Saudi General Authority for Zakat and Tax; and IMF staff calculations.Note: Data do not capture Zakat contributions through other channels, such as direct person-to-person donations. WBG is the annual average for 2007–11. Data labels use International Organization for Standardization (ISO) country codes.
Annex Figure 4.
Annex Figure 4.

Funding of Education

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF, FAD Expenditure Assessment Tool; UNESCO; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes.
Annex Figure 5.
Annex Figure 5.

Net Official Development Assistance and Official Aid Received, 2018

(Percent)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF, World Economic Outlook; World Bank; and IMF staff calculations.Note: Very high values in APD are for small island states.
Annex Figure 6.
Annex Figure 6.

Social Spending

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Social spending financed from abroad. Services which fall under “social spending” may be financed and/or directly provided by official donors or international non-governmental organizations (NGOs). On-budget aid for social spending should already be covered by our measures of social spending, while at least some aid (whether on- or off-budget) may be for purposes other than social spending purposes. However, off-budget aid or

NGO spending for social programs could bias our results if they affect the outcome variables which we consider but are not reflected in the measures of social spending used in our analysis. Nonetheless, Middle East and Central Asia countries occupy the middle ground in terms of overall net official development assistance and official aid received compared to other regions (Annex Figure 5), suggesting that any omitted spending is, on average, similar to that in other regions.

Annex 2. Data Sources and Coverage

Data on government key economic variables were sourced mainly from the IMF World Economic Outlook (WEO) database. The indicators included GDP (nominal, real, and PPP), government expenditure (total, current, capital, and compensation of employees), inflation, trade-to -GDP ratio, compensation of public sector employees. The WEO aggregates for public wage bills and other fiscal indicators were for the general government.

Data on social spending were collected from several other sources. Public health and education spending came from the IMF Fiscal Affairs Department (FAD) Expenditure Assessment Tool. It was supplemented by data from IMF Government Finance Statistics (GFS) and World Bank World Development Indicators (WDI). The data on social spending are sometimes not available for certain time periods and countries as highlighted in Annex Table 1.

Data on socioeconomic indicators were collected from external databases. The World Bank WDI database was used to retrieve indicators of school enrollment and life expectancy at birth. Infant mortality rate, HDI, and expected years of schooling came from UNDP databases. Data on government effectiveness and the control of corruption came from the World Bank World Governance Indicators (WGI) database.

The emerging market and LIC economies country group corresponds to the WEO definition. The full list of countries can be located at the WEO portal on the IMF website: https://0-www-imf-org.library.svsu.edu/external/pubs/ft/weo/2017/01/weodata/groups.htm.

Country grouping used in the papers are in Annex Table 2. The table provides classification of countries in the region by LIC-MENAP, EM-MENAP, CCA, and GCC, and the ISO code used in the figures and tables.

Annex Table 1.

Data Sources and Coverage

article image
Source: IMF staff.Note: Coverage refers to the maximum number of countries in the data in a given year between 1990–2017.
Annex Table 2.

MCD Countries Classification

article image
Source: IMF staff.

Annex 3. Detailed Regression Results

Annex Table 3.

Regression Results for Human Development Index

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. 2SLS = two-stage least squares; FE = fixed effects; OLS = ordinary least squares; SGMM = systems generalized method of moments.

Instruments are share of agriculture in GDP and ethnic tensions index.

Includes lagged dependent variables, public social spending at time t, year dummies.

Annex Table 4.

Regression Results for Child Mortality Rate

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. 2SLS = two-stage least squares; FE = fixed effects; OLS = ordinary least squares; SGMM = systems generalized method of moments.

Instruments are share of agriculture in GDP and ethnic tensions index.

Includes lagged dependent variables, public social spending at time t, GNI per capita; and year dummies.

Annex Table 5.

Regression Results for HDI

(SGMM conducted on three-year averages)

article image
Sources: IMF staff estimates.Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. HDI = Human Development Index; PPP = purchasing power parity; SGMM = systems generalized method of moments.
Annex Table 6.

Drivers of Efficiency

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1.
Annex Table 7.

Drivers of Efficiency

(Output Efficiency Scores)

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1.
Annex Table 8.

Regression Results for Poverty Rate (FE)

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. FE = fixed effects.

Instruments are share of agriculture in GDP and ethnic tensions index.

Annex Table 9.

Regression Results for IHDI (FE)

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. FE = fixed effects; IHDI = Inequality-adjusted Human Development Index.
Annex Table 10.

Regression Results for Secondary School Enrollment (FE)

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. FE = fixed effects.
Annex Table 11.

Regression Results for Tertiary School Enrollment (FE)

article image
Source: IMF staff estimates.Note: Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. FE = fixed effects.

Annex 4. Social Spending Policies in Response to COVID-19 Crisis

Annex Table 12.

Key Social Spending Policies in Response to COVID-19 in the Middle East and Central Asia Region

(As of June 4, 2020)

article image
article image
Sources: National authorities, IMF Policy Tracker: Policy Responses to COVID-19, and IMF staff.

Annex 5. Technical Annex for Stochastic Frontier Analysis of Social Spending Efficiency

Stochastic Frontier Analysis (SFA) requires an explicit assumption of the functional form through which the inputs are generating the output, and an assumption about the distribution of the inefficiency term. In most practical applications, and in our case, the output frontier is estimated using Cobb-Douglas form:

lnyit=α+xitβ+ϵit

The error term, εit, is composed of two components: a white noise component that arises due to idiosyncratic shocks that the countries face, data errors, and/or measurement errors vit, and the inefficiency component uit.

ϵit=vituitvitN(0,σv2)uitF

While the white noise is normally distributed with variance s σv2, assumptions need to be made on the distribution of technical inefficiency for estimation. As the (in)efficiency term is positive, exponential, half-normal, truncated normal or gamma distributions are used for the inefficiency term, uit. With these assumptions on distribution of both the white noise and the inefficiency term, the combined error term εit is skewed, and this skewness is used to disaggregate the white noise from inefficiency using maximum likelihood estimations. To be specific, the outputs are produced using the following functional form, which also incorporates inefficiency component:

yit=eα+xitβeviteuit

The first exponential on the right hand side (RHS) is the deterministic component, the second is white noise, while the third component inefficiency. Bigger the efficiency as measured by e2-u, smaller the dampening effect of inefficiency on inputs in producing outputs.1

The ratio of standard deviation of inefficiency estimators σu, and white noise σn gives us an estimator l, that measures the relative contribution of inefficiency and white noise in the estimates of the regression standard error. Very small or very large values of l make inferences more difficult, as if λ → 0, there is no contribution of inefficiency (this reduces the estimation back to OLS) while when λ → ∞, everything not explained by inputs is inefficiency (non-stochastic, non-parametric estimations, for example, DEA).

Within SFA, depending on structure of data, we can use panel data for estimations or cross-sectional data for estimations. As the time series data are not sufficiently long, a “hybrid” of the two approaches is used, where the countries are grouped according to departments within IMF, and estimators are estimated using department and time dummies.

In the SFA estimations of efficiency of spending on education, health and on social safety nets, the (in)efficiency estimates would be assumed to follow a half normal distribution, and the efficiency will be estimated using the method proposed by Jondrow and others (1982).

As SFA is governed by a functional form, a benefit of SFA, particularly for policy advice, is that we can run counterfactuals (what-if type analysis). For example, once estimators are obtained, the counterfactual inefficiency values can be used to estimate the effect of reducing inefficiency on the output, or while keeping output constant, the effect of reducing inefficiency on input.

eui=Ei=yiteα+xitβevitΔyit=ΔEi(eα+xitβevit)

In percentage terms:

Δyityit=ΔEiEi

In case of economizing on inputs while keeping output the same,

ΔEi=yiteα+xitβevitβΔxitΔxijt=1βjΔEiEi

Here xit is the jth input while βj is the coefficient with jth input.

Annex 6. Case Studies: Social Spending Challenges in Selected Countries

Kingdom of Bahrain1

Bahrain’s social welfare programs have been quite generous and aim to share oil wealth and promote equity. The Bahraini economy is largely dependent on oil, and the derived wealth has been distributed over time among citizens through generous transfers and subsidized public sector jobs. The ultimate objective of Bahrain’s 2030 Economic is to improve living standards by promoting more attractive employment opportunities and higher wages. Generally, social spending in Bahrain covers a broad range of programs including health care, education, subsidized food and energy, and universal support for housing and employment in the private sector. Moreover, Bahrain has a variety of social welfare programs targeted to support low-income families, including unemployment benefits and insurance, disabled, elderly, and widowed, wage subsidies, and loans and grants. The public sector is a major employer of nationals, while the public pension system provides retirees with generous retirement benefits despite sustainability concerns.

Recent Trends in Social Spending

Public social spending has increased rapidly, from a high base. To promote inclusive and sustainable growth, social spending in Bahrain has been scaled up over the last decade, rising from 8.8 percent of GDP in 2008 to 13.1 percent in 2015 and remained about 10 percent of GDP in 2018 (Annex Figure 6).2 As described in Annex 1 and mentioned further below, these figures arguably understate the true extent of social support given fuel subsidies and generous public employment seen in many GCC countries including Bahrain. Spending on education and health consumes almost half of social spending related outlays, followed by subsidies. However, the urgent need to restore fiscal sustainability could limit available resources to fund existing social welfare programs.3

Social spending in Bahrain, however, remains low in some key areas when compared to other regions. For example, Bahrain’s health care spending per capita is close to US$1,200, slightly below the MENAP average and substantially below the OECD average, partly because of the young demographic structure of Bahrain’s population. While government expenditure per student is nearly twice the levels seen in MENAP and other emerging market economies, it remains 40 percent lower than the OECD average for primary education and 20 percent lower for tertiary education (Annex Figure 7). Other social spending outlays (especially unemployment benefits and family related spending) appear modest when compared to their average in OECD countries.

Annex Figure 7.
Annex Figure 7.

Public Expenditure on Health and Education

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: World Bank ASPIRE Database; World Bank, Education Statistics; World Health Organization, Global Health Expenditures Database; IMF, FAD Expenditure Assessment Tool; and IMF staff calculations.

Social spending is boosted by public sector employment. The government sector absorbs more than a third of employed nationals, with relatively high compensation compared to the private sector. Wages alone account for about 40 percent of government outlays, having increased to more than 10 percent of GDP in 2018, representing one of the highest public wage bills in the world. The wage bill is also more than quadruple the size of public development spending (Annex Figure 8).

Annex Figure 8.
Annex Figure 8.

Structure of Expenditures

(Percent of total expenditures, 2018)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Pension in Bahrain is a regressive transfer system, albeit it remains a powerful social protection tool. As of the end of September 2019, the number of pensioners has reached 80,000, compared to 150,000 government and private sector employees. Early retirement is prevalent in Bahrain, where 30 percent of pensioners are younger than 50, and 65 percent are younger than 60. The ratio of pension-to-wage before retirement, known as the replacement rate, is high by international standards, with workers receiving about 80 percent of their gross monthly salary upon retirement. The unfunded actuarial liabilities of the system are above 35 percent of GDP and represent the largest public contingent liability.

The government financially supports SMEs and employment in the private sector. A dedicated public authority, Tamkeen, was established in August 2006 to foster the development and growth of enterprises and provide support to enhance the productivity and training of the national workforce. Several innovative programs are provided to Bahraini individuals and businesses which include training, financing, grants, advisory, entrepreneurship support, and others. The Training and Wage Support Program provides financial supports for enterprises wishing to hire, train and/or increase the salaries of their Bahraini employees. Since its inception, Tamkeen has invested BD1.5 billion (10 percent of 2018 GDP) with more than 200,000 citizens and 50,000 companies being financially supported.

External grants remain an important complement for social spending. In 2011, the GCC countries announced an aid package (GCC Development Fund) worth US$10 billion, more than one-third of Bahrain’s GDP, to support higher social spending in Bahrain by upgrading the country’s housing and infrastructure and creating jobs over 10 years. Projects for an amount of US$7.5 billion have so far been committed in the area of housing, social development, health, education, and infrastructure.

Private spending on health and education has been rising. Private spending on health nearly doubled between 2005 and 2016, reaching 1.8 percent of GDP and accounting for about 40 percent of total health spending. Spending on education also increased by more than 20 percent during the same period to 1.2 percent of GDP.4 These trends indicate an increasing willingness by Bahrainis to invest in their own (or their children’s) health and education. While spending gaps remain, private spending on health and education has increased from 7 percent of total household spending in 2005 to 11.5 percent in 2015. The lowest two income quintiles of Bahrainis account for less than 25 percent of total health and education spending, while the top two quintiles’ share was close to 60 percent for health and 70 percent for education spending (Annex Figure 9).

Annex Figure 9.
Annex Figure 9.

Bahraini Equivalized Spending by Income Group and Quintile, 2015

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Social Spending Outcome: Preliminary Assessment

Bahrain scores high on the Human Development Index (HDI). Bahrain ranked at the 45th position in the 2019 HDI out of a total of 189 countries, which places Bahrain in the “Very High Human Development” group. Between 1990 and 2018, Bahrain’s HDI value increased by 21 percent (from 0.694 to 0.838), which is above the average for Arab countries (0.703) and close to the 0.875 average of the very high human development group. This high ranking reflects Bahrain’s continued quality improvement in the areas of health, education, and standard of living.

This increased spending on health has been accompanied by major improvements in health outcomes. Over the last 25 years, life expectancy has increased by about 4.5 years to 77 years. This compares to 73 years in EMs and 81 years in OECD countries. Infant mortality at birth has declined significantly in Bahrain, dropping from more than 19 per 1,000 births in 1990 to 6 in 2015, compared to 17 in EMs and 4 in advanced economies. However, total per capita health spending (in PPP-adjusted terms) in Bahrain is more than double the EMs’ level, suggesting spending inefficiencies relative to the EMs (Annex Figure 10).

Annex Figure 10.
Annex Figure 10.

Health Indicators and Outcome

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF, FAD Expenditure Assessment Tool; World Health Organization, Global Health Expenditures Database; and IMF staff calculations.

Education outcomes also reflect major progress. Net primary and secondary enrollments in Bahrain are exceptionally high, exceeding the respective ratios in OECD countries. Primary completion rate in Bahrain is also high at 98 percent by end 2018, above the 91.3 percent average for the MENA region and almost at par with the OECD average. The literacy rate among the youth (age 15–24) has jumped from 86 percent in 1980 to 99.7 percent in 2018, or 10 percent above the average for the MENA region. However, the teacher-student ratio for Bahrain is about 9 per 100 students, much higher than the MENAP average, while in OECD and in EM countries the ratio is about 8 and 5, respectively (Annex Figure 11). This finding indicates that there is scope to enhance education spending efficiency in Bahrain by reducing the teacher-student ratio.

Annex Figure 11.
Annex Figure 11.

Education Indicators and Outcome

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: IMF, FAD Expenditure Assessment Tool; World Bank ASPIRE Database; World Bank, Education Statistics; and IMF staff calculations.

Dividends from education are substantial and on the rise. Annex Figure 12 shows that households’ earnings in Bahrain increase with their education level, though education attainment is only one factor in determining an individual’s income. The education premium has increased significantly between 2006 and 2015 for people with graduate degrees, while the average monthly income of a bachelor’s degree holder remained close to BD1,600. Graduates of tertiary education earn on average more than double the people who completed only up to upper secondary education. The demand for higher education in Bahrain continued to grow in line with the authorities’ vision to produce graduates with skills and knowledge required in the global knowledge economy. The proportion of the adult population with tertiary education is particularly growing fast. In 2016, the gross enrollment ratio in the tertiary education has reached 50 percent of tertiary school-age population, up from 28 percent in 2005.

Annex Figure 12.
Annex Figure 12.

Average Household Income by Educational Attainment of Householder

(Bahraini dinars)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Social spending has reduced income inequality and poverty.5 The Lorenz curve shows that the Gini coefficient for equivalized household income declined 3 percentage points since 2005, to 32 percent in 2015.6 This was four points below the median for emerging and developing economies, slightly higher than for emerging Europe, and on par with several countries in the MENA region. The reduction in inequality was driven by labor income growth at the bottom of the income distribution as well as the increase in social programs (Annex Figure 13). Excluding current transfers (mainly pension, social security, and unemployment insurance) from household’s income, income inequality would have been higher by 4.2 percent in 2015 and 1.9 percent in 2005, indicating higher dependence of the bottom of income distribution on social insurance.

Annex Figure 13.
Annex Figure 13.

Contributions to Household Equivalized Income Growth, 2005–15

(Cumulative percentage change)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Higher wages and employment support consumption. Wages and salaries account for 54 percent of average household income in 2015, down from 62 percent in 2005. While employment in the public sector has grown by 1.5 percent annually during the period, average pay was higher by 6.5 and 3.0 percent in nominal and real terms, respectively (Annex Figure 14). Wages in the private sector increased by about one-third during the same period. These developments appear to be in line with Bahrain’s Economic Vision which aims to ensure that every Bahraini household has at least twice as much real disposable income by 2030 compared to 2008. Moreover, the wage of a public sector employee accounts for 75 percent of total income and is 40 percent higher than average wage in the private sector. Higher wages for civil servants were in part reflected in 7 percent higher consumption relative to other households. With the relatively high growth in private sector employment, about 2.5 percent annually over the last decade, Bahrain continues to have one of the lowest unemployment rates in the region (below 4 percent over the last decade).

Annex Figure 14.
Annex Figure 14.

Government Employment of Nationals and Average Wage Rate

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: National authorities; and IMF staff calculations.

Social spending boosted gender equality. Consistent with the authorities’ vision to empower Bahraini females and consolidate the principle of equal opportunities, gender income inequality appears moderate in Bahrain as a female income is on average 10 percent less compared with the income of a household headed by a male, after adjusting for the size of the household. Bahraini female workers in the public sector also represent 48 percent of total Bahraini workers. While a female earns on average 30 percent less in wages and salaries than a male, she receives more from public social protection spending. In particular, only 47 percent of a female gross income comes from wages and salaries compared to 60 percent for a male.

Conclusion and Policy Implications

Social spending in Bahrain has provided high-quality socioeconomic outcomes. Although social spending is largely at par with the MENA average, it remains low when compared to OECD countries. Bahrain’s fiscal sustainability concerns constrain increases to social spending. Introducing a direct taxation regime—corporate income tax, property tax, and personal income tax—could better insulate current programs, expand social spending plans going forward, and enhance the redistributive role of fiscal policy to promote equity.7 Consideration could also be given to improve the efficiency of government social spending, especially on health and education. Reducing the high teacher-student ratio, for example, could help achieve a sizable reduction in the education wage bill which accounts for more than 87 percent of the education current expenditure. Considering the very limited expenditure on education-supporting goods and services, part of the wage bill saving could be allocated to support better performance of teaching and non-teaching staff, finance learning materials, and boost capital investment in the sector. An effective implementation of the recently approved medical insurance law would also boost efficiency and encourage competition among public hospitals and between public and private hospitals and improve health care quality.

Republic of Armenia8

Armenia’s social protection programs have played a key role in promoting equality and reducing poverty rates. While the majority of poverty reduction can be attributed to strong growth and an improvement in the standards of living, Armenia’s public social spending has been consistent, and has contributed to increases in inclusive growth in the country. Poverty rates declined by 30 percentage points from 2004 to 2018, accompanied by a decrease in the Gini coefficient from 37.5 to 34.4. During the period of 2005–18, public social spending averaged about 11 percent of GDP. As part of Armenia’s precautionary Stand-By Arrangement (SBA) with the IMF, the authorities maintain an indicative target floor on social spending.9

Education and health spending are low compared to peers. Education spending averages 2.7 percent of GDP in 2017, compared to 4.3 percent in the CCA, and 4 percent in the MENAP and emerging market economies. Education spending is also low across the board in primary, secondary, and tertiary per capita levels. Health spending is only 1.9 percent of GDP, on the same level as the CCA, but lower compared the MENAP at 3 percent and 4 percent in emerging markets. Out-of-pocket health expenditure is the highest among peers, at 80 percent of total health expenditure in contrast to 32 percent in emerging market counterparts.

Expenditure on social assistance and pensions in Armenia is higher than peers. On average, social assistance spending in Armenia stands at 2.4 percent of GDP, on par with OECD levels, and higher than in MENAP and emerging market countries. Pension spending reached 4.9 percent of GDP in 2018, and authorities have introduced pension reforms. Such spending has helped Armenia make progress with reducing poverty and inequality.

Armenia spends less than peers on education and health, but achieves good outcomes, suggesting that spending is relatively efficient. Despite comparatively low spending on education, Armenia performs better in PISA/TIMSS10 than the average in the CCA, MENAP, and other emerging market peers. Net enrollment in primary and secondary school is also comparable, albeit slightly lower, to OECD levels. Enrollment in Armenia stands at 92 and 88 percent in primary and secondary school respectively, while OECD enrollment is 97 and 93 percent, and emerging market are at 91 and 76 percent.

Life expectancy at birth is in line with emerging market average at 74 years, while infant mortality of 11.6 per 1,000 people is half of the MENAP average of 25.3, and below the CCA average of 19.8 and emerging markets at 16.3. This suggests relative efficiency in education and health spending.

While education spending can be considered relatively efficient in comparison to peers, it is important to note that there remains room for improvement. While PISA/TIMSS scores are higher in Armenia contrasted to the averages in MCD and emerging market peers, results have stagnated over the years and remain below OECD levels. The highest TIMMS score for math and science of 470 was obtained in 2003 for Armenia,11 and later results in 2011 and 2015 (452 and 466, respectively) have not recovered to that level. Studies have also shown that there is a widening achievement gap in TIMMS score over time related to the socioeconomic background of students in Armenia (Caro and He 2018). Furthermore, the expected years of schooling of a child at the age of 4 in Armenia is 11.1 years, but the learning-adjusted years of schooling12 is only 7.9 years, suggesting some learning inefficiency. There is also a distinct gap in preschool enrollment between urban and rural areas. Overall, 30 percent of children under the age of 5 in Armenia are enrolled in preschool. This number drops to 17 percent in rural villages, as compared to 35 percent in urban settings.

Social safety net system is well-targeted but does suffers from insufficient coverage. Armenia’s Family Benefit Program (FBP) is a well-targeted cash-based social safety net system that accords priority to the very poor and the most vulnerable social groups such as the elderly, persons with disabilities, single mothers, orphans, and poor families with multiple children. This program is means-tested on income and other proxies for poverty risk factors. Targeting of the FBP is done using the household poverty and vulnerability scoring formula to rank applicants in terms of their expected poverty. The FBP achieves a good targeting performance—about 72 percent of the program resources go to the poor. However, the program coverage of the poor is low as less than one-third of the poor and about 12 percent of the population are covered (World Bank 2011). Increasing budget allocation to the program would extend benefits to the poor. In addition to the FBP, there are other small social assistance programs and benefits. These include universal cash transfers to expectant mothers and working mothers with infants younger than two, free access to health care for the poor, and social care services.

To support the economy and lessen the short-term impact of COVID-19, the authorities have taken several measures to preserve progress on inclusive growth and safeguard existing social spending (Annex Figure 15). Armenia’s drawing on the augmented precautionary SBA13 provided additional financial support to mitigate the pandemic and support affected households and businesses. This includes direct social assistance transfers to the most vulnerable, labor subsidies to SMEs to retain employees, and short-term subsidized government-sponsored loans to selected enterprises heavily affected by the crisis.

Annex Figure 15.
Annex Figure 15.

Social Spending in Armenia

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: ASPIRE Database; IMF, Expenditure Assessment Tool; national authorities; OECD; Social Snapshot and Poverty in Armenia 2019; UNESCO; World Health Organization; and IMF staff calculations.Note: In panel 1, the poor are defined as those with consumption per adult falling below $US88 a month, moderately poor are those who fall below $US73 a month, and the extremely poor are below $US51 a month.

Republic of Tunisia14

Despite remarkable improvements in Tunisians’ living standards over the past three decades, wide economic and social disparities persist with negative effects on inclusive growth and risks to economic stability. To address this challenge, the authorities have reinforced efforts to improve the adequacy, efficiency, and sustainability of social policies since the mid-2010s. Specifically, they have: (1) strengthened social assistance by scaling up benefit levels, widening coverage, and building up administrative capacity for better targeting; (2) improved the financial viability of the social security system by adopting a pension reform and shoring-up the funding of the health care fund; and (3) pursued institutional and governance reforms. These initiatives were supported by two IMF arrangements during 2013–19. Moreover, the authorities have started reflections on a new comprehensive social safety system, which could be implemented over the medium term.

The Challenge

Tunisia saw improvements in living standards over the past three decades (Annex Figure 16). Its gross national income (GNI) per capita grew on average above 5 percent per year over 1990–2010, stronger than that in regional and EM peer groups. Poverty, as measured by the headcount ratio at US$5.50 per day, fell by two-thirds to 18 percent over the past three decades; and inequality receded as measured by the Gini coefficient that fell to a reading of 0.33.15 Over the same period, the HDI increased by 30 percent, putting Tunisia in the high human-development category and at rank 91 out of 189 countries.

Annex Figure 16.
Annex Figure 16.

Socioeconomic Indicators, 1990–20181

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: World Bank WDI; UNDP; and IMF staff calculations.1 Simple averages using country data for the indicated year or the last available observation within -/+ three-year window.

Progress, however, has slowed after the Revolution and has remained uneven (Figure 1). Growth fell dramatically in the 2010s relative to the preceding decade and unemployment persisted at 15 percent, mainly affecting the young and women. Together with uneven access to quality public services, these trends have not helped alleviate social conditions for many Tunisians, particularly among low-income households and in the interior regions.16 The World Bank Human Capital Index (HCI) shows that a child born in Tunisia in 2018 will only be 51 percent as productive when she grows up as she could be if she enjoyed complete education and full health.17 This is below what would be predicted for Tunisia’s income level.

Improving social spending thus remains a crucial challenge for Tunisia. Better social protection and public services could help address today’s most pressing issues:18 low and not sufficiently inclusive growth,19 elevated social tensions, and weak trust in the government amid domestic security pressures and regional instability. The authorities have acknowledged that meeting this challenge entails (1) more and better-targeted social spending; (2) a financially viable social security system; and (3) institutional and governance reforms to improve spending quality. This note discusses this challenge in some detail: it will assess the performance of social spending, present the authorities’ reform agenda, and offer some lessons learned.

Social Spending Performance

On the surface, Tunisia enjoys both relatively high social spending and good socioeconomic outcomes (Annex Figure 17). Total public social spending— comprising current and capital spending, including on wages and subsidies— amounted to 14 percent of GDP in 2010. This was well above the average in MENAP and EM peers, mainly on account of more outlays for education. The spending helped Tunisia achieve better socioeconomic outcomes than its peers by 2018: expected years of schooling rose beyond 15 years, secondary school enrolment reached more than 90 percent, life expectancy climbed to almost 76 years; and the infant mortality rate fell below 1.2 percent.

Annex Figure 17.
Annex Figure 17.

Tunisia: Social Spending Performance, 1990–20181

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: World Bank, World Development Indicators; UN Development Programme; and IMF staff calculations.1 Simple averages using country data for the indicated year or the last available observation within -/+ three-year window.2 Comprising current and capital spending, that is, including wages and subsidies.

Looking more closely though, the performance of social spending remains an issue. Tunisia’s social protection rests on two pillars: (1) three contributory schemes, including the public and private pensions funds (CNRPS and CNSS), the public health fund (CNAM),20 and (2) several noncontributory public programs, notably the direct cash transfer scheme (PNAFN) and two health care programs (AMG1 and AMG2).21 This system of interventions suffers from resource constraints, fragmentation, as well as governance weaknesses. As a result, it has not been able to span an adequate, efficient, and sustainable social safety net over those in need.

  • Public services. Their effectiveness in addressing supply-side constraints remains limited, especially those emanating from deficiencies in education, health care, and labor market regulations and programs.

  • Social security. The system’s coverage remains too narrow. Only 37 percent of Tunisians contribute to the pension system and only half are covered by public health insurance (World Bank 2015a).22 A national unemployment insurance scheme does not exist. Moreover, social security suffers from deficits and arrears. Demographic change and financing gaps are posing threats to its sustainability. Declining fertility and increased life expectancy have resulted in an aging population, and unfavorable economic conditions over much of the 2010s have made it difficult for social security to collect sufficient contributions from employers and employees to maintain current levels of pension and health benefits (for example, up to 80 percent replacement income). The situation is likely to worsen in the years ahead with growing life expectancy and the weak cash flow of many state-owned enterprises. Moreover, the pension funds’ arrears to the public health fund, which otherwise could cover its costs, undermine the provision of basic health services in hospitals and pharmacies.

  • Social assistance programs. Existing programs are fragmented and face difficulties in raising sufficient funding to cover their needs (Annex Figure 18 and Annex Table 13). They also fail to cover a significant part of the low-income population and informal sector employees, and disproportionately benefit the better-off in urban areas. In fact, nearly a quarter of Tunisians are net beneficiaries of rather generous social transfers that represent up to one-fifth of total income. However, only two in five of these beneficiaries live below the national poverty line. This mainly reflects sizeable subsidies on food and energy (about 4.3 percent of GDP in 2019) that mostly accrue to the better-off: energy subsidies benefit rich households up to 30 times more than those with lower income. Moreover, about 15 percent of households live below the national poverty line, yet only 9 percent receive cash transfers under the country’s main social assistance program PNAFN and free health care under the AMG1 program. Household survey data further indicate substantial leakage from these programs to non-poor households (more than 50 percent of covered households may not be poor). An additional 20 percent of the population receive subsidized health care under the AMG2 program.

Annex Figure 18.
Annex Figure 18.

Tunisia: Social Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: Tunisian authorities; and IMF staff calculations.Note: PNAFN = Programme National d’Aide aux Familles Nécessiteuses.1 Under the IMF programs, defined as spending on social transfers and key ministries’ capital expenditures.
Annex Table 13.

Social Transfers and Capital Investment, 2019

(Millions of Tunisian dinar)

article image
Sources: Tunisian authorities and IMF staff calculations.Note: PNAFN = Programme National d’Aide aux Familles Nécessiteuses.

Under the IMF programs, defined as spending on social transfers and key ministries’ capital expenditures.

Ongoing Reform Agenda

The Tunisian authorities have accelerated social protection reforms over the past decade. This effort has resulted from a social dialogue that followed the immediate post-Revolution era, when the government—faced with high unemployment, social pressures, and inadequate social safety nets—had used the public wage bill and subsidies for energy and food products as inefficient substitutes for targeted social policies. The social protection reforms— supported by the IMF Stand-By Arrangement (SBA) and Extended Fund Facility (EFF) over 2013–19—have focused on the immediate improvement of coverage for low-income households and, in parallel, on reforms that enhance the resource allocation, sustainability, and efficiency of the social safety net in a context of significant resource constraints (Annex Figure 18). It is worth noting that the SBA and EFF programs have both included a floor on social spending (which comprises spending on social transfers as well as key ministries’ capital expenditures). This floor was elevated from an indicative target to a quantitative performance criterion starting in September 2018—the first in an EM program case.

  • 2013–15: First steps toward more adequate and sustainable social spending. The authorities started several multiyear reforms, notably: (1) an increase in the level of social spending from 2015 and (2) a dialogue with social partners on a pension reform that would eliminate the need for sizeable budget transfers to the pension funds.

  • 2016–19: The implementation of a more comprehensive reform agenda. The four main workstreams have included:

    • Increasing social spending. Spending on social programs (excluding general subsidies, key ministries’ wage bills, and transfers to the social security system) increased from 1.6 percent of GDP in 2016 to 2.8 percent of GDP in 2019, mainly to finance a scaling-up of social assistance to low-income households. Specifically, the authorities raised the benefits levels for PNAFN recipients (0.1 percent of GDP) in January 2018 and broadened its coverage from 250,000 to 285,000 households (about 10 percent of the total population compared to 15 percent of the population below the poverty line) since June 2018; improved the supply of free and subsidized health care (0.1 percent of GDP) since March 2019; and provided financial support for low-income households’ unpaid energy bills and for investment in health care infrastructure (0.2 percent of GDP) in 2019. In addition, the authorities augmented seasonal cash transfers to low-income families at various occasions (for example, Ramadan and the beginning of the school year), reduced social tariffs for low-volume electricity users, and augmented social integration programs.

    • Improving infrastructure for a better targeting of social assistance programs. This strand of work has involved (1) adopting legislation (that is, the “AMEN” law) that guides the transition toward a targeting system by early 2019; (2) building and validating a database of low-income households (registering more than 800,000 households, a quarter of all Tunisian households, half of whom were already surveyed over 2016–19); and (3) issuing electronic cards for medical care with a unique social identifier and a payment card for cash transfers to the current beneficiaries of the health care and PNAFN programs. In parallel, the work has focused on establishing the administrative capacity and infrastructure necessary for a targeting system (including the interoperability with the social security registries, a scoring model, and modular administration software).

    • Addressing liquidity pressures in the social security funds. A first-stage reform of the public pension fund (CNRPS) became effective in May 2019, involving higher contribution rates for employers and employees and a gradual increase by two years in the retirement age from 60 to 62. A government decree applying the same reform elements to the private pension plan (CNSS) remains pending. As a result of these reforms, from 2020 onward, the authorities expect no further need for transfers to the public pension fund beyond the yield of the social solidarity contribution.23 They also started discussions with social partners on a second-round pension reform that could involve deeper parametric change to ensure long-term financial viability. Further changes in the contribution system,24 reinforced recovery efforts by the pension funds, and some arrears clearance by the government helped address short-term liquidity pressures in the social security funds.

    • Boosting spending efficiency through institutional and governance reforms. The authorities have intensified their fight against corruption in the past five years, mainly by advancing anti-corruption legislation (including laws to protect whistleblowers and improve access to information, combined with stronger social accountability and more space for civil society). Challenges remain in making these laws effective, devoting more financial and human resources to the prosecution of corruption, and improving the independent judiciary (Transparency International 2019). Moreover, the authorities work on improving the quality and effectiveness of the public administration, notably through strengthening institutional capacity and digitalization (including that for targeting, see above).

The authorities have also started reflections on a new comprehensive social safety system, which could be implemented over the medium term. They intend to introduce a social protection floor (“socle social”) as advocated by the International Labour Organization (ILO).25 The objective is a nationally defined set of basic social security guarantees to alleviate and prevent poverty, vulnerability, and social exclusion through (1) universal access to essential health care and income security at least at a nationally defined minimum level (horizontal dimension) and (2) the progressive achievement of higher levels of protection within comprehensive social security systems (vertical dimension). This project would involve unifying under one roof Tunisia’s existing contributive and non-contributive schemes; and could allow for a more efficient delivery of social security guarantees in a three-tier system.26 However, progress has been slow amid a fierce debate about the adequate level of a social protection floor and in the presence of limited fiscal space.

Lessons Learned

Tunisia’s remarkable, yet uneven socioeconomic progress has recently slowed. Over the past three decades, living standards and human development indicators have improved and fare above levels seen in peer countries. At the same time, wide economic and social disparities persist across income groups and regions, with adverse effects on social stability and inclusive growth. Moreover, needs increased in the post-Revolution era, amid a slowdown in growth, stubbornly high unemployment, and persistent structural deficiencies.

Improving social spending is critical for addressing this challenge. The associated efforts need to ensure (1) adequacy of spending, which calls for spending increases given the population’s growing social needs; (2) efficiency of the various programs in achieving the desired socioeconomic outcomes, which often calls for better targeting of beneficiaries; and (3) the financial sustainability of the programs in a context of demographic change and budget consolidation.

The authorities have already made important progress, but more work lies ahead. Recent achievements include (1) first steps in strengthening social assistance by scaling up benefit levels, widening coverage, and building up administrative capacity for better targeting; (2) an improvement in the financial viability of the social security system that will eliminate or significantly reduce the need for ad hoc budget transfers, mainly through the public pension reform and the shoring-up of funding for the health care fund; and (3) some progress on the efficiency of social spending by pursuing institutional and governance reforms.

Further efforts are needed to achieve a better social safety net. Tangible progress, however, will take time. Tunisia’s experience shows that building consensus around reforms in the area of social policy is a complex challenge, especially in the presence of powerful vested interests, limited fiscal space,27 and large gaps in infrastructure and technical capacity.

Tunisia: Social Spending Efficiency1

  • Education. Schools and vocational training fail to address the growing skills mismatch among the low- and high-skilled workers in the face of the evolving needs of the private sector. Relative to regional and EM peers, Tunisia produces weak, and since the Revolution even deteriorating educational outcomes as measured by the Program for International Student Assessment (PISA, Figure 2.1). This reflects weaknesses in learning processes and content, and in the use of education spending. The wage share is high and has further grown from 88 to 93 percent over 2012–17, leaving only 4 percent for investment. Since 2005, the teacher headcount and their real wages have grown on average by 1.1 and 3.1 percent per year, respectively, while the number of students dropped by 0.5 percent in primary and by 1.7 percent in secondary education. This made the teacher-student ratio fall to levels seen in high-income countries, while teachers’ hourly work declined below that in peer countries. School administration has also become a payroll cost driver.

  • Health. Regional disparities persist in terms of access, headcount deployment, and management. Besides, vulnerabilities arise from out-of-pocket expenditures, notably for the less well-off. Technical and allocative inefficiencies weigh on health care inputs and output choices. They mainly emanate from (1) high and rigid wage spending (with more than three-fourths directed to permanent staff), (2) a subsidy system for pharmaceutical products (burdened by deficits and arrears accumulation), (3) little room for preventive care (with curative in- and outpatient care assuming three-quarters of health expenditures already), and (4) deficiencies in the referral system.

  • Labor market insertion programs. They perform poorly (with an average placement rate of 20 percent) owing to weaknesses in targeting, governance, and implementation. Further challenges arise from rigid labor market regulations, insufficient job creation in the formal private sector, high labor taxes, and large disparities between public and private sector compensation. At the same time though, precarious employment in the informal sector has grown (providing no coverage by social security and thus little protection from risks and shocks), fueled by poor access to finance and difficulties in crossing over into a highly regulated formal private sector.

Figure 2.1.
Figure 2.1.

PISA Test Results, 2010 and 2018

(Average score, 0–600)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A999

Sources: 0ECD; PISA; and IMF staff calculations.Note: PISA = Programme for International Student Assessment.
1 See World Bank (2015, 2015b, and 2018).

References

  • Afonso, António, Ludger Schuknecht, and Vito Tanzi. 2005. “Public Sector Efficiency: An International Comparison.” Public Choice 123 (3–4): 32147.

    • Search Google Scholar
    • Export Citation
  • Aisa, Rosa, and Fernando Pueyo. 2006. “Government Health Spending and Growth in a Model of Endogenous Longevity.” Economics Letters 90 (2): 24953.

    • Search Google Scholar
    • Export Citation
  • Albino-War, Maria, and others. 2014. “Making the Most of Public Investment in MENA and CCA Oil-Exporting Countries.” IMF Staff Discussion Note SDN 14/10, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alper, F. Ozlem, and Mehmet Demiral, 2016. “Public Social Expenditures and Economic Growth: Evidence from Selected OECD Countries.” Research in World Economy 7 (2): 4451.

    • Search Google Scholar
    • Export Citation
  • Atolia, Manoj, Bin Grace Li, Ricardo Marto, and Giovanni Melina. 2017. “Investing in Public Infrastructure: Roads or Schools?IMF Working Paper WP/17/105, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Baldacci, Emanuele, Benedict Clements, Sanjeev Gupta, and Qiang Cui. 2008. “Social Spending, Human Capital, and Growth in Developing Countries.” World Development 36 (8): 131741.

    • Search Google Scholar
    • Export Citation
  • Biolsi, Christopher. 2017. “Nonlinear Effects of Fiscal Policy over the Business Cycle.” Journal of Economic Dynamics and Control 78 (May): 5487.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bouhlila, Donia Smaali. 2015. “The Quality of Educational Performance of Students in the Middle East and North Africa: Much Remains to be Done.” ERC Policy Brief, Economic Research Forum, Cairo.

    • Search Google Scholar
    • Export Citation
  • Brixi, Hana, Ellen Lust, and Michael Woolcock. 2015. “Trust, Voice, and Incentives: Learning from Local Success Stories in Service Delivery in the Middle East and North Africa.” World Bank, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brockmeyer, Anne, Maha Khatrouch, and Gaël Raballand. 2015. “Public Sector Size and Performance Management: A Case-Study of Post-Revolution Tunisia.” World Bank Policy Research Working Paper 7159, World Bank, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Caro, Daniel, and Jia He. 2018. “Equity in Education in Armenia: Evidence from TIMSS 2003–2015.” Open Society Foundations, New York.

    • Search Google Scholar
    • Export Citation
  • Clements, Benedict, Sanjeev Gupta, and Gabriela Inchauste. 2004. “The Effectiveness of Government Spending on Education and Health Care in Developing and Transition Economies.” In Helping Countries Develop: The Role of Fiscal Policy. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Clements, Benedict J., David Coady, Stefania Fabrizio, Sanjeev Gupta, Trevor Serge Coleridge Alleyne, and Carlo A. Sdralevich. 2013. Energy Subsidy Reform: Lessons and Implications. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Dreher, Axel, Peter Nunnenkamp, and Rainer Tiele. 2008. “Does Aid for Education Educate Children? Evidence from Panel Data.” World Bank Economic Review 22 (2): 291314.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Easterly, William, and Sergio Rebelo. 1993. “Fiscal Policy and Economic Growth.” Journal of Monetary Economics 32 (3): 41758.

  • Erdoğan, Engin, Meliha Ener, and Feyza Arıca. 2013. “The Strategic Role of Infant Mortality in the Process of Economic Growth: An Application for High Income OECD Countries.” Procedia-Social and Behavioral Sciences 99: 1925.

    • Search Google Scholar
    • Export Citation
  • Filmer, Deon, and Lant Pritchett. 1999. “The Impact of Public Spending on Health: Does Money Matter?Social Science and Medicine 49: 130923.

    • Search Google Scholar
    • Export Citation
  • Gaspar, Vitor, David Amaglobeli, Mercedes Garcia-Escribano, Delphine Prady, Mauricio Soto. 2019. “Fiscal Policy and Development: Human, Social, and Physical Investment for the SDGs,” IMF Staff Discussion Note 19/03, International Monetary Fund, Washington.