Social Spending for Inclusive Growth in the Middle East and Central Asia
  • 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

This paper examines the role of social spending in improving socioeconomic outcomes in the Middle East and Central Asia. In particular, it addresses the following questions: (1) how large is social spending across the region? (2) how do countries in the region fare on socioeconomic outcomes? (3) how important is social spending as a determinant of these outcomes? and (4) how efficient is social spending in the region?

Abstract

This paper examines the role of social spending in improving socioeconomic outcomes in the Middle East and Central Asia. In particular, it addresses the following questions: (1) how large is social spending across the region? (2) how do countries in the region fare on socioeconomic outcomes? (3) how important is social spending as a determinant of these outcomes? and (4) how efficient is social spending in the region?

Chapter 1 Introduction

The Middle East and Central Asia region faces an urgent need for more inclusive growth. Demographic pressures, youth unemployment, poverty, and high and rising inequality are challenging policymakers to create opportunities for all. At the same time, several countries in the region are dealing with internal conflicts, large inflows of refugees, and heightened security risks. In addition to domestic conditions, a more challenging external environment, with slowing global growth, uncertainty related to trade, and geopolitical risks, weigh further on economic prospects and hinder the ability of countries in the region to meet the SDGs. The recent COVID-19 pandemic has magnified these challenges and exposed significant vulnerabilities both in health infrastructure and social safety nets in the region. Public finances have been significantly stretched to deal with existing needs as well as the human cost of the pandemic and to contain its economic fallout. In many countries, financing constraints limit the availability of budgetary resources.

Against this background, this paper examines the role that social spending can play in improving social and growth outcomes in the Middle East and Central Asia. Previous IMF work has explored the effects of overall fiscal spending and of infrastructure spending on socioeconomic outcomes (IMF 2017, 2018), as well as the importance of reorienting fiscal policy toward promoting inclusive growth (IMF 2019b). This paper builds on those earlier studies by focusing on social spending. We show econometrically that increased public expenditure on education, health, and social protection lead to better education and health outcomes, reduced poverty and inequality, and stronger growth overall. We also document that—despite impressive, albeit uneven, progress over the past decades—Middle East and Central Asian countries generally lag global comparators in socioeconomic outcomes. Moreover, levels of social spending are typically lower than in global peers. This highlights the case for (1) increasing budgetary allocations and, given fiscal sustainability considerations, (2) improving the efficiency of social spending. The first objective requires a reprioritization of the existing expenditure envelope and/or enhanced revenue mobilization, with the precise mix varying across countries and dependent on a deeper analysis of fiscal space across the region that is beyond the scope of this paper.1 The second objective requires policy efforts to address the factors underpinning relative spending inefficiency, such as institutional weaknesses, governance problems, and poor financial inclusion.

The COVID-19 crisis has underscored the need for strong health systems and effective frameworks to channel cash transfers to vulnerable households.2 The crisis has forced a swift and concerted national and multilateral response to ensure adequate public spending on health and social protection, so as to cushion the human and economic toll. The pandemic is still unfolding, but the region has already demonstrated its capacity to quickly mobilize and deploy additional resources for health and social protection including through greater use of technology, to reach the most vulnerable. Adequate social protection can help reduce poverty and inequality and ensure the welfare of the most vulnerable. Beyond the crisis, more equitable access to education andhealth care can contribute to human capital accumulation and inclusive growth. Ensuring that a country has enough fiscal space to undertake adequate investment in human capital and can do so efficiently is key to preserving fiscal sustainability and enabling long-term growth.

The paper is organized as follows. Chapter 2 discusses the definition of social spending used in the paper and its possible limitations. Chapters 3 and 4 document how countries in the region fare in terms of the level of social spending as well as socioeconomic outcomes. The paper next presents econometric analyses of the extent to which social spending affects socioeconomic outcomes. The final sections analyze the efficiency of this social spending and explore factors that may be driving inefficiency in the region. The paper concludes with policy recommendations, drawing also on the annex that presents three case studies from the region.

Chapter 2 Defining Social Spending

This paper uses the traditional definition of “social spending” adopted in the literature. Consistent with IMF (2019b) social spending is hence defined as on-budget government spending on social protection, education, and health (Figure 1).

Figure 1.
Figure 1.

Definition of Public Social Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Source: IMF (2019b).

Other types of spending may also have a social component but are often seen as inferior to well-designed public spending (Annex 1). A large public wage bill due to high public-sector employment and/or a public-private wage premium can be seen as a form of social protection, albeit one that may be poorly targeted, adds to budget rigidities, and—in the case of a high wage premium—also disincentivizes private-sector employment (Tamirisa and Duenwald 2018).1 Likewise, some subsidies may have a social-protection element, as they amount to a universal transfer to households, though here too the benefits are seen mostly by the rich, at least in absolute monetary terms, and incentives again are distorted (toward overconsumption) (Figure 2). Countries in the region spend considerably on these (Figure 3). Private outlays on education2 and both domestic and foreign charitable spending in these areas (including off-budget, foreign-aid-funded spending) may also have a material impact on social outcomes, though it is worth noting that private spending cannot substitute for public when it comes to serving poorer segments of the population. Comprehensive cross-country data on these types of expenditure are not available, but to the extent that they are broadly similar, overall, as in the rest of the world, their omission is unlikely to bias our results. Thus, in this paper, we focus mainly on the literature’s standard definition of social spending, for which systematic, cross-country data are available (Annex 1).

Figure 2.
Figure 2.

Wage Bill, Subsidies, and Social Spending

(Percent of GDP, 2018 or latest available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, World Economic Outlook; IMF, FAD Expenditure Assessment Tool; and IMF staff calculations.
Figure 3.
Figure 3.

Selected Budgetary Spending with Social Aspects

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, FAD Country-level Energy Subsidies by Energy Product and Externality Component; IMF, Public-Private Sector Wage Premium Dataset; IMF, Government Compensation and Employment Dataset, 2016; World Bank, Worldwide Bureaucracy Indicators; The World Bank; 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. “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. Data labels use International Organization for Standardization (ISO) country codes.

Chapter 3 The Level of Social Spending in the Middle East and Central Asia

This section compares the relative size of social spending of countries in the Middle East, North Africa, Afghanistan, and Pakistan (MENAP) and the Caucasus and Central Asia (CCA) regions with global peers. As the Middle East and Central Asia (MCD) region includes low-income, emerging market, and high-income economies, it is important to ensure proper benchmarking across global peers.1 As income levels are an important determinant of social spending levels, we compare low-income countries in MENAP with other low-income countries (LICs), emerging markets in MENAP (EM-MENAP) as well as the CCA with other emerging markets (EMs), and GCC countries with advanced economies (AEs).

While there is significant cross-country diversity, social spending in the region is generally lower than in other parts of the world (Figure 4).2 Governments in the region devote 10.4 percent of GDP on average to social spending, compared to an EM average of 14.2 percent. LIC-MENAP countries’ level of social spending is particularly low, averaging 8 percent, compared to the global LIC average of 14 percent of GDP. GCC countries spend less than AEs. The difference is also striking in terms of purchasing power parity (PPP) per capita spending, where, for example, EM countries in the MENAP region spend an average of US$1,220 on social outlays compared to US$1,978 spent by EMs globally. It is worth emphasizing that simple cross-country comparisons like these are just a starting point—a full analysis of the adequacy of social spending would need to account carefully for country-specific circumstances. Nonetheless, the relatively low levels of social spending across country groupings in the region are notable and suggest the need for further, bottom-up sectoral analyses of spending needs.

Figure 4.
Figure 4.

Public Social Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Bank ASPIRE Database, World Bank, Education Statistics; World Health Organization, Global Health Expenditures Database; IMF, World Economic Outlook; IMF, FAD Expenditure Assessment Tool; and IMF staff calculations.Note: PPP = purchasing power parity.

The additional spending needed to reach the SDGs underscores the scale of the challenges faced by the region. The median country in the Middle East and Central Asia needs to spend an additional 5.3 percent of GDP per year by 2030 to achieve five critical SDGs covering human, social, and physical capital, and many MCD countries would need even more spending (Figure 5). Indeed, this estimate is a lower bound, as it assumes that spending efficiency is at the frontier—for less-efficient countries, even larger additional spending would be needed.3

Figure 5.
Figure 5.

Additional Spending Needs in 2030 to Meet Selected SDGs

(Percentage points of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Source: IMF staff calculations based on Gaspar and others (2019).Note: Additional annual spending required on education, health, roads, electricity, and water and sanitation to meet the corresponding UN Sustainable Development Goals (SDGs), relative to a baseline of current spending to GDP in those sectors.

The level of public health care expenditure is generally lower than in global comparators, while private expenditure on health is relatively large (Figure 6). On average, countries in the region spend 6 percent of GDP on health care, of which 3 percent is public expenditure and 3 percent is private. Private health expenditure in the CCA and LIC-MENAP comprises about 71 percent of overall health expenditure, perhaps reflecting the unavailability of extensive public medical services. This composition of expenditure, tilted toward private sources, raises concerns about the access of poorer individuals to health services. In particular, low public health spending often implies a larger financial burden for individuals due to high out-of-pocket expenses, which in turn is a significant barrier to accessing health care, especially among the poor and vulnerable.

Figure 6.
Figure 6.

Public and Private Health Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, FAD Expenditure Assessment Tool; IMF, World Economic Outlook; World Health Organization, Global Health Expenditures Database; and IMF staff calculations.Note: The data do not reflect the recent scaling-up of health expenditure due to the COVID-19 pandemic. Data labels use International Organization for Standardization (ISO) country codes. STD = standard deviation.

Public education spending in the region is also lower than in global peers (Figure 7). On average, governments in the region spend 3.5 percent of GDP on education, whereas global EMs are at 4.2 percent. This pattern is observable across all country groupings and their peers. GCC countries spend less on education, relative to GDP, than their AE counterparts, but they spend relatively more in terms of PPP US dollars per capita spending. LIC-MENAP countries spend considerably less than any other group, at only 2.6 percent of GDP. Comprehensive data on private education spending in the region are not available.

Similar patterns are observable with regard to social-protection spending (Figure 8). On average, countries in the region spend 4.9 percent of GDP on social protection, compared to 6.6 percent in EMs. Social spending in the CCA is comparable to that of EMs (6.6 percent of GDP for both), but EM-MENAP countries spend less (5.7 percent). The most striking difference is between GCC countries and their AE peers, with a spending gap of 11.7 percentage points. However, the variability of social protection spending in PPP$ per capita terms in the GCC is high, ranging from US$280 in Qatar, to US$7,200 in Kuwait.

Figure 7.
Figure 7.

Public Education Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Bank, Education Statistics; IMF, World Economic Outlook; IMF, FAD Expenditure Assessment Tool; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes. PPP = purchasing power parity.
Figure 8.
Figure 8.

Social Protection Spending

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, FAD Expenditure Assessment Tool; IMF, World Economic Outlook; World Bank ASPIRE Database; and IMF staff calculations.Note: The data do not reflect recent scaling up of social spending expenditure due to the COVID-19 pandemic. STD = standard deviation.

Most countries in the region substantially increased social spending in response to the COVID-19 crisis. Given the varying states of preparedness across the region, including in health care infrastructure, and the varying speed at which the pandemic spread, responses differed across countries (Table 1). An important part of the response was to support health care spending, and at least half of the countries in the region announced plans for targeted support to lower-income and vulnerable households and informal workers.4 IMF emergency financing operations helped many countries in the region to achieve these goals. Some countries (Egypt) also increased education spending. Most of the social protection assistance is directed through cash transfer programs, and many countries have made effective use of technology in delivering these programs. Although country heterogeneity is substantial, the fiscal cost5 of responding to the pandemic has generally been significant—on average above 2 percent of GDP, although somewhat smaller than in global peers (Figure 9; Box 1).

Table 1.

Social Protection Responses to COVID-19 in MCD Countries

article image
Sources: Gentilini (2020); national authorities; and IMF staff.
Figure 9.
Figure 9.

Fiscal Cost of COVID-19, 2020

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: National authorities; IMF COVID-19 Country Surveys; and IMF staff calculations.

Chapter 4 Socioeconomic Outcomes in the Middle East and Central Asia

We focus on standard socioeconomic indicators used in the literature for which we have the widest data availability, both in time and country coverage. For health outcomes, we focus on infant mortality rate and life expectancy at birth, which are two standard indicators widely used in the empirical literature to assess the relationship between health and economic progress as well as to measure the effectiveness of health expenditure (Erdoğan, Ener, and Arıca 2013; Aisa and Pueyo 2006). For education outcomes, we focus primarily on secondary school enrollment and expected years of schooling. These are also standard indicators used in the literature (Clements, Gupta, and Inchauste 2004; Afonso, Schuknecht, and Tanzi 2005) and have the widest data availability, both in time (since 1990) and country coverage. While these indicators may provide little information on the quality of education, they seem to be positively correlated with Program for International Student Assessment (PISA) and Trends in International Mathematics and Science Study (TIMMS) scores for countries where the data are available (Figure 10). We supplement this analysis with other educational outcome indicators, such as learning poverty (LP).1 For the aggregate socioeconomic outcomes, we look at the HDI,2 the inequality-adjusted HDI (IHDI),3 income per capita, poverty rates, and the World Bank’s Human Capital Index (HCI).

Figure 10.
Figure 10.

Education Outcomes

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Bank ASPIRE Database; TIMSS and PISA Evaluations; IMF, World Economic Outlook; IMF, FAD Expenditure Assessment Tool; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes.PISA = Program for Internation Student Assessment; TIMSS = Trends in International Mathematics and Science Study.

We find that socioeconomic outcomes in the region have improved substantially over the last two decades (Figures 10 and 11). Nearly all countries in the region made impressive absolute gains in health and education outcomes over the past two decades, as well as in poverty reduction. Even in relative terms, except for low-income countries, MENAP economies posted larger-than-average socio-economic gains. Tunisia (Annex 6) and Oman, for instance, are among the top 20 countries worldwide in terms of increasing secondary-school enrollment since 1990, and Morocco is in the top 20 countries for improvements in the HDI. Notably, LIC-MENAP countries reduced the secondary-school enrollment gap with other LIC peers, mostly because of closing the gender gap. Saudi Arabia also made significant progress in eliminating gender gaps in access to education and encouraging greater female labor force participation. Higher female secondary-school enrollment is associated with lower fertility rates (Figure 12), improved female literacy, and lower infant mortality. In the CCA, secondary school enrollment and expected years of schooling dropped following the collapse of the Soviet Union but have since recovered.

Figure 11.
Figure 11.

Improvement in Socioeconomic Indicators

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Development Indicators, Learning Poverty (October, 2019); World Bank and UNESCO Institute of Statistics; and IMF staff calculations.1 Years a current 2-year-old is expected to spend in school based on current enrollment rates of 2- to 29-year-olds.2 Children enrolled in secondary schools as a share of that age group; can exceed 100 due to repeaters and late/early enrollments.3 Number of deaths in the first year of life per 1,000 live births.4 How long, on average, a newborn can expect to live, if current death rates do not change.
Figure 12.
Figure 12.

Fertility Rate vs. Female Secondary Enrollment1

(Averages from 1980 to 2018)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Bank Development Indicators; and IMF staff calculations.1 Each dot represents average for each group in a specific year, starting from 1980.

However, the rate of progress in improving socioeconomic outcomes appears to be slowing down. In reducing infant mortality and achieving higher life expectancy, MENAP countries made progress at the same rate or higher rate than global peers during 1990–2000. However, the rate of progress has been slowing down, particularly over the last decade, perhaps reflecting in part conflicts in several countries which lead to internal displacement, increased refugee flows, and rising poverty. For example, life expectancy at birth and secondary school enrollment fell after the start of the conflict in Syria.

As a result, the region still lags its peers on health and education indicators (Figure 13). For example, despite their higher income levels, the GCC’s infant mortality rate is twice that of advanced economies. Infant mortality is also higher in CCA countries and in EM-MENAP and LIC-MENAP compared to their global counterparts. Similar trends are also visible in education, where MENAP emerging markets lag their peers in secondary school enrollment rates by 13.9 percentage points and in expected years of schooling by 1.4 years. Other educational outcomes tell a similar story: adult literacy rates in EM-MENAP are at 76 percent, which is 17 percentage points lower than the EM average.4 Data from the United Nations Development Programme (UNDP) Education Index reveal similar results, as countries in the region, with the notable exception of CCA, fare worse in educational outcomes than their global peers.

Figure 13.
Figure 13.

Socioeconomic Outcomes in Middle East and Central Asia Countries and Relevant Global Peers

(2018 or latest available value)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: The Human Capital Project; World Development Indicators, Learning Poverty (October 2019); World Bank and UNESCO Institute of Statistics; and IMF staff calculations.Note: STD = standard deviation.1 Number of deaths in the first year of life per 1,000 live births.2 How long, on average, a newborn can expect to live, if current death rates do not change.3 Years a current 2-year-old is expected to spend in school based on current enrollment rates of 2- to 29-year-olds.4 Children enrolled in secondary schools as a share of that age group. The percent can exceed 100 due to repeaters, late enrollments, or early enrollments.

Most of the region also lags on aggregate indicators of wellbeing (Figure 14). For example, despite much higher gross national income (GNI) per capita, GCC countries have lower HDI scores than advanced economies globally. Emerging and low-income MENAP countries lag their comparators in both GNI per capita as well as HDI scores. Inequality-adjusted HDI scores are also relatively low in emerging and low-income MENAP countries. CCA countries, however, have a higher average inequality-adjusted HDI score than EMs. In terms of the Gini coefficient, countries in the region score better (lower) than their global EM and LIC counterparts. Finally, poverty rates in the region are generally somewhat higher than in global peers, and the distribution of income somewhat more unequal.

Figure 14.
Figure 14.

Socioeconomic Indicators

(2017 or latest available value)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Development Indicators; United Nations Development Programme; and IMF staff calculations.Note: PPP = purchasing power parity; STD = standard deviation.

Chapter 5 Impact of Social Spending on Socioeconomic Outcomes

The extent to which social spending matters for socioeconomic outcomes remains a subject of discussion in the literature. Haile and Nino-Zarazua (2018) find a statistically significant impact of social spending on the IHDI and on child mortality. Alper and Demiral (2016) find that social spending boosts growth, and Gupta, Verhoeven, and Tiongson (2003) find that health spending improves health outcomes. Baldacci and others (2008) conclude that education and health spending have a significant impact on education and health capital, but that improving governance and taming inflation could help to achieve the same outcomes. On the other hand, Filmer and Pritchett (1999) find no effect of public health spending on child mortality on account of inadequate institutional capacity and market failures. Likewise, Rajkumar and Swaroop (2008) show that public spending has virtually no impact on health and education outcomes in poorly governed countries, whereas they find a positive impact of public spending in countries with good governance. Most prior empirical work finds that social spending, especially when accompanied by good governance, is associated with better social outcomes and higher growth.

We use a range of econometric methods applied to a global panel data set to tackle the question of whether social spending matters for socioeconomic outcomes. Our data cover 191 countries during 1990–2017. Data sources are described in detail in Annex 2. Socioeconomic outcomes enter the regressions as dependent variables.

We estimate the following equation:

Outcomeit α + B1(SSpending)it-1 + B2(Z)it + μr + μt + εi,t

Outcome refers to a set of socioeconomic outcomes. We estimate five different models using health-related outcomes (child mortality rate, life expectancy at birth), education-related outcomes (secondary school enrolment rate, expected years of schooling) and overall welfare outcome (Human Development Index) as dependent variables. We also consider specifications with poverty rates and inequality-adjusted HDI measures on the left-hand side.1 The term S_Spending denotes social spending as a percentage of GDP (or in PPP dollars per capita terms) in the previous year.2 Z refers to a vector of control variables; μr and μt denote unobserved region-specific effects and time effects, respectively; and εi,t represents the disturbance term.

We rely on standard controls used in the literature. To control for the structure of the economy, we use standard variables from the literature. Inflation is used to proxy macroeconomic stability, the sum of exports and imports to GDP to proxy trade openness, and the share of domestic credit to GDP is used to control for the level of financial development. To control for institutional quality, we include indices of government effectiveness and control of corruption from International Country Risk Guide (ICRG) and World Bank’s World Governance Indicators (WGI). Prior literature provides evidence of a strong correlation between health outcomes and access to safe drinking water and sanitation facilities (Rajkumar and Swaroop 2008), the degree of urbanization (Schultz 1993), and fertility rates (Mishra and New-house 2009). We add these as controls in the health-outcome regressions. We also control for external and domestic conflict.

Simple estimation methods suggest that there is a positive and statistically significant relationship between social spending and socioeconomic outcomes (Figure 15). We start with pooled OLS estimates with regional dummies and then add country fixed effects. Regressions are conducted on a global sample. The results suggest that there is a positive and statistically significant (at the 1 percent level) relationship between social spending and the HDI, IHDI, and poverty reduction (Annex 3). The results also suggest that public spending on education is associated with higher secondary-school enrollment and expected years of schooling, while higher public health expenditure is associated with greater life expectancy and lower infant mortality. In all these specifications we use lagged social spending to control for endogeneity.3 The results do not change regardless of whether we use social spending in percent of GDP or in PPP dollars per capita terms, or whether the analysis is conducted on annual data or on three-year or four-year averages. Detailed regression results, including various specifications, are presented in Annex 3.

Figure 15.
Figure 15.

Public Social Spending and Socioeconomic Outcomes

(1990–2017)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: United Nations Development Programme; World Development Indicators; World Bank ASPIRE Database; World Health Organization; and IMF staff calculations and estimates.

Addressing endogeneity is a key challenge.4 Some countries may choose to spend more on social objectives precisely because their outcomes are poor, and if this reverse causality is not accounted for, the estimates can be biased and wrongly suggest that social spending is effective. Although our OLS and fixed effect regressions always relate current outcomes to spending of the year before, this lag structure alone may not be sufficient to eliminate endogeneity.

We use instrumental variable (2SLS) estimation and the generalized method of moments for a system of equations (SGMM) to control for endogeneity. In our 2SLS estimation, we employ a range of instruments—variables correlated with spending but credibly unaffected by outcomes—to correct for endogeneity. These are standard instrumental variables (IVs) used in the literature. Easterly and Rebelo (1993) used log of population as an IV, arguing that smaller countries suffer from diseconomies of scale and have to spend more. The share of agriculture in GDP was used by Tanzi (1992) since agrarian societies have a weaker revenue base and tend to have lower spending. Von Hagen (2005) argued that ethnic tensions may result in suboptimal allocation of public spending by compounding a “common pool” problem. Recent papers by Haile and Nino-Zarazua (2018); Gisselquist, Leiderer, and Nino-Zarazua (2016); and Dreher, Nunnenkamp, and Thiele (2008) used the same set of instruments.5 The SGMM specification also allows us to control for the persistency of the dependent variable (HDI) and to demonstrate that the results are robust and the relationship between social spending and socio-economic outcomes is not spurious.

The econometric results suggest that public social spending has an appreciable effect on socioeconomic outcomes (Table 2). Higher public social spending is associated with a higher HDI, even after controlling for income, the degree of urbanization, macroeconomic stability, trade openness, domestic and external conflict,6 and the level of financial development. This conclusion holds regardless of estimation methods, use of different specifications, and whether the analysis is conducted on annual or three- and four-year averages (Annex Tables 3 and 5).7 Even the most conservative coefficient from the SGMM estimation points to an economically significant impact on the HDI, as it is scaled from 0 to 1.

Table 2.

Regression Results for HDI Outcome

(Annual data for 2SLS, and three-year averages for SGMM)

article image
Source: IMF staff estimates.Note: HDI = Human Development Index; SGMM = Systems Generalized Method of Moments; 2SLS = two-stage least squares.

Instruments are a share of agriculture in GDP and agriculture as a share of GDP. Bolded coefficients are significant at least at 5 percent level.

Includes lagged dependent variable; public social spending at time t.

Public spending on health and education has a significant impact on health and education outcomes (Table 3 and Annex 3). For the health-outcomes regression, besides the standard controls used above, we also use access to safe water as one of the explanatory variables. Since data on private health expenditure are available, we also use it as a regressor. Private health expenditure also matters for reducing child mortality, but it is less statistically significant, and the effect is smaller than for public health spending. While the private sector may be more efficient at delivering services for individual households, public health care spending seems to matter more for improving aggregate welfare indicators—lowering poverty rates, improving life expectancy, and reducing child mortality (Table 3). There are insufficient data on private education spending to include this variable in our regression analysis of determinants of education outcomes. 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 (see Annex 1).

Table 3.

Regression Results for Child Mortality Rate Outcome

article image
Source: IMF staff estimates.Note: FE = fixed effects; 2SLS = two-stage least squares.

Instruments are a share of agriculture in GDP and agriculture as a share of GDP. For 2SLS regressions public health spending is used at time t. Bolded coefficients are significant at least at 5 percent level.

We also find that social spending matters for both lowering poverty and boosting the IHDI. Both the aggregate measure of social spending and its health and education subcomponents come out as statistically significant and with the right sign, even after controlling for other macroeconomic and institutional variables and country heterogeneity. The quality of institutions, proxied by an index of government effectiveness, is found to help reduce poverty (Annex 3, Annex Table 6).

We also evaluate the relative importance of social spending components on aggregate socioeconomic indicators. Most of our empirical work considers health and education outcomes separately—public spending on education was one of the explanatory variables for education outcomes, while health spending (private and public) were explanatory variables for health outcomes.

Such analysis cannot reveal which type of social spending (health, education, social protection) is relatively more important. To do that, we looked at aggregate measures of well-being such as the HDI. We find that public spending on social protection has the largest, distinct, and most statistically significant impact on HDI relative to either health or education spending. One possible explanation for this result is that, perhaps, social protection schemes have the most immediate effect on lifting people out of poverty, while health and education spending take more time to bear fruit. When it comes to reducing poverty, spending on education seems to matter more than spending on health care (larger and more statistically significant estimated coefficient) (Annex 3, Annex Table 8).

We find that the quality of institutions matters for translating social spending into socioeconomic outcomes and for reducing poverty rates. For example, using the most conservative estimated coefficient in the SGMM, the results suggest that a 10 percent higher PPP dollars per capita social protection spending (if sustained for 10 years) can close 20–40 percent of the HDI gap between MENAP countries and their comparators and up to 65 percent of the gap between CCA countries and EM peer average (Figure 16). The analysis also reveals that governance matters for the impact of additional social spending on outcomes. For example, if MENAP countries can boost their survey-based governance indicators to those of their peers, the same increase in social protection spending can close 45–60 percent of the gap and completely close the gap for CCA countries and EM average.

Figure 16.
Figure 16.

Estimated Boost to HDI from Additional Social Protection Spending and Improved Governance

(2018 or latest available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: The International Country Risk Guide; United Nations Development Programme; World Development Indicators; and IMF staff calculations and estimates.Note: HDI = Human Development Index.

These conclusions are supported by case studies of countries from the region. Drawing country-specific policy advice from any econometric model has its limitations, as the exercise was conducted on a global sample, and, as noted in Annex 1, data coverage and quality can differ across countries. As shown in Chapter 6, the quality of social spending matters for translating limited fiscal resources into higher socioeconomic outcomes. To bridge the gap between empirical analysis and country specific developments, Annex 6 outlines the experience of Bahrain, Armenia, and Tunisia. This allows for a more granular analysis and policy recommendations.

  • Bahrain. Bahrain’s experience demonstrates how social spending has helped support inclusive growth and human development. Education outcomes, as reflected by net primary and secondary enrollments and literacy rate, put Bahrain at par with AE averages, while life expectancy increased by 4.4 years over the last 25 years to 77 years and infant mortality rate is getting closer to the AE average. In addition to better education and health outcomes, social spending in Bahrain contributed to a significant reduction in income and gender inequality and placed the country among the “Very High Human Development” group. However, social spending in Bahrain would benefit from further improvement in efficiency. For example, lowering the high teacher–student ratio8 would free resources to equip teaching and non-teaching staff with better educational materials, training, and other forms of professional support. The health sector would also benefit from enhanced competition between public and private hospitals.

  • Armenia. Armenia’s social protection programs helped reduce poverty rates by 30 percentage points from 2004 to 2018 and promote greater equality, as the Gini coefficient dropped from 37.5 to 34.4 in 2018. The targeting of social protection works well through a well-designed system of identification and selection of beneficiaries. However, only a small segment of the poor is reached by the program. This suggests that if Armenia had a greater budgetary allocation for social protection, such programs could more effectively help the poor.

  • Tunisia. Likewise, past social spending has helped improve socioeconomic outcomes in Tunisia. Over the past three decades, Tunisia’s HDI increased by 30 percent, putting the country in the high human-development category and the upper half of countries globally. By 2018, the expected years of schooling rose beyond 15 years, secondary-school enrollment reached more than 90 percent, and life expectancy climbed to almost 76 years. However, the level and performance of social spending remain a critical issue. Educational programs fail to address the growing skills mismatch with private sector requirements, PISA scores are still weak, and a rising share of the education spending goes to the payroll, leaving very little room for investment in latest technologies, training, and curriculum. In health, regional disparities persist in terms of access, headcount deployment, and management, while spending inefficiencies stem from a high and rigid wage spending, a subsidy system for pharmaceutical products, and not enough emphasis for preventive care. The social security system lacks adequate coverage, and social assistance programs remain fragmented, fail to cover a significant part of the low-income population and informal sector employees, and disproportionately benefit the better-off in the urban areas (Annex 6, Box 2). Policy priorities therefore are (1) more and better-targeted social spending, (2) a financially viable social security system, and (3) institutional and governance reforms to improve spending quality.

Chapter 6 Increasing the Efficiency of Social Spending in the Region

Most countries may not be able to permanently sustain higher levels of social spending without efforts to create fiscal space (Figures 17 and 18; Box 1). Public debt in MENAP EMs and LICs was high even before the pandemic and is now projected at close to 90 percent of GDP on average in 2020 (IMF 2020). Hence, sustaining the higher levels of social spending made necessary by the pandemic without further increasing the public debt burden will require efforts at reprioritizing current spending and/or mobilizing additional revenues (IMF 2018). Several governments in the region will also likely need to focus on increasing spending efficiency (Annex 4). Specific recommendations, however, will depend on individual country circumstances (see Annex 6).

Figure 17.
Figure 17.

Fiscal Space and COVID-19 Fiscal Cost, 2019–20

(Percent of GDP)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, World Economic Outlook; IMF, COVID-19 Country Surveys; national authorities; and IMF staff calculations.
Figure 18.
Figure 18.

Fiscal Balance, Debt, and Estimated Cost of COVID-19 Response

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, COVID-19 Country Surveys; IMF, World Economic Outlook; national authorities; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes.1 Response to the survey question “please specify above-the-line or on-budget measures in response to COVID-19 directly affecting the government budget balance or financing needs in gross terms: Additional spending or foregone revenue: Total estimated fiscal cost (and of which estimated cost in the health sector).”

In fact, the efficiency of social spending has generally been low in the region. The efficiency of social spending can be measured by a variety of techniques, both parametric and nonparametric. Nonparametric techniques, such as Data Envelopment Analysis (DEA) and Free Disposal Hull (FDH), simply plot countries according to their spending and their outcomes and draw an “upper envelope,” or frontier, that gives the best outcome that countries can achieve for any level of spending. Figure 19 shows an example of health and education spending efficiency frontiers using the nonparametric approach. The vertical distance by which countries fall short of this frontier is then taken as a measure of their inefficiency (strictly speaking, “output inefficiency”).1 This is a data-driven exercise, requiring no assumptions. On the other hand, “parametric” approaches—such as Stochastic Frontier Analysis (SFA)—enable a distinction between inefficiency and statistical noise, but require the imposition of a functional form on the input-output relationship. Both parametric and nonparametric approaches have their advantages and disadvantages and have been widely used in the literature.2 (Annex 5 for a technical discussion of SFA.)

Nonparametric approaches confirm that spending efficiency can be improved in the region. We report FDH estimates of public education and health efficiency scores from the international benchmarking study by Herrera and Ouedraogo (2018). In education spending, MENAP-EMs achieve less “bang for their buck” than their global EM peers; the same can be said for MENAP-LICs relative to the global LIC average and GCC countries relative to AEs (Figure 20). If spending efficiency in Mauritania, for example, were increased to the global frontier, the average years of schooling could double.3

Figure 19.
Figure 19.

Efficiency Frontiers in Nonparametric Approach

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, FAD Expenditure Assessment Tool (EAT); World Bank; and World Health Organization.1 Healthy life expectancy (HALE) is a measure of health expectancy that applies disability weights to health states to compute the equivalent number of years of life expected to be lived in full health.
Figure 20.
Figure 20.

Output Efficiency Scores from Nonparametric Approach1

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: Herrera and Ouedraogo (2018); and IMF staff calculations.1 Efficiency scores range from 0 to 100, where 100 represents the most efficient level.

The SFA approach also confirms this assessment (Figure 21). Regional LICs on average are somewhat less efficient than their global income peers in both health and education spending. The GCC economies appear to be somewhat less efficient than global AEs, and CCA countries are less efficient than EM peers. Efficiency in education spending for EMs in the region is broadly in line with global peers but exceeds global peers in health spending efficiency.

Figure 21.
Figure 21.

Spending Efficiency—Parametric and Nonparametric Approaches1

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: Herrera and Ouedraogo (2018); and IMF staff calculations.Note: DEA = data envelopment analysis; FDH = free disposal hull; SFA = stochastic frontier analysis.1 Efficiency scores range from 0 to 100, where 100 represents the most efficient level.

Therefore, even without increasing outlays, boosting the efficiency of spending would help significantly improve socioeconomic outcomes (Figure 22).4 For example, life expectancy at birth could increase by three years in Kuwait if the existing resource envelope was spent at the efficiency level of advanced economies. Under higher social spending efficiency, Afghanistan’s HDI could reach the level of the Kyrgyz Republic, and infant mortality in Iraq would drop from 31 per 1,000 live births to 27 per 1,000 live births. If the region could bring its social spending efficiency to the average efficiency level of advanced economies, even without any extra spending, it could close 34 percent of the HDI outcome gap, and 20 and 10 percent of the outcome gaps in secondary school enrollment and life expectancy, respectively. Of course, these are model-driven conclusions, and in reality, achieving such a boost in spending efficiency cannot happen overnight. The next section shows that better institutional quality and financial inclusion are associated with higher spending efficiency. Individual country experiences also show that better targeted social protection programs, good coverage of social safety nets, avoiding duplication of programs, and monitoring outcomes can help improve the efficiency of spending. Some countries in the region (for example, Iraq and Saudi Arabia) are making efforts on this front, but further improvement is still needed.5 Most countries in the region are working on improving financial inclusion.

Figure 22.
Figure 22.

Socioeconomic Outcomes under Higher Efficiency

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: UN Development Programme, World Development Indicators; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes. HDI = Human Development Index.

There is also scope to increase the efficiency of public health care spending in the region (Figure 23). This is especially the case in Kazakhstan, Kuwait, and Pakistan. Health care spending in Azerbaijan is particularly inefficient at attaining a lower infant mortality rate, but is doing a better job at improving life expectancy at birth. Afghanistan is able to achieve better-than-expected health outcomes for its level of spending. Increased investment in primary health care to enable early diagnosis and prevention of chronic illnesses is a more efficient way to spend fiscal resources, especially when compared to costly subsidies for medical treatment abroad. Investing in the human resources for primary health care will also offer a gender dividend, given that many primary health care workers in the region tend to be female.

Figure 23.
Figure 23.

Health Care Spending Adequacy and Efficiency

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Development Indicators; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes. ppts = percentage points.

In response to the COVID-19 pandemic, many countries have increased public outlays on health care and social protection.6 Much of this increase is temporary and will be rolled back once the health crisis abates, but an interesting thought experiment is to consider the impact on health indicators if the region sustainably spent at the current levels. Our SFA model suggests that life expectancy would increase substantially in certain countries, with the increases even larger if countries were able to raise their health spending efficiency at the same time (Figure 24). This is not to say that current levels of spending should be sustained—as discussed in Box 1, this depends on the available fiscal space and competing policy priorities—but again, additional spending can have a powerful effect on outcomes.

Figure 24.
Figure 24.

Increase in Life Expectancy if COVID-19 Health Spending Is Incorporated into Permanent Government Expenditure

(Years)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Development Indicators; and IMF staff estimates.Note: Data labels use International Organization for Standardization (ISO) country codes.

Although education spending in the Middle East and Central Asia is relatively higher and more efficient than health care spending, there is still scope for efficiency gains (Figure 25). Our estimates suggest that Kuwait, the Kyrgyz Republic, Saudi Arabia, and Uzbekistan, are spending in line with global peers on education, and should focus primarily on boosting efficiency. Lebanon, Mauritania, Pakistan, and Tajikistan are not spending enough and could see their gross school enrollment and expected years of schooling rise with larger education budgets. In countries that are spending both inefficiently and not enough (Mauritania, Pakistan, Qatar), priority should be given to addressing spending efficiency before increasing its level to maximize impact per dollar spent. As countries that are efficient but are not spending enough increase their level of spending, they may witness a decrease in their efficiency scores due to diminishing marginal returns. Indeed we find evidence of decreasing returns to scale. Therefore, any additional spending needs to be calibrated in a way that preserves efficiency and achieves better outcomes per dollar spent. Promoting high-quality education, starting in the early years and setting the right teaching policies, provides an opportunity for efficiency savings over time. The World Bank Human Capital Project demonstrates the benefit of investing in the early years to improve human capital.

The COVID-19 Crisis in the Middle East and Central Asia

The COVID-19 pandemic is still unfolding but appears to be disproportionately impacting the most vulnerable groups and threatening development achievements of recent decades. The unprecedented public health emergency and the associated lock-down measures have resulted in job losses and interrupted access to health and education services. According to World Bank estimates, the crisis has pushed 10 million households in the MENA region into poverty, of which 3 million were pushed into extreme poverty (Gerszon and others 2020).

The crisis has had a disproportionate impact on women. Job losses were predominately concentrated in the service sector, which tends to employ more women. More women also work in the informal sector, complicating their ability to claim unemployment benefits or access social protection schemes. The disproportionate burden of childcare and eldercare on women even in normal times has been further magnified by the pandemic, as schools closed, and family members got ill. More females are on the front lines of fighting the pandemic as well, as 69 percent of health professionals are female (Figures 1.1 and 1.2; Grown and Sánchez-Páramo 2020).

Figure 1.1.
Figure 1.1.

Female-to-Male Ratio of Time Devoted to Unpaid Domestic, Volunteer, and Care Work, 2014

(24-hour period)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: OECD Gender, Institutions, and Development Database (2014); and IMF staff calculations.Note: Data for GCC are not available.
Figure 1.2.
Figure 1.2.

Female and Male Employment on the Frontlines

(Percent, globe averages, 2018 or latest available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: International Labour Organization Statistics; and IMF staff calculations.

Governments in the region have announced sizeable increases in health and social protection spending aimed at mitigating the impact of COVID-19. As of June 2020, fiscal measures averaged 2.6 percent of GDP, including 0.6 percent of GDP in the health sector. While country heterogeneity is significant, the size of the fiscal response is generally lower than in global peers, perhaps reflecting limited fiscal space and, in the case of health spending, lower rates of infection due to successful containment measures. However, Iran’s additional health spending to combat the crisis, at 2.2 percent of GDP, is among the highest in the world, reflecting the high infection rate. Moreover, some countries in the region have extended social protection to previously uncovered groups during the crisis (ILO 2020), both through social insurance and tax-financed benefits, but additional efforts would be needed to sustain these measures and to transform emergency measures into sustainable elements of the national social protection system.

To prevent a deterioration in socioeconomic indicators, governments’ COVID-19 responses should proactively target vulnerable groups, including women, informal sector workers, and refugees. World Bank simulations of the pandemic suggest that children on average will lose 0.6 years of schooling, adjusted for quality, risking a deterioration in education outcomes and lifelong earnings. Given the uneven impact of the crisis on women, there is a high risk that gender inequality will widen, and progress achieved over the past two decades will be reversed. Encouragingly, Egypt, Mauritania, and Pakistan targeted financial support to vulnerable women through broader social assistance schemes, while Algeria gave priority for exceptional leave to pregnant women and women raising children (Gentilini and others 2020). There has also been insufficient focus on refugees in the pandemic response initiatives in most countries (International Rescue Committee 2020). Refugees in many Middle East and Central Asian countries are not eligible for national social assistance schemes, which often require documentation. While multilateral institutions such as the World Bank, United Nations, and EU have been trying to mobilize funding to fill in some gaps, special provisions for targeted support, both nationally and internationally, are necessary (Figure 1.3).

Figure 1.3.
Figure 1.3.

Fiscal Cost of COVID-19, 2020

(US dollars per capita)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: IMF, COVID-19 Country Surveys; national authorities; and IMF staff calculations.1 Fiscal cost in the health sector is available only for Bahrain.
Figure 25.
Figure 25.

Education Spending Level and Efficiency

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: World Development Indicators; and IMF staff calculations.Note: Data labels use International Organization for Standardization (ISO) country codes. ppts = percentage points.

Chapter 7 Drivers of Efficiency: Institutions and Governance

Institutional quality may be behind the region’s relatively low spending efficiency (Figure 26). Unpacking the drivers of efficiency is critical to offering practical advice to policymakers. There is a strong correlation between spending efficiency and indicators of institutional quality, such as government effectiveness, the control of corruption, and the rule of law. Stronger transparency and accountability over the use of public resources minimizes wasteful spending and promotes efficiency. This is consistent with findings in the literature as well (Rajkumar and Swaroop 2008; Albino-War and others 2014; IMF 2018).

Figure 26.
Figure 26.

Efficiency and Institutional Quality

(2018 or latest value available)

Citation: Departmental Papers 2020, 012; 10.5089/9781513553115.087.A001

Sources: Herrera and Ouedraogo (2018); World Governance Indicators; and IMF staff calculations.Note: DEA = data envelopment analysis.

A more formal analysis confirms the finding that improving institutional quality is key to improving spending efficiency (Table 4). Using efficiency scores for public health and public education from non-parametric techniques, we estimate the following equation using the Tobit technique, as the dependent variable is censored (ranges between 0 and 1).

Table 4.

Drivers of Efficiency

(Output Efficiency Score)

article image
Source: IMF staff estimates.Note: Bolded coefficients are significant at least the 5 percent level.

Ei α + B1(Inst)i + B2(Z)i + μr + εi

in which Ei refers to efficiency scores for country i. We estimate five models in which efficiency scores of two health-related outcomes (infant survival rate, life expectancy) and three education related outcomes (expected years of schooling, secondary-school enrollment, quality of math and science index) are used as dependent variables. These efficiency scores come from Herrera and Ouedraogo (2018) and are an average of the 2009–15 time period. The term Inst refers to institutional quality, proxied, in different specifications, by a range of World Bank indicators of government effectiveness, the control of corruption, the rule of law, and the strength of democracy.1 Z refers to a vector of control variables that are averaged at the 2009–15 time period; mrdenotes regional dummies; and represents the error term that follows a normal cumulative distribution function.

The results suggest scope for improving efficiency in the region. The MENAP region in particular is less efficient than the global EM average in its social spending, and this is consistent across many different social outcomes. Efficiency could be improved by strengthening governance, which is discussed at greater length in the upcoming IMF paper on governance in the Middle East and Central Asia (IMF, forthcoming). Spending efficiency is also positively correlated with the level of urbanization, given economies of scale. Addressing infectious diseases in some countries would also boost the productivity of the workforce and improve efficiency. Higher levels of inclusion and financial deepening are also associated with improved spending efficiency. Access to banking services allows households to both save for a rainy day and to borrow in emergencies. This ability to smooth consumption prevents deterioration in socioeconomic outcomes, as households as less likely to withdraw children from school, forego medical care, or cut down on nutritious intake. Financial deepening also helps facilitate delivery of social transfers and reduce opportunities for corruption thereby helping achieve bigger bang for the buck (Annex 3, Annex Table 7).

Standard metrics of either socioeconomic outcomes or efficiency may not fully capture the impact of social spending on the very poor. Aggregate measures of secondary-school enrollment and life expectancy do not reveal whether the improvements are observed across the population or instead disproportionately benefit relatively richer segments of the population. Previous work has shown that many socioeconomic programs in the region do not sufficiently benefit the poor, youth, women, refugees, and the rural population (Purfield and others 2018). Spending tends to be more efficient where it is more equally distributed and focused on achieving universal access.

Case studies allow us to explore in more detail why social spending efficiency is low in the region (see Annex 6). For example, schools and vocational training in Bahrain and Tunisia often do not adequately address the growing skills mismatch between what they teach and what is needed by the private sector. In Tunisia health care systems leave little room for preventive care; the administrative costs are relatively high, while a subsidy system for pharmaceutical products encourages overprescription. Existing programs suffer from fragmentation and insufficient coverage of low-income and informal sector employees, and they disproportionately benefit the better-off in urban areas. Some of these inefficiencies are being addressed in the ongoing social protection reforms.

The COVID-19 crisis has prompted many governments to come up with innovative and efficient solutions in administering social protection. Jordan has used mobile wallets to transmit transfers to recipients. Kazakhstan has allowed customers to open bank accounts via a mobile app, which could then be used to receive government cash transfers and make purchases. Morocco has been able to reach informal workers via SMS message to administer modest cash transfers. These new technologies have helped countries in the region improve the efficiency with which social assistance is delivered to its intended recipients.

Chapter 8 Takeaways and Policy Implications

This paper highlights the importance of increasing both the size and efficiency of social spending to achieve more inclusive growth in the Middle East and Central Asia. Although socioeconomic outcomes are determined by a number of factors, we find—using a variety of econometric techniques and a global sample spanning nearly 20 years of data—that public social spending can have an appreciable impact on outcomes. At the same time, we document that countries in the region generally lag their global income peers in socioeconomic outcomes. This reflects many factors, including the high incidence of conflict and fragility in the region, but also lower levels of public spending on health, education, and social protection, as well as a relative inefficiency of spending compared to global peers.

The current crisis has further underscored the importance of social spending and demonstrated the region’s ability to quickly mobilize additional outlays on health and social protection. The COVID-19 pandemic has brought to the fore the need for robust health care systems and frameworks for channeling targeted financial support to the vulnerable. Most countries in the region are expected to temporarily boost spending on health and social protection in 2020 to deal with the unfolding pandemic. They have also demonstrated ingenuity at delivering social protection through digital solutions.

Prioritization of social spending will need to continue post-COVID. While some COVID-related spending will likely be scaled back once the crisis abates, the need for adequate social spending more generally remains. Our estimates suggest that increases in social spending would result in sizeable improvements in outcomes. Sustaining—and potentially increasing— education spending is also important to mitigate the impact of the crisis on learning outcomes, especially for children most at risk of being left behind.

Efforts to create fiscal space for social spending should therefore continue. Given the region’s gaps with peers in socioeconomic outcomes, there is a need in many countries to create more fiscal space—including through budget reprioritization and enhanced revenue mobilization—to permit increased allocations for social spending while ensuring fiscal sustainability. Before the current crisis, many countries in the region had already started to take measures to create fiscal space for social spending, including by undertaking fiscal reforms together with strengthening targeted outlays on social safety nets (Armenia, Egypt, Tunisia, Jordan, Pakistan, Oman, Saudi Arabia), mobilizing and diversifying revenues (Bahrain, Saudi Arabia, United Arab Emirates), strengthening tax administration, and rationalizing tax exemptions (Djibouti, Morocco). These efforts will need to continue following the crisis.

Greater efforts are needed to boost social spending efficiency. Given the competing priorities for limited public resources, social spending should be used efficiently and targeted appropriately. This includes both countries that are able to generate fiscal space and countries that face a fixed spending envelope so that each dollar spent has a larger impact on socioeconomic outcomes. Our analysis suggests that efficiency can be raised by strengthening institutions, improving governance, and controlling corruption. Greater spending efficiency could deliver better inclusive growth outcomes even at the same spending levels. Innovative approaches adopted by governments during the COVID-19 crisis in administering social protection benefits should continue to fully capitalize on the benefits that digital solutions can offer in terms of spending efficiency and inclusion. Efforts to promote financial deepening and inclusion would also help strengthen spending efficiency, including by helping households withstand crises, simplifying payment delivery, and reducing opportunities for corruption.

Improving outcomes would also require identifying existing gaps that impede access to social services. This includes gender gaps that hinder access to education and health care and institutional factors that keep vulnerable groups outside the reach of formal social safety nets. It would also call for increased investments in primary health care, as early diagnosis and prevention of chronic illnesses is the least costly and most efficient way to improve health outcomes.

Finally, measurement issues discussed in this paper suggest a need for better data on non-public spending with a social component and socioeconomic outcomes. The traditional definition of social spending used in this paper—necessitated by data availability and allowing for better cross-country comparisons—may understate the amount of social spending individual countries engage in. While measurement issues are unlikely to bias comparisons between countries in the Middle East and Central Asia region and their global peers, more comprehensive data on social spending outside of the public sector and more comprehensive and timely data on socioeconomic outcomes will allow evidence-based approaches for richer and more tailored policy advice.

Social Spending for Inclusive Growth in the Middle East and Central Asia
Author: Mr. Koshy Mathai, Mr. Christoph Duenwald, Ms. Anastasia Guscina, Rayah Al-Farah, Mr. Hatim Bukhari, Mr. Atif Chaudry, Moataz El-Said, Fozan Fareed, Mrs. Kerstin Gerling, Nghia-Piotr Le, Mr. Franto Ricka, Mr. Cesar Serra, Tetyana Sydorenko, Mr. Sébastien Walker, and Mr. Mohammed Zaher
  • View in gallery

    Definition of Public Social Spending

  • View in gallery

    Wage Bill, Subsidies, and Social Spending

    (Percent of GDP, 2018 or latest available)

  • View in gallery

    Selected Budgetary Spending with Social Aspects

    (Percent of GDP)

  • View in gallery

    Public Social Spending

  • View in gallery

    Additional Spending Needs in 2030 to Meet Selected SDGs

    (Percentage points of GDP)

  • View in gallery

    Public and Private Health Spending

  • View in gallery

    Public Education Spending

  • View in gallery

    Social Protection Spending

  • View in gallery

    Fiscal Cost of COVID-19, 2020

    (Percent of GDP)

  • View in gallery

    Education Outcomes

  • View in gallery

    Improvement in Socioeconomic Indicators

  • View in gallery

    Fertility Rate vs. Female Secondary Enrollment1

    (Averages from 1980 to 2018)

  • View in gallery

    Socioeconomic Outcomes in Middle East and Central Asia Countries and Relevant Global Peers

    (2018 or latest available value)

  • View in gallery

    Socioeconomic Indicators

    (2017 or latest available value)

  • View in gallery

    Public Social Spending and Socioeconomic Outcomes

    (1990–2017)

  • View in gallery

    Estimated Boost to HDI from Additional Social Protection Spending and Improved Governance

    (2018 or latest available)

  • View in gallery

    Fiscal Space and COVID-19 Fiscal Cost, 2019–20

    (Percent of GDP)

  • View in gallery

    Fiscal Balance, Debt, and Estimated Cost of COVID-19 Response

  • View in gallery

    Efficiency Frontiers in Nonparametric Approach

  • View in gallery

    Output Efficiency Scores from Nonparametric Approach1

    (2018 or latest value available)

  • View in gallery

    Spending Efficiency—Parametric and Nonparametric Approaches1

    (2018 or latest value available)

  • View in gallery

    Socioeconomic Outcomes under Higher Efficiency

    (2018 or latest value available)

  • View in gallery

    Health Care Spending Adequacy and Efficiency

    (2018 or latest value available)

  • View in gallery

    Increase in Life Expectancy if COVID-19 Health Spending Is Incorporated into Permanent Government Expenditure

    (Years)

  • View in gallery

    Female-to-Male Ratio of Time Devoted to Unpaid Domestic, Volunteer, and Care Work, 2014

    (24-hour period)

  • View in gallery

    Female and Male Employment on the Frontlines

    (Percent, globe averages, 2018 or latest available)

  • View in gallery

    Fiscal Cost of COVID-19, 2020

    (US dollars per capita)

  • View in gallery

    Education Spending Level and Efficiency

    (2018 or latest value available)

  • View in gallery

    Efficiency and Institutional Quality

    (2018 or latest value available)