A USER’S GUIDE TO THE QUALITY OF EXPENDITURE HEATMAP (QEH) Santiago Herrera and Hironobu Isaka1 Abstract: The Quality of Expenditure Heatmap (QEH) is a tool that provides information to help assess the quality of public expenditure, comparing expenditure outcomes across countries or groups of countries. Using data from numerous sources (e.g., World Bank, IMF, OECD, CEQ, ND-GAIN) the tool provides raw data and transformations to facilitate comparisons across countries and over time on multiple dimensions. It is a user-friendly tool as it relies on an Excel-based template populated with data on different indicators, clustered around the three central objectives of fiscal policy: allocation, stabilization, and redistribution. The QEH scores the quality in the three pillars and ranks the countries accordingly. The QEH is not a substitute for in-depth examination of public spending at the project or sectoral level; it is intended to provide a starting point for the analysis by shedding light on where the major gaps of public expenditure exist. Key words: Allocation, Stabilization, Redistribution, Objectives of Fiscal Policy JEL codes: E63, H11, H5 1.The authors acknowledge comments by Emilia Skrok, Chadi Bou-Habib, Sanja Madzarevic-Sujster, and Eduardo Olaberria to a previous version. This paper was funded by the KDI School of Public Policy and Management. The findings, interpretations and conclusions expressed in this paper are entirely those of the authors and should not be attributed to the World Bank, its Executive Directors, or the countries they represent. Santiago Herrera Email: sherrera@worldbank.org. Hironobu Isaka Email: hisaka@worldbank.org. 2 Contents I. Introduction ............................................................................................................................................... 3 II. Overview of the Quality of Expenditure Heatmap (QEH) ......................................................................... 3 III. Description of individual indicators in each pillar. ................................................................................... 5 A. Long -Term Growth ............................................................................................................................... 5 1. Sustainability ..................................................................................................................................... 5 2. Technical and Allocative Efficiency ................................................................................................... 9 B. Stabilization......................................................................................................................................... 17 1. Volatility of spending .................................................................................................................. 17 2. Rigidity of spending. ........................................................................................................................ 20 3. Procyclicality of spending and fiscal impulse .................................................................................. 21 C. Redistribution ..................................................................................................................................... 22 1. Income Inequality ........................................................................................................................... 23 2. Gender Inequality ........................................................................................................................... 23 3. Intergenerational Inequality ........................................................................................................... 23 IV. Aggregation of indicators ...................................................................................................................... 24 A. Aggregation Methodology .................................................................................................................. 24 B. Aggregate indicators of quality of expenditure .................................................................................. 25 V. Visualization tool and summary of information: Three case studies ..................................................... 30 A. Chile .................................................................................................................................................... 31 B. Cameroon ........................................................................................................................................... 33 C. Poland ................................................................................................................................................. 34 References .................................................................................................................................................. 36 Appendix A. ................................................................................................................................................. 40 Appendix B ................................................................................................................................................ 50 Appendix C. ................................................................................................................................................. 52 3 I. Introduction Public debt has risen since the global pandemic and the war in Ukraine; this has tightened fiscal space as pressure for additional public spending from various sources mounts: climate change, food and energy inflation which is pushing up demand for subsidies, a higher debt burden, and an ageing population. In this context of resource scarcity, the quality of public spending needs to be excellent. Two features that describe good spending are: 1) it meets its objective; and 2) it does so at the lowest possible social and economic cost. Public spending is justified if the marginal benefits of meeting the objectives exceeds the marginal social cost of attaining the objective. Public spending, and fiscal policy in general, have multiple objectives: ensuring an efficient allocation of resources, stabilizing the economy, or redistributing resources across diverse groups of individuals.2 Fiscal policy also includes taxation and public debt management, the focus of this guide is public spending. To simplify matters, this Quality of Expenditure Heatmap (QEH) uses the three Musgrave objectives as an organizing principle. Governments use multiple tools to achieve the three objectives, which are considered complementary ˗ in the long term and in theory; The objectives may clash with each other over the short term in the presence of hard budget constraints. The Quality of Expenditure Heatmap (QEH),is a tool that provides information to help assess the quality of public expenditure, comparing different benchmarks within a country or examining its evolution over time. Using data from numerous sources (e.g., World Bank, IMF, OECD, CEQ, ND-GAIN) the tool provides raw data and transformations to facilitate comparisons across countries and over time on multiple dimensions. It is a user-friendly tool as it relies on an Excel-based template populated with data on different indicators, clustered around the three central objectives of fiscal policy. The QEH is not a substitute for in-depth examination of public spending at the project or sectoral level; it is intended to provide a starting point for the analysis. II. Overview of the Quality of Expenditure Heatmap (QEH) The analysis starts by delineating the objectives of fiscal policy following Musgrave's “problems� of fiscal policy and defining setting the pillars of this heatmap based on them: (i) allocating resources to support long-term GDP growth; (ii) stabilizing employment and output remain at their potential; (iii) redistributing resources equitably across distinct groups of individuals and generations. Efficient expenditure that is supportive of long-term growth must be sustainable, and efficient from technical and allocative perspectives. The source (e.g., taxes or debt) and the amount of funding need to be clearly identified. Policymakers use public spending to maintain output and employment at their potential levels; in this context, spending must be countercyclical, and stabilization faces many hurdles. Time inconsistency is one such hurdle: a plan may be optimal at the point of design, but, in the absence of commitment mechanisms, policymakers deviate from it at the time of execution, making it incredible. Good budget institutions, such as a fiscal rule, may be an adequate commitment mechanism that makes it costly for policymakers to deviate from the original plan. 2 Musgrave (1959) classified fiscal policy into these three “branches�, and, since then, the classification has been used in Public Finance textbooks and by numerous practitioners as a useful apparatus to organize ideas. 4 Musgrave’s third objective involves principles of equity and fairness and is more subject to political debate. If there is a rationale for public intervention and the program is the most cost-effective method of achieving the objective, it should be fine. But as society is composed of multiple groups of individuals, the demands for redistribution grow exponentially and prioritizing projects becomes a major problem. In theory, these three objectives—supporting long-term growth, stabilizing the economy around the potential levels of output and employment, and ensuring equity—are all complementary in the long-run, and governments typically strive to achieve all three objectives. However, in practice, numerous complications hinder straightforward policymaking, one of which involves balancing short-term or long- term needs. For example, during the pandemic, short-term equity and health objectives were prioritized in most countries while long-term considerations related to funding sources were relegated and governments issued debt or used central bank financing. After the pandemic, countries scrambled to consolidate and rearrange priorities to reduce the cost of debt. Strategies may also change with external circumstances. For instance, issuing debt when the Fed is in an expansionary phase as in 2020-21 is different from issuing debt when the Fed is mopping up liquidity as it did in 2022-23. Policymakers strive to achieve these three objectives subject to budget constraints. This report presents different indicators of outcomes related to the three objectives and benchmarks individual country results with a large set of countries, which can be clustered by region, income group, or natural-resource wealth. The tool also constructs a heatmap to help visualize the information. Figure 1 summarizes the contents of the Quality of Expenditure Heatmap (QEH).3 Figure 1. The three pillars to assess the quality of public expenditure are structured around the objectives of fiscal policy. 3Some analysts may argue that public services, like national security or street lighting, do not fall under these three categories. The clustering used here is a simplification, and alternative clustering arrangements could be used. However, following the Musgrave approach, national security could be classified under the resource allocation or long-term growth pillar because of its impact on investment, which shifts resources into the future, and the future is more certain in a secure environment. 5 III. Description of individual indicators in each pillar. A. Long -Term Growth (blue tabs in excel sheet) To assess the role of spending on achieving its long-term growth objective, the heatmap clusters multiple indicators into three main sub-components: sustainability, technical efficiency, and allocative efficiency. The three components are interrelated: spending that is not efficient is not sustainable and unsustainable spending is not efficient. Each sub-component comprises numerous indicators. Figure 2 summarizes the contents of this section of the heatmap. Figure 2. The indicators under the pillar of long-term growth Note: the number in parentheses shows the number of indicators in the Excel sheet. 1. Sustainability Sustainability of spending is associated with multiple variables, the most prominent being the size of government spending and the sources of financing. Funding is a strong determinant of good (or bad) spending (Bird, 2005). Additional variables may have information on the sustainability (Kose et. al.2018), which include: the debt level (% of GDP) and its composition, the primary balance, the difference between the observed primary balance and the debt-stabilizing primary balance, and the debt to tax ratio. The interest coverage ratio, measured as the tax revenue ratio to interest payments is also reported, as a recent IMF paper found it to be a good predictor of fiscal stress episodes (Comelli, et al. 2023). Below follows a brief discussion of the variables included in this section of the heatmap, the full description and source of all indicators can be found in Appendix 1. a) Size of spending When comparing the size of government across countries or over time, it is important to control for the country's level of development, as Wagner’s Law shows that government spending increases with GDP because of higher prices and wages in richer countries; richer countries also have older populations. 6 Hence, certain components of spending, such as the wage bill, pensions payments, and health spending, increase with the level of GDP (Figure 3, Figure 4, and 5). The heatmap presents the data in levels as well as the component of spending that is independent of the GDP level, which is calculated as the residual of a regression between public spending and GDP. Figure 3. Public wages per capita and GDP per capita Source: WEO and authors’ calculations Figure 4. Public pension payments per capita and GDP per-capita Source: WEO and authors’ calculations 7 Figure 5. Public health spending per capita and GDP per-capita Source: WEO, WDI, and authors’ calculations b) Debt-stabilizing primary balance. The debt stabilizing primary balance (DSPB) is the level of the primary balance that is compatible with a stable or a target debt level over time, i.e., the level of the constant primary balance that will bring the debt ratio to a desired target level in a pre-specified time horizon. A target debt level, which can be assumed to be the average of historical values, is assumed for calculating the DSPB; other inputs in the calculation are the duration of the horizon to achieve the target, and the spread between the real interest rate and the GDP growth rate. The larger the gap, the less sustainable the policy. More details can be found in Escolano (2010), and Butron-Calderon (2022) who extended the model to incorporate foreign- currency debt. In the heatmap, the levels of the DSPB are sourced from the Kose, et al (2018) database. c) Revenue to interest payments. A recent IMF paper finds that the ratio of revenues to interest payments is strongly correlated to episodes of fiscal stress (Comelli, et al., 2023). The ratio expresses the number of yearly interest payments that current revenue can cover. This is a modification of the concept of the cover ratio used in corporate finance. Comelli et al. find that this indicator is a robust predictor of fiscal stress.4 d) Funding sources. When evaluating public spending, it is key to know the sources of funding (Bird, 2005) as they are related to growth, social welfare, and redistribution. Taxation affects consumption, investment, and labor supply, and the observed levels of these variables are different from those that would prevail in a competitive equilibrium without taxation. These departures from the competitive equilibrium are distortions induced by taxation and necessarily imply welfare costs. Raising one dollar of tax revenue induces these distortions 4 Fiscal stress periods are defined when one or more of the 4 events are present: credit events (default, restructuring); loss of market confidence (loss of access to market or spike in spreads); recourse to large scale IMF finance; implicit domestic default via high inflation or arrears. 8 which must be included when computing the marginal cost of public funds (MCPF). Taxation is also a redistribution of resources from the private sector to the government. The heatmap reports several indicators related to funding sources: 1) tax collection over time (as a percentage of GDP); 2) the “tax gap�, which is the difference between the observed tax collection and the expected collection for a country of a given GDP level; 3) the change in tax revenue as a percentage of GDP over a user-defined period, to capture the rise or fall of the tax burden; and 4) an efficiency measure of tax collection, the VAT C-efficiency. The C-efficiency parameter is the ratio of actual VAT collection to a notional amount derived from a perfect enforcement of a uniform tax rate to all consumption, while actual collection depends on the nominal tax rate, exemptions, and administrative efficiency. A ratio of 1 indicates that all consumption is taxed at the nominal rate and there is no administrative inefficiency. The level of a country’s VAT C-efficiency is highly correlated with quality of governance indicators from the World Bank’s Worldwide Governance Indicators (Figure 6). Figure 6. VAT C-Efficiency and Governance The volatility of revenues (general government revenue, tax revenue, grants) is also reported in the QEH, as this variable is associated with pro-cyclicality of spending, as discussed in the stabilization section below (Talvi and Vegh, 2005; Herrera, Kouame, Mandon, 2018). Spending is also financed by issuing debt (deficit financing) with the burden shifted to future generations. Present generations pay only the cost of the interest, but the budget becomes more rigid, with the associated interest rate costs. The heatmap reports the debt levels (as a percentage of GDP), as well the ratio to general government revenues and tax collections, variables that are used in the analysis of fiscal space (Kose et Al, 2018). Debt levels affect sustainability as does debt composition due to differential risk implications. Foreign currency and domestic currency issues have different risk profiles, the former exposes the budget to exchange rate risk and, if issued in international markets, to external capital flows volatility. The composition of debt by maturity is also relevant, as short-term debt exposes the budget to refinancing 9 and repricing risks. Riskier debt structures may be associated with stop-go funding and associated project interruption which will reduce the effectiveness of public projects. Inflation tax and seigniorage are other financing sources, although inflation became less relevant after many central banks achieved independence from government. During the pandemic, rising inflation and the flexibility of central bank financing allowed public finances to adjust to extraordinarily high expenditure without defaulting. Inflation is not considered a desirable permanent source of financing given its detrimental impact on growth and on the budget via higher interest rates; and its regressivity. Still, the flexibility of the tool requires a quantitative approximation, for which we follow Kiguel and Neumeyer (1989): Seigniorage (S) and inflation tax are calculated as follows: (𝑀𝑡 − 𝑀𝑡−1 ) 𝑆 = 𝑃𝑡 𝑌𝑡 𝑡 𝑡 𝑀𝑡 𝑀𝑡−1 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛 𝑡𝑎𝑥 = 𝑆 − ( − ) 𝑃𝑡 𝑌𝑡 𝑃𝑡−1 𝑌𝑡 where Yt is real output at time t, Pt is the price level at t, 𝑃𝑡 𝑌𝑡 is the nominal gross national product, Mt money supply at period t. e) Perception of sustainability. The market’s assessment of a country’s credit risk informs the perception of its sustainability. In the heatmap, this is proxied by foreign currency long-term sovereign debt ratings from Kose et al. (2018) and EMBI spread from JP Morgan. 2. Technical and Allocative Efficiency Efficiency is another way of determining how supportive spending is to long-term growth. Farrell (1957) identified two ways in which productive agents could be inefficient: (i) they could use more inputs than technically required for a given level of output; or (ii) they could use a sub-optimal input combination given input prices and their marginal productivities. The first type of inefficiency is called technical inefficiency while the second one is allocative inefficiency. Technical efficiency. Technical efficiency examines the input/output relationship of a specific good or service. For a given output level, technical input-efficiency compares the distance between actual input utilization (production cost) with the cost minimizing bundle of inputs (production cost). Alternatively, for a given level of input utilization, output-efficiency is measured as the distance between the actual output (revenue) production and the maximum level that can be achieved with the given resource utilization. There are multiple ways of measuring technical efficiency of public spending. The Data Envelopment Analysis (DEA) is a non-parametric method that uses linear programming methods to construct the efficiency frontier based on actual observations of input and output combinations. The Free Disposal Hull (FDH) is also used to estimate the unobservable efficient frontier and does so without imposing convexity or smoothness assumptions, providing a piecewise linear frontier. Both DEA and FDH have been used by Gupta and Verhoeven (2001), Herrera and Pang (2004), Herrera and Ouedraogo (2018). The main 10 limitations of this analysis arise from the lack of control for error-in-measurement, the sensitivity to the sample selection, and sensitivity to outliers. To control for these limitations in DEA and FDH, the heatmap also uses a bootstrapping approach, following Simar and Wilson (1998). Bootstrapping mimics the data generation process to create replicate samples for estimating efficiency, it also corrects for biases by generating artificial datasets to estimate efficiency more accurately. The stochastic frontier analysis (SFA) is an econometric estimation of the frontier presented by Greene (1980, 1990, 1993, 2003, 2005) and used by Grigoli and Kapsoli (2013) and Grigoli (2014). The econometric methods incorporate uncertainty into the analysis and assume a functional form for the production function relating inputs and outputs. The residuals of the estimation are assumed to be composed by random noise and country-level inefficiency. Both components can be separated with specific assumptions on the statistical distribution of the inefficiency component. Efficiency can be estimated with this regression framework and statistical inference can be performed on the estimates. The QEH examines technical efficiency of public spending in infrastructure, education, and health, employing all parametric and non-parametric methodologies: FDH, DEA, Bootstrapped FDH, Bootstrapped DEA, and SFA. 5 In addition to technical efficiency calculations, the QEH also reports the quality of overall infrastructure from the Global Competitiveness Index (GCI) of the World Economic Forum and the Logistics Performance Index (LPI) of the World Bank. The WEF indicators of the overall infrastructure or the quality of trade and transport infrastructure are highly correlated with the LPI (Figure 7 and 8). Given delays in reporting the WEF measure, the LPI may be used as a substitute; the QEH reports the WEF indicators for historical comparative purposes and uses the LPI to ensure continuity in future updates.6 Figure 7 Relationship between the WEF Quality of Overall Infrastructure and the WB Logistics Performance Index 5 For education, the output is proxied by the literacy rate; for health, output is proxied by life expectancy at birth. Both output measures are described in Table A ((1.b) Technical Efficiency) in Appendix A and the list of education and health output variables can be extended as discussed in Herrera and Pang (2005) and Herrera and Ouedraogo (2018). 6 Most technical work on the efficiency of infrastructure used the WEF indicators and was the basis for numerous PERs. 11 Figure 8 Relationship between the WEF Quality of Trade and Transport Infrastructure and the WB Logistics Performance Index Allocative efficiency of public spending- Gauging allocative efficiency in the context of the Farrell (1957) framework used to gauge technical efficiency is difficult because the analysis now requires cross-country comparisons of input prices and demand-side information that is often unavailable.7 To examine the allocative efficiency of public resources to an activity or a project, the QEH follows the two central principles of cost benefit analysis, as described by Harberger (2003), Jenkins (2006), Devarajan (1997) and others: (i) there must be a rationale for the intervention; and (ii) the social marginal benefit of the project must exceed the marginal cost of public funds. This analysis must be done at the project or program level, aggregation at the national level is impossible from a practical perspective, and the concept of net present value of total spending is also meaningless from the conceptual viewpoint. However, the QEH infers allocative efficiency of spending from three different angles: (i) by exploring the quality of the country’s budget institutions that carry out this cost benefit analysis and that execute the budget,; (ii) by investigating the impact of spending on growth, assuming that the larger spending multipliers indicates adequate project selection and execution; 7Allocative efficiency requires that the marginal rate of transformation (MRT) equals the marginal rate of substitution (MRS) in consumption. On the production side, MRT must equal to the ratio of input prices and on the demand side, the MRS must equal the ratio of goods prices. It also requires that the production bundle composition matches the consumer demand for those goods. The analysis of allocative efficiency of spending requires General Equilibrium models built for individual countries or groups of countries that have similar characteristics. 12 and (iii) by examining the benefit/cost ratio of specific programs where a large country coverage exists, and assuming that the country’s performance quality may be extrapolated to the rest of public spending. The quality of budget institutions The country needs institutions and strong capacity to build a pipeline of projects that have been evaluated, to monitor budget execution, and to evaluate whether spending meets the assigned objectives. Indicators on the quality of budget institutions and frameworks are useful in inferring the quality of resource allocation. The heatmap uses The Public Expenditure and Financial Accountability (PEFA) Framework (IMF) and Open Budget indicators to capture the relationship between budget institutions and economic outcomes, such as efficiency. The PEFA is a comprehensive review of budget institutions using 31 indicators clustered into seven categories: (i) budget reliability, (ii) transparency of public finance, (iii) assets and liabilities management, (iv) policy-based fiscal strategy and budgeting, (v) predictability and control of budget execution, (vi) accounting and reporting, (vii) external scrutiny and audit. Countries are rated in each category and the QEH reports the scores and rankings in each of them. With this, a clear correlation was found between the quality of budget institutions and the efficiency of public spending (Figure 9). 8 There is also a clear correlation between the aggregate indicator of the quality of budget institutions and the governance indicators from Worldwide Governance Indicators. ( Figure 10). Figure 9. Public Spending Efficiency and Quality of Budget Institutions Synthetic Index 8The aggregate indicator of quality of budget institutions is obtained by Principal Component Analysis (PCA) taking the first principal component of the 31 indicators. 13 Figure 10. Quality of Budget Institutions and Governance Indicators The linear relationship between the synthetic PEFA indicator and technical efficiency (Figure 9) is a simplification of an overly complex relationship between the budget institutions and economic outcomes. This relationship between 31 PEFA indicators and multiple indicators of technical efficiency (described in previous subsection) using data for more than 60 countries needs to be explored with machine learning (ML) methods to better capture the non-linearities. Preliminary results indicate strong non-linearities in the relationship between the quality of institutions and the measures of technical efficiency; they also suggest that different sectors (e.g., health and education) are correlated with different subsets of the 31 budget quality indicators (Gomez and Kamm, 2023). This complex relationship needs to be explored further to define the statistical inference and help inform policy on budget reform and PFM institutions as part of the expenditure heatmap (QEH) agenda. In public resource allocation, capital spending is preferred given its long-term impact on growth. However, the return on new public investment may depend on the existing stock of public capital, as there are decreasing marginal returns to scale. The heatmap therefore reports different indicators on the size of public investment and the public capital stock: (i) public investment as a percentage of total expenditure; (ii) the stock of public capital as a percentage of Gross Domestic Product (GDP); (iii) and the ratio of the public capital stock to public employment to capture the factor intensity K/L ratio. The QEH also includes the stock of private capital as a percentage of GDP, as private capital complements public capital in delivery of services and tends to be low in developing countries. Figure 11 shows the positive association between a country’s income level and its ratio of private to public stock. The ratio of expenditure on goods and services, including maintenance, to the wage bill or to capital spending will be included in later versions of the heatmap. 14 Figure 11. GDP per capita and Private to Public Capital Stock Ratio The size of fiscal multipliers The literature on the size of fiscal multipliers is vast and mostly from the angle of stabilization. The QEH classifies the topic in the allocative efficiency section, which is a long-term context, as a proxy for quality of the project evaluation and implementation. This variable is not included in the stabilization section to avoid duplication. This subsection uses four main sources of data: (A) Bou-Habib, Francois, Utz (2023); (B) Comelli et Al,(2023), (C) Geli et Al,(2023) and (D) Batini et Al, (2014). The first three sources (A, B, C) report econometric estimates at country level or for country groups, while the last one provides a preliminary calculation based on individual country data for variables considered determinants of the size of fiscal multipliers. A, B, and C are results from published papers and cannot be updated regularly, the fourth calculation (D) can be adjusted and allows for user discretion as described below. Another difference across the studies is that estimate B only presents short-term fiscal multipliers and C only provides long- term multipliers, the other two have both short-term and long-term estimates. For a final estimate of the fiscal multiplier at the country level for short- and long-term, the QEH averages the three estimates (B, C, and D) for short-term fiscal multiplier and another three estimates (A, C, and D) for long-term multiplier. The back-of-the-envelope (Batini et. al.) method D considers several variables that capture structural characteristics of the economy associated with the size of multipliers: (i) the degree of openness of the economy as more closed economies have larger multipliers; (ii) labor market rigidities, with higher rigidity associated with larger multipliers; (iii) the size of automatic stabilizers, with small stabilizers associated with larger multipliers; (iv) the exchange rate regime with fixed exchange rate systems having larger multiplier effects; (v) the size or safety of public debt, with lower and safer public debt associated with larger multipliers; and (vi) the effectiveness of expenditure management and revenue administration systems, with more effective management associated with larger multipliers. Each of these structural characteristics is proxied by several variables where there is room for user discretion to add new variables9 9 The openness of the economy is proxied by the ratio of imports to GDP and the ratio of exports plus imports to GDP; labor market rigidity is proxied by the strictness of labor market regulations from the Fraser Institute Economic Freedom Index; the size of automatic stabilizers is proxied by three variables : the size of spending as a share of GDP (Battini), the change in the 15 Each of these is percentile-ranked, and the average percentile rank is calculated for each country. Countries are then clustered into four categories according to their average percentile ranking: the bottom quarter of are classified as low multiplier countries (with multipliers between 0.1 and 0.3); the second quarter is classified in the lower-middle multiplier category (multipliers between 0.3 and 0.5; the third quarter is classified in the upper-middle multiplier category (multipliers between 0.5 and 0.7); and the top quarter of the countries is classified in the high multiplier category with multipliers between 0.7 and 1.0. Batini et al propose adjusting these multipliers depending on two factors that evolve over time: a) the positioning of the economy along the business cycle. If the economy is at or near the trough, the multiplier can be adjusted upward by 60%, but if it is near the peak, the multiplier needs to be revised downward by 40%. The asymmetric revision and magnitudes are based on the Batini. b) the degree of monetary accommodation, found to be a key component of successful stabilization plans. Batini et al suggest symmetrically adjusting the value of the multipliers up to 30%, depending on the fully accommodative case of zero lower bound or the fully restrictive monetary policy. The positioning of the economy varies over time and from country to country, as does the degree of monetary policy accommodation, and so the adjustment is left to the user’s discretion. Such specific adjustments are more related to the stabilization role of policy which is not the focus of this section. To derive a single short-term multiplier, the QEH averages estimates from the Comelli, Geli, and Batini approaches. While the Batini method provides country-level estimates, Comelli and Geli do not, but rather regional or income group estimates. To obtain a final country-level short-term multiplier, we take an average of three estimates: one country-level estimate from the Batini method; and two estimates of a corresponding regional or income group from Comelli10 and Geli. To obtain the long-term fiscal multipliers, the QEH draws on Bou-Habib et al, (2023), Geli et Al,(2023), and Batini methods. For the latter, the authors suggest increasing the first-year impact of the fiscal shock by 10-30 percent in the second year and tapering off the impact to zero by the fifth year. Similarly, we assume that the impact of the first-year shock (the final short-term multiplier) increases by 20%—the midpoint of the proposed range—and then gradually decreases to zero over the next four years.11 The QEH then takes an average of the three estimates to obtain a comprehensive estimate of the long-term fiscal multipliers. Since Bou-Habib et al. (2023) and Geli et al. (2023) only provide regional or income group estimates, those are used in the averaging at the country-level. The Benefit/Cost ratio of public programs Ensuring allocative efficiency depends on how well the cost-benefit analysis has been performed. Although this information is not captured at the country level, the Social Protection and Jobs Practice of the World Bank includes a measure of the benefit/cost ratio of social protection and labor programs as cyclical primary balance (equation 3 from Fedelino), and by the change in the output gap multiplied by the share of expenditure in GDP (Fedelino et al equation 13a). The exchange rate regime classification is from the IMF; the safety of public debt is proxied by the credit rating score from 1-21; and the effectiveness of spending and revenue is proxied by: the technical efficiency of infrastructure, health, and education spending, the tax capacity gap, and the C-efficiency score described in the funding section. 10 If a country-specific estimate is available, it is also used, the average of that country’s estimate is taken as is that from the corresponding regional or/and income group estimates. 11 The impact in the third and fourth year is linearly interpolated between the second year and fifth year. For example, if the impact of the first-year fiscal shock is 1 (case with unity of first-year fiscal multiplier), the second-year impact is 1.2 (a 20% increase), 0.8 for third year, 0.4 for the fourth year, and zero in fifth year. The cumulative effect over five years (3.4 for this example) is then used as the Batini method’s estimate for long-term multipliers. 16 part of the Atlas of Social protection Indicators of Resilience and Equity (ASPIRE). Figure 12 illustrates the benefit-cost ratio by region, measured as the reduction in poverty gap12 obtained for each $1 spent in SPL programs; this result suggests that SPL programs in ECA countries alleviate poverty about twice as efficiently as in other regions. Figure 12. Benefit- cost ratio of SPL programs by region 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 Europe & South Asia Middle East East Asia & Sub-Saharan Latin Central Asia & North Pacific Africa America & Africa Caribbean Source: ASPIRE, World Bank The QEH includes this measure to infer how well cost-benefit analysis has been undertaken in the country. The close correlation between the Benefit/Cost ratio and the quality of budget institutions supports this assumption (Figure 13 )13 Figure 13: Benefit to Cost Ratio of SPL Programs and the Quality of Budget Institutions 12 “The poverty gap index is the average percentage shortfall in income of poor people from the poverty line and it is measured assuming the absence of programs (pre-transfer welfare distribution). Specifically, poverty gap reduction is computed as (poverty gap pre-transfer-poverty gap post transfer)/poverty gap pre-transfer.� (ASPIRE: https://www.worldbank.org/en/data/datatopics /aspire/documentation). 13 The figure reports the synthetic measure of the 31 PEFA indicators in its horizontal axis, obtained from the first principal component analysis (PCA) discussed in the subsection of budget institutions. The vertical axis is the B/C ratio of the ASPIRE indicators. 17 The QEH also includes a series of indicators related to adaptation and mitigation of the impact of climate change. The more resources devoted to this public good indicates a stronger planning and technical capacity, as well as a solider political cohesion to mobilize resources for future generations. The heatmap reports the following indicators related to climate change: fuel subsidies, with larger spending being less preferable, and public expenditure on environmental protection (more spending being preferable). Also, the Notre Dame - Global Adaptation Initiative (ND-GAIN) produces indices of readiness and vulnerability to climate change, which are also included in the heatmap.14 The volume of resources is important, as is the efficiency with which they are deployed. The QEH reports input and output efficiency scores calculated with the DEA and FDH efficiency frontiers for the readiness and resilience indicators from the ND-GAIN (Figure 14). The red line represents an efficiency frontier estimated by deterministic FDH, which shows that developing countries tend to spend less, and are farther way from the efficiency frontier. Figure 14. Efficiency of climate change spending i) Readiness to climate change vs. ii) Resilience to climate change vs. Environment protection spending Environment protection spending B. Stabilization (green tabs in excel sheet) The attainment of the objective of stabilization can be proxied through multiple indicators, clustered into three main categories or subcomponents: (i) those related to volatility of spending and public finances; (ii) those associated with budget rigidity; and (iii) those related to procyclicality of spending and behavior of the fiscal impulse. Figure 15 summarizes the contents of this section of the heatmap. 1. Volatility of spending Volatility of spending is undesirable for several reasons. From the macro perspective, it is associated with volatility of household consumption, which implies welfare losses for the risk-averse households; from the micro perspective, it causes delays in project execution and raises the cost of funding for public 14 We transformed the original ND-GAIN vulnerability index to convert it to a resilience index so that more is better. 18 projects. There is a clear negative association between the volatility of investment spending and the infrastructure technical efficiency measure described in the previous section (Figure 16). Figure 15. The indicators under the pillar of stabilization Note: the number of indicators in parentheses Figure 16. Public spending efficiency and volatility of public investment 19 The heatmap therefore includes measures of the volatility of public consumption spending, investment spending, and total spending. In addition to reporting the volatility of spending aggregates, the QEH also controls for the volatility of GDP growth, given the positive association between the two (Figure 17). Another proxy for policy volatility follows the Fatas and Mihov framework (2003, 2005), where the discretionary fiscal policy is defined as the residual εi,t of the following model: �G i, t = � i + � i �Yi, t + � i �G i, t -1 + � i Wi, t + � i ,t (1) where: - ΔG is the growth rate of government spending. - ΔY is the current output growth. - W is a vector of control variables including inflation and an oil price index. The volatility of fiscal policy is calculated as the standard deviation of εi,t for any country i.15 This measure of volatility is positively associated with household consumption volatility (Figure 18) and is the source of welfare losses when consumers are risk-averse and prefer a stable and predictable level of consumption; this motivates the importance of stabilization as a role for fiscal policy.16 Figure 17. Fiscal volatility, government consumption expenditure and GDP growth (1980-2020) 0.4 Eritrea consumption expenditure growth Congo, Rep. Zimbabwe SD of general government final 0.35 Nigeria Congo, Dem. Rep. Somalia Lebanon 0.3 Sierra Leone Guinea-Bissau 0.25 Chad Ghana y = 0.8452x + 0.0424 0.2 Guinea R² = 0.1195 0.15 Iraq 0.1 Equatorial Guinea 0.05 Macao SAR, China Georgia Bosnia and 0 Herzegovina 0 0.05 0.1 0.15 0.2 0.25 Standard Deviation of GDP growth Source: WDI. Note: t-stats for the intercept and coefficient are 3.93 and 4.80, respectively. 15 Current output growth was also estimated using as instrumental variables two lags of output growth, an index of real oil prices, lagged inflation, and lagged government spending growth. The country rankings do not change. 16 The original Lucas estimate of this cost for the US economy was less than one twentieth of one percent of consumption. Later, Obstfeld (1994) obtained larger estimates, but still below or around one percent of consumption. Most quantifications of this cost refer to advanced economies, and the few for developing economies find substantial costs. (Athanasoulis and van Wincoop, 2000; Pallage et. al. 2003; Herrera, 2007) 20 Figure 18. Public spending volatility and household consumption volatility (1980-2020) 2. Rigidity of spending. Public spending needs to be flexible to be reallocated as the government’s priorities change. This will happen when overall macro conditions change as key variables like the interest rate and the exchange rate will be affected. Flexibility/rigidity of the budget needs to be determined as the more rigid the budget, the lower its quality. There is evidence of a negative association between rigidity and efficiency of spending (Figure 19A), and that countries with lower efficiency of spending have higher spending levels, higher tax rates, and higher public debt (Herrera-Olaberria, 2020). Higher rigidity also hampers stabilization, as it is associated with higher financing needs and a lower likelihood of initiating fiscal consolidation if needed (Herrera-Olaberria, 2020). Figure 19B shows that countries with more rigid expenditure also have lower fiscal space. The heatmap presents two alternative measures of public expenditure rigidity: parametric and non- parametric. The parametric approach consists of econometric estimates of a function of spending determined by the level of income, population, and demographic characteristics,17 and taking the regression coefficients to estimate a country’s expected level of spending, given its level of income, population, and demographic characteristics. The expected level is the structural level of spending beyond the policymaker’s control and therefore, the rigid component of spending (Herrera-Olaberria, 2020 and Herrera-Velasco 2019). The non-parametric approach takes observed country data from the BOOST dataset or the IMF for wages, pensions, and interest payments, and aggregates them. There is no filtering or transformation to the data. 17 For example, young age dependency ratio (for wages), old age dependency ratio (for pensions and wages). 21 Figure 19. Rigidity of Spending and Fiscal Outcomes A. Public sector efficiency and rigid B. Budget rigidity and fiscal space19 expenditure18 3. Procyclicality of spending and fiscal impulse To achieve stabilization, fiscal policy must be countercyclical; therefore, this is the first dimension explored in this section of the QEH (Vegh, 2014; Frenkel, Vegh and Vuletin, 2011). Procyclicality of spending is commonly measured by regressing the cyclically adjusted component of spending vis à vis the cyclically adjusted component of output; a negative coefficient indicates a countercyclical expenditure policy. An alternative measure of procyclicality of policy is based on the cyclically- adjusted fiscal balances by regressing this variable on the output gap (Jalles et Al, 2023; IMF, 2022); a positive coefficient indicates a countercyclical policy stance. Both measures are reported in the heatmap. The procyclicality of fiscal policy is correlated with the quality of institutions (Fig. 18 and Frankel, Vegh, Vuletin, 2013). The alternative method with the cyclically adjusted budget balance yields identical results. The fiscal impulse is defined as the change in the cyclically-adjusted primary balance (CAPB), and it approximates the impact of fiscal policy on aggregate demand (Blanchard, 1990). A decline in the CAPB is considered an economic stimulus and an increase indicates a withdrawal of demand for resources from the economy. Since the fiscal impulse changes on a yearly basis, it is a volatile indicator; to evaluate whether fiscal policy is stabilizing, the heatmap computes the average over the business cycle. The duration of the business cycle is specific to each country, and the user can change this, but the default is based on an average duration of years, as reported by Calderon and Fuentes (2010). 18 Orthogonalized rigid expenditure is estimated by controlling actual rigid expenditure (log) for log of GDP per capita. The residuals are taken and denoted as orthogonalized rigid expenditure. 19 The index of fiscal space is derived from the first principal component of government debt, fiscal balance, and credit ratings. 22 Figure 20. Procyclicality of fiscal policy and governance Source: Worldwide Governance Indicators and authors’ calculation C. Redistribution (orange tabs in excel sheet) Redistribution of resources can be across different income groups, across gender, or across generations. These three topics are depicted in the three subcomponents of this pillar. Figure 21 summarizes the contents of this section. Figure 21. The indicators under the pillar of redistribution Note: the number of indicators in parentheses 23 1. Income Inequality The income redistribution effect of fiscal policy is analyzed in the work of Lustig (2018) and the Commitment to Equity (CEQ) Foundation who developed methodologies to quantify the impact of fiscal policy on inequality and poverty. The redistributive effect of taxes and transfers is quantified by computing the difference between the Gini coefficient before taxes and transfers, and the Gini coefficient resulting after taxes and transfers. The larger the difference the greater the impact of policy in redistribution. The heatmap reports this gap, sourced from the CEQ foundation and the OECD. The heatmap also reports indicators that describe the scope and performance of social protection and labor programs (SPL); these indicators are constructed by the World Bank and reported in the Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE). Specifically, the heatmap includes SPL Program Coverage (Poorest quintile), SPL Program Benefit Incidence (Poorest 20%), poverty headcount reduction, and poverty gap reduction by SPL programs (See the detail in Appendix). 2. Gender Inequality Redistribution across groups of individuals can also happen across gender. Fiscal policy may affect income distribution between males and females, and a recent proposal to examine the fiscal incidence of spending and tax burdens across genders is innovative (Jellema, et.al. 2022). It expands the CEQ approach by including additional information from household surveys and examines the impact of fiscal policy on gender distributional effects, at a point in time. The heatmap incudes a measure of the wage gap by gender from Worldwide Bureaucracy Database, as well as an indicator of the female employment in the public sector from ILOSTAT. 3. Intergenerational Inequality Redistribution can take place across generations, such as would happen with pension payments or climate change spending. Pension payments redistribute income to the old (from the young) and climate change spending redistributes resources to future generations. Both can be considered objectives of fiscal policy in a wider perspective that goes beyond the impact of fiscal policy on growth. In analyzing spending on the elderly population, we include pension as a share of total spending and the share of structural components in the actual pension payment, following the methodologies described in III B 2. To examine spending across generations, QEH calculates the ratio of government pension payment per young person to education expenditure per elderly person. As shown in Figure 22 the allocation of spending across generations depends on demographics, with aging countries spending more on the elderly. 24 Figure 22. The ratio of government pension payments to expenditure on education and old age dependency (avg. 2015-2019) 8.0 7.0 Ratio of pension and education 6.0 GRC JPN 5.0 spending 4.0 LUX AUTSRB BEL FRA BGR SVK PRT 3.0 RUS POL SVNHRV URY DEU BLR CZE HUN CYP IRLNZL UKR CHE LTU FIN 2.0 ARG ARM ALBAUS ROU USA GBR SWE COL TUR MDA JORKGZ BRA NOR NLD 1.0 IRQ NIC UZB SYC PRY PAN MUS ISR ZAF AZE PHLNPLSLV CHL SAU CIVMDV JAM LSOMYS ZWE CRI PER 0.0 ZMB SLE GHA CAF GNB GAB COD MWI SGP THA DNK 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% -1.0 Old age dependency Source: WEO and WDI IV. Aggregation of indicators Previous sections discussed numerous indicators included in the heatmap (Appendix Table A). The multiplicity of indicators and therefore the multi-dimensionality of the issue means that is it challenging to describe the evolution of a country’s quality of spending, or to compare quality across countries. To help synthesize the volume of information for each country, the heatmap aggregates the indicators at the pillar level (long-term growth, stabilization, redistribution), and then presents an overall aggregate measure. This section discusses the procedures. A. Aggregation Methodology The aggregation process in the QEH starts at the individual pillar level. First, each individual indicator (raw data) is converted to a percentile rank (0-1; 1 is performing better than others and 0 is under-performing all others). The heatmap then calculates a simple average of the percentile ranks in each subcomponent and each subcomponent is aggregated into the Pillar level weighting each subcomponent by the volume of information, or the number of variables used in its construction. For example, the Long-Term Growth pillar has three subcomponents: sustainability, technical efficiency, and allocative efficiency, each with a different number of indicators: 23 for sustainability, 10 for technical efficiency, and 27 for allocative efficiency for a total of 60 indicators. The sustainability subcomponent is weighted by 38% (23/60), the technical efficiency by 17% (10/60), and the allocative efficiency by 45% (27/60). The stabilization pillar has 3 subcomponents: volatility (4), rigidity (3), and cyclicality and fiscal impulse (3), where the aggregation process starts by calculating the simple average at the subcomponent level, and then aggregating the subcomponents, weighting each by the number of indicators in each. The Stabilization and Redistribution Pillars are aggregated following the same procedure: first, at the subcomponent level the individual indicators are averaged, and then, at the Pillar level the 3 subcomponents are averaged with each subcomponent weighted by the number of indicators involved in its construction. Once the 25 quality indicators (weighted average) are calculated for the three main pillars, these are aggregated to obtain an overall index for the quality of expenditure. The three pillars are averaged using alternative methods: with equal weights (by default), using the number of indicators used in each pillar as weights, or arbitrary user-selected weights that reflect evolving preferences of the policymaker on the priority of the objectives of long-term growth, stabilization, and redistribution. B. Aggregate indicators of quality of expenditure Table 1 reports the rankings in each pillar, by income group and by region. High-income countries have a higher quality of expenditure in all pillars, except for redistribution, where upper-middle income countries are in the same rank. The largest dispersion in quality between top-ranked and bottom-ranked regions happens in the long-term growth objective. In Table 1B, North American and European countries tend to perform better, while SSA countries tend to have the lowest quality of spending in most of the pillars, except redistribution, where MENA countries rank lower. Figure 23 provides a geographical representation of the overall index of quality of expenditure, with SSA and MENA having most of the countries with lower quality of expenditure. Table 1. Percentile ranks for each pillar and overall score. 1. Long-term 2. 3. Overall growth Stabilization Redistribution A. Income Groups High income 0.61 0.58 0.60 0.59 Upper middle income 0.51 0.48 0.56 0.51 Lower middle income 0.42 0.43 0.41 0.42 Low income 0.33 0.27 0.32 0.30 B. Regions East Asia & Pacific 0.50 0.51 0.46 0.48 Europe & Central Asia 0.61 0.60 0.60 0.61 Latin America & Caribbean 0.51 0.48 0.51 0.50 Middle East & North Africa 0.47 0.39 0.56 0.47 North America 0.70 0.63 0.65 0.66 South Asia 0.42 0.48 0.42 0.44 Sub-Saharan Africa 0.37 0.33 0.38 0.36 26 Figure 23. Overall index for quality of expenditure20 Source: Authors’ calculation The ranking differences across pillars at the regional level and across income groups leads to an examination of the correlation of the rankings across the different pillars to see if countries that perform well in one objective, rank equally well in other objectives. They have been found to rank equally well, with some caveats. Rank-correlation coefficients are positive and highly significant, confirming that countries that achieve long-term growth objectives are also ranked high in achieving stabilization and redistribution objectives (Table 2). The positive association between stabilization and redistribution is weaker. To evaluate the robustness of the relationships over time, we shortened the sample to 2016-2019 and results did not change (Table B1 in Appendix B). Table 2. Correlation across pillars, full sample, 2005-2019. The problem with using different sample periods is that several variables do not change because of how the dataset is built or because some information is unavailable.21 When these variables are removed from the dataset, the correlations across the pillars change. The positive association between the first two pillars remains, but there is a weak negative correlation between long-term growth and redistribution, and a strongly negative association between redistribution and stabilization (Table 3). These conclusions do not change in a shorter horizon, or if AEs are included in the sample (Table B5 in Appendix B). 20In the process of aggregation, QEH takes a five-year average from 2015-2019. 21For instance, the Gini coefficient is available for a point in time, following the CEQ methodology. The databse includes the last available datapoint. Similarly, the database includes the last datapoint in the credit ratings or the EMBI spreads, so these do not change over time. 27 Table 3. Correlation across pillars, excluding Advanced Economies (AE), 2016-19, and reduced set of variables. The correlation coefficients need to be interpreted with caution, as they are computed based on cross- sectional data. Consequently, it is not feasible to account for unobservable, time-invariant factors or the so-called "fixed effects" for individual countries. The correlation might be capturing country-specific unobserved effects from a third (omitted) variable. The redistribution pillar is also reduced to only one variable when time-invariant variables are removed. Hence, these correlations are simply suggestive and motivate a more in-depth analysis using panel data or time-series data, as well as a complete economic model to understand the interactions across expenditure policy objectives using different instruments. The rankings of the three pillars (growth, stabilization, and redistribution) are correlated with outcomes that are exogenous to the ranking exercises, indicating that policy orientation is related to outcomes, and may be interpreted as a measure of policy effectiveness. For instance, the rankings in Pillar I (allocation & growth) are correlated with growth; when controlling for the convergence effect there is a positive relationship between the ranking in Pillar I and average growth in 2005-2019 across countries (Figure 24). A cross-section regression of the growth level, controlling for convergence effect verifies that Pillar I rankings are highly significant and positively associated with growth (Table 4). Figure 24. Pillar I rankings and average GDP Growth, 2005-2019 28 Table 4. Pillar I rankings are positively associated with average growth rates, 2005-2019. OLS VARIABLES Coef std err Dependent variable: Growth rates . Long-term Growth Pillar (score) 6.8*** 1.654 GDP per capita (log) -1.339*** 0.175 Constant 12.21*** 1.049 Observations 182 R-squared 0.283 Adj R-squared 0.275 F-stat 34.91 *** p<0.01, ** p<0.05, * p<0.1 The rankings in Pillar II (stabilization) also show interesting correlations with inflation and growth volatility. First, a country-specific rate of inflation was estimated by removing the worldwide rate of inflation (estimated as the first principal component vector) showing a negative association between the ranking in Pillar II and the country-specific inflation rate (Figure 25). The country ranking is also negatively associated with the volatility of GDP growth (Figure 26). Figure 25. Rankings in Stabilization (Pillar II) and Average Inflation across countries 29 Figure 26. Rankings in Stabilization (Pillar II) and volatility of Growth Across Countries Finally, we examine the Pillar III (redistribution) ratings with a redistribution outcome variable which the QEH has not already used. It is difficult to come up with an exogenous variable to the ratings exercise and so an indirect approach was adopted, using a subjective well-being (SWB) measure, as the literature has found a positive relationship between egalitarian distribution and happiness (Dwyer et Al, 2022; Ferrer-i- Carbonell, 2012). There is a considerable debate on the use of SWB in economics (Benjamin, et Al, 2023); still the QEH uses the index of the World Happiness Report, used in numerous economics papers (Deaton, 2018)22. There is a high and significant positive association between the countries that are ranked highest in Pillar III and the rankings of subjective well-being (Figure 27). As in the other two pillars, the aim is not to explain the degree of subjective satisfaction (ladder-score), but simply to verify a correlation. Pillar IIl ratings are also correlated with a variable of the Varieties of Democracy dataset that measures equality of distribution of resources23; the countries with higher Pillar III rankings also have a more equal distribution of resources per the Varieties of Democracy construction (Figure 28).24 22 Subjective well-being (SWB), or happiness index, is derived from Gallup World Surveys. The survey asks individuals to imagine a ladder, with steps numbered from 0 at the bottom to 10 at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst Source: https://worldhappiness.report/ed/2019/. 23 The variable is v2xeg_eqdr. The index is formed by taking the point estimates from a Bayesian factor analysis model of the indicators for particularistic or public goods v2dlencmps, means-tested vs. universalistic welfare policies v2dlunivl, educational equality v2peedueq and health equality v2pehealth. Sigman et al. (2015, V-Dem Working Paper Series 2015:22); Pemstein et al. (2023, V-Dem Working Paper Series 2023:21); V-Dem Codebook 24 The close association between the Pillar III rankings and the numerous variables described above must be interpreted with caution, as there may be an omitted variable that could explain the correlation, this could be the GDP in levels. Table C in Appendix C shows the statistical positive association vanishes when GDP is included, reflecting the multicollinearity. 30 Figure 27. Pillar III Rankings (redistribution) and Subjective Well Being (SWB) Index Figure 28 Pillar III rankings and Equality of Distribution of Resources from V-Dem Database. V. Visualization tool and summary of information: Three case studies To condense the information and facilitate comparison of individual country data with different country clusters (regional, income-group, FCV status, or natural-resource dependence), the QEH has a built-in "Summary" tab and visualization tool, which shows each country’s ranking in each of the three objectives 31 (long-term growth, stabilization, and redistribution) using radar charts (Figure 29). Users have the option of selecting their country of interest (C4 cell, highlighted in light orange to indicate that it can be changed). Once the country is selected, the tool automatically sets benchmarks based on income group, region, Fragility, Conflict, and Violence (FCV) status, and whether the country is resource-rich. In addition to country income level, the tool also reports the data for the income level immediately above, to facilitate comparisons with aspirational peers. In column C, starting from row 10, all indicators used for the comparisons are listed (see Table A for details), allowing users to choose the indicators to include in the analysis by ticking/unticking boxes. On the right-hand side of the “Summary� tab (Figure 29), there are snapshots that show where the selected country is positioned compared to its peers. This is represented through radar charts, with the selected country highlighted in red. The radar chart in the upper-left section represents the three main pillars, while the three radar charts at the bottom provide a breakdown of these pillars. These visualizations allow users to easily understand how the selected country performs in relation to its peers. The subsequent section of the tool focuses on three example countries: Chile, Cameroon, and Poland. The county selection is for illustrative purposes only. Figure 29. Quality of Expenditure Heatmaps: “Summary� Tab A. Chile The overall index for quality of expenditure in Chile is 0.66, which is higher than the average for countries in the same income group (HIC) and region (LAC): 0.59 and 0.50 respectively (Table 5).Figure 30A provides a breakdown of achievement in the three main pillars: long-term growth, stabilization, and redistribution. Performance in these categories is consistently high, with scores above 0.6. In all the main pillars, Chile ranks higher than the income and regional comparators. 32 Table 5. Overall index for quality of expenditure (Chile and its peers) CHL HIC LAC Overall score 0.66 0.59 0.50 Long-term growth 0.64 0.61 0.51 Stabilization 0.69 0.58 0.48 Redistribution 0.65 0.60 0.51 The breakdown of each pillar reveals details that are interesting. For long-term growth, Chile’s ranking is above the LAC aggregate and similar to the HIC (Figure 30B). In the stabilization pillar, it ranks above all comparators in the Rigidity of the Budget and the Fiscal Impulse indicators but lags the HIC in the cyclicality and fiscal impulse measure, suggesting that Chile employs counter-cyclical fiscal policies to a lesser extent than HICs on average (Figure 30C). In the redistribution pillar, Chile lags in addressing gender inequality and income inequality, but performs better in intergenerational indicators. Figure 30. QEH radar charts - Chile A. Three core objectives of fiscal policies B. Breakdown of the long-term growth pillar Long-term Sustainability growth 1.0 1.0 0.8 0.6 0.5 0.4 0.2 0.0 0.0 Allocative Technical Redistribution Stabilization Efficiency Efficiency High income High income Latin America & Caribbean Latin America & Caribbean CHL CHL C. Breakdown of the stabilization pillar D. Breakdown of the redistribution pillar Income Volatility equality 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Cyclicality and Rigidity of Intergen Gender Fiscal Impulse Budget equality equality High income High income Latin America & Caribbean Latin America & Caribbean CHL CHL 33 B. Cameroon Cameroon is a LMIC, resource-dependent, and fragile, conflict-affected country25, providing a wealth of comparator clusters (Figure 31), alongside the average of the income group one level higher (UMICs). The overall expenditure quality score is 0.34, compared to 0.43 in LMICs and 0.38 in SSA (Table 6). The lower overall score primarily stems from the limited impact of expenditure on redistribution (Figure 31A). Table 6. Overall index for quality of expenditure (Cameroon and its peers) CMR LMIC UMIC SSA FCV Resource-dependent Overall score 0.33 0.42 0.51 0.36 0.34 0.42 Long-term growth 0.34 0.42 0.51 0.37 0.36 0.40 Stabilization 0.52 0.43 0.48 0.33 0.28 0.40 Redistribution 0.13 0.41 0.56 0.38 0.37 0.42 Regarding income equality, Cameroon's score is 0.1, significantly lower than its counterparts (Figure 31D). This disparity can be attributed to government interventions, such as social protection and labor programs, which have not yielded substantial reductions in poverty or income disparity. It is worth noting that the intergenerational equality score appears as zero, but it represents missing data, not included in the computation of the score; the same applies to rigidity of budget in Figure 31C. The user must bear in mind missing values when aggregating indicators, as outlined in section IV. 25 We use the same definition of resource-dependent countries as Herrera, S. W. Kouame, and P. Mandon (2019) do and FY23 List of Fragile and Conflict-affected counties (World Bank) for FCVs. 34 Figure 31. QEH radar charts - Cameroon A. Three core functions of fiscal policies, B. Breakdown of the long-term growth pillar Musgrave Long-term growth Sustainability 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Allocative Technical Redistribution Stabilization Efficiency Efficiency Lower middle income Upper middle income Lower middle income Upper middle income Sub-Saharan Africa FCV Sub-Saharan Africa FCV Resource-dependent CMR Resource-dependent CMR C. Breakdown of the stabilization pillar D. Breakdown of the redistribution pillar Income Volatility equality 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Cyclicality and Rigidity of Intergen Gender Fiscal Impulse Budget equality equality Lower middle income Upper middle income Lower middle income Upper middle income Sub-Saharan Africa FCV Sub-Saharan Africa FCV Resource-dependent CMR Resource-dependent CMR C. Poland Poland shows uniform ranking with its regional and income group comparators at the overall level. The scores for these objectives range from 0.58 to 0.67, resulting in an overall index for the quality of expenditure of 0.63 (Table 7). However, the breakdown of the stabilization pillar indicates higher rigidity in the budget than the income and regional comparators (see Section III B2 or Table A for detailed information)26. Poland's high percentage of rigid expenditure as a share of government total expenditure, contributes to a lower overall score for the rigidity of the budget (Figure 32C). Nevertheless, the estimation of structural expenditure reveals that a high proportion of rigid expenditure can be attributed to structural determinants like demographics. This factor increases the overall score for the rigidity of the budget, as countries with the greatest deviation from the structural 26Users can go to an individual sheet assigned to every indicator to see the actual value of indicators (For example, they can see the actual values of “Rigidity of Budget� indicator in “(2.b) Rigidity� tab. 35 level are considered less performing. However, this increase does not fully offset the decrease in the actual rigid expenditure variable. Table 7. Overall index for quality of expenditure (Poland and its peers) POL HIC ECA Overall score 0.63 0.59 0.61 Long-term growth 0.62 0.61 0.61 Stabilization 0.59 0.58 0.60 Redistribution 0.67 0.60 0.60 Figure 32. QEH radar charts - Poland A. Three core functions of fiscal policies, B. Breakdown of the long-term growth pillar Musgrave Long-term growth Sustainability 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Allocative Technical Redistribution Stabilization Efficiency Efficiency High income High income Europe & Central Asia Europe & Central Asia POL POL C. Breakdown of the stabilization pillar D. Breakdown of the redistribution pillar Income Volatility equality 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Cyclicality and Rigidity of Intergen Gender Fiscal Impulse Budget equality equality High income High income Europe & Central Asia Europe & Central Asia POL POL 36 References Batini, N., L. Eyraud, L. Forni, A. Weber (2014) Fiscal multipliers: Size, determinants, and use in macroeconomic projections. IMF. Fiscal Affairs Department. Technical Notes and Manuals. Benjamin, Daniel J., Kristen B. Cooper, Ori Heffetz, and Miles S. Kimball. 2023. "From Happiness Data to Economic Conclusions." Annu. Rev. Econ. 3: Submitted. DOI: https://doi.org/10.1146/annurev-economics-081623- 021136.Bird, R. 2005. Evaluating Public Expenditures: Does it Matter How they are Financed? in Fiscal Management ed by Anwar Shah. Public Sector Governance and Accountability Series. World Bank. Blanchard, O. (1990): «Suggestions for a New Set of Fiscal Indicators», OECD Working Paper No. 79. Buffie, E.F., A. Berg, C. Pattillo, R. Portillo, and L. F. Zanna. 2012. “Public investment, growth, and debt sustainability: putting together the pieces.� IMF Working Paper 12/144. International Monetary Fund, Washington, Butron-Calderon, L. (2022) Publiblic debt-stabilizing primary balance and debt decomposition excel tool. World Bank. Mimeo. Calderon, C. and R. Fuentes, 2010. Characterizing the Business cycles of emerging economies. Policy research Working Paper 5343. World bank. Causa and Hermansen. 2019. Income redistribution through taxes and transfers across OECD countries ECONOMICS DEPARTMENT WORKING PAPERS No. 1453. OECD. Comelli, F. A. David, L. Eyraud, P. Kovacs, J. Montoya, A.Sode (2023) Navigating Fiscal Challenges in Sub-Saharan Africa. International Monetary Fund. September. Deaton, A, 2018. What do self-reports of well-being say about the life-cycle theory and policy? Journal of Public Economics 162 18-25. doi: 10.1016/j.jpubeco.2018.02.014 Devarajan, S, V. Swaroop. H. Zou, 1996. “The composition of public spending and growth ,� Journal of Monetary Economics, Vol. 37, Issue 2, pp.313-344 Devarajan, S. L. Squire, S. Suthiwart-Narueput 1997. Beyond Rate of Return: Reorienting Project Appraisal. The World Bank Research Observer. Vol. 12 no. 1 Diener, R. Lucas, and S. Oishi (2018) Advances and Open Questions in the Science of Well Being. Collabra: Psychology https://doi.org/10.1525/collabra.115 Dwyer, R., and E. Dunn (2022) Wealth redistribution promote happiness. Procedures of the National Academy of Sciences. DOI: 10.1073/pnas.2211123119 Escolano, J. (2010). A Practical Guide to Public Debt Dynamics, Fiscal Sustainability, and Cyclical Adjustment of Budgetary Aggregates. Technical Notes and Manuals 10/02. International Monetary Fund. Washington D.C. Farrell, M. 1957 “The measurement of productive efficiency�, Journal of the Royal Statistical Society, Series A. Vol. 120, No. 3, pp. 253-290 Fatas A., Mihov I. 2003, The case for restricting fiscal policy discretion, QJE, March Fatas A., Mihov I., Fiscal discipline, volatility, and growth. Mimeo. The World Bank. Fatas A., Mihov I, 2005 Policy volatility, institutions, and economic growth, CEPR Discussion Paper Series, December 37 Fedelino, A., A. Ivanova, and M. Horton.2009.Computing cyclically adjusted balances and automatic stabilizers.IMF. Fiscal Affairs Department. Technical Notes and Manuals. Ferrer-i-Carbonell, A., and X. Ramos. 2012 Inequality and Happiness: A survey. Gini Discussion Paper 38 J. Frankel, C. Vegh, and G. Vuletin, 2013. "On Graduation from Fiscal Procyclicality," Journal of Development Economics, 100(1), , pp. 32–47. Garcia-Escribano, M, C. Yue Liu. 2017. Expenditure Assessment Tool. Technical Notes and Manuals. IMF. Geli, J.F. and A. Moura (2023) Getting into the nitty- gritty of fiscal multipliers: small details, big impacts. IMF Working Papers wp 23/29. February. Glomm, G., Ravikumar, B., 1994. Public investment in infrastructure in a simple growth model. J. Econ. Dyn. Control 18, 1173–1188. Glomm, G., Ravikumar, B., 1997. Productive government expenditures and long-run growth. J. Econ. Dyn. Control 21, 183–204. Gomez, JF and C. Kamm 2023. Using Machine Learning Models to Assess the Relationship Between Measures of Efficiency and Budget Institutions. Mimeo. EMFTX. World Bank. Green, W. 1980. Maximum likelihood estimation of frontier functions. Journal of Econometrics, 13. Pp 27-56. Greene, W. 1993. Econometric Analysis (Mc Millan, New York). Greene, W. 2003a. Distinguishing between heterogeneity and inefficiency: Stochastic frontier analysis of the World Health Organization’s panel data on national health care systems�. Mimeo. NYU. www.stern.nyu.edu/~wgreene. Greene, W. 2003b Fixed effects and bias due to the incidental parameters problem in the Tobit model. Mimeo. NYU. www.stern.nyu.edu/~wgreene Greene, W. 2005. Efficiency of Public Spending in Developing Countries: A stochastic Frontier Approach. Mimeo. World Bank. Grigoli, F., and J. Kapsoli. 2013. Waste not, want not: the efficiency of health expenditure in developing economies. IMF Working paper WP/13/187 Grigoli, F. 2014. A hybrid approach to estimating the efficiency of public spending on education in emerging and developing economies. IMF Working Paper WP/14/19 Gupta, S., and M. Verhoeven. 2001 . The efficiency of government expenditure: experiences from Africa. Journal of Policy Modelling Gupta, S., A. Kangur, C. Papageorgiou, and A. Wane. 2014. “Efficiency -Adjusted Public Capital and Growth,� World Economic Development, Vol. 57, Issue C: pp. 164–78. Haque, M. E. (2004). ‘The composition of government expenditure and economic growth in developing countries. Global Journal of Finance and Economics, Vol. 1, pp. 97–117. Harberger, A. 2003. Some recent advances in economic project evaluation. In Cuadernos de Economia. Nº 121, pp. 579-588 December. Harberger, A. 2004. Project evaluation as a development policy priority. Mimeo. Notes for a Public Expenditure Seminar. The World Bank. October. 38 Herrera, S., and G. Pang. 2005. Efficiency of Public Spending in developing Countries: An Efficiency Frontier Approach. Policy Research Working Paper 3645. World Bank Herrera, S. 2007. Public Expenditure and Growth. Policy Research Working Paper 4372. May. Herrera, S. W. Kouame, P. Mandon (2019) Why some countries can escape the fiscal pro-cyclicality trap and others can’t? Policy Research Working Paper 8968. World Bank. August. Herrera, S., and A. Ouedraogo. 2018. Efficiency of Public Spending in Education, Health, and Infrastructure: An International Benchmarking Exercise. Policy Research Working Paper 8586 World Bank . Herrera S. and E. Olaberria (2020) Budget Rigidity in Latin America and the Caribbean: Causes, Consequences, and Policy Implications. International Development in Focus. Washington DC. World Bank. Herrera, S., and J. Velasco (2019), “Budget Rigidity in LAC: A proposed new measure for empirical analysis.� Background Report for Regional Study on Budget Rigidities in Latin America. Mimeo. May. Herrera, Santiago; Vincent, Bruno. 2008. Public Expenditure and Consumption Volatility. Policy Research Working Paper No. 4633. © World Bank, Washington, DC. IMF (2022). Fiscal Monitor. October. Online Annex 1.1 Countercyclicality of Fiscal Policies Izquierdo, A. C. Pessino, G. Vuletin 2018. Mejor Gasto para Mejores Vidas. Banco Interamericano de Desarrollo. Washington. Jalles, J., Y. Kiendrebeogo, W.R. Lam., and R. Piazza. 2023. “Revisiting Fiscal Countercyclicality,� IMF Working Paper No.23/89, Washington, D.C. Jellema, J., C, Gown, A. Fuchs, M. Wai-Poi, S.Tiwari, M. Sosa. 2022. Gender and Fiscal Policy : A Methodological Proposal and its Application to Jordan and Armenia. Mimeo. Jenkins, G. 2006. Cost-Benefit Analysis; A tool for Public Expenditure Reviews (PER). Lecture Notes. Economic Policy and Debt Department. May. The World Bank. Johansson, A. (2016), "Public Finance, Economic Growth and Inequality: A Survey of the Evidence", OECD Economic Department Working Papers, No. 1346, OECD Publishing, Paris. Kiguel, M. and P. Neumeyer. 1989. Inflation and seigniorage in Argentina. Working Papers WPS 289. Policy Planning and Research. World bank. Kose, M. Ayhan, Sergio Kurlat, Franziska Ohnsorge, and Naotaka Sugawara. 2022. "A Cross-Country Database of Fiscal Space." Journal of International Money and Finance 128 (November): 102682. Loayza, Norman V. and Pennings, Steven Michael. 2022. “The Long-Term Growth Model: Fundamentals, Extensions, and Applications.�. World Bank Group. Lucas, R. (1987) Models of Business Cycles Lucas, R. (2003) “Macroeconomic priorities�, The American Economic Review, March 2003 Lustig, N. 2018 CEQ Handbook The Impact of Fiscal Policy on Inequality and Poverty. https://commitmentoequity.org/publications-ceq-handbook 39 Melina, G., S. C. S.Yang, and L. F. Zanna. 2014. “Debt Sustainability, Public Investment and Natural Resources in Developing Countries: The DIGNAR Model.� IMF Working Paper 14/50. International Monetary Fund, Washington, D.C. Musgrave, R. 1959. The Theory of Public Finance. McGraw Hill, New York. Pallage, S. and M. Robe (2003) On the welfare cost of economic fluctuations in developing countries, International Economic Review Vol. 44, No. 2, pp. 677-998 Ravallion, M. 2008. Evaluation in the practice of development. Policy Research Working Paper 4547. The World Bank. Simar, L., & Wilson, P. W. (1998). Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models. Management Science, 44(1), 49 –61. Stokey and Rebelo (1995) Suescun 2005 Fiscal Space for infrastructure in Colombia. Policy Research Working Paper 3625. The World Bank. June. Suescun 2007.The role of fiscal policy in human development and growth. Policy Research Working Paper 78499. The World Bank. March Suescun, R. 2020 A tool for fiscal planning in a medium-term fiscal framework. Economic Modelling 88 pp. 431–446 Talvi, E., Vegh, C. A. 2005. "Tax Base Variability and Procyclical Fiscal Policy." Journal of Development Economics, 78 156-190. Van Wincoop, E. (1994) “Welfare gains from international risk sharing� Journal of Monetary Economics, 34 pp. 175- 200 Vegh, C. 2015. Fiscal Policy in Emerging Markets: Procyclicality and Graduation. NBER The Reporter Zanna. F., E. Buffie, R. Portillo, A. Berg,(2019) Borrowing for Growth: Big Pushes and Debt Sustainability in Low- income Countries. The World Bank Economic Review, 33(3), 2019, 661 –689 Appendix A. Table A. List of all indicators in Quality of Expenditure Heatmaps27: Indicators Definition and Calculation Source Ranking (percentile rank)28 (1) Long Term Growth (1. a) Sustainability (1.a.1) Size of government: Expenditure controlling Residuals obtained from the below estimation: The authors' 0 = a country with deviation of observed for GDP (% of GDP) Random effect model with dummies of year and small calculation the largest spending from the structural island developing state (SIDS). Public expenditure as expenditure level (determined by GDP dependent variable and log of GDP per capita (PPP, controlling for and demographics) international dollars) as independent variable. This is GDP per capita because public expenditure tends to vary by income level. (1.a.2) Sustainability Gap: Gap between actual The gap is the difference between the primary balance A cross-country 0 = a country with difference between primary and debt stabilizing and the debt-stabilizing primary balance (pbsusgap1), database of fiscal the largest gap balance and debt-stabilizing primary balance from "A cross-country database of fiscal space (2022)." space (2022), World between actual primary balance Bank and debt stabilizing primary balance (1.a.3) Funding of public General Government General Government Gross Debt as share of GDP WEO April 2023 0 = a country with expenditure: debt burden, Debt (%GDP) the largest debt to debt composition, tax gap, GDP ratio grants, seignoriage, and inflation tax. Debt (% of General General Government Gross Debt as share of General WEO April 2023 0 = a country with Government Revenue) Government Revenue the largest debt to GG revenue 27 In the excel sheet of QEH, “metadata� tab displays Table A. 28 Each country is ranked in 0-1: 1 as performing well and 0 as performing less. 41 Debt (%Tax Revenue) (General Government Debt Stock)/(Tax Revenue), from The Macro Poverty 0 = a country with the Macro Poverty Outlook (MPO) Outlook (MPO) the largest debt to UNU-WIDER GG tax revenue Government Revenue Dataset (General Government) External public debt (General Government External Debt Stock)/(General ibid. 0 = a country with (%total) Government Debt Stock) the largest external debt to total debt External Debt/Domestic (General Government External Debt Stock)/(General ibid. 0 = a country with Debt Government Domestic Debt Stock) the largest external debt to domestic Debt duration Debt Duration: sovereign debt average maturity, years A cross-country 0 = a country with from a cross-country database of fiscal space database of fiscal the shortest debt space (2022) duration Interest Coverage Ratio (General Government Revenue - Grants)/(Interest Paid) WEO 0 = a country with lowest ICR Tax Revenue Gap (% of See revenue dashboard for the details World Bank 0 = a country with GDP) Revenue Dashboard the largest tax gap (Tax Capacity - Actual Value) Tax Revenue (% of GDP) General government tax revenue as share of GDP UNU-WIDER 0 = a country with Government the smallest tax Revenue Dataset revenue as % of GDP Change in Tax Revenue Change in general government tax revenue as share of ibid. 0 = a country with (% of GDP) GDP in selected years the largest decline in tax revenue to GDP ratio 42 VAT C efficiency VAT C-efficiency measures how effectively a Value- World Bank Staff 0 = a country with Added Tax is collected compared to an ideal scenario. Calculation the least It's a ratio of actual VAT revenue to a theoretical efficiency maximum, considering tax rates, exemptions, and administrative performance. A ratio of 1 means the tax is perfectly collected, with no gaps or inefficiencies. Grants, % of GDP Grants received as share of GDP UNU-WIDER 0 = the largest Government grants-receiving Revenue Dataset country as % of GDP Grants, % of Tax Grants received as share of tax revenue ibid. 0 = the largest grants-receiving country as % of tax revenue Standard deviation of Standard deviation of grants received as share of GDP UNU-WIDER 0 = a country with grants (% of GDP) during the specified period Government the largest Revenue Dataset volatility of grants and the authors' calculation Standard deviation of Standard deviation of tax revenue as share of GDP Ibid. 0 = the most tax revenue (% of GDP) during the specified period volatile country in tax revenue Standard deviation of Standard deviation of GG revenue as share of GDP WEO 0 = the most general government during the specified period volatile country in revenue (% of GDP) GG revenue Seigniorage (% of GDP) Seigniorage = (change in Money supply, M1)/(Nominal Haver Analytics and 0 = a country with GDP) the authors' the largest calculation seigniorage as % of GDP Inflation Tax (% of GDP) Inflation tax’ = (M’-M/GDP’) - (M’/GDP’ – M/P*Y’) ibid. 0 = a country with where ‘ denotes t+1, GDP is nominal, P is the GDP the largest deflator and Y’=GDP’/P’. inflation tax as % of GDP (1.a.4) EMBI spread EMBI Spread A. EMBI Spread as of the end of Dec 2022. J P Morgan 0 = a country with the largest spread 43 EMBI Spread Change B. EMBI Spread Change between the end of Dec 2022 J P Morgan 0 = a country with and the end of Dec 2021. the largest decline in the spread Foreign currency long- An annual average of foreign currency long-term A cross-country 0 = a country with term sovereign debt sovereign debt ratings by Moody’s, Standard & Poor’s, database of fiscal lowest credit ratings and Fitch Ratings. See A cross-country database of fiscal space (2022) rating. space (2022) for more details. (1.b) Technical Efficiency (1.b) Efficiency of public Quality of overall 2nd pillar: Infrastructure from "the Global The Global 0 = a country with spending infrastructure and Competitiveness Index 4.0 2019 Dataset (WEF)" Competitiveness the smallest value Logistics Performance Infrastructure score from Logistics Performance Index Index 4.0 2019 of the index Index Dataset (WEF) Logistics Performance Index (World Bank) Efficiency of Public The efficiency of public spending is calculated through Efficiency of Public 0 = the least Spending in Education, FDH, DEA, and Stochastic Frontier Analysis (SFA). The Spending in efficient country Health, and higher score is better. The scores of Infrastructure uses Education and Infrastructure public investment as the input, and quality of overall Health from infrastructure and Logistics Performance Index as the Santiago Herrera outputs. For Education, spending on education as the and Abdoulaye input, and literacy rate as the output. For health, health Ouedraogo (2018) public spending as the input, and life expectancy at and the authors' birth as the output. Input and output variables are the calculation average value during 2017-2019. otherwise (1.c) Allocative Efficiency (1.c.1) Budget Institutions PEFA scores Average of the 31 indicators of budget institution from The Public 0 = a country with PEFA 2016 Framework by pillars Expenditure and the smallest value Financial of the index Accountability (PEFA) Framework (IMF) 44 Budget transparency Budget transparency and Budget Oversight from Open The International 0 = a country with scores (Open Budget Budget Index Budget Partnership the smallest value Index) (IBP) of the index (1.c.2) Worldwide Worldwide Governance Percentile ranks for the four worldwide governance Worldwide 0 = a country with Governance Indicator Indicator indicators are used: Government Effectiveness, Governance the smallest value Regulatory Quality, Rule of Law, and Control of Indicators (WB) of the index Corruption. (1.c.3) Public Investment: Public Investment (% of General government investment (gross fixed capital IMF Investment and 0 = a country with public investment (% of total total expenditure) formation), in billions of constant 2017 international Capital Stock the smallest expenditure) and Capital dollars, divided by general government expenditure Dataset (ICSD) 2021, investment as % Scarcity and general of total government expenditure expenditure from WEO October 2022 Ratio of Public Capital General government capital stock, in billions of ICSD 0 = a country with Stock to GDP constant 2017 international dollars as share of real GDP the smallest public capital stock as % of GDP Ratio of Public Capital The above defined public capital divided by private ibid. 0 = a country with to Private Capital Stock capital stock, in billions of constant 2017 international the largest ratio of dollars as share of real GDP. public capital K/L: Ratio of Public General government capital stock divided by public capital stock from 0 = a country with Capital Stock to Public employment ICSD and public the smallest Employment employment from public capital ILOSTAT stock to public employment ratio (1.c.4) Spending for climate Fuel Subsidy (% of total Explicit subsidies divided by general government Explicit subsidies 0 = a country with change: fuel subsidies, expenditure) expenditure from Energy Subsidy the largest public expenditure on Template (IMF), Sep spending on fuel environmental protection, 2021 Version, and subsidy as % of and readiness to climate government total expenditure change. expenditure from WEO October 2022 45 Fuel Subsidy (% of GDP) Explicit subsidies as share of GDP ibid. 0 = a country with the largest spending on fuel subsidy as % of GDP Expenditure on Expenditure on environment protection as share of Expenditure on 0 = a country with Environment Protection total expenditure environment the smallest (% of total expenditure) protection from spending on IMF, Statistics environment Department. 2021. protection as % of Government total expenditure Finance Statistics (GFS) Database, and government expenditure from WEO October 2022 Expenditure on Expenditure on environment protection as share of ibid. 0 = a country with Environment Protection GDP the smallest (% of GDP) spending on environment protection as % of GDP Readiness to Climate Readiness: a country’s ability to leverage investments ND-GAIN database 0 = a country with Change and convert them to adaptation actions (University of Notre the least Dame) readiness to climate change Efficiency scores for Four scores are calculated by (1) method of efficiency Expenditure on 0 = the least readiness index vs score calculation (free disposal hull analysis and data environment efficient country environment protection envelopment analysis) by (2) input- and output- protection and GDP spending oriented measure. Input is orthogonalized expenditure per capita from IMF, on environment protection per capita controlling for and readiness GDP per capita and output is the readiness to climate indicator from change. Input and output variables are the average University of Notre value during 2016-2020. Dame 46 Efficiency scores for Three scores are calculated by (1) method of efficiency Expenditure on ibid. resilience index vs score calculation (FDH and DEA) by (2) input- and environment environment protection output-oriented measure. Input is orthogonalized protection and GDP spending expenditure on environment protection per capita per capita from IMF, controlling for GDP per capita and output is the and vulnerability resilience to climate change, defined as (1-vulnerability indicator from to Climate Change). Input and output variables are the University of Notre average value during 2016-2020. *DEA input-oriented Dame was not able to be calculated. (1.c.5) Fiscal multipliers First-year fiscal The average of three different estimates obtained from The authors' 0 = country with multiplier Comelli et al. (2023), Geli et al. (2023), and Batini (2014) calculation the lowest methods. multiplier Long-term fiscal The average over the three estimates from Bou-Habib ibid. ibid. multiplier et al. (2023), Geli et al. (2023), and the Batini method. (1.c.6) Benefit- cost ratio of Benefit- cost ratio of Reduction in poverty gap obtained for each $1 spent in ASPIRE: the atlas of 0 = country with SPL programs SPL programs SPL programs from ASPIRE indicators. See ASPIRE for social protection the lowest the details. indicators of efficiency of SPL resilience and programs equity. (2) Stabilization (2.a) Volatility: volatility of Volatility of Three variables: standard deviation of log difference of World Bank 0 = the most GDP growth, public government general government expenditure, public investment, Development volatile country expenditure, public expenditure, and general government final consumption Indicators (WDI), investment, and government expenditure. ICSD, and WEO government consumption. consumption, and public investment. Volatility of Three variables: standard deviation of residuals from WDI, ICSD, and WEO ibid. government OLS regression of the three different dependent and the authors' expenditure, variables (log difference of general government calculation government expenditure, public investment, and general consumption, and government final consumption expenditure) on real public investment GDP growth. 47 controlling for GDP growth Fatas and Mihov Three variables: Fatas and Mihov (2003)’s model of See Fatas and Mihov ibid. (2003)’s model of fiscal fiscal volatility focusing on general government (2003) or Herrera volatility expenditure, public investment, and general and Vincent (2008). government final consumption expenditure. (2.b) Rigidity of the Budget Rigid expenditure (% of Rigid expenditure is the sum of wages, interest Extended WEO April 0 = a country with total expenditure) payments, and pensions as a ratio to total expenditure. 2022 (as per the largest rigid request) expenditure (% of total expenditure) Structural expenditure Estimated rigid expenditure controlling for a set of The authors' 0 = a country with (% of total expenditure) variables based on the estimation equation "Budget estimation the largest Rigidity in Latin America and the Caribbean" (2019, structural Santiago et al.) employs. expenditure (% of rigid expenditure) Structural ratio of rigid Structural component of rigid expenditure, calculated The authors' 0 = a country with expenditure by dividing structural rigid expenditure by rigid estimation the largest expenditure deviation from structural level (2.c) Procyclicality of Procyclicality of See Herrera, Kouame, and Mandon (2019) for the Herrera, Kouame, 0 = a country with spending and fiscal impulse spending over 2011- methodologies. and Mandon (2019) the largest 2022 procyclical spending Procyclicality We estimate the coefficient b for procyclicality The authors' 0 = a country with coefficient measure from the following equation: estimation largest CAPB = a + b*output gap + c*lagged CAPB + u procyclicality in where CAPB is cyclicality-adjusted primary balance, spending output gap is estimated by applying HP filter to GDP, and lagged CAPB is CAPB t-1. 48 Fiscal Impulse average- Fiscal impulse is defined as the change in the cyclically MPO 0 = the most deviation from zero adjusted primary balance (CAPB) from the previous deviated country year. A decline in the CAPB is considered a positive fiscal impulse as a positive one should stimulate the economy. The indicator measures the absolute deviation from zero of fiscal impulse during the specified period. (3) Redistribution (3.a) Income Inequality: Difference in Gini Difference between Gini coefficients of market income OECD and CEQ (if 0 = a country with Difference in Gini before and after fiscal and of post-fiscal policy intervention. OECD data is not smallest change Coefficient between market policy intervention available) by the policy income and disposable (only latest) intervention income, and social protection system performance ASPIRE indicators of SPL Program Coverage (Poorest quintile), SPL Program ASPIRE 0 = a country with SPL system Benefit Incidence (Poorest 20%), Poverty headcount the smallest performance reduction, and Poverty gap reduction from ASPIRE coverage of the indicators. See ASPIRE for the details. poor and efficiency of SPL (3.b) Gender Inequality: Female to male wage Female to male wage ratio in the public sector (using Worldwide 0 = a country with income share by women ratio in the public sector mean) Bureaucracy the largest gap in Indicators (WWBI) pay Female employment in Deviation from female employment in the public sector ILOSTAT 0 = a country with the public sector (% of total) from female labor force (% of total) the largest gap in the female employment (3.c) Inter-generational Pension as share of General government social security expenditure as Government 0 = a country with Inequality: pension payment total spending share of general government expenditure Finance Statistics, the largest (actual and structural) IMF pension spending as % of total expenditure 49 The share of structural Estimated social security payment controlling for a set The authors' 0 = a country with components of variables based on the estimation equation "Budget estimation the largest Rigidity in Latin America and the Caribbean" (2019, structural Santiago et al.) employs. components (% of pension payment) Ratio of Pension and Social security payment per elderly person over Social security 0 = a country with Education Spending government expenditure on education per young payment from GFS the largest ratio of person and education pension to expenditure from education WDI 50 Appendix B. Correlation coefficients across the three main pillars, different samples, and different time horizons Table B1. Correlation Table Including Advanced Economies (2016-2019) Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.604 1 Redistribution 0.467 0.227 1 Table B2. Correlation Table Including Advanced Economies (2005-2019) Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.688 1 Redistribution 0.500 0.310 1 Table B3. Correlation Table Excluding Advanced Economies29 (2016-2019) Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.460 1 Redistribution 0.483 0.168 1 Table B4. Correlation Table Excluding Advanced Economies (2005-2019) Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.561 1 Redistribution 0.527 0.256 1 29 Please note that “Excluding Advanced Economies� means that the correlation coefficients are calculated excluding AEs but they are used in the aggregation process or computation of the indicators. 51 Table B5. Correlation Table Including Advanced Economies (2016-2019), with single-point estimate indicators30 excluded. Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.530 1 Redistribution -0.168 -0.219 1 Table B6. Correlation Table Including Advanced Economies (2005-2019), with single-point estimate indicators excluded. Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.589 1 Redistribution -0.137 -0.195 1 Table B7. Correlation Table Excluding Advanced Economies (2016-2019), with single-point estimate indicators excluded. Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.407 1 Redistribution -0.238 -0.363 1 Table B8. Correlation Table Excluding Advanced Economies (2005-2019), with single-point estimate indicators excluded. Long-term growth Stabilization Redistribution Long-term growth 1 Stabilization 0.478 1 Redistribution -0.262 -0.348 1 30 It means indicators that remain unchanged regardless of the years selected. 52 Appendix C. Table C. The Relationship Between Pillar III Rankings, The SWB Index , and V-Dem Indicators. (1) (2) (3) (4) (5) (6) VARIABLES Eq. 0 Eq. 1 Eq. 2 Eq. 3 Eq. 4 Eq. 5 Ladderscore 0.078*** 0.064*** 0.047*** 0.009 0.009 (0.011) (0.013) (0.015) (0.018) (0.018) FCV -0.161*** -0.087** -0.070* -0.047 -0.047 (0.030) (0.036) (0.036) (0.035) (0.036) Equal distribution of resources index 0.134** 0.012 (0.060) (0.068) Log of GDP per capita 0.076*** 0.074*** (0.018) (0.022) Constant 0.539*** 0.075 0.170** 0.182** -0.257** -0.245* (0.016) (0.065) (0.075) (0.074) (0.125) (0.144) Observations 180 134 134 134 134 134 R-squared 0.137 0.258 0.289 0.316 0.372 0.372 Adj R-squared 0.132 0.252 0.279 0.300 0.357 0.352 F-stat 28.29 45.86 26.69 19.98 25.62 19.08 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1