DISCUSSION PAPER MTI Global Practice No. 15 July 2019 Christine Richaud Arthur Galego Mendes Galego Firmin Ayivodji Samer Matta Sebastian Essl This series is produced by the Macroeconomics, Trade, and Investment (MTI) Global Practice of the World Bank. The papers in this series aim to provide a vehicle for publishing preliminary results on MTI topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For information regarding the MTI Discussion Paper Series, please contact the Editor, Ivailo Izvorski, at iizvorski@worldbank.org. © 2019 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved ii MTI DISCUSSION PAPER NO. 15 Abstract The paper updates the analysis of the fiscal policy response over the recent commodity cycle, contributes to the analysis of key drivers of fiscal policy procyclicality, and provides a stock- tacking of current fiscal vulnerabilities. Countercyclical fiscal policy during good times has been a key factor affecting the ability of commodity exporters to sustainably support economic activity when prices started declining. Fiscal space to withstand the next shock has narrowed in many EMDEs and may also be constrained by contingent liabilities stemming from exposure of state-owned enterprises and public and systemic banks to the commodity-sector. Fiscal consolidation is still necessary in many commodity-exporting EMDEs to reduce debt risks, rebuild fiscal and external buffers, and facilitate access to affordable financing. Fiscal policy should particularly aim at reducing the high volatility of public investment spending in commodity-exporting countries, both in good and bad times. Corresponding authors: crichaud@worldbank.org JEL Classification: E52, E62, H63, Q31, Q33 Keywords: EMDEs, Fiscal Policy, Contingent Liabilities, Debt Management, Commodity Prices 3 Table of Contents Abstract .......................................................................................................................................... 3 Table of Contents .......................................................................................................................... 4 List of Figures ................................................................................................................................ 5 List of Tables ................................................................................................................................. 6 List of Boxes................................................................................................................................... 7 Executive Summary ...................................................................................................................... 8 Framing the Issues: Economic and Fiscal Exposure to Commodity Price Shocks in EMDEs ......................................................................................................................................................... 1 2.1. Commodity Dependence and Price Volatility.................................................................................... 1 2.2. Commodity Shocks and the Business Cycle ...................................................................................... 5 2.3. Commodity Shocks and Fiscal Revenue Volatility ........................................................................... 7 2.4. Managing Diamond Revenue in Botswana: An Example of Success .............................................. 10 Fiscal Policy Response to the Commodity Cycle, 2000–2017 .................................................. 15 3.1. Did Commodity Exporters Take Advantage of the Last Boom to Prepare for the Downturn? ....... 15 3.2. Fiscal Policy Response to the Commodity Downturn of the Early 2010s ....................................... 19 3.3. Understanding the Fiscal Policy Response to the Recent Commodity Cycle .................................. 26 Overview of Existing Fiscal Vulnerabilities and Risks in Commodity-Exporting EMDEs . 35 4.1. The Commodity Price Outlook, 2018-2025 ..................................................................................... 36 4.2. Assessing Readiness for the Next Shock ......................................................................................... 37 4.3. Simulating the Impact of the Next Commodity Price Shock in Ghana, Angola and Kyrgyz Republic ................................................................................................................................................................ 50 Appendix 1: Overview of Commodity Exporters .................................................................... 55 Appendix 2: Structure of Commodity Exports ........................................................................ 58 Appendix 3: Commodity Booms and Busts Episodes 1960-2017 ........................................... 61 Appendix 4: Cyclicality of Fiscal Policy in Non-Oil Commodity-Exporting EMDEs : 2002- 07;2008-11 and 2015-16 .............................................................................................................. 64 Appendix 5: Estimating the Efficiency of Public Investment ................................................. 66 Appendix 6: Drivers of Procyclicality: Results of Bilateral Regressions ............................... 69 Appendix 7: Selected Indicators to Measure Fiscal Space in Commodity-Exporting EMDEs ....................................................................................................................................................... 71 References .................................................................................................................................... 72 4 List of Figures Figure 1: Commodity Price Index in Selected EMDEs. ............................................................................... 2 Figure 2: Oil Exporters are More Exposed to Volatility in Commodity Prices ............................................ 3 Figure 3: Commodity Boom Episodes, Selected Countries 1960–2016 Commodity Price Index (2010=100) ...................................................................................................................................................................... 5 Figure 4: Commodity Prices are Closely Related to Commodity Revenue… .............................................. 7 Figure 5: … and Total Fiscal Revenue ......................................................................................................... 7 Figure 6: GDP per Capita (Constant 2010 US$)......................................................................................... 11 Figure 7: Correlation Between Government Expenditures and Metal Prices in Botswana: 1990-2016 ..... 12 Figure 8: Correlation Between Government Expenditures and Metal Prices in SSA Metal Exporters: 1990- 2016 ............................................................................................................................................................ 12 Figure 9: Fiscal Aggregates (% of GDP) .................................................................................................... 13 Figure 10: Accumulated Financial Savings of GoB (%GDP) .................................................................... 13 Figure 11: Governance Indicators in SSA Countries that are Metal Exporters and Have Fiscal Rules ..... 14 Figure 12: The Business and the Commodity Cycles in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group (1990-2017) ................................................................................................... 16 Figure 13: Evolution of Institutional Quality Index by Income Group in Commodity-Exporting EMDEs, Cross-Country Average by Income Groups ................................................................................................ 17 Figure 14: Government Revenues and Expenditure in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group (1990-2017) ................................................................................................... 18 Figure 15: Debt-to-GDP Ratio in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group and Oil versus Non-Oil Exporters (1990-2017)............................................................................... 18 Figure 16: GDP Elasticity of Government Expenditure, 1996-2017 (as % Deviation from the Trend) ..... 20 Figure 17: Evolution of Cyclicality in Oil Exporting EMDEs 2002-2007 and 2015-2016 ........................ 22 Figure 18: Evolution of Cyclicality in Oil Exporting EMDEs 2008-2011 and 2015-2016 ........................ 22 Figure 19: Country Correlations Between the Cyclical Components of Real Government Expenditure and Real GDP, 2000-2017 versus 1980-1999 by Export Category (Left) and Income Group (Right) ............. 24 Figure 20: Country Correlations Between the Cyclical Components of Real Government Expenditure and: (i) Non-Commodity Real GDP (comparison with real GDP) 2002-2016 (Left); and (ii) Commodity Prices 2000-2017 versus 1980-1999 (Right) ......................................................................................................... 24 Figure 21: Country Correlations Between the Cyclical Components of Real Government Expenditure and Real GDP, Boom Episodes versus Bust Episodes. By Export Category (Left), and Income Groups (Right) .................................................................................................................................................................... 25 Figure 22: Procyclicality of Real Government Expense versus Investment Over the Business Cycle, 1980- 2017 (Correlations Between Cyclical Components). Oil and Non-Oil Exporters on the LHS, Income Groups on the RHS .................................................................................................................................................. 26 Figure 23: Commodity-Exporting Countries with Formal Fiscal Rules in Place (%) ................................ 30 Figure 24: Institutional Quality in OECD and CE’s EMDE Countries 2000-16 ........................................ 34 Figure 25: A Simple Framework for Assessing Fiscal Risks from Exposure to the Commodity-Export Sector .......................................................................................................................................................... 35 Figure 26: Commodity Prices have Stabilized Since 2016, but they Remain Generally Above their Long- Term Trends, Except for Oil Price .............................................................................................................. 36 Figure 27: Among Commodity Exporters, the Current Account Balance Worsened ................................. 38 Figure 28: … and the Debt Levels Climbed up After the Commodity Downturn ...................................... 38 Figure 29: Among commodity exporters, the fiscal accounts have deteriorated steadily since 2008 ........ 39 Figure 30: The Cyclically-Adjusted Fiscal Balance and…......................................................................... 40 5 Figure 31: … the Fiscal Sustainability Gap have Worsened After the Recent Price Downturn ................. 40 Figure 32: Concessional Debt Remains Prevalent among Low-Income Commodity Exporters, but its Share has Slightly Declined Compared to the Early 2000s, and… ....................................................................... 41 Figure 33: … the Share of Non-Paris Club Creditors in Bilateral Debt has Become Predominant for Commodity Exporters… ............................................................................................................................. 41 Figure 34: … Rollover Risks are Increasing, Particularly in Low-Income Commodity Exporting Countries, and … .......................................................................................................................................................... 41 Figure 35: … and Interest Payments on Public Debt Represent a Growing Burden for the Budget .......... 41 Figure 36: Credit to the Public Sector has Sharply Increased Since the Global Crisis ............................... 42 Figure 37: The Share of Commodity Exporters at High Risk of Debt Distress Increased Since 2013… ... 43 Figure 38: … and the Share of Commodity Importers at High Risk of Debt Distress Remained High ..... 43 Figure 39: Overview of Debt Risks in Commodity-Exporting EMDEs, 2017 ........................................... 44 Figure 40: Among Commodity Exporters, Debt is Lower in Net Terms … ............................................... 45 Figure 41: … but Depletion of Natural Resources Matters. ....................................................................... 45 Figure 42: All the Macroeconomic Indicators Worsened After the Drop in Oil Prices in 2014................. 46 Figure 43: Sonangol‘s Financial Performance Worsened Since 2014… .................................................... 47 Figure 44: … Resulting in a Higher Accumulated Debt ............................................................................. 47 Figure 45: NPLs Increased Rapidly After the Price Downturn, Especially for Oil… ................................ 48 Figure 46: … and Middle and High-Income Commodity Exporters .......................................................... 48 Figure 47: Unaccounted Residuals, Including Contingent Liabilities, Pushed Debt Up in Several Commodity-Exporting Countries................................................................................................................ 49 Figure 48: Stochastic Public Debt to GDP Paths ........................................................................................ 53 List of Tables Table 1: Price Volatility and Persistence of Selected Commodity Prices: 1960-2016 ................................. 4 Table 2: Average Standard Deviation of Key Macroeconomic Variables, 1960-2016 ................................ 6 Table 3: Cumulative Impulse Response to a One Standard Deviation Shock to the Commodity Price Index (% Deviation from the Growth Trend) – 1960-2014 .................................................................................... 6 Table 4: Standard Deviation of Government Budget Aggregates, 1960 - 2016 ........................................... 8 Table 5: Source, Mean and Standard Deviation of the Independent Variables .......................................... 31 Table 6: Country Fixed Effects Panel Regressions. All Commodity-Exporting EMDEs 2000-2016 (Dependent Variable: Cyclical Component (HP-Filter) of Log Real Government Expenditure) .............. 33 Table 7: The Commodity Price Outlook in 2018-2025 (Expected % Change in the Country-Specific Commodity Price index in Real Terms) ..................................................................................................... 37 Table 8: The Percentage of Commodity Exporters that Experienced a Deterioration in their Foreign- Currency Long-Term Sovereign Debt Ratings Increased After the Commodity Downturn ...................... 43 Table A9-1: Country Fixed Effects Panel Regressions. All Commodity-Exporting EMDEs 2000-2016 (Dependent Variable: Cyclical Component (HP-Filter) of Log Real Government Expenditure) ............... 70 Table A1.1: The Size of the Commodity Sector in Oil-Exporting EMDEs (Average 2008 – 2012).......... 55 Table A1.2. The Size of the Commodity Sector in Non-Oil-Exporting EMDEs (Average 2008 – 2012) . 56 Table A1.3: Economic Exposure to Commodity Cycles: Average 2002 - 2016. ....................................... 56 Table A2.1: Top 3 Commodity Imports for Non-Oil Commodity Exporting EMDEs (% of total imports - average 2008 - 2012) .................................................................................................................................. 58 Table A2.2: Top 3 Commodity Exports for Oil-Exporting EMDEs (% of total exports - average 2008 - 2012) ........................................................................................................................................................... 59 6 Table A2.3: Top 3 Commodity Exports for Non-Oil Commodity Exporting EMDEs (% of total exports - average 2008 - 2012) .................................................................................................................................. 59 Table A3.1: Oil exporting EMDEs ............................................................................................................. 61 Table A3.2: Non-Oil Commodity Exporting EMDEs ................................................................................ 62 Table A5.1: Investment Efficiency Levels per Commodity Exporters ....................................................... 67 Table A5.2: Relationship between Cyclicality of Fiscal Policy vs. Efficiency of Public Investment ........ 68 List of Boxes Box 1: Features of a Coherent Fiscal Framework in Resource-Rich Countries ........................................... 9 Box 2: Sovereign Wealth Funds (SWF) ..................................................................................................... 10 Box 3: A Long-Term Perspective on Procyclicality of Fiscal Policy ......................................................... 23 Box 4: Experience of Fiscal Rules in Commodity Exporting Countries .................................................... 29 Box 5: Taking into Account Financial and Natural Assets When Assessing Fiscal Space in Commodity- Exporting Countries .................................................................................................................................... 44 Box 6: Contingent Liabilities from Exposure to the Commodity Sector Weigh on Fiscal Space: The Experience of Angola ................................................................................................................................. 46 Box 7: Stochastic Projections of the Fiscal and Public Debt Positions with Commodity Price Volatility . 51 7 Fiscal Vulnerabilities in Commodity Exporting Countries and the Role of Fiscal Policy Christine Richaud, Arthur Galego Mendes Galego (Pontifical Catholic University of Rio de Janeiro), Firmin Ayivodji (Montreal University), Samer Matta, and Sebastian Essl Executive Summary With the end of the commodity boom in the early 2010s, threats to fiscal and debt sustainability have increased substantially in many commodity-exporting emerging and developing economies (EMDEs). Vulnerabilities have been fueled by widening twin deficits, increasing debt burdens and rising contingent liabilities. In this context, interest in the literature on the role of fiscal policy in addressing the challenges stemming from a country’s dependence on commodity exports has re- emerged. These well-documented challenges include commodity price volatility and unpredictability, as well as the exhaustibility of natural resources.1 Fiscal vulnerability may be broadly defined as potential failure to meet key fiscal policy objectives (Hemming and Petrie, 2000), including macroeconomic stabilization in the short-term and supporting inclusive, sustainable growth in the longer run2. In commodity exporting countries, a key goal of fiscal policy is shielding the economy from commodity price volatility through countercyclical fiscal policy, while preserving long term fiscal and debt sustainability (e.g., Villafuerte, Lopez-Murphy and Ossowski, 2010). These two objectives are consistent in the long run- because output volatility may be detrimental to long-term growth (International Monetary Fund (IMF), 2015) and thus macroeconomic sustainability. In the short run, however, fiscal vulnerabilities may arise when the government is not in a position to use fiscal policy in response to future negative price downturns because of short-term liquidity constraints, or threats to macroeconomic sustainability. Among other factors, this lack of fiscal space to withstand negative commodity busts may be the result of an inadequate response to past booms, if a government does not build the necessary buffers in good times (by saving commodity revenue stemming from taxes, royalties, and dividends or reducing public debt) to effectively respond to future downturns (IMF, 2018). Actual fiscal space also depends on the government’s fiscal exposure to the commodity sector through contingent liabilities arising from the financial position of state-owned enterprises operating in the commodity sector, as well as public and systemic banks which may be affected by a commodity downturn. While data availability represents a major constraint in a cross-country analysis, these elements may be considered in depth when assessing fiscal vulnerabilities in any given commodity-exporting country. Finally, for exporters of non-renewable natural resources, an important objective -which is not addressed in detail in this paper - includes ensuring inter- 1These include for example Baunsgaard et al., 2012; IMF, 2012, 2015; Schmidt-Hebbel, 2012; Cespedes and Velasco, 2014; Bova et al., 2016; Frankel, 2017, Izvorksi, Coulibaly and Doumbia, 2018. 2 For example, objective of fiscal policies are described as avoiding excessive fiscal deficits and debt, contributing to effective demand management, maintaining reasonable and stable tax rates (Hemming and Petrie, 2000); macro stabilization, allocation and redistribution (Arezki, 2010). 8 generational equity in the use of the commodity rent (Solow, 1986). An abundant literature analyzes how to improve fiscal management in commodity exporting countries, particularly resource-rich exporters of energy, mining and minerals. In this context, this paper focuses specifically on fiscal vulnerabilities associated to a country’s exposure to volatile commodity export prices, and aims to contribute to the analysis in three main ways: • It updates the analysis of the fiscal policy response over the recent commodity cycle (2000- 2017) in EMDEs. In particular, the paper assesses whether the commodity boom of the 2000s has been used as an opportunity to prepare for the commodity downturn of the early 2010s, and whether the shift towards less procyclical responses in the early 2010s observed in many EMDEs (Frankel et al., 2013; Carneiro et al., 2016) has been sustained in recent years. • It contributes to the analysis of key drivers of fiscal policy procyclicality in EMDEs in recent years, taking into account the persistence of commodity price shocks, which may span several years. • Finally, it provides a stock-tacking of current fiscal vulnerabilities in commodity-exporting EMDEs across all income groups, following the end of the commodity super cycle, in terms of readiness for the next price shock. In doing so, and building on abundant literature in this area, the paper proposes a simple framework to assess related risks, along the three dimensions of dependence on the commodity export sector, commodity price outlook for the country, and fiscal space. The analysis uses a country-specific commodity price index constructed for the period 1960-2025. It is worth noting that the paper does not intend to address broader aspects of macroeconomic management for commodity exporting countries, not does it constitute a handbook for fiscal management in these countries.3 Instead, it aims to review and explain the evolution of fiscal policy response to price volatility during the recent commodity cycle, and to provide a stock-taking of current vulnerabilities and risks in the face of future potential price downturns. Commodity prices are a significant determinant of the business cycle in EMDEs, compared to advanced commodity-exporting economies. Our analysis confirms that commodity price volatility remains associated to substantial volatility of GDP growth, consumption, investment and fiscal revenue in commodity-exporting EMDEs, in contrast to advanced economies. Furthermore, commodity price shocks are often persistent, which complicates the fiscal policy response to shocks given the potential trade-offs between ensuring short-term macroeconomic stabilization and long-term fiscal and debt sustainability. The commodity price downturn of the early 2010s has been accompanied by a return to increased fiscal procyclicality in most countries, as fiscal consolidation became necessary. Our analysis indicates that most commodity exporters, especially low-income countries, have historically 3A detailed discussion of these aspects is provided for example in Fiscal Management in Resource-Rich Countries, Rolando Ossowski and Håvard Halland, WBG 2016. 9 tended to run procyclical policy in response to temporary output fluctuations. Nevertheless, many commodity-exporters – particularly middle-income and high income EMDEs- managed to reduce procyclicality and rebuild buffers during the boom years of the 2000s, and this helped withstand the global crisis. This is in line with previous findings in the literature. However, we find that this trend has generally been reversed in the wake of the recent commodity price downturn as fiscal consolidation became necessary on the wake of enduring shocks. Countercyclical fiscal policy during good times has been a key factor affecting the ability of commodity exporters to sustainably support economic activity when prices start declining. We find that the few countries in the sample that were able to use fiscal policy to support aggregate demand in the wake of the recent price downturn while maintaining a low risk of debt distress are high-income countries that had run countercyclical policy during the boom (e.g., Chile). On average, however, commodity-exporters did not take full advantage of the boom years to improve their capacity to withstand future shocks, notably by increasing the quality of their institutions. High-income countries – mostly oil exporters – have achieved better results than lower-income countries in this respect. Shock persistence has been a key driver of procyclicality in 2000-2016: Enduring shocks in the second decade have triggered an adjustment to “a new normal”. Procyclicality declines sharply before the global crisis and starts rising again after 2013. We find that in the wake of the global crisis, many countries engaged in counter-cyclical fiscal policy to respond to the recessive effects of the global financial crisis. The fiscal consolidation that took place after 2012 was the response to two structural (non-transitory) shocks: the normalization of monetary policy in the US that drained liquidity in international financial markets, and the commodity downturn. This finding is consistent with the permanent income hypothesis, stating that only transitory shocks should be smoothed. Access to finance and the level of external buffers are a major determinant of countercyclicality. When controlling for shock persistence, we find that access to domestic and external financing and the level of external buffers have been a major determinant of procyclicality, which may have contributed to amplify output volatility (as financing is generally procyclical). At odds with the literature, we also find that institutional quality has not been a key direct determinant of cyclicality over the last cycle. This result may be explained by the stability of the indicator across our sample. Furthermore, institutional quality appears strongly correlated with the level of financial buffers during the period. On the other hand, fiscal rules, when formally enforced, induced higher procyclicality over the last cycle – a reason may be that, in EMDEs, fiscal rules tend to support debt sustainability, and thus supported the required adjustment to the prevalent global conditions, more than output stabilization. Fiscal vulnerabilities have increased in the wake of the recent downturn. The recent price downturn has eroded the relative gains achieved by commodity-exporting EMDEs during the boom years in terms of current account balances and debt ratios (compared to commodity importers). Furthermore, this deterioration shows several non-cyclical features which, compared to the situation observed at the end of the commodity bust of the 1990s, point to heightened vulnerabilities in terms of readiness to withstand the next shock. First, debt levels are lower 10 (following a decline in the 2000s that was largely driven, in low-income countries, by the HIPC and MDRI initiatives). However, external debt has become less concessional and the share of non- Paris Club creditors has gradually gone down. Secondly, cyclically-adjusted balances have steadily deteriorated. Furthermore, the burden of interest payments on public debt has gone up, in relation to increased reliance on non-concessional financing and domestic borrowing. Overall, risks of external debt distress have increased following the commodity price downturn, with fewer countries showing low risks. Fiscal space to withstand the next shock may also be constrained by contingent liabilities stemming from exposure of state-owned enterprises and public and systemic banks to the commodity-sector. Estimates for low-income countries show that the materialization of contingent liabilities (related or not to the commodity sector) may have driven debt accumulation by 2-5 percent of GDP or more in over half of the countries in the past decade. A comprehensive coverage of debt statistics, including contingent liabilities arising from the exposure of SOEs and public and systemic banks to the commodity export sector, would be necessary to adequately assess fiscal vulnerabilities in commodity-exporting EMDEs. This will require further analysis and data collection efforts. Fiscal consolidation is still necessary in many commodity-exporting EMDEs to reduce debt risks, rebuild fiscal and external buffers, and facilitate access to affordable financing. Using the country- specific commodity export price indexes, we find that most non-oil exporters may continue to face an enduring downturn in the near future, marked by deteriorating terms of trade and a less conducive global environment. Hence, increasing fiscal space to withstand the next shocks remains an important priority. In addition, strengthening fiscal frameworks may help facilitate the management of short-term price volatility while protecting long-term fiscal and debt sustainability. In conjunction with an adequate macroeconomic policy mix, improved debt management geared toward reducing the cost of borrowing and extending maturities of the portfolio will be important to complement fiscal consolidation and reduce debt risks. In parallel, reforms aimed at further improving government effectiveness, rule of law, regulatory quality and control of corruption, are a priority to support access to affordable financing and help withstand future shocks without further threatening debt sustainability. Fiscal policy should particularly aim at reducing the high volatility of public investment spending in commodity-exporting countries, both in good and bad times. We find that variations in public investment spending are more procyclical than current spending. If not managed efficiently, this may be a source of waste and may harm long-term growth and fiscal sustainability. In the same vein, stochastic analysis also underscores the importance for supporting debt sustainability of reforms aimed at reducing macroeconomic volatility, by creating fiscal space for countercyclical fiscal policy. Together with reforms aiming at further improving public investment management and efficiency, this is important to maintain macroeconomic stability, protect fiscal space and help withstand future downturns. 11 Framing the Issues: Economic and Fiscal Exposure to Commodity Price Shocks in EMDEs 2.1. Commodity Dependence and Price Volatility Among commodity exporters, dependence on commodity exports is particularly high for oil exporters (Appendix 1).4 On average, commodity exports account for 35 percent of total exports in non-oil exporting EMDEs that are considered moderately dependent, and 58 percent among highly dependent ones.5 Tables A1.1 and A1.2 provide a list of countries by commodity dependence. This percentage climbs to 41 percent and 78 percent respectively for oil exporters. Besides having large commodity sectors, commodity-exporting EMDEs are generally specialized in a small number of commodities. Oil and natural gas exports represents on average 67 percent of commodity exports in oil-exporting counties, and account for more than 90 percent of export revenues in Angola, Azerbaijan, Iraq, Libya and Venezuela. Non-oil exporters are more diversified but still rely on a relatively small number of products. Commodity exporters are exposed to highly volatile commodity prices. To assess exposure to price volatility at the country level, a country-specific commodity price index is constructed for 69 commodity exporters for the period 1960-2025, based on real global prices and the average structure of commodity exports per country.6 As an illustration, Figure 1 displays the index for Angola, Ethiopia, Chile and Brazil. The evolution of the index informs on a country’s specific exposure to price fluctuations. Angola is a highly dependent oil exporter and therefore the index reflects the dynamics of oil prices, with two hikes in the 1970s caused by the two oil price shocks of that decade (Figure 1). Driven by booming oil prices, the index started soaring in the early 2000s with peaks in 2007. There was a brief sharp decline in prices during 2008 and early 2009 in the wake of the global crisis, but the index rebounded quickly and reached an all-time high in 2012. With the collapse of oil prices, the index plummeted in 2015. Like Angola, Chile is a highly specialized commodity exporter and its 4 Commodity-exporters are defined following Global Economic Prospects (January 2018) as countries which, on average in 2008- 2012, met the following criteria: (i) commodities exports > 30% of total exports; or (ii) exports of any single commodity > 20 % of total exports. The full sample includes 91 countries. The size of the sample may vary according to data availability. Oil exporters are also classified following the Global Economic Prospects (January 2018): a country is classified as oil exporter when, on average in 2008–14, exports of crude oil and natural gas accounted for 20 percent or more of total exports. 5Countries that are highly dependent on the commodity export sector are defined are those in which net commodity exports account for at least 10 percent of GDP. All other commodity exporters in the sample are considered moderately dependent. 6 The index is a particular case of the more general index used in Cespedes and Velasco (2014). It is a weighted average of global commodity prices that each country exports. Each commodity is assigned weight based on its average share of exports value out of total exports in the country in 2008-2012, using trade data reported by UN Comtrade. (http://comtrade.un.org/data/). Only commodities for which price data are available in the Pink Sheet are included. Prices are mapped to these commodities by assigning commodities in the Pink Sheet with the corresponding 4-digit HS2002 classification code. The base year for prices is 2010. The set of commodities contains 49 products: [Energy] crude oil, coal, natural gas; [Beverage] cocoa, coffee, tea; [Fats and oils] copra, groundnuts, fishmeal, groundnut oil, palm oil, palm kernel oil, soybeans, soybean oil, soybean meal; [Grains] barley, maize, sorghum, rice, wheat; [Other food] Banana, orange, beef, meat chicken, meat sheep, shrimps, sugar; [Raw material] logs, sawn wood, plywood, wood pulp, tobacco, cotton, rubber; [Fertilizers] phosphate, dap, tsp, urea, potassium chloride; [Metals and Minerals] aluminum, Iron, copper, lead, tin, nickel, zinc, gold, platinum, silver. 1 index largely reflects movements in copper prices. The dynamics of Chile’s index during the past two decades is very similar to Angola’s but the latest boom episode ended earlier (peaking in 2011) and less abruptly. Brazil and Ethiopia are more diversified exporters. Hence their indices reflect the dynamics of various commodity prices (iron, oil, soybean and sugar for Brazil and coffee, gold and sheep meat for Ethiopia). For the remainder of the paper, it should be underlined that periods of booms and busts are country specific thanks to the use of the index. Figure 1: Commodity Price Index in Selected EMDEs. Angola Ethiopia 140 250 120 200 100 80 150 60 100 40 50 20 0 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 2024 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 2024 Chile Brazil 120 120 100 100 80 80 60 40 60 20 40 0 20 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 2024 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 2024 Forecast Index Linear (Forecast) Source: Authors’ calculations. Prices: Work Bank Pink Sheet (including forecasts up to 2025) (Real 2010 US dollars). Volatility of commodity prices is particularly high among oil exporters which also account for the least diversified economies.7 The magnitude of the annual volatility is remarkable (Figure 2). The standard deviation of the index annual growth rate ranges from 10 percent to 41 percent, and its cross-country average among oil and non-oil exporters reaches 27 percent and 16 percent, respectively. The least diversified oil exporters – such as Iraq, Venezuela, The Republic of Congo, Kuwait, and Azerbaijan- face a volatility that exceeds 35 percent a year. 7 An extensive literature has argued that commodity prices shown greater volatility than manufactures and are extremely volatile since Blandford (1983) and Deaton and Laroque (1992). 2 Figure 2: Oil Exporters are More Exposed to Volatility in Commodity Prices Standard Deviation of the Commodity Price Index (Average annual growth rate: 1960-2016) 45 40 35 Percent (%) 30 25 20 15 10 5 0 Malaysia Botswana Paraguay Qatar Libya Congo Canada Central African Rep Argentina Myanmar Burkina Faso CostaRica Lao Mali Togo Colombia Russia Oman Burundi Uruguay Ethiopia Norway Kuwait Guinea Indonesia Armenia Gambia Suriname Venezuela Iraq Chile Madagascar Cote d' Ivoire Kazakhstan Arzeibajan Kenya New Zealand Oil Exporter Non-Oil Exporter Source: Authors’ calculations. Combined with short-term price volatility, commodity exporters also need to manage long-lasting price shocks, with episodes of commodity booms and busts spanning several years, especially for oil exporters8. For example, using monthly data over the period 1957 and 1998, Cashin, McDermott, and Scott (1999) find that the half-life of shocks to coffee, cotton, nickel, sugar and rice range between 9 to 18 years. When commodity prices are highly persistent, a price shock is expected to last for several periods. This high persistence of global commodity prices is reflected in the country-level commodity price index. The average half-life is 10 and 5.2 years among oil and non-oil exporters. However, the variability of the half-life is large across countries. While highly dependent oil exporters face significant shock persistence, more diversified economies like Costa Rica, Rwanda and Kenya face fairly transitory price shocks. 8 Cf. Appendix 3 for the definition of episodes of commodity booms and busts. 3 Table 1: Price Volatility and Persistence of Selected Commodity Prices: 1960-20169 Volatility Persistence* Persistence** (SD of annual % (Half-life in <1 1-4 5-8 9-18 growth rate) years) Oil 37.9 11.2 . . . . X Beef 12.2 6.6 . . X . . Natural Gas 24.2 6 . . . . X Cooper 22.6 6 . . X . . Gold 17.8 6 . . . . X Coffee (Robusta) - 5 . . . . X Soy bean 16.6 3 . X . . . Coffee (Arabica) 30.5 2.6 X . . . . Sugar 48.4 2.3 . . . X . Source: Authors’ calculations from World Bank’s Pink sheet. Persistence from Borensztein, Jeanne, and Sandri (2013) and Cashin, McDermott, and Scott. (1999). * Borensztein et al (2002) ** Cashin et al (1999) Because enduring shocks require a different policy response, analyzing the fiscal policy response requires therefore going beyond short-term fluctuations, and identifying booms and busts episodes, (Table 1).10 All oil exporting EMDEs experienced two commodity booms since 1960: one starting in the early 1970s that lasted until the mid-1980s, and another running from the early 2000s to 2014. Non-oil exporters present more dispersion across the dates and duration of commodity booms, but importantly, despite the downturn experienced after a 2010-2011, prices are generally still above their long-term trends (Figure 3) – which suggest that the outlook may not be as favorable. 9The half-life is calculated as ln(0.5)/ln(rho), where rho is the autoregressive coefficient of an AR(1) regression on simulated commodity price data. Time-series of each commodity price were simulated using the persistence estimated in Cashin, McDermottand, and Scott (1999) and Borensztein and Jeanne (2012). 10Based on Cespedes and Velasco (2014), we define the start of a commodity boom when the commodity price index reaches a level of at least 25 percent above its trend and the end of a commodity boom when the index comes back to a level lower than 10 percent above its trend. The trend is calculated as centered moving average with a 25-years window. Adler and Magud (2013) define a terms-of-trade episode as an event where the terms of trade increase at least 15 percent from start to peak and the annual average increase in terms of trade is of at least three percent. The start (peak) date is identified as a local min (max). Comparing with Adler and Magud 2013, the start dates of the boom episodes are similar. However, the duration of the boom tends to be higher under our definition. 4 Figure 3: Commodity Boom Episodes, Selected Countries 1960–2016 Commodity Price Index (2010=100) Kuwait Malaysia 150 150 Commodity Price Index 100 100 50 50 0 0 1960 1970 1980 1990 2000 2010 2020 1960 1970 1980 1990 2000 2010 2020 Peru South Africa 50 100 150 150 Commodity Price Index 100 50 0 1960 1970 1980 1990 2000 2010 2020 0 Booms Index Trend 1960 1970 1980 1990 2000 2010 2020 Source: Authors’ calculations based on the UN-ComTrade dataset. 2.2. Commodity Shocks and the Business Cycle Commodity price fluctuations are an important determinant of the business cycle in commodity- exporting EMDEs,11 in contrast to advanced commodity-exporting economies, and are associated with high investment and consumption volatility. Volatility in commodity prices can have harmful implications for long-term growth and overall welfare, especially in economies with poorly developed financial systems. Over 1960-2016, volatility of commodity prices has been passed on to output, consumption, and investment (Table 2). In the sample, the average standard deviation of GDP growth in highly dependent oil exporters is particularly large, 8.1 percent a year, i.e., four times larger than the average observed in commodity-exporting advanced economies. Moreover, the volatility of investment growth is very high in EMDEs, especially when compared to the advanced economies. In highly dependent economies, both oil and non-oil exporters, it ranges between 27 and 34 percent per year.12 In this context, Drechsel and Tenreyro (2017) find that commodity prices can explain 38 percent of the variance of GDP growth in the post-1950 data in 11 Cf. Mendoza 1995; Kose 2002; Cespedes and Velasco 2014. 12 The literature has documented various channels through which commodity price fluctuations can trigger vigorous growth episodes as well as disruptive economic cycles. Many studies have focused on the financial aspects of commodity cycles. Drechsel and Tenreyro (2017) argue that commodity fluctuations affect both the competitiveness of the economy and its borrowing terms in global financial markets. This results in strong positive effects of commodity price increases on GDP, consumption, and investment, but a negative effect on the trade balance - which may be a consequence of overspending during commodity booms in EMDEs that is arguably boosted by pro-cyclical fiscal policy and counter-cyclical cost of borrowing. 5 Argentina. Fernandez et al. (2017) find that commodity and interest rate shocks explain on average 33 percent of output fluctuations on a panel of 138 countries over the period 1960-2015. Table 2: Average Standard Deviation of Key Macroeconomic Variables, 1960-2016 Trade Commodity GDP Consumption Investment Balance price index (% growth) (% growth) (% growth) (% GDP) (% growth) Oil exporters Moderate dependency 4.5 6.4 15.3 6.4 20.2 High dependency 8.1 13.4 27.1 14.8 28.8 Non-oil exporters Moderate dependency 5.1 8.4 22.7 8.5 16.1 High dependency 4.9 9.1 39.4 6.4 17.8 Developed CEs 2.2 2.0 6.4 3.8 15.6 Source: Authors’ calculations based on WEO-IMF (default) and World Bank WDI (when WEO not available). GDP, consumption and investment are expressed in constant local currency. Developed commodity exporters are Australia, Canada, New Zealand and Norway. A shock to the commodity price index induces a significant cumulated change in the GDP growth rate (Table 3). The estimated cumulated deviation from the growth trend induced by a one standard deviation price shock over 1960-2014 ranges between 0.34 and 1.25 percentage point. Not surprisingly, price shocks induce higher volatility in GDP, consumption and investment growth in countries that are more commodity-dependent. Moreover, commodity price shocks appear to drive particularly high investment volatility in EMDEs, for both oil and non-oil commodity exporters. This may be associated with the impact of price shocks on public investment and foreign direct investment in the natural resource sector. Advanced commodity exporters, on the contrary, have faced significantly lower output and investment volatility over the same period, despite similarly volatile commodity prices. Table 3: Cumulative Impulse Response to a One Standard Deviation Shock to the Commodity Price Index (% Deviation from the Growth Trend) – 1960-201413 Source: IMF’ WEO as default, WDI and Authors’ calculations 13 Developed CEs include Australia, Canada, New Zealand and Norway. 6 2.3. Commodity Shocks and Fiscal Revenue Volatility In EMDEs, country-specific commodity price indexes are highly correlated with total fiscal revenues14 across the sample, which suggests that prices are a key driver of the amount of all revenues collected by the average government, not only from the commodity sector, but also from economic activity and external aid in general (Figure 4 and Figure 5).This underscores that commodity prices remain on average a major determinant of the resources that are available to the government for implementing its fiscal policy objectives. Figure 4: Commodity Prices are Closely Figure 5: … and Total Fiscal Revenue Related to Commodity Revenue… Source: Authors’ calculations, COMTRADE, WEO. The period is 1990-2016. In this context, exposure to volatile commodity prices imposes substantial volatility on fiscal revenues compared to advanced economies. This is partly due to greater volatility of the business cycle in EMDEs. The observed volatility in consumption induced by price shocks may play an important role, because of the predominance of consumption taxes in the tax base of EMDEs. Overall, economies that are highly dependent on the commodity sector tend to face substantially more revenue volatility (Table 4), which underscores the importance of enhancing economic diversification over time to smooth revenue. In turn, commodity exporting EMDEs present much higher volatility in government expenditures over 1960-2016 than more advanced economies. 14 Based on a sub-sample of countries where data on commodity revenue is available (WEO). 7 Table 4: Standard Deviation of Government Budget Aggregates, 1960 - 2016 Revenues (% growth) Expenditure (% growth) Oil exporters Moderate dependency 9.7 21.5 High dependency 16.4 22.2 Non-oil exporters Moderate dependency 11.3 18.2 High dependency 18.7 26.7 Developed CE's 3.3 3.3 Source: Authors’ calculations based on WEO-IMF (default) and World Bank WDI (when WEO not available). All variables are expressed in constant local currency. Developed commodity exporters are Australia, Canada, New Zealand and Norway. Fiscal revenue volatility and unpredictability may hinder the implementation of fiscal policy and be a source of vulnerability if not adequately managed. A key objective of fiscal policy in commodity exporting countries is to smooth the impact of short-term commodity price volatility on output through countercyclical fiscal policy aimed at stabilizing aggregate demand, while maintaining long-term fiscal sustainability in the context of persistent and unpredictable shocks (Villafuerte, Lopez Murphy, and Ossowski 2010, IMF 2012a, IMF 2015)). This is particularly important for EMDEs which may not have access to effective hedging mechanisms (Frankel, 2017). Highly volatile fiscal revenue may represent an important constraint to achieving these objectives if the government does not have access to the necessary institutional and financial resources to smooth revenue volatility over the cycle and manage fiscal and debt sustainability. Shock persistence complicates the fiscal policy response to price fluctuations because the persistent component of a price shock is not easily predictable and almost impossible to disentangle from its cyclical component. Hence, fiscal vulnerability may also arise from difficulties to anticipate and withstand enduring periods of booms or busts. Striking the right balance between smoothing short-term volatility and adjusting to longer positive or negative shocks may therefore be a challenge. For example, Mendes and Pennings (2017) conclude that in presence of enduring negative shocks, counter-cyclical fiscal policy based on external borrowing may generate adverse sustainability effects and volatile cost of financing. Since prices are highly unpredictable, policy recommendations are generally tilted toward cautiousness (IMF, 2012a). 8 Box 1: Features of a Coherent Fiscal Framework in Resource-Rich Countries A coherent fiscal framework for a resource-exporting economy needs to ensure that it: • Promotes short-term macroeconomic stability by delinking fiscal policy from the volatility of oil prices; • Helps to achieve long-term fiscal sustainability by generating sufficient savings for future generations (when the natural resources will be exhausted); • Allocates sufficient resources to meet development needs and promotes the structural transformation and diversification of the economy. The fiscal framework should be internally consistent, with clear guidance and communication on the course of fiscal policy over the short- and medium-term and underpinned by strong institutional arrangements to bolster policy credibility and investor confidence. Such institutional arrangements typically include fiscal rules, sovereign wealth funds, fiscal responsibility laws, and fiscal advisory councils. Norway and Chile stand out among the countries with the most successful fiscal frameworks and institutions to manage natural resource wealth. Three critical choices have to be made when designing a fiscal policy framework. • A fiscal anchor (or indicator) around which fiscal rules or guidelines will be framed; this may include: (i) The non-resource fiscal balance, defined as the difference between public expenditure and non-resource fiscal revenue; and (ii) The structural fiscal balance, the result of a budget formulated using a smooth, benchmarked commodity price over time. Such price-based rules are typically coupled with targets for the overall fiscal balance. • An appropriate target for the fiscal anchor; (i) If the non-resource deficit is the anchor of the rule that underpins the fiscal framework, the target should be set using a conservative estimate of annual resource revenue over a long-term horizon. Under such a rule, the targeted nonoil deficit is financed by an annual transfer from a sovereign wealth fund (SWF). (ii) Alternatively, the target for the non-resource deficit could be set at a level equal to the annual income stream derived from the accumulated financial assets of an SWF. This is a “bird-in-hand” approach which is considered very conservative. (iii) When price-based rules are used, commodity reference prices for the structural fiscal balance can be calculated on the basis of various formulas. Price formulas can be a moving average of either past prices or of past spot and futures (market) prices. A structural balance rule can also include an adjustment of non-resource revenue to remove the impact of the economic cycle, as in Chile. However, if the economic cycle is not well defined, this additional adjustment may introduce unnecessary complexity into the fiscal framework. Moreover, the target for the structural budget balance should more appropriately be formulated in terms of the primary balance, because it imposes an effective constraint on borrowing needs and thus on the dynamics of public debt. • The desired degree of flexibility for achieving the target for the fiscal anchor when unexpected economic developments occur. In complement, fiscal councils are non-partisan agencies with a mandate to assess fiscal policies, plans, rules, and performance. Such a council is particularly helpful if the fiscal rule involves the calculation of either long-term commodity price benchmarks (as in the price-based rule for the structural fiscal balance) or a Permanent Income Hypothesis-type of annuity (as in the non-resource deficit rule), as it helps ensure that such calculation benefits from an independent oversight. Source: World Bank (2017b) “Kazakhstan: Enhancing the Fiscal Framework to Support Economic Transformation,” Public Finance Review, The World Bank Group, August 2017. 9 Commodity exporters reliant on exhaustible natural resources (energy, mining and metals) also face the challenge of ensuring intergenerational equity through adequate mechanisms for saving and investing the resource rent for future generations (Hartwick 1977; Khadour 2011). Complex fiscal policy and governance issues may therefore arise from potential trade-offs between supporting short-term macroeconomic stabilization, long-term fiscal and debt sustainability, and intergenerational equity (IMF 2012a, Go et. Al., 2013). This is an important challenge in countries with relatively weak fiscal frameworks and fiscal management capacity (IMF, 2012b). Literature on these issues is abundant, and this section only aims to briefly summarize and illustrate key findings. Box 1 highlights important elements of an adequate fiscal framework in a resource-rich country. As a complement, a number of commodity-exporting countries have set up resource funds to support their policy objectives (Box 2). The experience of Botswana (Section 2.4) provides insights on how these challenges can successfully be managed.15 Box 2: Sovereign Wealth Funds (SWF) Resource funds may complement policy tools to deal with resource revenue management. Sovereign Wealth Funds include: Stabilization funds which are set up to insulate the budget and economy from commodity price volatility. Examples are the Economic and Social Stabilization Fund in Chile and the Oil Stabilization Fund in Russia. Typically, the inflows and outflows are contingent on whether revenues are deemed “high” or “low” compared to a benchmark level. The strategic asset allocation tends to be relatively conservative and favors liquid asset classes. Savings funds which are set up to ensure inter-generational equity in countries producing exhaustible natural resources, by allowing future generations to benefit from a windfall stemming from the current generation’s exploitation of natural resources. Saving funds include for example the State Oil Fund in Azerbaijan, the Petroleum Fund in Timor- Leste and the Pula Fund in Botswana. These funds typically have fixed inflows and discretionary outflows and reflect a higher tolerance for short-term volatility and a focus on long-term returns. The objectives of Sovereign Wealth Funds depend on country-specific circumstances, which may evolve over time. In practice, many funds have multiple objectives, such as both stabilization and savings for the future (Azerbaijan, Timor-Leste, and Trinidad & Tobago). A well-designed SWF can help support the successful implementation of fiscal policy, but it cannot serve as substitute for a sound fiscal framework (Box 1). SWFs should also be integrated into the budget to ensure its integrity and protect its role as the mechanism to set expenditure priorities and efficiently allocate public financial resources. 2.4. Managing Diamond Revenue in Botswana: An Example of Success Botswana is one of the few commodity exporters that successfully managed its natural resources in support of macroeconomic stability and growth, thus it could offer important insights for other countries.16 At independence in 1966, Botswana was one of least developed countries in the world with a real GDP per capita of only US$ 83.7. Over the years, it has benefited from the onset of diamond production and succeeded in outperforming richer and resource-rich African countries (Figure 6). Nevertheless, in the literature, Botswana is often referred as the only African country (alongside Mauritius) that managed its resource windfall properly (Arezki et al., 2011; Gylfason 15While it is an important fiscal policy objective, the issue of intergenerational equity is not discussed in detail in this paper, which focuses on fiscal vulnerabilities associated to commodity price fluctuations. 16Another example, would have been to look at the case of Chile. However, Chile had a much higher GDP of about US$ 4,500 in 1966. 10 2001).17 While Botswana is still dependent on diamonds, the economy has significantly diversified over the years. The share of diamonds out of real GDP dropped from about 60 percent in the early eighties to less than 16 percent in 2013. Similarly, the share of diamond revenues out of total fiscal revenues decreased from 60 percent in the late eighties to about 30 percent in 2013 (World Bank 2015). Figure 6: GDP per Capita (Constant 2010 US$) 8000 7000 6000 5000 4000 3000 2000 1000 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Botswana Sierra Leone Nigeria South Africa Source: WDI. Botswana’s economic success was partly due to a sustainable and pro-growth fiscal policy which linked non-productive spending to non-diamond revenues, while directing diamond revenues towards productive expenditures. Botswana is one of the few resource-rich countries in the world, together with Norway, to have successfully applied the Hartwick rule (Hartwick 1977; Solow 1986) which states that resource revenues must be invested in financial, human or capital assets. The country introduced the Hartwick rule in its fiscal policy by creating the Sustainable Budget Index (SBI) which denotes the ratio of recurrent spending to recurrent revenues: = In the above equation applied to Botswana, recurrent (or non-development) spending excludes productive expenditures such as spending on health, education and infrastructure,18 while recurrent revenues represent non-mining revenues. In essence, Botswana’s fiscal rule was to maintain an SBI of 1 or less which indicates that current government consumption is being financed entirely by diamond revenues, and thus the fiscal position is sustainable over the long-run (World Bank 2010). Except for the 2001-2004 period, when the Southern African Customs Union (SACU) revenues dropped significantly, the SBI was constantly lower than 1 suggesting that Botswana’s 17According to Gylfason (2001), Botswana, Indonesia, Malaysia and Thailand were the only four resource-based countries that could be considered success stories in terms of economic growth at the time. Arezki, Gylfason, and Sy (2011) also identified Chile and Mauritius as other successful resource-rich countries. 18 Non-recurrent spending “includes not only the capital budget (referred to in the public finance data as ‘d evelopment expenditure’), but also that portion of the recurrent budget used for education and health, interpreted as investment in huma n capital” (Lange and Wright 2004). 11 fiscal policy was sound and sustainable. In fact, Botswana was the only metal exporter country in SSA that conducted a countercyclical fiscal policy in relation to metal prices (Figure 7 and Figure 8).19 Figure 7: Correlation Between Figure 8: Correlation Between Government Expenditures and Metal Government Expenditures and Metal Prices in Botswana: 1990-2016 Prices in SSA Metal Exporters: 1990-2016 y = -5.8x + 59.2 y = 9.4341x + 4.7236 50 30 Gov. expenditure (% GDP) Gov. expenditure (% GDP) 28 26 45 24 22 40 20 18 16 35 14 12 30 10 1.5 1.6 1.7 1.8 1.9 2.0 2.1 1.5 1.7 1.9 2.1 Log Metal Price Index Log Metal Price Index Source: Authors’ calculations. Source: Authors’ calculations. The government’s strategy to allocate most of its diamond revenues to physical and human capital was coupled with an implicit rule to save any remaining fiscal surpluses into financial assets. In addition to satisfying the SBI rule, Botswana enjoyed 24 years of fiscal surpluses between 1980 and 2015, particularly during the eighties and nineties (Figure 9). These surpluses were accumulated as savings into the Government Investment Account (GIA) which became part of the Pula Fund created by the Central Bank of Botswana (BoB) in the mid-nineties (Figure 10).20 The Pula fund, which was tasked to invest its assets in long-term instruments overseas to achieve higher returns, served as a buffer against short-run declines in diamond prices revenues as well as long- run reductions in diamond depletion. For instance, the Pula fund served as a revenue and output stabilization fund amidst the 2007-08 global financial crisis. During that time, the government ran large fiscal deficits (9.4 percent of GDP on average between 2008 and 2010) as mining revenues dropped and expenditures soared due to an increase in infrastructure-related spending aimed at offsetting the adverse effects of the global economic downturn and boosting long-term productivity. As a result, the government financed this deficit by drawing upon savings from the Pula fund and issuing new debt (World Bank 2016). 19Central African Republic, Guinea, Liberia, Mauritania, Mozambique, Niger, Rwanda, South Africa, Zambia and Zimbabwe. Diamonds are considered as metals. 20The Pula Fund, is composed of the Government Investment Account (GIA), which reflects savings from accumulated fiscal surpluses, and the BoB’s reserve accumulation above the target for liquid reserves. Under the Bank of Botswana Act (1996), the Pula finds invest all its assets in long-term instruments abroad. However, the legal framework of the Pula fund does not specify how the generated returns relate to the overall fiscal policy. 12 Figure 9: Fiscal Aggregates (% of GDP) Figure 10: Accumulated Financial Savings of GoB (%GDP) 75 120 65 100 55 45 80 35 60 25 40 15 5 20 -5 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 -15 Fiscal balance Revenues Expenditures Source: World Bank (2010) and IMF WEO. Source: World Bank 2016. The effectiveness of Botswana’s fiscal rule stems from the insulation of recurrent spending from short-run fluctuations in the price of diamonds and strong political commitment. In contrast to Chile and Norway (two successful commodity exporters) where fiscal rules had a clear statutory basis, the rule of maintaining an SBI below 1 was not legally binding in Botswana. Thus, the reason for Botswana’s success in adhering and implementing its fiscal rule was the genuine political will to satisfy the SBI threshold. In fact, the government of Botswana has continuously measured and referred to the SBI in its internal reports to assess the economic performance, and most importantly to ensure budget sustainability (Lange and Wright 2004; World Bank 2016). This fiscal discipline would not have been possible without good governance and strong institutions. Botswana outperforms most SSA countries on the various measures of institutional quality (Figure 11) for three main reasons. First, strong institutions are related to the pre-colonial Tswana culture. This culture played a key role part in building effective institutions, because it developed an inclusive political environment. Tribal chiefs were held accountable to the people, who were allowed to speak during meetings and voice their opinions. According to a famous Tswani proverb: “the king is king by the grace of the people” (Gulbrandsen, 1995, p.1). Second, the country was largely free of kleptocracy and armed conflicts because the Tswana groups controlled the nation and influenced the institutional set-up of the country without conflict (Robinson, 2009). Third, the post-independence leadership focused on building capacity in the public sector and asserted the state’s right over mineral resources (Acemoglu, Johnson, and Robinson 2001; Lewin 2011). In addition, elites who were part of the leadership had vested interests in ranching, Botswana’s most important sector during the 60s (Robinson, 2009). As a result, they invested heavily in this industry, thus boosting growth in the early stages of development. The example of Botswana illustrates the features of sound management of natural resources in a context of volatile (and exhaustible) fiscal revenues. 13 Figure 11: Governance Indicators in SSA Countries that are Metal Exporters and Have Fiscal Rules Control of Corruption Government Effectiveness 2 2 1 1 0 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 -1 -1 -2 -2 Botswana Angola Mauritius Botswana Angola Mauritius Nigeria South Africa Senegal Nigeria South Africa Senegal Cote d'Ivoire Liberia Cote d'Ivoire Liberia Regulatory Quality Rule of Law 2 2 1 1 0 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 -1 -1 -2 -2 Botswana Angola Mauritius Botswana Angola Mauritius Nigeria South Africa Senegal Nigeria South Africa Senegal Cote d'Ivoire Liberia Cote d'Ivoire Liberia Source: World governance indicators. Note: The score for each indicator ranges from -2.5 (worst) to 2.5 (best). 14 Fiscal Policy Response to the Commodity Cycle, 2000–2017 The third section of the paper focuses on analyzing the fiscal policy response to the commodity cycle over the last two decades. While exact dates of booms and busts vary across countries, these decades have overall been marked by an enduring price boom of the 2000s, followed by a price downturn in the early 2010s. 3.1. Did Commodity Exporters Take Advantage of the Last Boom to Prepare for the Downturn? Reducing fiscal procyclicality in good times to maintain macroeconomic stability and rebuild fiscal space will support sustainable countercyclical responses to commodity downturns. Procyclical fiscal policy during a boom may contribute to well-documented Dutch Disease phenomena, including inflation and real exchange rate appreciation, and this may in turn affect long-term growth and macroeconomic sustainability. Furthermore, this weighs on a country’s ability to borrow in bad times, compounding the lack of sufficient fiscal buffers when prices start declining. A key constraint to using countercyclical fiscal policy in response to a commodity price shock may also be the political space to do so (Arezki and Brückner 2012).21 Hence to what extent did commodity exporting EMDEs managed the commodity boom of the 2000s to prepare for the last shock? The last commodity cycle was characterized by robust economic growth in the 2000s fueled by booming commodity prices. The average real GDP growth increased substantially during the first part of the 2000s until the global crisis hit in 2008, reducing the pace of economic expansion (Figure 12). Low-income countries benefited relatively less from the booming global economy and, consequently, were less exposed to the effects of the global crisis. Average annual GDP growth rates hiked from 2 percent in the 1990s to 6 percent in 2000-07 in high-income, upper- middle income, and lower-middle income commodity-exporting EMDEs. Upper-middle-income countries were the most affected by the slowdown of the global economic activity in 2008 onwards, growing only 2.2 percent on average in 2000-16. For commodity exporting EMDEs, an important channel driving the economic downturn after 2007 was a temporary decline in commodity prices in the wake of the global crisis (Figure 12). In 2009, high-income – mostly oil - exporters, recorded a decline in commodity price index as sharp as 40 percent. However, this decline was short-lived. Commodity prices recovered in 2010 and growth in commodity-exporting EMDEs rebounded, while most advanced economies were experiencing a sluggish recovery from the crisis. It is only in the early 2010s that commodity-exporters were hit by the sharp drop in commodity prices – with non-oil commodity prices declining broadly after 2011, and oil prices plummeting after 2014.22 21 These elements, which are considered important drivers of vulnerability, are explored in more depth in Section 4 of this paper. 22 As the paper uses country-specific commodity price indexes, the exact date of the boom and downturn periods vary across countries. Overall, however, the years 2010-2011 can be considered as the last years of the boom period across the whole sample and are therefore used accordingly in the analysis. 15 Figure 12: The Business and the Commodity Cycles in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group (1990-2017) Source: Authors’ calculations based on IMF WEO data. Notes: the dark line refers to the average growth rates over the periods 1990-99, 2000-07, 2008-2011, 2012-2017. On average, the boom period was not accompanied by significant improvements in the quality of institutions, except in high-income countries. Following Frankel, Vegh, and Vuletin 2013, a country-specific institutional quality index (IQ) is computed23 (Figure 13). The evolution over 2000-2017 shows only modest improvements during the boom episode, except for high-income commodity exporters. These evolutions may have constrained their ability to prepare for and withstand the shock when commodity prices fell. 23 The variables used to measure institutional quality include government effectiveness, rule of law, regulatory quality and control of corruption, all provided by the International Country Risk Guide. Following Frankel, Vegh, and Vuletin 2013, a country-specific institutional quality index (IQ) is computed as the average of these four variables. The index is normalized to range between 0 (lower quality) and 1 (highest quality). 16 Figure 13: Evolution of Institutional Quality Index by Income Group in Commodity- Exporting EMDEs, Cross-Country Average by Income Groups Low Income Lower-Middle Income 0.45 0.445 0.44 0.44 0.43 0.435 0.42 0.43 0.41 0.425 0.4 0.42 0.39 0.415 0.38 0.41 2000 2002 2004 2006 2008 2010 2012 2014 2016 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Upper-Middle Income High Income 0.5 0.68 0.49 0.66 0.48 0.47 0.64 0.46 0.62 0.45 0.6 0.44 0.43 0.58 2000 2002 2004 2006 2008 2010 2012 2014 2016 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Source: Authors’ calculation, based on the International Political Risk database. Note: The sample of countries include only emerging markets and developing economies (EMDEs) as per the World Bank classification. Nevertheless, during the boom, commodity-exporting EMDEs managed on average to build up reserves and reduce public debt, and this helped them withstand the global crisis. As shown in Figure 14, the annual growth rate of real government revenues rose from around 5 percent during the 1990s to 8 percent in 2000-07. That expansion was fueled by fast GDP growth and the boom in commodity revenues. During that period, real government expenditure grew slower than revenues: high and middle-income countries kept annual expenditure growth rate below 8 percent, accumulating surpluses and reducing fiscal deficits. Improved fiscal balances, coupled with the benefits of the HIPC and MDRI initiatives for low-income countries, helped rebuild buffers (foreign exchange reserves and sovereign wealth funds) and reduce public debt. Commodity exporters reduced significantly their debt-to-GDP ratios across all income groups (Figure 15). In 2008, the cross-country average public debt represented only 35 percent of GDP. In relatively good financial shape, governments stood ready to pursue counter-cyclical spending at the onset of the financial crisis in 2008, and despite the drastic drop in fiscal revenues and accelerating expenditure, public debt remained contained at the turn of the 2010s. 17 Figure 14: Government Revenues and Expenditure in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group (1990-2017) 24 Source: Authors’ calculations based on IMF WEO data. Notes: the dark line refers to the average growth rates over the periods 1990-99, 2000-07, 2008-2011, 2012-2017. Figure 15: Debt-to-GDP Ratio in Commodity-Exporting EMDEs, Cross-Country Averages by Income Group and Oil versus Non-Oil Exporters (1990-2017) Source: Authors’ calculations based on IMF WEO data. Notes: the dark line refers to the average debt to GDP ratios over the periods 1990-99, 2000-07, 2008-2011, 2012-2017. 24The black full line in Figures 4.1, 4.2 and 4.3 refer to the indicator averages over the periods 90-1999, 2000-07, 2008-11, and 2012-2017. 18 3.2. Fiscal Policy Response to the Commodity Downturn of the Early 2010s How has fiscal policy response evolved in the wake of the commodity price downturn? To understand more precisely the evolution of the fiscal policy response to the recent shock, a country- specific, time-varying measure of fiscal cyclicality is computed using the Local Gaussian Weighted Least Squares method (LGWLS). This method has been used by Aghion et al. (2007), Guerguil et al. (2017) and Keita and Turcu (2018) and allows the government reaction to the business cycle to fluctuate over time and differ between good and bad times.25 Despite gains achieved earlier, fiscal policy became more procyclical in the wake of the commodity downturn, signaling the need to adjust to the enduring negative shock. As underlined in 3.1, there was a clear downward trend in the cross-country average of fiscal procyclicality prior to 2008 for high and middle-income countries (Figure 16), which is in line with the literature that presents evidence of reduced fiscal procyclicality in a number of countries during the 2000s (e.g. Frankel, Vegh, and Vuletin 2013, Cespedes and Velasco 2014, Carneiro et al., 2016). In 2008-11, many countries engaged in counter-cyclical fiscal policy to respond to the recessive effects of the global financial crisis. During this period, further reduction in the measure of fiscal cyclicality was observed, especially in lower-middle and low-income countries – suggesting reduced procyclicality or increased countercyclicality. Accommodative monetary policy in the advanced economies increased liquidity in international financial markets, and, to some extent, these funds were channeled to EMDEs to finance counter-cyclical government expenditure. However, after 2011, except from high-income countries (i.e., mostly oil exporting EMDEs), fiscal policy has become more procyclical: the GDP elasticity of government expenditure has increased drastically from 2011 onwards, especially in -middle income countries. The combination of the global crisis and the commodity downturn weighed on global liquidity and strained external and fiscal buffers, and many commodity-exporting EMDEs had to undertake fiscal consolidation to adjust to these enduring shocks.26 25 We estimate the following equation: Where ̂, correspond to the cyclical components of government expenditure and GDP in period in country , ̂, and respectively, expressed as percent deviation from the trend. The method uses all the observations for each year and the closest observations to the year considered are given a greater weight. To compute each , , the method weights all observations prior to t by a half-Gaussian distribution centered at t. We follow Aghion et al. (2007), Guerguil et al. (2017) and Keita and Turcu (2018) and use a value of equal to 5. 26 Van Doorn R., V. Suri, and S. Gooptu. 2010, find that a number of middle-income countries, in the aftermath of the global crisis, needed to undertake severe fiscal adjustment or face a more favorable economic outlook than forecaseted at the time. 19 Figure 16: GDP Elasticity of Government Expenditure, 1996-2017 (as % Deviation from the Trend) Source: Authors’ calculations based on IMF WEO data. Notes: the dark line refers to the average growth rates over the periods 1990-99, 2000-07, 2008-2011, 2012- 2017. However, this evolution masks significant differences across countries over the cycle.27 Among oil exporters, many countries ran countercyclical or (mostly) a-cyclical policies during the boom years of 2002-2007 (Figure 17). Angola, Cameroon, Iran, Venezuela and Yemen were among the few countries where fiscal policy remained procyclical, with positive changes in the output gap triggering overspending. In the aftermath of the global crisis (2008-2011), large fluctuations across oil-exporting countries can be observed in terms of the fiscal policy response to the negative output shock (Figure 18). While some countries adjusted through spending restraint amplifying the shock, others undertook countercyclical fiscal policy in support of aggregate demand (e.g., Algeria, Bahrain, Saudi Arabia, Republic of Congo, Malaysia and UAE). When oil prices fell, many oil exporters had to adjust as vulnerabilities emerged (yellow dots): fiscal policy in 2015- 2016 became more decisively procyclical. Nevertheless, oil exporters that had run countercyclical fiscal policy during the boom years were able to maintain a countercyclical or a-cyclical policy stance (although with some degree of spending restraint) in the aftermath of the commodity shock, except Nigeria and the Republic of Congo where debt risks were increasing. The Republic of Congo, in particular, also shows a strong countercyclical response to the global crisis, but this was 27Izvorksi, Coulibaly and Doumbia (2018) find that most resource rich African countries did not take advantage of the commodity boom, and boosted procyclical spending. 20 not sustainable. It threatened long-term sustainability and triggered a need for fiscal consolidation when oil prices fell. Graphs for non-oil exporters are displayed in Appendix 4. Nigeria exemplifies the adverse economic effects of not building fiscal buffers during good times. In response to the drop in oil revenues - which account for about half of total revenues-, the government cut spending from 12.7 percent of GDP in 2014 to 9.5 percent in 2016. This sharp fiscal contraction contributed to a sharp slowdown in GDP growth ‒ from 5.3 percent in 2012- 2014 to 0.6 percent in 2015-2017. It also exposed several issues with Nigeria’s fiscal policy during the commodity super-cycle. First, the authorities did not replenish their sovereign wealth fund (called Excess Crude Account) in 2010-2013, leaving a limited fiscal space to react to shocks. Second, the oil price assumption used in the preparation of the budget cycle was not independently formulated, which made it vulnerable to political influences. Third, the fiscal rule that was designed to insulate spending from oil price volatility did not account for variables affecting oil production (Izvorski et al., 2018). Only a few non-oil exporting countries undertook counter-cyclical fiscal policy during the boom years, most of them spending the windfall or running an a-cyclical policy. Like oil exporters, the countries that were able to build sufficient buffers during good times are also the ones that maintained a counter-cyclical policy stance in the aftermath of the price downturn. This was done, however, at the cost of rising debt distress risks in many of these countries, such as Zambia and Ethiopia, where the continuation of countercyclical policy in the wake of the global crisis should have come to a halt when the price downturn hit, compounding the global crisis shock. Among non-oil exporters in the sample,28 only Chile was able to maintain simultaneously a countercyclical policy stance throughout the period and a low risk of debt distress.29 The counter- cyclical fiscal policy manifested itself by an increase in government expenditures which grew from 23.3 percent of GDP in 2012-2014 to 25.2 percent of GDP in 2015-2017. This helped attenuate output volatility: GDP growth dropped by 2 percentage points between 2012-2014 and 2015-2017, compared to 2.6 percent, on average, across the other metal exporters. In parallel, Chile’s fiscal sustainability was maintained thanks to its fiscal rule, which allowed the country to save a substantial proportion of commodity revenues into sovereign wealth funds during the commodity super-cycle (World Bank, 2017a). Part of these savings were drawn down to boost the economy in the wake of the crisis. 28 Some countries, such as Botswana, are not included in the sample because of lack of data. 29 Section 4 provides a detailed overview of current fiscal vulnerabilities. 21 Figure 17: Evolution of Cyclicality in Oil Exporting EMDEs 2002-2007 and 2015-2016 GDP Elasticity of Government Expenditure (as % Deviation from the Trend) 8 6 4 2 0 -2 -4 -6 ALB DZA AGO AZE BHR BOL CMR COL COG ECU IRN KAZ KWT LBY MYS MMR NGA OMN QAT RUS SAU TTO UAE VEN YEM Cyclicality in 2002-2007 Cyclicality in 2015-2016 Source: Authors’ calculations based on IMF WEO data Figure 18: Evolution of Cyclicality in Oil Exporting EMDEs 2008-2011 and 2015-2016 GDP Elasticity of Government Expenditure (as % Deviation from the Trend) 8 6 4 2 0 -2 -4 ALB DZA AGO AZE BHR BOL CMR COL COG ECU IRN KAZ KWT LBY MYS MMR NGA OMN QAT RUS SAU TTO UAE VEN YEM Cyclicality in 2008-2011 Cyclicality in 2015-2016 Source: Authors’ calculations based on IMF WEO data Public investment often bore the weight of the fiscal response to changes in output. Because increasing public investment in periods of positive output gap - notably to finance infrastructures – and cutting or postponing projects to help close negative gaps is often politically easier than adjusting current spending, capital spending appears historically more procyclical than other categories of expenditures (Ardanaz and Izquierdo, 2017). The procyclicality of public investment spending is shown in Box 3. However, procyclical public investment spending may be associated with lower long-term growth and sustainability. If not managed efficiently, accelerated increases in public investment spending in good times may lead to rapid debt build-up and waste of resources. Similarly, investment 22 spending cuts in bad times may generate detrimental volatility in project implementation. The paper reviews the correlation between the cyclicality of fiscal policy response to short-term output fluctuations and public investment efficiency. The efficiency of public investment is measured as the long-run relationship between public investment and output. To examine this relationship for commodity exporters during the period 1990-2016, a Common Correlated Effects Mean Group (CCEMG) model is used (Pesaran 2006; Pesaran and Smith 1995), described in detail in Appendix 5. Countries are then classified according to three categories regarding long-term public investment efficiency: (i) very inefficient, (ii) inefficient and (iii) efficient. Overall, the results indicate that public investment is efficient in only 23 out of 84 commodity exporters in the sample. Furthermore, after controlling for debt levels and the quality of institutions, results indicate that 60 percent of commodity exporters with a positive long-run elasticity between public investment and growth (1990-2016) have implemented an a-cyclical or countercyclical fiscal policy in 1990-2016 (Appendix 5). This underscores the importance of stabilizing public investment spending, while enhancing efficiency of public investment management - to support long-term growth and protect fiscal space.30 Box 3: A Long-Term Perspective on Procyclicality of Fiscal Policy A long-term perspective over 1980-2017 shows a large heterogeneity of fiscal policy behavior over income levels, commodity boom and bust episodes, and more widespread procyclicality of public investment spending. 31 Frankel, Vegh, and Vuletin 2013 analyzed the evolution of fiscal policy in 96 developing and industrial economies and found that many countries had “graduated” from fiscal procyclicality since the early 2000s. Figure 19 replicates their methodology to commodity-exporting EMDEs and extends it to 2017. It depicts the correlation between the cyclical components of real government expenditure and real GDP (output gap) for the periods 1980 – 1999 and 2000-2017. Like Frankel Vegh, and Vuletin (2013), the plot is divided into four quadrants and the countries are classified as (1) Established graduates (bottom-left), countries that have always had counter-cyclical fiscal policy; (2) Still in school (top-right), countries that have always been procyclical, (3) Recent graduates (bottom-right), countries that were procyclical during 1980-99 but “graduated” from pro-cyclicality and have been a-cyclical or even countercyclical since 2000; (4) Back to school, countries that used to be a-cyclical or counter- cyclical during 1980-1999 but have been downgraded to procyclical fiscal policy since 2000. 30 It should be noted that the observed relationship between the efficiency of public investment and the a(counter)cyclicality of fiscal policy is possibly endogenous as these two measures could be driven by similar factors, such as the quality of institutions. 31 The cyclical components have been estimated using the HP filter with a smoothing parameter equal 100. Sample: Countries listed as commodity exporting EMDEs by (World Bank, GEP 2018) and data on GDP and government expenditure available before 2000. Final sample contains 65 countries: Albania, Algeria, Angola, Argentina, Azerbaijan, Bahrain, Belize, Benin, Bolivia, Brazil, Brunei, Burkina Faso, Burundi, Central African Republic, Chad, Chile, Colombia, DR of Congo, Congo, Rep. of, Costa Rica Cote d’Ivoire, Ecuador, Equatorial Guinea, Ethiopia, Gabon, Ghana, Guatemala, Guinea, Guinea Bissau, Guyana, Honduras, Indonesia, Iran, Kenya, Kuwait, Kyrgyzstan, Libya, Madagascar, Malaysia, Morocco, Mozambique, Myanmar, Namibia, Niger, Oman, Papua New Guinea, Paraguay, Qatar, Russia, Rwanda, Saudi Arabia, Senegal, Sudan, , Suriname, Tajikistan, Tanzania, Togo, Trinidad and Tobago, Turkmenistan, Uganda, Ukraine, United Arab Emirates, Uzbekistan, Venezuela, Yemen. 23 Figure 19: Country Correlations Between the Cyclical Components of Real Government Expenditure and Real GDP, 2000-2017 versus 1980-1999 by Export Category (Left) and Income Group (Right) Source: Authors’ calculations based on IMF WEO data. Overall, commodity-exporting EMDEs have been less able to reduce procyclicality than other EMDEs in the last two decades. Only 5 percent of the sampled countries hold the status of established graduates, and only 14 percent of the sample has recently graduated. On the contrary, 66 percent of the sample is still in school and 14 percent is back to school compared to the period 1980-1999. In contrast, the overall graduation rate for EMDEs is 34 percent, as reported in Frankel, Vegh, and Vuletin (2013). While the distribution of countries across the graduate status is fairly homogenous across oil and non-oil EMDEs, 100 percent of established graduates are either high or upper- middle income commodity exporters. Furthermore, most countries remained procyclical with respect to non- commodity GDP over the period under review (Figure 20). The majority of high-income and several upper middle- income economies, including Bahrain, Chile, Kuwait, Guyana, Brunei and Algeria, ran countercyclical fiscal policy in response to changes in output gaps, but procyclical policy with respect to fluctuations in the non-commodity sector. Figure 20: Country Correlations Between the Cyclical Components of Real Government Expenditure and: (i) Non-Commodity Real GDP (comparison with real GDP) 2002-2016 (Left) 32; and (ii) Commodity Prices 2000-2017 versus 1980-1999 (Right) Procyclicality of Govt Expenditure vis-a-vis the Commodity Price Index, 1980-1999 versus 2000-2017 Source: Authors’ calculations based on IMF WEO data. 24 During the last two decades, fiscal policy was much less procyclical over the commodity cycle than over the business cycle. Figure 20 (Right) plots the country correlation between the cyclical components of real government expenditure and the country-specific commodity price index, comparing the period 1980-1999 with 2000-2017. In general, 82 percent of the commodity-exporting EMDEs were less procyclical over the commodity cycle than over the business cycle. Moreover, 39 percent are classified as established graduates and 17 percent as recent graduate, and only 24 percent and 18 percent are back in school and still in school, respectively. Why such difference? An explanation may be that global commodity prices, although largely unpredictable, have higher visibility than output gaps, and a significant impact on fiscal revenue; hence price fluctuations may represent a key driver of fiscal policy. In this respect, fiscal rules established by commodity exporters have often focused on managing commodity revenue based on price levels and fluctuations (as prices still represent, as seen in section 2, a major driver of fiscal revenue), saving a fraction of that income during good times and spending the savings in recessions for macroeconomic stabilization purposes. For example, Chile’s structural surplus rule involves saving copper revenues that are above their perceived long-run level in a Sovereign Wealth Fund, and drawing upon these funds when copper prices are lower (Box 4 provides a brief overview of the types and experience of fiscal rules). The fiscal policy response to output fluctuations may differ depending on whether the country is going through an episode of commodity boom or a bust. The literature (e.g., Spatafora and Samake, 2012) points out that governments have different reasons and incentives for pursuing more procyclical policy during economic booms. In general, the view is that positive price shocks are more willingly seen as permanent – thus requiring adjustment to this “new normal” and greater procyclicality-, while bust episodes are seen as temporary. Overall, our analysis indicates that 9 percent of commodity exporting EMDEs consistently implement countercyclical policy during both booms and busts (Figure 21). Conversely, 68 percent of countries are consistently procyclical. Except Kyrgyzstan, all consistently countercyclical countries are either high or upper-middle income countries, including Malaysia, Bahrain, Kuwait, Guyana, Costa Rica and Chile. Figure 21: Country Correlations Between the Cyclical Components of Real Government Expenditure and Real GDP, Boom Episodes versus Bust Episodes. By Export Category (Left), and Income Groups (Right) 33 Source: Authors’ calculations based on IMF WEO data. The composition of the fiscal policy response to shocks matters, because highly procyclical public investment, if not well-managed, may be quite harmful to long-term economic growth and thus fiscal sustainability. Increasing investment disproportionally during positive shocks and then cutting back in recessions might translate into inefficient capital allocation and waste in good times, and incomplete implementation during bad times. Figure 22 displays a comparison between the cyclicality of government expense (total expenditure less public investment) 32In the absence of data, non-commodity GDP is proxied as GDP – commodity exports. Commodity exports is provided by CommTrade for 2002-16. 33Following Frakel et. al (2013), the cyclical components have been estimated using the HP filter with a smoothing parameter equal 100. Commodity boom begins: Commodity price Index reaches a level of at least 25% above the trend. Commodity boom ends: Commodity price index comes back to a level lower than 10% above the trend. 25 and investment over the business cycle in the period 1980-2017. Commodity exporters have generally implemented procyclical policy with respect to both expense and investment (upper right quadrant) in 1980-2017. Furthermore, investment tends to be more procyclical than government expense: 55 percent of the sampled countries lie below the 45 degrees line. Lower income commodity-exporting EMDEs have generally been more procyclical during episodes of booms than busts in 1980-2017. In 61 percent of the countries under consideration, fiscal policy tends to be more decisively procyclical during good times than bad times. This is particularly the case for 72 percent of low and lower-middle income countries, and, among those, for 78 percent of low-income countries. This may suggest a higher propensity to spend temporary windfalls during commodity boom periods, and less focus on supporting aggregate demand when the output gap becomes temporarily negative - and thus overall optimism. On the contrary, a majority of upper middle income and high income EMDES have been less procyclical during commodity booms than busts. This may suggest stronger macroeconomic and fiscal frameworks (with macroeconomic management geared towards avoiding overheating and Dutch disease from temporary windfalls during good times and towards investing a higher fraction of such windfalls during episodes of commodity downturns), and possibly better access to finance to support temporary downturns in periods of favorable commodity prices compared to episodes of lower prices. Drivers of procyclicality are analyzed in detail in Section 3.3. Figure 22: Procyclicality of Real Government Expense versus Investment Over the Business Cycle, 1980- 2017 (Correlations Between Cyclical Components). Oil and Non-Oil Exporters on the LHS, Income Groups on the RHS Source: Authors’ calculations based on IMF WEO data. The cyclical components have been estimated using the HP filter with a smoothing parameter equal 100. Sample: Countries listed as commodity exporting EMDEs by (World Bank, GEP 2018) and data on real government expenditure and real government investment 1980-2017. 3.3. Understanding the Fiscal Policy Response to the Recent Commodity Cycle What were the main drivers of fiscal procyclicality in the past two decades? Reducing procyclicality of fiscal policy in response to temporary shocks may allow to rebuild buffers, support macroeconomic stability and prepare for future downturns. On the other hand, enduring shocks may require fiscal adjustment to a “new normal”, especially during downturns if financing constraints are present or if increased spending could jeopardize macroeconomic sustainability. In order to understand what has driven the fiscal policy response in the last commodity cycle of the 2000s and 2010s, the following regression is estimated: 26 (3.3.1) Where ̂, correspond to the cyclical components of government expenditure and GDP ̂, and in period in country , respectively, expressed as percent deviation from the trend. The vector , contains candidate variables for drivers of fiscal procyclicality. It includes sets of variables related to: (i) structural macroeconomic characteristics linked to the level of development and commodity dependence, (ii) political factors (iii) the quality of the institutional framework, (iv) access to financing (v) public debt and external sustainability considerations, and (vi) fiscal and monetary policy frameworks. This specification is based on Frankel et. al (2013). The vector contains the main parameters of interest. They capture the marginal effect of each independent variable on the overall procyclicality. For example, suppose that is the marginal effect of the quality of the institutional framework. A negative (positive) value means that improvements in institutional quality over time reduces (increases) procyclicality. Country fixed effects, , are included to allow country-specific characteristics to affect the dynamics of public expenditure. An annual panel data of commodity exporting EMDEs for the period representing the recent commodity cycle (2000- 2016) is employed for the estimation of equation (3.3.1). Countries are classified according to the GEP (2018) criteria, yielding an unbalanced sample of 68 countries. Table 5 (below) describes the variables included in the vector , and reports their source, mean and standard deviation. Table A9 in Appendix 6 displays the impact of each candidate variable, taken individually, on the procyclicality of government expenditure for our sample of commodity-exporting EMDEs during the last commodity cycle. Exogenous variables and their expected effect are described below. • Structural macroeconomic characteristics (Panel A). Real GDP per capita measures the level of development as a determinant of the cyclicality of fiscal policy. Commodity exports as a share of GDP measures the dependence of each country on the commodity export sector. Since commodity prices are usually more volatile than the non-resource business cycle, it is expected that countries that are highly dependent on commodity exports would rely more on countercyclical fiscal policy to stabilize output. In this panel, GDP volatility is a proxy for the variability of tax revenues.34 The motivation to include GDP volatility in the vector , is based on the channel emphasized by Talvi and Vegh (2005), who argue that, in the presence of political distortions, the larger the variability of tax revenues, the more pro-cyclical fiscal policy tends to be. 34We follow Keita and Turcu (2018) and measure GDP volatility as the square of the cyclical component of real GDP from an HP (100) filter. 27 • Commodity price shock persistence. An enduring shock may trigger the need for fiscal adjustment. There are two channels through which shock persistence can affect procyclicality. First, due to financial frictions, financial institutions may not provide funding for countercyclical fiscal policy for extended periods. Second, based on the permanent income hypothesis, Mendes and Pennings (2017) argue that governments may choose a procyclical response to enduring shocks in order to smooth consumption of financially-constrained households. These considerations are incorporated in the analysis by controlling for the persistence of the country-specific commodity price index.35 • Political factors. Abundant literature has highlighted political variables as leading drivers of procyclicality, since Alesina and Tabellini (1990) who found that a tendency to overspend in good times can result from election uncertainty. Elected governments may have a tendency to use their time in office to provide a certain type of preferred public goods, building up public debt which will constraint the next administration, in turn, to provide other types of public goods. Velasco (2000) argues that public resources are a common property out of which interest groups can finance expenditures on their preferred goods. Interest groups fail to internalize the cost of the government expenditures on their behalf, leading to excessive indebtedness and procyclicality. To test if these channels are at work in commodity-exporting EMDEs, we include in vector , a measure of political checks and balances from the IDB Database of Political Institutions (Panel B in Table 5). Stronger checks and balances are assumed to constrain politicians to influence different branches of the government, reducing rent-seeking activities. • Quality of institutions. Calderón and Schmidt-Hebbel (2008), Frankel, Vegh, and Vuletin (2013), Cespedes and Velasco (2014), Carneiro (2016) stress that the ability of countries to conduct countercyclical fiscal policy is notably affected by the quality of their institutions. As indicated in panel B of Table 5, the variables used to measure institutional quality include government effectiveness, rule of law, regulatory quality and control of corruption, all provided by the International Country Risk Guide.36 • Access to domestic or international financial markets (Panel C). Gavin and Perotti (1997), Riascos and Vegh (2003) and Caballero and Krishnamurthy (2004) argue that imperfect international credit markets prevent developing countries from borrowing during economic downturns. To test this channel, we follow Loayza and Ranciere (2006) and measure the development of domestic financial markets using liquid liabilities as a share of GDP, provided by the World Development Indicators database, and financial integration using the Chinn-Ito financial openness index (Chinn and Ito 2006). 35Following Mendes and Pennings (2017), we measure persistence as the autoregressive coefficient of an AR(1) regression on the country commodity price index. 36Following Frankel, Vegh, and Vuletin 2013, we calculate a country-specific institutional quality index (IQ) as the average of these four variables. The index is normalized to range between 0 (lower quality) and 1 (highest quality). 28 • Public debt and external buffers. Debt levels and external buffers influence the country’s default risk premia and its ability to borrow without threatening debt sustainability (e.g., Cuadra, Sanchez, and Sapriza (2010), IMF (2015, 2016a), Kose et al. (2017)) and withstand negative shocks. Addressing these concerns, two stock variables are added to , : (i) government debt-to GDP ratio provided by World Economic Outlook, and (ii) foreign currency reserves in terms of monthly imports, provided by the World Development Indicators. To mitigate concerns of reverse causality running from procyclicality to public debt, we consider the one-year lagged debt-GDP ratio instead of the contemporaneous value of debt. • Fiscal rules may help to strengthen fiscal discipline and accountability, contributing to macroeconomic stabilization and making fiscal policy more resilient to government corruption (Kumhof and Laxton 2013; Elbadawi, Schmidt-Hebbel, and Soto 2015). However, the experience of EMDEs is mixed (Box 4). We test if the adoption of fiscal rules by commodity exporting EMDEs have contributed to reducing procyclicality during the last commodity cycle. We gather information about fiscal rules from the IMF-FAD Fiscal Rules Dataset (2016)37. A dummy variable indicates if each country’s general government formally operates at least one of the following fiscal rules: budget balance rules, debt rules, revenue rules or expenditure rules. Moreover, we interact the fiscal rules dummy with a second dummy indicating if the rule comes along with formal enforcement procedures (IMF-FAD Fiscal Rules Dataset (2016)). Box 4: Experience of Fiscal Rules in Commodity Exporting Countries As of 2015, 29 out of the 89 countries in the sample of commodity exporters had fiscal rules that aimed at supporting fiscal credibility and discipline and help insulate spending from commodity prices volatility (Figure 24). Fiscal rules include: • Budget balance rules (overall balance, current balance, primary balance) aiming to control the year-by- year evolution of the fiscal balance, such as in Indonesia, Mexico, Nigeria. • Structural or cyclically-adjusted balance (structural fiscal balance, non-resource balance, non-resource primary balance, structural non-resource primary balance, cyclically adjusted primary balance) to accommodate the fiscal balance to price volatility and economic cycle, such as Chile, Columbia, Ecuador, and Timor Leste. • Expenditure rules (ceiling on government spending) to limit the procyclicality of fiscal policy, such as in Botswana, Chad, and Ecuador. • Debt rules focusing on long-term sustainability. 37 Schaechter, A., T. Kinda. N. Budina, and A. Weber (2012); Lledo. V, S. Yoon, X. Fang, S. Mbaye, and Y. Kim (2017). 29 Figure 23: Commodity-Exporting Countries with Formal Fiscal Rules in Place (%) 90% 81.8% 80% 69.4% 68.0% 70% 61.1% 60% 50% 38.9% 40% 30.6% 32.0% 30% 18.2% 20% 10% 0% Source: IMF The experience of commodity exporters with fiscal rules has been mixed. While fiscal rules have helped ensure fiscal discipline in some countries (Chile, Botswana, Norway, and Timor-Leste), they have had limited success in many other commodity exporters (Nigeria, Equatorial Guinea). A recent study of resource-rich countries finds that fiscal rules have not reduced the procyclicality of government expenditure in a statistically significant way (Bova et al., 2016). Fiscal rules alone are not enough to promote counter-cyclical fiscal policy and should be combined with better institutions (Keita and Turcu, 2018). Country experiences show that successful fiscal rules generally are flexible and easy to monitor, grounded on strong legal basis, tightly linked to fiscal sustainability objectives, and support countercyclical fiscal policy. Broad institutional coverage (general versus central government) and supporting institutions, including fiscal councils, are also important. • Exchange rate regime. Fiscal policy has a major role in output stabilization at the country- level in fixed exchange rate regimes, where variations in the exchange rate cannot help to absorb shocks. Fiscal policy would therefore be expected to be more proactively countercyclical (Gali and Monacelli, 2005). Mendes and Pennings (2017) also argue that when monetary policy is effective in absorbing structural shocks, fiscal policy can focus on smoothing the consumption of financially-constrained households over the business cycle. Both views imply that fiscal policy in commodity exporters (facing persistent shocks) will be more procyclical (or less countercyclical) as exchange rates become more flexible. We introduce these considerations in the analysis by adding an exchange rate regime index based on Ilzetzki et al (2017). Panel D in Table 5 shows that the classification ranges from 1 to 15, with lower values representing more fixed exchange rate regimes and higher values representing more flexible regimes. • Monetary framework. Independence of monetary authorities along with stronger monetary frameworks may limit the scope of seignorage revenue and therefore constrain government expenditure and a countercyclical fiscal stance during downturns. An index of monetary policy independence provided by Aizenman et al. (2009) is added to equation (3.3.1). Panel D in Table 5 shows that the index ranges from 0 (lowest independence) to 1 (highest independence). 30 Table 5: Source, Mean and Standard Deviation of the Independent Variables When estimating the model, shock persistence appears to be a key driver of procyclicality over the cycle. Econometric results are provided in Table 6. As emphasized in Mendes and Pennings (2017), there is a strong link from the persistence of the commodity price index to fiscal pro- cyclicality (column 11). A 0.1 increase in the autoregressive coefficient increases the GDP- elasticity of government expenditure by 0.3. This effect is sizeable and statistically significant at 5 percent. This result suggests that enduring positive and negative price shocks have triggered an adjustment – increasing spending in good times and consolidating in bad times – with less emphasis on short-term stabilization. On the contrary, per capita income is not a significant variable once institutional quality is accounted for (Column 3), while political checks and balances, although consistently negatively related to pro-cyclicality (Columns 4 through 11), is not a 31 statistically significant variable. Similarly, according to the results, the exchange rate regime and the independence of monetary authorities do not appear to have played a significant role. When formally enforced, fiscal rules are associated with less counter-cyclical fiscal policy. Columns 9 through 11 show the coefficient on enforced fiscal rules is consistently positive and significant at 5 percent. As documented in Box 4, the explanation for this surprising result may lie in the fact that fiscal rules in commodity-exporting EMDEs often aim at ensuring debt sustainability and do not prioritize output stabilization. Access to financing and the level of foreign exchange buffers are major determinants of countercyclicality, even after controlling for shock persistence and debt ratios. After controlling for all elements in , Column 11), higher financial depth and financial integration into global financial markets reduce the procyclicality of government expenditure, being both statistically significant at 5 percent. Higher levels of international foreign exchange reserves also appear as an important driver of countercyclicality, with a 10 percent significance. The point estimates are remarkably large: a build-up of foreign reserves equivalent to one month of imports reduces pro- cyclicality by 0.07; expanding liquid liabilities by 10 percent of GDP is associated with a 0.13 decrease in pro-cyclicality; and a 10 percent improvement in the capital account towards more openness decreases pro-cyclicality by 0.07. This underscores that while enduring shocks will trigger fiscal adjustment - as seen above-, financing availability is also in itself a key determinant of the fiscal policy response to shocks. The procyclicality of capital flows (Kaminsky, Reinhart, and Vegh, 2005; Frankel, 2017) fuels the procyclicality of fiscal policy, contributing to overspending in good times when resources are abundant - and spending cuts in bad times when resources dry up. This may impose more drastic responses to shocks than warranted in any given circumstances. 32 Table 6: Country Fixed Effects Panel Regressions. All Commodity-Exporting EMDEs 2000-2016 (Dependent Variable: Cyclical Component (HP-Filter) of Log Real Government Expenditure) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Cyclical log real GDP 2.12** 2.25** 3.38*** 3.59** -1.13 -1.11 -0.20 -0.26 -0.36 -0.47 -0.44 (0.92) (0.93) (1.30) (1.46) (1.37) (1.48) (1.48) (1.43) (1.42) (1.57) (1.59) Interaction between Cyclical log real GDP with Per capita GDP -0.01* -0.01* -0.00 -0.00 -0.01* -0.02*** -0.01 0.00 -0.00 0.01 0.01 (0.00) (0.00) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) Commodity exports -1.44*** -1.37*** -0.89** -0.98* -1.45** -0.53 -0.93 -0.23 -0.08 -0.13 -0.12 (0.36) (0.36) (0.39) (0.59) (0.63) (0.72) (0.74) (0.71) (0.71) (0.74) (0.74) Commodity index persistence -0.72 -0.89 -1.13 -1.22 2.97** 2.67* 2.11 2.65* 2.87** 3.00** 2.99** (1.08) (1.08) (1.46) (1.52) (1.41) (1.52) (1.50) (1.42) (1.42) (1.46) (1.47) GDP volatility 0.00 0.00* 0.00 -0.01** -0.01** -0.00 -0.00 -0.00 -0.00 -0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) Institutional Quality index -2.60*** -2.76*** -0.31 -0.02 -0.00 -0.47 -0.71 -0.76 -0.79 (0.67) (0.70) (0.72) (0.85) (0.88) (0.92) (0.92) (1.30) (1.32) Checks and balance -0.02 -0.06 -0.04 -0.05 -0.01 -0.05 -0.03 -0.03 (0.09) (0.08) (0.09) (0.09) (0.08) (0.08) (0.09) (0.09) Foreign reserves 0.00 0.00 -0.00 -0.05** -0.06** -0.07** -0.07** (0.00) (0.01) (0.01) (0.02) (0.02) (0.03) (0.03) Financial depth -0.23 -0.01 -1.40** -1.22** -1.39** -1.39** (0.43) (0.43) (0.59) (0.59) (0.63) (0.63) Financial integration -0.94*** -0.87*** -0.77*** -0.70*** -0.70** (0.27) (0.25) (0.26) (0.27) (0.27) Debt-to-GDP ratio (one year lag) 0.28 0.19 0.19 0.20 (0.27) (0.27) (0.28) (0.30) Enforced fiscal rules dummy 0.63** 0.65** 0.66** (0.31) (0.32) (0.33) Monetary independence 0.08 0.09 (0.50) (0.50) ER regime -0.00 (0.03) Constant -0.00 -0.00 -0.00 -0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) Observations 1,180 1,180 983 953 790 710 699 670 670 603 603 R-squared 0.23 0.23 0.23 0.23 0.22 0.24 0.27 0.25 0.25 0.25 0.25 Number of countries 67 67 59 59 51 49 48 47 47 46 46 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations Higher levels of the debt-to-GDP ratio in the previous year do not lead to higher procyclicality. Although the coefficient on the one-year lagged debt is positive as expected across all specifications, it is not statistically significant at 10 percent. Despite the fact that higher debt levels could have triggered fiscal consolidation to protect or restore macroeconomic sustainability, this effect seems to have played a small role in inducing procyclicality once the other variables are accounted for. 33 At odds with the literature, institutional quality is not among the main direct drivers of countercyclicality once financial variables are included. Consistently with the literature, the coefficient of the IQ index is negative and statistically significant at 1 percent in columns (3) and (4). However, we find that this result essentially reflects the high correlation between IQ and variables related to access to financing and debt. The significance of the IQ index vanishes as these variables are added in column (5) and onwards. This result is at odds with the literature that stresses the importance of institutions in preventing fiscal mismanagement. This branch of the literature suggests nevertheless that the quality of institutions is likely a critical factor for building fiscal policy credibility that will support access to more affordable financing during downturns, allowing thereby countercyclical fiscal responses to volatility. Figure 24: Institutional Quality in OECD and CE’s EMDE Countries 2000-16 Government Effectiveness 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Rule of Law 0 Control of Corruption Source: Political Risk Services International Country Risk Guide (PRS) Regulatory Quality Average in OECDs Average in EMDEs Norway Chile Source: Political Risk Services, International Country Risk Guide (PRS). Furthermore, the quality of institutions in commodity-exporting countries remains generally low and this may influence the results. As shown earlier, IQ has not, on average, significantly improved during the boom. Figure 24 displays the average score of each variable used to construct the IQ index during the period 2000-16 for OECD and commodity-exporting EMDEs. In each dimension, the average score of EMDEs appears well below the score of OECD countries, notably Norway, which is recognized for the high-quality of its institutions and its success at promoting fiscal discipline through the implementation of a sophisticated fiscal structural-surplus rule. Chile is also one of the few EMDEs which managed to implement a sound fiscal framework with a successful rule that involves saving copper revenues that are above their estimated long-run level and drawing upon these savings when copper prices are low, along with independent commodity price forecasting (Frankel, 2011). Overall, as shown in Figure 25, the quality of institutions in commodity-exporting EMDEs remains relatively weak on average over the period. The mean of the IQ index is 0.47 and its dispersion is small, 0.11. Therefore, small improvements in relatively low-quality institutions may have not brought about the changes required to constrain politicians to reduce overspending in good times or to convince market players to lend funds to governments when the downturn happened. 34 Overview of Existing Fiscal Vulnerabilities and Risks in Commodity- Exporting EMDEs Do commodity exporters stand ready to withstand the next shocks? In this paper, fiscal risks linked to the commodity export sector are assessed as a function of a country’s dependence38 on commodity exports as a source of foreign exchange and fiscal revenue; the commodity price outlook; and its capacity to respond to negative commodity price shocks (vulnerabilities). Risks are particularly important for those commodity exporters that are highly dependent on the commodity export sector, are facing a negative price outlook and may be unable to undertake countercyclical fiscal policy in the wake of a (temporary) shock in a fiscally sustainable manner. This framework (Figure 25) builds on abundant work on measuring fiscal vulnerability in advanced economies and EMDEs (Hemming and Petrie, 2000; Van Doorn, Suri, and Gooptu. 2010; Baldacci, McHugh, and Petrova, 2011). Although cautiousness is warranted given the unpredictability of price fluctuations, the commodity price outlook is considered in the analysis, because it allows to distinguish between those commodity exporters which have little buffers but are likely to benefit from favorable price evolutions in the near future, and those which, in the same situation, are on the brink of a price decline. Using the country-specific price index allows to do this distinction with some granularity. Figure 25: A Simple Framework for Assessing Fiscal Risks from Exposure to the Commodity-Export Sector Dependance Commodity price outlook Fiscal vulnerabilities Commodity-related fiscal risks 38 As defined in footnote 5. 35 4.1. The Commodity Price Outlook, 2018-2025 Commodity prices have stabilized since 2016, but on average remain above their long-term trends, except for oil prices.39 After the end of the super cycle of the 2000s, the commodity price downturn of the early 2010s has severely affected commodity exporters. Since 2016, global prices for minerals, mining and agricultural products have generally stabilized (Figure 26). However, as indicated earlier, they remain largely above their long-term trends. Oil prices, on the other hand, are projected to further increase moderately in the coming years, suggesting a more favorable outlook for energy exporters, but posing terms of trade risk for commodity exporters that are also net oil importers. This underscores that a precautionary fiscal stance is warranted. Figure 26: Commodity Prices have Stabilized Since 2016, but they Remain Generally Above their Long-Term Trends, Except for Oil Price Fossil Fuels: Crude Oil (real 2010 USD) Agriculture: Soy beans (real 2010 USD) $120.0 $1,000.0 $900.0 $100.0 $800.0 $700.0 $80.0 $600.0 $60.0 $500.0 $400.0 $40.0 $300.0 $200.0 $20.0 $100.0 $0.0 $0.0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 Forecast Crude oil, average Linear (Forecast) Forecast Soybeans Linear (Forecast) Source: Authors’ calculations based on World Bank Source: Authors’ calculations based on World Bank Commodities Price Data. Commodities Price Data. Minerals: Copper (real 2010 US dollars) Precious Metals: Gold (real 2010 USD) $9,000.0 $1,600.0 $8,000.0 $1,400.0 $7,000.0 $1,200.0 $6,000.0 $1,000.0 $5,000.0 $800.0 $4,000.0 $600.0 $3,000.0 $400.0 $2,000.0 $200.0 $1,000.0 $0.0 $0.0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 Forecast Gold Linear (Forecast) Forecast Copper Linear (Forecast) Source: Authors’ calculations based on World Source: Authors’ calculations based on World Bank Commodities Price Data. Bank Commodities Price Data. 39 Global commodity prices are extracted from the World Bank pink sheets. 36 Out of 69 commodity exporting countries for which the specific index could be computed, less than a third ― mostly oil exporters ― face a positive price outlook for 2018-2025.40 As prices are difficult to forecast (Bowman and Husain 2004), these results need to be interpreted with caution. Nevertheless, they suggest that non-oil exporters may face further declines in prices in the next few years (Table 7). This may limit a country’s ability to rebuild fiscal buffers and withstand future downturns. Table 7: The Commodity Price Outlook in 2018-2025 (Expected % Change in the Country- Specific Commodity Price index in Real Terms) Below -20% Below -10% Below 0% 0 - 10% Malaysia Cameroon Malawi Albania Madagascar Algeria CAR Ghana Mali Cameroon Malawi Angola Costa Rica Malaysia Mauritania Colombia Mali Azerbaijan Indonesia Burkina Faso Mozambique Ecuador Mauritania Bahrain Kyrgyzstan Burundi Namibia Ghana Morocco Bolivia Namibia CAR Nicaragua Malaysia Mozambique Congo Nicaragua Costa Rica Niger UAE Namibia Iran Ethiopia Peru Argentina Nicaragua Iraq Guatemala Rwanda Armenia Niger Kazakhstan Guyana South Africa Belize Paraguay Kuwait Honduras Tanzania Benin Peru Libya Indonesia Uganda Brazil Rwanda Myanmar Kenya Uruguay Burkina Faso Senegal Nigeria Kyrgyzstan Burundi South Africa Oman Lao CAR Tanzania Qatar Madagascar Chile Togo Russia Costa Rica Uganda Saudi Arabia Cote d'Ivoire Ukraine TTO Ethiopia Uruguay Venezuela Guatemala Zambia Yemen Guyana Zimbabwe Suriname Honduras Indonesia Kenya Kyrgyzstan Lao Source: Authors’ calculations. Price forecasts from World Bank Pink Sheet 2018. Note: Prices: Forecasts up to 2025 (Real 2010 US dollars) 4.2. Assessing Readiness for the Next Shock A range of indicators can be used to measure the fiscal space available to commodity exporters to withstand future price downturns, but the significant lack of data represents a major caveat. The proposed indicators described in Appendix 7 build on the literature (Botev, Fournier, and Mourougane 2016; Ghosh et al. 2013) and the recent work conducted by the IMF (2016) and the 40 The price index could not be constructed for all commodity exporters due to data limitations. 37 World Bank (Kose et al. 2017) to examine fiscal space. These variables may be of varying importance depending on each country’s characteristics and data availability. They aim to capture specific features shared by commodity exporters, including notably (i) fiscal revenue exposure; (ii) long-term fiscal and debt sustainability; (iii) short-term liquidity constraints; (iv) size of contingent liabilities stemming from exposure to the commodity export sector; (v) expenditure rigidities; and (vi) the quality of institutions. In the following review, a subset of these indicators is used for cross-country analysis, based on data availability. 4.2.1. Twin Deficits and Debt are Growing The recent commodity price downturn has eroded the relative gains achieved by commodity exporters during the super cycle. On average, the current account balance returned to a deficit after the 2008 global financial crisis and deteriorated steadily in the wake of the non-oil and oil commodity price downturn of the early 2010s. In fact, since 2014, the average current account deficit (CAD) of commodity exporters has exceeded the average CAD of non-exporters for the first time since 2002 (Figure 27). Similarly, the overall fiscal balance of the average commodity exporter has also been significantly impacted by the recent commodity downturn, weakening progressively in the early 2010s, in contrast to the path observed across commodity importers. Deteriorating fiscal positions have led to a steep increase in debt ratios among commodity exporters, both in gross and net terms (Figure 28 and Figure 41),41 thus eroding the relative gains achieved in the past decade thanks to favorable commodity prices and the HIPC and MDRI initiatives. Figure 27: Among Commodity Exporters, Figure 28: … and the Debt Levels Climbed the Current Account Balance Worsened up After the Commodity Downturn Source: IMF WEO April 2017, Authors’ Source: IMF WEO April 2017, Authors’ Calculations. Calculations. Fiscal balances have steadily deteriorated across the four main commodities: fuel, metals, agriculture raw materials and food. Only oil exporters managed to accumulate a small fiscal 41 Based on data availability. 38 surplus on average during the price boom. However, this surplus eroded in the wake of the global financial crisis and turned into high primary and overall deficits after oil prices plummeted. Likewise, exporters of metals, agricultural and food products suffered from a steady - albeit less pronounced - deterioration in their overall fiscal balances due to the combination of the global crisis and the commodity price downturn (Figure 29). Figure 29: Among commodity exporters, the fiscal accounts have deteriorated steadily since 2008 Source: IMF WEO April 2017 and Authors’ Source: IMF WEO April 2017 and Authors’ calculations. calculations. Source: IMF WEO April 2017 and Authors’ Source: IMF WEO April 2017 and Authors’ calculations. calculations. Fiscal positions have deteriorated beyond cyclical fluctuations, which points to heightened vulnerabilities. On average, the cyclically-adjusted fiscal balance has worsened, especially among non-oil and middle-income commodity exporters (Figure 30). This points to a significant erosion of fiscal space going forward. Furthermore, in 2017, cyclically-adjusted balances in percentage of potential GDP are, on average, substantially lower than at the end of the last commodity downturn episode of the 1990s – with a less favorable outlook. The deterioration of primary balances is weighing on the public debt dynamics. The primary balance fiscal sustainability gap, based on the 39 indicator computed by Kose and et al. (2017)42 under current market conditions, has become negative after the commodity downturn for all categories of commodity exporters (Figure 31). This deterioration is more pronounced for oil exporters, but importantly, these countries are facing a more positive price outlook: this could result in improved market conditions and thus a reduced fiscal sustainability gap going forward. Low-income economies, on average, show a negative sustainability gap, which points to unfavorable debt dynamics. The gap is still substantially lower than at the end of the last bust episode of the 1990s, as debt levels have been brought down with the support of the HIPC and MDRI initiatives. Figure 30: The Cyclically-Adjusted Fiscal Figure 31: … the Fiscal Sustainability Gap Balance and… have Worsened After the Recent Price Downturn Source: DEC fiscal space database, authors’ calculations. Public debt vulnerabilities in commodity exporting countries have risen substantially over the past years. While public debt levels on average remain below the levels of the early 2000s, they have increased sharply in many commodity-exporting countries, particularly oil exporters.43 Between 2011 and 2017, average public debt in oil-exporting countries increased by 26 percentage points, compared to 15 percentage points for non-oil commodity exporters, and 3 percentage points for commodity importers (Figure 41). In addition, the composition of public debt has changed away from traditional, concessional sources of financing towards riskier and more expensive sources of financing. There has been a sharp increase in the weight of bilateral external debt owed to non- Paris Club creditors for all commodity exporters, particularly oil exporters, at levels much higher 42 Kose et al. 2017. In their paper and the associated database (referred to as the DEC Fiscal Space database in this paper), the primary balance sustainability gap is calculated as the difference between the primary balance and the debt-stabilizing primary balance (pbsusgap) with this equation: pbsusgap = p – (r-g)/(1+g).d*, where p is the primary balance (in percent of GDP), g is the real GDP growth rate, r is the real interest rate (defined as the nominal long-term interest rate deflated by the GDP deflator), and d* the target debt ratio (in percent of GDP). The primary balance sustainability gap under current market conditions is computed by using GDP growth and interest rates at their current levels. The target debt ratio, d*, is defined as being equal to the historical median value for EMDEs. The target (and median) debt ratios resulting from this assumption for EMDEs is 45.2 percent of GDP. 4332 commodity exporting developing countries benefitted from substantial debt relief under the Heavily Indebted Poor Country (HIPC) and the Multilateral Debt Relief Initiatives (MRDI). 40 than those observed among commodity importers (Figure 33). In addition, domestic debt has increased substantially and more low- and lower-middle income commodity exporters have issued bonds in international markets. The shift towards new lenders and more market-based debt has contributed to a fall in the average concessionality of public debt portfolios. Concessional debt as a share of public external debt has declined for all commodity exporters since 2011, but especially for oil-exporting LICs. Figure 32: Concessional Debt Remains Figure 33: … the Share of Non-Paris Club Prevalent among Low-Income Commodity Creditors in Bilateral Debt has Become Exporters, but its Share has Slightly Predominant for Commodity Exporters… Declined Compared to the Early 2000s, and… Concessional Debt (in % of Total External Debt) 100 80 60 40 20 0 Oil Non Oil Oil Non Oil exporters exporters exporters exporters LIC LIC 2000 2011 2016 Figure 34: … Rollover Risks are Figure 35: … and Interest Payments on Increasing, Particularly in Low-Income Public Debt Represent a Growing Burden Commodity Exporting Countries, and … for the Budget Central Government Debt Maturing in 12 Months or Less (Percent of GDP) Government Interest Payment to 10 Revenue (in Percent) 8 20 6 15 4 10 2 5 0 0 2011 2017 2000 2011 2017 Source: WEO, WDI, DEC Fiscal Space database, Authors’ calculations. The burden of interest payments on external and domestic public debt is growing and rollover risks have increased, particularly for low-income commodity exporters. The burden of interest payments 41 on public debt has increased in recent years with the gradual shift to less concessional borrowing. For some countries, this is directly linked to increased reliance on domestic borrowing (Figure 34), which has driven T-bill yields up in the context of narrow domestic financial markets and increasing sovereign exposure of the domestic financial sector. In the Gambia, for example, interest payments absorb over 50 percent of government revenue. Overall, domestic credit to the central government and state-owned enterprises has increased steadily since the global crisis (Figure 36). Furthermore, rollover risks have increased. The share of central government debt maturing in the next 12 months has climbed up significantly between 2011 and 2017, notably in low-income commodity exporting countries (Figure 34). This deterioration of financing conditions has constrained the ability of commodity exporters to undertake countercyclical fiscal policy and/or smooth the fiscal adjustment in the wake of recent price fluctuations, and points to increased fiscal vulnerabilities going forward. Figure 36: Credit to the Public Sector has Sharply Increased Since the Global Crisis Domestic Credit to Goverment and SOEs in Commodity-Exporting EMDEs 2000-2016 15 Percent of GDP 10 5 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 oil Non Oil Source: FinStat. 4.2.2. More Commodity-Exporters Present a High-Risk of Debt Distress The share of commodity exporting countries covered under the Low-Income Country Debt Sustainability Framework (LIC DSF) at high risk of debt distress has increased substantially over the past years (Figure 38). Based on the LIC DSAs prepared for 68 countries, the share of commodity exporters with LIC DSA facing a low risk of debt distress has more than halved since 2011. At the same time, the share of commodity exporters at high risk of debt distress or in debt distress has increased substantially over the same time. While the share of commodity importers at high risk of debt distress has remained high over the past years, LIC DSA risk rating downgrades have been less frequent than for commodity exporters, for which risk dynamics have been substantially worse (Figure 37 and Figure 38). As LIC DSAs were not available in the early 2000s, one cannot compare the present situation with the one that prevailed at the beginning of the super cycle; however, even though a higher number of low-income commodity exporters may have been at high risk of debt distress at the end of the commodity bust of the 1990s, the HIPC and MDRI initiatives were instrumental in bringing debt down in the following years. 42 Figure 37: The Share of Commodity Figure 38: … and the Share of Commodity Exporters at High Risk of Debt Distress Importers at High Risk of Debt Distress Increased Since 2013… Remained High Commodity Exporters: Evolution of Debt Distress Risk Commodity Importers: Evolution of Debt Distress Risk Ratings Ratings (in percent of commodity exporters with LIC DSAs) (in percent of commodity importers with LIC DSAs) low moderate high low moderate high 100 100 18 16 20 34 27 27 80 36 42 42 80 40 40 40 39 45 50 57 61 61 60 51 60 39 50 32 56 32 53 40 40 40 40 40 40 35 43 44 27 26 22 17 20 20 32 34 34 32 33 24 20 20 20 20 20 23 17 17 17 22 18 13 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: WBG-IMF LIC DSA Database, as of September 2018. This includes 45 commodity exporters and 23 importers. For market-access countries, long-term sovereign debt ratings have also deteriorated substantially following the downturn in commodity prices. Using the ratings computed by Kose et al. (2017), column 1 of Table 8 shows that among oil and non-oil exporters, more than half of the countries experienced a deterioration in their sovereign debt ratings between 2011 and 2017. However, in the countries for which ratings are available for the whole period, there is an improvement compared to the levels observed at the end of the commodity bust of the 1990s (column 2). Between 2000 and 2017, only about a quarter of these commodity exporters44 have experienced a downgrade. Nevertheless, they are mostly non-oil exporters, facing a less favorable price outlook. Table 8: The Percentage of Commodity Exporters that Experienced a Deterioration in their Foreign-Currency Long-Term Sovereign Debt Ratings Increased After the Commodity Downturn % of countries having % of countries having experienced a experienced a degradation degradation between 2011 and 2000 between 2011 and 2017 (1) (2) Among all commodity exporters 57.1 23.5 Among oil exporters 60.0 13.3 Among non-oil exporters 55.2 31.6 Source: DEC Fiscal Space database and authors’ calculations.45 Overall, the prevalence of higher debt risks in EMDE commodity-exporters in 2017 points to heightened fiscal vulnerabilities and risks going forward (Figure 39). Many commodity-exporters may find it more challenging to access affordable financing to withstand future short-term price 44 A deterioration is observed in Venezuela and Bahrain, as well as Argentina, Belize, Costa Rica, Mongolia, Papua New Guinea, and a marginal decline in South Africa. 45 Rating is averages of foreign-currency long-term sovereign debt ratings by Moody’s, Standard& Poor’s, and Fitch Ratings. Each rating is first converted to a numerical scale ranging from 1 to 21 (higher, better rating). Based on 20 (15) oil exporters and 29 (19) non-oil commodity exporters at end 2017 (2000). 43 volatility, which may result in detrimental output volatility. Furthermore, many non-oil exporting countries, which are likely to face an enduring deterioration in the terms of trade going forward, will need to adjust further through the continuation of procyclical policy in order to maintain or restore fiscal and debt sustainability. In these countries, further fiscal consolidation will need to be carefully managed to minimize social and economic costs. Figure 39: Overview of Debt Risks in Commodity-Exporting EMDEs, 2017 Source: LIC DSF risk of external public debt distress; when not available, the DEC Fiscal Space Database Index of Sovereign Foreign currency long-term sovereign debt ratings is used (ranging 1-21), after classification of countries in 3 categories (Index<7= low; 714, high). It should be underlined that, particularly in commodity-rich countries, the approach to sustainability could usefully take into account SWF and the depletion of natural resources, although data constraints limit this analysis on a cross-country basis (Box 5). Box 5: Taking into Account Financial and Natural Assets When Assessing Fiscal Space in Commodity-Exporting Countries In assessing fiscal space, net debt may be as relevant as gross debt because of the magnitude of assets held in sovereign wealth funds and other financial instruments, particularly for oil and metals exporters. In oil-exporting countries, net debt was almost nil in the wake of the commodity boom but has started increasing again after the price downturn (Figure 40). As pointed out in the literature (e.g., Spatafora and Samake, 2012), in natural-resource rich countries, a measure of net wealth which also includes assets in the ground could be used to assess long-term debt sustainability, but data availability represents a significant constraint. Similarly, many natural resource-rich countries are characterized by the importance of foreign investment in the natural resource sector, and as a result, outflows of profits and income may be significantly larger than inflows, which may lead to high discrepancies between GDP and national income. Furthermore, because of the exhaustibility of natural resources, the depletion of natural resources associated to the production process is an important dimension of sustainability (Hamilton and Ley, 2010). As shown in Figure 41 for a few countries where data are available, public debt burdens are higher when accounting for this depletion. 44 Figure 40: Among Commodity Exporters, Debt is Figure 41: … but Depletion of Natural Resources Lower in Net Terms … Matters. General Government Debt (% of Gross Public Debt, 2000-2016 150 GDP) 90 80 100 70 Percent (%) Percent (%) 60 50 50 40 30 0 20 2000 2011 2017 2000 2011 2017 2000 2011 2017 2000 2011 2017 10 -50 0 Average Oil Average Average Average 2000 2011 2016 2000 2011 2016 2000 2011 2016 2000 2011 2016 Exporters Non-Oil High High Exporters Dependent Dependent Oil Non-Oil Argentina Brazil Indonesia Russia Gross Debt Net Debt Public debt/GDP % Public Debt/ Net GNI % Source: WEO, WDI. Source: WEO, WDI For commodity exporters, collateralized debt contracted by the central government and state- owned enterprises may also be an important factor weighing on future revenue streams and thus fiscal space. Commodity exporting countries, particularly state-owned enterprises operating in the oil and gas sectors, have historically been the main users of collateralized borrowing.46 Unfortunately, data scarcity limits the ability to examine the fiscal vulnerabilities that may arise from high amounts of collateralized debt. Available data suggests that by end-2017, 21 percent of the stock of commercial loans (excluding project financing) was collateralized in oil exporting countries, compared to 23 percent in other commodity exporters. This percentage is higher for low-income countries (29 percent for non-oil exporting countries), but commercial borrowing amounts to a smaller fraction of external financing in these countries.47 Yet, borrowing against future commodity revenue streams on commercial terms may be a major source of fiscal vulnerability in the event of a negative shock, as illustrated by the fiscal and debt crisis that hit Chad following the reversal in global oil prices after 2014. 4.2.3. Contingent Liabilities Weigh on Fiscal Space Fiscal space may also be constrained by contingent liabilities arising from the exposure of state- owned enterprises (SOEs) and the financial sector to the commodity export sector. Data scarcity on SOEs’ financial position, exposure to the commodity export sector, and debt in many countries48 (beyond external debt explicitly guaranteed by the central government) is a significant constraint to incorporating this in this assessment of fiscal vulnerabilities. Nevertheless, a negative commodity price shock may directly affect the fiscal position of the government through various 46 This is starting to evolve, with more non-commodity exporting countries restoring to collateralized borrowing in recent years (e.g., China). Source: World Bank Global Macro and Debt Analytics Unit. Coverage is 50 commodity exporters. 47 Source: World Bank Global Macro and Debt Analytics Unit. 48An ongoing survey launched by the Global Macro and Debt Analytics Unit of the World Bank Group (2018) is expected to address the issue of data scarcity on SOE debt. 45 channels in relation to the SOE and public financial sector. These include lower tax revenue and dividend payments, but also growing needs for financial support, notably through budgeted and unbudgeted transfers, debt servicing on account of SOEs, and recapitalization needs. In turn, the fall in commodity revenue may trigger the accumulation of domestic arrears by the central government which may directly impact the financial situation of some SOEs and banks in the sector. Hence substantial contingent liabilities may arise from the exposure of state-owned enterprises, public banks and systemic banks to the commodity export sector, notably in highly dependent countries, as illustrated by the recent case of Angola (Box 6). Box 6: Contingent Liabilities from Exposure to the Commodity Sector Weigh on Fiscal Space: The Experience of Angola Given its significantly large size and systemic role in Angolan economy, any debt accumulated by Sociedade Nacional de Combustiveis de Angola (Sonangol) can be considered an Implicit Contingent Liability (ICL) to the sovereign. Sonangol, Angola’s national oil company and SSA’s largest SOE is a major player in the domestic economy with total assets valued at US$ 45.9 billion in 2016 (about half of the country’s GDP). In addition, Sonangol has a wide network of subsidiaries and related companies in the non-oil sector (Corkin and Bank 2017). For instance, it is a major shareholder in Banco Económico (owning about 40% of total shares) and Banco Angolano de Investimentos (owning about 8.5% of total shares), two of the biggest private banks in Angola. As such, any debt this SOE amasses can be categorized as an implicit contingent liabil ity for the central government, i.e., “a moral obligation or expected responsibility of the government that is not established by law or contract but instead is based on public expectations, political pressures, and the overall role of the state as society understands it” (Polackova 1998, 2). The Angolan economy was significantly impacted by the 2014-2015 drop in oil prices. During the 2011-2013 period, Angola was heavily dependent on oil which accounted for 98 percent of exported goods, 79 percent of fiscal revenues and about a third of its GDP. As a result of this dependence, the end of the commodity super-cycle wreaked havoc on the Angolan economy: the fiscal and current account balances deteriorated substantially, the non-oil GDP growth decelerated significantly and inflation accelerated rapidly (Figure 42). Figure 42: All the Macroeconomic Indicators Worsened After the Drop in Oil Prices in 2014 46 Source: IMF WEO (October 2017). Amidst the decelerating economy, the implicit fuel subsidies, and the failed restructuring plans, Sonangol’s financial accounts deteriorated. Sonangol’s revenues dropped from US$41.7 billion in 2013 to US$14 billion in 2016 and net profits decreased from US$6.7 billion in 2014 to only about US$75 million in 2016 (Figure 43). Several factors were behind these worsening financial conditions. First, the decelerating economic activity and low oil prices adversely affected the company’s revenues after 2014. Second, Sonangol had to sell domestic fuel below its cost-recovery prices, without being compensated by the government because subsidized fuel prices had not been adjusted upward after end-2015 despite the gradual recovery in international oil prices in 2015. Third and finally, Sonangol announced restructuring plans for the first time in June 2016 aimed at increasing revenue, efficiency and transparency. However, these plans did not materialize resulting in underinvestment, low maintenance, arrears with suppliers and oil majors and contractors, among other issues. Figure 43: Sonangol‘s Financial Performance Figure 44: … Resulting in a Higher Worsened Since 2014… Accumulated Debt Source: Sonangol consolidated annual reports. Source: IMF article IVs (2015 and 2016). Sonaongol’s financial woes increased fiscal risks on the sovereign, thus inducing the government to intervene to avoid its collapse. As a result of its financial problems, the stock of debt accumulated by Sonangol increased rapidly since 2014, adding pressure on the central’s government growing debt (Figure 44). To prevent such risks of materializing, the government intervened and injected about US$ 3.8 billion in financing to Sonangol despite its own fiscal deficit (IMF 2016b). Consequently, the debt of Sonangol decreased significantly in 2017 based on preliminary data, with media outlets reporting that it dropped from US$ 9.8 billion in 2016 to US$ 4.8 billion in 2017.49 This was achieved at a significant cost for the government. 49https://allafrica.com/stories/201803010216.html; http://www.angop.ao/angola/en_us/noticias/economia/2018/1/9/Sonangol- owes-USD-billion-State,4ff038b1-2b70-49d8-9771-9ed9f99108bc.html 47 In the wake of the global crisis and the commodity downturn, there has been a substantial deterioration in the situation of the financial sector of the average commodity exporter. While detailed data by bank are not readily available on a cross-country basis, the percentage of non- performing loans (NPLS) out of total loans has increased steadily, especially in oil exporting countries (Figure 45 and Figure 46). The lack of data prevents a more granular cross-country analysis of the direct impact of the price shock on the situation of financial institutions, and the magnitude of associated fiscal risks. Nevertheless, these findings suggest that available fiscal space may be eroded by growing contingent liabilities from public and systemic banks. This needs to be taken into account in a country-specific analysis of fiscal vulnerabilities associated to a country’s exposure to the commodity-export sector. Figure 45: NPLs Increased Rapidly After Figure 46: … and Middle and High- the Price Downturn, Especially for Oil… Income Commodity Exporters Non performing loans in percent of total gross loans Non perfoming loans in percent of total gross loans 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Low income Middle and high income Oil Non Oil Source: FinStat. To illustrate the magnitude of the fiscal cost of contingent liabilities in the absence of detailed data, a stock-flow adjustment calculation is used. Stock-flow adjustments, measured as the discrepancy between the change in public debt, and the primary fiscal deficit and other identified drivers of debt accumulation, is used as a proxy to measure the realization of contingent liabilities (Abbas et al. 2011; Weber 2012). In this study, these adjustments are calculated for 27 commodity exporters based on data from the LIC-DSF between 2006 and 2017. 48 Figure 47: Unaccounted Residuals, Including Contingent Liabilities, Pushed Debt Up in Several Commodity-Exporting Countries Average Debt Accumulation and Unexplained Residual in Selected Commodity Exporting Countries, 2006-2017 South Sudan Burundi Togo Zimbabwe Mauritania Liberia Benin Chad Rwanda Malawi Guinea-Bissau Ethiopia Niger Mali Uganda Gambia Congo, Democratic Republic Cote d'Ivoire Afghanistan Guinea Burkina Faso Madagascar Mozambique Senegal Central African Republic Tanzania Sierra Leone -10 0 10 20 30 Average residual Percent (% of GDP) Average change in debt Source: IMF-WB LIC DSF, Author’s calculations. In half of these countries, estimated materialization of contingent liabilities has accounted for debt accumulation equivalent to 2 to 5 percent of GDP in 2006-2017. Most countries have registered positive debt accumulation, explained by changes in the primary deficit, real GDP growth, exchange and interest rate movements, and other factors (such as sales of assets). The residual is assumed to be a reasonable proxy for the materialization of contingent liabilities or other sources of borrowing needs that are unaccounted for, beyond statistical discrepancies. As shown in Figure 47, this residual is positive in most countries, and can be large, driving debt substantially up, notably in South Sudan, Burundi, Togo, Zimbabwe and Mauritania. While this remains a rough proxy, this indicator suggests that contingent liabilities are not a negligible source of fiscal vulnerabilities in commodity exporting countries, driving debt accumulation and thus weighing on fiscal space. 49 4.3. Simulating the Impact of the Next Commodity Price Shock in Ghana, Angola and Kyrgyz Republic While fiscal vulnerabilities have increased, a number of commodity exporters face substantial downside risk to macroeconomic and debt stability from potentially lower-than-forecasted commodity prices. Price projections, as discussed in Section 4.1, suggest favorable price developments for oil-exporters, but downside price adjustments for non-energy commodities. Nevertheless, oil exporters have historically faced greater volatility in prices than non-oil exporters. The section aims to shed light on the impact of potential lower-than-expected prices on fiscal and debt positions as described in 4.2. in selected countries. Stochastic projections of the fiscal and public debt position, in contrast to deterministic scenarios, account for the uncertainty and volatility underlying future commodity prices. Aside from deterministic approaches to projecting future macroeconomic and fiscal stability, several stochastic approaches, such as vector auto-regression (VAR) models have been discussed in the literature. The approach used in this study follows Berti (2013). Using the Choleski decomposition of a historical variance-covariance matrix, random shocks, assumed to follow a joint normal distribution, are created and applied to a baseline and shock scenario going forward. The methodology is described in Box 7. The stochastic methodology used in this paper considers country-specific historical experience, projecting future debt paths based on historical shocks and the correlation of the shocks of different variables (commodity price index, real GDP growth, total government revenue, and nominal exports). In the model50, shocks are applied to four variables, i.e. the country specific commodity price index (ComPI), real GDP growth, nominal exports, and total government revenues. The methodology is applied to three countries which are representative of exposure to different types of commodity exports and degrees of commodity dependence, i.e. Angola (oil), Ghana (non- oil), and Kyrgyz Republic (non-oil).51 The elasticity of output to the country-specific price index is highest in Ghana and Angola, and substantially lower for Kyrgyz Republic. The country-specific historical commodity price volatility, measured as the standard deviation of the index annual growth rates (as displayed in Figure 2 of Section 2), is used for the definition of future shocks, and varies across the three countries. Angola historically has faced the highest price volatility with 38 percent, in line with the finding that oil exporters are exposed to more volatile prices than non- energy exporters. For Ghana and Kyrgyz Republic, price volatilities have reached 16 and 15 percent respectively. 50 The World Bank’s Fiscal Sustainability Analysis Tool was used for the analysis. 51To estimate the elasticity of GDP to commodity prices, a linear regression of the natural logarithm of the commodity price index on the natural logarithm of output was performed. To estimate the elasticity of government revenue to output, a linear regression of the natural logarithm of output on the natural logarithm of revenue was performed. Those elasticities reflect the impact of commodity prices fluctuations on output and government revenues (via export earning, royalties and taxes), including the government’s fiscal policy reaction to commodity price cycles. 50 Box 7: Stochastic Projections of the Fiscal and Public Debt Positions with Commodity Price Volatility In a first step, the model transforms the historical data series, covering the period 2000 – 2017, of those variables into annual shocks δ , where t represents the year, and x the respective variable to be shocked: δ = − −1 , δ = − −1 ,δ = − −1 ,δ = − −1 Then a variance-covariance matrix of historical shocks is generated: Compi Growth Exports Revenue Covar (Compi, Compi Var (Compi) Covar (Compi, Growth) Covar (Compi, Exports) Revenue) Covar (Compi, Covar (Growth, Covar (Growth, Growth Var (Growth) Growth) Exports) Revenue) Covar (Compi, Covar (Growth, Covar (Revenue, Exports Var (Exports) Exports) Exports) Exports) Covar (Compi, Covar (Growth, Covar (Revenue, Revenue Var (Revenue) Revenue) Revenue) Exports) A Monte Carlo simulation generates 1000 random vectors of uncorrelated shocks εx t (where t represents the year, and x the respective variable) over the period 2018 – 2023 from a joint normal distribution with zero mean a and variance- covariance identical to the historical shocks. The Choleski method decomposes the matrix and correlates the previously uncorrelated shocks by finding a triangular matrix A, such that the transposition of A x A is equal to the matrix of our variables. If matrix A is then multiplied with a vector of random uncorrelated shocks, they become correlated according to the original variance – covariance matrix. The stochastic shocks generated through the methodology described above are then applied to the baseline and one alternative scenario of gross public debt in percent of GDP : = ̅̅̅̅̅̅̅̅̅̅̅ + ε ̅̅̅̅̅̅̅̅̅̅̅ , with baseline commodity price index; = ̅̅̅ baseline GDP real growth rate at year t; = EXP + ε , with ̅̅̅ ̅̅̅̅̅ + ε ̅̅̅̅̅̅ , with ̅̅̅̅̅̅̅ ̅̅̅̅̅̅ = baseline nominal exports at year t; = + ε , with = baseline revenues at year t. The new variables obtained over the projection period, 2018 – 2023, are then plugged into the model/tool and generate a new path for the evolution of gross public debt relative to GDP. Finally, the steps described above were repeated 1000 times, therefore creating 1000 alternative paths for the public debt ratio going forward. From these paths, the tool derived percentiles and based on those constructed fan charts for the projected variables. In addition to the baseline scenario based on current price forecasts, the analysis assumes an alternative scenario in which the commodity price index will be one standard deviation below the projection. The shock in the annual change of commodity prices feeds into the model by decreasing output and revenue according to estimated price elasticities of those variables (footnote 47). Section 2 shows that price shocks historically have been persistent and long-lasting, especially for oil-exporters. Hence the shock, while not large in magnitude, is applied over the entire projection period. The results of the stochastic projections - the fans around the baseline and alternative scenario – need to be interpreted taking into account the substantial reduction in public debt ratios and debt 51 servicing costs of the 2000s. All three countries under consideration displayed public debt ratios of more than 100 percent relative to GDP in 2000. With the commodity boom, as well as debt relief provided under the HIPC and MDRI to Ghana, public debt levels had decreased substantially, below 50 percent of GDP for the three countries by 2011, and as low as 16 percent of GDP in Angola by 2007. The associated reduction in the cost of servicing lower debt created fiscal space for productive growth-enhancing public spending, such as infrastructure investment. Hence the random shocks drawn from the period 2000 – 2017 and used for the stochastic projections reflect a high variance and covariance in public debt, real GDP growth, government revenues, and nominal exports. This may lead to overestimating these linkages to some extent when exploring future shocks. Not surprisingly, lower than expected commodity prices in the alternative scenario compared to the baseline scenario lead to higher than projected public debt levels in all three countries (Figure 48). Angola records the largest increase in public debt relative to GDP in the wake of a one standard deviation shock to their country specific commodity prices index growth. This result is in line with the finding that commodity price elasticity of output and revenue has been substantially higher for Angola than for Ghana and Kyrgyz Republic. Hence a one standard deviation shock to the country specific commodity price index leads to a higher debt ratio relative to GDP. Ghana’s public debt relative to GDP increases at a lower rate in the alternative scenario compared to Angola, but steadily over the projection period. In the Kyrgyz Republic, which has both the lowest commodity price elasticity of output and the lowest historical volatility in the annual change of the commodity price index among the three countries, debt to GDP under the alternative scenario changes only marginally compared to the baseline. Stochastic projections for Angola, Ghana and Kyrgyz Republic allow to assign probabilities to future paths of public debt relative to GDP. The fan charts in Figure 48 show the probabilities for debt to GDP increasing above or decreasing below a certain level or remaining within a certain span. The large variance in the distribution of public debt ratios, as mentioned above, is evident in each of the countries. 52 Figure 48: Stochastic Public Debt to GDP Paths In the case of Angola, there is a 75 percent probability that the debt ratio will be higher than currently forecasted. Over the projection period, the probability of a path reflecting the baseline scenario, which assumes a decrease in public debt in Angola, is only around 25 percent, while the chance of a higher debt ratio than the baseline assumption is 75 percent. A negative price shock would bring the debt path in the center of the distribution. While this result for Angola must be interpreted with caution because of the large variance of macroeconomic variables over the past, Angola has historically faced substantial volatility of the commodity price index. Hence stochastic projections point out that the likelihood of future shocks to public debt is therefore high, and a cautious fiscal policy stance is warranted. In Ghana, the probability of public debt reaching a level relative to GDP higher than projected in the baseline scenario is above 50 percent. Furthermore, under the downside alternative scenario of a price shock, the debt ratio would be higher and the probability of exceeding that level by 2023 would be about 50 percent. Even though the probability of an increase of public debt above 80 53 percent of GDP by 2023 is estimated to be low, at around 10 percent, these results underscore the importance of rebuilding fiscal and external buffers. For the Kyrgyz Republic, the baseline scenario reflecting a moderate decrease in public debt relative to GDP seems to reflect the most likely path by 2023, and a negative price shock would not derail this path significantly. The probability of public debt increasing above 70 percent of GDP by 2023 is only 10 percent, while at the same time the probability of a substantial reduction shows a similarly low likelihood. Based on historical movements, lower commodity prices would not drastically modify the forecasted debt path, in a context of lower commodity-induced macroeconomic instability and lower dependence on the commodity sector. Future stochastic debt paths highlight debt sustainability risks and underscore the importance of reforms aimed at reducing macroeconomic volatility by creating fiscal space for countercyclical fiscal policy. For Angola and Ghana, the probability of public debt exceeding baseline assumptions over the medium-term is high, 75 percent and 50 percent respectively. Prudent fiscal policy and debt management will be crucial for protecting macroeconomic and debt sustainability. These countries are highly dependent on the commodity sector and have been facing high macroeconomic volatility in the wake of past shocks, hence the high level of downside risks to debt. This underscores the importance of reforms aimed at reducing macroeconomic volatility by creating fiscal space for countercyclical fiscal policy. While this is important for Kyrgyz Republic as well, risks of divergence from the baseline scenario seem to be less likely. Data limitations for low and middle-income commodity exporting countries are substantial and impede comprehensive analysis and projections. The historical and projected data used as the baseline scenario for the stochastic analysis are taken from the IMF WEO database. Data on commodity revenue is only available for a subset of countries. In addition, data on the size and composition/characteristics of public debt is limited. Gross debt numbers are in many cases only available from the early 2000s. Market based debt as well as debt from new bilateral creditors has substantially increased in many developing and emerging countries, however, little information is available on the terms and costs of these new instruments. Such information would improve the quality of deterministic and stochastic projections, allowing for more accurate projections of the maturity profile of debt and debt servicing costs. 54 Appendix 1: Overview of Commodity Exporters Following GEP (2018), an EMDE is classified as an oil exporter when, on average in 2008-12, exports of crude oil accounted for 20 percent or more of total exports. A country is classified as non-oil commodity exporter when, on average in 2008-2012, either (i) total commodities exports accounted for 30 percent or more of total exports, or (ii) exports of any single commodity other than oil and natural gas accounted for 20 percent or more of total exports. Table A1.1 shows that there are 26 oil-exporting EMDEs and that the commodity sector represents a large share of their economies. Columns (1) and (2) show that, in the period 2008-2012, the cross-country average of commodity exports as a percent of exports of goods and total exports are 75.9% and 71.9%, respectively. Columns (3) and (4) show that most of these countries are also net commodity exporters (commodity exports minus imports). We further classify a country as highly oil- dependent when net commodity exports represent more than 10% of GDP (column (5)). Oil exporters that do not meet that criteria are classified as moderate oil exporters. On average, net exports represented 24.1% of oil-exporters GDP in the period 2008-2012. Table A1.1: The Size of the Commodity Sector in Oil-Exporting EMDEs (Average 2008 – 2012) Table A1.2 displays the 43 non-oil-exporting EMDEs. Non-oil exporters are in general less dependent on the commodity sector than oil exporters. However, columns (1) and (2) show that the commodity sector is still a key sector in these economies, representing 54.2% of exports of goods and 40.9% of total exports. Columns (3) and (4) show that 20 non-oil commodity exporters are not net commodity exporters. Moreover, while most oil-exporters are highly dependent on the oil sector, only 8 non-oil exporters are highly dependent on the commodity sector. Table A2.1 shows that crude oil or natural gas represented the largest share of total imports in 40 out of 43 non-oil exporters in 2008-12, and crude oil alone accounts on average for 17% of total imports in non-oil exporters. 55 Table A1.2. The Size of the Commodity Sector in Non-Oil-Exporting EMDEs (Average 2008 – 2012) Table A1.3 summarizes the average economic exposure of commodity-exporting EMDEs to commodity cycles during 2002-2016 for the entire sample (26 oil exporters and 43 non-oil exporters). Commodity-exporting EMDEs are remarkably exposed to commodity cycles. Commodity exports represents 78.5% (41.1%) of total exports in highly (moderately) dependent oil exporters, and 58.2% (35.6%) in highly (moderately) dependent non-oil exporters. Net commodity exports represent 33.1% (4.7%) of GDP in highly (moderately) dependent oil exporters, and 13.7% (-1.3%) in highly (moderately) dependent non-oil exporters. Table A1.3: Economic Exposure to Commodity Cycles: Average 2002 - 2016. Commodity Exports Net Commodity Exports (% of total exports) (% of total trade) (% of GDP) Oil Exporters High dependency 78.5 40.5 33.1 Moderate dependency 41.1 7.6 4.7 Non-Oil Exporters High dependency 58.2 15.9 13.7 Moderate dependency 35.6 -1.1 -1.3 Source: Authors’ calculations based on the UN-ComTrade database. 56 Commodity-exporting EMDEs have a highly concentrated export structure. Besides having large commodity sectors, commodity-exporting EMDEs are highly specialized in a small number of commodities. Tables A2.2 and A2.3 describe the top three commodity exports for oil and non-oil exporters during 2008-2012, respectively. Oil and natural gas exports represents on average 67% of commodity exports. More than 90% of export revenues are composed by oil receipts in Angola, Azerbaijan, Iraq, Libya and Venezuela. Non-oil exporters are more diversified but still rely on a relatively small number of products. In the period 2008-12, the top 3 commodities represented 43% of total exports. Gold is a particularly important export for Ghana (47.8%), Guyana (38.4%), Kyrgyzstan (40.5%), Tanzania (29.1%) and Mali (72.2%). Other important non-oil commodities are copper and iron. Copper constitutes 24.7%, 54.4%, 55.6%, 24.5% and 74.8% of exports in Armenia, Chile, Lao, Peru and Zambia, respectively. Iron is an important commodity for Brazil (12%), Mauritania (38.5%), South Africa (6.4%) and Ukraine (4.2%). 57 Appendix 2: Structure of Commodity Exports Table A2.1: Top 3 Commodity Imports for Non-Oil Commodity Exporting EMDEs (% of total imports - average 2008 - 2012) TOP 1 TOP 2 TOP3 Argentina crude oil 5.3 natural gas 2.4 iron 1.3 Armenia natural gas 9.0 crude oil 8.4 diamond 2.7 Belize crude oil 14.8 natural gas 1.5 wheat 1.0 Benin crude oil 11.7 rice 10.5 chicken 8.6 Botswana diamond 14.6 crude oil 12.3 sugar 0.7 Brazil crude oil 12.9 natural gas 2.2 coal 2.0 Burkina Faso crude oil 22.4 rice 3.8 dap 2.1 Burundi crude oil 9.8 soybeans 7.0 rice 1.2 CAR sugar 3.8 palm oil 1.9 aluminum 1.4 Chile crude oil 19.7 natural gas 2.7 coal 1.5 Costa Rica crude oil 11.3 maize 1.0 copper 0.8 Cote d'Ivoire crude oil 27.9 rice 7.2 wheat 2.0 Ethiopia crude oil 18.5 wheat 4.1 palm oil 2.9 Ghana crude oil 5.4 rice 2.8 wheat 1.5 Guatemala crude oil 16.7 natural gas 1.7 maize 1.3 Guyana crude oil 28.4 natural gas 2.1 wheat 1.5 Honduras crude oil 21.8 maize 1.4 soybean 1.0 Indonesia crude oil 20.5 wheat 1.2 aluminum 1.0 Kenya crude oil 22.8 palm oil 3.7 maize 1.9 Kyrgyzstan crude oil 12.7 natural gas 2.1 wheat 1.9 Lao crude oil 22.3 gold 2.0 dap 0.8 Madagascar crude oil 16.0 rice 2.4 sugar 2.0 Malawi crude oil 10.0 urea 6.9 dap 3.7 Mali crude oil 25.7 urea 2.3 rice 1.7 Mauritania crude oil 24.9 wheat 4.5 rice 1.8 Morocco crude oil 16.6 natural gas 4.4 wheat 3.0 Mozambique crude oil 15.5 aluminum 4.3 rice 2.5 Namibia crude oil 9.8 diamond 3.1 copper 2.1 Nicaragua crude oil 20.7 rice 1.2 wheat 1.0 Niger crude oil 11.0 rice 5.9 sugar 1.7 Paraguay crude oil 13.4 dap 3.4 tobacco 1.2 Peru crude oil 14.8 wheat 1.6 maize 1.5 Rwanda crude oil 5.9 sugar 2.4 palm oil 1.8 Senegal crude oil 25.2 rice 7.2 wheat 2.7 South Africa crude oil 19.7 diamond 0.7 coal 0.6 Suriname crude oil 18.7 chicken 1.3 aluminum 1.2 Tanzania crude oil 28.0 wheat 3.0 palm oil 2.1 Togo crude oil 15.3 wheat 1.6 dap 1.5 Uganda crude oil 19.6 palm oil 3.7 wheat 2.3 Ukraine natural gas 15.7 crude oil 12.4 coal 3.0 Uruguay crude oil 21.8 dap 1.4 urea 1.1 Zambia copper 12.5 crude oil 11.1 urea 2.1 Zimbabwe crude oil 13.3 dap 7.0 maize 3.0 Source: Authors’ calculations based on UN-ComTrade. 58 Table A2.2: Top 3 Commodity Exports for Oil-Exporting EMDEs (% of total exports - average 2008 - 2012) TOP 1 TOP 2 TOP3 Albania crude oil 13.5 copper 3.6 aluminum 1.7 Algeria crude oil 59.9 natural gas 37.5 sugar 0.2 Angola crude oil 96.7 diamond 1.9 coffee arabica 0.0 Azerbaijan crude oil 92.8 natural gas 1.5 sugar 0.6 Bahrain crude oil 66.0 aluminum 12.2 iron 7.0 Bolivia natural gas 42.4 zinc 10.6 soybean meal 4.6 Cameroon crude oil 21.0 cocoa 19.8 sawn wood 11.3 Colombia crude oil 40.7 coal 14.5 gold 4.6 Congo, Rep crude oil 74.5 natural gas 1.4 logs 0.3 Ecuador crude oil 56.1 banana 10.7 shrimps 4.8 Iran crude oil 65.7 natural gas 4.0 iron 1.0 Iraq crude oil 99.7 potassium 0.0 banana 0.0 Kazakhstan crude oil 65.0 copper 4.7 natural gas 3.6 Kuwait crude oil 89.8 natural gas 3.4 urea 0.5 Libya crude oil 91.6 natural gas 5.9 urea 0.4 Malaysia crude oil 9.8 natural gas 7.4 palm oil 6.6 Myanmar natural gas 32.9 rice 3.3 rubber 2.8 Nigeria crude oil 83.9 natural gas 4.6 rubber 2.9 Oman crude oil 62.3 natural gas 8.2 urea 1.8 Qatar natural gas 53.1 crude oil 33.4 urea 1.6 Russia crude oil 50.1 natural gas 13.5 coal 2.3 Saudi Arabia crude oil 84.0 natural gas 2.6 urea 0.5 TTG crude oil 36.7 natural gas 25.6 iron 3.8 UAE crude oil 38.3 gold 6.5 diamond 5.1 Venezuela crude oil 89.5 aluminum 0.7 iron 0.6 Yemen crude oil 79.8 natural gas 9.6 banana 0.2 Source: Authors’ calculations based on UN-ComTrade. Table A2.3: Top 3 Commodity Exports for Non-Oil Commodity Exporting EMDEs (% of total exports - average 2008 - 2012) TOP 1 TOP 2 TOP3 Argentina soybean meal 12.1 soybean oil 6.1 crude oil 5.9 Armenia copper 24.7 diamond 9.1 aluminum 6.8 Belize crude oil 33.8 sugar 12.4 banana 11.9 Beninys cotton 29.7 chicken 11.2 rice 6.4 Botswana diamond 71.5 nickel 9.1 copper 1.6 Brazil iron 12.1 crude oil 9.5 soybeans 6.4 Burkina Faso gold 57.0 cotton 22.8 crude oil 1.2 Burundi coffee arabica 37.6 gold 30.6 tea 6.5 CAR diamond 56.1 sawn wood 10.2 cotton 5.1 Chile copper 54.4 gold 1.6 iron 1.4 Costa Rica banana 6.7 coffee arabica 3.3 palm oil 1.4 Cote d'Ivoire crude oil 27.0 cocoa 23.2 rubber 6.6 Ethiopia coffee arabica 30.2 gold 5.9 meat sheep 2.0 Ghana gold 47.8 cocoa 17.4 crude oil 8.5 Guatemala coffee arabica 9.0 sugar 6.9 banana 5.2 Guyana gold 38.4 rice 14.7 aluminum 12.1 59 TOP 1 TOP 2 TOP3 Honduras coffee arabica 26.6 natural gas 5.5 palm oil 5.4 Indonesia coal 12.3 natural gas 10.2 palm oil 8.8 Kenya tea 20.4 coffee arabica 3.9 crude oil 3.8 Kyrgyzstan gold 40.5 crude oil 3.7 cotton 1.7 Lao copper 55.6 gold 9.2 maize 5.6 Madagascar shrimps 6.4 crude oil 5.7 nickel 1.2 Malawi tobacco 55.9 sugar 7.3 tea 6.1 Mali gold 72.7 cotton 10.6 dap 2.9 Mauritania iron 38.5 gold 19.8 copper 12.0 Morocco dap 6.8 crude oil 2.3 orange 2.1 Mozambique aluminum 36.6 tobacco 6.7 natural gas 4.5 Namibia diamond 20.3 zinc 6.9 copper 5.3 Nicaragua coffee arabica 15.4 gold 9.5 shrimps 5.0 Niger gold 8.8 crude oil 4.6 sugar 1.3 Paraguay soybeans 22.8 soybean meal 5.6 maize 4.6 Peru copper 24.6 gold 21.5 crude oil 7.9 Rwanda tea 21.0 tin 19.3 coffee arabica 16.8 Senegal crude oil 19.8 gold 8.7 groundnut oil 1.9 South Africa platinum 11.1 coal 7.0 iron 6.4 Suriname crude oil 9.1 rice 2.0 sawn wood 0.1 Tanzania gold 29.1 tobacco 3.5 coffee arabica 3.3 Togo cotton 4.7 dap 3.8 cocoa 3.8 Uganda coffee arabica 19.2 crude oil 4.9 tea 3.4 Ukraine iron 4.2 crude oil 3.6 maize 2.7 Uruguay soybeans 10.1 rice 6.8 beef 4.3 Zambia copper 74.8 maize 1.7 sugar 1.7 Zimbabwe nickel 20.6 tobacco 13.7 gold 7.8 Source: Authors’ calculations based on UN-ComTrade. 60 Appendix 3: Commodity Booms and Busts Episodes 1960-2017 Table A3.1: Oil exporting EMDEs 1st Boom Duration 2nd Boom Duration Albania 1974-1986 12 2004-2015 11 Algeria 1974-1988 14 2000-2015 15 Angola 1973-1986 13 2004-2015 11 Azerbaijan 1973-1986 13 2004-2015 11 Bahrain 1974-1986 12 2004-2015 11 Bolivia 1973-1986 13 2000-2012 12 Cameroon 1976-1981 5 2005-2018 13 Colombia 1974-1987 13 2004-2018 14 Congo, Rep. 1973-1986 13 2004-2015 11 Ecuador 1974-1986 12 2005-2016 11 Iran 1974-1986 12 2003-2015 12 Iraq 1973-1986 13 2004-2015 11 Kazakhstan 1973-1986 13 2004-2015 11 Kuwait 1973-1986 13 2004-2015 11 Libya 1973-1986 13 2000-2015 15 Malaysia 1974-1986 12 2003-2015 12 Myanmar 1974-1987 13 2000-2009 9 Nigeria 1973-1986 13 2003-2015 12 Oman 1974-1986 12 2000-2015 15 Qatar 1974-1988 14 2000-2012 12 Russia 1974-1986 12 2000-2015 15 Saudi Arabia 1973-1986 13 2004-2015 11 TTO 1974-1987 13 2000-2015 15 UAE 1973-1986 13 2004-2016 12 Venezuela 1974-1986 12 2004-2015 11 Yemen 1974-1986 12 2000-2015 15 Note: As in Cespedes and Velasco (2014), a commodity boom is defined as an episode during which the index reaches a level of at least 25% above its trend. An episode ends when the commodity price index comes back to a level lower than 10% above its trend. 61 Table A3.2: Non-Oil Commodity Exporting EMDEs 1st Boom Dur 2nd Boom Dur 3rd Boom Dur 4th Boom ur Argentina 1973-1975 2 2007-2018 11 . . . Armenia 1964-1967 3 2005-2018 13 . . . . Belize 1963-1965 2 1973-1978 5 1979-1985 6 2006-2018 12 Benin 1973-1975 2 2010-2015 5 . . . . Botswana 1969-1974 5 1988-1990 2 2004-2016 12 . . Brazil 1973-1978 5 2006-2015 9 . . . . Burkina Faso 1973-1978 5 1980-1985 5 2008-2018 10 . . Burundi 1976-1981 5 2009-2018 9 . . . . CAR 1970-1971 1 1973-1974 1 1980-1981 1 1985-1986 1 Chile 1964-1971 7 2004-2018 14 . . . . Costa Rica 1977-1978 1 2008-2018 10 . . . . Cote d'Ivoire 1969-1970 1 1973-1986 13 2005-2018 13 . . Ethiopia 1976-1981 5 2010-2013 3 2014-2018 4 . . Ghana 1969-1970 1 1973-1988 15 2005-2018 13 . . Guatemala 1963-1965 2 1974-1978 4 1980-1981 1 2006-2018 12 Guyana 1963-1964 1 1973-1976 3 1980-1982 2 2006-2018 12 Honduras 1976-1981 5 2010-2013 3 2014-2018 4 . . Indonesia 1974-1975 1 1977-1986 9 2003-2015 12 . . Kenya 2009-2018 9 . . . . . . Kyrgyzstan 1973-1989 16 2006-2018 12 . . . . Lao 1964-1971 7 2005-2018 13 . . . . Madagascar 1972-1986 14 2011-2015 4 . . . . Malawi 2010-2018 8 . . . . . . Mali 1973-1986 13 2006-2018 12 . . . . Mauritania 1965-1971 6 1973-1975 2 1979-1986 7 2005-2018 13 Morocco 1967-1978 11 1979-1985 6 2005-2018 13 . . Mozambique 2006-2009 3 . . . . . . Namibia 1964-1967 3 1973-1975 2 2005-2018 13 . . Nicaragua 1974-1975 1 1976-1985 9 2010-2018 8 . . Niger 1963-1964 1 1973-1986 13 2006-2018 12 . . Paraguay 1973-1975 2 2008-2018 10 . . . . Peru 1964-1967 3 1973-1975 2 1979-1982 3 2005-2018 13 Rwanda 1977-1978 1 2008-2018 10 . . . . Senegal 1973-1986 13 2005-2018 13 . . . South Africa 1979-1982 3 1984-1991 7 2004-2018 12 . . Suriname 1967-1969 2 1973-1986 13 2005-2015 10 . . Tanzania 1974-1975 1 1977-1978 1 1980-1985 5 2007-2018 11 Togo 1973-1981 8 2008-2018 10 . . . . 62 1st Boom Dur 2nd Boom Dur 3rd Boom Dur 4th Boom ur Uganda 1976-1980 4 2010-2018 8 . . . . Ukraine 1973-1976 3 2005-2015 10 . . . . Uruguay 1973-1975 2 2008-2018 10 . . . . Zambia 1964-1971 7 2005-2018 13 . . . . Zimbabwe 2006-2018 12 . . . . . . Note: As in Cespedes and Velasco (2014), a commodity boom is defined as an episode during which the index reaches a level of at least 25% above its trend. An episode ends when the commodity price index comes back to a level lower than 10% above its trend. 63 Appendix 4: Cyclicality of Fiscal Policy in Non-Oil Commodity-Exporting EMDEs : 2002-07;2008-11 and 2015-16 Evolution of Cyclicality in Non Oil Commodity Exporting EMDEs MDG NAM RWA MAR GTM MOZ MRT ZMB GHA GUY HND MWI UGA ARG UKR URY BRA KEN KGZ TOG BEN CHL NER BFA CAF ETH PRY SUR TZA BEL SEN ZAF MLI PER IND BDI CIV NIC CRI 8.00 6.00 4.00 2.00 0.00 -2.00 -4.00 -6.00 -8.00 Cyclicality in 2002-2007 Cyclicality in 2015-2016 64 Evolution of Cyclicality in Non Oil Commodity Exporting EMDEs MDG NAM RWA MAR GTM MOZ GHA GUY HND MWI MRT UGA ZMB ARG UKR URY BRA KEN KGZ TOG BEN CHL NER BFA CAF PRY SUR ETH TZA BEL SEN ZAF MLI PER IND BDI CIV NIC CRI 8.00 6.00 4.00 2.00 0.00 -2.00 -4.00 -6.00 Cyclicality in 2008-2011 Cyclicality in 2015-2016 65 Appendix 5: Estimating the Efficiency of Public Investment In this appendix, we estimate the long-run relationship between public investment and output. It is not feasible to use individual time-series models to examine this relationship for commodity exporters because the time dimension is very short (less than 20 years for most countries). To overcome this limitation, we rely on the CCEMG panel data model (Pesaran 2006) that has two main advantages in our context. First, it allows for a country-by-country assessment of the relationship between investment and output, a feature that is very useful to account for country- specific (heterogenous) characteristics. Second, it accounts for unobserved common factors (cross- sectional dependence) among commodity exporters which tend to be affected similarly by changes in commodity prices (Cavalcanti, Mohaddes, and Raissi 2015). The econometric specification is: ∆, = 0, + 1, ,−1 + 2, , + 3, ∆, + 4, ,−1 + 5, ∆, + , + , (1) Where , represents the log of real GDP per capita for country at year ; , is the log of real public investment expenditures per capita; ,−1 denotes the debt-GDP-ratio to account for the fact that a higher stock of public debt in the previous year could hinder GDP growth as it may constraint the government from investing more in the current year (Checherita-Westphal and Rother 2012); , denotes the rule of law index that measures institutional capacity and captures significant political events that might affect growth (armed conflicts, revolutions, etc.); and , represents the cross sectional averages. In equation (1) the coefficient of interest is 2, , and can be interpreted as the per capita GDP growth rate related to a 1 percent increase in per capita real public investment for country . The corresponding estimation results are presented in Table A5 1. Data for all variables are from the IMF’s World Economic Outlook (WEO) database, except for the rule of law index which is taken from the World Governance Indicators database. The Maddala and Wu test as well as the Fisher test show that real GDP per capita and investment per capita are non-stationary across all countries. Meanwhile, the Kao and Westerlund tests show that both series are cointegrated for all countries in the sample. Results are provided in Table A5.1. 66 Table A5.1: Investment Efficiency Levels per Commodity Exporters Country β2,i Efficiency level Country β2,i Efficiency level Algeria -0.019*** Very inefficient Kuwait 0.252 Inefficient Angola 0.402 Inefficient Laos -0.063*** Very inefficient Antigua and Barbuda -0.147 Inefficient Liberia -0.206* Inefficient Argentina 0.412 Inefficient Madagascar 0.05*** Efficient Azerbaijan -0.869 Inefficient Malawi 0.158* Inefficient Bahrain 0.065** Efficient Maldives 0.059** Efficient Barbados 0.016*** Efficient Mali 0.04 Inefficient Belize 0.092 Inefficient Mauritius -0.066*** Very inefficient Benin -0.82* Inefficient Moldova 1.027 Inefficient Bolivia 0.094*** Efficient Mozambique 0.032*** Efficient Botswana -0.108 Inefficient Myanmar 0.292*** Efficient Brazil 0.158 Inefficient Namibia 0.153 Inefficient Burkina Faso 0.117* Inefficient Nicaragua 0.31 Inefficient Burundi 0.019*** Efficient Niger 0.03*** Efficient Cameroon 0.171 Inefficient Nigeria 0.042*** Efficient Cape Verde -0.013*** Very inefficient Oman -0.024*** Very inefficient Central African Republic 0.243 Inefficient Panama 0.539 Inefficient Chad 0.195 Inefficient Paraguay 0.435 Inefficient Chile 0.319*** Efficient Peru 0.04*** Efficient Colombia -0.056*** Very inefficient Qatar -0.118 Inefficient Comoros -0.022*** Very inefficient Russia 0.14*** Efficient Congo -0.007*** Very inefficient Rwanda 0.621 Inefficient Costa Rica -0.067* Inefficient Saint Lucia -0.004*** Very inefficient Cote d'Ivoire 0.277 Inefficient Saint Vincent and the Grenadines 0.005*** Efficient Democratic Republic of Congo 0.17 Inefficient Sao Tome and Principe -0.009*** Very inefficient Dominica 0.271 Inefficient Saudi Arabia -0.056*** Very inefficient Ecuador 0.074 Inefficient Senegal 0.312*** Efficient Egypt -0.134** Very inefficient Seychelles 0.045** Efficient Equatorial Guinea -0.687 Inefficient Sierra Leone 0.738 Inefficient Ethiopia 0.068*** Efficient South Africa 0.053*** Efficient Gabon -0.004*** Very inefficient Sudan 0.359 Inefficient Gambia 0.063 Inefficient Tajikistan -0.016*** Very inefficient Georgia 0.482 Inefficient Tanzania -0.033*** Very inefficient Ghana -0.3 Inefficient Trinidad and Tobago 0.029*** Efficient Guinea 0.013*** Efficient Turkmenistan -0.065*** Very inefficient Guinea-Bissau 0.042 Inefficient Uganda 0.263* Inefficient Guyana 0.335 Inefficient United Arab Emirates -0.122* Inefficient Honduras 0.225 Inefficient Uruguay 0.217 Inefficient Indonesia 0.158*** Efficient Vanuatu -0.003*** Very inefficient Iran 0.09*** Efficient Venezuela 0.025 Inefficient Kazakhstan 0.243 Inefficient Yemen 0.549 Inefficient Kenya -0.459 Inefficient Zambia 0.014*** Efficient Source: Authors’ calculations. Note: , is the estimated coefficient from equation (1). 67 Table A5.2: Relationship between Cyclicality of Fiscal Policy vs. Efficiency of Public Investment All commodity exporters Counter- or a-cyclical capital spending* country Pro-Cyclical capital spending* Very inefficient Inefficient Efficient Very inefficient Inefficient Efficient Algeria Belize Namibia Bahrain Colombia Angola Equatorial Guinea Bolivia Laos Benin Paraguay Chile Congo Argentina Guinea-Bissau Niger Oman Democratic Republic Qatar of Congo Ethiopia Gabon Azerbaijan Kazakhstan Nigeria Saudi Arabia Gambia Sierra Leone Guinea Turkmenistan Brazil Liberia Peru Tajikistan Ghana Uganda Indonesia Burkina Faso Mali Senegal Tanzania Guyana UAE Madagascar CAR Nicaragua South Africa Honduras Venezuela Mozambique Cameroon Rwanda Zambia Kenya Myanmar Chad Sudan Kuwait Russia Costa Rica Uruguay Malawi Trinidad and Tobago Cote d'Ivoire Ecuador Source: Authors’ calculations. Note: * Cyclicality is the correlation between cyclical components of real investment and the real GDP. 68 Appendix 6: Drivers of Procyclicality: Results of Bilateral Regressions When considered individually, most variables in the econometric analysis provided in Section 3.3. are statistically significant and display the expected sign. Table A9-1 displays the impact of each candidate variable on the procyclicality of government expenditure for our sample of commodity- exporting EMDEs during the last commodity cycle. Among structural macroeconomic variables, higher per capita GPD and commodity dependency are associated with more counter-cyclical expenditure, while higher GDP volatility and shock persistence are correlated with a more procyclical fiscal stance. In line with the literature, institutional quality is significant at 1 percent and economically important: a 10 percent increase in the IQ index is associated with a 0.3 reduction in the GDP elasticity of government expenditure. At odds with the literature, checks and balance is positively related to procyclicality. Financial integration is the most important factor among financial-related variables: a 10 percent increase in the index is associated with a 0.09 decrease in procyclicality. In line with Gali and Monacelli (2005) and Mendes and Pennings (2017), more rigid exchange rate regimes are related to more countercyclical fiscal policy. 69 Table A9-1: Country Fixed Effects Panel Regressions. All Commodity-Exporting EMDEs 2000-2016 (Dependent Variable: Cyclical Component (HP-Filter) of Log Real Government Expenditure) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) Cyclical log real GDP 1.04*** 1.53*** 4.10*** 0.90*** 2.22*** 0.52*** 0.82*** 1.01*** 1.19*** 1.37*** 0.85*** 0.60*** 0.59*** (0.07) (0.11) (0.82) (0.05) (0.25) (0.10) (0.08) (0.10) (0.08) (0.09) (0.06) (0.21) (0.12) Interaction between Cyclical log real GDP with Per capita GDP -0.02*** (0.00) Commodity exports -1.73*** (0.28) Commodity index persistence -3.49*** (0.89) GDP volatility 0.00* (0.00) Institutional Quality index -3.12*** (0.56) Checks and balance 0.21*** (0.05) Foreign reserves -0.00 (0.00) Financial depth -0.21 (0.19) Financial integration -0.91*** (0.19) Debt-to-GDP ratio (one year lag) -0.32** (0.13) Enforced fiscal rules dummy 0.63*** (0.18) Monetary independence 0.56 (0.43) ER regime 0.05*** (0.02) Constant -0.00 -0.00 -0.00 -0.00 0.00 -0.00 0.00 -0.00 0.00 -0.00 -0.00 -0.00 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) Observations 1,198 1,202 1,184 1,202 1,002 1,162 938 998 1,120 1,154 1,202 876 1,134 R-squared 0.20 0.22 0.21 0.20 0.21 0.21 0.18 0.22 0.21 0.24 0.21 0.20 0.20 Number of countries 68 68 67 68 60 68 58 66 67 67 68 63 68 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations 70 Appendix 7: Selected Indicators to Measure Fiscal Space in Commodity- Exporting EMDEs A. Fiscal revenue exposure to commodity prices • Correlation between total revenue and commodity price index (-1 to +1) • Correlation between commodity revenue and commodity price index (-1 to +1) B. Long-term fiscal and debt sustainability • Long term sovereign debt rating (1 to 21) • Distance of primary deficit to sustainability (debt-stabilizing) level (Percent of GDP, %) • Cyclically-adjusted primary and overall fiscal balance (Percent of GDP, %) • Gross private sector debt (Percent of GDP, %) • Gross and net public debt (Percent of GDP, %) • Gross and net public debt (Percent of net GNI (GNI net of depletion of natural resources), %) • Gross and net public debt in % of average (cyclically-adjusted) annual tax revenue • Risk of external debt distress (Low Risk, Moderate Risk, High Risk, In Distress) • Concessional debt (Percent of external public debt, %) C. Liquidity constraints • Sovereign exposure of domestic banking sector (credit to central government and state-owned enterprises (Percent of GDP, %) • Central government debt maturing in 12 months or less (Percent of GDP, %) • Projected fiscal financing needs (Percent of GDP, %) • 5-year sovereign CDS spreads (basis points) • Domestic arrears (Percent of GDP, %) • International reserve coverage (Months of imports) D. Contingent liabilities (from economic exposure to the commodity export sector) • Debt of state-owned enterprises (guaranteed and not guaranteed) (Percent of GDP, %) • Non-performing loans in state-owned banks or systemic banks exposed to commodity sector (Percent of total loans, %) • Residual (stock flow adjustment) of drivers of debt accumulation (as a proxy for materialization of past contingent liabilities and other unidentified debt drivers) E. Expenditure rigidities • Wage bill (Percent of total expenditures, %) • Interest payments (Percent of total expenditures, %) F. Institutional quality • Presence of fiscal rules (Yes; No) • Enforcement of fiscal rules (Yes; No) • Indicator of government effectiveness, Rule of Law (-2.5 to 2.5) • Indicators of Fiscal Transparency, Quality of fiscal policy (CPIA). • Index of Institutional Quality. G. Other • De jure and de facto exchange rate regime • Persistence of past price shocks Source: WEO, WDI, DEC database, Fiscal rule database, Tax rate database, Finance database, Authors’ calculations 71 References Abbas, A., N. Belhocine, A. El-Ganainy, and M. Horton. 2011. “Historical Patterns and Dynamics of Public Debt—Evidence From a New Database.” IMF Economic Review 59 (4): 717–742. https://doi.org/10.1057/imfer.2011.24 Acemoglu, D., S. H. Johnson, and J. A. Robinson. 2001. “An African Success Story: Botswana.” MIT Working Paper Series 01-37. https://doi.org/10.2139/ssrn.290791 Adler, G., and Magud, N. E., 2013. “Four Decades of Terms-of-Trade Booms: Saving-Investment Patters and a New Metric of Income Windfall” IMF Working Papers 13/103, International Monetary Fund. Aghion, Philippe, and Ioana Marinescu. 2008. “Cyclical Budgetary Policy and Economic Growth: What Do We Learn from OECD Panel Data?” NBER Chapters. In NBER Macroeconomics Annual 2007. Vol. 22, 251–278. National Bureau of Economic Research, Inc. Aizenman, Joshua, Menzie David Chinn, and Hiro Ito. 2009. “Assessing the Emerging Global Financial Architecture: Measuring the Trilemma's Configurations over Time.” Santa Cruz Department of Economics, Working Paper Series qt840728sc, Department of Economics, UC Santa Cruz. Alesina, Alberto, and Guido Tabellini. 1987. “A Positive Theory of Fiscal Deficits and Government Debt in a Democracy.” NBER Working Papers 2308, National Bureau of Economic Research, Inc. Ardanz M., and A. Izquierdo. 2017. “Current Expenditure Upswings in Good Times and Capital Expenditures Downswings in Bad Times?” IDB Working Paper Series 838, Inter- American Development Bank. Arezki, R. 2010. “Fiscal Policy in Commodity Exporting Countries: Stability and Growth.” High-Level Seminar on Natural Resources, Development, and Finance: Existing and New Challenges. Arezki, R., and M. Brückner. 2012. “Commodity Windfalls, Democracy and External Debt.” Economic Journal 122 (561): 848–866. https://doi.org/10.1111/j.1468-0297.2012.02508. Arezki, R., T. Gylfason, and A. Sy. 2011. Beyond the Curse: Policies to Harness the Power of Natural Resources. Washington, DC: International Monetary Fund. Baldacci E., J. McHugh, and I. Petrova, 2011. “Measuring Fiscal Vulnerability and Fiscal Stress: A Proposed Set of Indicators.” IMF Working Paper WP 11/94, International Monetary Fund. Barro, Robert J., and David B. Gordon. 1983. “Rules, Discretion and Reputation in a Model of Monetary Policy.” Journal of Monetary Economics 12 (1), 101–121. Berti, Katia. 2013. “Stochastic public debt projections using the historical variance-covariance matrix approach for EU countries.” European Economy, Economic Papers 480, European Commission. 72 Blandford, D. “1983. Instability in World grain MARKETS.” Journal of Agricultural Economics. Borensztein, E., O. Jeanne, and D. Sandri. 2013. “Macro-hedging for Commodity Exporters.” Journal of Development Economics 101 (1): 105–116. https://doi.org/10.1016/j.jdeveco.2012.08.005 Botev, J., J.-M. Fournier, and A. Mourougane. 2016. “A Re-assessment of Fiscal Space in OECD Countries.” OECD Economics Department Working Paper WKP(2016)76. https://www.oecd.org/eco/A-re-assessment-of-fiscal-space-in-OECD-countries.pdf Bova, Elva, Nathalie Carcenac, and Martine Guerguil. 2014. “Fiscal Rules and the Procyclicality of Fiscal Policy in the Developing World.” IMF Working Papers 14/122, International Monetary Fund. Bowman, C., and A. Husain. 2004. “Forecasting Commodity Prices: Futures versus Judgement.” IMF Working Paper 04/41, Washington, DC. https://doi.org/10.5089/9781451846133.001 Caballero, Ricardo J., and Arvind Krishnamurthy. 2004. “Fiscal Policy and Financial Depth.” NBER Working Papers 10532, National Bureau of Economic Research, Inc. Calderón, César, and Klaus Schmidt-Hebbel. 2008. “Business Cycles and Fiscal Policies: The Role of Institutions and Financial Markets.” Working Papers Central Bank of Chile 481, Central Bank of Chile. Cashin, Paul, C. John McDermott, and Alasdair Scott. 1999. “Booms and Slumps in World Commodity Prices.” Reserve Bank of New Zealand Discussion Paper Series G99/8, Reserve Bank of New Zealand. Cavalcanti, T., K. Mohaddes, and M. Raissi. 2015. “Commodity Price Volatility and the Sources of Growth.” Journal of Applied Econometrics 30: 857–873. https://doi.org/10.1002/jae Cespedes, L. F., and A. Velasco. 2014. “Was this Time Different? Fiscal Policy in Commodity Republics.” Journal of Development Economics 106: 92–106. https://doi.org/10.1016/j.jdeveco.2013.07.012 Checherita-Westphal, C., and P. Rother. 2012. “The Impact of High Government Debt on Economic Growth and its Channels: An Empirical Investigation for the Euro Area.” European Economic Review 56 (7): 1392–1405. https://doi.org/10.1016/j.euroecorev.2012.06.007 Chinn, Menzie D., and Hiro Ito. 2006. “What Matters for Financial Development? Capital Controls, Institutions, and Interactions.” Journal of Development Economics 81 (1): 163– 192. Corkin, L., and R. Bank. 2017. After the Boom: Angola’ s Recurring Oil Challenges in a New Context. WPM 72, Oxford: Oxford Institute For Energy Studies. Cuadra, Gabriel, Juan Sanchez, and Horacio Sapriza. 2010. “Fiscal Policy and Default Risk in Emerging Markets.” Review of Economic Dynamics 13 (2): 452–469. 73 Elbadawi, Ibrahim, Klaus Schmidt-Hebbel, and Raimundo Soto. 2015. “Why do Countries have Fiscal Rules?.” Journal Economía Chilena (The Chilean Economy) 18 (3): 28–61. Deaton, Angus, and Guy Laroque. 1992. “On the Behaviour of Commodity Prices.” Review of Economic Studies 59 (1): 1–23. Drechsel, Thomas, and Silvana Tenreyro. 2017. “Commodity Booms and Busts in Emerging Economies.” Discussion Papers 1723, Centre for Macroeconomics (CFM). Fernández, Andrés, Stephanie Schmitt-Grohé, and Martín Uribe. 2016. “World Shocks, World Prices, and Business Cycles: An Empirical Investigation.” NBER Chapters. In NBER International Seminar on Macroeconomics 2016. National Bureau of Economic Research, Inc. Frankel, Jeffrey A. 2011. “A Solution to Fiscal Procyclicality: The Structural Budget Institutions Pioneered by Chile.” Working Paper 16945, National Bureau of Economic Research. http://www.nber.org/papers/w16945 Frankel, Jeffrey A., Carlos A. Vegh, and Guillermo Vuletin. 2013. “On Graduation from Fiscal Procyclicality.” Journal of Development Economics 100 (1): 32–47. Frankel, Jeffrey A. 2017. “How to Cope with Volatile Commodity Export Prices: Four Proposals.” Faculty Research Working Paper Series RWP17-033, Harvard Kennedy School. Gali, Jordi, and Tommaso Monacelli. 2008. “Optimal Monetary and Fiscal Policy in a Currency Union.” Journal of International Economics 76 (1): 116–132. Garcia-Cicco, Javier, Roberto Pancrazi, and Martin Uribe. 2010. “Real Business Cycles in Emerging Countries?” American Economic Review 100 (5): 2510–2531. Gavin, Michael, and Roberto Perotti. 1997. “Fiscal Policy in Latin America.” NBER Chapters. In NBER Macroeconomics Annual. Vol. 12, 11–72. National Bureau of Economic Research, Inc. Ghosh, A. R., J. I. Kim, E. G.Mendoza, J. D. Ostry, and M. S. Qureshi. 2013. “Fiscal Fatigue, Fiscal Space, and Debt Sustainability in Advanced Economies.” Economic Journal 123 (566): 4–30. https://doi.org/10.1111/ecoj.12010 Gill, I.S., I. Izvorksi, W. van Erghen, and D. De Rosa. 2014. “Diversified Development: Making the Most of natural Resources in Eurasia”. Europe and Central Asia Studies. World Bank. Go, D. S. Go, S. Robinson, K. Thierfelder, and R. Utz. 2013. “ Dutch Disease and Spending Strategies in a Resource-Rich Low-Income Country : The Case of Niger”, Policy Research Working Paper 6691, World Bank. Gulbrandsen, Ornulf. 1995. “The King is King by the Grace of the People: The Exercise and Control of Power in Subject-Ruler Relations”, Comparative Studies in Society and History, Vol. 37, No. 3, pp. 415-444. 74 Kaminsky, Graciela, Carmen Reinhart, and Carlos Vegh, 2005, "When It Rains, It Pours: Procyclical Capital Flows and Macroeconomic Policies," NBER Macroeconomics Annual 2004, Vol. 19, pages 11-82. Gylfason, T. 2001. “Natural Resources, Education, and Economic Development.” European Economic Review 45 (4–6): 847–859. https://doi.org/10.1016/S0014-2921(01)00127-1 Hamilton, K. and E. Ley. 2010. “Measuring National Income and Growth in Resource-Rich, Income-Poor Countries”, Economic Premise 28, World Bank. Hartwick, J. 1977. “Intergenerational Equity and the Investing of Rents from Exhaustible Resources.” American Economic Review 67 (5): 972–974. Hemming, R., and M. Petrie. 2000. “A Framework for Assessing Fiscal Vulnerability.” IMF WP/00/52, Washington, DC. Ilzetzki, Ethan, Enrique G. Mendoza, and Carlos A. Végh. 2013. “How Big (Small?) Are Fiscal Multipliers?” Journal of Monetary Economics 60 (2): 239–254. International Monetary Fund (IMF). 2012a. World Economic Outlook, Chapter 4: 125-169. ———. 2012b. ” Fiscal Frameworks for Resource-Rich Developing Countries”. IMF Discussion Note SDN/12/04. ———. 2015. World Economic Outlook, Chapter 1: 1-33. ———. 2016a. “Assessing Fiscal Space: An Initial Consistent Set of Considerations.” Washington, DC. ———. 2016b. Angola: 2016 Article IV Consultation. IMF Country Report No. 17/39. IMF Article IV (Vol. 139), Washington, DC. https://www.imf.org/external/pubs/ft/scr/2016/cr16139.pdf ———. 2018. “Capitalizing on Good Times”. IMF Fiscal Monitor April 2018, Chapter 1:1-41.. Izvorksi I., S. Coulibaly, and D. Doumbia (2018). “Reinvigorating Growth in Sub-Saharan Africa: The Role of Resource-Rich Countries.” Washington, D.C. : World Bank Group. Keita, and Turcu. 2018. “On Promoting Fiscal Discipline: The Role of Exchange Rate Regimes, Fiscal Rules and Institutions.” Working Paper. Khadour, H. 2011. “Primary Commodity Price Series: Lessons for Policymakers in Resource- Rich Countries.” In Beyond the Curse: Policies to Harness the Power of Natural Ressources, edited by R. Arezki, T. Gylfason, and A. Sy. Washington, DC: International Monetary Fund. Kontopoulos, Yianos, and Roberto Perotti. 1999. “Government Fragmentation and Fiscal Policy Outcomes: Evidence from OECD Countries.” NBER Chapters. In Fiscal Institutions and Fiscal Performance, 81–102. National Bureau of Economic Research, Inc. 75 Kose, M. A., S. Kurlat, F. Ohnsorge, and N. Sugawara. 2017. “A Cross-Country Database of Fiscal Space.” World Bank Policy Research Working Paper WPS8157, Washington, DC. http://documentos.bancomundial.org/curated/es/601211501678994591/pdf/WPS8157.pdf Kumhof, Michael, and Douglas Laxton. 2013. “Simple Fiscal Policy Rules for Small Open Economies.” Journal of International Economics 91 (1): 113–127. Kydland, Finn E., and Edward C. Prescott. 1977. “Rules Rather Than Discretion: The Inconsistency of Optimal Plans.” Journal of Political Economy 85 (3): 473–491. Lange, G. M., and M. Wright. 2004. “Sustainable Development in Mineral Economies: The Example of Botswana.” Environment and Development Economics 9 (4): 485–505. https://doi.org/10.1017/S1355770X04001469 Lledo. V, S. Yoon, X. Fang, S. Mbaye, and Y. Kim. 2017. “Fiscal Rules at a Glance”, Background Paper for the publication of the IMF Fiscal Ruile Database. http://www.imf.org/external/datamapper/FiscalRules/map/map.htm. Lewin, M. 2011. “Botswana’s Success: Good Governance, Good Policies, and Good Luck.” In Yes Africa Can: Success Stories from a Dynamic Continent. Washington, DC, 81–90. https://doi.org/10.1596/978-0-8213-8745-0 Loayza, Norman V., and Romain Ranciere. 2006. “Financial Development, Financial Fragility, and Growth.” Journal of Money, Credit and Banking 38 (4): 1051–1076. Mendes, Arthur, and Steven Michael Pennings. 2017. “Consumption Smoothing and Shock Persistence: Optimal Simple Fiscal Rules for Commodity Exporters,” Policy Research Working Paper Series 8035, World Bank. Mendoza, Enrique G. 1991. “Real Business Cycles in a Small Open Economy.” American Economic Review 81 (4): 797–818. Pesaran, M. H. 2006. “Estimation and Inference in Large Heterogeneous Panels with Cross Section Dependence.” Econometrica 74 (4): 967–1012. https://doi.org/10.1111/j.1468- 0262.2006.00692.x ———. 2015. “Testing Weak Cross-Sectional Dependence in Large Panels.” Econometric Reviews 34 (6–10): 1089–1117. https://doi.org/10.1080/07474938.2014.956623 Pesaran, M. H., and R. Smith. 1995. “Estimating Long-Run Relationships from Dynamic Heterogeneous Panels.” Journal of Econometrics 68. https://doi.org/10.1016/0304- 4076(94)01644-F Pieschacón, Anamaría. 2012. “The Value of Fiscal Discipline for Oil-Exporting Countries.” Journal of Monetary Economics 59 (3): 250–268. Polackova, H. 1998. “Government Contingent Liabilities: A Hidden Risk to Fiscal Stability.” World Bank Policy Research Working Paper No. 1989, Washington, DC. Robinson, James. 2009. “Botswana as a Role Model for Country Success”. UNU-WIDER Research Paper No. 2009/40. 76 Roubini, Nouriel, and Jeffrey Sachs. 1989. “Government Spending and Budget Deficits in the Industrial Economies.” NBER Working Papers 2919, National Bureau of Economic Research, Inc. Schaechter, A., T. Kinda. N. Budina, and A. Weber. 2012. “Fiscal Rules in Response to the Crisis – Toward the Next Generation Rules: A New Dataset”. IMF Working Paper 12/187. International Monetary Fund. Solow, R. M. 1986. “On the Intergenerational Allocation of Natural Resources Author.” Scandinavian Journal of Economics 88 (1): 141–149. Spatafora N., and I. Samake. 2012. “Commodity Price Shocks and Fiscal Outcomes”, IMF Working Paper WP/12/112. Talvi, Ernesto, and Carlos A. Vegh. 2005. “Tax Base Variability and Procyclical Fiscal Policy in Developing Countries.” Journal of Development Economics 78 (1): 156–190. van der Ploeg, Frederick, and Steven Poelhekke. 2008. “Volatility, Financial Development and the Natural Resource Curse.” OxCarre Working Papers 003, Oxford Centre for the Analysis of Resource Rich Economies, University of Oxford. Van Doorn R., V. Suri, and S. Gooptu. 2010. Do Middle-Income Countries Continue to Have the Ability to Deal with the Global Financial Crisis?”. Policy Research Working Paper 5381, World Bank. Vegh, Carlos A., Luis Morano, Diego Friedheim, and Diego Rojas. 2017. “Between a Rock and a Hard Place.” World Bank Other Operational Studies 28443, World Bank. Villafuerte, M., P. Lopez Murphy, and R. Ossowski. 2010. “Riding the Roller Coaster: Fiscal Policies of Nonrenewable Resource Exporters in Latin America and the Caribbean.” IMF Working Papers WP/10/251, Washington, DC. https://doi.org/10.5089/9781455209514.001 Weber, A. 2012. “Stock-Flow Adjustments and Fiscal Transparency: A Cross-Country Comparison.” IMF Working Papers WP/12/39, Washington, DC. https://doi.org/10.5089/9781463933821.001 World Bank. 2010. Botswana: Public Expenditure Review. Washington, DC. ———. 2015. Botswana: Systematic Country Diagnostic. Washington, DC. ———. 2016. Public Finance and Mineral Revenues in Botswana Technical Report. Washington, DC. ———. 2017 (a). “The Republic of Chile: Systematic Country Diagnostic,” Washington, DC: World Bank Publications, World Bank. ———. 2017 (b). “Kazakhstan: Enhancing the Fiscal Framework to Support Economic Transformation,” Public Finance Review, Europe and Central Asia Region. ———. 2018. Global Economic Prospects, January 2018. Washington, DC: World Bank Publications, World Bank. 77