Taming Volatility: Fiscal Policy and Financial Development for Growth in the Eastern Caribbean Francisco Galrao Carneiro • Rei Odawara Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. Report No: AUS15511 June 2016 Taming Volatility: Fiscal Policy and Financial Development for Growth in the Eastern Caribbean Macroeconomic and Fiscal Management Global Practice Caribbean Countries Management Unit Latin America and the Caribbean Region Acronyms and Abbreviations AMC Asset Management Company AML Anti-Money Laundering BAICO British Insurance Company CLI CLICO International Life CLICO Colonial Life Insurance Company CLF Trinidad and Tobago based CL Financial Ltd. ECAMC Eastern Caribbean Asset Management Corporation ECCB Eastern Caribbean Central Bank ECCU Eastern-Caribbean Currency Union ECHMB Eastern-Caribbean Home Mortgage Bank EM Emerging Market FATCA U.S. Foreign Account Tax Compliance Act FATF Financial Action Task Force FINDEX Household Financial Exclusion Challenge FD Financial Development FDI Foreign Direct Investment GDP Gross Domestic Product IADI International Association of Deposit Insurance ICRG International Country Risk Guidance IFC International Finance Corporation IQ Institutional Quality IMF International Monetary Fund IV Instrumental Variable LAC Latin America and the Caribbean LCR Latin America and the Caribbean Region NFA Net Foreign Assets NPL Nonperforming Loans OAS Organization of American States OECS Organization of Eastern Caribbean States OLS Ordinary least square estimations PCG Partial credit guarantee RGSM Regional Government Securities Market SIDS Small-Island Development States SME Small- and Medium-Sized Enterprises TFP Total Factor Productivity TOT Terms of Trade US United States WDI World Development Indicator WEO World Economic Outlook Contents Executive Summary 7 Chapter 1. Growth in the Caribbean: Vulnerable and Volatile 15 1. Introduction 15 2. Growth in Latin America and the Caribbean – A Tale of Two Regions 15 3. The Recent Economic Performance in the OECS 17 4. A History of Vulnerabilities to Economic Shocks 19 5. Is Economic Growth Excessively Volatile in the OECS? 24 Chapter 2. Output Volatility and the Cyclicality of Fiscal Policy in the OECS 29 1. Introduction 29 2. Why Should Countries Avoid a Procyclical Fiscal Policy? 29 3. How Does Fiscal Policy Cyclicality Differ in Rich and Developing Countries? 31 4. Pro-Cyclicality and Macroeconomic Volatility in the OECS 35 5. How to Make Fiscal Policy More Effective and Less Procyclical 40 6. Concluding Remarks 42 Chapter 3. Comprehensive Financial Development for Growth and Lower Volatilty in the OECS 53 1. Introduction 53 2. The OECS Financial Sector: Large and Frail 53 3. Financial Development in the OECS: A Partial and Incomplete System 56 4. Entry Points to Reorient Financial Development in the OECS 59 4.1 Reorienting Financial Development to Reduce Volatility 59 4.2 Reorienting Financial Development for Enhanced Economic Growth 63 4.3 Reorienting Financial Development for Effective Fiscal Policy 68 5. Conclusions 70 Chapter 4: The Effects of Volatility, Fiscal Policy Cyclicality, and Financial Development on Growth 73 1. Introduction 73 2. Empirical Approach 73 2.1 Graphical Analysis 74 2.2 Regression Results 74 3. Impulse Response Analysis 80 4. Conclusions and Directions for Policy 88 Taming Volatility List of Figures Figure ES1: Growth & Financial Development 10 Figure ES2: Volatility & Financial Development 10 Figure 1.1: Actual and Forecasted GDP Growth in LAC – 2010 to 2015 16 Figure 1.2: Growth Deceleration in LAC 16 Figure 1.3: Heterogeneity in Growth in Latin America 17 Figure 1.4: Caribbean Growth is Stronger Now than in the Past 18 Figure 1.5: Output growth decomposition for selected OECS countries: 1971–2011 22 Figure 1.6: FDI Inflows to OECS, percentage of GDP 23 Figure 1.7: Backward Linkages Increase with Size 23 Figure 1.8: Partial Correlation Between Backward Linkages and Average Years of Education Attained Controlling for Labor Force Size 23 Figure 1.9: Government Debt in 2013 23 Figure 1.10: The OECS countries face higher volatility compared to regions with similar economies 24 Figure 1.11: Growth Volatility in the OECS 24 Figure 1.12: Countries that are more diversified have higher growth rates and less volatility 25 Figure 1.13: The Negative Relationship between Economic Growth and Volatility 25 Figure 1.14: Impact of Damages from Disasters are Significant 26 Figure 2.1: A substantive number of developing countries follows a pro-cyclical fiscal policy 30 Figure 2.2: Graduating Classes for Fiscal Policy Cyclicality 32 Figure 2.3: Cyclicality of Real Government Expenditures under Country-Specific Year Breaks 33 Figure 2.4: Fiscal Cyclicality in Booms and Downturns (1990–2011) 34 Figure 2.5: Caribbean Countries: Fiscal Cyclicality in Booms and Downturns (1990-2011) 35 Figure 2.6: Change in Fiscal Stance Proxy vs Change in Institutional Quality Index 35 Figure 3.1: Private Sector Credit 55 4 Figure 3.2: Private Sector Credit Growth 55 Figure 3.3: NPLs & Provisioning 55 Figure 3.4: Return on Average Assets 55 Figure 3.5: Components of the Financial Development Index 57 Figure 3.6: Private Credit to GDP 57 Figure 3.7: Financial Development Index 57 Figure 3.8: Growth & Financial Development 58 Figure 3.9: Volatility & Financial Development 58 Figure 3.10: ECHMB Bonds Outstanding by Tenor 64 Figure 3.11: Loans Requiring Collateral 65 Figure 3.12: Collateral Value to Loan (%) 65 Figure 3.13: A2F Main Constraint (%) 67 Figure 3.14: DB Getting Credit Ranks 67 Figure 3.15: RGSM Outstanding by Tenor (2002-2015) 69 Figure 3.16: RGSM Outstanding by Tenor (last five years) 69 Figure 3.17: Amounts Issued 2011-2016 69 Figure 3.18: Amounts Issued Adjusted for Short-Term Rollovers 2011-2016 69 Figure 4.1: Economic Growth and Volatility 75 Figure 4.2a: Impulse Responses to a 1% Positive Shock to Government Expenditures 81 Figure 4.3a: Impulse Responses after a 1% Positive Productivity Shock: Macro Aggregates 82 Figure 4.3b: Impulse Responses after a 1% Positive Productivity Shock: Financial Variables 82 Figure 4.4: Impulse Responses after a 1% Positive Shock to International Interest Rate 84 Taming Volatility List of Tables Table ES1: This Report’s Main Policy Recommendations 13 Table 1.1: Correlation of OECS real GDP per capita growth with the U.S. and the rest of the world (1980 to 2014) 19 Table 1.2: Regional Comparison of Correlation of Real GDP Per Capita 20 Table 1.3: OECS’ Export and World GDP 20 Table 1.4: Travel and Tourism Contribution (2011, %) 21 Table 1.5: Tourism and World GDP 21 Table 1.6: FDI Inflows and World Output 22 Table 1.7: Growth Volatility, OECS 26 Table 2.1: OLS and Two-Stage Least Square for Instrumental Variable Estimation regarding the effect of Institutional Quality on fiscal cyclicality stance without control variables 36 Table 2.2: OLS and Two-Stage Least Square for Instrumental Variable Estimation regarding the effect of Institutional Quality on fiscal cyclicality stance with control variables 37 Table 2.3: Averages Values for Selected Key Macroeconomics Variables: 1980-2014 38 Table 2.4: Volatility of Key Macroeconomic Variables: 1980–2014 38 Table 2.5: The Correlation of GDP with Key Macroeconomic Variables: 1980–2014 39 Table 2.6: The Correlation of Interest Rates with Key Macroeconomic Variables: 1980–2014 39 Table 3.1: Number of Insurers 61 Table 4.1: Economic Growth and Volatility 75 Table 4.2: Economic Growth and Volatility 75 Table 4.3: Economic Growth and Volatility 77 Table 4.4: Economic Growth, Financial Development, and Volatility 77 Table 4.5: Economic Growth, Fiscal Procyclicality, and Volatility 78 Table 4.6: Economic Growth, Financial Development, Fiscal Procyclicality, and Volatility 78 Table 4.7. Economic Growth, Financial Development, Fiscal Procyclicality, and Volatility: Heterogeneity OECS 79 5 Table 4.8: Simulated and Actual Business Cycles in the OECS Countries: Volatilities 85 Table 4.9: Simulated and Actual Business Cycles in the OECS Countries: Comovement with Interest Rate 86 Table 4.10: Sensitivity Analysis: Working Capital Requirement 88 Taming Volatility Acknowledgments This report was written by a team led by Francisco Carneiro, Lead Economist and Program Leader for the Caribbean Region at the World Bank, and Rei Odawara, Senior Country Economist for the OECS countries in the Macroeconomics and Fiscal Management Global Practice. The core team included Markus Brueck- ner, University of Queensland, Australia; Viktoria Hnatkovska, University of British Columbia, Canada; Ha Nguyen, Economist at the Macroeconomics and Growth Unit of the Development Research Group; Nadeem Karmali, Financial Economist, and Caroline Cerruti, Senior Financial Sector Specialist at the Finance and Markets Global Practice; Kirsten-Anne Elizabeth Roach, Junior Professional Associate in the Macroeconom- 6 ics and Fiscal Management Global Practice; Leonardo Garrido, consultant, Robert H. Singletary, consultant and Meir G. Kohn, consultant. The team is grateful to Daniel Lederman, Lead Economist and Deputy Chief Economist, Latin America and the Caribbean Region, and the peer reviewers Alfonso Garcia Mora, Director at the Finance and Markets Global Practice, and David Rosenblatt, Economic Adviser at the Department of Development Economics for insightful advice on the concept and structure of the report. Inputs received from Alessandro Legrottaglie, Cecile Niang, Raha Shahidsaless, and other meeting participants in two brainstorming sessions held at the World Bank during the preparation of the report are also acknowledged. The work was carried out under the overall supervision of Miria Pigato, Practice Manager, Macroeconomics and Fiscal Management Global Practice. Sophie Sirtaine, Country Director for the Caribbean Region, pro- vided guidance and supported the team. Miriam Beatriz Villarroel formatted the final version of the report. Elisabeth Mekonnen and Patricia Chacon provided logistic support to the team, organizing meetings and brainstorming sessions. Taming Volatility Executive Summary T his report assesses the sources of macroeco- including the erosion of trade preferences; the decline in nomic volatility in the Eastern Caribbean and official foreign assistance; turbulence in the business cy- discusses policy options to deal with the ef- cles of the countries that matter the most for the region fects of volatility and that could have positive effects in terms of tourism revenues and foreign direct invest- on long-term growth. In doing so, the report contrasts ments; and recurrent natural disasters. With the incipi- the performance of the member countries of the Orga- ent recovery, the OECS economies have an opportunity nization of Eastern Caribbean States (OECS) with that to build resilience to exogenous shocks. of other developing economies in the Caribbean and other regions of the world. More specifically, the anal- The report is structured in four chapters that outline 7 ysis focuses on the interplay between terms of trade the main sources of volatility in the region and sug- volatility, fiscal policy (pro) cyclicality, and financial gest ways to mitigate the impacts of that volatility development on growth in the OECS region. By consid- on growth. Chapter 1 presents stylized facts associated ering the interaction between these factors, the report with the growth performance of the Eastern Caribbean investigates whether (i) an increase in volatility results over the last 40 years. It contrasts the growth perfor- in lower GDP growth; (ii) the negative impact of vola- mance of the OECS with the rest of the Latin America tility on growth is more pronounced in less financially region and shows that the two groups of countries have developed countries, and (iii) in countries where fiscal shown significant heterogeneity over the business cy- policy is more pro-cyclical. cle. The chapter also highlights some of the factors that might be responsible for the volatility of growth in the Understanding the sources of macroeconomic vol- OECS, including the region’s exposure to natural disas- atility and how to deal with them is important for a ters, high debt, and adverse developments in the finan- region that is prone to face exogenous shocks. Eco- cial sector. Chapter 2 provides new evidence on output nomic growth rates in the OECS are now higher than in volatility and the cyclicality of fiscal policy in the OECS the recent past and this represents an opportunity for and discusses why countries are better off avoiding a the region to strengthen its ability to manage the im- pro-cyclical fiscal policy stance. Chapter 3 assesses pacts of unexpected shocks. Output growth rates over the level of financial development in the region as well the last three years are at a higher level than the av- as the relationship between financial development, erage for the period 2009-13 period and there is hope growth, and volatility. The chapter also explores criti- that this growth momentum can be sustained. However, cal policy options to strengthen financial development history suggests that the small, open economies of the in the OECS. Chapter 4 assesses empirically the com- OECS will remain vulnerable to considerable volatility bined effects of terms of trade volatility, fiscal policy stemming from terms of trade shocks. In the past, the (pro) cyclicality, and financial development on growth in region has been affected by a series of adverse events, the OECS and other countries using two complementa- Taming Volatility ry modeling approaches. First, through an econometric have contributed to the volatility of economic growth model using panel data for 175 countries over the period in the region. This report finds that economic growth 1980-2010. Second, by using impulse-response analy- volatility in the OECS is higher than the average for the sis based on a structural model of the business cycle in world. Although output volatility was more pronounced the OECS region. in the early half of the 1990s and more recently during the global financial crisis than it is now, several factors Growth in Latin America make the region prone to exogenous shocks that can and the Eastern Caribbean be a source of volatility. These include small size, open- ness to trade, and the synchronization with major global While economic growth in the Latin America region growth poles (such as the U.S. and EU), as well as the has plummeted over the last five years, the econo- region’s exposure to natural hazards (such as hurricanes mies of the Eastern Caribbean are showing signs of and severe storms and flooding, for example). a growth rebound. With the end of the “commodity super-cycle”, the Latin America region has been experi- Output Volatility and the Cyclicality of encing a significant growth deceleration for the last five Fiscal Policy years or so. Countries in the south of Latin America have followed a different cycle when compared with those in The large majority of developing countries, including the north and the Caribbean. Since the beginning of the the ones in the Eastern Caribbean, have yet to gradu- “great deceleration” in 2011, the commodity-dependent ate from procyclical fiscal policies. Although the num- south has followed closely the business cycle observed ber of developing countries following a more procyclical in China. The countries in the north, on the other hand, fiscal policy stance has declined over time, they are still have followed more closely the business cycle of the outnumbered by the more industrialized economies. In U.S., which has started to show signs of some recovery. the Caribbean, and the OECS in particular, the majori- In the Caribbean, however, the story is slightly different ty of the countries also tend to follow procyclical fiscal as growth rates are now much higher than immediately policies. The reasons for such behavior are various. This 8 after the global financial crisis. could be due to frictions in international credit markets that might prevent countries from borrowing in bad The close ties with the U.S. economy and its business times and forcing them to lower spending during reces- cycle means that the OECS remains vulnerable to the sions. It could also be seen as a signal of weak and un- fluctuations of output in the U.S. as well. Evidence pre- derdeveloped institutions in governments where there is sented in this report confirms that the economic perfor- limited technical capacity to develop the means to save mance of the small islands of the OECS is much more in good times to spend in bad times. closely associated with developments in its main trade partners, especially the United States and the European Governments should avoid reinforcing the business Union, than with its peers in the Caribbean. Tourism is cycle because this behavior jeopardizes their abili- the single-most important industry for the OECS coun- ty to lean against the wind in periods of hardship. A tries, ranging from 26 percent of GDP in St Vincent and pro-cyclical fiscal policy can hamper government efforts the Grenadines to 74 percent of GDP in Antigua and Bar- to reduce the effects of volatility on growth. In general, buda. Most of the tourists who visit the Caribbean with government spending can mitigate the negative effect frequency come from the U.S. and Europe. As tourism that volatility has on growth if: (i) government spending is very income-elastic, when these two countries face a is counter-cyclical; and (ii) its impact on output is pos- shock and the income of their residents drops, tourism itive. From a risk management point of view, a count- trips to the Caribbean and elsewhere are likely among er-cyclical fiscal policy can be useful for at least three the first items to be cut. The pro-cyclicality of tourism compelling reasons. First, by leaning against the wind, is also mimicked by foreign direct investment (FDI) in- governments can continue to provide goods and services flows, which have also been found to follow closely the and to maintain public investment even in the event of a business cycle observed in the U.S. economy. drop in public revenues. Second, in a downturn, a coun- tercyclical fiscal policy can help governments increase The small size, degree of trade openness, and oth- social assistance. Third, as witnessed during the global er intrinsic characteristics of the OECS economies financial crisis of 2008-09, a countercyclical fiscal pol- Taming Volatility icy can help countries stimulate the economy and cope financial sector as a driver of growth is further enforced. better with the effects of a prolonged recession. Financial development also leads to financial stability to the extent that deep and liquid financial systems with Most of the OECS economies have adopted a fiscal more diverse instruments can help alleviate the impact policy stance that seems to exacerbate output vola- of shocks. Financial development also helps countries to tility. That is, they tend to behave in a pro-cyclical way manage better the impact of terms of trade volatility, during booms and downturns. The only exception seems especially in the case of small, open economies such as to be St. Kitts and Nevis which appears together with in the case of the OECS. Trinidad and Tobago, and Belize, in the group of coun- tries that show a counter-cyclical behavior in booms, The association between financial development and which is a good indication that they are saving for rainy economic growth is well recognized in both the growth days. However, these countries still display a pro-cycli- and finance literature. Economic theory suggests that cal behavior in downturns. In doing so, they tend to am- well-functioning financial intermediaries and markets plify upswings and worsen recessions – what Kaminsky, are the conduit to reduce information asymmetries, Reinhart, and Vegh (2004) termed as the “when it rains, facilitate risk sharing and mobilize savings. This then it pours” phenomenon. This behavior is in stark contrast leads to a more efficient resource allocation and, thus, with that of the large majority of industrialized coun- may foster long-term growth. More recently, evidence tries that have largely adopted a counter-cyclical or points to a more complex relationship between finance even a-cyclical fiscal policy stance. and growth that may be non-linear or bi-directional depending on whether the financial market is equity or The volatility of output observed in the OECS region bank based. is also present in many of its relevant macroeconom- ic aggregates. An assessment of the volatility of major In the OECS countries, there is clearly a non-linear as- macroeconomic aggregates, including interest rates, sociation between financial development and growth trade balance, terms of trade, investments, government as well as volatility. The analysis is this report has con- spending, and net foreign assets (NFA), shows that all firmed empirically the non-linearity in the association 9 of them are volatile. The observed volatility in the OECS between financial development, growth as well as vola- is quite high compared to developed economies and tility for the OECS countries (Figure ES 1 and Figure ES even compared to a few other developing countries. 2). Using an index of financial development produced by The highest GDP volatility is observed in Antigua and the International Monetary Fund (IMF), our results are Barbuda and in Grenada. Interest rates are also quite unequivocal that financial development impacts growth volatile in these countries but the magnitudes are sim- positively. However, this positive effect weakens at ilar across them, with the exception of St. Lucia. The higher levels of financial development and then eventu- volatility of trade balance and terms of trade shows ally turn negative portraying a bell-shaped relationship much more dispersion across the OECS countries. In between growth and financial development. We have this context, the report finds that the OECS countries also confirmed the existence of a non-linear relationship exhibit very volatile business cycles, even compared to between financial development and volatility. As a mir- developing economies. ror image of the dynamics between financial depth and growth, financial development initially lowers volatility Financial Development in the OECS up to a certain point where it starts to create additional volatility (Figure ES 2). Financial development is an important driver of growth and can increase a country’s resilience to The OECS should continue to pursue ways to strength- external shocks and volatility. Financially developed en its financial sector as this will contribute to reduce economies find it easier to mobilize savings, share in- volatility and create appropriate conditions for sus- formation, improve resource allocation, and implement tainable growth. In both charts above, the position of more effective diversification and risk management each of the six OECS countries is well before the inflexion strategies. In the OECS context of a currency board that point, beyond which further development of the finan- provides for a monetary anchor, limited fiscal space, and cial sector may have detrimental effects to economic GDP that is highly dependent on services, the role of the growth and volatility. Given the high level of credit to Taming Volatility Fi ur ES1 Growth & Fin nci l D v lopm nt Fi ur ES2 Vol tilit & Fin nci l D v lopm nt 0.010 0.0000 DMA GRD ATG 0.005 ATG -0.0025 LCA Economic Growth Growth Volatility KNA VCT 0.000 -0.0050 VCT DMA GRD KNA -0.005 -0.0075 LCA -0.010 -0.0100 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Financial Development Financial Development Source: Authors’ own calculations using the financial deepening Source: Authors’ own calculations using the financial deepening index developed by the IMF in Sahay and others (2015a). index developed by the IMF in Sahay and others (2015a). GDP, the OECS should strive to reorient its financial credit infrastructure are critical for enhancing economic sector to improve its comprehensive level of financial growth. Finally, this chapter highlighted the importance development. This reorientation – on both the supply of reversing the short-termism in sovereign debt mar- and demand side –will positively contribute to economic kets in the OECS and striving for a more active second- growth and stability. ary market as a way to improve the efficacy of fiscal policy in the OECS. To reorient financial development in the region to be 10 more comprehensive, a number of challenges need to The Effects of Volatility, Fiscal Policy be addressed. Over the last decade, the OECS countries Cyclicality, and Financial Development have made progress to further develop their financial on Growth systems. However, there is scope to reorient financial development in the region so that it less prone to high Econometric estimates show that the negative effect collateral-low productivity projects, less likely to create of terms of trade volatility on growth is mediated by asset price bubbles and is able to contribute to enhanc- cross-country differences in financial development. ing economic growth and reducing volatility. Using an econometric model that includes an interac- tion term between terms of trade volatility and the GDP There are feasible policy options that could help re- share of domestic credit to the private sector we were orient financial development in what that would con- able to confirm that terms of trade volatility has a sig- tribute to lower volatility, higher growth, and a more nificant negative effect on economic growth; however at effective fiscal policy. Improving savings instruments, higher values of financial development the effect loses strengthening the regional supervision of insurance, and significance. We have also confirmed through our econo- establishing deposit insurance could all help econom- metric model the results discussed above that the re- ic agents better manage volatility. Restoring banking lationship between GDP p.c. growth and the GDP share stability can reduce the systemic volatility that has of domestic credit to the private sector is an inverted emanated from the heightened stress that the bank- U-shaped. The average marginal effect of financial de- ing sector has experienced over the past five years. The velopment on economic growth remains, however, posi- new Banking Act that has been passed provides the tive and significant. foundation for improved banking supervision and future consolidation. Developing long term finance for infra- Terms of trade volatility has a particularly large neg- structure and more housing in addition to developing ative effect on economic growth in countries where tools for more effective SME finance against the back- fiscal policy is procyclical. This relationship was con- drop of improvements to the enabling environment and firmed by the empirical results of the econometric Taming Volatility model that includes an interaction term between the cal responsibility laws and fiscal rules. These are widely standard deviation of the terms of trade growth rate recognized as effective mechanisms that can increase and the country-specific coefficients that measure the the discipline and credibility of the fiscal authorities. Not response of government spending to the business cy- only would these policy tools help in making fiscal policy cle. The coefficient on the interaction between terms less pro-cyclical in the OECS, but they would also help of trade volatility is negative in all specifications. In the countries in the region to make significant progress addition, the results show that the coefficient on the in reigning in fiscal expenditures and implementing ef- interaction between terms of trade volatility, fiscal fective fiscal consolidation programs. The introduction procyclicality, and the OECS indicator is significantly of fiscal rules would need to be supported by expen- negative. Further, the interaction between terms of diture reforms in the context of a medium term fiscal trade volatility and fiscal procyclicality is also negative framework to signal the authorities’ commitment to and significantly different from zero. These results sug- fiscal sustainability. Given that natural disasters are gest that: (i) fiscal procyclicality exacerbates the neg- common across the region, OECS countries would do ative growth effect of terms of trade volatility – for the well to integrate the likelihood of a disaster in their fiscal OECS region and for other regions; and (ii) the mediat- programming exercises. Many countries in similar situ- ing role of fiscal cyclicality is particularly pronounced ations have benefitted from the parallel creation of an in the OECS region. It is also noteworthy that the coef- independent fiscal council that monitors macroeconom- ficient on the interaction between terms of trade vola- ic projections underlying the budgeting process and the tility, financial development, and the OECS indicator is compliance with the fiscal rule. significantly positive. This report argues that OECS countries can also A structural model of business cycles in the OECS draw on existing good practices on how to strength- confirmed that further financial development could en their fiscal positions and be better equipped to help reduce volatility in the region. By using a com- adopt counter-cyclical fiscal policies. Moving towards plementary modeling approach, we have looked at the a full blown FRL, with or without a formal fiscal rule, impacts of different types of shocks to fiscal policy in requires some preparation and building technical ca- 11 the presence of financial frictions on the economic per- pacity in order to design, implement, and monitor the formance of the region. Our simulations (based on im- new policy tool. There are some well-established good pulse-response analysis) suggest that eliminating fiscal practices in that regard that could help the Eastern Ca- policy shocks could reduce the volatility of consumption ribbean states to become less procyclical. In addition to and trade balance, but without the volatility of GDP. internalizing the likelihood of a natural disaster in their We also show that domestic financial markets devel- financial programming exercises, countries in the region opment plays an important role in buffering the effects should consider establishing savings funds with a strong of interest rate shocks on the economy. Eliminating the institutional framework, a solid governance structure working capital constraint, while keeping all shocks in and clear operational rules for the allocation of the place, for example, could reduce the volatility of GDP, fund’s resources. The proceeds accumulated in such a consumption, employment and government spending fund could be used for emergency situations, following significantly. a disaster or a protracted economic shock, for example. The OECS countries should also strive to ensure that Directions for Policy to Reduce fiscal policy guides the budgetary process and not the Volatility and Sustain Growth in the other way around; this could be done through the adop- OECS tion of medium-term fiscal frameworks (MTFFs) and medium-term debt management strategies. Other pol- Moving toward a more counter-cyclical fiscal policy icy options that could help Eastern Caribbean countries should be a priority for all the countries in the OECS. respond in a more symmetric way to the business cycle This will have particularly high payoffs in terms of re- include, for example, expenditure ceilings, cyclical deficit ducing the adverse growth effects of terms of trade targets, and rules-based stabilization funds. volatility in the region. One way of strengthening the region’s ability to shift toward a more counter-cyclical Reducing volatility and sustaining growth will require fiscal policy stance would be through the adoption of fis- more stable financial markets and stronger financial Taming Volatility sector institutions. This report discusses a few options last decade, the OECS countries have made progress to for designing policies to help the OECS economies to further develop their financial systems. However, there achieve that objective. First, greater openness to in- is scope for further financial development in the region. ternational financial markets is important as it could The report identified specific measures that can be ad- help the OECS economies to hedge fluctuations in fun- opted to further enhance financial sector development. damental shocks, such as shocks to technology, terms A new Banking Act has been passed which provides the of trade, and shocks associated to natural hazards. foundation for improved banking supervision and future Second, greater openness must be accompanied by consolidation, and the insurance sector would benefit improvements in domestic financial markets and gov- from similar harmonization. NPL management should ernment’s efforts to stabilize domestic risk-premium. be strengthened, and if an AMC is adopted, its scope By reducing the frictions in the domestic financial mar- and nature should follow leading global examples. Fi- kets, these economies can cushion the negative effects nancial access can be improved through better credit in- of interest rate shocks on domestic economic activity, formation tools and institutions that help to reduce the and achieve lower volatility. Third, if pro-cyclical fiscal elevated levels of SME credit risk such as a guarantee policies induce higher country risk-premium in the inter- scheme. Finally, there are legal and regulatory improve- national markets, governments of the OECS countries ments in the secured transactions framework and in the can stabilize their country’s risk-premium by switch- foreclosure legislation that would greatly help the sec- ing to counter-cyclical policies. Fourth, if government tor. Careful coordination across all of the above will en- consumption is strongly complementary with private sure that a strengthened and more nimble financial sec- consumption, switching to an independent or counter- tor is better able to provide for the needs of the economy cyclical fiscal policy stance can reduce the volatility of of the OECS. Once the financial sector is strengthened in consumption in the economy. All of these should have a the region, a deposit insurance scheme can be developed positive effect on long-term growth. to provide a formal financial sector safety net. To strengthen financial development in the region, a A summary of the main recommendations of the report 12 number of challenges need to be addressed. Over the are presented in Table ES 1. Taming Volatility Table ES1: This Report’s Main Policy Recommendations Policy Objectives Priority Areas to Reduce Volatility and Strength of Policy Leverages and Buttress Growth What Has Been Done to Strengthen What Could Be Done to Make Fiscal Discipline Fiscal Policy Less Pro-cyclical Grenada remains the only country in region to have adopted a rules- A savings funds with clear operational rules based fiscal policy framework and its example should be emulated in its for the allocation of the fund’s resources neighboring countries. could be a good complement to this initiative. Many OECS countries have defined fiscal surplus targets, Countries in the region are strongly advised to include in ceilings on public spending, or limits on debt creation, but their fiscal targets buffers to cover future disaster-re- the region remains largely vulnerable to natural disasters. lated expenses. Adoption of medium-term fiscal frameworks (MTFFs) and medium-term Homegrown fiscal consolidation programs debt management strategies could help strengthen fiscal discipline and are being deployed across the region with a ensure that fiscal policy guides the budgetary process and not the other view to reign in expenditures and reduce debt. way around. Fiscal policy remains Adoption (FRLs) and/or fiscal rules and fiscal councils of formal fiscal responsibility laws largely pro-cyclical could help OECS countries strengthen fiscal discipline and move toward a more counter-cycli- in the OECS. cal fiscal policy stance. 13 What Has Been Done to Strengthen What Could Be Done to Increase Financial Development Depth and Stability in the Financial Sector The OECS has a regional sovereign debt market (RGSM) There is a need to reverse the short-termism in the RGMS, which has recently been de-materialized and is fully elec- strengthen the secondary market in order for fiscal policy tronic. to be more effective. Restoring the stability in the banking sector will help reduce this systemic A new banking act has been passed which volatility. Further, improving savings instruments, strengthening the re- will reduce some of the systemic volatility gional supervision of insurance, establish deposit insurance could all help that the recent banking stress has created. reduce volatility. The OECS needs to work on developing more long-term finance (including infrastructure and Financial development housing) and sme finance while working in parallel to improve the enabling environment and has not been growth credit infrastructure. This will help reorient financial development to be more comprehensive focused in the OECS. and enhance economic growth. Taming Volatility Chapter 1. Growth In The Caribbean: Vulnerable And Volatile Introduction contributed to a history of growth vulnerability in the region and Section 4 highlights some factors that might T he Organization of Eastern Caribbean States contribute to the volatility of growth in the OECS, in- (OECS) region has suffered the effects of cluding the region’s exposition to natural disasters, high the global financial crisis and is starting an debt, and adverse developments in the financial sector. encouraging recovery. Economic growth rates over the last three years are now higher than their average Growth in Latin America and the during the 2009–13 period and there is hope that this Caribbean – A Tale of Two Regions 15 growth momentum can help the region reduce poverty faster than in the past. However, history demonstrates Economic growth in Latin America has been on a that this region remains vulnerable to considerable downward spiral for the last five years. After sever- volatility. This volatility has many sources and can be al years of high commodity prices and strong regional manifested in different forms. In the past, the region has growth during a period that was commonly referred to been affected by a series of adverse exogenous shocks, as the “commodity super-cycle,” commodity prices have including the erosion of trade preferences; the decline in been decreasing since 2011, along with Chinese econom- official foreign assistance; turbulence in the business cy- ic activity, weakening the region’s terms of trade. As a cles of the countries that matter the most for the region result, the region has witnessed its most pronounced in terms of tourism revenues and foreign direct invest- and long-lasting growth deceleration since 2009, the ments; and recurrent natural disasters. While there is year that marked the height of the global financial crisis. optimism with the incipient recovery in the OECS, there The deceleration has frustrated most growth forecasts is no reason to believe that the latent sources of vulner- due to its magnitude (Figure 1.1) and seems to reflect ability and volatility in the region have subsided. underlying weaknesses in both aggregate demand and supply in an unfavorable external environment. While This chapter presents some stylized facts associat- growth has decelerated across the region, the region- ed with the growth performance of the Eastern Ca- al average has been dragged down by developments ribbean over the last 40 years. In the first section, we in Argentina, Brazil and Venezuela, which have had the contrast the growth performance of the OECS with the highest rates of growth deceleration in the region and rest of the Latin America region and show that the two recorded their worst economic performances in decades groups of countries have shown significant heteroge- (Figure 1.2). neity over the business cycle. Section 2 focuses on the main factors that explain the growth trajectory of OECS While growth has decelerated across the whole re- countries, including the role of Total Factor Productivity. gion, the countries in the south have followed a dif- Section 3 discusses some of the aspects that may have ferent cycle when compared to those in the north. The Taming Volatility Fi ur 1.1 Actu l nd For c st d GDP Fi ur 1.2 Growth D c l r tion in LAC Growth in LAC – 2010 to 2015 South North 7% 4 6% 2 0 5% -2 4% -4 3% -6 -8 2% -10 1% -12 0% -14 2010 2011 2012 2013 2014 2015e Brazil Honduras Venezuela El Salvador Argentina Chile Bolivia Costa Rica Mexico Nicaragua Jamaica SA Ecuador Peru Uruguay Colombia Paraguay Trin. & Tob. Panama MCC Dom. Rep. Guatemala Haiti Guyana Actual Growth December Previous Year Forecast Source: LCR Chief Economist’s Office estimates. Source: LCR Chief Economist’s Office estimates. Fi ur 1.3 H t ro n it in Growth in L tin Am ric P n l A: Th South C cl P n l B: Th North C cl 14% 6% 12% 10% 4% 8% 2% 16 6% 4% 0% 2% 0% -2% -2% -4% -4% -6% -6% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 South YoY Growth China YoY Growth US YoY Growth North YoY Growth Note: Real GDP growth rates smoothed by the Hodrik-Prescott filter. South includes Argentina, Bolivia, Brasil, Chile, Colombia, Ecuador, Peru, and Uruguay. North includes Panana, Costa Rica, Dominican Republic, El Salvador, Guatemala, Jamaica, and Mexico. Source: LCR Chief Economist’s Office estimates. region’s main commodity exporters, including, for exam- tion” in 2011. The commodity-dependent south appears ple, Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, to follow closely the same business cycle as China’s but Peru, and Uruguay are concentrated in the south of Lat- with a much more pronounced drop in output in the re- in America. This is the part of the region which has been cent period, even when compared with the growth slow- hit the hardest by the fall in global commodity prices down observed in China. On the other hand, the north (Figure 1.2). The countries in the north, including Pana- of South America seems to follow very closely the U.S. ma, Costa Rica, Dominican Republic, El Salvador, Gua- business cycle, but differently from what has happened temala, Jamaica, and Mexico, among others, have also in the most recent years in the south, growth seems to shown significant growth decelerations, but the magni- be recovering in the northern countries mimicking the tude of this phenomenon has not been as pronounced pickup in the United States. as in the south. What is interesting to note is that the north and the south have followed very different busi- In the Caribbean, the story is slightly different from ness cycles since the beginning of the “great decelera- the rest of the region as growth has been much better Taming Volatility Fi ur 1.4 C ribb n Growth is Stron r Now th n in th P st Real GDP Growth (in percent) 2014 2015 2016 2009-2013 8 7 6 5 4 3 2 1 0 -1 -2 -3 -4 ev ts ud a ba rism r t e ty iti ob dad e s s a ia t lize da a Su a cen ma do arb tigu n am aic nic d N Kit c po di Ha ya na Lu is Co ed o St. a rs Be rba d T ini ex o Jam ag ha mi Vin rin Tou Gu Gre an St. mm d B An s St. an Tr Ba Do Ba e Th an Source: International Monetary Fund, World Economic Outlook. now than in the past. Perhaps boasted by the recov- natural hazards. After growing faster than the rest of ery in the U.S. economy in the most recent years, the the world in the 1980s at an annual average of 6 per- tourism-dependent economies of the region have shown cent, the OECS countries have experienced a significant growth rates in 2014–2015 that are much more robust growth slowdown since the 1990s with annual growth than the ones observed over the 2009–13 period (Fig- rates of 2 percent or less on average. More recently, the ure 1.4). More specifically, in 2014–15 the tourism sector region was severely hit by the effects of the global finan- was the main engine of growth in the tourism-intensive cial crisis of 2008-09 because of their close ties with economies of the Caribbean (The Bahamas, Barbados, the economies of the U.S., Canada, and Europe which Jamaica, and the countries of the Eastern Caribbe- are their main source of tourist arrivals.1 Some common 17 an Currency Union, ECCU), due to rising visitor arriv- challenges faced by the small economies of the OECS als (buoyed by the U.S. recovery). Tourism contributed include exposure to frequent natural disasters; vulner- to stronger-than-expected growth in these countries, ability to external shocks; high debt; and lack of econo- except in the case of Jamaica, where a drought had mies of scale. deleterious effects on growth. The performance of the Caribbean commodity exporters (Suriname, Guyana, Total Factor Productivity (TFP) has been an import- and Trinidad and Tobago), on the other hand, has been ant determinant of growth in the OECS over the last 4 in line with that of the other commodity exporters from decades but its importance has declined more recent- the south of Latin America where past growth has been ly. Results from a growth accounting exercise reveal higher than current growth. In addition, since 2011, there that TFP explains the bulk of the variation in economic has been a positive terms-of-trade shock with the de- growth in the ECCU during the last 40 years (Figure 1.5). cline in commodity prices – one that turned sharper with TFP’s contribution to growth for the OECS on average the more recent decline in oil prices. The recent evolution has been declining after reaching its peak in the 1970s of the terms of trade might be another important factor when it explained 86 percent of regional growth; this that could contribute to robust growth in the region. contribution has come down to just 8 percent in the ear- ly 2000s. The downward trend in TFP was mainly due The Recent Economic Performance in to reduced productivity in Dominica, St. Lucia and more the OECS recently Grenada, with negative TFP growth observed in the cases of Dominica and Grenada (see Schipke et Historically, the economic performance of the mem- al. (2013)). More recently, however, a few countries in bers of the OECS has been particularly uneven. This the OECS, including Antigua and Barbuda, St. Kitts and is due to reasons that range from the need to reinvent themselves after the end of preferential trade agree- 1  See Kouame and Reyes (2015) for evidence of how the Carib- ments with Europe in the 1980s to the frequency of bean’s economic growth rates relate to those of key drivers of the global economy. Taming Volatility Fi ur 1.5 Output Growth D composition for S l ct d OECS Countri s: 1971–2011 Anti u nd B rbud Gr n d 0.14 0.25 0.12 0.20 0.10 0.15 0.08 0.06 0.10 0.04 0.05 0.02 -0.00 0.00 -0.05 -0.02 -0.04 -0.10 -0.06 -0.15 1971-1980 1981-1990 1991-2000 2001-2005 2006-2011 1971-1980 1981-1990 1991-2000 2001-2005 2006-2011 Output growth Capital growth Labor growth TFP Growth St Luci St. Vinc nt nd th Gr n din s 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 -0.00 -0.05 -0.05 -0.10 -0.10 18 -0.15 -0.15 1971-1980 1981-1990 1991-2000 2001-2005 2006-2011 1971-1980 1981-1990 1991-2000 2001-2005 2006-2011 Output growth Capital growth Labor growth TFP Growth Nevis, and St. Vincent and the Grenadines, showed im- and weak institutions. Another explanation for negative provements in productivity growth between 2000 and growth is the underutilization of resources, such as cap- 2005, but turned to negative TFP growth in the period ital and labor. Negative TFP growth in the latter half of that coincided with the global financial crisis (2006–11). 2000s could be caused by the underutilization of capital While it may be difficult to precise the exact causes be- in particular, or resources in general, during the Great hind this trend, it might be the case that these coun- Recession. tries were successful in gradually shifting towards more capital intensive tourism and away from labor intensive The performance of the OECS economies appears to agriculture throughout the region during the first half of follow closely the economic performance of the Unit- the 2000s and before suffering the impacts of the glob- ed States and the European Union. In a recent paper, al crisis. Kouame and Reyes (2015) studied the relation between growth in the Caribbean and some global growth en- Updated data available for Antigua and Barbuda, gines, namely the United States, the European Union, Grenada, St. Lucia, and St. Vincent and the Gren- Brazil, and China. They first assessed empirically the adines reveal that TFP growth in the latter half of relationship between growth rates in the Caribbean and 2000s turned negative for all of these four countries. the global growth poles, and then tried to discern the As pointed out by Barro (1998), negative TFP growth is effect of business cycle movements in the global growth hard to interpret because it suggests a decline in techni- poles on the Caribbean. They found a statistically sig- cal progress, reflecting a reduction in the efficiency with nificant and positive relationship between growth in the which the other factors of production are used either be- United States and the Caribbean, as well as between cause of complementary factors or due to bad policies growth in the European Union and the Caribbean. The Taming Volatility evidence of growth synchronization between Caribbean small states. In what follows, we briefly discuss some of states with Brazil and China, however, was weak and the sources of these vulnerabilities. not statistically significant. That finding is consistent with the good performance of the Caribbean, and espe- The high synchronization of the region with the U.S. cially the tourism-dependent economies of the OECS, in business cycles means that the OECS remains vulner- the most recent period. able to exogenous shocks. We were able to confirm this by calculating the average correlation of the change in A History of Vulnerabilities to GDP per capita (in constant 2005 US$) between OECS Economic Shocks countries and the U.S. as well as the rest of the world between two consecutive periods, 1980-1999 and Small states are subject to a number of vulnerabilities 2000–2014. The underlying assumption is that since that can hamper their long-term growth. A recurring OECS countries are very small, they are not expected question commonly asked by those who live and work to affect world GDP. The correlation, therefore, only re- on small states is whether small economy states ac- flects the impact of the world’s economic volatility on tually suffer from their smallness? Easterly and Kraay the OECS countries. With the exception of probably (2000) addressed this question head on and argued that Dominica, all other OECS countries’ GDP per capita are there are good theoretical reasons to believe that they highly correlated to the U.S and world output. The most do. According to these authors, the provision of public vulnerable country is Antigua and Barbuda, with the services may be subject to indivisibilities that lead to correlation to the U.S. output of 0.66, which means that increasing returns to scale, especially fiscal institutions if the U.S. GDP were to drop by 1 percentage point in any and defense. They also raised theoretical arguments given year, the GDP of Antigua and Barbuda would be suggesting that increasing returns to scale in the private expected to decline by 0.66 percentage point (see Table economy may be difficult to realize in small states and 1.1). The high correlation of Antigua and Barbuda with that small economies may also be at a disadvantage the US and world output might be related to the coun- because their size prevents them from diversifying into try’s high dependence (the highest among the OECS a wide range of activities, making them more vulnera- countries – see Table 1.4) on tourism. While this high 19 ble to terms of trade shocks than large states. To com- dependence represents an important source of vulnera- plicate matters even further, many small states suffer bility on the downward side of the business cycle, it also from poor location in that they are remote and/or land- means that Antigua and Barbuda would be the OECS locked, and many are located in regions prone to hurri- country standing to benefit the most from a growth ac- canes and volcanic activity. Easterly and Kraay (2000) celeration in the U.S. economy. argued further that public officials in small states may be much more likely to be subjected to conflicting pres- The performance of the small islands of the OECS sures, and that it may be difficult to recruit a high-qual- seems to be much more closely associated with ex- ity civil service given the limited pool of candidates in ternal developments than its peers in the Caribbe- Table 1.1: Correlation of OECS Real GDP Per Capita Growth with the U.S. and the Rest of the World (1980 to 2014) Growth in real GDP per capita (con- 1980–1999 2000–2014 stant 2005 US$) World USA World USA Grenada 0.5 0.21 0.41 0.5 Dominica 0.1 0.11 0.24 0.07 St. Vincent and the Grenadines 0.24 0.15 0.13 0.21 St. Lucia 0.45 0.46 0.39 0.39 St. Kitts and Nevis 0.64 0.49 0.47 0.46 Antigua and Barbuda 0.38 0.44 0.68 0.66 Source: World Bank staff calculations based on World Development Indicators. Taming Volatility an and elsewhere. To put OECS’ external vulnerability amination of the region’s exports. Overall, we find that into perspective, we compare their external dependence the OECS’s export performance is positively correlat- with that of other Caribbean countries and other small ed with U.S. and the world’s GDP, with the exceptions islands. From Table 1.2, two observations are notewor- of Dominica during the 1980s and 1990s, and St Lucia thy. First, the correlation between growth in OECS’ GDP between 2000 and 2014. Among the other countries, St. per capita and the world’s as well as the U.S.’ appears Kitts and Nevis stands out as the country where export consistently in the range between 0.55–0.65 over time. revenues are the most dependent on world GDP: its cor- This suggests that the OECS’s output growth is consis- relations to the U.S. and world GDP are 0.69 and 0.77, tently dependent on world output. Second, OECS econo- respectively (Table 1.3). mies are significantly more dependent on the world than other Caribbean islands and other non-Caribbean small Tourism revenues in the OECS are highly vulnerable islands. This finding reinforces the view that OECS coun- to external economic volatility. Tourism is the sin- tries are more vulnerable to external economic shocks gle-most important industry for the OECS countries. than their comparators. The total contribution of international tourism in 2011, for example, ranged from 26.2 percent (for St Vincent The export performance of the OECS countries is also and the Grenadines) to 74.2 percent of GDP (for Antigua closely associated with the business cycles in the and Barbuda) (see Table 1.4). Tourism, unfortunately, is U.S. and the rest of the world. To gain a better under- very income-elastic. That is, when income falls, tour- standing of the OECS’s vulnerability to external shocks, ism trips are likely among the first items to be cut in a we investigate three factors that could be responsible household’s budget planning. Not surprisingly, tourism for this external dependence: export revenues, tourism revenues are highly vulnerable to external economic vol- revenues, and FDI inflows. In particular, we examine how atility. This can be seen from Table 1.5 which presents dependent OECS export and tourism revenues are to the the correlation between the change in the log of real world and the U.S. GDP per capita. We start with an ex- international tourism revenue per capita and world real 20 Table 1.2: Regional Comparison of Correlation of Real GDP Per Capita Growth in real GDP per capita 1980–1999 2000–2014 (constant 2005 US$) World USA World USA OECS average 0.63 0.56 0.64 0.65 Other Caribbean 0.10 0.00 0.42 0.44 Other small islands 0.52 0.58 0.29 0.15 Source: World Bank staff calculations based on World Development Indicators. Table 1.3: OECS’ Export and World GDP 1980–1999 2000–2014 Growth in real export per capita Growth in real GDP Growth in real GDP Growth in real GDP Growth in real GDP per capita World per capita USA per capita World per capita USA Grenada 0.240 0.256 0.22 0.08 Dominica -0.248 -0.221 0.29 0.13 St. Vincent and the Grenadines 0.291 0.300 0.35 0.51 St. Lucia 0.270 0.165 -0.13 -0.02 St. Kitts and Nevis 0.436 0.491 0.69 0.77 Antigua and Barbuda 0.177 0.159 0.29 0.31 Source: World Bank staff calculations based on World Development Indicators. Taming Volatility Table 1.4: Travel and Tourism Contribution (2011, %) Direct Contribution Total contribution Direct Total Exports to GDP to GDP Employment employment Antigua & Barbuda 17.8 74.2 18 69 74.4 Dominica 7.5 24.8 6.9 22.9 45.3 Grenada 7.3 24.2 6.8 22.4 66.7 St Kitts & Nevis 7.8 28.2 7.7 26.7 44.3 St Lucia 15.4 45.8 17.5 45.4 60.8 St Vincent & Grenadines 7.7 26.2 7.1 23.9 51.8 Notes: Direct contribution: Tourist transportation, food and leisure industries. Indirect contribution: capital investment in travel and tourism; government spending to support tourism, and supply chain effects. Source: World Travel and Tourism Council; OECS; Caribbean Tourism Organization. Table 1.5: Tourism and World GDP Growth in real international tourism 2000–2014 revenue per capita (2005 US$) Growth in real GDP per capita World Growth in real GDP per capita USA Grenada 0.18 0.00 Dominica 0.23 0.05 St. Vincent and the Grenadines 0.29 0.48 St. Lucia 0.04 0.24 St. Kitts and Nevis 0.52 0.64 21 Antigua and Barbuda 0.31 0.29 Source: Authors ‘own elaboration based on data from the World Travel and Tourism Council; OECS; Caribbean Tourism Organization. GDP per capita between 2000 and 2014.2 Table 1.5 re- FDI inflows (as a share of GDP) and the change in the veals that international tourism revenues for the OECS log of world real GDP per capita, as well as U.S.’s real are very dependent on the U.S.’s and world GDP. This is GDP per capita, for the period between 1980 and 2014. especially true for St. Vincent and the Grenadines, and Overall, we observe a positive correlation between the St. Kitts and Nevis. change in FDI inflows to the OECS countries (as a share of output) and real growth in real GDP per capita of the FDI inflows to the OECS are highly pro-cyclical with world and the U.S. external developments. Figure 1.6 shows that FDI in- flows to the OECS vary a great deal, ranging from 5 to Differentiated performance among the OECS coun- 37 percent of GDP. Not only that, they are also highly tries are evident. As shown in Figure 1.4, some OECS dependent on the world economy (which serves as a countries. push factor). In the boom time prior to the Great Reces- sion (2000–2005), FDI inflows increased sharply for all The Caribbean region has experienced low-long term OECS countries, peaking in 2006 and 2007. However, growth despite large inflows of FDI. One would ex- when the major growth poles suffered the effects of the pect that foreign direct investment inflows should be Great Recession, FDI inflows to the OECS also collapsed associated with higher growth rates due to knowledge sharply. Table 1.6 confirms the analysis with longer time transfers and spillover effects. However, this is not the series. It shows the correlation between the change in rule in the Caribbean region, and in particular in the ser- vice-oriented economies of the OECS. Antoine, de Piniés 2  Unfortunately there were not enough data points for the and Martin-Sanchez (2015) show that the service ori- 1980–1999 period to yield meaningful analyses. Taming Volatility Fi ur 1.6 FDI Inflows to OECS, P rc nt of GDP 35 30 25 20 15 10 5 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Antigua & Barbuda Dominica Grenada St Kitts & Nevis St Lucia St Vincent & Grenadines Table 1.6: FDI Inflows and World Output Growth in real GDP per capita ∆FDI/GDP*100 World USA Antigua & Barbuda 0.24 0.143 Dominica 0.068 0.041 Grenada 0.057 0.082 St Kitts & Nevis 0.221 0.244 22 St Lucia 0.191 0.021 St Vincent & Grenadines 0.147 0.111 Source: Authors’ own elaboration based on World Development Indicators. ented economies of the Caribbean have received more Along with other small economies, Caribbean coun- foreign direct investment than other countries in the tries also exhibit high levels of public debt relative to region but have not seen significantly elevated growth GDP. Figure 1.9 shows the median debt to GDP ratio in rates. According to these authors, this lack of positive 2013 over several country groupings. While it is not en- growth spillovers from FDI inflows is associated with the tirely clear that debt is directly related to size, it is im- small size of these economies. Figure 1.7 shows that the mediately apparent that the Caribbean stands out as a percentage of inputs of domestic origin in foreign affil- particularly high debt region. This may be partly related iates, an indication for backward linkages, is higher for to the constant need to rebuild infrastructure after fre- countries with larger labor forces. In addition, they note quent natural disasters and partly due to the fact that that the foreign firms investing in the service-oriented government revenues as a share of GDP are consistently countries in the Caribbean tend to rely mostly on foreign low. Haque et al. (2016) find that small states, in general, technologies which they find is associated with a lower have accumulated public debt at a faster rate than larg- propensity for backward linkages. While this could be an er countries. For example, over the more recent period, indication that these economies do not have the skilled lower-middle and upper-middle income small states ac- labor force required to develop backward linkages in more cumulated public debt at an average rate of 1.3 percent advanced services, Lederman and Lesniak (2016) do of GDP and 1.8 percent of GDP per annum, respectively, not find evidence in support of this argument and show over 2009–2014. According to Haque et al. (2016), this that the small size of these economies is a more com- compares to slightly negative debt accumulation for pelling reason behind this phenomenon (see Figure 1.8). other states. Taming Volatility Fi ur 1.7 B ckw rd Link s Incr s with Si Fi ur 1.18 P rti l Corr l tion B tw n B ckw rd Link s nd Av r Y rs of Educ tion Att in d Controllin for L bor Forc Si .5 5 LCA SUR BLZ BLZ TTO JAM GUY TTO VCT BHS JAM 4 Controlling Size GUY DOM 0 BRB BRB 3 DOM -.5 2 -4 -2 0 2 4 6 -2 -1 0 1 Log (Labor Force) Resid Educaction Controlling Size Fit Line Service Oriented Rest of World Fit Line Service Oriented Rest of World Fit Line Manufacturing Oriented Commodity Export Oriented Fit Line Manufacturing Oriented Commodity Export Oriented Note: Graph shows fit line of the follwing linear regression: Log(Per- Note: Graph shows the fit line of an ordinary least squares regression. centage of inputs of domestic origin)= log(labor force) + error. The It is calculated by first estimating the following two equations: (1) relationship is positive and statistically significant at the 1% level Log(percentage of inputs of domestic origin)= log(labor force) + error1; with a coefficient of .09. Source: Antoine, de Pinies, and (2) Log(average years of education)=log(labor force) + error2. Then the Sanchez-Martin (2015) based on World Bank Enterprise Survey, plot represents the regression of error1 on error2. This can be thought World Bank World Development Indicators database. of as representing the relation between the % of domestic inputs a firm uses and the average years of education attained by workers in a country controlling for the size of the labor force. The relationship is positive with a coefficient of .01 but not statistically significant. Source: Lederman and Lesniak (2016) based on data from Barro Lee Education Data set, World Bank enterprise Survey, World Bank World Development Indicators database, and author’s calculations. 23 Fi ur 1.9 Gov rnm nt D bt in 2013 Labor Force Size Income level Region Economy type 100 80 60 40 20 0 Hig ome le EC e Up om le 1 t llion n 5 t illio n on Lo lion n Low ome CD an ca CS y eri ral Co ted e M a r i en o d i t y i en i n g xed y ua ua n-O om i en i c i a Mi tha Inc idd Inc idd eri OE 1 M ss th be illi Am Cent OE O r er v pe e n u ted ted D Pa a O r c tur Mi ug rag c no h Inc Inc Am rib rM M M 0M O m 20 re S Ur l Le Mo m fa er o5 Ca w tin o2 La Notes: Bars represent the median value of the Gross Government debt to GDP ratio in 2013 for the grouping. Labor force groups are: less than 1 million, 1 million to 5 million, 5 million to 20 million, and larger than 20 million (upper border included in higher category). Country Income groups/levels defined as in the World Bank WDI. Caribbean includes: Antigua and Barbuda, The Bahamas, Belize, Barbados, Domini- ca, Dominican Republic, Grenada, Guyana ,Haiti, Jamaica, St. Kitts and Nevis, St. Lucia, Suriname, Trinidad and Tobago, and St. Vincent and the Grenadines. OECS includes: Antigua and Barbuda, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, and St. Vincent and the Grenadi- nes. Service Oriented Economies includes: Antigua and Barbuda, The Bahamas, Barbados, Dominica, Grenada, Jamaica, St. Kitts and Nevis, St. Lucia, and St. Vincent and the Grenadines. Manufacturing/Services oriented includes: Belize, Dominican Republic, and Haiti. Commodi- ty Export Oriented includes: Guyana, Suriname, and Trinidad & Tobago. Latin America includes: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador. El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. Source: Lederman and Lesniak (2016) based on data from the IMF World Economic Outlook database April 2015, and author’s calculations. Taming Volatility The main drivers of public debt accumulation in small external debt accumulation for small states was the size states appear to be rising fiscal deficits. Haque et al. of the current account deficit (11.5 percent of GDP for (2016) present evidence showing that, for lower-middle lower-middle income small states, 10.4 percent of GDP income countries, debt accumulation is driven by larg- for upper-middle income small states, and 5.6 percent er primary deficits (3.2 percent of GDP compared to of GDP for larger states). Slower growth in smaller states 1.9 percent of GDP for larger states) and slower growth also exacerbated negative external debt dynamics. (-1.6 percent of GDP, compared to -1.8 percent for larg- er states). While upper-middle income small state have Is Economic Growth Excessively run smaller primary deficits than larger countries, high Volatile in the OECS? existing debt burdens including significant commercial debt have led to a strong contribution from real inter- The OECS countries have suffered historically with est rates (1.6 percent of GDP per annum, compared to volatility of economic growth. Economic growth vola- slightly negative contribution for larger states). Haque tility in the OECS has been higher than that observed et al. (2016) conclude that higher average primary in other groups of countries with similar characteristics deficits are associated with public debt accumulation such as other small states of the Caribbean, the broad- across small states and that the impact of primary defi- er group of small-island development states (SIDS), and cits dominates growth effects, with several small states Central American countries (Figure 1.10). Output volatil- experiencing reasonable rates of growth combined with ity in the region was particularly pronounced in the ear- increasing debt stocks in the presence of large primary ly half of the 1990s and 2000s and then more recent- deficits. ly during 2008–2010 as a result of the global financial crisis. This is not surprising given these countries’ small Small states also tend to accumulate external debt at size, high degree of openness, dependence on tourism a faster pace than larger countries. According to the from a not so diversified range of countries, and prone- evidence presented by Haque et al. (2016), lower-mid- ness to natural hazards which make them highly ex- dle and upper-middle income small states accumulated posed to external shocks. 24 debt at 1.1 percent and 2.9 percent of GDP per annum respectively, compared to debt accumulation of just 0.1 The most volatile countries in the OECS are Antigua percent for larger countries. The major driver of increased and Barbuda and Grenada and to a lesser extent are Fi ur 1.10 Th OECS countri s f c hi h r vol tilit Fi ur 1.11 Growth Vol tilit in th OECS comp r d to r ions with simil r conomi s high growth; high growth; R l GDP St nd rd D vi tion b R ion 1/ low volatility high volatility 5 7 Panama 6 4 Cabo Verde 5 Mauritius Belize Average (1990-2015) 3 4 Costa Rica Dominican Rep. St. Kitts 2 3 and Nevis Grenada Dominica St. Vicent Antigua 2 St. Lucia and Barbuda 1 1 0 0 1 2 3 4 5 6 7 8 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 low growth; low growth; low volatility high volatility OECS 2/ Small Island Developing States 3/ Standard Deviation (1990-2015) Caribbean Small States 4/ Central America 5/ Note: 1/ 5 year moving standard deviation; 2/ Antigua and Barbuda, Dominica, Grenada, St. Lucia, St. Kitts and Nevis, and St. Vincent and the Grenadines; 3/ Comoros, Djibouti, Fiji, Guinea-Bissau, Kiribati, Maldives, Marshall Islands, Mauritius, Micronesia, Fed. Sts, Palau, Papua New Guinea, Samoa, Sao Tome and Principe, Seychelles, Singapore, Solomon Islands, Timor-Leste, Tonga, and Vanuatu; 4/ The Bahamas, Barbados, Belize, Guyana, Suriname, Jamaica, and Trinidad and Tobago; 5/ Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, and Panama. Source: IMF; WEO and World Bank staff calculations. Taming Volatility St. Lucia and St. Kitts and Nevis. Dominica and St. Fi ur 1.12 Countri s Th t Ar Mor Div rsifi d Vincent and the Grenadines, on the other hand, seem H v Hi h r Growth R t s nd L ss Vol tilit to experience the lowest levels of volatility in the region (Figure 11). It is also worth noting that the countries 100 that exhibit the highest levels of volatility in the OECS are also the ones that are less diversified and that rely 80 the most on services related to tourism as their major 60 growth driver. By contrast, the countries which exhibit the lowest volatility in the region are those which have 40 become more diversified and where the contribution of other sectors GDP growth, such as agriculture and in- 20 dustry, mainly construction activity, is also important (Figure 1.12). 0 ATG LCA PAN GRD VCT KNA CPV DMA MUS BLZ CRI DOM Agro Industry Services The relationship between terms of trade volatility and economic growth is negative for a large sample of Source: IMF; WEO and World Bank staff calculations. countries, including the OECS. Figure 1.13 plots on the y-axis countries’ average GDP per capita growth (over A. Statistically, we cannot reject the hypothesis that the five years); on the x-axis is the standard deviation of the slope coefficient in Panel A is equal to the slope coeffi- terms of trade growth rate (also computed over a five cient in Panel B (t-value 0.91). year period). We see from Panel A in Figure 1.10 that for a sample of 175 countries terms of trade volatility has a Nevertheless, growth volatility in the OECS, while negative average effect on economic growth. The coef- substantially higher than the world average, is within ficient from a bivariate regression that corresponds to the range of their comparators. Most notably, growth the plot in Panel A of Figure 1.10 is -0.31; this coefficient volatility for the OECS countries is in the same range is significant at the 1 percent level (p-value 0.007). Panel of other small island countries throughout the previous 25 B shows that a negative relationship between terms of decades (3.68 percent versus 3.70 percent in 1980s; trade volatility and economic growth is visible also with- 1.99 percent versus 2.03 percent in 1990s; 3.53 percent in the sub-sample of OECS countries. The coefficient versus 2.87 percent in 2000s, and 2.42 percent versus from a bivariate regression that corresponds to the plot 2.11 percent since 2011). Interestingly, OECS growth is in Panel B of Figure 1.10 is -0.36; thus it is quantitatively more volatile than in other Caribbean countries in pre- very close to the slope coefficient that emerges in Panel vious decades, but since after the Great Recession, Fi ur 1.13 Th N tiv R l tionship B tw n Economic Growth nd Vol tilit P n l A: All Countri s P n l B: OECS Onl BIH .22 AZE .4 ARM BLR GNQ LVAGEOCHN KNA SGP MAC KOR BWA KAZ GRD .2 HKG QAT LTU KNA MLT MDV ROM VNM THAMYSGRD JPN .2 GRC PRT CYP ESP BGR IRL ATG LAO ESTDMA LKA EGY VCT IDN TTO AUT LUXISR FIN ITA UKR PAN BTN SYC DOM INDCPV LCA NOR TKM OMN growth BEL FRASVN DNK GBR POL TUN SWZ NLDMUS TON TUR ISL MAR PAK BRA MRT LSOCHL CUB IRN KWT SVK BLZ .18 HUNSWE ALB CZE USA CAN TZA AUS MEX CRI SYR PHLMOZCOL ECU MDA CHENPL KGZ BRB NZL KHMSDN JOR URY FSM MLI MNG PER ARG MWI JAM BMU FJI LBYGAB PRY SLV SAU GTM COG AGO MKD MHL HRV BHSGUY SURTJK SLE BGD BFA NAM UZB HNDBENGHA ETHTCD UGASLBPNG RUS IRQVEN DZA PLW KEN BOL ARE CIV COM CMR BDI NGARWA STP GMB LBN AFG SEN BHR GIN NIC TGOGNB HTI BRN 0 MDG NER ATG DMA .16 SOM KIR ERI CAF DJI LBR VCT LCA .14 −.2 0 .05 .1 .15 .2 .25 .02 .04 .06 .08 .1 .12 volatility volatility Source: Brueckner and Carneiro (2015). Taming Volatility Table 1.7: Growth Volatility, OECS 1980–1999 1980-89 1990–99 2000–09 2010–14 Antigua and Barbuda 3.96 3.09 7.04 4.81 Dominica 3.78 1.80 3.44 1.85 Grenada 4.23 3.88 6.11 2.70 St. Kitts and Nevis 4.05 2.97 4.21 4.33 St. Lucia 10.27 6.48 3.86 1.17 St. Vincent and the Grenadines 4.00 2.61 3.60 2.06 OECS average 3.68 1.99 3.53 2.42 Other Caribbean 2.03 2.34 2.62 5.91 Other small islands 3.70 2.03 2.87 2.11 World average 1.32 0.84 1.91 0.73 Fi ur 1.14 Imp ct of D m s from Dis st rs r Si nific nt 250% Hurricane: Hurricane: Hattie Ivan 200% Hurricane: 150% Hurricanes: Georges 26 David & Frederick Hurricane: 100% Hurricane: Hurricane: Hurricane: Hurricanes: Luis Allen Luis Luis & Marilyn Gilbert Flood Hurricane: Tomas 50% 0% Belize Dominica Saint Lucia Jamaica AB Dominica Saint Kitts Saint Kitts Grenada Guyana St Lucia 1961 1979 1980 1988 1995 1995 and Nevis and Nevis 2004 2005 2010 1995 1998 Source: EM-DMT. other Caribbean countries suffered significantly higher position of the governments, exacerbating existing ex- volatility (5.91 percent versus 2.42 percent). The finding penditure pressures, redirecting public resources away suggests that small islands, including the OECS, other from long-run development plans and increasing in- Caribbean, and other small islands around the world, are debtedness. Between 1993 and 2012, average OECS an- facing significantly higher growth volatility. nual losses from natural disasters reached 4.3 percent of GDP. In addition, Hurricanes Ivan (2004) and George Substantive exposure to natural disasters contrib- (1998) caused damages amounted to over 200 percent utes to output volatility in the OECS. The region con- of GDP in Grenada and 140 percent of GDP in St. Kitts centrates the countries in the world that are most and Nevis, respectively (Figure 1.14). exposed to natural hazards. Natural disasters such as hurricanes and tropical storms have seriously impacted The level of financial development is also believed the countries’ growth performance by severely affecting to affect the impact that volatility has on economic the productive sectors of the economy, such as agri- growth. Recent studies suggest that beyond a certain culture and tourism, and communities and households. level financial development may generate decreasing Natural disasters have also added pressure on the fiscal returns to growth and stability (Arcand, Berkes, and Taming Volatility Panizza (2012); Sahay et al. (2015a)). A few arguments access to credit and the quality of the banking sector, might explain why this is so. One such view is that too especially after the 2008 global crisis. As a result of much finance may increase the frequency of booms and the economic slowdown following the crisis, nonper- busts thus increasing volatility and affecting economic forming loans (NPLs) in the banking sector worsened, growth negatively. Excessive availability of finance can increasing to 18.0 percent of total loans in 2014 from also cause a diversion of talent and human capital away 7.7 percent in 2008, and domestic banks have experi- from productive sectors and toward the financial sec- enced losses region wide. Credit growth to the private tor without a clear net positive impact on growth. Also, sector continued to fall and since 2013 is now negative. excessive leverage and risk taking can lead to increased Most of the credit tightening occurred in services (such economic and financial volatility, with potentially neg- as tourism-related services) although the highest share ative consequences for long-term growth, especially if of loans are to individuals (54 percent) for acquisition of regulation and supervision are inadequate (IMF (2003); property and home construction and renovation, which Reinhart and Rogoff (2011); Sahay et al. (2015a) and has been increasing in recent years. Provisioning levels (2015b)). have increased for these NPLs but continues to provide inadequate coverage of bad debts and these are still Financial soundness in the OECS region has deterio- below international standards for provisioning. Insuffi- rated post crisis. A formal assessment of the impact of cient access to credit to the private sector, in particular financial sector development on volatility and ultimate- small and medium enterprises in the tourism sector, ly growth is presented in Chapter 4, but it is worth men- has weakened the fragile recovery in the region and re- tioning that the region has suffered with inadequate main a challenge. 27 Taming Volatility REFERENCES Antonine, Kassia, de Pinies, Jaime, and Sanchez-Martin, International Monetary Fund (IMF). 2003. “Financial As- Miguel Eduardo. 2015. Measuring the Determinants set Price Volatility: A Source for Instability.” Chapter of Backward Linkages from FDI in Developing Econo- 3 in Global Financial Stability Report. Washington. mies. Is it a Matter of Size?” World Bank. Forthcom- Lederman, D. and J. Lesniak. 2016. Economic Develop- ing. ment with Limited Supplies of Labor: Common Chal- Arcand, J. L., E. Berkes and U. Panizza. 2012. “Too Much lenges, Shared Solutions for the Caribbean. Office of Finance?” IMF Working Paper 161, International Mon- the Chief Economist for the Latin America and Ca- etary Fund, Washington. ribbean Region. The World Bank. Washington, D.C. Barro, R. 1998. “Notes on Growth Accounting”. NBER (forthcoming). Working Paper No. 6654. Cambridge, Massa- Reinhart, C. M., and K. S. Rogoff. 2011. This Time Is Differ- chusstts. ent: Eight Centuries of Financial Folly. Princeton, New Easterly, W. and A. Kraay. 2000. Small States, Small Jersey: Princeton University Press. Problems? Income, Growth, and Volatility in Small Sahay, R., M. Čihák, P. N’Diaye, A. Barajas, R. Bi, D. Ayala, States. World Development Vol. 28, No. 11, pp. 2013- Y. Gao, A. Kyobe, L. Nguyen, C. Saborowski, K. Sviry- 2027. dzenka, and Seyed Reza Yousefi. 2015a. “Rethinking Haque, T., Z. Partow, S. Varma, and L. Oliveira. 2016. Financial Deepening: Stability and Growth in Emerg- Addressing Debt Vulnerabilities of Small States: The ing Markets.” IMF Discussion Note SD 15/08. Potential Role of New Financing Instruments. Macro- Sahay, R., M. Cihak, P. N’Diaye, A. Barajas, S. Mitra, A. economics and Fiscal Management Discussion Paper Kyobe, Y. N. Mooi, S. R. Yousefi. 2015b. “Financial In- N. 10. The World Bank. clusion: Can It Meet Multiple Macroeconomic Goals?” Kouame, A. T.; Reyes, M. I. 2016. Before and after the IMF Staff Discussion Note 15/17, International Mone- global financial crisis: evaluating the Caribbean’s tary Fund, Washington. synchronization with global engines of growth. Pol- Schipke, A.; A. Cebotari; and N. Thacker. 2013. The Eas- 28 ter Caribbean Economic Currency Union: Macroeco- icy Research working paper; no. WPS 7544. Wash- ington, D.C.: World Bank Group. nomics and Financial Systems. International Mone- tary Fund. Washington, D.C. Taming Volatility Chapter 2. Output Volatility and the Cyclicality of Fiscal Policy in the OECS Introduction cyclicality differs in rich and developing countries and presents new evidence on the “when it rains it pours” T here is a wealth of evidence on the pro-cy- phenomenon. Section 4 presents stylized facts on the clicality of fiscal policy in industrialized and cyclicality and volatility of key macroeconomic vari- emerging market economies. The latter group ables for the OECS countries. Section 5 discusses good is usually associated with a more pro-cyclical behav- practice policy options that could help the countries in ior of fiscal policy. That is, developing countries tend to the region to strengthen fiscal discipline and respond orient government consumption and investment in the more symmetrically to the business cycle. Concluding remarks are discussed in Section 6. The chapter in- 29 same direction as that of the cycle in general economic activity. In doing so, they tend to amplify upswings and cludes an appendix with methodological details of the worsen recessions – what Kaminsky, Reinhart, and Vegh econometric analysis used to investigate the main de- (2004) termed as the “when it rains, it pours” phenome- terminants of fiscal policy cyclicality. non. Industrialized countries, on the other hand, are be- lieved to behave largely in a counter-cyclical or even in Why Should Countries Avoid a a-cyclical fashion.3 Procyclical Fiscal Policy? In this chapter, we provide new evidence on the cycli- From a theoretical point of view, the pro-cyclical- cality of fiscal policy across countries placing a spe- ity of fiscal policy remains a puzzle. Governments cial emphasis on the Eastern Caribbean. This analy- should be able to exercise discretion and avoid making sis is important and sets the stage for the following a bad situation turn worse. From that perspective, it chapter where we test the hypothesis that a pro-cy- does not seem to be optimal to exacerbate the busi- clical fiscal policy in the context of weaknesses in the ness cycle by adopting an expansionary fiscal poli- financial sector exacerbate the negative effects of out- cy in booms and a contractionary fiscal stance in a put volatility on economic growth in the Caribbean and downturn. In what economists call the neoclassical elsewhere. After this introduction, Section 2 discusses world, marked by less intervention of the state in the why countries are better off avoiding a pro-cyclical fis- economy, the optimal fiscal policy stance is either cal policy stance. Section 3 assesses how fiscal policy a-cyclical (Barro (1979)) or counter-cyclical (Baxter and King (1993)). In contrast, in a Keynesian frame- 3  Evidence on the pro-cyclical pattern of fiscal policy in devel- oping countries was first found by Gavin and Perotti (1997) who work, in the presence of price and wage rigidity and showed that Latin American tended to adopt policies that were a greater presence of the state in the economy, the expansionary in good times and contractionary in bad times. Talvi optimal fiscal policy stance is counter-cyclical (Chris- and Vegh (2000) then showed that such behavior was far from be- ing a trademark of Latin America alone as many other developing tiano et al. (2011) and Nakata (2011)). countries across the world espoused a pro-cyclical fiscal policy stance. Taming Volatility Irrespective of theoretical prescriptions, fiscal policy to reduce political rents. During a boom, voters demand remains rather pro-cyclical in a number of countries. more public goods and lower taxes in order to prevent The reasons for such behavior are various. One expla- corrupt politicians from appropriating tax revenues. nation is that frictions in international credit markets might prevent countries from borrowing in bad times A very clear economic reason as to why governments forcing these countries to lower spending during reces- should avoid reinforcing the business cycle is associ- sions (Gavin and Perotti (1997)). A pro-cyclical fiscal pol- ated with their ability to lean against the wind in pe- icy stance could also be seen as a signal of weak and riods of hardship. In addition, a pro-cyclical fiscal policy underdeveloped institutions in governments where there can hamper government efforts to reduce the effects of is limited technical capacity to develop the means to volatility on growth. In general, government spending save in good times to spend in bad times (e.g., Calderon can mitigate the negative effect that volatility has on and Schimdt-Hebel (2008)). Another explanation for growth if: (i) government spending is counter-cyclical; pro-cyclical government expenditures relies on political and (ii) its impact on output is positive. For conceptual economy reasons which suggest that during good times clarity it is useful to note that: (i) refers to the behavior governments face political pressures and temptations of government spending, i.e. the response of govern- to keep spending high and run fiscal deficits. In addition, ment spending to the business-cycle; (ii) refers to the delays in the implementation and execution of fiscal pol- government spending multiplier, i.e. the effectiveness of icies could contribute to fiscal policy pro-cyclicality. government spending with regard to changing output in the short-run (and possibly the long-run). If government A more recent strand of the literature has argued spending is pro-cyclical and the government spending that the cyclicality of government spending depends multiplier is positive, this will then increase the variance on the quality of institutions. Alesina et al. (2008), for of exogenous shocks which will tend to have a more neg- example, argue that the pro-cyclicality of government ative effect on growth. spending is related to corruption. Based on a sample of 83 countries during 1960-2003, they find that in de- From a risk management point of view, a counter-cy- 30 mocracies with higher levels of corruption government clical fiscal policy can be useful for at least three com- spending tends to be more pro-cyclical. The authors ex- pelling reasons. First, by leaning against the wind, gov- plain their finding through the lens of a political econo- ernments can continue to provide goods and services my model where voters “starve the Leviathan” in order and to maintain public investment even in the event of a Fi ur 2.1 A Subst ntiv Numb r of D v lopin Countri s Follows Pro-C clic l Fisc l Polic Corr l tion co ffici nts b tw n r l ov rnm nt xp nditur s nd r l GDP: 1980-2013 0.8 0.6 0.4 0.2 -0.0 -0.2 -0.4 -0.6 United Arab Emirates Switzerland United Kingdom Brazil United States Australia Guatemala Bangladesh Uganda Oman Germany Congo, Dem. Rep. Belgium Congo, Rep. Jordan India Norway Argentina Egypt, Arab Rep. Korea, Rep. Pakistan Turkey Portugal Kenya Uruguay Mozambique Azerbaijan Philippines Trinidad and Tobago Sierra Leone Dominican Republic Gabon Source: Authors’ own calculations based on the IMF’s World Economic Outlook (WEO) for real GDP and for real government expenditures over the period 1980-2013. Note: The information reported by WEO on real government expenditures refers to General Government Net Lending (GGXCNL) defined as the difference between General Government’s revenues and expenditures. Taming Volatility drop in public revenues. Second, in a downturn, a coun- oping countries across the world espoused a pro-cyclical tercyclical fiscal policy can help governments increase fiscal policy stance. social assistance and insurance to a large number of cit- izens affected by more adverse macroeconomic condi- There is a number of different explanations as to why tions. Third, as witnessed during the global financial cri- developing countries tend to behave in that fashion sis of 2008-09, a countercyclical fiscal policy can help vis à vis industrialized economies. Some of the rea- countries stimulate the economy and cope better with sons most commonly found in the literature include the effects of a prolonged recession (see World Bank credit constraints faced by developing countries, which (2014)). In that regard, as argued by Mollick et al. (2011), would prevent them from raising money in international in the aftermath of the global financial crisis it became capital markets in bad times and would force them to clear that the countries that weathered the effects of adopt a contractionary fiscal policy in downturns Gavin the crisis better were those that had followed some sort & Perotti (1997). Political economy considerations would of concerted macro-fiscal responses that helped them also seem to play a role as good times could encourage build resilience to exogenous shocks. When the crisis fiscal profligacy (Tornell & Lane (1999), Alesina & Tabel- hit, these countries were able to have more favorable lini (2005)). access to credit in international financial markets, and resist speculative attacks. Not only they had stronger A more interesting and elucidating exercise requires fundamentals entering the crisis, which allowed them to investigating whether countries that have followed a quickly deploy counter-cyclical fiscal policies, they were pro-cyclical fiscal policy have been able to switch to a also better positioned to adopt exit strategies faster by more counter-cyclical fiscal policy stance over time. raising interest rates, controlling domestic credit growth To address this question, we have constructed a proxy and reverting to more orthodox fiscal policies. for fiscal policy cyclicality based on correlation coeffi- cients for time series of real government expenditures How Does Fiscal Policy Cyclicality and real GDP smoothed by the Hodrick-Prescott filter, Differ in Rich and Developing following the same methodology as in Frankel at al. Countries? (2013). This allowed us to classify countries according 31 to their ability to move from a pro-cyclical to a count- A significant number of authors have documented a er-cyclical fiscal policy stance. A negative (positive) cor- more pro-cyclical behavior of fiscal policy in develop- relation coefficient between the cyclical component of ing countries. Industrialized countries, in turn, tend to government spending and GDP indicates a counter-cy- behave largely in a counter-cyclical or at worst a-cycli- clical (pro-cyclical) fiscal policy stance. Figure 2.2 plots cal fashion (see Figure 2.1). An idea put forward by Ka- countries according to their “graduating classes”, which minsky, Reinhart, & Végh (2004) was that, for develop- are defined as follows: ing countries, and in particular for upper middle-income countries, macroeconomic policies and in special fiscal i) Established graduates (EG) as those with count- policy tend to reinforce the business cycle. These au- er-cyclical fiscal policies in the first and second peri- thors coined this behavior as the when-it-rains-it-pours ods; syndrome and we will shed some light on its relevance ii) Recent graduates (RG) as those with pro-cyclical in the context of the Caribbean and the OECS further policies in the first period and counter-cyclical in the down below. second; iii) Still in school (SS) as those with pro-cyclical fiscal More broadly, emerging markets in Latin America policies in the first and second periods; and, and elsewhere have a long track-record of pro-cycli- iv) Back to school (BS) as those with counter-cyclical cal government spending. Evidence on the pro-cyclical fiscal policies in the first period and pro-cyclical in pattern of fiscal policy in developing countries was first the second. found by Gavin & Perotti (1997) who showed that Latin American was much more expansionary in good times The large majority of Caribbean countries, including and contractionary in bad times. Talvi & Vegh (2000) the ones in the OECS sub-region have yet to graduate then showed that such behavior was far from being a from procyclical fiscal policies. That is, they are either trademark of Latin America alone as many other devel- “still in school” or “back to school”. Not surprisingly, most Taming Volatility Fi ur 2.2 Gr du tin Cl ss s for Fisc l Polic C clic lit C clic lit of R l Gov rnm nt Exp nditur s. 1990-2005 vs 2005-2013 1 0.8 Back to School Still in School 0.6 DMA 0.4 HTI Correlation 2000-2010 ATG LCA KNA 0.2 DOM BRG BHS GUY JAM 0 TTO VCT BLZ -0.2 GRD SUR -0.4 -0.6 -0.8 -1 Establish Graduate Recent Graduate -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Correlation 1990-2000 Note: Proxy for fiscal cyclicality based on correlation coefficients for time series of real government expenditures and real GDP smoothed by the Hodrick-Prescott filter. We use a sub-period, 1990-2010, from the overall 1980-2013 dataset because for many countries, data on real government expenditures and/or on GDP are missing, mostly for the early 1980s. Caribbean countries are highlighted in red. Source: Carneiro and Garrido (2015) based on IMF’s World Economic Outlook (WEO). of the countries classified as “established graduates” in correlation coefficients (not the Boolean classification are industrialized economies, as noted earlier (see Fig- of countries in graduating classes) the evidence on Ca- ure 2.1). In Figure 2.2, most of the countries in the up- ribbean countries moving away from pro-cyclical fiscal per quadrants are developing economies. Interestingly, policies is more compelling under this country-specific a number of those in the “recent graduates” class are approach. Twelve out of fifteen economies now show emerging economies. Among the Caribbean countries, a declining correlation coefficient for the fiscal stance 32 Guyana is the only one which made it to the “established proxy (see Figure 2.3). graduates” class. Grenada and Jamaica, which have been implementing fiscal adjustment programs under Does it really pour when it rains? an IMF program, have made it to the “recent graduate” class along with Suriname, Belize, and St. Vincent and We make two implicit assumptions as to why coun- the Grenadines. Note, however, that these results are tries might change their fiscal stance during booms sensitive to the time period in which correlation coeffi- and downturns. First, we consider changes as generally cients are computed, as we discuss below. not random. That is, they are mostly associated to policy shifts within given administrations, which may or may The robustness of these results was tested by inves- not be politically motivated (as incumbent administra- tigating whether the graduating classes changed sig- tions tend to spend more ahead of elections), or across nificantly if the sample period was split in different administrations after elections, influenced by ideological points in time. We did this by running tests of structur- principles. Second, we assume that those changes are al break for the time series of real government expendi- generally driven or motivated by observed trends in eco- tures for each country and splitting the sample when a nomic activity and not the other way around. This as- structural break was identified (see Carneiro and Garri- sumption is not uncontroversial. For instance, Rigobon do (2015) for details). After following this approach we (2004) argues that fiscal policy shocks drive output and confirmed only one Caribbean country actually classi- not the other way around, while Ilzetzki & Vegh (2008), fied as an established graduate (Guyana) and 3-4 coun- on the other hand, find causality running both ways. tries moving away from pro-cyclical to counter-cyclical policies, compared to 1-3 countries falling back into Keeping these assumptions in mind we have re-com- pro-cyclical fiscal policies.4 When one looks at changes puted correlation coefficients for the fiscal cyclicality proxy for 180 countries over the period 1990-2011 by 4  Only Jamaica and Trinidad and Tobago are recent graduates differentiating what happens in fiscal policy in differ- under all filtering methods; and only St. Kitts and Nevis is consis- ent parts of the business cycle. This is an important tently back in school for all methods. Taming Volatility Fi ur 2.3 C clic lit of R l Gov rnm nt Exp nditur s und r Countr -Sp cific Y r Br ks 1 Back to School LCA Still in School 0.8 HTI 0.6 0.4 BRB DMA BLZ KNA Correlation 2000-2010 ATG DOM 0.2 BHS VCT 0 GRD SUR -0.2 JAM -0.4 TTO GUY -0.6 -0.8 -1 Establish Graduate Recent Graduate -1 -0.8 -0.6 -0.2 0 0.2 0.4 0.6 0.8 1 Correlation 1990-2000 Note: For this chart we use data for real government expenditures and real GDP covering 180 countries for the period 1980-2013. We also apply a criterion for splitting the sample, by country, at potentially different points in time based on potential structural breaks in (per capita GDP) data. The intuition is that changes in macroeconomic policies and performance generally reflect on changes in per capita income. A structural break on the series for the latter may be indicative of changes in fiscal stance. Source: Carneiro and Garrido (2015) based on IMF, WEO. exercise because it allows one to have an indication of fiscal policies in both booms and downturns. Other the impact of fiscal policy (pro) cyclicality might have things equal, such stance contributes to exacerbate on output volatility, for example. Countries that are, on output volatility. Not surprisingly, one finds many re- average, pro-cyclical in booms and downturns, would source-rich economies in this category. In addition, tend to exacerbate their business cycle; those that are many upper middle-income countries appear promi- counter-cyclical in both, booms and downturns, have a nently in that group. fiscal policy that contributes to stabilize the cycle. • Upper left quadrant: Those that exhibit counter-cy- 33 clical fiscal policies in booms and pro-cyclical fiscal Because countries may not always be pro-cyclical or policies in downturns. Other things equal, such fiscal counter-cyclical we have considered different scenar- behavior improves a country’s fiscal sustainability ios. Whenever a country exhibits an average counter-cy- profile. clical fiscal stance in booms, and a pro-cyclical stance in • Lower left quadrant: Those that exhibit counter-cycli- downturns, other things equal, it will likely improve its cal fiscal policies in both booms and downturns. Oth- medium to long term fiscal sustainability profile. A coun- er things equal, such stance contributes to stabilize try that is pro-cyclical in booms and counter-cyclical in output around its long-term trend. Expectedly, most downturns would, ceteris paribus, deteriorate its fiscal of high-income countries fall under this category. sustainability profile. • Lower right quadrant: Those that exhibit pro-cyclical fiscal policies in booms and counter-cyclical fiscal An interesting pattern emerges when we classify policies in downturns. Other things equal, such be- countries according to their fiscal policy stance over havior deteriorates a country’s fiscal sustainability the different phases of the business cycle. Figure 2.4 profile. plots the value of the fiscal stance proxy in periods of expansion (when the cyclical component of real GDP is We have confirmed earlier findings that emerging and positive) against that registered in downturns. We have industrialized countries tend to behave in a different identified high-income countries in red while developing fashion over the business cycle. Most of the countries economies appear in blue. By quadrants5, we identify in the upper and lower right quadrants in Figure 2.4 are four groups of countries: developing economies (in blue) and most important- ly upper middle-income countries. In contrast to that, • Upper right quadrant: Those that exhibit pro-cyclical most of the high income countries appear on the upper and lower left quadrants with fiscal stances that largely 5  Initial debt levels might have a big impact on which quadrant a contribute to long-term fiscal sustainability. country ends up falling into. Taming Volatility Fi ur 2.4 Fisc l C clic lit in Booms nd Downturns (1990–2011) 1 0.8 Positive e ect in L-T Fiscal Exacerbate volatility Sustainbility; Counter-cyclical in ATG Pro-cyclical in booms 0.6 booms; pro-cyclical in downturns QAT and downturns TTO Fiscal Cyclicality Downturns 0.4 KOR GRD ISL 0.2 CHE PRT NZL AUT SVN HUN 0 ARE SAU CZE -0.2 JPN NOR AUS NLD DEU NOR ITA OMN BEL BHS EST -0.4 SGP FRA LUX ESP -0.6 BHR FIN Contribute to Stability: Negative e ect in L-T Fiscal -0.8 Counter-cyclical in DNK GBR Sustainbility; Pro-cyclical in booms; -1 booms and downturns USA counter-cyclical in downturns -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Fiscal cyclicality in Booms Note: Proxy for fiscal cyclicality based on correlation coefficients for time series of real government expenditures and real GDP smoothed by the Baxter-King filter. Source: Carneiro and Garrido (2015) based on IMF’s World Economic Outlook (WEO). Some of these results may seem counter-intuitive but OECS countries can rely only on fiscal policy as an eco- they actually reveal the benefits associated with fis- nomic stabilization tool, which makes a counter-cyclical cal discipline. For instance, one may be surprised to see fiscal policy even more important to control output vol- Chile, a country that has earned a reputation of fiscal atility. prudency and good overall macro management, in the fourth quadrant. As it turns out, Chile is, on average, for When we looked at the determinants of fiscal policy the period 1990-2011, moderately pro-cyclical in booms cyclicality, we found robust evidence of the relevance and markedly anti-cyclical in downturns. Their ability to of institutional quality for a country’s ability to move 34 sustain a strong fiscal position arises from having a sys- to counter-cyclical fiscal policies. The results are ro- tem of buoyant tax revenues and the great contribution bust to different definitions of fiscal cyclicality, different of the private sector to economic activity, so the coun- sample periods, alternative ways of measuring institu- try is able to register solid, positive fiscal balances both tional quality, and to the inclusion of other control vari- in booms and recessions with marked improvements ables in the estimations.6 However, for the sub-sample in its overall fiscal stance. Compare this performance of Caribbean countries, the evidence in that regard is with that for Greece, for example, especially during the more mixed. The three Caribbean countries that became post-financial crisis period when it showed deteriorating more pro-cyclical (St. Kitts and Nevis, St. Lucia and Bar- fiscal balances, faster increase in expenditures relative bados) are among those with the highest improvement to revenues, and poor economic performance, with an in institutional quality (see Figure 2.6).7 exacerbated contribution to volatility derived from a more pro-cyclical fiscal stance. With this, there is no The importance of institutional quality as an import- doubt that Greece has yet to earn enough stars to join ant determinant of a country’s ability to behave in the same status as Chile’s. a counter-cyclical way was confirmed after we ad- dressed potential endogeneity concerns. Endogeneity Most of the OECS economies have adopted a fiscal concerns are common in the context of analysis involv- stance that seems to exacerbate output volatility. ing institutional quality and other macroeconomic vari- That is, they tend to behave in a pro-cyclical way during 6  See Appendix for methodological details and Carneiro and booms and downturns. The only exception seems to be Garrido (2015) for a full set of results. St. Kitts and Nevis which appears together with Trinidad 7  Compare an average institutional quality index, based on the and Tobago, and Belize, in the “back to school” quadrant, Kunčič (2013) dataset, equal of 0.54 (in a 0-1 scale; with higher values indicating better institutions) for the Caribbean countries where countries show a counter-cyclical behavior in relative to an average index equal to 0.39 for the sample of lower booms, which is a good indication that they are saving middle income countries and 0.56 for the sample of upper middle income countries in the period 1990-2000. The variance for Ca- for rainy days, but still display a pro-cyclical behavior in ribbean countries is 0.01. All countries, except Haiti have an initial downturns (Figure 2.5). As members of the ECCU, the IQ index of 0.4 or higher, and 10 out of 15 have initial IQ of 0.5 or higher. Taming Volatility Fi ur 2.5 C ribb n Countri s: Fisc l C clic lit in Booms nd Downturns (1990-2011) 1 Positive e ect in L-T Fiscal Exacerbate volatility: 0.8 Sustainability: Counter-cyclical in KNA ATG BRB Procyclical in booms 0.6 HTI Fiscal Cyclicality in Downturns booms; pro-cyclical in downturns and downturns 0.4 TTO DMA VCT GRD 0.2 0 BLZ DOM LCA -0.2 JAM SUR -0.4 GUY BHS -0.6 Contribute to Stability: Negative e ect in L-T Fiscal -0.8 Counter-cyclical in booms Sustainability: Pro-cyclical in booms; and downturns counter-cyclical in downturns -1 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Fiscal cyclicality in Booms Source: Carneiro and Garrido (2015) based on IMF’s World Economic Outlook (WEO). Note: Proxy for fiscal cyclicality based on correlation coefficients for time series of real government expenditures and real GDP smoothed by the Baxter-King filter. ables because it may be difficult to establish the direc- Fi ur 2.6 Ch n in Fisc l St nc Prox vs tion of causality. We addressed this issue by following Ch n in Institution l Qu lit Ind x the approach used by Acemoglu, Johnson, & Robinson 1.5 Change Correlation (Second -First Period) (2001) to look for potential endogeneity problems (see Appendix 1 for methodological details).8 We used alter- 1 native definitions of institutional quality (International LCA KNA Country Risk Guidance (ICRG) and Kunčič (2013)) and 0.5 BRB alternative fiscal stance proxies. Instruments for insti- DMA GRD VCT 0 tutional quality are European settlers’ mortality rates HTI ATG SUR 35 BLZ JAM (in logs) and latitude. We have also added a set of con- -0.5 BHS TTO trols, including financial depth, financial integration, -1 DOM debt to GDP ratios, government accountability, output volatility and the reserves to import ratio. Tables 2.1 and -1.5 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 2.2 show the results of ordinary least square estima- Change Institutional Quality Index: 2010 - 1990 tions (OLS) and instrumental variable (IV) specifications under the alternative institutional quality (IQ) and fiscal Note: A positive change in correlation indicates the country has become more pro-cyclical. The size of the bubbles indicate the stance proxies. In all cases we were able to find evidence initial quality of institutions. Source: Authors’ calculations based on IMF, WEO, and Kuncic (2013). of causality running from IQ to fiscal stance confirming the robustness of our results. over the past several decades; the business cycle facts Pro-Cyclicality and Macroeconomic in these economies, such as the volatility of output, con- Volatility in the OECS sumption, investment, trade balance, and real interest rates; the cyclicality of these variables with output and In what follows, we focus on stylized facts of the interest rate, as well as their persistence; and the prop- OECS economies for a number of key economic vari- erties of fiscal policy in these countries, with a particular ables over the business cycle. In that regard, we docu- focus on its cyclical characteristics.9 ment in detail the growth experience of OECS countries 9  All data are from the International Monetary Fund’s World Eco- 8  Acemoglu et al. (2001) provided an excellent framework for the nomic Outlook (WEO) database and cover the period 1980-2014 at analysis of institutional quality on current economic performance, annual frequency. To transform the data into real terms, all nomi- by tracing historical relationships of European settler mortality to nal quantities are deflated by the GDP deflator. Interest rate is real the type of settlements and quality of early institutions in former lending rate obtained as the difference between the lending rate colonies, to current quality of institution and level of income. From and the consumer price index (CPI) inflation rate. We also consid- their methodological point of view, the authors were able to iden- ered an effective interest rate on government debt computed as tify a source of exogenous variation in institutions that affected a the ratio of general government interest expenditures and gross country’s current per capita income. debt. The two rates are highly correlated, with the average correla- Taming Volatility Table 2.1: OLS and Two-Stage Least Square for Instrumental Variable Estimation regarding the effect of Institutional Quality on fiscal cyclicality stance without control variables Dependent Variable is the Correlation Coefficient of Real Government Expenditures and Real GDP, 1990-2011 (Alternative filtering methods) OLS IV OLS IV OLS IV OLS IV VARIABLES Hodrick-Prescott Baxter-King Christiano-Fitzgerald Butterworth IQ: ICRG Avg. -1.119*** -1.284*** -0.941*** -1.170*** -0.996*** -1.017*** -0.838*** -1.039*** 1990-2011 (-7.096) (-4.550) (-5.771) (-4.112) (-6.056) (-3.469) (-4.951) (-3.305) Constant 0.718*** 0.751*** 0.618*** 0.701*** 0.651*** 0.619*** 0.530*** 0.596*** (7.633) (5.103) (6.351) (4.724) (6.636) (4.049) (5.249) (3.630) Observations 123 71 123 71 123 71 123 71 R-squared 0.294 0.360 0.216 0.305 0.233 0.288 0.168 0.219 r2_a 0.288 0.351 0.209 0.295 0.226 0.277 0.162 0.207 F 50.36 20.71 33.31 16.91 36.67 12.04 24.51 10.93 Dependent Variable is the Correlation Coefficient of Real Government Expenditures and Real GDP, 1990-2011 (Alternative filtering methods) OLS IV OLS IV OLS IV OLS IV VARIABLES Hodrick-Prescott Baxter-King Christiano-Fitzgerald Butterworth IQ: Kuncic Avg. -0.973*** -1.429*** -0.787*** -1.292*** -0.853*** -1.233*** -0.738*** -1.143*** 1990-2011 (-5.805) (-4.058) (-4.595) (-3.652) (-4.998) (-3.594) (-4.231) (-3.143) Constant 0.611*** 0.813*** 0.509*** 0.745*** 0.547*** 0.718*** 0.455*** 0.639*** (6.642) (4.633) (5.405) (4.209) (5.844) (4.195) (4.749) (3.524) 36 Observations 145 84 144 83 145 84 145 84 R-squared 0.191 0.130 0.129 0.065 0.149 0.128 0.111 0.084 r2_a 0.185 0.119 0.123 0.0537 0.143 0.117 0.105 0.0724 F 33.70 16.47 21.12 13.34 24.98 12.92 17.9 9.881 Note:Instruments are log of settler mortality following Acemoglu, Johnson & Robinson (2001). t-statistics in parentheses *** p< 0.01, ** p<0.05, *p<0.1 Government spending can mitigate the negative ef- run (and possibly the long-run). If government spending fect that volatility has on growth in different ways. is pro-cyclical and the government spending multiplier This can happen, for example, if government spending is is positive, then increases in the variance of exogenous counter-cyclical, and when its impact on output is pos- shocks will have a more negative effect on growth. itive. For conceptual clarity it is useful to note that the first effect refers to the behavior of government spend- Over the last 35 years or so, the OECS region record- ing, i.e. the response of government spending to the busi- ed average positive growth rates albeit while keeping ness-cycle. The second effect refers to the government low investment levels. We begin by reporting the key spending multiplier, i.e. the effectiveness of government properties of some of the variables that best character- spending with regard to changing output in the short- ize the region’s economic performance. Table 2.3 pres- ents the GDP growth rates for the OECS countries, the tion across countries equal to 0.89. As a result, the stylized facts average real interest rate as well as the average invest- are very similar for the two interest rates, with the main difference being the lower average effective interest rate as compared to the ment and government expenditures as a share of GDP. lending rate. We choose to proceed with the lending rate, as it bet- We also report the net foreign asset (NFA) position of ter reflects the cost of borrowing to the private firms. While this the OECS member countries. The OECS countries were rate does not capture the cost of borrowing internationally for the domestic firms, the fact that it has very similar dynamics to the growing at an average rate of 3 percent per year over interest rate on government debt (both domestic and foreign) gives the 1980-2014 period. The average interest rate was us some confidence in its appropriateness for our analysis. Taming Volatility Table 2.2: OLS and Two-Stage Least Square for Instrumental Variable Estimation regarding the effect of Institutional Quality on fiscal cyclicality stance with control variables Dependent Variable is the Correlation Coefficient of Real Government Expenditures and Real GDP, 1990-2011 (Al- ternative filtering methods) OLS IV OLS IV OLS IV OLS IV VARIABLES Hodrick-Prescott Baxter-King Christiano-Fitzgerald Butterworth IQ: ICRG Avg. 1990-2010 -1.028*** -1.650* -0.732** -1.178 -1.130*** -1.448 -0.697** -1.516 (-3.903) (-1.926) (-2.586) (-1.394) (-3.929) (-1.602) (-2.411) (-1.513) Chinn-Ito Index of capital -0.009 0.021 -0.041 0.009 -0.001 0.005 -0.022 0.034 openness, average 1990- (-0.348) (0.553) (-1.507) (0.230) (-0.036) (0.106) (-0.785) (0.768) 2011 Financial depth (M2 Ratio 0 0.001 0 0 0.001 0.002 0 0.002 to GDP), average 1990- (-0.295) (0.462) (0.046) (-0.086) (0.477) (0.576) (-0.297) (0.513) 2011 GDP volatility (Squared Cy- 36.180** 78.034 33.693 105.036** 0.376 -2.008 27.232 185.837 clical Component of GDP), (2.450) (1.625) (1.110) (2.082) (0.411) (-1.188) (0.607) (1.043) average 1990-2011(x) Proxy for political Checks -0.003 0.002 -0.011 -0.010 -0.002 -0.009 -0.006 -0.002 and Balances, average (-0.132) (0.064) (-0.489) (-0.377) (-0.089) (-0.298) (-0.253) (-0.063) 1990-2011 Debt to GDP ratio, average -0.001 -0.001 -0.001 -0.002** -0.001 -0.001 0 -0.001 1990-2011 (-1.106) (-1.593) (-1.118) (-2.440) (-1.142) (-0.766) (-0.555) (-1.551) Reserves to Imports Ratio, -0.012 -0.019 -0.007 -0.019* -0.001 -0.015 -0.006 -0.024 average 1990-2011 (-1.595) (-1.599) (-1.058) (-1.696) (-0.221) (-1.228) (-0.793) (-1.782) Constant 0.775*** 1.016** 0.616*** 0.926* 0.757*** 0.946*** 0.550*** 0.949*** (5.223) (2.550) (3.694) (2.382) (4.793) (2.389) (3.361) (2.032) 37 Observations 122 71 122 71 122 71 122 71 R-squared 0.342 0.428 0.258 0.432 0.249 0.349 0.187 0.296 r2_a 0.301 0.364 0.212 0.369 0.203 0.277 0.137 0.218 F 8.450 5.133 5.660 5.608 5.405 2.449 3.744 2.917 Note:Instruments are log of settler mortality following Acemoglu, Johnson & Robinson (2001). t-statistics in parentheses *** p< 0.01, ** p<0.05, *p<0.1 at 7.22 percent, with the lowest average interest rate pared to developed economies and even compared to a observed in Barbados at 5.72 percent and the highest few other developing countries (see Neumeyer and Perri, average rate – in Antigua and Barbuda at 8.39 percent. 2005). The highest GDP volatility is observed in Antigua Investment stood at 27 percent of GDP on average, and Barbuda and in Grenada. Interest rate is also quite while government expenditures at 30 percent of GDP volatile in these countries but the magnitudes are simi- across this group of countries over our sample period. lar across them, with the exception of St. Lucia. The vol- All OECS countries had positive NFA, with the average atility of trade balance and terms of trade shows much NFA-to-GDP ratio equal to 12.6 percent. Thus, all these more dispersion across the OECS countries, although countries were net lenders to the rest of the world during the main result stands – the OECS countries exhibit very 1980-2014, on average. volatile business cycles, even compared to developing economies. The region has seen high volatility in many of its rel- evant macroeconomic aggregates. Table 2.4 reports Consumption and government expenditures are the percentage standard deviation of the key variables, more volatile in the Caribbean than elsewhere. As is as well as the relative standard deviation. It shows that commonly observed in the business cycles literature, all variables are quite volatile with the average real GDP investment is the most volatile variable among the ex- volatility equal to 3.72 percent. This is quite high com- penditure components of GDP, with the relative volatil- Taming Volatility Table 2.3: Averages Values for Selected Key Macroeconomics Variables: 1980-2014 GDP growth Int rate Inv/GDP Gov exp/GDP NFA/GDP Antigua and Barbuda 3.23 8.39 33.73 26.35 9.16 Barbados * 0.84 5.72 15.63 37.13 8.61 Dominica 2.71 6.94 20.20 32.34 14.82 Grenada 3.37 6.98 29.87 27.65 11.20 St. Kitts and Nevis 3.80 7.19 35.65 30.32 23.96 St. Lucia 3.52 7.94 24.85 26.24 2.60 St. Vincent and the Grenadines 3.50 7.34 25.70 27.79 18.07 Average 3.00 7.22 26.52 29.69 12.63 Notes: (*) We note that Barbados is not a member of the OECS region, nor the ECCU, but given its proximity and similarity with the OECS countries we have included it in the analysis. Int rate is the real interest rate computed as the lending rate minus inflation rate. Source: World Economic Outlook. Table 2.4: Volatility of Key Macroeconomic Variables: 1980–2014 % Standard deviation % Standard deviation of x % Standard deviation of GDP GDP TB/GDP TOT Int rate Inv Gov exp Cons Antigua and Barbuda 5.38 6.98 6.76 2.87 3.42 2.64 4.56 Barbados 3.56 2.31 7.95 2.28 4.82 1.22 1.06 38 Dominica 2.49 6.17 4.03 2.90 6.88 3.96 2.82 Grenada 4.28 4.61 7.24 2.72 4.19 2.36 1.64 St. Kitts and Nevis 3.55 6.07 4.01 2.20 5.26 3.69 3.87 St. Lucia 3.63 5.04 6.20 3.23 4.07 1.46 2.50 St. Vincent and the Grenadines 3.11 5.32 6.04 2.63 3.33 2.50 2.17 Average 3.72 5.21 6.03 2.69 4.57 2.55 2.66 Notes: TOT is terms of trade; TB/GDP is exports minus imports over GDP; Int rate is the real interest rate computed as the lending rate minus inflation rate. All series (except trade balance, terms of trade and real interest rate) were log-transformed. To obtain the cyclical components of the key variables, they were Hodrick-Prescott (HP) filtered with the smoothing parameter of 100. Source: World Eco- nomic Outlook. ity in the OECS countries equal to 4.6 on average. This all OECS countries the relative volatility of consump- number is comparable to that found for other countries. tion to GDP is well above 1. Government expenditures also exhibit higher volatility than GDP in the OECS countries by a factor of 2.55. The examination of the cyclical properties of the key This is in line with the findings in Male (2010) for devel- variables reveals important insights. Table 2.5 reports oping countries. In contrast, in developed countries the the correlations between GDP and various macroeco- ratio of volatility of government expenditure and GDP nomic aggregates. We notice, for example, that invest- tends to be closer to 1. We found that the volatility of ment and private consumption are pro-cyclical, with consumption in the OECS countries is strikingly high, the average correlations equal to 0.61 and 0.36, respec- equal to 2.66 times that of GDP volatility. This is quite tively. We also find that government expenditures are high compared to both developing countries and de- pro-cyclical with a correlation coefficient of 0.39. This veloped economies. The highest consumption volatility finding confirms earlier results in the literature that de- is observed in Antigua and Barbuda and St. Kitts and veloping countries tend to follow pro-cyclical fiscal poli- Nevis, while the lowest – in Barbados and Grenada. In cies. In addition, we find that interest rates in the OECS Taming Volatility Table 2.5: The Correlation of GDP with Key Macroeconomic Variables: 1980–2014 Correlation of GDP with Inv Gov exp Cons TB/GDP TOT Int rate Antigua and Barbuda 0.48 0.31 0.63 -0.44 -0.27 -0.33 Barbados 0.79 0.20 0.50 -0.33 0.39 -0.27 Dominica 0.56 0.65 0.23 -0.32 -0.23 -0.30 Grenada 0.56 0.49 0.32 -0.10 0.19 -0.21 St. Kitts and Nevis 0.65 0.38 0.04 -0.33 0.31 -0.32 St. Lucia 0.72 0.22 0.44 -0.27 -0.32 -0.55 St. Vincent and the Grenadines 0.55 0.52 0.38 -0.10 -0.08 -0.51 Average 0.61 0.39 0.36 -0.27 0.00 -0.36 Notes: TOT is terms of trade; TB/GDP is exports minus imports over GDP; Int rate is the real interest rate computed as the lending rate minus inflation rate. Source: World Economic Outlook. Table 2.6: The Correlation of Interest Rates with Key Macroeconomic Variables: 1980–2014 Correlation of Interest rate with GDP Inv Gov exp Cons TB/GDP TOT Effective int rate Antigua and Barbuda -0.33 -0.30 0.04 -0.21 -0.08 0.00 0.79 Barbados -0.27 -0.21 -0.17 -0.40 0.39 0.11 0.93 Dominica -0.30 -0.50 -0.18 -0.22 0.72 0.43 0.87 39 Grenada -0.21 -0.01 0.01 -0.35 -0.06 0.04 0.92 St. Kitts and Nevis -0.32 -0.06 0.24 -0.37 0.03 -0.04 0.94 St. Lucia -0.55 -0.53 0.02 -0.27 0.31 0.14 0.86 St. Vincent and the Grenadines -0.51 -0.43 -0.23 -0.51 0.40 0.37 0.92 Average -0.36 -0.29 -0.04 -0.33 0.24 0.15 0.89 Notes: TOT is terms of trade; TB/GDP is exports minus imports over GDP; Int rate is the real interest rate computed as the lending rate minus inflation rate. Source: World Economic Outlook. countries are countercyclical – another important result investment, government expenditures, and consump- that distinguishes the business cycles in developing tion. Indeed, we find that this is the case for investment, countries – with the correlations ranging from -0.55 in where the correlation with interest rate is equal to -0.29, St. Lucia to -0.21 in Grenada. With the currency board, on average; and for consumption, where the correlation an external shock generally translates into a decline in ranged from -0.51 in St. Vincent and the Grenadines to international reserves and monetary contraction that -0.21 in Antigua and Barbuda, with the average of -0.33. leads to an increase in interest rates. Lastly, we show The correlation of the interest rate with government that net exports are countercyclical, in line with the expenditures was close to 0, on average, although with findings for developing economies in other studies. a significant spread ranging from -0.23 for St. Vincent and the Grenadines and 0.24 for St. Kitts and Nevis (Ta- Finally, we look at the correlation of real interest rates ble 2.6). Our finding that real interest rates are volatile with various macroeconomic variables. The fact that and countercyclical in the OECS countries mirrors the real interest rates are countercyclical with GDP sug- results in Neumeyer and Perri (2005), Uribe and Yue gests that they may also be negatively correlated with (2006), and others for developing economies. Taming Volatility How to Make Fiscal Policy More in advanced and emerging economies. Factors that Effective and Less Procyclical motivated their adoption range from reigning in debt excesses that resulted from banking and economic Effective fiscal policies respond symmetrically to the crises in the early 1990s (e.g., Finland, Sweden) and business cycle and help countries build positive rep- debt crises in Latin American countries (e.g., Brazil, utational capital. As this chapter has argued, discre- Peru), consolidation needs to qualify for the euro area tionary fiscal policy is quite often applied asymmetrical- (e.g., Belgium), and more generally attempts to reduce ly over the business cycle. In that vein, many countries trends of rising deficits and debts (e.g., the Nether- tend to raise government expenditures in a recession, lands, Switzerland). In some cases, the introduction but fail to lower it sufficiently in good economic times to of the rules coincided with large fiscal adjustments, balance the budget over the business cycle. This asym- in others (e.g., in Finland) it followed an improvement metric response is thus associated with unsustainable in fiscal positions to ensure continued fiscal discipline growth in government expenditures and debt mostly after the crisis. because the increase in expenditures tends to impose permanent consequences on public finances. However, It is common to find countries with more than one fis- there are various options available for countries that cal rule in place today. The trend to multiple fiscal rules seek a more balanced and prudent path for their fiscal is in part the result of the introduction of supranational policy, ranging from the adoption of fiscal responsibili- fiscal rules, but also reflects decisions to broaden nation- ty laws (FRLs) to the adoption of fiscal councils. These al fiscal rules arrangements, in particular in emerging options, when adopted effectively, help countries build economies. Adopting two or more fiscal rules also help reputational capital over time. to define the anchor for fiscal policy, and in that con- text having different rules could help a country address There are several aspects that need to be taken into different policy objectives under particular economic cir- account when a country considers adopting a fiscal cumstances. The most frequently used rules constrain responsibility law as a policy tool. First, FRLs by defi- debt and the budget balance, often in combination. In 40 part, this reflects that supranational rules for members nition sacrifice discretion and replaces it by rules. The implication of this tradeoff is that it makes the careful of monetary unions and the EU include these two types design of fiscal legal frameworks extremely important. of rules, except the ECCU which only has a budget bal- Rules can be rigid and can limit the ability of policy mak- ance rule. Across national fiscal rules, expenditure rules ers to adjust fiscal policy when needed to respond to are often combined with budget balance or debt rules changes in economic circumstances. If there are no pro- to provide a greater anchor for debt sustainability. Rev- visions for such unforeseen circumstances in the FRL, enue rules play a much more limited role, likely because governments may find it difficult to adjust in the face of they are less well suited to ensure the sustainability of an exogenous shock, for example. Second, the effective- public finances. ness of FRLs in supporting fiscal discipline strengthens with their successful implementation for extended pe- There are several reasons that make fiscal rules an riods of time and vice-versa. Third, to work effectively, attractive option for countries that seek to strength- a legal framework for fiscal responsibility requires ade- en their fiscal discipline and manage better the busi- quate PFM systems aligned with the framework’s level ness cycle. A significant weakness of fiscal principles of sophistication (see van Eden, Khemani, and Emery and procedural rules is their lack of specificity about the Jr, 2013). actual fiscal policy stance a country should adopt. Nu- merical fiscal rules, however, provide a clear anchor for Fiscal rules represent another policy tool that has fiscal policy. Ideally, they determine, given the present helped many countries to strengthen fiscal disci- economic and fiscal situation of the country, the levels pline and behave less procyclically. According to re- at which fiscal aggregates should be set. Fiscal rules—in search by the IMF, fiscal rules have spread worldwide the ideal situation—can provide automatic and objec- over the last 20 years or so (Schaechter, Kinda, Budi- tive answers that would otherwise require a great deal na, and Weber, 2013). The more prevalent use of na- of economic analysis, judgment, and political compro- tional fiscal rules reflects responses to different pres- mise. They thus decrease policy transaction costs and sures on public finances and are now more prevalent provide better fiscal outcomes. Taming Volatility It is important to embed flexibility in the design of fiscal discipline in the region. Many ECCU countries either a FRL or a fiscal rule to enable governments have embarked on homegrown fiscal consolidation to respond to unexpected events. As argued in a re- programs to reign in public spending, but the adoption cent World Bank report (World Bank, 2013), pragma- of a rules-based fiscal policy framework remains rare tism calls for the development of well-designed escape in the region, with the exception of Grenada (see Box clauses in FRLs, allowing the fiscal rule framework to be 2.1). In many instances, guided by the IMF or by virtue waived or adjusted during exceptional economic circum- of necessity, ECCU member states have defined fiscal stances. A second solution, combining rules and discre- surplus targets, ceilings on public spending, or limits tion, would be for the FRL to define the fiscal rules that on debt creation, as a way to address concerns with must be adhered to but leave the numerical values to debt sustainability and revenue shortfalls. For exam- be determined by government on a recurring basis (van ple, Antigua and Barbuda targeted a primary surplus Eden, Khemani, and Emery Jr., 2013). of 3 percent of GDP under an IMF Stand By Agreement (SBA) in 2010; Dominica is currently implementing a Countries that seek the path of fiscal discipline will fiscal adjustment equivalent to 6 percent of GDP over inevitably have to develop a sound medium-term fis- the next five years. Other common measures adopted cal framework to operationally either a FRL or fiscal throughout the region include negotiated wage freezes rule. An MTFF requires a more disciplined approach to to control the growth and the size of wage bills, revision fiscal policymaking by formalizing the development of a of tax holidays and tax exemptions to improve tax rev- medium-term fiscal strategy and medium-term orien- enue collections, and enhanced monitoring of the op- tation within the budget process through a combination erations and performance of state-owned enterprises of (1) a medium-term macroeconomic framework that (SOEs). provides multiyear projections of macroeconomic vari- ables such as GDP, inflation, exchange rates, and the The OECS countries can draw on existing good prac- balance of payments; (2) a framework that produces a tices on how to strengthen their fiscal positions and set of multiyear targets or ceilings on fiscal aggregates, be better equipped to adopt counter-cyclical fiscal such as overall government expenditure, borrowing, and policies. Moving towards a full blown FRL, with or with- 41 debt; and (3) identification of concrete policy measures out a formal fiscal rule, requires some preparation and that translate the projected overall resource envelope building technical capacity in order to design, imple- and the government’s fiscal objectives into a set of ment, and monitor the new policy tool. There are some credible and binding multiyear expenditure ceilings and well-established good practices in that regard that policies. could help the Eastern Caribbean states to become less procyclical. First, when establishing primary fiscal In the Eastern Caribbean, there are non-binding su- surplus targets, countries in the region are strongly ad- pranational fiscal guidelines that establish fiscal vised to include in their targets buffers to cover future principles that should be adopted by member coun- disaster-related expenses. Second, the creation of a tries. The fiscal guidelines of the Eastern Caribbean savings fund associated with the fiscal buffers men- Currency Union (ECCU) members aim to put the coun- tioned above could be a good and transparent way to tries on a path consistent with the reduction of their discipline the use of these set-aside resources. This public debt-to-GDP ratios to 60 percent by 2020. ECCU would require the development of a strong institutional members also had in place for some time an overall framework and governance structure with clear oper- deficit target of 3 percent of GDP for which compliance ational rules for the allocation of the fund’s resources. was weak because of various shocks. This overall defi- Third, the adoption of medium-term fiscal frameworks cit target was dropped in 2006 and the level of the pri- (MTFFs) and medium-term debt management strate- mary balance consistent with the debt target in 2020 gies could help strengthen fiscal discipline and ensure has been used to guide fiscal policy, but it has not been that fiscal policy guides the budgetary process and not strictly followed. the other way around. Finally, pursue symmetric fiscal policy responses through the use of fiscal rules aimed The experience of the ECCU countries with fiscal at stabilizing macroeconomic shocks by adopting, for consolidation has been mixed and only more recent- example, expenditure ceilings, cyclical deficit targets, ly there has been a more explicit move to strengthen and rules-based. Taming Volatility Box 2.1 Grenada’s Rules-Based Fiscal Policy Framework Grenada’s fiscal responsibility law set an important step in the transition to a rules-based fiscal framework in the OECS region. In June 2015 Grenada’s Parliament was the first in the OECS to approve a landmark fiscal responsibility law. The new legal framework establishes a 2 percent cap on real expen- diture growth and a debt ceiling of 55 percent of GDP to be achieved by 2022. The framework includes escape clauses in the event of unpredictable shocks such as natural disasters, public health epidemics, and economic and financial crises. The rule-based fiscal framework will be implemented in three phases. Phase 1. Under the Government’s homegrown program, a primary balance of 3.5 percent of GDP should be achieved by 2016, aiming to reduce the debt-to-GDP ratio and restore debt sustainability. Phase 2. During Phase 1, the primary balance is maintained at 3.5 percent of GDP on average until public debt is lowered to 55 percent of GDP. Phase 3. Once the debt target has been achieved, the debt is maintained at this level, while the primary balance is maintained at its debt-stabilizing level, currently estimated around 1 percent of GDP. In addition, the rules-based fiscal framework contains guidelines on public sector wage negotiations, a ceiling on PPP-related contingent liabilities, and the use of CBI receipts. The law also sets a public wage-bill target of 9 percent of GDP, establishes a cap on PPP related contingent liabilities at 5 percent of GDP, and provides for a share of monthly Citizenship by Investment (CBI) inflows to be saved and used to finance future budgetary contingencies or natural-disaster and response efforts. This framework is expected to provide a strong basis for counter-cyclical fiscal policy and enduring medium-term fiscal and debt sustainability. It is more binding than previous fiscal targets used by OECS 42 member countries, which lacked clear operational guidelines and were not fully adhered to. Source: Authors’ elaboration based on IMF Grenada-Second Review under the Extended Credit Facility (June 2015) Concluding Remarks terized by (i) higher volatility of most macroeconomic variables when compared to other more advanced econ- We have confirmed earlier findings in the literature omies as exemplified, for example, by the volatility of the showing that a number of developing countries have key variable affecting welfare – consumption – which is graduated from fiscal policy pro-cyclicality. In com- above the volatility of output in developing countries, parison with industrialized countries, however, develop- but below output volatility in advanced economies; ing countries tend to behave in a way that contributes (ii) countercyclical real interest rates in developing coun- to exacerbate the effects of the business cycle; that tries as opposed to mildly pro-cyclical or a-cyclical real is, they tend to exhibit more often than industrialized interest rates in advanced economies; net exports are economies pro-cyclical fiscal policies in both booms and also found to be much more countercyclical in develop- downturns, contributing to exacerbate output volatility. ing countries relative to the developed economies; and This result coincides with the findings from Kaminsky et (iii) fiscal policy is pro-cyclical, while it tends more fre- al. (2004) and represent additional evidence in support quently to be countercyclical in advanced economies. of the “when-it-rains-it-pours” phenomenon. A result worth mentioning is that we have also found We have also shown that the business cycles in the evidence in support of the idea that institutional qual- OECS countries exhibit the properties that are typical ity is an important determinant of a country’s fiscal of developing countries. In particular, these countries, stance. This is an important finding that suggests that very much like other developing countries, are charac- efforts to graduate from fiscal policy pro-cyclicality Taming Volatility need to be accompanied by policy reforms that seek to would be the optimal solution, fiscal discipline can be strengthen the ability of countries to save in good times buttressed in the OECS by the adoption of intermedi- to generate fiscal buffers that could be used in bad ate but nevertheless good practice policy options. These times. In that regard, initiatives such as the establish- would include (i) building fiscal buffers, that could be em- ment of fiscal councils and the adoption of fiscal rules, bedded in primary fiscal balance targets, to cover future the development of sound debt management strategies disaster-related expenses; (ii) creating savings funds that reinforce fiscal discipline, and the strengthening of with a strong institutional framework and governance macro prudential regulations appear to be necessary structure with clear operational rules for the allocation conditions for graduation from pro-cyclicality. of the fund’s resources; (iii) adopting medium-term fis- cal frameworks (MTFFs) and medium-term debt man- Finally, we have argued that the countries in the re- agement strategies to strengthen fiscal discipline and gion would have much to gain in terms of their ability ensure that fiscal policy guides the budgetary process to respond in a symmetric way to the business cycle and not the other way around; and (iv) pursuing sym- by following well-established good practices. While metric fiscal policy responses through the use of fiscal the adoption of full-blown FRLs and formal fiscal rules rules aimed at stabilizing macroeconomic shocks . 43 Taming Volatility APPENDIX Methodological Approach Used to Investigate the Determinants of Fiscal Policy Cyclicality We begin by estimating equation (1) below (using a country panel fixed effects specification) linking the cyclical component of real government expenditures ( g ci,t ) to the cyclical component of real GDP ( y ci,t ) and a proxy for insti- tutional quality (IQi,t) where we denote a given country by “i” in a given year “t”. The interaction variable is meant to show how institutional quality increases (a positive sign for α3, in the event of α2>0) or decreases (α3<0, in the event of α2>0) fiscal pro-cyclicality. (1) We then expand the analysis by including the role of initial institutional quality (IQ initiali,t ) and changes of the variable (Δ IQi,t) in equation (2). One expects both signs of the parameters of the interaction terms included in the equation, α3 and α4, to be negative and significant in explaining the relationships between cyclical components of real govern- ment expenditures and real GDP (in the event of α2>0). (2) As a next step, we added to this panel equation a vector Z of “q” additional controls to look for possible omitted vari- able bias problems. One expects to observe significant values for the institutional quality variable after controlling for the variables included in Zq vector. (3) 44 Finally, we tried a different approach with equation (4) by switching to a cross section specification where the de- pendent variable (ρiavg) is the correlation coefficient of real government expenditures and real GDP for a given time period. The specification is first computed including the average value of the institutional quality variable (IQiavg) for the same period as sole control, and then adding additional explanatory variables (vector Z). Again, the value for a2 is expected to be negative and statistically significant. (4) In all cases, we used alternative definitions of fiscal cyclicality, namely by applying different filters to smooth the series including the Hodrick-Prescott, Baxter-King, Christiano-Fitzgerald and Butterworth filters. The specifications were applied for the following periods and samples: i) First, for the period 1984-2009, using the Frankel, Vegh, & Vuletin (2014) sample of real government expenditures and real GDP, as well as the set of controls identified by these authors, including the institutional quality proxy, based on the International Country Risk Guide (ICRG) data.10 ii) Next, for the period 1990-2011, using a sample of 180 countries and including the same controls defined by Fran- kel, Vegh, & Vuletin (2014). A problem we found was that of availability of data for the ICRG variable. In particular, the source includes information for 138 countries up to 2010 only. To get around this limitation, we considered an alternative proxy for IQ based on Kunčič (2013), which is available for 193 economies for the period 1990-2011. This is also motivated out of our interest to test whether results found by Frankel, Vegh, & Vuletin (2014) are robust to the choice of IQ variable and hold for a different (larger) country sample.11 10  While the data on real government expenditures and real GDP was kindly provided by Frankel, Vegh, & Vuletin (2013), we compiled the remaining information from the original sources. 11  An additional reason for using the Kunčič (2013) dataset is that it includes data on the IQ proxy for 14 out of 15 Caribbean countries that are of special interest in this research, whereas ICRG data only covers 6 of those countries. Taming Volatility iii) Third, we generated results for additional specifications, not included in Frankel, Vegh, & Vuletin (2014), and in- troduced additional controls. Tables A3.1 and A3.2 show the results of econometric estimations that confirm the relevance of institutional quality as an important determinant of fiscal policy cyclicality. The full set of results can be found in Carneiro and Garrido (2015) and will not be repeated here. Our results show that IQ is strongly significant as an explanatory variable of movements away from cyclicality for all specifications using ICRG as a proxy for IQ and for all but one specification that rely on the Kunčič (2013) dataset, following Equation 1. When we separate the effects of initial IQ from changes in IQ (Equation 2) we observe that the former variable is strongly significant under all filtering methods when using ICRG as proxy for IQ, but less so in the specifications using Kunčič (2013) data. On the other hand, the latter spec- ifications show higher statistical significance for the changes in IQ compared to the former. When the same panel specification was expanded to account for the role of additional controls (Equation 3), one can observe how robust the significance of IQ is when ICRG is used as a proxy under all filtering methods. The cross-country regression results for the period 1990-2011 through the estimation of equation (4) considers the correlation between cyclical components of real government expenditures and real GDP as a function of average IQ and average values for other controls, as in Frankel, Vegh, & Vuletin (2013). This specification also shows robust results for institutional quality under all filtering methods and for alternative IQ definitions. 45 Taming Volatility Table A2.1: Panel Regression: Determinants of Cyclical Component of Real Gov. Expenditures (alternative filters). Additional Controls. Period 1984-2009 Cyclical Component Real Gov. Exp. VARIABLES 1 2 3 4 HP filter BK filter CF filter BU filter Cyclical Component Real GDP, alternative filters (x) 2.899*** 3.061*** 1.940*** 3.218*** (8.337) (7.631) (7.280) (7.273) Interaction Cyclical Component Real GDP -3.021*** -3.042*** -1.319*** -3.222*** filter and Institutional Quality (x) (-5.042) (-4.406) (-3.218) (-4.064) Interaction Cyclical component of Real GDP 0.086 0.099 0.043 0.100 and Chinn-Ito Index of Capital Openness (x) (1.440) (1.407) (1.106) (1.207) Interaction Cyclical component of Real GDP -0.003 -0.005 -0.003* -0.004 and financial depth (M2/GDP) (x) (-1.230) (-1.505) (-1.888) (-1.040) Interaction Cyclical component of Real GDP and -5.571 -5.200 17.303 -8.466 Volatility (Squared cyclical component GDP) (x) (-0.319) (-0.208) (0.848) (-0.294) Interaction Cyclical component of Real GDP 0.072 0.107* 0.047 0.146** and Proxy Political Checks and Balances (x) (1.437) (1.791) (1.380) (2.086) Interaction Cyclical component of Real GDP -0.005*** -0.007*** -0.005*** -0.009*** and Debt Ratio to GDP (x) (-3.855) (-4.443) (-5.247) (-5.527) Interaction Cyclical component of Real GDP -0.079*** -0.065*** -0.026** -0.060*** and Reserves Ratio to Imports (x) (-5.777) (-3.833) (-2.355) (-3.035) Institutional Quality Proxy 0.000 0.001 0.052* 0.014 (0.001) (0.049) (1.953) (0.823) Chinn-Ito Index of Capital Openness 0.001 0.001 0.001 0.001 (0.590) (0.466) (0.479) (0.785) 46 Financial Depth (M2 Ratio to GDP) 0.000 0.000* 0.000* 0.000 (1.513) (1.750) (1.904) (1.444) GDP volatility (Squared Cyclical Component -0.489 0.278 0.443 0.201 of GDP) (-0.354) (0.193) (0.322) (0.183) Proxy for Political Checks and Balances -0.001 -0.001 -0.003 -0.000 (-0.332) (-0.513) (-1.382) (-0.262) Debt to GDP ratio 0.000 0.000 0.000*** 0.000*** (1.136) (1.246) (5.493) (4.178) Reserves to Imports Ratio -0.000 -0.000 0.001 0.000 (-0.604) (-0.849) (1.046) (0.008) Constant -0.010 -0.013 -0.060*** -0.023** (-0.707) (-0.911) (-3.534) (-2.168) Observations 1,952 1,768 1,768 1,768 R-squared 0.085 0.087 0.082 0.077 Number of ccode1 91 90 90 90 r2_a 0.0328 0.0302 0.0243 0.0195 F 11.41 10.60 9.863 9.272 Notes: t-statistics in parentheses. t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1 *** p<0.01, ** p<0.05, * p<0.1. (x) The filter used to com- pute the cyclical component of GDP is the same as that used for computing the dependent variable. Source: Authors calculations. Taming Volatility Table A2.2: Cross country Regression. Fiscal Cyclicality Proxy under alternative filtering methods. 1984-2009. Role of Instit. Quality and Determinants Correl. Cyclical Component Real Gov. Exp. And Real GDP 1 2 3 3 4 5 6 7 VARIABLES HP filter BK filter CF filter BU filter HP filter BK filter CF filter BU filter Institutional Quality (from ICRG), -0.946*** -0.763*** -0.812*** -0.645*** -1.118*** -0.883*** -1.187*** -0.778*** average 1984-2009 (-6.420) (-5.263) (-5.267) (-4.318) (-4.460) (-3.504) (-4.515) (-3.165) Chinn-Ito Index of Capital Openness, 0.001 -0.001 -0.006 -0.038 average 1984-2009 (0.023) (-0.021) (-0.217) (-1.388) Financial Depth (M2 Ratio to GDP), 0.001 0.000 0.000 0.001 average 1984-2009 (0.566) (0.367) (0.408) (0.833) GDP volatility (Squared Cyclical Component 95.657** 118.368* 3.916* 345.230** of GDP), average 1984-2009 (x) (2.109) (1.925) (1.971) (2.294) Proxy for Political Checks and Balances, 0.029 0.027 0.038 0.061** average 1984-2009 (1.213) (1.129) (1.579) (2.606) Debt to GDP ratio, average 1984-2009 -0.001 -0.001 -0.002** -0.002** (-1.386) (-1.577) (-2.292) (-2.362) Reserves to Imports Ratio, -0.020*** -0.013* -0.017** -0.014* average 1984-2009 (-2.706) (-1.671) (-2.195) (-1.851) Constant 0.648*** 0.567*** 0.587*** 0.476*** 0.742*** 0.627*** 0.844*** 0.459*** (7.502) (6.652) (6.492) (5.433) (4.921) (4.133) (5.852) (3.088) Observations 92 91 92 92 91 90 91 91 R-squared 0.314 0.237 0.236 0.172 0.413 0.313 0.364 0.334 r2_a 0.306 0.229 0.227 0.162 0.364 0.254 0.311 0.278 F 41.22 27.69 27.75 18.65 8.358 5.325 6.793 5.945 47 Notes: t-statistics in parentheses. t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1 *** p<0.01, ** p<0.05, * p<0.1. (x) The filter used to compute the cyclical component of GDP is the same as that used for computing the dependent variable. Source: Authors calculations Taming Volatility Table A2.3: Panel Regression: Determinants of Cyclical Component of Real Gov. Expenditures (alternative filters and two alterntive proxies for Instit. Quality). Additional Controls. Period 1990-2013. Institutional Quality proxy is that of Kuncic Institutional Quality proxy is ICRG (2013) Cyclical Component Real Gov. Exp. 1 2 3 3 4 5 6 7 HP filter BK filter CF filter BU filter HP filter BK filter CF filter BU filter Cyclical Component Real GDP, Alternative 1.025*** 1.086*** 0.825*** -0.191 2.150*** 2.653*** 1.267*** 2.130*** filters (x) (3.853) (3.193) (5.193) (-0.614) (8.195) (7.004) (8.396) (6.005) Interaction Cyclical Component Real GDP filter -0.564 -0.143 -0.031 1.812** -2.296*** -3.092*** -0.792*** -2.274*** and Institutional Quality, Alternative filters (x) (-0.867) (-0.184) (-0.103) (2.179) (-4.404) (-4.389) (-3.628) (-3.330) Interaction Cyclical component of Real GDP -0.042 -0.070 -0.014 -0.058 -0.007 0.002 -0.035** 0.009 and Chinn-Ito Index of Capital Openness, (-0.794) (-1.104) (-0.822) (-0.892) (-0.155) (0.035) (-2.142) (0.145) Alternative filters (x) Interaction Cyclical component of Real GDP -0.007*** -0.007** -0.002*** -0.010*** -0.008*** -0.003 -0.002*** -0.008*** and financial depth (M2/GDP), Alternative (-2.884) (-2.547) (-2.671) (-3.668) (-3.924) (-1.216) (-2.612) (-2.953) filters (x) Interaction Cyclical component of Real GDP -3.047 -9.535* -1.397** 4.028 1.705 8.989 -2.020*** -2.993 and Volatility (Squared cyclical component (-1.562) (-1.945) (-2.124) (0.921) (0.569) (0.768) (-2.770) (-0.697) GDP), Alternative filters (x) Interaction Cyclical component of Real GDP 0.007 -0.019 -0.035** -0.015 0.013 -0.001 -0.012 0.019 and Proxy Political Checks and Balances, (0.145) (-0.337) (-2.252) (-0.241) (0.301) (-0.016) (-0.747) (0.345) Alternative filters (x) Interaction Cyclical component of Real GDP 0.002** 0.002** 0.000 0.004*** -0.001 -0.002** -0.000 -0.001 and Debt Ratio to GDP, Alternative filters (x) (2.380) (2.021) (0.251) (3.412) (-0.913) (-1.976) (-0.612) (-1.018) Interaction Cyclical component of Real GDP 0.004 -0.016 0.006 0.006 0.005 -0.016 0.008*** 0.003 48 and Reserves Ratio to Imports, Alternative (0.646) (-1.578) (1.627) (1.027) (0.847) (-1.473) (3.290) (0.493) filters (x) Institutional Quality Proxy 0.010 0.009 0.016 0.014 0.013 0.039 -0.069** 0.011 (0.244) (0.198) (0.345) (0.460) (0.476) (1.273) (-2.037) (0.532) Chinn-Ito Index of Capital Openness 0.000 0.002 0.001 0.001 -0.001 0.001 -0.002 0.000 (0.064) (0.801) (0.275) (0.338) (-0.320) (0.241) (-0.677) (0.030) Financial Depth (M2 Ratio to GDP) 0.000*** 0.000*** 0.000 0.000* 0.000** 0.000** 0.000 0.000 (2.856) (2.584) (1.146) (1.708) (2.536) (2.345) (0.562) (1.348) GDP volatility (Squared Cyclical Component -0.804 -0.684 0.228 -0.287 1.137 0.674 0.513 -0.129 of GDP), Alternative filters (x) (-1.485) (-0.732) (0.505) (-0.668) (1.499) (0.592) (1.572) (-0.404) Proxy for Political Checks and Balances -0.002 -0.002 -0.006*** -0.001 -0.001 -0.001 -0.003 -0.001 (-1.067) (-1.107) (-2.627) (-1.028) (-0.457) (-0.815) (-1.060) (-0.649) Debt to GDP ratio 0.000* 0.000 0.000* 0.000** 0.000*** 0.000 0.000 0.000* (1.931) (1.067) (1.948) (2.040) (2.864) (0.381) (1.635) (1.719) Reserves to Imports Ratio -0.000 -0.000 0.000 -0.000 -0.001 -0.000 0.000 -0.000 (-0.528) (-0.185) (0.741) (-0.418) (-1.249) (-0.295) (0.900) (-0.537) Constant -0.019 -0.020 -0.017 -0.015 -0.022 -0.035* 0.034 -0.011 (-0.925) (-0.856) (-0.675) (-0.929) (-1.321) (-1.814) (1.564) (-0.893) Observations 2,538 2,318 2,538 2,538 2,264 1,978 2,264 2,264 R-squared 0.069 0.056 0.055 0.029 0.122 0.083 0.082 0.059 Number of ccode1 155 153 155 155 125 124 125 125 r2_a 0.00256 -0.0168 -0.0123 -0.0399 0.0643 0.0143 0.0224 -0.00244 F 11.70 8.577 9.213 4.782 19.63 11.11 12.72 8.900 Notes: t-statistics in parentheses. t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1 *** p<0.01, ** p<0.05, * p<0.1. (x) The filter used to com- pute the cyclical component of GDP is the same as that used for computing the dependent variable. Source: Authors calculations Taming Volatility Table A2.4: 2-period Panel Country Regression. Changes Fiscal Cyclicality Proxy under alternative filtering methods. Role of Instit. Quality and Determinants. Instit. Quality proxy is that of Kuncic (2013) (*) Correl. Cyclical Component Real Gov. Exp. And Real GDP 1 2 3 3 4 5 6 7 HP filter BK filter CF filter BU filter HP filter BK filter CF filter BU filter Initial Correlation Real Gov. Exp-GDP, alterna- -0.937*** -0.837*** -0.978*** -1.028*** -0.968*** -0.848*** -1.010*** -1.016*** tive filters (Avg. 1990-2000) (x) (-10.292) (-9.897) (-11.085) (-11.313) (-10.561) (-9.784) (-11.300) (-10.761) Initial Institutional Quality (Avg. 1990-2000) -0.491** -0.772*** -0.482** -0.464** -0.440 -0.681* -0.340 -0.454 based on Kuncic (2013) (-2.158) (-3.781) (-2.122) (-2.022) (-1.010) (-1.742) (-0.783) (-0.987) Initial Chinn-Ito Index of Capital Openness 0.001 0.006 -0.008 0.027 (Avg. 1990-2000) (0.025) (0.188) (-0.204) (0.659) Initial M2 Ratio to GDP (Avg. 1990-2000) -0.002 -0.002 -0.003* -0.002 (-1.494) (-1.186) (-1.717) (-1.375) Initial GDP volatility (Squared Cyclical Compo- 6.178 7.928 7.507 11.952 nent of GDP), alternative filters (Avg. 1990- (0.316) (0.597) (0.396) (0.713) 2000) (x) Initial Checks and Balances (Avg. 1990-2000) -0.013 0.000 -0.017 0.001 (-0.376) (0.006) (-0.476) (0.018) Initial Debt to GDP ratio (Avg. 1990-2000) 0.000 0.001 -0.000 0.000 (0.026) (0.455) (-0.162) (0.045) Initial Reserves to Imports Ratio (Avg. 1990- -0.014 -0.012 -0.016 -0.015 2000) (-0.901) (-0.864) (-1.046) (-0.935) Change in Institutional Quality (2000-2010 vs 0.224 -0.247 0.009 -0.285 0.613 0.386 0.295 0.048 1990-2000) based on Kuncic (2013) (0.323) (-0.403) (0.013) (-0.397) (0.825) (0.584) (0.396) (0.061) Change in Chinn-Ito Index of Capital Openness -0.003 -0.006 0.002 0.073 (2000-2010 vs 1990-2000) (-0.058) (-0.134) (0.053) (1.459) 49 Change in M2 Ratio to GDP (2000-2010 vs -0.003 -0.004* -0.002 -0.003 1990-2000) (-1.160) (-1.869) (-1.077) (-1.367) Change in GDP Volatility, alternative filters 8.767 30.415** 4.395 11.544 (2000-2010 vs 1990-2000) (x) (0.575) (2.378) (0.304) (0.501) Change in Checks and Balances (2000-2010 vs -0.062 -0.045 -0.058 -0.075 1990-2000) (-1.265) (-1.018) (-1.167) (-1.441) Change in Debt to GDP Ratio (2000-2010 vs 0.000 0.000 0.000 -0.000 1990-2000) (0.148) (0.160) (0.251) (-0.040) Change in Reserves to Imports Ratio (2000- -0.023** -0.016** -0.019** -0.013 2010 vs 1990-2000) (-2.473) (-2.023) (-2.188) (-1.310) Constant 0.212 0.368*** 0.217 0.184 0.476** 0.475** 0.485** 0.368 (1.571) (3.036) (1.613) (1.367) (2.019) (2.237) (2.046) (1.490) Observations 131 131 131 131 128 128 128 128 R-squared 0.456 0.437 0.493 0.503 0.531 0.507 0.564 0.563 r2_a 0.443 0.424 0.481 0.491 0.469 0.440 0.506 0.504 F 35.43 32.85 41.16 42.81 8.471 7.665 9.658 9.615 Notes: t-statistics in parentheses. t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1 *** p<0.01, ** p<0.05, * p<0.1. (x) The filter used to com- pute the cyclical component of GDP is the same as that used for computing the dependent variable. Source: Authors calculations Taming Volatility REFERENCES nal of Political Economy, Vol. 119, No. 1 (February 2011), 78-121. De Jong, R. M., & Sakarya, N. (2013). The Econometrics Acemoglu, D., Johnson, S., & Robinson, J. (2001). The of the Hodrick-Prescott filter. Columbus, OH, USA: Colonial Origins of Comparative Development: An Department of Economics, Ohio State University. Empirical Investigation . The American Economic Re- Drukker, D. M. (2011). New Methods for Filtering Times view, Vol. 91, No. 5 (Dec., 2001), , pp. 1369-1401. Series in STATA 12. Stata Conference. Chicago: STA- Acevedo, S., Cebotari, A., & Turner-Jones, T. (2013). Ca- TA. ribbean Small States: Challenges of High Debt and Frankel, J. A., Vegh, C. A., & Vuletin, G. (2013). On grad- Low Growth. IMF Policy Papers, February 2013. uation from fiscal procyclicality. Journal of Develop- Alesina, A., & Tabellini, G. (2005). Why is Fiscal Policy ment Economics. Often Procyclical? National Bureau of Economic Re- Gavin, M., & Perotti, R. (1997). Fiscal Policy in Latin search. America. National Bureau of Economic Research. Amo-Yartey, C., Narita, M., Nicholls, G. P., Okwuokei, J. Gonzalez-Garcia, J., Lemus, A., & Mrkaic, M. (2013). Fis- C., Peter, A., & Turner-Jones, T. (2012). The Challeng- cal Multipliers in the ECCU. (W. H. IMF Working Paper, es of Fiscal Consolidation and Debt Reduction in the Ed.) Washington D.C.: International Monetary Fund. Caribbean. IMF Working Paper WP/12/276. Harvey, A. C. (1989). Forecasting Structural Time Series Andrews, D. (1993). Tests for Parameter Instability and Models and the Kalman Filter. Cambridge University Structural Change with Uncertain Change Point. Press. Econometrica July, 61:4, 821-56. Ilzetzki, E., & Vegh, C. A. (2008). Procyclical Fiscal Pol- Andrews, D., & Plogeber, W. (1994). Optimal Test when icy in Developing Countries: Truth or Fiction? NBER a Nuisance Parameter is Present Only under the Al- Working Paper Series, Working Paper 14191. ternative. Econometrica November 62:6, 1383-414. IMF. (2013). Antigua and Barbuda: Staff Report for the Asonuma, T., Peraza, G., Vitola, K., & Tsuda, T. (2014). 2012 Article IV consultation. IMF Country Report No. Sovereign Debt Restructurings in Belize: Achieve- 13/76. 50 ments and Challenges Ahead . IMF Working Paper IMF. (2013). Dominica: Article IV Consultation. IMF Coun- WP/14/132 . try Report No. 13/31. Barro, R. (1979). On the Determination of Public Debt. IMF. (2014). 2014 St. Kitts and Nevis: Article IV Consul- Journal of Political Economy 87, 940-971. tation. IMF Country Report No. 14/86. Baum, C. (2006, July). Time Series Filtering Techniques IMF. (2014). Annual Report on Exchange Arrangements in Stata. Retrieved from http://www.stata.com/ and Exchange Restrictions. meeting/5nasug/TSFiltering_beamer.pdf IMF. (2014). Belize: 2014 Article IV Consutation, Staff Baxter, M., & King, R. (1993, June). Fiscal Policy in Gen- Report. IMF Country Report No. 14/280. eral Equilibrium. The American Economy Review. Vol IMF. (2014). Chile: Article IV Consultation - Staff Report. 83. #3. , 315-334. IMF Country Report No. 14/218. Baxter, M., & King, R. G. (1995). Measuring Business IMF. (2014). Guyana: 2013 Article IV Consultation - Staff Cycles Approximate Band Pass Filters for Economic Report. IMF Country Report No. 14/294. Times Series. NBER Working Paper Series, Working IMF. (2014). St. Vincent and the Grenadines: Request for Paper 5022. Disbursement Under the Rapid Credit Facility and Butterworth, S. (1930). On the theory of filter amplifi- Purchase Under the Rapid Financing Instrument - ers. Experimental Wireless & the Wireless Engineer, Staff Report. IMF Country Report No. 14/360. 7, 536-451. IMF. (2015). Grenada: 2014 Article IV Consultation and Calderon, C., & Schmidt-Hebbel, K. (2008). The Choice of Request for an Extended Credit Facility Arrange- Fiscal Regimes in the World. Working Papers Central ment. IMF Country Report No. 14/196. Bank of Chile 487. Juglar, C. (1862). Des Crises commerciales et leur retour Christiano, L. J., & Fitzgerald, T. J. (1999, July). The Band periodique en France, en Angleterre, et aux Etats- Pass Filter. NBER Working Paper Series, Working Pa- Unis. per 7257. Kaminsky, G. L., Reinhart, C. M., & Végh, C. A. (2004). Christiano, L., Eichenbaum, M., & Rebelo, S. (2011). When When It Rains, It Pours: Procyclical Capital Flows Is the Government Spending Multiplier Large? Jour- and Macroeconomic Policies. NBER Macroeconomics Taming Volatility Annual 2004, Volume 19. Ravn, M. O., & Uhlig, H. (2010). On Adjusting The Ho- Kitchin, J. (1923). Cycles and Trends in Economic Fac- drick-Prescott Filter for the Frequency of Obser- tors. Review of Economics and Statistics, 5(1), 10-16. vations. London, UK: London Business School and Korotayev, A., &  Tsirel, S. (2010). A Spectral Analysis of Centre for Economic Policy Research, and Humboldt World GDP Dynamics: Kondratieff Waves, Kuznets University and Centre for Economic Policy Research. Swings, Juglar and Kitchin Cycles in Global Economic Rigobon, R. (2004). Comments on When It Rains It Development, and the 2008–2009 Economic Crisis. Pours: Pro-cyclical Capital Flows and Macroeconom- Structure and Dynamics: eJournal of Anthropologi- ic Policies. In NBER Macroeconomics Annual. Mark cal and Related Sciences. Gertler and Kenneth Rogoff, editors. MIT Press. Kunčič, A. (2013). Institutional Quality Dataset. Sahay, R. (2006). Stabilization, Debt, and Fiscal Pol- Kuznets, S. (1930). Secular Movements in Production icy in the Caribbean. IMF Working Papers Series and Prices. Their Nature and their Bearing upon Cy- WP/05/26. clical Fluctuations. Schaechter, A., T. Kinda, N. Budina, and A. Weber (2012). Lane, P. R., & Tornell, A. (1999). Why Aren’t Savings Fiscal Rules in Response to the Crisis - Toward the Rates in Latin America Procyclical? National Bureau “Next Generation” Rules. A New Dataset. IMF Work- of Economic Research. ing Paper Series WP/12/187. Maravall, A., & del_Río, A. (2001). Time Aggregation and Sterman, J. (2000). Business Dynamics: Systems Think- the Hodrick Prescott Filter. Banco de Espana. Docu- ing and Modeling for a Complex World . Irwin Mc- mento de Trabajo No 0.108. Graw-Hill. Mollick, A. V., Cabral, R., & Carneiro, F. G. (2011). Does in- Stock, J., & Watson, M. (2014). Introduction to Econo- flation targeting matter for output growth? Evidence metrics. Prentice Hall (3rd Edition). from industrial and emerging economies. World Talvi, E., & Vegh, C. A. (2000). Tax Base Variability and Bank Policy Research Working Papers 4791. Procyclical Fiscal Policy. NBER Working Paper 7499. Mpatswe, G. K., Tapsoba, S. J.-A., & York, R. C. (2011). van Eden, H., P. Khemani, and R. Emery Jr. (2013). Devel- The Cyclicality of Fiscal Policies in the CEMAC Re- oping Legal Frameworks to Promote Fiscal Respon- gion. IMF Working Papers WP/11/205. sibility: Design Matters. Chapter 2 in Cangiano, M., T. 51 Nakata, T. (2013). Optimal Fiscal and Monetary Policy Curristine, and M. Lazare (editors). Public Financial with Occasionally Binding Zero Bound Constraints. Management and its Emerging Architecture. Wash- Finance and Economics Discussion Series Divisions ington, D.C.:International Monetary Fund. of Research & Statistics and Monetary Affairs, Fed- Villafuerte, M., Lopez-Murphy, P., & Ossowski, R. (2010). eral Reserve Board. Riding the Roller Coaster: Fiscal Policies of Nonre- Park, J. (2012). Determinants and consequences of fiscal newable Resource Exporters in Latin America and procyclicality and sustainability. the Caribbean. IMF Working Papers WP/10/251. Pollock, D. (1999). A Handbook of Time-Series Analysis, World Bank (2013). Oil Rules: Kazakhstan’s Policy Op- Signal Processing and Dynamics. London: Academic tions in a Downturn. Report No.: 82509-KZ. Wash- Press. ington, D.C.: The World Bank Group. Quandt, R. (1960). Test of the Hypothesis that a Linear Re- World Bank. (2014). World Development Report 2014: gression System Obeys Two Separate Regimes . Jour- Risk and Opportunity. World Bank WDR. nal of American Statistic Association 55, , 324-30. Taming Volatility Chapter 3. Comprehensive Financial Development for Growth and Lower Volatilty in the OECS Introduction concludes by exploring specific policy options that could help reorient financial development so that it could be C omprehensive financial development can be supportive of; (i) lower volatility; (ii) enhanced economic an important driver of economic growth and growth, and (iii) a more effective fiscal policy. can increase a country’s resilience to exter- nal shocks and volatility. Fully financially developed The OECS Financial Sector: economies find it easier to mobilize savings, share in- Large and Frail formation, improve resource allocation, and implement The financial sector in the OECS is dominated by the 53 more effective diversification and risk management strategies. In the OECS context of a currency board that banking sector and is large as a percentage of GDP. provides for a monetary anchor, limited fiscal space, and The region has a high level of credit and is overbanked: GDP that is highly dependent on services, the role of the in 2015, banking sector assets represented 166 percent financial sector as a driver of growth is further enforced. of the region’s overall GDP. There are some 40 licensed Comprehensive financial development also leads to less commercial banks (among them, 13 are indigenous volatility to the extent that deep and liquid financial sys- banks and 26 are branches or subsidiaries of foreign tems with more diverse instruments can help alleviate banks, primarily Canadian-owned). Total commercial the impact of shocks. Comprehensive financial devel- banks’ assets for the OECS amounted to US$10.4 billion opment also helps countries to manage better the im- as of December 2015 while total bank deposits stood at pact of terms of trade volatility, especially in the case of US$6.2 billion for the same period. Further, the number small, open economies such as in the case of the OECS. of commercial bank branches per 100,000 people is 27 in the OECS compared with a LAC average of just 20. This chapter explores critical policy options to The commercial banks are regulated by the Eastern strengthen financial development in the region. In Caribbean Central Bank (ECCB), and indigenous banks what follows, we first discuss the structure and evo- make up about half of the banking system as measured lution of the financial sector, in particular the banking by assets, deposits, as well as loans. sector in the OECS region over the past 15 years. Sec- ond, we assess the level of financial development in the The non-banking sector is gaining importance in the region as well as the global relationship between com- OECS with credit union membership on the rise. As of prehensive financial development, growth, and stability. 2015, there were 51 active credit unions, which manage We do this by using a comprehensive index of financial total assets of about US$880 million (13 percent of the development recently created by the IMF that captures region’s GDP) and have a membership base of 304,699, different aspects of financial development. The chapter which is just under 50 percent of the region’s popula- tion. Credit union membership in the OECS is high as Taming Volatility compared with other countries, and its membership The OECS has a regional capital market that remains increased steadily during 2001-10 despite a decline in underdeveloped. The Regional Government Securities the number of credit unions. Credit unions have come Market (RGSM) was established in 2002 for the prima- into prominence in light of stringent credit conditions in ry issuance of government debt securities for 8 ECCU the banking sector following the global financial crisis. member countries to create a single regional financial Its assets and deposits experienced steady growth, es- space for government debt. Since the establishment of pecially for 2005-10, a period over which its total asset the RGSM, the holding of sovereign debt across borders size almost doubled. In particular in 2015, credit union has also increased. The increased exposure of banks to assets account for as much as 44 percent GDP in Dom- sovereign debt increases their liquidity and solvency inica. For Grenada, St. Vincent and the Grenadines, and risks of other governments than their own. Only five St. Lucia, credit union assets vary between 14 to 16 per- of the eight member countries – Antigua and Barbuda, cent of GDP. Grenada, St. Lucia, St. Kitts and Nevis, and St. Vincent and the Grenadines – have thus far issued securities in The insurance sector in the OECS is relatively large the regional market, with St. Lucia and St. Vincent and and is characterized by regional conglomerates. The the Grenadines being the most active issuers. However, two major insurance companies are the Sagicor Finan- the ECCB has not succeeded in creating a proper sov- cial Group and Guardian Holdings. Assets for the sector ereign debt market, as there is no secondary markets, constitute approximately 17 percent of regional GDP. and as a result financial institutions buy and hold debt The January 2009 collapse of Trinidad and Tobago to maturity. Secondary market activity in government based CL Financial Ltd. (CLF) and related companies securities has also been precluded under the RGSM on (such as CLF’s insurance subsidiaries, the Colonial Life account of a broker/dealer system that is not condu- Insurance Company (CLICO), CLICO International Life cive to secondary market trading. There is also an In- (CLI), and the British Insurance Company (BAICO)) af- terbank Market and Repo Market administered by the fected the Caribbean, but hit the OECS region the hard- ECCB and Eastern Caribbean Home Mortgage Bank est because of its high exposure to CLICO and BAICO (ECHMB) which also comprise of the long term debt 54 that was estimated at 15 percent of GDP in 2009. capital market. Offshore financial sectors in the region have been de- Before the 2008 global financial crisis, credit in the veloped as a means to increase fiscal revenues, while OECS expanded rapidly that was not supported by not taxing their population. The OECS member states underlying economic growth. The banking sector in have a relatively small share of worldwide offshore ac- the OECS expanded rapidly in the early 2000s due to tivity, while they offer financial services ranging from a credit boom during the run-up to the 2007 Cricket international banking for corporations and individuals, World Cup, and rapid credit expansion to the public sec- to FDI, to insurance. Within the OECS, Antigua and Bar- tor to help finance governments’ public sector invest- buda and St. Kitts and Nevis have the most active off- ment programs. Between 2003 and 2008, credit to the shore financial sectors. However the offshore sector has private sector across the OECS grew at an average an- been under threat due to: (i) the OECD Global Forum to nual rate of 13.7 percent, as show in Figure 3.1 and Figure fight tax evasion; (ii) the Anti Money laundering (AML) 3.2. In St. Lucia private credit to the private sector grew regulation of the Financial Action Task Force (FATF); and at 19 percent a year. Over 50 percent of this credit was (iii) more recently U.S. Foreign Account Tax Compliance for the acquisition of personal property and the growth Act (FATCA).12 of lending for services also grew sharply. The high pre- vailing level of credit to the private sector to GDP (which reached 85 percent in 2013) combined with GDP growth 12  The original FATF Forty Recommendations were drawn up in 1990 aimed at combating the misuse of financial systems through of only 1.4 percent annually between 2003 and 2013, money laundering. After several rounds, the revised FATF AML reg- was not sufficient to support this level of credit growth. ulation addresses new and emerging threats. The FATF Standards have also been revised to strengthen the requirements for higher This then led to a deterioration in asset quality. risk situations. The Foreign Account Tax Compliance Act (FATCA) is a United States federal law that requires United States per- At the onset of the global financial crisis, economic sons, including individuals who live outside the United States, to have yearly reported themselves and their financial accounts held activity contracted sharply leading to a sudden in- outside of the United States to the Financial Crimes Enforcement crease in non-performing loans. After a significant Network. Taming Volatility Fi ur 3.1 Priv t S ctor Cr dit Fi ur 3.2 Priv t S ctor Cr dit Growth 15 20% 15% Private Credit (Bn. XCD) 10 10% 5% 5 0% -5% 0 -10% 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 Personal Non-personal Ag. & shing Industry 1/ Services 2/ Personal Total loans Source: ECCB. Source: ECCB. 1/ Includes manufacturing and construction. 2/Includes distributive trades, tourism, entertainment, transport, public utilities, professional services, and financial institutions. Fi ur 3.3 NPLs & Provisionin Fi ur 3.4 R turn on Av r Ass ts 20% 60% 3% 15% 45% 2% 10% 30% 1% 55 5% 15% 0% 0% 0% -1% 03 04 05 06 07 08 09 10 11 12 13 14 15* 03 04 05 06 07 08 09 10 11 12 13 14 15* NPLs Provisioning Rate (rhs) Source: ECCB. Data from 2015 is from September 2015. Source: ECCB. Data from 2015 is from September 2015. credit expansion to the private sector during 2000-07, sector. Commercial banks’ NPLs rose from 7.7 percent regional economic activity experienced a sharp con- in 2008 to 18.0 percent in 2014 (see 3.3). The over- traction as a result of the 2008 global financial crisis. extension of credit proved to be unsustainable, and a This has put significant stress on the banks’ balance credit overhang has been created. Protracted foreclo- sheets and exposed vulnerabilities in the banking sys- sure processes which in some countries require judicial tem. Credit growth to the private sector continued to reviews have meant that the NPLs have remained on fall and since 2013 is now negative (Figure 3.3). Most of bank’s balance sheets for the past six to eight years. the credit tightening has occurred in services (such as Although provisioning levels have increased for these tourism-related services) although the highest share of non-performing loans, these are still below internation- loans are to individuals (54 percent) for acquisition of al standards for provisioning. As show in Figure 3.4, the property and home construction and renovation, which deteriorating asset quality has weighed down signifi- has been rising in recent years. cantly on profitability. As a result, there has been a deterioration in asset The new Banking Act has now been passed in each quality evidenced by a sharp increase in the share of of the eight OECS member states, and this provides nonperforming loans (NPLs) in the regional banking the framework for improved supervision and reso- Taming Volatility lution, if needed. Important efforts have been made We use a comprehensive index of financial develop- to improve the regulatory and supervisory framework. ment that captures the multidimensional compo- The new Banking Act, which is harmonized across the nents of financial sector development. Traditionally, ECCU, has strengthened the regulatory and supervisory the literature has used the ratio of private credit to frameworks, including by introducing higher minimum GDP as a proxy to measure the level of financial de- capital requirements, a more effective resolution of velopment. However, this indicator does not fully cap- failed banks, and a stronger depositor protection. Look- ture the full set of financial sector attributes, such as ing ahead, the currency union will need to operationalize development of non-bank financial sector which con- the new Banking Act towards enhanced supervision and sists of pension funds, credit unions, insurance com- eventually strive to implement risk-based supervision panies, mutual funds, or securities markets. Over the and Basel II regulation for banks and non-banking finan- past decade, however, the non-banking financial sec- cial institutions. tor has significantly contributed to the overall finan- cial sector development in the Latin America and the Against this backdrop and the efficiency of the sector, Caribbean. This is also true for the OECS. The IMF’s there is need to build a stronger banking and financial newly constructed comprehensive index of financial system that will be able to contribute towards growth development (Sahay et al. (2015)) – the Financial De- and reducing volatility. In addition to the high number velopment (FD) Index - aims to capture different as- of branches per capita, commercial bank costs struc- pects of financial development rather than relying on tures in the OECS are expensive. The average 2014 cost- a single proxy, such as, credit to GDP. The index con- to-income ratio at the indigenous banks in the OECS is sists of two major components: financial institutions 79 percent, which is well above the emerging market and financial markets. Each component is broken average of 45 percent. This combined with the level of down into three sub-components: depth, access, and NPLs suggests that efforts towards strengthening and efficiency (see Figure 3.5).13 consolidating the sector will significantly aide its ability to contribute towards growth and development. Although the OECS scores well in the credit-to-GDP 56 ratio, its index of comprehensive financial develop- Financial Development in the OECS: A ment is low. Figure 3.6 shows the credit to GDP ra- Partial and Incomplete System tio for the OECS as well as key comparator countries and country groupings.14 Compared to the LAC and The ratio of credit to GDP and credit growth have re- Emerging Market (EM) average, which are each 47 and cently been questioned as their comprehensiveness 49 percent, the OECS is well above these two bench- of measuring financial development, as they may not marks. The OECS indicator is above both the level in capture the ability of financial sectors to contribute Slovenia and Sri Lanka. However, as Figure 3.7 shows, towards growth and reducing volatility. The previous the level of its comprehensive financial development as section showed that commercial bank assets and credit measured by the index construction referenced above, in the OECS expanded rapidly in the 2000s. However, is lower than these country groupings and peer coun- high bank credit to the private sector may be a mis- tries. The index is scaled from 0 to 1, and the higher leading indicator to measure financial development. For the index the better the level of financial development: a country with a relatively high credit to GDP ratio, for the U.S. scores 0.8, for example. The OECS scores 0.30 instance, financial depth alone may only be a necessary on this index, while the LAC average is 0.31. Compared but not sufficient condition to guarantee financial de- to the comparator countries and regions, the OECS velopment. More importantly, if credit is not used for scores much lower, the difference in score between the productive purposes, it can contribute to financial sec- tor instability. Or it could have a limited contribution 13  These subcomponents are constructed based on a number to economic development through excessive personal of underlying variables that track development in each area. The database includes 123 countries for 1995-2013. lending or in the creation of asset price bubbles. This 14  The rationale for selecting these comparator countries are: (i) section assesses the current level of financial develop- Mauritius for its tourism and service-based economy (though Mau- ment in the OECS region from the perspective of a more ritius does have a large offshore financial sector that is aggregated in domestic measures; (ii) Slovenia for a small open-economy that comprehensive measure of financial development. This was able to navigate the financial crisis through careful fiscal man- measure captures depth, access, as well as efficiency. agement, and (iii) Sri Lanka for a service-based economy that has been able to increase SME lending over the past two decades. Taming Volatility Fi ur 3.5 Compon nts of th Fin nci l D v lopm nt Ind x FD Financial Development FI FM Financial Institutions Financial Markets FID FIA FIE FMD FMA FME Depth Access E ciency Depth Access E ciency Source: Sahay et al. (2015a). Fi ur 3.6 Priv t Cr dit to GDP Fi ur 3.7 Fin nci l D v lopm nt Ind x 120% 0.5 100% 0.4 80% 0.3 60% 0.2 40% 57 20% 0.1 0% 0.0 OECS Mauritius Slovenia Sri Lanka EM LAC OECS Mauritius Slovenia Sri Lanka EM LAC Source: WDI (2015). Source: Sahay et al. (2015a). OECS and Slovenia is 0.18, which is approximately one and finance literature. Economic theory suggests that standard deviation of the 2013 index across the world. well-functioning financial intermediaries and markets are the conduit to reduce information asymmetries, With reference to the components of the comprehen- facilitate risk sharing and mobilize savings. This then sive financial development index, the OECS performs leads to a more efficient resource allocation and, thus, less well in the financial market indicators. Figure 3.5 may foster long-term growth.15 A large empirical litera- shows the composition of the financial development in- ture provides evidence that financial development mat- dex used herein. As expected, the OECS performs well ters for growth. However, there is less consensus as to in the depth of financial institutions measure, as one whether the effect is mainly due to banks, stock mar- of inputs to this sub-measure is private credit to GDP. kets or both. The “finance-led growth” hypothesis states However, in financial intuitions in the areas of access that financial development exerts a positive and causal and efficiency it performs less well. The main factor that effect on real economic output. This is mainly supported lowers the score of the OECS is the financial market in cross-country studies that focus on bank develop- scores. These are measures of the demand side – includ- ment proxies.16 When stock market development is also ing the levels of equity and debt market developments. considered, either the direction of causality becomes 15  See (Greenwood J. a., 1990), (Levine R. , 1991), (Bencivenga, The relationship between financial development and 1991), (Greenwood J. a., 1997) and (Blackburn, 1998). economic growth is well recognized in both the growth 16  (King, 1993), (Levine R. N., 2000), (Calderón, 2003) , (Chris- topoulos, 2004), (Rioja, 2004) and (Loayza, 2006). Taming Volatility Fi ur 3.8 Growth & Fin nci l D v lopm nt Fi ur 3.9 Vol tilit & Fin nci l D v lopm nt 0.010 0.0000 DMA GRD ATG 0.005 ATG -0.0025 LCA Economic Growth Growth Volatility KNA VCT 0.000 -0.0050 VCT DMA GRD KNA -0.005 -0.0075 LCA -0.010 -0.0100 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Financial Development Financial Development Source: Authors’ own calculations using the financial deepening index developed by the IMF in Sahay and others (2015a). Note: The curve in Figure 3.8 shows the predicted effect of financial deepening on growth for each level of the index, holding fixed other controls. The curve in Figure 3.9 shows the predicted effect of financial deepening on growth volatility, holding fixed other controls. Growth volatility is measures as the standard deviation of GDP growth rates over a five-year moving average. more difficult to assess or the impact of banking sector the cost of long-term investment is lower in recessions. development on growth is negative.17 Moreover, Carneiro and Hnatkovska (2016) argue that when domestic financial markets are under-developed, More recently, evidence points to a more complex domestic households and firms face binding financial relationship between finance and growth that may constraints that become tighter in bad times, and this be non-linear. Sahay et al. (2015a) develop a new and could amplify the effects of interest rate fluctuations 58 broader measure of financial development and find that on domestic activity. To the extent that higher volatility the effect of financial development on economic growth leads to lower investment rates, output and consump- is bell-shaped. The authors also present evidence that tion, it will result in lower economic growth and welfare. increased financial development is positively related to economic growth at lower levels of financial develop- Following the literature above, the non-linear rela- ment, but that economic growth is weakened at higher tionship between comprehensive financial develop- levels of financial development. Cecchetti and Kharroubi ment and growth as well as volatility also holds for (2015) show that an exogenous increase in finance may the OECS countries. We have confirmed empirically the disproportionately benefit high collateral/low productiv- non-linearity in the association between comprehensive ity projects and affect total factor productivity growth financial development, growth as well as volatility for negatively. These authors also show that, in a model with the OECS countries (Figure 3.8 and Figure 3.9). Using the skilled workers and endogenous financial sector growth, FD index, our results are unequivocal that comprehen- skilled labor will tend to absorb labor talent with the fi- sive financial development impacts growth positively. nancial sector growing more quickly at the expense of the However, this positive effect weakens at higher levels of real economy. Based on these findings, the authors con- comprehensive financial development and then eventu- clude that financial growth may disproportionately harm ally turn negative portraying a bell-shaped relationship financially dependent and R&D-intensive industries. between growth and financial development. We have also confirmed the existence of a non-linear relationship The level of financial development has also been between comprehensive financial development and vol- shown to affect the level of volatility in an economy. atility. As a mirror image of the dynamics between fi- Aghion et al. (2010) identified a transmission channel nancial depth and growth, financial development initial- through the existence of credit constraints in the econ- ly lowers volatility up to a certain point where it starts omy. In the absence of financial frictions, long-term to create additional volatility (Figure 3.9). Given a low investment is believed to be counter-cyclical because FD index of 0.3 for the OECS on average out of 1.0, this suggests that there are additional net benefits to be de- 17  (Beck, 2004), (Shen, 2006), (Arcand, 2012). Taming Volatility rived amongst OECS countries from strengthening their restoring the stability of the banking sector itself. As financial sector in order to foster economic growth and shown above, the sector has come into a period of stress lower volatility, given the minimal scope of policy design. – and this has directly increased volatility. Tackling the source of the stress directly can help reduce volatility in The results above strongly demonstrate that there is the region. sufficient room for the OECS to improve its compre- hensive financial development, which will in turn con- Financial Instruments and Institutions for tribute positively to economy growth and in reducing Volatility Reduction volatility. In both Figure 3.8 and Figure 3.9 the position of each of the six OECS countries is well before the in- It is well noted that financial instruments such as flexion point, beyond which further comprehensive de- savings and insurance, and institutions such depos- velopment of the financial sector may have detrimental it insurance help economic agents better manage effects to economic growth and volatility. Given the high shocks and smooth consumption. This sub-section level of credit to GDP, the OECS should strive to reorient provides an overview of savings, insurance and deposit its financial sector to improve its comprehensive level of insurance in the OECS to understand the policy options financial development. This reorientation – on both the that can be adopted to reorient financial development supply and demand side –will positively contribute to towards improved volatility reduction. economic growth and stability. The subsequent section explores specific policy options to reorient financial so The cost of financing is largely determined by the that it could be more supportive of: (i) lower volatility; bank lending rate. As we have seen, banks are the (ii) enhanced economic growth, and (iii) a more effective principal source of financing for larger enterprises in fiscal policy. the OECS. For smaller enterprises, too, bank lending is important: directly and also indirectly as a source of fi- Entry Points to Reorient Financial nancing for the firms that provide them with trade cred- Development in the OECS it. The ECCU has a statutory minimum savings rate that all regulated entities must pay to savings deposits. In 59 To reorient financial development in the region to be a competitive market, the bank lending rate is equal to more comprehensive, a number of challenges need to the cost of lending. The cost of lending consists of the be addressed. Over the last decade, the OECS countries cost of the necessary funds—for simplicity, we will take have made progress to further develop their financial this to be the rate the bank pays on deposits—and on systems. However, there is scope to reorient financial other costs. In a competitive market the ‘spread’—the development in the region so that it less prone to high difference between the lending rate and the deposit rate collateral-low productivity projects, less likely to create is equal to the sum of these other costs. If the market is asset price bubbles and is able to contribute to enhanc- not competitive, banks will be able to raise their lending ing economic growth and reducing volatility. This sec- rates and/or lower their deposit rates, and the spread tion discusses entry points across three areas: (i) reduc- will be larger than the sum of the costs. ing volatility; (ii) enhancing economic growth, and (iii) effective fiscal policy. If the demand for loans increases then both the lend- ing rate and the deposit rate will rise. If the supply of Reorienting Financial Development to Reduce deposits increases then both the deposit rate and the Volatility lending rate will fall. An increase in ‘other costs’ will cause the spread, and so the lending rate, to increase. More comprehensive and reoriented financial de- In the ECCU, this simple picture is altered by the imposi- velopment can significantly contribute to volatility tion of the Minimum Savings Deposit Rate (MSDR). This reduction in two main areas in the OECS. The first is is a regulation that sets the minimum rate that banks through financial instruments and institutions that are may pay on savings deposits. Its level is determined volatility reducing. Better and more sustainable savings periodically by the Monetary Council of the ECCB. The instruments, a stronger insurance market and the de- MSDR was changed most recently in April, 2015, when velopment of a deposit insurance scheme can all con- it was lowered to 2% from 3% (where it had been since tribute to reducing volatility. The second is through August 2012). Taming Volatility The effect of MSDR is that by setting the deposit rate Although the MSDR is good for savers in the short– at a level above the market-clearing rate it creates an run, it has and will have deleterious long-term effects excess supply of funds. The supply of deposits at that on the financial system, and less distortionary sav- rate is greater than the demand for loans at that rate ings instruments and sustainable welfare programs plus the spread. The difference between the two is ‘ex- should be developed for the OECS population. Savings cess liquidity’—funds the bank is borrowing but unable instruments should not be held captive to the remain- to lend profitably. der of the financial system – as this too can actually increase systemic volatility. Rather less distortionary The MSDR has two effects on the lending rate. The instruments are needed – perhaps even Unit Investment first is direct: if the lending rate is the deposit rate plus Trusts – to access foreign markets domestically could the spread, and if the spread is unchanged, then raising be developed. From the welfare perspective, more tar- the deposit rate to the level of the MSDR will raise the geted programs should be developed to the extent that lending rate. The second effect is indirect. The excess li- the MSDR plays a strong welfare role. quidity created by the MSDR represents a fixed cost to the banks: they have to pay interest on funds they take In addition to changes to the savings market, in but are unable to lend profitably. This increase in fixed strengthening regulation is critically needed in the costs increases the spread, further raising the lending insurance sector. Drawing the lessons from the failure rate, reducing the amount of lending and exacerbating of CLICO/BAICO, a new insurance law was drafted in the excess liquidity. A study by the ECCB found empiri- 2013. One innovation of the law is to create a single cal evidence of this effect. Of course, the MSDR will have financial space, that is insurance companies would get these effects only if it is above the market-clearing rate. a single license to operate in the 8 ECCU countries. In If it is below the market-clearing rate, it will be irrelevant return they would have to increase their minimum cap- and have no effect at all. ital and be subject to stronger prudential standards. The bill provides the supervisory with stronger correc- The decrease in the demand for loans and the increase tive action and intervention powers. The insurance bill 60 in the supply of deposits should have led to a signifi- was drafted together with the bill establishing a single cant fall in the rate on savings deposits. In the Unit- supervisor for non-bank starting with insurance: East- ed States, the rate on savings deposits fell, in the same ern Caribbean Financial Services Commission (ECFSC). period, to close to zero and has remained there ever The purpose is to optimize supervisory resources and since: today, the average rate on MMAs (a popular type make them more efficient and better able to super- of savings deposit) is 0.11%. The market-clearing rate in vise multi-country insurers. However the insurance the OECS today is probably similarly close to zero. So bill faces opposition from small insurers, which would a MSDR of 2% is effective, and it is probably raising the face a consolidation process, and the bill on the single lending rate by significantly more than 2% by creating non-bank supervisor has encountered opposition from excess liquidity. the country supervisors. Both bills should be urgently approved to ensure financial soundness and more effi- This is an example of ‘financial repression’ and the cient supervision. precise motivation for the MSDR is unclear. Perhaps it is a politically popular subsidy for savers or an ex- The overcrowding of financial institutions has ad- tension of the welfare system. Perhaps the MSDR is a verse implications for supervision and harmonizing way of bailing out the national social security schemes this should be a long term goal. The reality of eight dif- which deposit a large part of their funds in the banks. ferent jurisdictions means that the required superviso- Nevertheless, the MSDR is an example of the broader ry apparatus is subject to an 8 times multiplier effect, phenomenon of ‘financial repression’—the suppression being applied to a number of licensees that is already of market-driven financial activity as a result of govern- much larger than necessary. Based on data provided by ment involvement in the financial system and of poor the ECCB, the 61 regionally licensed insurers actually management of government finance. The effects of fi- comprise a group of 161 separate registrants because nancial repression in general are to raise borrowing rates most of the insurers are licensed in more than one of for the private sector and to reduce the availability of the eight jurisdictions (see Table 3.1). The creation of a financing. single non-bank supervisor should also be considered, Taming Volatility Table 3.1: Number of Insurers Country/ Region Population No. of insurers Insurers per 100,000 ECCU 600,000 61 10.17 Bahamas 383,054 29 7.57 Barbados 283,380 26 9.17 Belize 351,706 15 4.26 British Virgin Islands 32,680 1 3.06 Trinidad 1,354,483 24 1.77 UK 64,559,135 911 1.41 Puerto Rico 3,548,397 54 1.52 Canada 35,543,658 233 0.66 Panama 3,867,535 28 0.72 Germany 80,970,732 560 0.69 Jamaica 2,720,554 18 0.66 Dominican Republic 10,405,943 31 0.30 Source: Individual country Insurance Commissions, WDI. and made a long term goal. In 2013, as a lesson learned and could even be used as an incentive for banks to from the failures of BAICO and CLICO, the ECCU coun- strengthen their balance sheets. tries agreed to set up a single non-bank supervisor for insurance, credit unions and offshore businesses, which A new deposit insurance scheme should follow the 61 would start with insurance. In addition, the quality of International Association of Deposit insurance (IADI) supervision of credit unions is uneven across countries, elements, in addition to precluding legacy issues from and a regional body could implement the best practices affecting the new scheme. The IADI essential elements in the least advanced countries. These changes in har- for an effective deposit insurance provide an excellent monized supervision, will help strengthen the insurance framework for devising new schemes. The four elements market particularly, and by doing so will contribute to a are: (i) limited but credible limits for insured deposits; reduction in volatility. (ii) adequate funding for the deposit insurer and explicit back-up funding to allow the fund to respond to a num- The lack of an explicit financial safety net is a glaring ber of institution failures at once; (iii) an effective pub- gap from both the stability and depositor protection lic awareness campaign to make sure that depositors perspective. The OECS countries are amongst the few know the rules on coverage and are aware of the extent middle-income countries without explicit deposit insur- of and limits to existing compensation arrangements, ance.18 Instead they have an implicit guarantee on the and (iv) an effective deposit insurance system requires deposits, which, in the absence of fiscal space, is not ful- certain prerequisites including the presence of a working ly credible. The tight fiscal situation severely constrains supervisory framework and good coordination among the governments’ capacity to underwrite the deposit all members of the financial safety net. Finally, in the base. In addition, the provision of emergency liquidity context of the ECCU it must be the case that legacy is- assistance by the ECCB is restricted by the currency sues do not impair the viability of the scheme. board. Therefore it is critical that the foundations for a formal deposit insurance scheme should be put in place Restoring Banking Sector Stability to Reduce once the system has been sufficiently strengthened, Volatility 18  Out of the 28 middle-income countries without explicit de- The banking sector is currently in a position of insta- posit insurance, 6 are in the OECS (WB staff calculations form bility, and restoring the stability of the banking sec- Demirgüç-Kunt, Kane, & Laeven (2014)). Taming Volatility tor will itself contribute to reduced volatility. In the along these lines, and this should be regionalized. Fur- aftermath of the financial crisis, the banking sector in ther, when fully operationalized, the new Banking Act the OECS has experienced a period of prolonged stress. will reduce NPLs classification challenges and raise loan Given its size, this stress has increased volatility in the loss provisioning standards. Finally, the new Appraisal region. This has been manifested by increased NPLs, Institute that has been established, that will harmonize lower profitability, higher bank fees and three resolved real estate appraisals across the region and reduce arbi- banks over the past five years. Addressing this very trage. On the other hand, provisions should be assigned stress will contribute to more comprehensive financial to debtors (rather than to loans), with more frequent re- development and reducing volatility. views and cash flow analysis to determine categories; in addition provisioning levels should be more stringent, The new Banking Act is a significant legislative im- while write-off standards could also be strengthened.19 provement across the ECCU and its operationaliza- tion will be critical for the future stability of the re- For current NPLs, more concerted efforts will be re- gion. As of early 2016 all eight member countries of the quired to swiftly dispose of and thereby recognize OECS have now passed the new Banking Act. This is these losses. The World Bank has recently released a very welcome as it provides the foundation for a stron- toolkit for public asset management companies (AMC).20 ger banking sector. This has strengthened the regulato- An AMC is a statutory body or corporation that assumes ry and supervisory frameworks, including by introducing the management of distressed assets and strives to re- higher minimum capital requirements, a more effective coup the public cost of resolving them. Other options resolution of failed banks, and a stronger depositor pro- are to improve the framework for debt enforcement so tection. Looking ahead, enhancing and establishing su- as to facilitate the enforcement of collateral. This goes pervisory procedures to meet the new legislation will be beyond insolvency reform but does take considerable critical to underwrite the stability of the banking system time, as demonstrated in Latvia. On the other hand, if and thereby reduce volatility that has been created from distressed assets are confined to limited number of in- banking sector weaknesses. stitutions, special purpose vehicles can be established 62 as a subsidiary of the bank or its holding function in or- The diagnostic work to assess the strength of the der to work out the distressed assets. A final option is banking system has now been completed, and com- to attract private distressed asset funds. These usually mensurate supervisory as well as regulatory actions have a minimum size threshold (and are therefore not will be required to strengthen the commercial banking suited to small loan sizes) and require higher rates of re- system. An extensive diagnostic exercise was undertak- turn. They do, however, require no public funds and are en to assess the degree and spread of commercial bank usually more expedient than public agencies. strengthening needs. Now that this has been completed, a comprehensive strategy is now needed to collectively The World Bank public AMC toolkit highlights import- buttress the foundation of the financial sector. Given the ant preconditions that must be in place. Public AMCs prevailing efficiency and business models of the sector, a require assessing whether there exist the conditions to focus on consolidation and leaner commercial banks will ensure that the objective of the effective management augur well for both the strength of the banking system and sale of distressed assets through a public entity. as well as its ability to serve the needs of the economy. The preconditions relate to (i) a commitment to com- prehensive reforms, (ii) a systemic problem and public Aside from strengthening the banking system, a funds at risk, (iii) a solid diagnostic and critical mass of number of key steps have been taken to better man- impaired assets, (iv) a tradition of institutional indepen- age future NPLs. In the long run, a stronger insolvency dence and public accountability, and (v) a robust legal framework will be a key enabler to helping banks bet- framework for bank resolution, debt recovery, and cred- ter manage collateral collections. However foreclosure itors’ rights. The analysis of nine case studies of AMCs in is last resort and yields the lowest recovery rates, and the toolkit shows that AMCs have a mixed track record. therefore efforts should also be made to help the system consider liquidation, receiverships and reorganizations 19  For instance a prudential requirement that after a certain time past due the loan be written off; or raising risk-weights on within the insolvency frameworks. There is incipient impaired assets beyond a certain vintage. reform in St. Lucia and St. Vincent and the Grenadines 20  World Bank (2016) Public Asset Management Companies: A Toolkit Taming Volatility AMCs are costly to establish and to operate; therefore The outcome of the process of managing and resolv- their costs and benefits should be assessed carefully be- ing existing NPLs should be favorable if the right tools fore they are established. to address existing NPLs are used. International best practice has shown that resolving NPLs is tricky, espe- A regional Asset Management Company (AMC) could cially once political economy implications are consid- help clear the vast amount of NPLs. A regional public ered. What is critical is that the management and reso- AMC would offer various advantages including: lution of existing NPLs is swift and non-reversible. This could be a combination of a public AMC, working with • Address the lack of scale and depth of property mar- private distressed asset funds as well as improving the kets at the national level: a regional AMC would be framework for debt enforcement. Provided that these able to bundle loans from different banks and coun- tools are designed well, have sufficient scale and use tries to facilitate their sale to investors. public resources efficiently, the banking sector and the • Address constraints in country insolvency and fore- inherent volatility therein would strongly benefit from closure legislation: the regional AMC has been grant- the removal of this heavy strain on its balance sheet. ed special powers to expedite the management of bad loans, which draw on the most advanced coun- Reorienting Financial Development for try legislation and would benefit all the ECCU region. Enhanced Economic Growth • Ensure transparency: the regional AMC has been set up as a statutory body owned by all governments. As has been shown above, the financial sector of the It would be subject to strong transparency require- OECS is overly reliant on the high collateral low pro- ments and the scrutiny of all country Parliaments. ductivity model and is prone to asset price bubbles, • Minimize corruption: the distressed asset manage- and therefore needs to reorient towards more compre- ment business is prone to corruption. A regional body hensive financial development for achieving enhanced would be less prone to local political interference and economic growth. This section covers three areas that pressures from the banks. could help reorient financial development and support • Attract specialized expertise: a regional institution growth: (i) long-term finance; (ii) SME finance, and (iii) 63 with a critical mass of assets would be more able to the enabling environment and credit infrastructure. attract international and expert staff than a local one. Currently there is very limited expertise in dis- Long-Term Finance tressed asset management in the ECCU. The scarcity of long-term credit in developing market As of early 2016, Eastern Caribbean Asset Manage- economies is recognized as an obstacle to productiv- ment Corporation (ECAMC) Agreement was signed ity growth and a source of volatility which further by seven OECS member countries21 and the ECAMC hampers growth. Investments in infrastructure, facto- legislation had been passed in seven territories.22 This ries and equipment, new housing and commercial busi- legislation provides for: time bound safeguards that lim- ness, research and development and education are all it the period for purchasing assets and a fixed sunset necessary to expand the productive frontier. Firms and clause; financial safeguards for initial equity, a leverage households need to fund these investments through fi- ratio; a minimum capital ratio of 2 percent; and, a dual nancial markets. The scarcity of long-term credit is one mandate: for the AMC to (i) purchase NPLs from solvent of the most important impediments to the operations institutions; and (ii) act as a receiver and thus managing of firms in emerging markets; rollover risks prevent NPLs of insolvent institutions. This in addition to its spe- borrowers from undertaking long-term investments cial powers to expedite the management of bad loans, with short-term loans which in turn negatively affects which draw on the most advanced country legislation productivity.23 Scarcity of long-term financing has also and would benefit all the OECS region. been argued to increase the pro-cyclicality of invest- ment, amplifying macroeconomic volatility and lower- ing growth. According to the 2014 Financial Develop- 21  Antigua and Barbuda, Dominica, Grenada, Montserrat, St ment Barometer, 75 percent of respondents indicated Kitts and Nevis, St. Lucia and St Vincent and the Grenadines. 22  St. Lucia which await assent, Anguilla, Antigua and Barbu- da, Dominica, Grenada, Montserrat and St Vincent and the Gren- 23  (Demirgüç-Kunt, 1999), (Schiantarelli, 1997), (Schiantarelli F. adines. a., 1997) , (Jaramillo, 2002), (Almeida, 2009) Taming Volatility that low use of long-term finance in their countries was Fi ur 3.10 ECHMB Bonds Outst ndin b T nor primarily a supply problem.24 Extending the maturity structure of financing is at the core of sustainable finan- $250 cial development. $200 Banking systems are the main providers of long-term $150 EC$ million financing to the private sector around the world. Long-term funding is typically defined as debt (loans or $100 bonds) over one year maturity (following the definition of fixed investment in National accounts). The Group of $50 20 (G-20, 2013) defines it, however, as debt of over 5 years maturity. Equity, as it is not repayable, can also $0 be considered long-term finance. Firms around the 07 08 09 11 12 13 14 15 16 0 -1 n- n- n- n- n- n- n- n- n- Jun Ju Ju Ju Ju Ju Ju Ju Ju Ju world obtain most of their funds to fund fixed assets from banks, regardless of their size. While firms in high 1 year 2 year 3 year 4 year 8 year income countries raise more long-term and debt from capital markets that firms in developing countries, over 60 percent of their investments is funded by bank loans. funding liquidity for the Primary Lenders. Only recently Households’ main long term investment, housing, is also has the ECHMB sold back mortgages. Thus for most of overwhelmingly funded by banks. Capital markets are its existence the ECHMB has not operated as a two-way also important indirect source of long-term financing to liquidity buffer, instead it has only generated liquidity. the extent they fund banks provision of long-term credit The ECHMB’s operations are by definition limited to the through purchases of long-term bank bonds and equi- mortgage market. While the Primary Lenders may have ty or through investments in long-term deposits in the recycled the funds received from sales of mortgages case of pension funds and insurance companies. into other lending lines, providing general liquidity to the 64 banks is not part of the ECHMB’s mandate. In the OECS there is a need for more long-term finance for infrastructure and housing finance. Given the re- Of late, the same patterns of moving to shorter tenors gion’s susceptibility to natural disasters, the disaster hold true for the ECHMB as it does for the sovereign risk mitigation (DRM) agenda is steep. Long-term fi- debt market, although on a smaller scale and involv- nancing is required for such projects. However, the mar- ing fewer tenors. ECHMB began its bond funding in ear- ket for infrastructure finance is not very well developed. ly 2008 using a combination of 8-year and-2 year debt. Granted it is a small market, however given the amount Beginning in mid-2011 it shifted to using 3-year and of liquidity in the market it would be worth exploring to 4-year debt. Then in late 2013 it moved to using 1-year what extent long-term infrastructure projects can be and 2-year maturities. The overall picture is a shift to securitized either by banks or the capital markets. How- short-term bonds, with the attendant lower rates. Late- ever, in housing more has been done through the cre- ly the ECHMB seems to be relying on 1-year securities ation of the Eastern Caribbean Home Mortgage Bank almost exclusively. (ECHMB). Therefore the long-term finance agenda needs to Over most of its existence the ECHMB has acted as a be reinvigorated to enhance economic growth. One liquidity provider for mortgage lending. The ECHMB’s clear departure from the high collateral low productiv- stated objective is promoting the development of the ity model is greater long-term finance for productive secondary mortgage market within the Member States. uses. This is both for banking as well as the capital Its principal activity is buying and selling mortgage markets. Indeed, it is worrisome that the capital mar- loans on residential properties, in order to develop and kets for sovereign debt have become more short term. maintain a secondary market in these instruments. As Efforts to develop infrastructure finance, improve the such, the Bank aspires to be the principal provider of ECHMB mandate and the market for housing as well as work on long terms corporate financing will help to 24  Financial Development Barometer. www.worldbank.org/fi- reorient financial development into being more com- nancialdevelopment. Taming Volatility Fi ur 3.11 Lo ns R quirin Coll t r l Fi ur 3.12 Coll t r l V lu to Lo n (%) 100% 140% 120% 80% 100% 60% 80% 40% 60% 40% 20% 20% 0% 0% OECS Mauritius Slovenia Sri Lanka EM LAC OECS Mauritius Slovenia Sri Lanka EM LAC Source: OECS (2014 Compete Caribbean Survey), Mauritius Source: OECS (2014 Compete Caribbean Survey), Mauritius (2009 Enterprise Survey), Slovenia (2009 Enterprise Survey), (2009 Enterprise Survey), Slovenia (2009 Enterprise Survey), Sri Lanka (2011 Enterprise Survey), EM and LAC are Enterprise no data for Sri Lanka, EM and LAC are Enterprise Survey averages Survey averages 2009-2011. 2009-2011. prehensive and thereby contribute to enhanced eco- of loans to enterprises (including large enterprises) re- nomic growth. quired collateral in the OECS. This is much higher than the LAC average of 60 percent. Further, loan size in the SME Finance OECS have much larger discounts with respect to the underlying collateral. In 2014, the collateral to loan value Credit markets for Small- and Medium-Sized Enter- was 144 percent. This certainly reflects the same con- prises (SME) are notoriously characterized by mar- cern about inflated asset prices, however the data from ket failures and imperfections, and this is the case in other peer countries and regions shows that for its level 65 the OECS as well. These include information asymme- of income, this ratio is very high. tries, inadequacy or lack of recognized collateral, high transaction costs of small-scale lending and percep- The second structural challenge in SME lending is the tion of high risk. This is also because commercial banks complete lack of credit information systems. Infor- are restricted by the central bank in types of collateral mation asymmetries contribute to higher credit spreads they consider to be acceptable. In weak informational and lower financing for SMEs. The OECS neither has environments, commercial banks traditionally display a credit bureaus nor credit registries. This lack of informa- preference for consumer lending which is perceived as tion on borrowers is a key structural challenge. In fact all less risky. As can be seen in Figure 3.1 and Figure 3.2 the six OECS countries have a global rank of 152 out of 185 stock of credit has begun to decrease over the recent countries in the World Bank’s getting credit indicator of past. This decrease has affected non-personal lending its Doing Business rankings, and this is primarily due to more than personal lending. no credit bureau or registry coverage, but also because of creditor rights (Figure 3.13). One main structural challenge in SME lending in the OECS is represented by the over reliance on collateral- More recently, there have been signs of a cycli- ized lending. Given that lending to SMEs is risky, many cal stress in intermediation to SMEs as a result of financial institutions require collateral. However, when the financial crisis and the subsequent increase in the lending culture becomes over collateralized, then few NPLs. The prevailing trends from Figure 3.1 and Figure economically viable projects get funded as lending deci- 3.2 show that credit has begun to contract across the sions are over determined by the quality of the collater- OECS. As asset quality has deteriorated and in the con- al, and less so on the strength of the business plan. Fig- text of poor information on borrowers, financial insti- ure 3.10 and Figure 3.11 show the percentage of granted tutions have pivoted their portfolios towards personal loans that require collateral and the collateral value to lending. Figure 3.2 shows that the only lending segment loan value for loans to enterprises. In 2014, 88 percent that experienced positive credit growth in 2012, 2013, Taming Volatility and 2014 was personal lending. Interviews with finan- leaving less room for distortions in credit markets than cial institutions have confirmed that credit risk for SMEs more direct forms of intervention such as state-owned has now increased as a result of the past experience of banks. The recently released principles provide sixteen SME portfolios. guidelines to maximize their success.26 There is a risk that this cyclical stress could further There is a continued need for training and upgrading worsen the structural challenges in SME lending if left of skills of SMEs to help facilitate their financing. A unaddressed. This evolving credit market imperfections short 2015 World Bank survey of SMEs across the OECS requires attention, and there is a small possibility that showed that 40 percent of firms lacked business plan SMEs financing comes to stall, which would have worry- development skills. There is a critical gap in that many ing consequences for economic growth and job creation. financial institutions do not have the full expertise to as- Some initiatives have already been attempted, however sess credit for SMEs, while many SMEs do not have the a review of 11 regional and national interventions (main- requisite training or skills in accounting or business plan- ly grant-based) showed that they suffered implemen- ning to secure financing. There are a number of region- tation and funding challenges, and have been unable to al initiatives which have shown promise – including the fully close the gap.25 Organization of American States (OAS) Small Business Development Centers, the OECS Commission’s Compet- As the engines of economic growth, countries respond itive Business Unit as well as the recent due diligence in four main ways to addressing such market imper- conducted by the German Savings Bank. It is import- fections. The first is through liquidity support, the sec- ant that these efforts are harmonized and coordinated ond through directed lending, the third through building with the supply side instruments chosen to address the credit infrastructure and the fourth through guarantee evolving credit market imperfection for SMEs. schemes. Given the lack of fiscal space, the limited num- ber of state banks and no historical precedent of direct- Enabling Environment & Credit Infrastructure ed lending, as well as the level of liquidity in the system 66 suggests that the latter two would be instruments to Access to financing remains a key obstacle for enter- consider to reverse the tide of credit away from SMEs. prises across the OECS, despite the high number of in- This would directly help in the reorientation of financial stitutions and the high level of credit to GDP. Between development towards being more comprehensive, and 2010 and 2014 the percentage of enterprises that re- contribute to enhanced economic growth. port access to finance as the main obstacle to doing business has increased from an average of 23 percent On credit guarantees, the World Bank has recent- of 28 percent. Figure 3.12 shows the number of firms ly released a set of principles as well as necessary in each OECS country reporting access to finance as pre-conditions that are required when setting up such their main obstacle. This has increased in every country schemes. Partial credit guarantee (PCG) schemes pro- except for Dominica. Figure 3.13 shows that obtaining vide third-party credit risk mitigation to financial inter- credit remained an important deterrent to the business mediaries with the objective of increasing access to fi- environment in the Doing Business surveys of 2007 and nance for SMEs. This is done through the absorption of a 2015. For all the countries in the region, their ranks have portion of the lender’s losses on the loans made to SMEs worsened – this is a result of other countries improving in case of default, in return for a fee from the financial their credit bureau and registry coverage as well as their institution. The appeal of PCGs is partly due to the fact depth of credit information. that they typically combine a subsidy element with mar- ket-based arrangements for credit allocation, therefore Credit reporting systems are critical components to reduce information asymmetries and allow lenders to 25  Dominica Small Business Unit, Dominica Youth Business more accurately evaluate risks. Good and reliable in- Trust, Waitukubuli Entrepreneurs Lévé, Dominica Women Entre- preneurs Fund and Startup Stars: Youth Special Fund ,Grenada formation can eases the adverse selection problem and Industrial Development Corporation Technical Assistance ,Grena- lower the costs of credit for good borrowers. For exam- da Ministry of Agriculture & Fisheries Lending Program, Grenada ple in Egypt, the average percent of sub-standard loans Small Business Development Fund, St. Lucia National Competi- tiveness and Productivity Council, Antigua Small Business Credit decreased by 0.94 percentage points between 2009-11 Guarantee Scheme, ECCB Eastern Caribbean Enterprise Fund, and the ECCB Export Credit Guarantee Scheme 26  World Bank (2015) Principles for Public Credit Guarantees Taming Volatility Fi ur 3.13 A2F M in Constr int (%) Fi ur 3.14 DB G ttin Cr dit R nks 50% VCT 40% LCA 30% KNA 20% GND DMA 10% ATB 0% ATB DMA GRD KNA LCA VCT 0 50 100 150 2010 2014 2007 2015 Source: 2010 Enterprise Survey and 2014 Compete Source: World Bank Doing Business Survey. Caribbean Survey. after credit bureau was established in 2007. Further transactions and in the property that can be used as there is evidence that this information can also increase collateral. In well-functioning markets this can increase credit volumes and improves access to credit. Data the credit capital available in the local economy, creating show that there are 89 percent higher approvals in cas- a virtuous legal / economic environment. If the collater- es where full-file information sharing exists, 11 percent al registries also include movables, this allows SMEs to higher approvals where information is collected from use movable assets – which comprise of 80 percent of traditional and non-traditional lenders; and in another SME assets in general – as collateral for a loan. There is case 12 percent more firms have access to bank financ- evidence that this can increase the competitiveness of ing when a credit bureau exists.27 Finally, these systems domestic economic actors: in China, the total number of 67 support the introduction of credit scoring and automat- commercial loans involving movable assets grew by 21% ed underwriting, which lowers lender’s operational costs per year over 2008-2010, in comparison to a flat rate over and improves profitability. the preceding period.28. Finally, secured transactions pro- vides certainty and transparency in the priority of cred- A draft credit reporting bill has been prepared for the itors (including insolvency proceedings) and certainty in OECS region and once passed this can facilitate the the rights of third parties, reducing risk and cost of credit. beginning of a new credit bureau. Although the legisla- tion is ready, it is important to finalize a thorough con- There is an ongoing initiative in St. Lucia on the legal sultation process to share the changes that this new leg- framework for movable secured transactions that islation will bring. As with any new legislation there are could potentially be regionalized. As with the impor- uncertainties and these need to be fully understood be- tance of any credit reporting system to be regional, fore embarking on its implementation. Given the OECS so is the case for secured transactions. An OECS-wide member states are small island economies, maximizing secured transactions initiative could build on the work scale through the development of a single registry for that has already commenced in St. Lucia. This could be- the entire union is certainly preferable. Further, ongoing gin with the development of an OECS legal framework financial literacy and awareness raising for the general to support the implementation of a modern system of public and for all other stakeholders should be perused. financing secured by movable assets, or the drafting and harmonization of local legal frameworks. Other op- The second pillar of credit infrastructure is secured tions could include the possible development and design transactions which provides greater flexibility in loan of an OECS security interest registry, and the building of local capacity to benefit from the new system, including 27  Love and Mylenko (2003): Credit Reporting and Financing stakeholder training. Constraints, World Bank Policy Research Working Paper 3142, Oc- tober 2003; Barron and Staten (2003): The Value of Comprehen- 28  Secured Transactions Advisory Project in China. IFC Advisory sive Credit Reports: Lessons from the U.S. Experience. Services. International Finance Corporation. 2012. Taming Volatility On non-movable secured transactions – land – there Reorienting Financial Development for is a need to work on land reform across the OECS and Effective Fiscal Policy this could have substantial impacts for the finan- cial and non-financial sector. Land reform is lengthy, There is an important link between financial devel- cross-cutting and is not politically straight forward. opment and fiscal policy: sovereign debt markets. However there is a strong need to harmonize and digi- Around the world, governments finance fiscal deficits tize land registries. The OECS has both a registered sys- through public borrowing from the financial markets. tem and an unregistered system. A registered system The effectiveness of fiscal policy from this perspective provides conclusive evidence of the holders of rights in depends on two key variables. The first is the interest a particular property. While an unregistered system rate of public debt and the second is the tenor of public does not provide conclusive evidence of land ownership. debt. The confluence of financial market development In four territories there is a cadastral system (Anguilla, and macroeconomic variables (such as inflation, country Antigua, St. Lucia and Montserrat), in St. Vincent and risk, and economic growth) determine the range of these the Grenadines there is a deed system, while in Dominica two variables. Low interest and longer tenor debt allow and St. Kitts and Nevis, and Grenada there is convey- governments to attain cheaper (in present value terms) ancing. public finance and facilitate better planning of medium to long terms public finance. The last pillar of credit infrastructure is insolvency and creditor and debtor rights, and data from Doing The Regional Government Securities Market (RGSM) Business 2016 show that reorganizations present the handles both the primary market (offerings) and best proceedings of insolvency. Studies have shown secondary market (trading) of the debt instruments that effective insolvency reform is associated with: issued by the OECS Member States. Today, five juris- lower credit costs; increased access to credit; improved dictions participate: Dominica, Antigua and Barbuda, creditor recovery; strengthened job preservation; pro- Grenada, St. Lucia and St. Vincent and the Grenadines. motion of entrepreneurship, and other benefits for small The RGSM handles both treasury bills (91 days to 1 year 68 businesses.29 Insolvency is often associated only with maturities) and treasury bonds (maturities up to 10 foreclosure, but there are four possible proceedings of years). The market was established in November 2002 insolvency: foreclosure, liquidations, receivership and and utilizes the ECSE’s fully electronic platform. Auc- reorganization. Data from Doing Business show that re- tions and trading are performed through the ECSE’s coveries are 79 percent in reorganizations and only 30 trading system with transactions cleared and settled percent in foreclosure. through its depository / registry subsidiary. In St. Lucia there are efforts underway at reforming At first glance the government securities market the insolvency framework and the OECS would bene- offers a variety of tenors in growing numbers. Using fit from regionalizing this initiative. The new legislation auction data provided by the RGSM we have calculated addresses insolvency regulation of corporates, SMEs the amount of government securities issued and out- (such as sole proprietorships) and individuals. The new standing through the ECSE, for all participating coun- law is intended to be a model for the whole OECS region tries combined. Figure YY shows outstanding govern- and beyond, and follows many practices already initiat- ment securities in excess of EC$ 1.6 billion, divided fairly ed by Jamaica and Trinidad & Tobago. Once the law is evenly among the tenors. This would seem to indicate a passed, outreach and training will need to be conducted wide choice of investment maturities. Indeed focusing to ensure effective implementation. Regionalizing this on the outstanding tenor amounts over the nearer term initiative across the OECS and ensuring coherence with (i.e., the last 5 years) reinforces the idea that there are a the secured transitions reform would contribute greatly variety of longer term tenors available. to the progression towards the single financial space. However, if one focuses on the levels of recent offer- ings among the available tenors it is clear that only a minimal amount of longer-term securities are being 29  Araujo, Ferreira and Funchal (2012), Gamboa-Cavazos and issued. The vast majority of auctions conducted on the Schneider (2007), Armour, Hsu and Walters, (2012), Gine and Love (2010) and Frisby (2007). RGSM are to rollover existing short term T-Bills (91 and Taming Volatility Fi ur 3.15 RGSM Outst ndin b T nor Fi ur 3.16 RGSM Outst ndin b T nor (2002-2015) (l st fiv rs) 2000 2000 1500 1500 Million EC$ Million EC$ 1000 1000 500 500 0 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 Mar Sep Mar Sep Mar Sep Mar Sep Mar Sep Mar 11 11 12 12 13 13 14 14 15 15 16 15 year 10 year 6-8 years 5 year 15 year 10 year 6-8 years 5 year 2-4 years 1 year 180 days 91 days 2-4 years 1 year 180 days 91 days Source: RGSM. Source: RGSM. Fi ur 3.17 Amounts Issu d 2011-2016 Fi ur 3.18 Amounts Issu d Adjust d for Short-T rm Rollov rs 2011-2016 15 year 15 year 10 year 10 year 8 year 8 year 7 year 7 year 6 year 6 year 69 5 year 5 year 4 year 4 year 3 year 3 year 2 year 2 year 1 year 1 year 180 days 180 days 91 days 91 days 0 1000 2000 3000 0 400 800 Million XCD Million XCD 180 day securities and 1 year maturities). Even adjust- excess liquidity in the banking system alone. This also ing the offering amounts to recognize the rollover nature ignores the investable instrument needs of the credit of the shorter-term offerings it is still clear that there a unions, insurance companies, and pension funds (both few medium- to long-term securities being created. The Pillars I and II, and the private schemes). ‘inventory’ of longer-term government securities is not growing appreciably. This impression of a short supply of longer-term se- curities is borne out by a comparison of the second- Ignoring, for the sake of the argument, that some of ary trading in OECS government securities compared the offerings of 2-year to 10-year bonds were to roll to the available amount outstanding. In a developed over previous longer-term debt, the total amount is- capital market, short-term securities are known as “buy sued between 2011 through 2016 for these tenors and die paper”. This means that the investor purchases has been EC$ 925 million. This compares to EC$ 5.1 the securities for investment, holds until maturity, and billion in idle funds within the banks themselves, or 18%. then perhaps purchases again in the rollover issuance. It is clear that, even under the most generous analyti- One does not expect much if any secondary market cal terms, the amount of offerings of longer-term gov- trading in T-Bills (91 days to 1 year). The opposite is true ernment securities cannot be expected to absorb the for longer term maturities. It is expected that there is Taming Volatility a positive correlation between the time left to maturity Through the exploration of a multidimensional mea- for any given security and the level of secondary trading sure of financial sector development, the results in that security. In other words, the longer the remaining found in the literature of a non-linear relationship for maturity the more secondary trading expected. financial sector development with growth and volatil- ity were confirmed by adding the OECS countries. Fol- Deviating from this norm, the secondary market for lowing the approach in Sahay et al. (2015) it was shown OECS government securities shows almost no turn- that each of the OECS countries lies below the inflection over in government securities for all tenors. Again, point for economic growth and for volatility. This implies while this might be expected for T-Bills, it is contrary to that reorienting financial sector development to be more the normal market dynamic for longer term bonds. The comprehensive can contribute to reducing volatility, en- investors buying T-Bonds intend (at least today) to buy hancing economic growth and facilitating more effective and hold. The turnover rates for T-Bonds are extremely fiscal policy. This is particularly salient in the context of low compared to standard expectations; and thus the the currency board that anchors the exchange rate, the data reinforce the idea that longer-term government se- limited fiscal space as well as the service-based econo- curities are in high demand / short supply for the finan- mies and implies that the financial sector is critical for cial sector’s needs. Once acquired they are not traded continued economic growth and the enhanced manage- out again in the secondary market. ment of risks. Therefore, reversing the recent trend towards issuing Finally, the chapter identified policy options that shorter term securities and further developing the could help reorient financial development. Improving secondary market will have material effects on the savings instruments, strengthening the regional super- efficacy of fiscal policy. Public debt markets have a vision of insurance, and establishing deposit insurance range of policy levers and comprise of many stakehold- could all help economic agents better manage volatili- ers. Nevertheless exploring options that will lengthen ty. Restoring banking stability can reduce the systemic the yield curve across the Member States and also lower volatility that has emanated from the heightened stress 70 interest rate profiles can provide enhanced stability to that the banking sector has experienced over the past both the private and public sectors. With more predict- five years. The new Banking Act that has been passed able estimates of the cost of public debt, policy makers provides the foundation for improved banking supervi- will be able to secure cheaper financing (in present value sion and future consolidation. Developing long term fi- terms) and be better able to plan medium to long term nance for infrastructure and more housing in addition to public financing needs. developing tools for more effective SME finance against the backdrop of improvements to the enabling environ- Conclusions ment and credit infrastructure are critical for enhancing economic growth. Finally, this chapter highlighted the This chapter begun by identifying the economic weak- importance of reversing the short-termism in sovereign nesses that emanate from a large and frail financial debt markets in the OECS and striving for a more active sector. With banking assets at 166 percent of the re- secondary market as a way to improve the efficacy of gion’s GDP, the banking sector is indeed large. Insurance fiscal policy in the OECS. assets, the number of insurers as well as the size and number of credit unions also point to a large non-bank- ing sector as well. Further, the aftermath of the 2008 financial crisis has put a strain on the banking sector. Asset quality has deteriorated and as a result bank prof- itability as well as credit growth have suffered. It would be remiss to conclude that, given the size of the sector, that further more comprehensive financial development is not warranted. Taming Volatility REFERENCES ruptcy Reform in Colombia. Journal of Law and Eco- nomics. 53(4) 833-864 Almeida, H. C. (2009). Corporate debt maturity and the Greenwood, J. a. (1990). Financial Development, Growth, real effects of the 2007 credit crisis. National Bureau and the Distribution of Income. The Journal of Politi- of Economic Research(No. w14990). cal Economy, 98 (5 Part 1). Araujo, A., R. Ferreira, and B. Funchal (2012), The Brazil- Greenwood, J. a. (1997). Financial markets in develop- ian bankruptcy law experience, Journal of Corporate ment, and the development of financial markets. Finance, 18(4), 994-1004 Journal of Economic Dynamics and Control, 145-181. Armour, J, Hsu, A. and Walters, A (2012). The Costs and Jaramillo, F. a. (2002). Acces to Long Term Debt and Ef- Benefits of Secured Creditor Control in Bankruptcy: fects on Firms’ Performance. World Bank Publications. Evidence from the UK. Review of Law & Economics. King, R. G. (1993). Finance and growth: Schumpeter Vol 8 might be right. he quarterly journal of economics, Arcand, J.-L. E. (2012). Too much finance?. IMF Working 717-737. Paper. Levine, R. (1991). Stock markets, growth, and tax policy’. Barron and Staten (2003): The Value of Comprehensive Journal of Finance, 46(4) 1445–65. Credit Reports: Lessons from the U.S. Experience. Levine, R. N. (2000). Financial intermediation and Credit reporting systems and the international econ- growth: Causality and causes. Journal of monetary omy, 8. 273-310 Economics 46.1, 31-77. Beck, T. a. (2004). Stock markets, banks, and growth: Loayza, N. V. (2006). Financial development, financial Panel evidence. Journal of Banking & Finance 28.3, fragility, and growth. Journal of Money, Credit and 423-442. Banking, 1051-1076. Bencivenga, V. R. (1991). Financial intermediation and Love and Mylenko (2003): Credit Reporting and Financ- endogenous growth. The Review of Economic Stud- ing Constraints, World Bank Policy Research Work- ies 58.2, 195-209. ing Paper 3142, October 2003 Blackburn, K. a. (1998). A theory of growth, financial de- Peia, O. a. (2014). Finance and growth: time series evi- dence on causality. Journal of Financial Stability. 71 velopment and trade. Economica 65.257, 107-124. Calderón, C. a. (2003). The direction of causality be- Rioja, F. a. (2004). Does one size fit all?: a reexamination tween financial development and economic growth. of the finance and growth relationship. Journal of Journal of development economics 72.1. Journal of Development economics 74.2, 429-447. development economics 72.1, 312-334. Sahay, M. R. (2015). Rethinking Financial Deepening: Cecchetti, S. G. and E. Kharroubi (2015). Why does finan- Stability and Growth in Emerging Markets. IMF. No. cial sector growth crowd out real economic growth? 15-18. Bank of International Settlements. Secured Transactions Advisory Project in China. IFC Ad- Christopoulos, D. K. (2004). Financial development and visory Services. International Finance Corporation. economic growth: evidence from panel unit root and 2012 cointegration tests. Journal of development Eco- Schiantarelli, F. a. (1997). Debt maturity and firm perfor- nomics 73.1, 55-74. mance: A panel study of Indian public limited compa- Demirgüç-Kunt, A. a. (1999). Institutions, financial mar- nies. World Bank, Policy Research Dept., Finance and kets, and firm debt maturity. Journal of financial Private Sector Development Division (1724). economics 54.3, 295-336. Schiantarelli, F. a. (1997). The Maturity Structure of Debt: Demirgüç-Kunt, Kane, & Laeven (2014). Deposit Insur- Determinants and Effects on Firms’ Performance? ance Database, World Bank Policy Research Work- Evidence from the United Kingdom and Italy. World ing Paper. Bank Policy Research Working Paper (1699). Frisby, Sandra (2007). A preliminary analysis of Shen, C.-H. a.-C. (2006). Same financial development pre-packaged administrations. Report to The Asso- yet different economic growth: why?. Journal of ciation of Business Recovery Professionals Money, Credit and Banking, 1907-1944. Gamboa-Cavazos, Mario and Schneider, Frank (2007). World Bank (2015). Principles for Public Credit Guar- Bankruptcy as a legal process. Available at SSRN antees schemes for SMEs. Washington D.C. : World 979614 Bank Group. Gine, Xavier and Love, Inessa. (2010). Do Reorganization World Bank (2016). Public Asset Management Compa- Costs Matter for Efficiency? Evidence from a Bank- nies: A Toolkit. World Bank Publications. Taming Volatility Chapter 4. The Effects of Volatility, Fiscal Policy Cyclicality, and Financial Development on Growth Introduction model of the business cycle in the OECS to simulate the impacts of fiscal policy pro-cyclicality and financial T his chapter assesses empirically the impacts frictions on growth and other relevant macroeconomic of terms of trade volatility, fiscal policy (pro) aggregates for the region. Section 3 summarizes our re- cyclicality, and financial development on sults and discusses some directions for policy. growth in the Eastern Caribbean. By considering the effects of the interaction between terms of trade vol- Empirical Approach atility and fiscal pro-cyclicality we intend to gauge (i) Our results are derived from a simple econometric mod- 73 whether fiscal pro-cyclicality exacerbates the negative growth effect of terms of trade volatility; and (ii) wheth- el that represents the behavior of our variables of inter- er the mediating role of fiscal policy cyclicality is par- est on growth. Our working hypotheses are that: (i) an ticularly pronounced in the OECS as compared to other increase in terms of trade volatility results in lower GDP countries. This is an important policy discussion for a growth; (ii) the negative impact of volatility on growth is region that adopts a strong peg against the U.S. dollar more pronounced in less financially developed countries; since this arrangement limits considerably their ability and (iii) in countries where fiscal policy is more pro-cy- to respond to shocks. clical. The econometric model for testing hypotheses (i)- (iii) is: The main message from the analysis is that a more counter-cyclical fiscal policy stance and further fi- Growthit = ai + bt + αVolatilityit + β(Volatilityit*FDi) + nancial development can mitigate the adverse growth γ(Volatilityit*Procyclicalityi) + φlnGDPit-1 + eit effects of terms of trade volatility in the OECS re- gion. We were able to arrive at this conclusion by using where Growth is the change in the natural logarithm of complementary empirical modeling strategies. First, in real GDP per capita in country i between period t and t-1. section 2, we follow Brueckner and Carneiro (2015) and Volatility is the country-specific standard deviation of estimate an econometric model using panel data on external economic. Following Rodrik (1998), we will use GDP per capita, financial development, and government the standard deviation of countries’ terms of trade as spending for a sample of 175 countries over the period the measure of externally induced economic volatility. 1980-2010. These data are available from the Penn FD is a measure of financial development. Following the World Tables and the World Development Indicators. finance literature, our main measure of financial devel- With this modeling strategy, we were able to benchmark opment is the GDP share of domestic credit to the pri- the behavior of OECS economies against other countries vate sector. Further, the chapter will present estimates in the world. Second, in section 3, following Carneiro and where FD is instrumented by legal origin.30 Hnakovska (2016), we used the results of a structural 30  The law and finance literature, see, for example, La Porta Taming Volatility In our empirical approach, Procyclicality is a measure tries are part of the Organization of Eastern Caribbean of the extent to which government spending responds States. Note that α2, β2, and γ2 capture the differences to business-cycle variation in GDP. Formally Git=θiG- in effects that the variables of interest have on econom- DPcyclit. The parameter θ captures the pro-cyclicality ic growth in OECS countries. This can be seen from dif- of government spending. That is, if θ is positive, then ferentiating equation (3) with respect to volatility: government spending increases when the cyclical com- ponent of GDP is positive (meaning, when GDP is above (4) d(Growth)/d(Volatility) = α1 + β1*FDi trend). We obtain data on θ from Frankel et al. (2013). + γ1*Procyclicalityi +(α2 + β2*FDi + Frankel et al. (2013) provide values of θ for a sample of γ2*Procyclicalityi)*OECSi 93 countries; we will use this data in the estimation of equation (1). In order to maximize coverage of countries (5) d(d(Growth)/d(Volatility))/dOECSi= α2 + β2*FDi + when estimating equation (1), we apply the methodology γ2*Procyclicalityi of Frankel et al. and Carneiro and Garrido (2015) to gen- erate values of θ for countries not covered by Frankel et Table 4.1 provides a list of the variables used in the al. (2013).31 econometric analysis and their data sources. Table 4.2 shows bi-variate correlations. Summary statistics In order to estimate the model and test our working of the variables’ first and second moments can be found hypotheses, we need to modify the original equation in Table 4.3. slightly. Differentiating equation (1) with respect to vol- atility yields: Graphical Analysis (2) d(Growth)/d(Volatility) = α + β*FDi + A graphical examination of the data shows a negative γ*Procyclicalityi relationship between volatility and economic growth. Figure 4.1 plots on the y-axis countries’ average GDP per In practical terms, this modification implies that for capita growth (over five years); on the x-axis is the stan- 74 hypotheses (i)-(iii) to hold then the three coefficients in dard deviation of the terms of trade growth rate (also equation (2) will be negative, or α<0, β<0, γ<0. computed over a five year period). We see from Panel A in Figure 4.1 that for a sample of 175 countries terms In addition, in order to examine whether the coefficients of trade volatility has a negative average effect on eco- α, β, and γ differ for the Eastern Caribbean, we need to nomic growth. The coefficient from a bivariate regres- augment the econometric model: sion that corresponds to the plot in Panel A of Figure 4.1 is -0.31; this coefficient is significant at the 1 per- (3) Growthit = ai + bt + α1Volatilityit + cent level (p-value 0.007). Panel B shows that a nega- β1(Volatilityit*FDi) + γ1(Volatilityit*Procyclicalityi) tive relationship between terms of trade volatility and + (α2Volatilityit + β2(Volatilityit*FDi) + economic growth is visible also within the sub-sample γ2(Volatilityit*Procyclicalityi))*OECSi + φ1lnGDPit-1 + of OECS countries. The coefficient from a bivariate re- e1it gression that corresponds to the plot in Panel B of Fig- ure 4.1 is -0.36; thus it is quantitatively very close to the where OECS is an indicator variable that is unity if coun- slope coefficient that emerges in Panel A. Statistically, we cannot reject the hypothesis that the slope coeffi- et al. (1998), has documented that countries’ legal origin is an cient in Panel A is equal to the slope coefficient in Panel important determinant of cross-country differences in financial development. La Porta et al. argued that common law ensures a B (t-value 0.91). greater protection of private property (relative to civil law). Pro- tection of private property is paramount for investment and the development of the financial sector. La Porta et al. document for Regression Results a cross-section of countries that financial development is signifi- cantly higher in countries with common law. 31  The econometric model includes country fixed effects, ai. The econometric results suggest that countries with These fixed effects capture time-invariant country characteristics lower terms of trade volatility tend to have systemat- that affect economic growth, i.e. variables related to history and ically higher GDP per capita growth. Table 4.1 reports geography. Note that the country fixed effects account for the lin- ear effect that FDi and Procyclicality have on economic growth. estimates of the average effect that terms of trade vol- The time fixed effects, bt, capture common shocks that affect atility has on economic growth, based on a static panel countries’ GDP per capita growth in a given time period. Taming Volatility Fi ur 4.1 Economic Growth nd Vol tilit (Cross-Countr R l tionship) P n l A: All Countri s P n l B: OECS Onl BIH .22 AZE .4 ARM BLR GNQ LVAGEOCHN KNA SGP MAC KOR BWA KAZ GRD .2 HKG QAT LTU KNA MLT MDV ROM VNM THAMYSGRD JPN .2 GRC PRT CYP ESP BGR IRL LAO ESTDMA LKA IDN TTO LUXATG AUT FIN ITA UKR PAN ISR EGY VCTBTN SYC DOM INDCPV LCA NOR TKM OMN growth BEL FRASVN DNK GBR POL TUN SWZ NLDMUS TON TUR ISL MAR PAK BRA MRT LSOCHL CUB IRN KWT SVK BLZ .18 HUNSWE ALB CZE USA CAN TZA AUS MEX CRI SYR PHLMOZCOL ECU MDA CHE KGZ NPL BRB NZL SDN JOR KHM URY FSM MLI MNG PER ARG MWI JAM BMU FJI PRY SLVETHTCD LBY GTM GAB SAU COG AGO MKD MHL HRV BHSGUY SURTJK SLE BGD BFA NAM UZB HNDBENGHA UGASLBPNG RUS IRQVEN DZA PLW KEN BOL ARE CIV COM CMR BDI NGARWA STP LBNGMB BHR GIN NIC TGO GNB AFG SEN HTI BRN 0 MDG NER ATG DMA .16 SOM KIR ERI CAF DJI LBR VCT LCA .14 −.2 0 .05 .1 .15 .2 .25 .02 .04 .06 .08 .1 .12 volatility volatility Table 4.1: Economic Growth and Volatility (Average Effect, Static Model) GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.44*** -0.34*** -0.10 -0.01 -0.05 (0.08) (0.08) (0.10) (0.09) (0.09) Terms of Trade Growth 0.11*** (0.03) 75 R-Squared 0.06 0.15 0.06 0.12 0.15 Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Table 4.2: Economic Growth and Volatility (Average Effect, Dynamic Model) GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS (6) SYS-GMM Terms of Trade Volatility -0.39*** -0.33*** -0.10 0.04 0.00 -0.02 (0.08) (0.09) (0.09) (0.09) (0.09) (0.07) Terms of Trade Growth 0.10*** 0.08*** (0.03) (0.02) R-Squared 0.01** 0.00 -0.05 -0.15*** -0.15*** -0.17 Lagged GDP per capita (0.01) (0.01) (0.04) (0.05) (0.05) (0.11) Country Fixed Effects No No Yes Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Taming Volatility data model. Column (1) reports unconditional estimates; economic growth is significantly different for the OECS column (2) adds to the regression time fixed effects; col- region. This is true regardless of whether or not country umn (3) includes in the econometric model country fixed fixed effects are included in the econometric model.32 effects; and column (4) shows estimates based on an econometric model that includes both time and country Additional estimates show that the negative effect fixed effects. Model specifications that do not include of terms of trade volatility on growth is mediated by country fixed effects, i.e. columns (1) and (2), show a sig- cross-country differences in financial development. nificant negative average relationship between terms of Table 4.4 reports estimates from an econometric mod- trade volatility and economic growth. The coefficient of el that includes an interaction term between terms of -0.34 suggests that the cross-country relationship be- trade volatility and the GDP share of domestic credit to tween terms of trade volatility and economic growth is the private sector. Following the finance literature (La sizable. For example, the model predicts that econom- Porta et al., 1998), we use variation in the GDP share of ic growth is lower by around 0.6 percentage points per domestic credit to the private sector that is predicted annum when taking a country from the 25th percentile by British legal origin. As an additional variable we use (0.034) of terms of trade volatility to the 75th percentile distance to the equator in order to generate variation (0.125). in the GDP share of domestic credit that is exogenous to economic growth.33 From columns (1) and (2) of Table The average negative relationship between econom- 4.4 we see that the coefficient on terms trade volatility ic growth and terms of trade volatility holds for dif- is around -0.7 and significant at the 1 percent level; the ferent specifications of the econometric model. Table interaction term between terms of trade volatility and 4.2 shows that a dynamic model yields similar results the GDP share of domestic credit to the private sector is to the static model that was estimated in Table 4.1. The around 1.6 and significant at the 5 percent level. These dynamic panel model includes the lag of GDP per cap- values imply that at median levels of financial develop- ita on the right-hand side of the estimating equation; ment (GDP share of domestic credit to the private sector a negative coefficient on lagged GDP per capita means equal to 35 percent), terms of trade volatility has a sig- 76 that there is a convergence. In columns (1) and (2) we see nificant negative effect on economic growth; however at that the coefficient on lagged GDP per capita is not sig- higher values of financial development the effect loses nificant. Hence, there is no cross-country convergence significance.34 in GDP per capita; this is a well-known result (see, for example, Mankiw et al., 1992). Columns (3) and (4) show Financial development has a positive effect on eco- that there is significant convergence of GDP per capi- nomic growth on average. This can be seen from the ta to country-specific steady states. This can be seen coefficient on the GDP share of domestic credit to the from the negative coefficient on GDP per capita in the private sector. The relevant coefficient is around 0.2 and model specifications that include country fixed effects. is significantly different from zero at the 1 percent level, The estimated coefficient on lagged GDP per capita in see columns (1) and (2). This is in accordance with the model specifications with country fixed effects suggests finance literature where it is found that a greater GDP that the per annum convergence rate to country-specif- share of domestic credit to the private sector is on av- ic steady state is around 3 percent. erage growth enhancing. Sahay et al. (2015) present ev- idence that the growth effect of financial development The results for the OECS region follow the same pat- may be non-linear. Indeed, we see from columns (1) and tern as the results for the rest of the world. In order to 32  Note that in columns (3)-(5) the model does not explicitly in- explore whether the average relationship between terms clude the OECS dummy; this is because the country fixed effects of trade volatility and economic growth is different for fully take into account the average growth of OECS countries. 33  Both British legal origin and distance to the equator have the OECS region the econometric model is augmented a highly significant positive effect on the GDP share of domestic to include an interaction term between terms of trade credit to the private sector. The F-statistic on the joint test that British legal origin and distance to the equator have no significant volatility and a dummy that is unity for countries that effect on the GDP share of domestic credit to the private sector is belong to the OECS. Table 6 reports the estimation re- 34.77; British legal origin and distance to the equator explain about sults from this augmented econometric model. The one-quarter of the variation in the sample of countries’ average GDP shares of domestic credit to the private sector. main finding is that there is no evidence that the aver- 34  The interquartile range of the GDP share of domestic credit to age relationship between terms of trade volatility and the private sector predicted by British legal origin and distance to the equator is [0.25, 0.46]. Taming Volatility Table 4.3: Economic Growth and Volatility (Is the Effect Different in the OECS?) GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.44*** -0.34*** -0.10 -0.01 -0.05 (0.08) (0.08) (0.10) (0.09) (0.09) Terms of Trade Volatility*OECS 0.48 0.34 0.19 -0.05 -0.16 (0.39) (0.32) (0.73) (0.37) (0.42) OECS -0.02 -0.04* (0.03) (0.02) Terms of Trade Growth 0.11*** (0.03) Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Table 4.4: Economic Growth, Financial Development, and Volatility GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility [A] -0.68*** -0.71*** -0.33 -0.30 -0.38* (0.17) (0.16) (0.25) (0.20) (0.19) 77 Terms of Trade Volatility*Credit-to-GDP ratio [B] 1.60** 2.17*** 0.69 1.28 0.89 (0.63) (0.58) (0.96) (0.89) (0.66) Credit-to-GDP ratio 0.26*** 0.23*** (0.06) (0.06) Terms of Trade Growth 0.08*** (0.02) Test [A]=[B]=0, p-value 0.00 0.00 0.16 0.33 0.12 Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. (2) of Appendix Table 1 that the relationship between Terms of trade volatility has a particularly large neg- GDP p.c. growth and the GDP share of domestic credit to ative effect on economic growth in countries where the private sector is an inverted U-shaped. The average fiscal policy is procyclical. This is shown in Tables 4.5 marginal effect of financial development on economic and 4.6. Table 4.5 reports estimates from an economet- growth remains, however, positive and significant. Fur- ric model that includes an interaction term between the ther, we have explored using the index of financial de- standard deviation of the terms of trade growth rate velopment developed in Sahay et al. (2015); this yields and the country-specific coefficients that measure the results similar to those presented in Table 7, see Appen- dix Table 2.35 (ability of individuals to access financial services), and efficiency of financial markets and institutions (ability of institutions to provide 35  The index of financial development developed in Sahay et financial services at low cost and with sustainable revenues, and al. (2015) measures depth (size and liquidity of markets), access the level of activity of capital markets). Taming Volatility response of government spending to the business cycle. between terms of trade volatility is negative in all spec- Table 4.6 reports estimates from an econometric mod- ifications. In Table 4.5, it is significantly different from el that includes in addition to the interaction between zero in specifications that include country fixed effects terms of trade volatility and fiscal procyclicality an in- (columns (3)-(5)). If country fixed effects are excluded, teraction term between terms of trade volatility and fi- see columns (1) and (2), the interaction between terms nancial development. The coefficient on the interaction of trade volatility and fiscal procyclicality is individually Table 4.5: Economic Growth, Fiscal Procyclicality, and Volatility GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility [A] -0.46*** -0.38*** -0.12 -0.02 -0.03 (0.08) (0.08) (0.09) (0.09) (0.09) Terms of Trade Volatility*Fiscal Procyclicality [B] -0.31 -0.32 -0.75** -0.68** -0.66** (0.28) (0.23) (0.30) (0.27) (0.28) Fiscal Procyclicality -0.00 0.01 (0.03) (0.03) Terms of Trade Growth 0.10*** (0.03) Test [A]=[B]=0, p-value 0.00 0.00 0.04 0.04 0.06 Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes 78 Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Table 4.6: Economic Growth, Financial Development, Fiscal Procyclicality, and Volatility GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.46*** -0.37*** -0.38* -0.06 0.05 (0.08) (0.18) (0.21) (0.21) (0.22) Terms of Trade Volatility*Fiscal Procyclicality -0.52* -0.41* -0.70** -0.67** -0.67** (0.24) (0.21) (0.29) (0.27) (0.27) Terms of Trade Volatility*Credit-to-GDP ratio 0.55 0.37 0.98 0.16 -0.30 (0.56) (0.51) (0.65) (0.61) (0.66) Fiscal Procyclicality 0.04 0.01 (0.04) (0.01) Credit-to-GDP ratio 0.27*** 0.20*** (0.05) (0.05) Terms of Trade Growth 0.10*** (0.03) Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Taming Volatility not significantly different from zero; however, the F-test cal policy pro-cyclicality. The estimates reported in rejects the null hypothesis that the interaction term is Table 4.7 speak to the question of whether the medi- jointly equal to zero with the linear effect that terms of ating role that fiscal (counter-)cyclicality and financial trade volatility has on economic growth. In Table 4.6, we development have with regard to the effect of terms of see that the interaction between terms of trade volatili- trade volatility on economic growth differ for the OECS ty and fiscal procyclicality is significantly different from region. Specifically, Table 4.7 reports estimates from zero in all specifications. an econometric model that includes, in addition to the variables in Table 4.6, two further interaction terms: The difference in effect of terms of trade volatility on one interaction term is constructed as the interaction economic growth across differences in countries’ fis- between terms of trade volatility, fiscal procyclicality, cal procylicality is sizable. For example, according to and the OECS indicator; and another interaction term the estimates in column (5) of Table 4.6, for a country that is constructed as the interaction between terms at the 25th percentile of fiscal procyclicality the implied of trade volatility, financial development, and the OECS marginal effect (standard error) of terms of trade vola- indicator. The coefficient on the first (second) inter- tility on economic growth is 0.04 (0.20); for a country action term gives the difference in the mediating role at the 75th percentile the corresponding marginal effect of fiscal procyclicality (financial development) for the (standard error) is -0.36 (0.21). OECS region. Table 4.7 shows that the coefficient on the interaction between terms of trade volatility, fiscal pro- Finally, the OECS region shows a strong negative in- cyclicality, and the OECS indicator is significantly neg- teraction between terms of trade volatility and fis- ative. Further, the interaction between terms of trade Table 4.7. Economic Growth, Financial Development, Fiscal Procyclicality, and Volatility: Heterogeneity OECS GDP per capita Growth Dependent Variable is: 79 (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.46*** -0.37*** -0.38* -0.07 0.03 (0.19) (0.18) (0.21) (0.20) (0.22) Terms of Trade Volatility*Fiscal -0.43* -0.42* -0.70** -0.67** -0.60** Procyclicality (0.25) (0.21) (0.29) (0.27) (0.27) Terms of Trade Volatility*Fiscal -1.95** -2.22*** -72.01* -78.00*** -73.01*** Procyclicality*OECS (0.77) (0.64) (42.04) (15.37) (26.31) Terms of Trade Volatility*Credit-to-GDP 0.55 0.36 0.99 0.17 -0.08 ratio (0.56) (0.51) (0.65) (0.61) (0.64) Terms of Trade Volatility*Credit-to-GDP 1.76** 2.07*** 9.07 15.08*** 12.66* ratio *OECS (0.86) (0.62) (10.56) (4.05) (6.73) Fiscal Procyclicality 0.04 0.04 (0.04) (0.03) Credit-to-GDP ratio 0.27*** 0.20*** (0.06) (0.06) OECS 0.03* 0.01 (0.01) (0.01) Terms of Trade Growth 0.12*** (0.03) Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Taming Volatility volatility and fiscal procyclicality is also negative and In this case we allow for both independent country risk significantly different from zero. and induced country risks, i.e. country risk that arises endogenously in response to changes in fundamentals. These results suggest that: (i) fiscal procyclicality In the second case, we study the effects of the fiscal exacerbates the negative growth effect of terms of policy that is procyclical to the economic conditions. trade volatility – for the OECS region and for other re- Here we are again interested in the possible interaction gions; and (ii) the mediating role of fiscal cyclicality is between fiscal policy and country risk, so we consider particularly pronounced in the OECS region. It is also both independent and induced country risk scenarios. noteworthy that the coefficient on the interaction be- In what follows, we use impulse response analysis to tween terms of trade volatility, financial development, understand the linkages in the model and the effects of and the OECS indicator is significantly positive. The in- various model features. In particular, we study how the teraction between terms of trade volatility and fiscal key macroeconomic aggregates respond to one-time procyclicality is not significantly different from zero. The shocks to productivity, government expenditures and results in Table 4.7 thus suggest that counter-cyclical risk-premium. fiscal policy and financial development can mitigate the adverse growth effects of terms of trade volatility in the We start by considering the effects of government OECS region. spending shocks under a-cyclical fiscal policy. The simulations in Figure 4.1 show that an increase in gov- Impulse Response Analysis ernment expenditures has a contractionary effect on the economy, with all macroeconomic aggregates declining We have also investigated the role played by finan- following the shock. The largest decrease is experienced cial frictions and different fiscal policy stances for by private consumption which falls by 0.05 percent af- the transmission of shocks to economic activity in ter a 1 percent increase in government expenditures. the OECS region by means of a structural model of The responses of output, employment and investment, business cycles. By using a complementary model- while all negative, are more muted. These dynamics can 80 ing approach, we have looked at the impacts of differ- be understood by looking at how firms and households ent types of shocks to fiscal policy in the presence of adjust their borrowing/lending behavior in response to financial frictions on the economic performance of the shocks. When government expenditures increase, it region. The analysis is based on a theoretical structural causes a decline in household’s lifetime income, leading model developed by Carneiro and Hnatkovska (2016) as to a fall in consumption. Since consumption decline is a background paper for this report. The details of the smaller than the fall in household’s disposable income model are omitted here for simplicity, but we highlight due to consumption smoothing, savings must fall as its two key features. The first is that domestic financial households decrease their bond holdings. Firms, faced markets are subject to a friction – firms have to pay a with lower demand, cut down on employment and in- share of the bill for the factors of production before pro- vestment, and thus borrow less from the international duction takes place and revenues are realized. This cre- markets. With lower employment, GDP also declines. ates a need for working capital by firms. The second key Due to increased household borrowing, both trade bal- feature of the model is the presence of a fiscal authority. ance and net foreign assets (NFA) deteriorate. It levies lump-sum taxes and uses the tax revenues to provide public consumption/investment. In the model, The simulations of shocks in productivity show an public expenditures have different cyclical properties. overall improvement in the economy. The impulse re- These two features of the model generate transmission sponses to a 1 percent positive productivity shock un- channels through which real interest rates and fiscal der a-cyclical and pro-cyclical fiscal policy stances are policy shocks affect the level of economic activity. shown in Figures .4.3a and 4.3b. We also include im- pulse responses under the scenario with induced coun- Our objective is to evaluate the expected impact of try risk-premium in the figures. Under the procyclical different fiscal policy stances on macroeconomic ag- fiscal policy, an increase in productivity triggers a rise gregates. We start with the case in which fiscal policy is in government expenditures. Focusing first on the di- a-cyclical. That is, where government expenditures are rect effects of productivity shocks, we see that a rise independent of the fundamental state of the economy. in productivity that is persistent has relatively standard Taming Volatility Fi ur 4.2 Impuls R spons s to 1% Positiv Shock to Gov rnm nt Exp nditur s x 103 1.0 0 -0.5 0.8 -1 0.6 % Dev from s.s. % Dev from s.s. -1.5 0.4 -2 0.2 -2.5 0.0 -3 2 4 6 8 10 12 2 4 6 8 10 12 years years Gov exp Consumption Output Employ Invest 73.0 -60.0914 2 13 72.5 -60.0916 Level Level 1 12 72.0 -60.0918 71.5 -60.0920 0 11 81 2 4 6 8 10 12 2 4 6 8 10 12 years years HH bond holdings Firm borrowing TB NFA Note: Impulse responses are computed under benchmark parameterization. effects – it leads to an expansion in the economy with balance. When the increase in productivity is accom- employment, investment, consumption and output all panied by a rise in government spending (pro-cyclical rising. Higher productivity raises the return to capital fiscal policy), the response of household consumption is and labor, so firms want to increase employment and more muted. This is because an accompanying increase investment. However, to hire more workers, firms must in government expenditures curtails the rise in the dis- finance a larger working capital, so firms’ borrowing posable income of the households after productivity im- goes up. This can be seen in Figure 2b, which displays provement. This limits the resources available for con- impulse responses of agents’ asset holdings/borrowing, sumption and investment. As a result, households must trade balance and NFA position. Since returns to invest- lower their savings by more under the procyclical fiscal ment increase following a positive productivity shock, policy, leading to a larger deterioration in the trade bal- households reduce their savings in international bonds ance and NFA in the economy. Thus, procyclical fiscal and invest more in domestic enterprises. The outcome policy acts to curtail the effects of productivity shocks of these adjustments is that the trade balance worsens on consumption, by amplifying the effects of these and NFA position declines. shocks on savings, net exports and NFA. Under a procyclical fiscal policy, the adjustments of The magnitude of the impact of a positive produc- the economy in response to a positive productivity tivity shock in the economy depends on the follow on shock are very similar. The key difference lies in the dy- impacts on the country risk premium. Next, we turn namics of consumption, household savings, and trade to the effects of productivity shocks when they also de- Taming Volatility Fi ur 4.3 Impuls R spons s ft r 1% Positiv Productivit Shock: M cro A r t s Productivit Gov xp Output 1.0 0.30 1.5 0.8 0.25 0.20 % Dev from s.s. % Dev from s.s. % Dev from s.s. 0.6 1.0 0.15 0.4 0.10 0.5 0.2 0.05 0.0 0.00 0.0 2 4 6 8 10 12 2 4 6 8 10 12 2 4 6 8 10 12 years years years Consumption L bor Inv stm nt 1.0 8 1.5 0.8 6 % Dev from s.s. % Dev from s.s. % Dev from s.s. 1.0 0.6 4 0.4 0.5 2 0.2 82 0.0 0.0 0 2 4 6 8 10 12 2 4 6 8 10 12 2 4 6 8 10 12 years years years Benchmark Procyc Induced Note: Impulse responses are computed under benchmark parameterization. termine the risk-premium in the economy (i.e., induced larger with induced risk-premium. Overall, in the pres- risk premium). The responses of various variables to a ence of endogenous country risk-premium, the effects productivity shock in such a case are shown as green of productivity shocks on the economy are amplified. dash-dot lines in Figures 4.3a and 4.3b. Under this sce- nario, an increase in productivity has the same effects Lastly, the effects of a shock to domestic interest as described above, except an increase in productivity rate arising as a consequence of a shock to the inter- also triggers a fall in country risk-premium, which in national interest rate tends to have a contractionary turn provides an additional boost to the economy. In- effect on the economy. Figure 4.4 presents the respons- deed, with induced risk-premium, all macroeconomic es of key variables. A rise in the international interest aggregates experience a greater expansion relative to rate triggers an increase in the domestic interest rate the scenario with independent risk-premium. This oc- which raises the cost of borrowing for working capital curs because lower risk-premium reduces the interest for domestic firms. Therefore, they reduce borrowing, rate faced by domestic agents, encouraging additional cut employment, which in turn lowers output. Consump- borrowing by firms and a greater reduction in savings tion also declines and this fall exceeds the drop in out- (bond holdings) by households. As a result, both employ- put. This is an important result of the model as it shows ment and investment are scaled up significantly. Not that fluctuations in the interest rate can help account surprisingly, the deterioration in the NFA position and for the high volatility of consumption in the OECS coun- trade balance (in fact, trade balance goes into deficit) is tries. The increase in interest rates also induces higher Taming Volatility Fi ur 4.3b Impuls R spons s ft r 1% Positiv Productivit Shock: Fin nci l V ri bl s HH bond holdin s Firm borrowin -60.1 73 -60.2 -60.3 72 -60.4 Level Level 71 -60.5 -60.6 70 -60.7 2 4 6 8 10 12 2 4 6 8 10 12 years years Tr d b l nc NFA 1.5 13 1 12 0.5 Level Level 11 0 10 -0.5 9 83 2 4 6 8 10 12 2 4 6 8 10 12 years years Benchmark Procyc Induced Note: Impulse responses are computed under benchmark parameterization. savings by domestic households, whose bond holdings productivity innovations, , is set so that the volatility of rise; and discourages investment. As a result, trade bal- GDP in the model matches the average volatility of GDP ance improves and so does the NFA. in the OECS countries. We now turn to the evaluation of the contribution of In all experiments we simulate the model economy for various shocks, financial frictions and fiscal policy to a random sequence of shocks to productivity, govern- the business cycles of the OECS countries by means of ment expenditures, international interest rate and several numerical experiments. In our first experiment country risk-premium. We then obtain the volatilities we assume that both fiscal policy and country risk-pre- and comovements among the key aggregates from this mium are a-cyclical, or independent of the state of the simulated data and contrast them with the actual data economy. This version is the benchmark case. In the sec- and across various versions of the model. When simulat- ond experiment we consider the case of independent fis- ing the model, we treat model series in exactly the same cal policy, but assume that country risk-premium is en- way as the data. In particular, we simulate 45 years of dogenous to productivity. The third experiment assumes data and remove the first 10 years to reduce the effects pro-cyclical fiscal policy and independent risk-premium. of initial conditions. This gives us model series of the The last, experiment studies the case where both fiscal same length as in the data. All series, except interest policy and country risk-premium respond to produc- rate and trade balance are log-transformed and HP-fil- tivity changes. In each case, the standard deviation of tered with a smoothing parameter of 100. Volatility Taming Volatility Fi ur 4.4 Impuls R spons s ft r 1% Positiv Shock to Int rn tion l Int r st R t 2 -0.2 0 -0.4 % Dev from s.s. % Dev from s.s. -2 -0.6 -4 -0.8 -1 -6 -1.2 0 -1.4 2 4 6 8 10 12 2 4 6 8 10 12 years years Int rate Investment Output Employ Invest 76 -59.4 4 16 75 -59.6 3 15 Level Level 74 -59.8 2 14 73 -60 1 13 84 72 -60.2 0 12 2 4 6 8 10 12 2 4 6 8 10 12 years years HH bond holdings Firm borrowing TB NFA Note: Impulse responses are computed under benchmark parameterization. and comovement statistics are then computed on each reason for this lower volatility is the counteracting ef- model series and averaged across 1000 simulations. fect that government expenditures have to productivity The benchmark model replicates the volatilities of the changes under the procyclical fiscal policy (see impulse macro variables in the OECS data quite closely. As it responses in Figure 4.2a). can be seen from panel 1.a in Table 4.3, the model match- es the volatilities of GDP, interest rate, investment, and Next we consider a scenario in which country risk-pre- government spending because these moments were mium is endogenous to productivity changes. This is targeted in the calibration. But we did not target the shown in Panel 2 in Table 4.3 under the label of “Induced volatility of consumption and trade balance: while the country risk”. Here, we calibrated the standard deviation model comes very close to replicating the volatility of TB of innovations to all shocks such that we replicate the to GDP ratio it under predicts the volatility of consump- volatility of GDP, interest rate, and government spend- tion in the OECS countries. At the same time, it is im- ing. All other parameters are set to their baseline val- portant to note that the model yields consumption that ues. We find that with induced risk-premium, the vola- is more volatile than GDP, in line with the data facts for tility of all non-targeted variables increases relative to the OECS countries (as seen in stylized facts discussed the benchmark scenario. For instance, the percentage in Chapter 1 of this report). Under a pro-cyclical fiscal standard deviation of investment goes up from 4.57 in policy, the model with all shocks (panel 3.a) predicts the benchmark model with independent risk premium to lower volatility of trade balance and consumption. The 7.74 under induced risk premium, rising by 70 percent. Taming Volatility Table 4.8: Simulated and Actual Business Cycles in the OECS Countries: Volatilities % Standard deviation % Standard deviation of x % Standard deviation of GDP GDP TB/GDP Int rate Inv Gov exp Cons Employment OECS data 3.72 5.21 2.69 4.57 2.55 2.66 n/a Independent fiscal policy 1. Independent country risk (a) all shocks 3.72 4.79 2.69 4.57 2.55 1.40 0.89 (b) no z shocks 1.81 4.46 2.69 7.09 5.25 2.03 1.44 (c) no G shocks 3.72 3.89 2.69 4.57 0.00 1.39 0.89 (d) no R*, D shocks 3.20 3.17 0.00 3.20 2.96 1.11 0.62 2. Induced country risk 3.72 6.55 2.69 7.74 2.55 1.41 0.94 (d) no R*, D shocks 1.91 2.89 0.00 3.16 4.97 1.11 0.62 Procyclical fiscal policy 3. Independent country risk (a) all shocks 3.72 4.68 2.69 4.57 2.55 1.31 0.89 (b) no z shocks 1.81 4.58 2.69 7.09 5.61 2.03 1.44 (c) no G shocks 3.72 3.89 2.69 4.57 1.02 1.31 0.89 (d) no R*, D shocks 3.20 2.99 0.00 3.20 2.96 0.97 0.62 4. Induced country risk 3.72 6.49 2.69 7.74 2.65 1.35 0.94 (d) no R*, D shocks 1.91 2.89 0.00 3.16 5.15 0.95 0.62 85 Source: Authors’ calculations. Similarly, employment volatility goes up by 5.2 percent, and government expenditures is close to zero both in the while the volatility of net exports rises by 37 percent. data and in the model. This is because the effects of productivity shocks on the economy are amplified in the presence of endogenous Several important results from this alternative mod- risk-premium due to the presence of working capital eling strategy can be emphasized. First, counter cy- constraint.36 clical fiscal policy curtails the effects of productivity shocks in the economy, and reduces the volatility of Lastly, we present the cyclical properties of key vari- consumption and net exports. Second, if risk-premium ables with the real interest rate. An important finding responds to the changes in the economy’s fundamen- in the OECS counties data was that real interest rates tals, the effects of productivity shocks on the economy are strongly countercyclical with main macroeconomic are amplified through the working capital channel, and aggregates. Table 4.5 shows that the model reproduces therefore the volatility of key macro aggregates rises. this result quite closely. In particular, in the benchmark This amplification effect would disappear if there was no scenario (panel 1.a), real interest rate commoves nega- spillover from productivity to country risk-premium or if tively with investment, consumption and employment. there was no need for working capital. The correlation with net exports is positive. The signs of the correlations remain unchanged both under a procy- We have also isolated the contribution of various clical fiscal policy and with induced country-risk scenar- shocks to the overall volatility in the economy re-com- io. Also note that the correlation between interest rates puted volatilities and correlations in the model while sequentially eliminating shocks to productivity, 36  The increase in volatility due to endogenous risk-premium is government expenditures, and interest rate shocks similar under the pro-cyclical fiscal policy, with the exception that (international interest rate and country risk-premi- consumption volatility is affected more. Taming Volatility um). The results are presented in panels b, c, and d of Given the importance of productivity shocks in the Tables 4.3, 4.4, 4.5. Without productivity shocks (pan- OECS countries, it becomes necessary to better iden- els b of Table 4.3; 4.4; 4.5), volatilities of several vari- tify the sources of these shocks. As we argued before, ables are significantly reduced relative to the bench- the productivity shocks in the model find a broad corre- mark case with all shocks. For instance, GDP volatility spondence to shocks in the data – these include tech- declines from 3.72 percent when all shocks are present) nology shocks, shocks to the terms of trade, unexpected to 1.81 percent when productivity shocks are switched changes in weather conditions, etc. Panels c) of Tables off, implying that productivity shocks account for about 4.3, 4.4, 4.5, report the business cycles statistics from 51 percent of GDP volatility in the Eastern Caribbean simulations where shocks to government expenditures economies. Volatility of net exports also declines when are eliminated. Without government spending shocks, productivity shocks are eliminated from the simula- the volatility of trade balance is reduced by 20 percent tions, although the decline is more muted. Specifical- (from 4.79 percent to 3.89 percent) and the volatility ly, productivity shocks account for about 7 percent of of consumption also decreases. Lastly, we simulate the volatility in next exports. In contrast, the volatilities of model without interest rate shocks (both to the inter- investment, consumption, employment, all rise when national interest rate and country risk-premium) and productivity shocks are eliminated. Similarly, the neg- report the resulting statistics in panels d) of Tables 4.3, ative correlation between trade balance and GDP, and 4.4, 4.5. Eliminating shocks to the interest rate lowers interest rate and GDP become exaggerated without the the volatility of all variables, except government spend- productivity shocks. Taken together, these results sug- ing. For instance, the percentage standard deviation gest that these shocks are important in explaining the of GDP declines from 3.72 percent to 3.20 percent — a business cycles in the OECS economies. 14 percent reduction; consumption volatility declines by 21 percent, while employment and investment volatili- 86 Table 4.9: Simulated and Actual Business Cycles in the OECS Countries: Comovement with Interest Rate Correlation of Interest rate with Inv Gov exp Cons TB/GDP Employment OECS data -0.29 -0.04 -0.33 0.24 -0.36 Independent fiscal policy 1. Independent country risk (a) all shocks -0.68 -0.01 -0.70 0.61 -0.77 (b) no z shocks -0.87 -0.01 -0.97 0.65 -0.97 (c) no G shocks -0.68 0.00 -0.70 0.75 -0.77 (d) no R*, D shocks -0.01 0.00 0.00 0.00 0.00 2. Induced country risk -0.41 -0.01 -0.81 0.36 -0.84 (d) no R*, D shocks -0.02 0.01 -0.02 0.00 -0.02 Procyclical fiscal policy 3. Independent country risk (a) all shocks -0.68 -0.01 -0.74 0.63 -0.77 (b) no z shocks -0.87 -0.01 -0.97 0.64 -0.97 (c) no G shocks -0.68 -0.01 -0.74 0.75 -0.77 (d) no R*, D shocks -0.01 0.00 0.00 0.00 0.00 4. Induced country risk -0.41 -0.03 -0.83 0.37 -0.84 (d) no R*, D shocks -0.01 0.0 -0.00 0.00 -0.00 Source: Authors’ calculations. Taming Volatility ty each decline by about 30 percent, and the volatility available to the agents. As a result, these shocks make of trade balance drops by 34 percent. We should note OECS economies more volatile. that the majority of this decline in volatilities in account- ed for by the absence of risk-premium shocks. For in- We have assessed how the business cycle is affected stance, eliminating just the default risk (and allowing by shocks to the financial sector channel. We did this for the shocks to the international interest rate) reduces by simulating the effects of a smaller value for the pa- the volatility of GDP by 8 percent, the volatility of con- rameter that determines the size of the working capital sumption by 15 percent and employment volatility by requirement – parameter φ. Table 4.6 presents the re- 20 percent. sults for volatilities and correlations in the case where only 50% of the labor cost has to be paid in advance In the absence of interest rate shocks, the cyclical (φ=0.5); and the case in which no labor cost has to be properties of the key variables in the economy are paid in advance (φ=0). In both cases we keep all other also affected. That is, the co-movement between GDP parameters unchanged at their benchmark values given and consumption, investment and employment all in- in Table 5. To quantify how the business cycle properties crease above their data counterparts; the trade balance change with φ in Table 4.6 we report the volatility of the becomes less countercyclical than in the data; and the key aggregates relative to their respective volatilities negative correlation between the interest rate and GDP, under the benchmark model scenario with φ=1. consumption, and investment disappears. We interpret this result as supportive of the importance of interest In a scenario with independent fiscal policy and a rate shocks in explaining the OECS business cycles. shock to interest rates, the volatility of GDP increas- es. In panel 1 of Table 4.6, when φ=1 the model generated The changes in volatilities and correlations are even volatility of GDP and investment that was very similar more pronounced when we consider the scenario with to that found in the OECS data. Similarly, it was able induced country risk-premium. This case is summa- to replicate the negative comovement between GDP rized in panels 2.d) and 4.d) of Tables 4.3, 4.4, 4.5. Elimi- and trade balance, and between GDP and the interest nating interest rate shocks when risk-premium is deter- rate. When the working capital parameter φ is reduced 87 mined by domestic fundamentals leads to a 49 percent to 0, the volatility of GDP, consumption, employment reduction in GDP volatility. Again, the majority of this and government spending, all decline significantly. For reduction is accounted for by eliminating the shocks to instance, GDP volatility is reduced by 14 percent, while risk-premium. Similarly, without interest rate shocks, that of consumption by 24 percent, and employment – consumption volatility in the OECS countries would de- by 30 percent. Similarly, the correlation coefficient be- cline by 21 percent, employment volatility by 33 percent tween output and interest rate turns from being large and investment volatility by 59 percent. and negative to being positive. The same is true for the correlation between consumption and interest rate. The volatility of GDP declines when either produc- Trade balance also becomes much less countercyclical, tivity or interest rate shocks are eliminated. In con- but its volatility rises (by around 20 percent relative to trast, the volatilities of consumption, investment and the benchmark case with φ=1). This is primarily driven employment rise when productivity shocks are absent, by higher volatility of investment. while those volatilities decline in the scenario without interest rate shocks. The differences in the behavior In the absence of financial sector shocks, it becomes of consumption, investment and employment in the easier for firms to adjust to productivity shocks. two scenarios can be understood through the lens of These changes occur because without the working cap- agents’ risk-sharing opportunities in the model. Con- ital requirement, the negative impact of interest rates sider first a productivity shock. In response to such on labor demand of firms is eliminated. As a result, it a shock, households and firms can borrow/lend in in- becomes easier for firms to adjust employment and in- ternational markets at a given interest rate, which al- vestment in response to productivity shocks. This leads lows them to smooth out the effects of the shock. The to higher volatility of employment, investment and, shocks to interest rate are harder to smooth out since therefore, trade balance. At the same time, firms’ em- they directly affect the cost of borrowing for working ployment decisions become less sensitive to interest capital, and no other mechanisms for risk-sharing are rate shocks when φ=0, which lowers the volatility of Taming Volatility Table 4.10: Sensitivity Analysis: Working Capital Requirement % Standard deviation of x Correlation between % Standard deviation of x, benchmark GDP TB/ Inv Gov Cons Employ GDP,R GDP,TB/ Cons,R Inv,R GDP exp GDP 1. Independent fiscal policy φ=0.5 0.90 1.05 1.36 0.62 0.84 0.80 -0.20 -0.21 -0.44 -0.69 φ=0 0.86 1.19 1.70 0.36 0.76 0.70 0.06 -0.05 0.05 -0.70 2. Procyclical fiscal policy φ=0.5 0.90 1.07 1.36 0.62 0.84 0.80 -0.20 -0.27 -0.46 -0.69 φ=0 0.86 1.21 1.70 0.36 0.78 0.70 0.06 -0.07 0.05 -0.70 Note: The table reports the volatility of various aggregates relative to the volatility of the corresponding aggregate under the benchmark parameterization of φ=1. All other parameters are set to their benchmark values given in Table 5. Source: Authors’ calculations. employment in the model. The second effect dominates of an independent fiscal council that monitors macro- and employment volatility is reduced with lower φ. Vola- economic projections underlying the budgeting process tility of investment, however, unambiguously rises. and the compliance with the fiscal rule (see Amo-Yartey et al., 2012). Although it is not an official fiscal council, Conclusions and Directions for Policy the member countries of the ECCU have fiscal targets, agreed under the ECCU Eight Point Stabilization and This chapter has assessed the effects of terms of Growth Program in 2009. The problem is, however, that trade volatility on real GDP per capita growth in the these targets are not binding. 88 OECS and other regions of the world. A key conclusion from the analysis is that counter-cyclical fiscal policy We have also shown that macroeconomic aggregates and stable and well developed financial markets and in- in OECS are quite volatile, with consumption exhib- stitutions will have particularly high payoffs in terms of iting higher volatility than GDP. We also find that in reducing the adverse growth effects of terms of trade these economies real interest rates are very volatile volatility in the OECS region. As discussed in greater de- and strongly countercyclical with GDP and other mac- tail in chapter 2, one way of strengthening the region’s roeconomic aggregates. Similarly, fiscal expenditures ability to shift toward a more counter-cyclical fiscal also show significant volatility, but are pro-cyclical with policy stance would be through the adoption of fiscal GDP. In that context, our simulations suggest that elim- responsibility laws (FRLs) and/or fiscal rules. These are inating fiscal policy shocks could reduce the volatility of widely recognized as effective mechanisms that can consumption and trade balance, but without the volatil- increase the discipline and credibility of the fiscal au- ity of GDP. Eliminating shocks to interest rates, however, thorities. Not only would fiscal rules help in making fis- could reduce the volatility of GDP by 14 percent, and the cal policy less pro-cyclical in the OECS, but they would volatility of consumption by 21 percent. We also show also help the countries in the region to make significant that domestic financial markets development plays an progress in reigning in fiscal expenditures and imple- important role in buffering the effects of interest rate menting effective fiscal consolidation programs. The shocks on the economy. Eliminating the working capital introduction of fiscal rules would need to be supported constraint, while keeping all shocks in place, for exam- by expenditure reforms in the context of a medium term ple, could reduce the volatility of GDP, consumption, em- fiscal framework to signal the authorities’ commitment ployment and government spending significantly. For to fiscal sustainability. Given that natural disasters instance, GDP volatility would be reduced by 14 percent, are common across the region, OECS countries would while that of consumption would decline by 24 percent. do well to integrate the likelihood of a disaster in their fiscal programming exercises. Many countries in simi- These results suggest a few directions for designing lar situations have benefitted from the parallel creation policies to help reduce the volatility experienced by Taming Volatility the OECS economies. First, greater openness to inter- of interest rate shocks on domestic economic activity, national financial markets is important as it could help and achieve lower volatility. Third, if pro-cyclical fiscal the OECS economies to hedge fluctuations in funda- policies induce higher country risk-premium in the inter- mental shocks, such as shocks to technology, terms national markets, governments of the OECS countries of trade, and shocks associated to natural hazards. can stabilize their country’s risk-premium by switch- Second, greater openness must be accompanied by ing to counter-cyclical policies. Fourth, if government improvements in domestic financial markets and gov- consumption is strongly complementary with private ernment’s efforts to stabilize domestic risk-premium. consumption, switching to an independent or counter- By reducing the frictions in the domestic financial mar- cyclical fiscal policy stance can reduce the volatility of kets, these economies can cushion the negative effects consumption in the economy. 89 Taming Volatility APPENDIX Appendix Table 1. Non-Linear Effects of GDP Share of Domestic Credit to Private Sector GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.20 -0.19 -0.88 -0.96** -1.05** (0.17) (0.35) (0.60) (0.45) (0.46) Terms of Trade Volatility*Credit-to-GDP ratio -1.29 -0.82 5.05 6.51** 6.52* (2.74) (2.62) (4.05) (3.22) (3.56) Terms of Trade Volatility*Credit-to-GDP ratio 3.63 3.43 -7.61 -9.15* -9.17* squared (4.45) (4.45) (5.97) (4.77) (5.59) Credit-to-GDP ratio 0.69** 0.79*** (0.28) (0.26) Credit-to-GDP ratio squared -0.59 -0.82** (0.40) (0.40) (0.06) (0.36) Terms of Trade Growth 0.08*** (0.02) Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes 90 Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Appendix Table 2. Financial Development Index from Sahay et al. (2015) GDP per capita Growth Dependent Variable is: (1) LS (2) LS (3) LS (4) LS (5) LS Terms of Trade Volatility -0.70*** -0.74*** -0.32 -0.30 -0.21 (0.19) (0.17) (0.27) (0.22) (0.23) Terms of Trade Volatility*FD Index 2.42** 3.29*** 0.93 1.81 1.07 (0.98) (0.93) (1.44) (1.34) (1.31) FD Index 0.42*** 0.36*** (0.09) (0.26) Terms of Trade Growth 0.10*** (0.03) Country Fixed Effects No No Yes Yes Yes Time Fixed Effects No Yes No Yes Yes Note: The method of estimation is least squares. Huber robust standard errors (shown in parentheses) are clustered at the country level. *Significantly different from zero at the 10 percent significance level, ** 5 percent significance level, *** 1 percent significance level. Taming Volatility References Amo-Yartey, C., M Narita, G.P. Nicholls, J.C. Okwuokei, A. Frankel, J., C. Vegh, and G. Vuletin 2013. “On Graduation Peter, and T. Turner-Jones. 2012. “The Challenges of from Fiscal Procyclicality.” Journal of Development Fiscal Consolidation”, IMF Working Paper WP/12/276, Economics 100: 32-47. International Monetary Fund, Washington, D.C. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vish- Brueckner, M. and F. Carneiro (2015). The Effects of ny (1998). “Law and Finance.” Journal of Political Econ- Volatility, Fiscal Policy Cyclicality and Financial De- omy 106: 1113–1155. velopment on Growth: Evidence for the Eastern Ca- Mankiw, G., D. Romer, and D. Weil (1992). “A Contribution ribbean. World Bank Policy Research Working Paper to the Empirics of Economic Growth.” Quarterly Jour- 7507. nal of Economics 107: 407-437. Carneiro, F., L. Garrido (2015). New Evidence on the Cy- Rodrik, D. (1998). “Why Do More Open Economies Have clicality of Fiscal Policy. World Bank Policy Research Bigger Governments.” Journal of Political Economy Working Paper 7293. 106: 997-1032. Carneiro, F. and V. Hnatkovska. 2016. Business Cycles Sahay, R., M. Čihák, P. N’Diaye, A. Barajas, R. Bi, D. Ayala, in the Eastern Caribbean Economies: The Role of Y. Gao, A. Kyobe, L. Nguyen, C. Saborowski, K. Sviry- Fiscal Policy and Interest Rates. World Bank Policy dzenka, and Seyed Reza Yousefi. 2015. “Rethinking Research Working Paper 7545. Financial Deepening: Stability and Growth in Emerg- ing Markets.” IMF Discussion Note SD 15/08. 91