主权债务数据指南.pdf

WP/19/195 A Guide to Sovereign Debt Data by S. Ali Abbas and Kenneth Rogoff, with Chengyu Huang and Kunxiang Diao IMF Working Papers describe research in progress by the authors and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the authors and do not necessarily represent the views of the IMF, its cutive Board, or IMF management. 2 IMF Working Paper Strategy, Policy, and Review Department A Guide to Sovereign Debt Data 1Prepared by S. Ali Abbas and Kenneth Rogoff, with Chengyu Huang and Kunxiang Diao Authorized for distribution by Mark Flanagan September 2019 Abstract The last decade or so has seen a mushrooming of new sovereign debt databases covering long time spans for several countries. This represents an important breakthrough for economists who have long sought to, but been unable to tackle, first-order questions such as why countries have differential debt tolerance, and how debt levels affect the scope for countercyclical policy in recessions and financial crises. This paper backdrops these recent data efforts, identifying both the key innovations, as well as caveats that users should be aware of. A Directory of existing publicly-available sovereign debt databases, featuring compilations by institutions and individual researchers, is also included. JEL-Classification Numbers H6, H63, N1, F3. Keywords sovereign debt, public debt, government debt, debt statistics, historical database Author’s E-Mail Address sabbasimf.org, krogoffharvard.edu, kdiaoimf.org and chuangimf.org 1 The contents of this working paper are expected to be included in the forthcoming book S. Ali Abbas, Alex Pienkowski, and Kenneth Rogoff eds, Sovereign Debt A Guide for Economists and Practitioners, Oxford University Press. IMF Working Papers describe research in progress by the authors and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the authors and do not necessarily represent the views of the IMF, its cutive Board, or IMF management. © 2019 International Monetary Fund WP/19/1953 Background Over the past decade, there has been a quantum leap in the development of long-dated cross- country sovereign debt databases that has provided both important resources for researchers, as well as a more complete frame of reference for policymakers and investors. To understand issues surrounding the risks and opportunities from sovereign debt, it is important to know not only current magnitudes, maturities, and currency of denomination, but also to have extensive historical data on these same variables. Otherwise, it is extremely difficult to meaningfully ask questions such as “At what levels of debt do emerging markets begin to face high risks of losing investor confidence, Does entering a deep recession or financial crisis with very high public debt impede a country’s ability to engage in countercyclical fiscal policy” This appendix showcases some of the resources now available. It should be noted that although this topic received little attention from official lenders including the International Monetary Fund IMF until the 2008 financial crisis, the IMF has since devoted considerable resources and moved to the forefront of this topic as of this writing. The IMF’s new work begins with the publication of the historical debt database of Abbas and others, 2 2010which importantly builds on the archival research of Reinhart and Rogoff 2009in This Time is Different Eight Centuries of Financial Follyand, more recently, the even more ambitious IMF 2018 Global Debt Database GDD which now provides the most comprehensive account of private and public debt since World War II see Mbaye, Moreno Badia, and Chae 2018. 3While it may seem hard to believe, prior to Reinhart and Rogoff’s This Time is Different work and their two earlier 2008 NBER papers, 4modern researchers had access to only extremely limited long-dated historical data on total public debt. While data on public debt issued abroad was widely available, data on domestically-issued debt was extremely sparse. Indeed, such data was not available for most advanced economies for the period before the 1980s. The problem was worse for emerging markets and low-income countries, with studies like Abbas and Christensen 2007 forced to use banking system claims on government as a 2 “A Historical Public Debt Database,” IMF Working Paper 10/245, also published as “Historical Patterns and Dynamics of Public DebtEvidence From a New Database,” IMF Economic Review, Volume 59, Issue 4. 3 “The Global Debt Database ology and Sources,” IMF Working Paper 18/111. For the data, see https//www.imf.org/en/Publications/WP/Issues/2018/05/14/Global-Debt-Database-ology-and-Sources- 45838. The GDD aims to address several shortcomings of previous databases. First, it takes a fundamentally new approach to compiling historical data. The GDD adopts a multidimensional approach by offering multiple debt series with different coverages, thus ensuring greater consistency across time. Second, it more than doubles the cross-sectional dimension of existing private debt datasets. Finally, the integrity of the data has been checked through bilateral consultations with officials and IMF country desks of all countries in the sample. 4 “This Time is Different A Panoramic View of Eight Centuries of Financial Crises,” NBER Working Paper 13882; and “The Forgotten History of Domestic Debt,” NBER Working Paper No. 13946. 4 proxy. 5This fact which was first emphasized in Reinhart, Rogoff, and Savastano 2003 may sound incredible to those who are not empirical researchers using macroeconomic data. 6The reasons why the data was not better kept are something of a mystery, but it is nevertheless true that long-dated historical data on domestically-issued public debt and, hence, total public debt, was not available in any standard source, even the debt time series published by the IMF or the World Bank. 7This lack of historical debt data was a huge omission that, as Reinhart and Rogoff 2009 and in several related papers showed, led to a blind spot in research in many areas, for example, the risks of sovereign debt crisis, and the genesis of very high inflation. Importantly, Reinhart and Rogoff 2009 contained extensive documentation of data sources for constructing long-dated time series on public debt, both domestically-issued and issued abroad, offering a blueprint for replication and extensions, including the IMF’s important research that followed. As Reinhart and Rogoff discuss, and as others have since replicated, a critical archival source turned out to be old volumes of the League of Nations. One fact Reinhart and Rogoff found was that, contrary to widespread belief, many emerging markets, including in Latin America, had been able to issue in earlier decades large quantities of long- term debt denominated in their own currencies. For many of these countries, therefore, the problem of “original sin” actually started with the hyperinflations of the 1980s and 1990s. Another major gap in the data concerned the currency composition of debt. Even for debt issued abroad, data on currency of denomination had been very limited, except for the modern period as for example the data employed in Bevilaqua, Bulow, and Rogoff, 1992. 8But, again, recent work by IMF economists notably Abbas and others, 2014 has taken large strides in remedying this situation by providing extensive data on both currency of denomination and maturity,data that allows far more nuanced analyses of debt vulnerabilities than was previously possible. 9More recently, there have been many other important advances, as embodied in the data sets listed below. Importantly, the databases listed in this appendix are all associated with careful scholarly research. Whereas they have been integral to construction of the IMF’s GDD, they are of particular use for researchers precisely because there are typically associated research 5 “The Role of Domestic Debt Markets in Economic Growth An Empirical Investigation for Low-Income Countries and Emerging Markets,” IMF Working Paper 07/127. 6 “Debt Intolerance,” NBER Working Paper No. 9908. 7 One explanation that is sometime presented is that, until the 1980s/1990s, many countries did not have modern debt management offices that would collect all debt data; and that these functions remained segregated in different parts of the finance ministries/central banks. However, this cannot fully explain why total public debt coverage was relatively good in the inter-war period, the lack of incentive to institute dedicated debt management offices till the 1980s, or why international financial institutions did not put greater priority on catalyzing the collection of historical domestic debt data. 8 “Official Creditor Seniority and Burden Sharing in the er Soviet Bloc,” Brookings Papers in Economic Activity 1. 9 “Sovereign Debt Composition in Advance Economies A Historical Perspective,” IMF Working Paper 14/162. 5 papers with detailed discussion and documentation that aid an understanding of the data, and potentially allow a check of the decisions made in constructing it by making reference to the original sources. Caveat emptor Tedious historical and archival research lies at the heart of many of these new data sets; it is hard enough to decide how to measure contemporaneous debt for example, should the central bank and central government balance sheet be integrated, how does one account for the debt of state-owned enterprises, how does one account for government assets, what about unfunded pension liabilities but historical ones are more difficult, particularly in the many cases where the data had effectively been lost or forgotten until recentlya widespread problem with public debt data. And the problems just increase when one wants to bring in other variables, for example, in ing debt/GDP ratios. Although some researchers seem blithely unaware of the fact, Gross Domestic Product is in fact a relatively modern construct, developed by Simon Kuznets and his collaborators in the mid-1930s. Economic historians have done very important work trying to artificially reconstruct historical data most famously, the work of the late Angus Madison who curated data from many other researchers, https//www.rug.nl/ggdc/historicaldevelopment/maddison/, but there are huge gaps and many assumptions are required. Indeed, even many of today’s advanced economies took many years after the Second World War to build up their statistical capabilities so that for quite a few states data before 1955 can be unreliable, and for some this is the case going into the 1960s. A second important caveat relates to cross-country and time series comparability of debt data. Because researchers have an interest in developing long time series for as many countries as possible, they often make expedient decisions regarding inclusion/exclusion of data that can result in statistical inconsistencies for instance, a long debt series for an individual country may concatenate central government debt in the early years with general government debt in later years an issue made transparent in Mauro and others, 2013; 10or public debt numbers can be reported for two sovereigns for the same year but without noting the differences in instrument coverage, valuation basis, and netting of assets. 11With this background, readers should be able to better appreciate both the major gaps in sovereign debt data, as well as the important advances recently made by institutions and researchers to address those gaps. The Directory of Sovereign Debt Databases that follows provides a state-of-play on these advances, by providing a listing of known sovereign debt datasets currently available for use by researchers and practitioners. 10 “A Modern History of Fiscal Prudence and Profligacy,” IMF Working Paper No. 13/5, International Monetary Fund, Washington, DC. 11 To be able to analyze and balance the competing considerations of larger sample and comparability, and to appropriately qualify their conclusions, researchers may find useful the international ological standards on compiling public sector debt e.g., the IMF’s Public Sector Debt Statistics Guide for Compilers and Users, 2013 and Government Finance Statistics Manual 2014. 6 Appendix Table 1 provides a tabular summary of the databases included in the Directory. These are listed in alphabetical order and subdivided into two broad categories Researcher databases, and Institutional databases. For each database, ination is provided on time and country coverage; as well on the types of variables included; these broadly fall into four categories debt-to-GDP levels, debt structure currency, maturity, holder profile; crises/restructurings; and other public finance. Appendix Table 1 Summary of Sovereign Debt Databases No. Researchers/ Institution Name of database Coverage years countries Debt- to- GDP Debt structure Crises/ Restruct- urings Other public finance Researcher Databases 1 Abbas, Belhocine, El-Ganainy US √ √ 13 Panizza Domestic and External Public Debt in Developing Countries 1990-2007 130 economies √ √ 14 Presbitero Domestic Debt in Low- income Countries 1970-2010 44 LICs √ 15 Reinhart Dates for Banking Crises, Currency Crashes, Sovereign Domestic or External Default or Restructuring, Inflation Crises, and Stock Market Crashes Varieties 1800-2010 70 economies √ √ 16 Reinhart Debt-to-GDP Ratios 1692-2010 70 economies √ √ 17 Reinhart EMs Emerging Markets; EMLICs emerging markets and low-income countries; LatAm Latin America; EC European Commission, ECB European Central Bank, BIS Bank for International Settlements; IDB Inter-American Development Bank; IMF International Monetary Fund. Directory of Sovereign Debt Databases as of September 2018 A. Researcher Databases 1. “A Historical Public Debt Database,” IMF Working Paper No. 10/245, International Monetary Fund, Washington, DC 12Authors S. M. Ali Abbas, Nazim Belhocine, Asmaa El-Ganainy, and Mark Horton Website https//www.imf.org//media/Websites/IMF/imported- datasets/external/pubs/ft/wp/2010/Data/_wp10245.ashx Description This database provides a very long time series of debt-to-GDP ratios for almost all IMF member countries. The earliest data point starts from 1692. Most G-7 countries and a few other advanced countries are covered since 1880 and the general coverage starts from 12 Also published as “Historical Patterns and Dynamics of Public DebtEvidence From a New Database,” IMF Economic Review Volume 59, Issue 4.