On Graduation from Default, Inflation, and Banking Crises: Elusive or Illusion?

Abstract

Previous articleNext article FreeOn Graduation from Default, Inflation, and Banking Crises: Elusive or Illusion?Rong Qian, Carmen M. Reinhart, and Kenneth RogoffRong QianUniversity of Maryland Search for more articles by this author , Carmen M. ReinhartUniversity of Maryland and NBER Search for more articles by this author , and Kenneth RogoffHarvard University and NBER Search for more articles by this author University of MarylandUniversity of Maryland and NBERHarvard University and NBERPDFPDF PLUSFull Text Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinked InRedditEmailQR Code SectionsMoreI. IntroductionThis paper addresses the concept of “graduation” from external default, banking, and inflation crises.1 Employing a vast data set cataloging more than 2 centuries of financial crises for over 60 countries developed in Reinhart and Rogoff (2009), we explore the risk of recidivism across advanced economies versus middle- and low-income countries. We show that 2 decades without a relapse (falling into crisis) is an important marker. After 1800, roughly two-thirds of recurrences of external default on sovereign debt and three-quarters of recurrence of inflation crisis occur within 20 years.2 However, crisis recidivism distributions have very fat tails, so that it takes at least 50 and perhaps 100 years to meaningfully speak of “graduation.” Indeed, in the case of banking crises in particular, it is hard to argue that any country in the world has truly graduated.Given that graduation (with its companion question, will this ever happen again?) is arguably one of the most important issues in macroeconomics and development, there has been remarkably little theoretical or empirical investigation of the subject. For example, the large theory literature on sovereign lending and default, while producing many important insights into the fundamental distinction between willingness to pay and ability to pay, largely treats a country’s basic developmental and political characteristics as parametric. There is very little on explaining the political, social, economic, and financial dynamics that ultimately lead a country to be less prone to certain types of crises.We acknowledge that the concept of graduation is a hard nut to crack. Many advanced countries had enjoyed a long hiatus from systemic banking crises after World War II and yet had huge problems during the recent global financial crisis. After 90 years of serial default running from 1557 to 1647, Spain did not default again until 1809. Even the advanced countries had high inflation as recently as the 1970s and early 1980s, while many emerging markets had hyperinflation less than 2 decades ago. Is the advent of modern independent central banks sufficient to guarantee that fiscal dominance never again reasserts itself? Have the rich countries that have supposedly “graduated” from serial default on external debt shifted the locus of risk to de jure or de facto (via inflation or financial repression) default on domestic debt? Does the theory of sovereign default or of financial development tell us that we should expect richer and more advanced countries to be immune? Or is graduation a mirage, with the “graduates” really being at best “star pupils,” and can graduates be distinguished from patients in remission?Our goals in this paper are fairly narrowly circumscribed. Most of our analysis is based on data on the dates and duration of the crises themselves. We speculate on underlying causal factors but do not approach them empirically here.3 Although the various types of crises often occur in clusters, our quantitative analysis mainly treats individual crises separately.We begin the paper in Section II by defining the crises that we will catalog. In Section III of the paper, we present a summary time line of crisis, followed by a brief overview of the early history of serial default on external debt. An interesting case is France, which defaulted on its external debt no fewer than nine times from the middle of the sixteenth century through the end of the Napoleonic War but has not defaulted on external debt since. France is a canonical case of what we define as an “external default graduate.” (This did not stop France from having numerous severe banking crises in the past 2 centuries.)In the main body of the paper, we provide a broad aggregative historical overview of the data across different types of crises, distinguishing between advanced countries and emerging markets, also taking into account the advent of International Monetary Fund (IMF) programs after World War II as another marker of a debt crisis.In Section IX of the paper, we speculate on links between graduation and development and the possibility for recidivism among richer countries. The fact that the canonical theory of sovereign default does not strongly predict smaller problems in richer countries (it does not strongly predict graduation) might be considered a flaw in theory. But it might also be taken as warning sign that graduation can be more difficult and take even more time than our data of “just” a few centuries can reveal. On banking crises, the theory needs to better explain why countries never seem to graduate.The main empirical results from our long-dated historical time series on financial crises may be described as follows. First, the process of “graduation,” that is, emergence from frequent crisis suffering status, is a long process. False starts are common and recurrent. This is especially true in the case of banking crises, for both high-income countries and middle- and low-income countries.Second, the vulnerability to crisis in high-income countries versus middle- and low-income countries differs mostly in external default crises, to a lesser extent in inflation crises, and surprisingly little in banking crises.4Third, the sequence of graduation for most countries is first to graduate from external default crisis, then from inflation crisis, and eventually from banking. The last stage of graduation is extremely difficult, even for high-income countries. Among high-income countries, even though most of them have graduated from external default crisis and inflation crisis, more than 20% recently experienced a banking crisis, and far more when weighted by size. Schularick and Taylor (2009) speculate that advanced countries continue to experience credit busts despite arguably advancing regulation and institutions, because as risks moderate, financial systems grow and restore them.Finally, the role of IMF programs in crises in the modern period is important. The availability of IMF bridge loans certainly has increased countries’ resilience to “sudden stops” but, even setting aside moral hazard problems, is by no means a cure-all. Countries entering IMF programs are still forced to undergo painful macroeconomic adjustments in an attempt to regain sound fiscal footing and regain access to private capital markets. The challenges of successfully implementing IMF programs are underscored by the fact that there are many significant cases in which countries default within 3 years of an IMF bailout. IMF programs may help facilitate orderly debt workouts but do not guarantee them. We also note that in its early history, many of today’s rich countries regularly drew on IMF resources, although there has been a 3-decade hiatus.II. Definition of CrisesExternal debt crisis. We distinguish between external and internal debt on the basis of the legal jurisdiction where the debt contracts are enforced. This is a convenient construct given the history and evolution of sovereign debt. Obviously it may be useful to parse the data in other ways for some exercises, and in principle our data set allows that.Although there are exceptions and there has been some evolution in recent years, typically in our long-dated historical data set, external debt is denominated in foreign currency and held by foreign creditors. There are certainly important examples, such as Mexico’s short-term Tesobono bonds in the mid-1990s, where the debt is domestic yet is denominated in foreign currency and held primarily by foreign creditors. Although we regard the U.S. abrogation of the gold clause in the early 1930s—when gold was revalued from $21 to $35 per ounce—to be a default on domestic debt, many non-U.S. residents were also holding the debt at the time. In general, following standard practice, we define an external debt crisis as any failure to meet contractual repayment obligations on foreign debts, including both rescheduling or repayments and outright default. (As both of these examples make clear, however, one ultimately needs to think carefully about whether graduation from external default may sometimes just mean a shift to episodic de facto and de jure internal default.)In practice, most defaults on external debt end up being partial, with creditors typically (but not always) repaying 30¢–70¢ or more on the dollar, admittedly not adjusting for risk. The rationale for lumping together defaults regardless of the ultimate “haircuts” creditors are forced to absorb is that, in practice, the fixed costs of external debt default (which include difficulties in obtaining trade credits and loss of reputation) tend to be large relative to the variable costs. In principle, one could parse episodes more finely here according to, say, output or tax revenue loss depending on data availability, although we do not undertake that exercise here. See, however, Tomz (2007) and Tomz and Wright (2007).Inflation crises. Following Reinhart and Rogoff (2004), we define inflation crises as episodes in which annual inflation exceeds 20%. This threshold is lower than the 40% we and others have used in related studies on postwar data but is a compromise reflecting that prior to World War I, average inflation rates were much lower, and 20% inflation generally represented a significant level of dysfunction. Indeed, since we are particularly interested here in inflation as a vehicle for partial default, one clearly would also want to consider lower levels of sustained unanticipated inflation such as many advanced countries experienced in the 1970s. Depending on the maturity structure of debt, sustained 10% inflation can certainly be tantamount to de facto default. A proper calibration, however, would require detailed data on the maturity structure of debt (as in Missale and Blanchard 1994) and, ideally, also on the evolution of inflation expectations. We do not attempt this here, though again, this is an important caveat to interpreting the concept of graduation from external debt crises.Banking crises. Our definition of banking crises follows standard practice (e.g., Kaminsky and Reinhart 1999; Caprio and Klingebiel 2003). Following our own earlier work, “We mark a banking crisis by two types of events: (1) bank runs that lead to the closure, merging or takeover by the public sector of one or more financial institutions and (2) if there are no runs, the closure, merging, takeover, or large-scale government assistance of an important financial institution (or group of institutions) that marks the start of a string of similar outcomes for other financial institutions” (Reinhart and Rogoff 2009, 11). We recognize that our listing of systemic (on a national scale) banking crises may be incomplete, especially prior to 1970, especially for crises outside the large money centers that attract the attention of the world financial press.5Having set out basic definitions, we are now ready to view some basic characteristics of the data. To provide context and motivation for the concept of graduation, we begin with a summary time line of financial crises since 1550, followed by a brief overview of the early history of sovereign defaults.III. A Time Line of Financial Crises and the Early History of Sovereign DefaultsTable 1 provides a summary historical perspective that helps show how the three different varieties of financial crisis have spread over time and across country groups. Between 1550 and 1800, sovereign defaults on external debt were relatively common in Europe, but they were relatively rare elsewhere if only because (a) there were few other independent nations in a position to default and (b) given the crude state of global capital markets, relatively few countries were wealthy enough to attract international capital flows. Thus defaults were relatively insignificant in the regions that constitute today’s emerging markets. Systemic banking crises, however, were relatively rare everywhere. The legal and technological underpinnings of modern private banking simply had not reached a stage of maturity and depth sufficient to cause systemic crises in most instances. (Of course, there are exceptions. Following Cipolla [1982] and MacDonald [2006], Reinhart and Rogoff [2009] discuss how England’s 1340 default to Florentine bankers triggered a financial crisis in Italy.) Similarly, inflation crises were relatively rare, although again there are many exceptions (see Reinhart and Rogoff 2009, chap. 12). Prior to the widespread adoption of paper currency, bouts of very high inflation were relatively difficult to engineer.Table 1. Time Line of Crises, 1550–2010 External DebtCrisesBanking CrisesInflation Crises1550–1815 (Napoleonic wars end)Frequent in advanced economies (including the “world powers” of the time); serial in some casesRareRare1826Frequent in “peripheral” advanced economies and most emerging marketsSerial in advanced; rare in emergingRare18501900Serial in advanced; more frequent in emerging1913 (WW1 begins)Frequent in advanced and emerging1945 (WWII ends)Rare in advanced and emergingPost-1945Rare19641973Serial in some emerging marketsMore frequent in advanced; serial in emergingFrequent in advanced and emergingEarly 1980sEarly 1990sFrequent in emerging2000Rare20092010??View Table ImageThe end of the Napoleonic War in the early 1800s marks a significant transition. The largest advanced countries were increasingly able to avoid external default, albeit partly by their ability to issue an increasing share of their debt domestically. Default, however, became common in “peripheral” advanced countries such as Spain and Portugal, while newly independent emerging markets such as Greece and Latin America entered a long period of serial default. Over the same period, as advanced countries developed more sophisticated banking systems, banking crises became far more common. Emerging markets were certainly affected by advanced country banking crises but did not have so many of their own, if only because their financial systems were dominated by foreign banks.By the turn of the twentieth century, emerging market financial institutions had developed to the point where domestic banking crises became more common. By the time of the Great Depression of the 1930s, banking crises were a worldwide phenomenon. Owing in no small part to the financial repression that followed in reaction to the Great Depression, banking crises were relatively rare during the period from the end of World War II until the early 1970s. As financial repression thawed, banking crises became more frequent in the advanced economies and serial in many emerging markets, bringing us to the recent financial crisis episode.Finally, Table 1 gives a time line of inflation crises, which of course were quite common in all countries in the 1970s and remained a problem in emerging markets until the past decade.We thus focus our early history on sovereign external defaults. As Reinhart et al. (2003) and Reinhart and Rogoff (2009) emphasize, many of today’s advanced economies had recurrent problems with default on sovereign debt during the period when they might arguably have been characterized as emerging markets. Table 2 illustrates the case of Europe for the 3-century period 1550–1850, with the years listed marking the beginning of a sovereign default episode.Table 2. External Defaults: Europe, 1550–1850CountryYears of DefaultNumberof DefaultsAustria-Hungary1796, 1802, 1805, 1811, 18165England*1594*1*France1558, 1624, 1648, 1661, 1701, 1715, 1770, 1788, 18129Germany: Prussia1683, 1807, 18133 Hesse18141 Schleswig-Holstein18501 Westphalia18121Netherlands18141Portugal1560, 1828, 1837, 1841, 18455Russia18391Spain1557, 1575, 1596, 1607, 1627, 1647, 1809, 1820, 1831, 184310Sweden18121Source: Reinhart et al. (2003), Reinhart and Rogoff (2009), and sources cited therein.Note: The table excludes Greece (which gained independence in 1829). Note that for some countries, even if there was a default on external debt, there may have been a default on domestic debt, as was the case for Denmark (1813).* Denotes our uncertainty at this time about whether England’s default involved external (as opposed to purely domestic) debt.View Table ImageAs one can see clearly from the table, serial default was quite common among the major European powers during the sixteenth through nineteenth centuries, with France defaulting on its external debt nine times and Spain defaulting 10 times (with three more to follow in the second half of the nineteenth century). One important observation, immediately apparent from the table, is that there is typically a substantial interval between defaults, typically decades, but sometimes centuries. (Note that we require at least 2 years between default episodes to regard them as independent events.) After defaulting in 1683, Prussia’s next default episode did not follow for more than a century in 1807. Portugal, after defaulting in 1560, did not default again until 1828, when the country lapsed into a period of serial default that did not end until 1890. At this writing, Portugal has not defaulted again since. (Importantly, during a significant portion of Portugal’s quiescent period, it had effectively lost its independence.)Figure 1 gives a measure of the duration of periods of recidivism during the pre-Napoleonic era for the independent (relatively) high-income countries of our sample. The figure captures the length of time between default episodes (including cases in which there was no recidivism). As one can see from the figure, fully half of all default recurrences occurred after a more than 20-year hiatus, with a significant percentage occurring even after a 60-year hiatus.Fig. 1. External default crises: duration of tranquil time, 1300–1799, high-income countries, frequency distribution (in %). Note: Duration of tranquil time is calculated as the number of years between two consecutive external defaults’ starting years. We first count the number of external default episodes, then calculate the duration of tranquil time if it was reversed, and finally calculate the frequency distribution. Sample coverage: 14 episodes of default crisis with reversal and two episodes with no reversal, six countries (United Kingdom, Spain, Germany [Prussia], Portugal, Austria, and France). Sources: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.View Large ImageDownload PowerPointAdvanced country external sovereign debt defaults have become much rarer events in the modern era. Germany’s most recent default occurred in 1939, Austria’s in 1940, and Hungary’s in 1941 (Reinhart and Rogoff 2009). Especially interesting are the cases of Sweden and France. France, despite a near record level of defaults in its pre-Napoleonic era, has not defaulted on external debt since. Sweden, too, has not defaulted on external debt since its default at the end of the Napoleonic War in 1812. It would be interesting to explore whether wartime defaults are less damaging to reputation than peacetime defaults, though of course over many episodes, it is precisely the propensity to wage war that motivates many countries to build up large debts (as in the tax-smoothing model of Barro [1979]). Later, we will consider the robustness of our recidivism results to the exclusion of wartime.Reinhart and Rogoff (2009) also show that the kind of long cycles illustrated in Table 2 are quite characteristic of some of today’s emerging markets, many of which have defaulted at least once during the past 2–3 decades. The number of emerging markets that have experienced external debt crises expands considerably if one includes “near-default” episodes in which countries averted technical default thanks to IMF bridge loans. In virtually all these cases, the countries still suffered massive recessions as governments were forced to tighten fiscal policy as borrowing options dried up. Importantly, we do not include these in our calculations below, although arguably from the point of view of understanding macroeconomic volatility and the dangers of excessive debt accumulation, they are equally important. We return to this issue later when we study IMF programs.IV. The Duration and Prevalence of Crises: The Post-1800 ExperienceWe now proceed to focus on the more “recent” period, 1800 to the present, at the same time expanding the analysis to include banking and inflation crises, which, as shown in Table 2, emerged as important in this era. The past 2 centuries also give a much broader sample of independent nations to study, as various regions of the world threw off the yoke of colonialization. In Table 3, we present measures of crisis probability. Each measure takes the number of years a country experienced each kind of crisis (including all years and not just the initial one) divided by the number of years since independence (or since 1800).Table 3. Summary Statistics of Crisis Probabilities External DefaultInflationBankingAverageStandardDeviationAverageStandardDeviationAverageStandardDeviationWorld.19.18.12.12.08.07High-income countries.07.13.06.05.07.04Middle- and low-income countries*.19.17.17.17.11.09Latin America.34.13.12.07.04.03Source: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.Note: Crisis probability is calculated as the number of years in crisis divided by the number of years since independence. Probabilities were calculated for each country since 1800 or the country’s independence year. Sample coverage: 66 countries for external default crisis and 67 countries for inflation and banking crises.* Excludes Latin America.View Table ImageTable 3 shows that the biggest difference between high-income countries and the rest of the world lies in exposure to external default crisis. The average external default crisis probability of the high-income group is less than half of that of middle- and low-income countries and almost one-fifth of that of Latin American countries. The difference would be even larger if we included only twentieth- and twenty-first-century defaults. Inflation crisis probabilities are also higher in the rest of the world than in high-income countries although the gap is smaller. Interestingly, the average probabilities of banking crises in high-income countries and in the rest of the world are similar.6 The results in Table 3 are, of course, complete consistent with the time line in Table 1.Note that inflation and banking crisis probabilities are lower in part because the average duration of these crises tends to be much shorter compared to external default crises. (Note also that we are counting years in crisis, as opposed to the number of independent events.)7Appendix Table A1, which gives the average duration of crises, shows the striking difference between the mean and median duration of external default crises versus inflation and banking crises. The median duration of banking crises is 3 years or less across all income classes, where the world median for default crises is 8 years. For inflation crises, the median is only 1 year across all income classes. Presumably this implies that a country can find ways to trudge on in a state of sovereign default far more easily than it can continue any semblance of business as usual during a banking or inflation crisis.Given the long duration of external default crises and their frequency, it is not surprising that large portions of the world have been in default over much of the last 200 years, as illustrated by Reinhart and Rogoff (2009, 72). Some of the major default episodes include the Napoleonic Wars in the early nineteenth century and then Latin American countries once independent, Greece, Spain, and Portugal in the first quarter of the century. The biggest default spike occurs during the era that bridged the Great Depression and World War II, when at the peak more than 40% of the world, weighted by GDP, was in default on external debt.Figure 2 gives the share of countries in inflation crisis over the same period. Note the huge rise in inflation crises starting after World Wars I and II and again in the 1980s and early 1990s. The very recent history of low inflation throughout most of the world indeed represents a major shift from the preceding 80 years. It remains to be seen whether inflation is a scourge that has been eradicated. As Rogoff (2003) has argued, institutional changes, including especially the advent of independent central banks with a strong anti-inflation commitment, have been an important factor in this dramatic fall in inflation, but so too was the precrisis boom that alleviated political pressures on central banks to engage in unanticipated inflation. It remains to be seen whether the current period will prove merely another lull (one sees many in fig. 2) as opposed to a permanent structural shift toward universal low and stable inflation.Fig. 2. Share of countries in inflation crisis, 1800–2008, world. Sample coverage: 66 countries that were independent in the given year. Sources: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.View Large ImageDownload PowerPointIndeed, if one truly believes that fiscal dominance will never again assert itself in most countries, then, arguably, historical measures of outright default may underestimate the true probabilities (if the option of default via surprise inflation has been effectively erased). The recent explosion of public debt globally underscores this concern.Figure 3 gives the share of the world experiencing banking crises since 1800. Note the remarkably small number of banking crises during the years of financial repression that began during World War II and continued in many countries well into the 1970s. By historical standards, this was a uniquely quiescent period. It is clear also from the figure that this era has been long but seems to be coming to an end.Fig. 3. Share of countries in banking crisis, 1800–2008, world. Sample coverage: 66 countries that were independent in the given year. Sources: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.View Large ImageDownload PowerPointThe next three figures contrast the experiences of high-income countries with those of middle- and low-income countries (including Latin America). They corroborate what we have already seen in Table 3 but give more detail. Figure 4 on external debt crises, for example, illustrates two points. First, as already noted, middle- and low-income countries are in technical default on external debt a significantly higher percentage of the time than high-income countries. Second, high-income countries had a dramatic drop in external defaults starting in the late 1960s with none (as of this writing!) since the advent of floating exchange rates in the 1970s. Later we shall look at evidence on distance since the last default crisis. (Note: We exclude from our middle- and low-income countries very low-income countries that do not have external default by virtue of the fact that they are not able to borrow at all on private markets.)Fig. 4. Share of countries in external default crisis, 1800–2008, high-income versus middle- and low-income countries. Sample coverage: 66 countries (23 high-income and 43 middle- and low-income countries) that were independent in the given year. Sources: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.View Large ImageDownload PowerPointHigh-income countries seem to have graduated from default crisis, or at least gone into deep remission. But most middle- and low-income countries have not yet graduated.Figure 5 shows the frequencies of inflation crises in middle- and low-income countries versus high-income countries. High-income countries have had inflation crises more recently than external default crises, but the frequency has dropped to zero since the early 1990s. For middle- and low-income countries, a spike in the 1990s has been followed by a sharp tapering during the 2000s. Whereas figure 5 is illustrative of the frequency of very high inflation episodes, we note that it does not capture episodes of sustained high inflation below 20% that, if significantly unanticipated and depending on the maturity structure of government debt, may represent a substantial de facto default on domestic debt.Fig. 5. Share of countries in inflation crisis, 1800–2008, high-income versus middle- and low-income countries. Sample coverage: 67 countries (23 high-income and 44 middle- and low-income countries) that were independent in the given year. Sources: Reinhart and Rogoff (2009), sources cited therein, and authors’ calculations.View Large ImageDownload PowerPointFigure 6 on banking crises tells a very different story. (Our data for developing countries begin more recently; hence the dashed line for middle- and low-income countries begins only in the 1860s. Of course, many of today’s developing countries did not gain their independence until later.) One can see that in sharp contrast to external default and inflation crises, banking crises are “an equal opportunity menace” (Reinhart and Rogoff 2009, chap. 10). Although banking crises have picked up dramatically in