This Time Is Different: Eight Centuries of Financial Folly Read online

  We now take up some important potential applications of the data.

  Episodes of Domestic Default

  Theoretical models encompass a wide range of assumptions about domestic public debt. The overwhelming majority of models simply assume that debt is always honored. These include models in which deficit policy is irrelevant due to Ricardian equivalence.7 (Ricardian equivalence is basically the proposition that when a government cuts taxes by issuing debt, the public does not spend any of its higher after-tax income because it realizes it will need to save to pay taxes later.) Models in which debt is always honored include those in which domestic public debt is a key input in price level determination through the government’s budget constraint and models in which generations overlap.8 There is a small amount of literature that aims to help us understand why governments honor domestic debt at all.9 However, the general assumption throughout the literature is that although governments may inflate debt away, outright defaults on domestic public debt are extremely rare. This assumption is in stark contrast to the literature on external public debt, in which the government’s incentive to default is one of the main focuses of inquiry.

  In fact, our reading of the historical record is that overt de jure defaults on domestic public debt, though less common than external defaults, are hardly rare. Our data set includes more than 70 cases of overt default (compared to 250 defaults on external debt) since 1800.10 These de jure defaults took place via a potpourri of mechanisms, ranging from forcible conversions to lower coupon rates to unilateral reduction of principal (sometimes in conjunction with a currency conversion) to suspensions of payments. Tables 7.2–7.4 list these episodes. Figure 7.5 aggregates the data, plotting the share of countries in default on domestic debt each year.

  Our catalog of domestic defaults is almost certainly a lower bound, for domestic defaults are far more difficult to detect than defaults on international debt. Even the widespread defaults on domestic debt during the Great Depression of the 1930s in both advanced and developing economies are not well documented. As a more recent example, consider Argentina. Between 1980 and 2001, Argentina defaulted three times on its domestic debt. The two defaults that coincided with defaults on external debt (in 1982 and 2001) did attract considerable international attention. However, the large-scale 1989 default, which did not involve a new default on external debt, is scarcely known outside Argentina.

  Some Caveats Regarding Domestic Debt

  Why would a government refuse to pay its domestic public debt in full when it can simply inflate the problem away? One answer, of course, is that inflation causes distortions, especially to the banking system and the financial sector. There may be occasions on which, despite the inflation option, the government views repudiation as the lesser, or at least less costly, evil. The potential costs of inflation are especially problematic when the debt is relatively short term or indexed, because the government then has to inflate much more aggressively to achieve a significant real reduction in debt service payments. In other cases, such as in the United States during the Great Depression, default (by abrogation of the gold clause in 1933) was a precondition for reinflating the economy through expansionary fiscal and monetary policy.

  TABLE 7.2

  Selected episodes of domestic debt default or restructuring, 1740–1921

  TABLE 7.3

  Selected episodes of domestic debt default or restructuring, late 1920s–1950s

  TABLE 7.4

  Selected episodes of domestic debt default or restructuring, 1970–2008

  Figure 7.5. Sovereign domestic debt: Percent of countries in default or restructuring, 1900–2008 (five-year moving average).

  Sources: League of Nations; Reinhart, Rogoff, and Savastano (2003a); Standard and Poor’s; and the authors’ calculations.

  Notes: Unweighted aggregates.

  Of course, there are other forms of de facto default (besides inflation). The combination of heightened financial repression with rises in inflation was an especially popular form of default from the 1960s to the early 1980s. Brock makes the point that inflation and reserve requirements are positively correlated, particularly in Africa and Latin America.11 Interest rate ceilings combined with inflation spurts are also common. For example, during the 1972–1976 external debt rescheduling in India, (interbank) interest rates in India were 6.6 and 13.5 percent in 1973 and 1974, while inflation spurted to 21.2 and 26.6 percent. These episodes of de facto default through financial repression are not listed among our de jure credit events. They count at all only to the extent that inflation exceeds the 20 percent threshold we use to define an inflation crisis.12

  Clearly, the assumption embedded in many theoretical models, that governments always honor the nominal face value of debt, is a significant overstatement, particularly for emerging markets past and present. Nevertheless, we also caution against reaching the conclusion at the opposite extreme, that governments can ignore powerful domestic stakeholders and simply default at will (de jure or de facto) on domestic debt. We will now proceed to explore some implications of the overhang of large domestic debt for external default and inflation.

  - 8 -

  DOMESTIC DEBT: THE MISSING LINK

  EXPLAINING EXTERNAL DEFAULT

  AND HIGH INFLATION

  Recognizing the significance of domestic debt can go a long way toward solving the puzzle of why many countries default on (or restructure) their external debts at seemingly low debt thresholds. In fact, when previously ignored domestic debt obligations are taken into account, fiscal duress at the time of default is often revealed to be quite severe.

  In this chapter we also show that domestic debt may explain the paradox of why some governments seem to choose inflation rates far above any level that might be rationalized by seignorage revenues leveraged off the monetary base. (Loosely speaking, if a government abuses its currency monopoly by promiscuously printing currency, it will eventually drive the demand for its currency down so far that it actually takes in less real revenue from currency creation than it would at a lower level of inflation.) Although domestic debt is largely ignored in the vast empirical literature on high inflation and hyper-inflation, we find that in many cases the hidden domestic public debt was at least the same order of magnitude as base money (currency plus private bank deposits at a government’s central bank) and sometimes a large multiple of that amount.

  Understanding the Debt Intolerance Puzzle

  We begin by revisiting the conventional wisdom on external debt default and its implications for debt sustainability exercises and debt default thresholds. Indeed, in the 250 episodes of default on external debt in our database, it is clear that domestic debt loomed large across the vast majority of them. Table 8.1 gives the ratio of both external debt and total debt (including domestic and external liabilities) relative to government revenues on the eve of many of the most notable defaults of the nineteenth and twentieth centuries. We normalize debt by means of government revenues because data on nominal GDP is sketchy or nonexistent for the nineteenth-century default episodes. (For many countries, standard sources do not provide anything close to a continuous time series for GDP for the nineteenth century.)1 Exports, which make sense as the main basis for assessing a country’s ability to service external debt owed to foreigners, are perhaps less important than government revenues once domestic public debt is added to the calculus of debt sustainability.

  Looking more broadly at our sample, figure 8.1 is based on the eighty-nine episodes of external default from 1827 to 2003 for which we have full data on external debt, total debt, and revenues. We see that in all regions except Latin America, external debt has typically accounted for less than half of total debt during the year a country has defaulted on external debt; for Latin America, the average ratio has been higher but still only 60 percent.

  TABLE 8.1

  Debt ratios at the time of default: Selected episodes

  Figure 8.1. Ratios of public debt to revenue during external default:
/>
  Eighty-nine episodes, 1827–2003.

  Sources: The League of Nations, Mitchell (2003a, 2003b), the United Nations, and others sources listed in appendixes A.1 and A.2.

  Thus, uncovering data on domestic debt suggests at least a partial answer to a basic question asked throughout the literature on international debt: Why do the governments of emerging markets tend to default at such stunningly low levels of debt repayment and ratios of debt to GDP?2 In chapter 2, for example, we discussed evidence that serial defaulters tend to default at ratios of debt to GDP that are below the upper bound of 60 percent set by the euro area’s Maastricht Treaty.3 In fact, taking into account domestic public debt, the anomaly largely disappears.

  Figures 8.2 and 8.3 give us a different perspective on the data by providing the frequency distribution (simple and cumulative) of external debt to GDP and of total debt to GDP across all the episodes of external debt in our sample for which we have full data. As the figures illustrate, the ratios of external debt to government revenue are massed at a much smaller average than are the ratios of total debt to government revenue during the year of an external default, with a mean of 2.4 versus 4.2. This order-of-magnitude difference is consistent across individual episodes (as table 8.1 highlights for some well-known cases). It is also consistent across regions and time.

  Figure 8.2. Ratios of public debt to revenue during external default: Frequency of occurrence, 1827–2003.

  Figure 8.3. Ratios of public debt to revenue during external default: Cumulative frequency of occurrence, 1827–2003.

  Sources: The League of Nations, Mitchell (2003a, 2003b), the United Nations, other sources listed in appendixes A.1 and A.2, and the authors’ calculations.

  Note: Kolmogorov-Smirnov, 31.46; significant at 1 percent.

  Obviously, if domestic debt were trivial, the frequency distribution of the total debt ratio at the time of default should overlap that of domestic debt. That is, we should find that domestic debt is quite a small share of total debt on the eve of default. But this is hardly the case, and a standard battery of tests rejects this hypothesis across the board.4

  Domestic Debt on the Eve and in the Aftermath of External Default

  Domestic debt is not static around external default episodes. In fact, the precrisis buildup in domestic debt often shows the same frenzied increases in the run-up to external default as foreign borrowing does. The pattern is illustrated in figure 8.4, which depicts debt accumulation during the five years before and during external default across all the episodes in our sample.

  Presumably, the co-movement of domestic and foreign debt is a product of the same procyclical behavior of fiscal policy documented by previous researchers.5 As has been shown repeatedly over time, the governments of emerging markets are prone to treat favorable shocks as permanent, fueling a spree in government spending and borrowing that ends in tears.6 Figure 8.4 does not continue past the default date. If it did, we would see that countries often continue to run up domestic public debt after they have been shut off from international capital markets.

  Figure 8.4. The run-up in government domestic and external debt on the eve of external default: Eighty-nine episodes, 1827–2003.

  Sources: The League of Nations, the United Nations, and other sources listed in appendixes A.1 and A.2.

  Note: The year of the default is indicated by t; t − 4 = 100.

  Precommunist China (figure 8.5) provides an interesting and instructive example of why domestic debt often builds up in the aftermath of external defaults. Prior to its two major defaults in 1921 and 1939, China’s government relied almost exclusively on external debt. But as access to foreign borrowing dried up, the government, still faced with the need to fund itself, was forced to rely on internal domestic borrowing, despite the underdeveloped state of China’s financial markets. So it is hardly surprising that public domestic debt exploded in the aftermath of both incidents. By the mid-1940s, China’s government relied almost exclusively on domestic debt.

  The Literature on Inflation and the “Inflation Tax”

  Another area of the literature that has, by and large, ignored domestic debt is the empirical work on high inflation and hyperinflation.

  Figure 8.5. Domestic public debt outstanding: China, 1895–1949.

  Sources: Huang (1919); Cheng (2003); the United Nations,

  Department of Economic Affairs (various years), Statistical Yearbook, 1948–1984; and the authors’ calculations.

  Note: Data for 1916–1919 are missing.

  Ever since Cagan, researchers have concentrated on a government’s incentives to gain seignorage revenues from the monetary base.7 Indeed, a recurring paradox in this literature has to do with why governments sometimes seem to increase inflation above and beyond the seignorage-maximizing rate. Many clever and plausible answers to this question have been offered, with issues of time consistency and credibility featuring prominently. However, we submit that the presence of significant domestic public debt may be a major factor overlooked, especially considering—as we have already discussed—that a large share of debt was often long term and nonindexed. We do not refer simply to the study of rare hyperinflation episodes but equally to the much more common phenomenon of high and moderately high inflation as studied, for example, by Dornbusch and Fischer and by many others since.8 Although there are literally hundreds of empirical papers on inflationary finance in developing countries and postconflict economies, domestic debt is rarely mentioned, much less employed in time series analysis.

  Defining the Tax Base: Domestic Debt or the Monetary Base?

  As in the literature on external debt, the implicit assumption is that domestic public debt is relatively unimportant. But is this a good approximation? Table 8.2 suggests that in many important episodes domestic debt has been a major factor in a government’s incentive to allow inflation, if not indeed the dominant one.9 Thus a comparison of actual inflation rates to any hypothetical “seignorage-maximizing rate,” calculated only off the monetary base, may often be beside the point.

  For example, we see from table 8.2 that when inflation first spiked to 66 percent in Germany in 1920 after World War I, domestic debt was almost triple the size of the monetary base. In the case of Brazil, domestic debt was almost 20 times the size of the money base.10

  The importance of domestic debt is hardly confined to episodes of hyperinflation. Table 8.2 lists a number of high-inflation episodes as well. Domestic public debt was almost 80 percent of Japan’s total domestic liabilities (including currency) in 1945, when inflation went over 500 percent. In all of the cases listed in table 8.2, domestic public debt has been at least the same order of magnitude as the monetary base (with the exception of the case of Norway, where it was slightly below that in 1918).

  TABLE 8.2

  Inflation and domestic public debt: Selected episodes, 1917–1994

  Sources: See appendixes A.1 and A.2.

  Notes: “Money” and “debt” refer to the levels at the beginning of each episode. The episodes of hyperinflation meet the classic Cagan definition.

  aThis episode does not meet the classic Cagan definition.

  The “Temptation to Inflate” Revisited

  Precise calculations of the gain to governments of inflating away the real value of their debt require considerably more information on the maturity structure of the debt and interest payments than is available in our cross-country data set. A critical piece of knowledge is the extent to which inflation is expected or not. In addition, one needs to understand bank reserve requirements, interest rate regulations, the degree of financial repression, and other constraints to make any kind of precise calculation. But the fact that domestic nominal debt has been so great compared to the base money across so many important episodes of high inflation suggests that debt needs to be given far more attention in future studies.11

  We have now discussed some of the potential links among external default, inflation, and domestic debt, and emphasized that default throu
gh inflation is an important component of the domestic default calculus. In the next chapter we turn our attention to some features of the domestic versus external default cross-country experience that have hitherto remained unexplored.

  - 9 -

  DOMESTIC AND EXTERNAL DEFAULT:

  WHICH IS WORSE? WHO IS SENIOR?

  We have shown that the amount of domestic debt is large in general, particularly in episodes of external default or high inflation. Clearly, in trying to understand how crises play out, it would be helpful to better understand the relative seniority of domestic and foreign debt. This section is an attempt to provide a first pass in looking at some key characteristics of the data. Clearly, the way crises play out is going to differ across countries and time. Many factors, such as the independence of the central bank and exchange rate regime, are likely to be relevant. Nevertheless, a few simple comparisons of the trajectory of output and inflation during the run-up to and the aftermath of domestic and external defaults are revealing.1

  Our calculations can be taken as only suggestive for several reasons. One is simply that there is no comprehensive database on overt domestic debt defaults prior to our own, much less on de facto defaults. Although we are confident that we have a relatively complete picture of external defaults and episodes of high inflation in our sample, we simply do not know how many episodes of domestic default we may have missed, even restricting attention to de jure defaults. In this chapter we provide a broad indication of how clear episodes of domestic default or restructuring are hidden in the historical archives. Thus, our list of domestic defaults is surely a lower bound on the actual incidence.