Evaluating debt sustainability requires determining what level of debt is high enough to trigger a default by a sovereign country.[1] This is relatively straightforward in the case of a private firm. The “default point” of a private company is the point at which its debt liabilities are equal to the total market value of its assets, that is, the point at which the equity value of the firm becomes zero (Merton, 1974). Finding the default point of sovereign debt is much more complicated, because a government’s assets, which include, for example, the ability to tax its citizens, do not have an observable market value.[2] Moreover, governments typically do not stop paying pensions or disband the military to make room for debt service payments—and are not expected to do so.

Rules of thumb are not likely to provide very useful approximations of the level of debt that triggers default. For example, the debt-to-GDP ratio, at the time of the events of default, of countries that have defaulted since the 1980s has had a wide range of values, from around 0.4 to more than 1.5. Figure 12.1 displays defaulting countries’ level of debt at the end of the year preceding the default, but a similar picture emerges if debt levels in the default year are used. Needless to say, many countries have had debt levels within the same range and have not fallen into default. This suggests that finding the default point of sovereign debt requires a more elaborate analysis.

Furthermore, creditor rights are less effectively enforceable for sovereign debt than private debts. If a private firm becomes insolvent, legal authorities have the means to enforce creditors’ claims on the company’s assets, even if those assets may be insufficient to cover the totality of the debt. By contrast, in the case of sovereign debt, despite the fact that the claim and the relevant legal authority are typically well-defined, the enforcement capacity is limited to assets in the same legal jurisdiction, which limits the efficacy of the legal recourse.[3]

This has led economists to posit that sovereign defaults are a reflection of a government’s “willingness to pay” rather than its “ability to pay.” The theoretical economic literature has traditionally seen the sovereign as calculating the cost implied by a debt default and comparing it to the burden of servicing its debt to decide whether to continue meeting its debt obligations. Defaults are then the result of a strategic decision to obtain a financial gain, rather than the result of a legitimate situation of bankruptcy. In fact, the existence of significant costs of default is considered the mechanism that makes sovereign debt possible in the first place. Otherwise, why would sovereigns repay their debts? If sovereigns did not suffer some type of cost in the event of default, no investor would be willing to lend any money to them (see, for example, Dooley, 2000).

Yet this type of strategic behavior is not in line with what has been observed in sovereign debt crises. Sovereign defaults have taken place after a country’s economy has gone through a serious downturn and other measures have failed. The precise timing seems to respond to economic considerations which are far removed from the strategic factors hypothesized by the sovereign debt literature. In fact, there is evidence that, rather than engaging in strategic considerations and trying to avoid repayments, in moments of crisis, countries go to great lengths to adopt policies aimed at preventing default.

WHAT IS A DEFAULT?

Identifying sovereign default episodes and measuring their duration is not a straightforward exercise. There are multiple definitions of what constitutes a default episode and different ways of determining the precise time at which a default event occurs. Moreover, the nature of a default event is clearly distinct depending on whether the underlying debt is with private creditors or an official lender, and within the latter group, a bilateral or a multilateral creditor. This issue is of consequence from the point of view of empirical research on the effects of default, as the correct identification of the default episode and its precise timing may be critical for dealing with causality and simultaneity issues that arise in econometric work.

The most commonly used catalogs of default events are defaults with official bilateral creditors that are members of the Paris Club (available in the Paris Club database); defaults on private bank loans and bond instruments, as classified by rating agencies like Standard & Poor’s (S&P); and defaults on commercial and official debt under Detragiache and Spilimbergo’s (2001) methodology, which is largely based on the database of the World Bank’s Global Development Finance.

While there is substantial coincidence among the three databases, the correspondence is far from perfect. Some mismatches can be traced to differences in the methodology used to measure the length of a default episode. For instance, the methodology used by Detragiache and Spilimbergo is based on the existence of arrears and rescheduling negotiations and considers as defaults several episodes that are not classified as such by S&P.[4] Conversely, S&P classifies as defaults Argentina’s 2001 and Uruguay’s 2003 exchanges on the basis of their less than voluntary nature, although no arrears were incurred at the time. Not even the well-documented Paris Club defaults are free from methodological ambiguity. In the case of the Paris Club database, only two dates are provided: the date of signature of the restructuring agreement and the cutoff date that determines the debt under renegotiation (where debt incurred after that date is excluded). While the former date has often been used as the starting point of the default episode, a case can be made that it signals the completion of negotiations and therefore the end of the default and that the starting point is likely to be more appropriately proxied by the cutoff date that separates pre- and postdefault debt. This is the criterion used in this report.

Figure 12.2 shows the number of defaults per decade between 1970 and 2004 and before 1970. Episodes tend to lump together at several points, most notably during the debt crisis of the early 1980s, suggesting a dependence on common external factors. However, the incidence of default episodes is by no means restricted to particular periods. On the contrary, there has hardly been a year in the recent period without a default event.

The duration of default episodes—the amount of time that passes between the default event and when the debt is restructured in one way or another—has tended to vary over time, particularly in recent years, declining from an average of about eight years in the period 1970–1990 to roughly four years since 1991. This reflects in part the fact that, for a growing group of emerging market countries, bonds have substituted for banks as the primary borrowing form and that, contrary to what was once thought, bonded debt restructuring through unilateral exchange offers has proven to be much faster to complete than bilateral bank debt restructurings. Additionally, there seems to have been a relative decline in the incidence of default with official lenders, possibly reflecting the diminishing importance of bilateral lending (see Chapter 6).

COSTS OF DEFAULT

The theoretical literature on sovereign debt has traditionally focused on two channels through which costs of default may materialize: reputation (that is, higher borrowing costs that, in the limit, could result in absolute exclusion from financial markets in the future) and direct sanctions (such as legal attachments of property and international trade sanctions imposed by creditors’ countries of residence).

The reputation argument assumes that sovereign debt is used primarily to insure against income shocks by increasing disposable income in bad states of the world, at the expense of a smaller income (resulting from repayment) in good states. In their seminal paper, Eaton and Gersovitz (1981) showed that such a debt contract can be sustained solely on the basis of a country’s credit record (reputation), inasmuch as the loss involved in being deprived of the insurance benefits provided by sovereign debt exceeds the maximum payment under the contract.[5] In a well-known critique, Bulow and Rogoff (1989) showed that a country can always use a fraction of the payment due under the original (defaulted) debt contract to pay in advance for an insurance policy that offers the same benefits, concluding that reputational concerns alone are not sufficient to ensure that sovereign debt can be sustained and that direct sanctions are needed to counteract default risk.[6] However, this critique hinges on three nontrivial assumptions: (1) that the new lender (the insurer) can commit to paying the sovereign as stipulated by the insurance policy (there is no lender commitment problem); (2) that the sovereign can find a new lender while in default (lenders do not collude); and (3) that the sovereign does not spend its debt payment money in the current period (investing in insurance is time consistent for the government). These assumptions have been recently questioned by Kletzer and Wright (2000), Wright (2002), and Amador (2002). Thus, the theoretical debate appears to have gone full circle back to the insurance view.

One problem with the insurance view is that it implies that, in the absence of financial imperfections, net transfers should be negatively correlated with deviation from expected long-run income. Hence, defaults should occur only in good times, when a debtor that has the funds to repay chooses to keep the payment entirely for itself. In other words, default should be, almost by definition, strategic: a deliberate opportunistic decision to repudiate previous obligations.[7] However, both the assumption and the implication appear to be starkly at odds with reality. Private cross-border debt flows are not countercyclical and, as a result, defaults tend to occur in the context of economic contractions, casting doubt on the relevance of the insurance benefits of international debt flows—and the reputational costs of losing access to those benefits (Levy Yeyati, 2006b).

While there is no evidence of the imposition of direct trade or economic sanctions following the latest sovereign default episodes, the literature has highlighted alternative channels through which a country may face immediate real economic cost as a result of default. One such channel is the presence of externalities on sovereign contracts that cannot be readily insured (Cole and Kehoe, 1996) or in the domestic private sector in the form of reduced access to financing (Sandleris, 2006).

More recently, recognizing that holders of government debt are often primarily resident investors, more attention has been paid to a third channel, namely, the immediate consequences of default for the domestic economy (in particular, the impact of default on the solvency of the banking sector and its income effect on domestic demand). This channel has been particularly relevant in recent defaults in emerging economies in which banks held significant amounts of government liabilities and the anticipation of default may have fueled, at least in part, a run on bank deposits.[8]

A natural corollary of this last channel is that a default may involve sizable political cost for the government. A declining economy and a banking system in crisis typically combine in this case with the effect of the default on domestic debt holders to undermine the image of incumbent policymakers, a channel that has been noted in the context of currency devaluations but has been overlooked in the case of debt defaults.

In light of this extensive theoretical literature on sovereign debt, there is surprisingly little empirical work to substantiate these alternative views of the reasons why a sovereign is expected to repay its obligations. Only recently a number of papers have attempted to assess the different channels suggested by the analytical models. These papers do not find much support for the traditional “willingness to repay” explanations of default. Sovereign defaults occur after a country’s economy has gone through a serious downturn and other measures have failed, and the precise timing responds to both economic and political considerations that are far removed from the strategic factors highlighted above.

Access to International Capital Markets

Studies that provide empirical evidence in support of the “market exclusion” view implicitly assume that access to international markets is valuable for a country. Starting from this premise, Tomz (2004) uses the case study method to argue that Argentina repaid its debt with the United Kingdom in 1930 in order to strengthen its reputation as a good debtor, rather than to avoid a trade embargo from the United Kingdom, as had been previously interpreted by, for example, Díaz-Alejandro (1983). In turn, English (1996) studies the evidence on U.S. states that defaulted between 1841 and 1843. While states that were in default were mostly excluded from the capital market, they were able to regain access after renegotiating debt payments, even when the latter involved partial write-offs of debt.

This is in line with the conventional view that a temporary default does not lead to permanent exclusion from the international capital market: a country is likely to lose access to this market while in default, but once the restructuring process is behind it and the country becomes current on its debt once again, the market does not discriminate, in terms of access, between defaulters and nondefaulters; nor does an absence of default in a country’s record guarantee it access to the market. Examples that lend support to this account can be found in the period that goes from the 1930s to the 1960s, in which all Latin American countries were largely excluded from the world capital market regardless of whether they had defaulted in the 1930s, and in the lending boom of the 1990s, which did not exclude countries that had defaulted in the 1980s. More recently, countries that defaulted in the late 1990s regained access to the international capital market almost immediately after their debt renegotiations were concluded.[9]

However, access cannot be analyzed solely as a binary variable according to whether a country is or is not excluded from borrowing. It may well be the case that countries regain access to markets after default, but at a higher financial cost or to a lesser degree.

In this regard, the studies that estimate the impact of past defaults on current sovereign spreads or credit ratings (which tend to correlate closely with spreads) find weak or short-lived effects. While defaults after 1970 are associated with a two-notch drop in a country’s credit rating (Cantor and Packer, 1996), the effect exhibits low persistence: only defaults in the previous five years are found to display any significant correlation with current ratings (Borensztein and Panizza, 2006a). Along the same lines, a number of studies have looked at the direct impact of default on borrowing costs and found generally small or extremely short-lived effects.[10]

By contrast, there appears to be some evidence that in recent years the volume of capital flows to a country has been correlated with its reputation as a debtor. Figure 12.3 illustrates this pattern. In the left panel, the figure compares net private capital flows (as a share of GDP) to two groups of sovereign borrowers over the 2000–2004 period: those that had been in default at some point since 1970, and those that had a completely clean record over the same period. The figure shows substantially higher net flows to the nondefaulting sovereigns. The same can be said about total private cross-border debt flows to these countries, comprising both sovereign and private borrowers. In the right panel of the figure, the same data are displayed after the effect of several variables, including the state of the business cycle and country- and time-specific factors, have been controlled for using econometric methods (see Levy Yeyati, 2006b).

This evidence, however, should be taken with caution, since the reduced inflows may well be reflecting a country’s policy decision to cut indebtedness that has proved excessive in the past, rather than a lessening of the country’s ability to borrow in international financial markets. Indeed, it has been shown in historical studies—such as English (1996) on U.S. states—that defaulting states that regained access to international markets have nonetheless been net payers in the postdefault years.

Trade Sanctions and Trade Credit

While the idea that defaults may lead to trade retaliation has been around for a long time in the economic literature (see, for instance, Díaz-Alejandro, 1983), there have been a limited number of empirical studies on the link between default and international trade. An influential recent study (Rose, 2005) focuses on Paris Club debt renegotiations. Using “gravity” models of bilateral trade (models that estimate natural levels of trade based on variables such as geographic proximity and historical affinity), the study finds that defaults on bilateral official loans appear to be associated with a persistent decline in bilateral trade that lasts for 15 years.[11] There is also some evidence that sovereign defaults with the private sector involve significantly more economic costs for export-oriented industries than for other manufacturing sectors, although in this case the effects have been found not to be persistent once the default is resolved (Borensztein and Panizza, 2006b).

This evidence does not shed light on the specific channels through which defaults may affect trade or, more specifically, exports. The sovereign debt literature has often assumed that reduction in trade following a default comes from restrictive measures imposed by the country of residence of the defaulted investors. However, there is not much evidence of countries’ imposing quotas or embargoes on—let alone initiating direct military actions against—defaulting countries in modern times. Moreover, bondholders do not appear to be an effective political lobby today, as perhaps was the case in the historical period of the nineteenth-century bond market.

A more realistic candidate for explaining the effect of default on a country’s trade is the deterioration in the credit quality of exporting firms in the defaulting country after the default—resulting from confiscation or convertibility risk—which could restrict access to trade credit. There is anecdotal evidence that international trade credit tends to be affected when markets become concerned with the creditworthiness of a government. In 2002, Brazil arranged financing from the IDB and the World Bank to offset the cutting off of international credit lines to Brazilian exporters in the context of country risk concerns (Financial Times, 2002). This support from the international community, together with intervention by Brazil’s central bank and banking system, was successful in protecting the country’s export sector from a credit crunch. At the aggregate level, OECD data on trade credit flows from private and official sources show that defaults have a negative effect on trade credit, but this decline seems to be small in magnitude (Borensztein and Panizza, 2006a; Love and Zaidi, 2004; World Bank, 2004b).[12]

Financial Sector

Possibly the most important collateral effects of government debt crises have been those that have occurred in the domestic financial sector. When banks are heavily exposed to government debt (as is often the case in Latin America; see Chapter 8), government defaults may cause a banking crisis or at least a period of weakening in banking credit to the private sector. This may happen for several reasons. First of all, default episodes may cause a collapse in confidence in the domestic financial system and may lead to bank runs, resulting in banking crises or at least a credit crunch. Second, even in the absence of a bank run, default episodes will have a negative effect on banks’ balance sheets, especially if the banks’ holdings of the defaulted paper are large, and lead banks to adopt more conservative lending strategies. Finally, default episodes are often accompanied by a weakening of creditor rights or at least more uncertainty about them, which may also have a negative effect on bank lending. This result will be a magnification of the economic recession associated with the default. A detailed analysis of four recent sovereign defaults—those of Ecuador (1999), Pakistan (1999), Russia (1998), and Ukraine (1998)—has highlighted this association between debt crisis and banking distress (IMF, 2002c).

There is a fairly close association between sovereign defaults and domestic banking crises. Based on data between 1975 and 2000, it has been estimated that the probability that a banking crisis will occur within one to two years of a default is as high as 14 percent. By contrast, banking crises do not precede defaults very often, despite the fact that defaults tend to be anticipated by the public and deposit runs are likely in such an event (Borensztein and Panizza, 2006a).[13]

Economic Growth

Whatever the specific channel through which a sovereign default affects the domestic economy, if defaults exert a significant influence on economic growth, this must be observed in a direct link between default events and GDP growth. Indeed, there is a strong association between defaults and recessions when annual GDP data are considered (Sturzenegger, 2004; Borensztein and Panizza, 2006a).

Looking at annual data, however, may not bring out the full picture of the relationship between debt crises and growth (Levy Yeyati and Panizza, 2006). Consider, for instance, two recent default episodes: those in Ecuador (1999) and Argentina (2001). Judging from annual data, Ecuador’s GDP contracted by 6 percent in 1999 (the default year), and Argentina’s output declined by 12 percent in 2002 (the official date of the Argentine default was December 2001). The quarterly evolution of GDP in these two countries suggests that the collapse in output occurred just before the default event. This is illustrated in Figure 12.4, which also portrays the case of Uruguay in 2003. Annual data may mask the timing of events, because the start of a recession may spill over from one year into the following year in the data, as annual GDP is an average of what happened during the year. Thus, for example, the sharp GDP contraction in Argentina in late 2001 is largely registered as an output decline in 2002, despite the fact that the economy started to recover in that year. Even more striking is the case of Uruguay, in which a recovery was already incipient when the government launched its debt exchange.

Figure 12.5 takes a somewhat longer view of the evolution of GDP around default episodes. The figure shows, for a sample of emerging economies that have experienced default, quarterly GDP levels in a six-year window centered on the default period.[14]Time 0 in the figure indicates the year of the default episode, time −4 indicates one year (four quarters) before the event, and time 4 indicates one year after the event. As the figure shows, GDP drops in the period preceding default and, while it still falls slightly in the quarter after the event, it reverses its trend in the following quarter. Recovery is quick, and long-term growth does not appear to suffer in the period after default.[15] Naturally, as in the case of the drop in output at the time of default, the growth improvement may be related to recovery from deep currency and financial crises. However, this does not appear to be the case here, as the same pattern emerges when the immediate postdefault recovery period is excluded.[16]

These findings do not imply that policies that lead to default have no cost; on the contrary, the large GDP decline that typically precedes a default may reflect in part the anticipation of the default decision, and the postdefault recovery observed in the data may not be independent of the cost paid in the preceding periods.[17] Moreover, to some degree, the fact that output contractions precede an event of default may be a consequence of the fact that the default is already widely anticipated, and residents may start to hoard financial assets abroad and postpone investment projects in view of the uncertain prospects. Likewise, the official timing of the default may in some cases be somewhat later than the time when the debt crisis that generates the default actually started. Bond payments, for example, typically have a grace period before the bonds are technically considered in default.

But the evidence on defaults, recessions, and recoveries flies in the face of stories of “strategic” sovereign default. A sovereign choosing to default on the assumption that it could obtain a financial gain by doing so, even if it faced exclusion from financial markets, would always default in a situation of economic strength, namely, when it did not need to resort to borrowing in the near future. Recessions are periods when incomes are low and financing needs are high, and far from being the result of punishments or sanctions imposed by creditors as retaliation for the default, as posited by the traditional theoretical models of sovereign debt, they often precede (and may cause) the default decision. Defaults coming after, or in the midst of, a period of economic weakness are an indication of a situation of financial insolvency on the part of the sovereign, not a sign of a calculated, strategic action.

Politics

Postponing a default that has been widely anticipated by the market may be costly for at least three reasons: (1) it requires adjustment policies that may be perceived to be excessively costly (at least in the short run) and hence not fully credible; (2) default risk, as long as it is not realized, translates into high interest rates and overall business uncertainty that hamper investment and deepen banking fragility through greater nonperformance ratios; and (3) the liquidity crunch typically associated with debt crises leads to fire sales of assets and to the validation of unreasonably high rollover costs, compromising the solvency of both public and private debtors.

And yet, politicians and technocrats in ministries of finance and central banks seem to go to great lengths to avoid default. In the case of Argentina in 2001, for instance, it has been reported that, prior to the default, even Wall Street bankers tried to persuade Argentine policymakers to face reality and initiate a debt-restructuring operation (Blustein, 2005).

Why the reluctance? A possible hypothesis is that defaults can be politically costly for the careers of finance ministers and top executive politicians. In fact, the evidence on the political impact of recent events of default reveals that in 18 out of the 19 cases studied, the ruling coalitions lost votes after the default. In addition, ruling governments in defaulting countries faced, on average, a 16 percent decline in electoral support and, in 50 percent of the cases, a change in the chief executive in the year of the default or the following year—more than twice the probability in normal economic times (Borensztein and Panizza, 2006a).[18]

This political cost may relate to the fact that electors may interpret a default as an explicit sign that the policies in place prior to the default were not working. However, this would explain reluctance to default only among politicians and technocrats who have been in charge for a long period of time and hence can be blamed for past policies. But new politicians and technocrats who are appointed with the explicit or implicit objective of saving a country from default may also try to delay default, because default would be a clear signal that they had not been able to accomplish their mission.[19] These politicians may have an incentive for “gambling for resurrection,” namely, taking extreme measures that have a low chance of success but, if they do succeed, will bring clear political gains to the ruling administration. A more benign interpretation is that politicians know that the market will severely punish a strategic default, and in delaying default, they go to great lengths to make sure that everybody agrees that a default is indeed inevitable and hence not strategic. According to this interpretation, politicians who delay default are actually maximizing social welfare and not just their own.

ABILITY TO PAY OR WILLINGNESS TO PAY?

There is a disconnect between the theoretical literature on sovereign debt and the empirical evidence on sovereign defaults and renegotiations. The theoretical literature presents a paradigm of sovereign debt as a contract in which the costs of default only slightly exceed the benefits of pocketing the money otherwise needed for repayment. Hence, according to this literature, default events occur in good times, when countries enjoy a strong financial position and do not anticipate the need of market financing in the near future. These are strategic defaults rather than bankruptcies such as those that occur in the business world. There is little evidence, however, of strategic sovereign defaults ever occurring, and time after time default events occur in situations in which a country has reached a condition that can be described as sovereign bankruptcy.[20]

A central finding of this chapter is that while defaults may have costs in terms of higher spreads, lower international trade, and more limited access to finance, these costs tend to be short lived. More interestingly, the chapter provides preliminary evidence that economic crises take place before defaults and that recoveries start soon after the event. This suggests that sovereigns may sometimes delay debt-restructuring decisions too long. Clearly, more work needs to be done in this area, but if further analysis confirms this conjecture, the next challenge will be to find out why this is the case and what the policy implications of this finding are.

The chapter provides two conjectures as to why a political administration may postpone the moment of reckoning. The first focuses on self-interested politicians who are worried about the effect on their careers, as there is clear evidence of accelerated political turnover following a debt default. The second interpretation assumes that, while strategic defaults would be very costly in terms of reputation—and that is why they are never observed in practice—“unavoidable” defaults carry limited reputation loss in the markets (Grossman and Van Huyck, 1988). Hence, policymakers may postpone default actions to ensure that there is broad consensus, prior to the actual occurrence of default, that the decision is unavoidable and not strategic. The idea is that politicians choose the lesser of the two evils and are willing to pay the additional cost brought about by the delayed default rather than subject the country to punishment by the market. This would be consistent with widely anticipated defaults that happen in situations when the economy is very weak.

These two interpretations have widely different policy implications. If the problem is self-interested politicians who do not maximize social welfare, then reforms should focus on the policymaking process (see IDB, 2005b). If the problem is that politicians delay default in order to guarantee that markets will perceive the default, when it does occur, as necessary, then part of the solution may be a better, faster understanding of the economic situation of countries that are headed for default, an area in which the international financial institutions could make a valuable contribution.