CR 8 – Country Risk – Determinants, Measures, and Implications

Country risk is the umbrella term for everything that can go wrong when you lend to, invest in or trade with a particular country. Of the many strands under that umbrella, the one that financial markets can see most clearly is the chance that the national government itself stops servicing its debt. That chance is called sovereign default risk. Because the government sits at the centre of its economy, the rate it pays to borrow feeds into the rates charged to banks, companies and households in the same country, so a sound estimate of sovereign default risk is needed not only to price government bonds and loans but to price almost every other asset there as well.

People have tried to put numbers on this risk since the nineteenth century, when investors in London and Paris poured money into newly independent states and were regularly left holding unpaid paper. The modern toolkit has three layers. The oldest and most widely quoted is the sovereign credit rating published by agencies such as Moody’s, Standard & Poor’s (S&P) and Fitch. The second is the market default spread, the extra yield a government pays on bonds issued in a foreign currency over a default-free bond in that same currency. The third and newest is the sovereign credit default swap (CDS) spread, the annual premium a buyer pays a seller for protection against a government credit event.

Any of these measures is easier to judge once you know how often governments have actually defaulted, on which kinds of debt, and what happened to them afterwards. The lesson follows that order. It begins with the historical record, moves on to the economic and political variables that push a government toward default, then looks at how ratings are built and where they have failed, and finishes with the two market-based measures and the evidence on which one predicts default best.

Foreign currency debt versus local currency debt

One distinction runs through the whole topic: the currency in which the government owes money. Foreign currency debt is owed in a currency the government cannot create, such as US dollars for an emerging market borrower, so repayment depends on earning, holding or borrowing enough of that currency. Local currency debt is owed in the government’s own money, which its central bank can in principle print without limit. Defaults on the first kind are easy to understand, because a government can simply run out of dollars. Defaults on the second kind look puzzling at first sight, and a later section explains why they happen anyway.

Keep in mind, too, that a default is rarely a clean event in which lenders lose everything. Most episodes end in a negotiation in which old debt is swapped for new debt with a longer maturity, a smaller principal, a lower coupon or some mix of the three. Rating agencies count both an outright missed payment and such a distressed exchange as a default, so the historical statistics that follow include both.

A government that borrows in dollars, euros or pounds is exposed in a very direct way. If its tax take, export earnings and reserves fall short of what is due in that currency, the treasury has no printing press to fall back on and a missed payment follows. Much sovereign borrowing has historically taken this form, either as loans from banks in lending countries or as bonds sold to foreign investors, and a large share of all sovereign defaults has come from it.

The table below is drawn from Moody’s records of sovereign defaults over the period 1983 to 2021. For each episode it gives the defaulting country, the month of default, the amount of debt in default in millions of US dollars, the Moody’s rating when the default happened, and whether the bonds involved were foreign currency (FC), local currency (LC) or both. The asterisks are carried over from the original listing, which does not explain them; "ongoing" marks episodes that had not been resolved when the data were compiled.

Source: Moody’s.

A few patterns stand out. First, governments almost never default from a comfortable rating: apart from Venezuela’s 1998 local currency default at Ba2, every rating at default sits at B3 or lower, and most are in the Caa band or below. Second, the same names keep reappearing. Belize defaults five times in fifteen years, Argentina four times, and Ukraine and Ecuador three times each, which is consistent with the idea that a past default makes future default more likely. Third, the amounts span three orders of magnitude, from $145 million for Moldova in June 2002 to $264,211 million for Greece in March 2012. The Greek and Cypriot episodes, which involved euro-denominated local debt, are also reminders that local currency defaults can happen even inside a rich-country monetary union, a theme picked up two sections later.

Over the last two centuries, sovereign defaults have not been spread evenly through time. They arrive in waves, with long quiet spells in between. A survey by Hatchondo, Martinez and Sapriza (2007) records the default years of European and Latin American governments in a set of windows running from 1824 to 2003. Counting the countries that defaulted at least once in each window gives the picture below.

Counts derived from the country-by-country default years in the survey; a country appears once per window however many defaults it had in that window.

Two windows dominate. The Great Depression of the 1930s pushed nine European governments, including Germany, Austria, Hungary and Italy, into default, alongside fourteen Latin American ones. The debt crisis of 1976 to 1989 was almost entirely a Latin American event, with sixteen countries of the region defaulting. Some borrowers turn up in nearly every window: Argentina and Ecuador each appear in six of the seven. The survey does not cover Asia or Africa, but governments in both regions have also defaulted over the last 50 years.

What the wider research says

Studies by the rating agencies and by academics agree on two further points. The first is that governments have been more likely to default on loans from banks than on bonds they issued. A sovereign default database maintained by the Bank of Canada and the Bank of England, which tracks the share of sovereign debt in default back to 1820, shows this gap. Until the 1960s, bank loans were the only practical way for most governments to borrow abroad; since then, bond markets have opened up access to many more sovereigns.

The second point is that, measured in dollars, Latin American governments account for the largest slice of defaulted sovereign debt over the last 50 years. Data from 1975 to 2004 split by region show the region’s share dominating, and the 1990s were the only decade of the last five in which Latin America did not make up 60% or more of the sovereign debt in default.

The nineteenth century template

Latin America’s history is worth a closer look because it reveals the mechanics of default so clearly. After independence, the new republics had little domestic saving but plenty of natural resources, and British, French and Spanish investors were eager to lend. The governments took on large debts, typically payable in gold or a foreign currency and running for more than 20 years. Brazil and Argentina also sold domestic debt carrying gold clauses that let the lender choose repayment in gold, which turned nominally local debt into a claim the government could not print. Defaults were typically triggered by wars with neighbours or coups at home, and weak institutions made every shock worse.

Of the 81 sovereign defaults recorded between 1820 and 1919, 58 were Latin American. Over the 115 years from 1825 to 1940, the countries of the region together spent 38% of the time in default. The table converts each country’s percentage into an approximate number of years.

Source: Taylor (2003), with default data from Tomz (2001). Years column derived.

Defaults on foreign currency debt attract the headlines, but a number of the countries in the historical record also defaulted on debt denominated in their own currency, often at the same time. An S&P survey of defaults since 1975 counts 23 issuers that defaulted on local currency debt. Examples include Argentina (2002 to 2004), Madagascar (2002), Dominica (2003 to 2004), Mongolia (1997 to 2000), Ukraine (1998 to 2000) and Russia (1998 to 1999). The pattern can even run the other way round: Kuwait in 1992 kept paying its foreign currency creditors even as it stopped servicing local currency debt.

The amounts can be very large. Russia’s default on $39 billion of ruble debt is the biggest local currency default in recent years, and Brazil defaulted on $62 billion of local currency debt in 1990. Moody’s data on the share of sovereigns defaulting between 1975 and 2004 show local currency defaults occurring throughout the period, though at a lower rate than foreign currency defaults.

Defaulting on both at once

When Moody’s split sovereign defaults by the type of debt involved, it found that governments are increasingly defaulting on local and foreign currency debt together rather than on one alone. The comparison of two periods makes the shift plain.

Source: Moody’s. The later column sums to 99% because of rounding in the source.

Joint defaults rose from a small minority to the largest single group. A more recent breakdown by the Bank of Canada, which classifies defaulted debt by currency and by type of creditor from 1976 to 2020, tells a related story: over time the debt in default has moved away from foreign currency toward local currency, and away from bank loans toward borrowing from other lenders, including bondholders and official creditors such as the IMF and the Paris Club.

Erce, Mallucci and Picarelli (2022) add a legal angle. They find that in recent decades sovereigns have increasingly defaulted on debt issued under domestic law rather than foreign law. Two forces lie behind this. Governments have become more willing to default on local currency debt, and restructurings of domestic-law debt are completed faster. The median duration of domestic-law defaults peaked in the 1990s and was shorter in the 2000s and the 2010s, with the 2010s median the shortest of the four decades shown, so restructuring has grown quicker over time, judged purely by its length.

If a central bank can create unlimited amounts of the local currency, a government should never have to miss a local currency payment. That argument is logically sound but ignores three real constraints. The first two are situations in which the government has lost the power to print freely; the third is a deliberate choice between two painful options.

Lost the power to print

Under the gold standard, which some countries followed in the decades before 1971, every unit of currency had to be backed by gold reserves. The size of those reserves capped how much money could be issued, so a government short of gold was effectively short of currency, however much its debts were denominated in its own unit.

A shared currency creates the same constraint in modern form. Countries that adopted the euro accepted a bargain: they gained a common market and the ease of one money, and in return each gave up control over how much of that currency it could create. When Greece faced its crisis, the Greek government in July 2015 could not simply print euros to pay its creditors, because only the European Central Bank can issue them. For debt purposes, euro-denominated borrowing behaves like foreign currency borrowing for an individual member state.

Default versus debasement

Even where the printing press is available, currency debasement is costly. Creating money to pay debts debases the currency, devalues it against other currencies and can drive inflation up at an accelerating pace. High inflation in turn shrinks the real economy: investors pull out of financial assets and markets and pile into real assets such as property and gold, while companies divert effort from productive investment into financial speculation. Default carries its own costs, discussed in the next section, so a government in distress has to weigh which route does less long-term damage. Many governments conclude that default is the cheaper option.

One explanation for why some countries default while others inflate focuses on the balance sheets of local companies. If firms have funded local currency assets with foreign currency debt, printing money is especially destructive: inflation and devaluation shrink the value of their assets while the foreign currency liabilities stay the same, and the corporate sector can be wiped out. In that setting the government has a strong reason to default on its own local currency bonds rather than devalue.

Before the twentieth century, sovereign default could bring gunboats rather than lawsuits. When Turkey defaulted in the 1880s, Britain and France intervened and appointed commissioners to oversee the finances of the Ottoman Empire. Around the same time, after Egypt defaulted, Britain sent troops and seized control of its government. A Venezuelan default early in the twentieth century led European powers to blockade the country, prompting a response from President Theodore Roosevelt, whose government saw the blockade as a challenge to US power in the region.

Since then the costs have become economic and political rather than military. Lenders obviously lose some or much of what they are owed, but the defaulting country pays in four further ways.

Four channels of damage

Reputation loss. A government that defaults carries a "deadbeat" label for years, which makes it harder and more expensive to borrow again. Capital market turmoil. Investors pull money out of the country’s equity and bond markets, so private companies there struggle to raise funds for projects. Real output. The uncertainty spreads to investment and consumption; households cut spending and firms hold back from long-term commitments, so defaults are usually followed by recessions. Political instability. Default can wound national pride and endanger the leadership. The wave of European defaults in the 1930s, which took in Germany, Austria, Hungary and Italy, helped the Nazis rise and set the stage for the Second World War, and in Latin America defaults and coups have travelled together for much of two centuries.

Default seldom means a government repudiates its debt entirely. Most episodes end with negotiations that produce a debt exchange or restructuring, giving the government more time, a lower principal, lower interest, or a combination. Rating agencies usually date a default episode from the missed payment to the completion of the restructuring. Governments that keep lenders’ losses small can limit the damage to their reputation and return to markets sooner.

What the evidence shows

Empirical research on the aftermath of default reaches five conclusions. Real GDP falls by between 0.5% and 2%, with most of the drop in the first year, after which the effect fades. Ratings and borrowing costs suffer for longer: a study of 1995 ratings found that countries with at least one default since 1970 were rated one to two notches below otherwise similar countries, and defaulters pay borrowing costs about 0.5 to 1% higher, though both effects dissipate over time. Trade can suffer retaliation: one study finds an 8% drop in bilateral trade after default lasting up to 15 years, and industry-level work shows export-oriented industries are hit hardest. Banking systems become more fragile: across 149 countries between 1975 and 2000, the probability of a banking crisis was 14% in countries that had defaulted, eleven percentage points above the rate for countries that had not. Finally, political change becomes more likely. Studies of the sharp devaluations that often accompany default, covering 1971 to 2003, find a 45% increase in the probability that the top leader (president or prime minister) is replaced and a 64% increase in the probability that the finance minister or central bank head is replaced.

Default is therefore costly, and governments are right not to treat it lightly. It is especially expensive when it triggers a banking crisis, which impairs firms’ ability to fund investment for years, or a currency collapse, which breeds political and institutional instability that also lasts a long time.

Governments default for much the same reason that households and companies do. In good times they borrow more than their assets and earning power can support, and when a downturn arrives they cannot keep up with the payments. An analyst assessing a country’s default risk would therefore examine six groups of variables.

1. Degree of indebtedness

The natural starting point is how much the government owes, both to foreign banks and investors and to its own citizens. Because larger economies can carry more debt in absolute terms, the total is scaled by gross domestic product. The table lists the 20 countries with the highest government debt relative to GDP in 2020, together with the 2010 figure and the change between the two.

Source: IMF. Change column derived.

The list is a useful screen but an incomplete measure. Venezuela, whose debt ratio rose from 25% to above 300% of GDP, clearly carries high default risk, as do Egypt and El Salvador. Yet the same list includes Japan, the United States and France, which ratings agencies and markets regard as among the most creditworthy borrowers in the world. US federal debt, tracked from 1966 to 2021, rose above 100% of GDP with surges after 2008 and again after 2020, the highest ratio since the Second World War, which explains why questions about US default risk have surfaced at all. Different data sources report somewhat different levels, but the trend is the same.

2. Pension and social service commitments

Many governments have also promised their citizens pensions and health care. These obligations compete with debt service for the same limited revenue, so countries with larger commitments should carry more default risk. Because both pension and health costs rise as people age, countries with ageing populations and a shrinking working-age share face extra pressure.

3. Revenues and inflows to government

Government revenue comes mostly from taxes, so its size depends on how taxes are designed (the tax code) and on what there is to tax (the tax base). Other things equal, a broader base generates more potential revenue that can be used to meet debt payments.

4. Stability of revenues

Debt creates fixed payments that fall due in bad years as well as good ones, so a government with steady revenue faces less default risk than one with volatile revenue. Revenue stability depends partly on how diversified the economy is. Peru, which leans on copper and silver production, and Jamaica, which depends on tourism, face more default risk than larger and more diversified economies such as Brazil or India. It also depends on the tax system: revenue from income taxes tends to be more volatile than revenue from sales taxes or value added taxes.

5. Political risk

The decision to default is political as well as economic. Because default puts the leadership under pressure, an autocracy, less exposed to voter anger, may find default easier than a democracy would. And because printing money is the main alternative to default, the independence and strength of the central bank also enter the assessment.

6. Implicit backing by stronger partners

On joining the European Union, Greece, Portugal and Spain were judged less likely to default by investors, analysts and ratings agencies, who quietly assumed that Germany, France and the Scandinavian countries, as stronger members, would come to the rescue of a weaker one. The danger with any implicit guarantee is that it is not written down, so lenders may find the support never arrives and that they have no legal claim to it.

A full assessment, in other words, means going beyond the debt numbers to understand how the economy works, how robust the tax system is and how far the governing institutions can be trusted.

Few investors have the time or resources to study small and unfamiliar countries, so third parties have filled the gap. Bond rating agencies entered this field with two advantages. They had spent a century or more assessing default risk in companies and could carry much of that skill across to governments. And bond investors already understood the rating scale from the corporate market, so it was easy to extend it to sovereign bonds, where a AAA rated country is seen as close to riskless and a C rated one as very risky.

From a handful of ratings to more than a hundred

Moody’s began rating corporate bonds in 1919 and added government bonds during the active market of the 1920s; by 1929 it covered almost fifty central governments. The Depression and the Second World War shrank interest in government bonds and in rating them, and the business revived only slowly from the 1970s. Even in the early 1980s only about fifteen mature governments had ratings, most at the top level. The table shows how quickly the list then grew.

Cumulative column derived.

The cumulative column confirms that fifteen sovereigns were rated by 1984. Between 1985 and 1994 the newly rated countries were mostly riskier borrowers, and the median rating of new entrants fell from AAA/Aaa to A/A2 and then to BBB-/Baa3. The growth continued after 1994 as sovereign bond markets expanded, and by 2022 Moody’s, S&P and Fitch each rated more than a hundred countries.

Local currency and foreign currency ratings

The ratings themselves now carry more detail. Moody’s and S&P each give a country two ratings: a local currency rating for debt in the domestic currency and a foreign currency rating for borrowing in other currencies. The outlook attached to each rating signals the agency’s view of the next move, with NEG marking a possible downgrade, POS a possible upgrade and STA a stable rating. The Moody’s ratings for Latin America in July 2022 illustrate the system.

Source: Moody’s. "none" means no local currency rating was assigned.

Ecuador, El Salvador and Panama carry only a foreign currency rating, and every other country in the list has identical local and foreign currency ratings. In principle the local currency rating should be at least as high as the foreign currency rating, because a government has more power to produce its own money than to obtain someone else’s. Exceptions exist in the other direction, however. In March 2010, for example, India was rated Ba2 on its local currency debt but Baa3 on its foreign currency debt, so its local currency rating was one notch lower.

Agencies broadly agree on most sovereigns but can differ sharply on individual countries, reflecting different judgements of political and economic risk. Some critics also see home bias, arguing that the three US-based agencies tend to rate the US too highly. A related complaint is regional bias, with agencies said to under-rate whole regions such as Latin America and Africa. The agencies reply that a record of past defaults helps forecast future ones, and that Latin America’s record is a long one.

Each agency adapted the template it had built for companies, and although the details differ, the processes share a great deal. Five elements define what a sovereign rating is and how it is produced.

What the rating measures

A sovereign rating addresses the government’s creditworthiness toward private creditors, meaning bondholders and private banks, and not toward official creditors such as the World Bank or the IMF. Agencies differ on scope: some rate only how likely default is, others also how severe it would be. S&P aims its ratings at the likelihood that default happens, not necessarily at how bad it will be. Moody’s ratings reflect both the probability of default and the expected loss severity, which it expresses through the expected recovery rate. All the agencies count two kinds of event as default: an outright default, which means missing an interest or principal payment when it falls due, and a restructuring default, in which debt is rescheduled, exchanged or otherwise restructured.

The determinants of the rating

S&P publishes the variables it weighs when rating a country, which span political, economic and institutional factors. They are grouped into the nine areas below; Moody’s and Fitch use their own lists but give similar weight to economic, political and institutional detail.

Source: S&P Global, summarised.

The rating process

The analyst primarily responsible for a country drafts a rating recommendation and report. A rating committee of five to ten analysts then debates each analytical category and scores it; after closing arguments, the committee votes on the final rating.

Notch-up and notch-down approaches

Since each sovereign usually receives a local currency and a foreign currency rating, the agencies need a rule linking the two. The notch-up approach treats the foreign currency rating as the main gauge of the sovereign’s credit risk and raises the local currency rating above it on the strength of domestic debt market factors. Under the notch-down approach, the local currency rating is the anchor and the foreign currency rating is set lower to reflect foreign exchange constraints. Either way, the size of the gap depends mainly on monetary policy independence. Countries with floating exchange rates that fund themselves in deep domestic markets show the widest gaps, while countries that have surrendered monetary independence, through dollarization or a monetary union, see their local currency rating converge on their foreign currency rating. This is consistent with the Latin American table, in which Ecuador, El Salvador and Panama, all of which use the US dollar, carry only a foreign currency rating.

Reviews and updates

Ratings are reviewed on a regular schedule and also in response to news. A coup or an economic disaster can trigger a review not only of the country concerned but also of its neighbours, which may suffer contagion. Gill (2015) gives a vivid account of how this process worked, flaws included, through the ratings assigned to the UK government in the 1970s.

An early criticism of sovereign ratings was that they barely moved. Ratings have become more active in recent years, and the clearest way to measure how much they change is a ratings transition matrix. Each row starts from a rating class and shows the percentage of sovereigns in that class that ended up in each class, including default and withdrawal of the rating, after a fixed period. S&P’s matrices for its foreign currency sovereign ratings cover 1975 to 2021.

The full one-year row for BBB rated sovereigns shows how a single row works. The percentages add up to 100% (to rounding) because every sovereign that starts the year at BBB must end it somewhere.

Source: S&P Global.

Reading every row in full is tiring, so the next two tables condense each row into five derived figures: the share upgraded to any higher class, the share unchanged, the share downgraded to a lower non-default class, the share that defaulted, and the share whose rating was withdrawn. The diagonal values and the default and not-rated columns are exactly as printed; the upgrade and downgrade columns are sums of the printed cells.

Source: S&P Global; upgrade and downgrade columns derived.

Source: S&P Global; upgrade and downgrade columns derived.

Over one year, stability dominates: a AAA rated sovereign kept its rating 96.82% of the time, and every class from AAA to B stayed put in more than 86% of cases. Over fifteen years the picture changes completely. Only 67.14% of AAA sovereigns and 25.25% of BBB sovereigns still held the same rating, and the default columns climb steeply as the starting rating falls, from 0.00% for AAA to 63.64% for CCC/CC. The printed fifteen-year row for A shows a higher default share (8.28%) than the BBB row (8.21%), a reminder that small samples of sovereigns make individual cells noisy.

The agencies’ central claim is that their sovereign ratings are good measures of default risk. They accept that some individual countries have been misjudged but argue that ratings and subsequent defaults are strongly correlated. The cleanest test is a table of cumulative default rates: the percentage of sovereigns in each rating class that defaulted within a given number of periods of being rated.

Source: S&P Global. Selected rows and horizons; the printed table runs from AAA to CC and from 1 to 15 periods.

A word on units. The printed table labels its horizon in months, and the commentary that accompanies it repeats the word, saying for example that a BBB sovereign has a 1.99% chance of default within 5 months. These figures come from S&P’s long-run default study, whose horizons are measured in years, and the first-period rates are of the same order as the one-year default cells of the transition matrix (0.00% for BBB, 0.38% for BB and 2.42% for B, against 0.00%, 0.10% and 2.27% here). The lesson therefore reads the horizon as years; the figures themselves are unchanged.

Read this way, the evidence supports the agencies. A AAA rated sovereign has never defaulted within fifteen years of its rating. A BBB rated sovereign has a 1.99% chance of default within 5 years, 3.31% within 10 years and 4.78% within 15 years. Investment grade sovereigns as a group default far less often than speculative grade ones at every horizon, and the CCC classes default at very high rates almost immediately. Moody’s and Fitch publish similar tables with similar messages, so on average the agencies have delivered what they promise. The ranking is not perfectly smooth class by class, though: the printed A row reaches 10.96% at fifteen periods, more than double the BBB figure, which again reflects the small number of sovereigns behind each cell.

Despite that broadly good record, sovereign ratings have drawn five lines of criticism.

Upward bias. Agencies are accused of being too optimistic about both corporate and sovereign borrowers. For companies the usual explanation is the conflict of interest created when issuers pay for their ratings. That argument is weaker for sovereigns, because the fees a government pays are small next to the reputational damage an agency would suffer among rating users if it were seen to over-rate countries. Herd behaviour. When one agency raises or lowers a sovereign, the others tend to follow, which reduces the value of having three agencies, since their views are no longer independent. Too little, too late. Banks setting loan rates and investors pricing bonds need timely risk assessments, and agencies have long been criticised for changing ratings slowly and too late to protect investors from a crisis. Vicious cycle. Once a crisis is under way, agencies are perceived to overreact, cutting ratings so far that they feed the crisis they are describing. Ratings failures. When an agency changes a sovereign’s rating several times within a short span, it is in effect admitting that the original rating was wrong.

Bhatia’s record of ratings failures

Bhatia (2004) studied sovereigns whose S&P or Moody’s ratings were changed several times within a year between 1997 and 2002. The table lists the rating judged to have failed, the corrected rating, the number of notches between them and the main factor behind the change.

Source: Bhatia (2004). Notch direction shown in words.

Most of these failures cluster around the Asian crisis of 1997 and the Russian default of 1998, and the dominant cause is the sudden loss of foreign exchange reserves. The Korean case is the most dramatic: S&P cut Korea ten notches in two months, then raised it four notches the following year when reserves were rebuilt.

Why ratings fail

Bhatia offers four explanations. Information problems: agencies get most of their data from the governments they rate, the quantity and quality vary widely, and governments may release good news while sitting on bad news, which could explain the upward bias. Limited resources: sovereign ratings generate modest revenue and must at least cover their costs, so agencies cannot hire many analysts. In 2003 each analyst was estimated to rate four to five sovereign governments, and some argue that such workloads push analysts to rely on common information and to herd. Revenue bias: because sovereign ratings are given away free to most users, revenue has to come from issuers or from related business. Agencies earn significant fees from rating sub-sovereign issuers such as regions and cities, whose ratings typically fall after a sovereign downgrade, so those issuers lobby against sovereign downgrades, again tilting ratings upward. Other incentive problems: analysts occasionally move to jobs with governments they rate, which is too rare to be a serious conflict, but the agencies’ other businesses, such as market indices, performance evaluation and risk management services, may be lucrative enough to influence ratings.

As sovereign borrowing shifted from bank loans to bonds in the 1980s and 1990s, the rating business grew, and so did a second source of information. Bond prices, and the yields they imply, give a measure of sovereign default risk that updates continuously as trading happens.

The idea is simple. Take a bond issued by a government in a foreign currency, for example US dollars, and compare its yield with the yield on a riskless bond in the same currency and of the same maturity. If the US Treasury is assumed to be default free, the difference is the market’s price for the extra default risk of the issuing government.

On 1 July 2022, Brazil’s 10-year dollar bond yielded 6.19% and the 10-year US Treasury yielded 3.02%, so Brazil’s default spread was 6.19% minus 3.02%, or 3.17%. (If the Brazilian yield is rounded to 6%, the spread becomes 2.98%.) The table repeats the calculation for several emerging markets, ordered from widest to narrowest spread.

Source: Bloomberg. Rows reordered by spread.

Spreads and ratings are positively correlated, with B2 rated Turkey paying the widest spread and A1 rated Chile the narrowest, but the market measure has two advantages. It is more granular: Colombia and Indonesia share a Baa2 rating, yet the market charges Colombia a noticeably wider spread. It is also more dynamic, because it moves in real time rather than waiting for a committee.

Brazil and Venezuela show how fast the market can separate two borrowers. In December 2005 their spreads were close, at 3.18% for Brazil and 3.09% for Venezuela. Over 2006 to 2009 Brazil’s spread fell to 1.32% while Venezuela’s widened to 10.26% by December 2009.

The measure has a precondition: a default-free bond must exist in the same currency as the sovereign bond. Local currency bonds of different governments cannot be compared directly, because differences in their yields may reflect differences in expected inflation rather than default risk. Even for dollar bonds, the method works only because the US Treasury is assumed to be default free.

Is the spread a better predictor than the rating?

Market spreads react quickly to new information, so they tend to flag trouble earlier than ratings. The price of that speed is noise: spreads are much more volatile than ratings and can move for reasons unrelated to default, such as changes in liquidity or investor demand. Studies of spreads as default measures reach three broad conclusions. Spreads are mostly correlated with both ratings and eventual default, since low-rated sovereign bonds trade at much higher yields and also default more often. The bond market leads the agencies, with spreads usually widening before a downgrade and narrowing before an upgrade. And a rating change is still news that moves prices when it happens. So it would be wrong to dismiss ratings: bond markets draw on ratings when pricing, just as agencies draw on market data when revising ratings.

A credit default swap lets an investor buy protection against default by a borrower. The protection buyer pays the seller a periodic premium, called the spread, expressed as a percentage of the face (notional) amount protected. In exchange, the seller promises to make the buyer whole if the bond’s issuer, the reference entity, suffers a credit event: failing to pay, restructuring, or going bankrupt. Alongside the growth of CDS written on companies, a market in sovereign CDS has developed, and its prices give a constantly updated market view of each country’s default risk.

When a credit event occurs, the seller can settle in one of two ways. Under physical settlement, the buyer delivers the defaulted bond to the seller and receives its par value. Under cash settlement, the seller pays the buyer the difference between par and the bond’s market price after the event, a price that reflects expected recovery from the issuer. Either way, the buyer is protected against credit losses over the life of the contract.

Two weaknesses of the protection

The protection has limits. First, it is triggered only by a credit event; if the bond’s price collapses without a failure to pay or restructuring, the buyer receives nothing. Second, the protection is worth no more than the seller’s own creditworthiness. If the seller defaults, the insurance fails, and on the other side the buyer may stop paying the premiums it has promised.

Market background

J.P. Morgan is credited with the first CDS, which it created after lending Exxon a $4.8 billion credit line and passing the credit risk to investors. The market then grew rapidly. The notional value of securities covered by CDS passed $60 trillion by end-2007, before the crisis pushed it back to about $39 trillion by December 2008.

The market can be divided by type of reference entity. Corporate CDS form the bulk, bank CDS come next, and sovereign CDS are a small share. For all its notional size, the market is narrow: a few investors account for most trading. Banks hedging their default exposure dominated at first, and portfolio managers and speculators joined later, but the number of active players remains small relative to the market’s size. That narrowness is a weakness, since the failure of one or two big players can throw spreads into turmoil, as the collapse of Lehman Brothers did for several weeks in 2008.

If counterparty risk and liquidity are set aside, CDS prices should be an up-to-date measure of a reference entity’s default risk, adjusting to new information in a way that infrequently revised ratings cannot. Greece in the years 2006 to 2010 illustrates the point. Monthly CDS spreads for Greece moved throughout the period, while ratings from Fitch stayed fixed for most of it and moved only late in 2009 and in 2010, when the downgrades came. Both market measures, the CDS spread and the bond default spread, were reassessing Greek default risk every month as information arrived.

What the evidence says about CDS spreads

Showing that CDS spreads are timelier than ratings is easy; the harder question is whether they predict default better. Four findings stand out. Changes in CDS spreads lead changes in both sovereign bond yields and sovereign ratings, giving earlier warning of downgrades (Ismailescu, 2007). Although the debate is not settled, evidence suggests that changes in sovereign CDS spreads predict sovereign default events better than ratings do (Rodriguez, Dandapani and Lawrence, 2019). Sovereign CDS spreads rise when economic policy uncertainty increases, and those rises come with currency depreciation. Finally, CDS prices cluster: groups of countries tend to move together, and one study identifies six clusters of emerging markets.

Difference column derived.

In cluster 1, for instance, the average correlation between member countries is 0.516, while its average correlation with countries outside it is just 0.210. The derived column shows that the gap is narrowest for cluster 3, at 0.124, and widest for cluster 2, at 0.376, so cluster 2’s members move together most distinctively.

Limits of CDS as a default measure

CDS prices carry two built-in distortions. Exposure to counterparty and liquidity risk means prices can move for reasons unrelated to the reference entity: much of the late-2008 spike in CDS prices, in the fourth quarter, came from Lehman’s failure and the surge in counterparty concerns that followed. Related to that, the narrowness of the market leaves individual contracts exposed to liquidity squeezes that move prices. Even so, changes in CDS prices do carry useful, if noisy, information about shifts in default risk. What the evidence does not yet show is that CDS spreads are better than bond default spreads at assessing that risk.

Summary of the evidence discussed in this lesson.

The sensible conclusion is that the three measures complement each other. Ratings provide a stable, widely understood anchor and still move prices when they change; bond spreads add granularity and speed wherever a default-free benchmark exists; CDS spreads add another real-time signal that tends to lead the other two, provided you discount the noise from counterparty and liquidity effects.