When sovereign debt makes the news, it is usually bad. You are unlikely to see the headline “Venezuelan Bond Market Running Smoothly” in The New York Times or Wall Street Journal. And that is the way it has always been.
The first global sovereign debt boom and bust, which included the Greek international bond (surprise, surprise), ended in panic in 1825. After 1826, Greece refused to service debts issued by the militias fighting for independence. More generally, the history of sovereign debt appears to be a history of default, repudiation, and limited debt enforcement. Why, then, do investors keep going back?
A recent NBER working paper by Josefin Meyer, Carmen M. Reinhart, and Christoph Trebesch goes beyond the headlines and popular narratives. The authors collected data on global sovereign debt markets going back to its humble beginning when a dozen new republics in Latin America sought financing in London after the Spanish War of Independence (1808-1814).
The paper’s major contribution is to calculate total sovereign bond returns across the globe for two hundred years. In particular, they calculate the return of privately-held foreign debt, such as when a German company holds Greek government bonds. With that data in hand, the authors show there is not much of a puzzle; the returns on sovereign bonds compensate investors for the risks.
The average real yearly return on a global portfolio of external sovereign bonds was 6.8%, which is about 4% higher than the risk-free benchmark of the US or UK government bonds over the same time horizon. Excess returns are driven by high coupons. Moreover, the risk-return properties are in line with other tradable assets, such as US and UK equities over the same time horizon. The debt of riskier countries earns a higher expected return, as expected. To visually demonstrate this relationship, the authors plot the number of defaults vs. expected return, while separating out 60 “serial defaulters”, countries that have defaulted on their external debt at least twice since 1815 and/or who were in default for protracted spells.
While calculating bond returns in 2018 is simple, calculating returns for two hundred years across many countries is not. First, you need the time-series data on prices. The authors collected monthly price quotations of 1,400 foreign-currency bonds issued and traded in London and New York with a total of 219,968 observations covering 91 countries. That step is relatively straightforward.
But that’s not enough since defaults and debt restructurings are a major issue for this asset class. Moreover, unlike organized, corporate debt restructuring within the United States under Chapter 11, external sovereign debt defaults can affect each bond in a unique way with no overarching principle of who gets paid what when. Even in the same episode, not all investors take the same haircut. To calculate returns you need bond-level default data; the authors painstakingly collected data on missed payments, renegotiations, and face value write-downs in 313 debt crisis episodes in 91 countries since 1815.
Since coupons are the main driver of total returns, the authors must measure coupon payments accurately. This is especially difficult during default spells. But they uncovered something interesting: sovereigns in default on principal payments often continued to service coupons in full or in part. That pushes up investor returns. However, coupon payments on the same bond can vary wildly over the timeline of the crisis. With all those complications, the fine-grained data that the authors collect is what ultimately allows them to build monthly total return series for each bond in a consistent manner.
In addition to resolving the puzzle of why investors are willing to continually reinvest in risky sovereign debt and providing immense data returns, the paper paints a picture of these markets over time that is interesting in itself. For example, the authors are able to look into the history of haircuts closely. Almost all of the defaults were resolved by exchanging old debt into new debt at a discount. Over the 200 years, the average haircut was 44%. Each restructuring is plotted below by year on the x-axis and the size of haircuts on the y-axis is averaged across all instruments involved and the size of the circles represents the inflation-adjusted amounts of debt affected by the restructurings (in real 2009 US$).
A few things are worth pointing out. First, haircuts are common throughout the whole time period, with the notable except of the Bretton-Woods period between WW2 and the 1970s. This is because there was very limited cross-border lending, so there was not a need to restructure privately-held debt. Second, there has been a dramatic increase since the 1980s, but the number of countries has also increased over that same time period.
Their full dataset makes a 200-year version of the widely used Emerging Market Bond Index (EMBI) by JP Morgan, at a monthly frequency that people can analyze at the global-, country-, and bond level. Building such datasets is a significant contribution that will greatly improve the quality of research on sovereign debt. The sovereigns will still have their own issues.