What is the interest rate premium – over and above the spread warranted by economic fundamentals – charged to governments that re-enter international capital markets after settling on defaulted debt? How rapidly does such a ‘default premium’ decay? Answers to these questions are important for gauging fiscal sustainability and the welfare of many countries. Yet, answers have been anything but consensual.
One answer is that such a premium has been historically small. A string of studies finds that countries that honoured their debts through the 1930s depression and World War II years benefited very little – if at all – in terms of lower spreads in subsequent decades relative to those that defaulted.1
Another answer, based on emerging market data from the 1990s onwards, is that the average default premium is very large upon market re-entry (around 400 basis points) but not significant after a couple of years (Borenzstein and Pannizza 2009), except for a few cases where exceptionally sizeable losses or ‘haircuts’ have been imposed on lenders (Cruces and Trebesch 2013).
Does the average investor really forget all that quickly after having forgiven the non-repayment on large chunks of her loans? The disparate results mentioned above call for a reappraisal of the broad historical evidence. Our new new research (Catão and Mano 2015) uses far more comprehensive historical data and – unlike previous studies -- the same measurement methodology across periods and countries to reappraise the evidence on the size and persistence of the sovereign default premium. A key finding is that the default premium is very sizeable in the first years after a country re-enters international capital markets (between 250 and 400 basis points) and it decays more slowly than previously found. This indicates that investors do not forget defaults so easily. Another main finding is that the longer a government remains in default, the higher the default premium. This provides motivation for avoiding delays in debt renegotiation.
Our reappraisal of the previous evidence rests on three pillars:
- A more general metric of the default premium that nests specifications used in previous studies;
- Crucially, a new and broad dataset built on extensive research with primary and secondary sources;
The new dataset encompasses and expands those of previous studies. It spans both advanced and emerging countries for two active periods of international bond markets – 1870-1938 and 1970-2011, for a total of about 3,000 annual observations of market interest rates paid by borrowing governments in international capital markets.2
- A broader set of indicators to measure the macroeconomic situation of a country and the memory of investors regarding its credit history.
We also consider the state of global financial markets and, following Cruces and Trebesch (2013), the size of past defaults (as measured by the actual ‘haircut’ on investors). These indicators include many standard indicators of macroeconomic ‘fundamentals’ (such as debt to GDP ratios, economic growth, etc.) as well as alternative empirical gauges of how investors’ memories decay.
We obtain a measure of default premium by running a panel regression of countries’ interest rate spreads over the global risk free rate of a similar maturity on a variety of macroeconomic fundamentals and on the following indicators of credit history:
- The ratio of the number of years that a country was in default to the total number of years in the sample up to the current year (call it MEM1);3
- The number of years since the last default (call it MEM2);
- Whether the year in question corresponds to the first, second, third, fourth or fifth year after the debt renegotiation. This is meant to capture the possibility that investors’ memories don’t decay linearly – for instance, they may decay more slowly in the first years after debt restructuring.
Our measure of the default premium is the sum of these three components, i.e. it is the component of the sovereign spread, over and above the country’s current macroeconomic fundamentals, that is due to credit history. The novelty is to use those three terms listed above together rather than separately, as in prior studies. In Catão and Mano (2015), we discuss how this metric is consistent with the view that investors’ information about a government’s account and motivations. In this case, investors try to gauge the missing information bits by observing the government’s main actions, such as borrowing patterns, prompt repayment or default, and renegotiation delays. The default premium should thus reflect the additional information contained in credit history behaviour, over and above the information contained in publicly available statistics on the country’s macroeconomic fundamentals.
Our econometric analysis shows that all memory indicators are jointly significant and that the memory effects are proportionally stronger in the aftermath of debt settlement, leading to a higher than usual premium in the first couple of years after the country resumes private borrowing. Thereafter, the decay in the premium is gradual. This is shown in Figure 1. Upon market re-entry, the default premium in the pre-World War II period is 250 basis points. This is substantially higher than earlier estimates (see Flandreau and Zummer 2004). It is also persistent – five years after debt settlement, the mean default premium is still 150 basis points. The respective estimates for the post-World War II default premium are even higher. Upon market re-entry we obtain 400 basis points, which is similar to Borenzsten and Panizza (2009) and about 150 basis points higher than Cruces and Trebesch (2013). Like both studies, the significance of the post debt settlement year effect drops from the second year. Yet, in our econometric model, the effects of the other memory variables (MEM1 and MEM2) then kick in, making the decay slower: five years after debt settlement, the default premium stands at 200 basis points.
Figure 1. Average default premium in the pre- and post-World War II samples
Note: Solid line is the mean default premium; dashed lines are 1.96 standard errors of the mean.
Importantly, Figure 1 shows a non-trivial dispersion around the mean default premium, as per the dotted lines that plot the 95% standard-error bands. This dispersion reflects differences in countries’ credit histories. A further probe into the data reveals that this is due in large part to the difference between one-time and serial defaulters. If we were to plot default premium estimates for individual countries, serial defaulters would show up closer to the upper standard error bands. In other words, serial defaulters typically pay a higher-than-average default premium. These estimates are robust to many changes in econometric specification. One of them is to assume that, instead of linearly forgetting past, investors discount the past in hyperbolic or quasi-hyperbolic fashion. This yields no lower default premium. The other is to add the number of defaults – as opposed to the total time in default gauged by the first memory variable (MEM1) — as an explanatory variable. The latter is not significant. So, what matters for the default premium is the total number of years a country is in default in its relevant history. This implies that a serial defaulter pays a higher default premium because it is in default during a lengthier fraction of its credit history – it also implies that delays in debt crisis resolutions carry a future interest cost.
Overall, the default premium accounts for up to 60% of the variations in the sovereign spread (see Figure 2). That is, much of the overall interest rate paid by a sovereign when it returns to private capital markets is not due to the state of its observable macro fundamentals but to faulty credit history.
Figure 2. Decomposing changes in spreads: fundamentals and default premium (pre-World War II sample on the left, post-World War II sample on the right)
How about the effect of the size of past defaults or ‘haircut’ on the default premium? While it may be surprising to hear, we find that the actual size of past haircuts is not a significant determinant of the average default premium. In theory, this is not so surprising since all defaults have a haircut and by measuring the average default premium, we are automatically measuring the default premium for the average haircut. Beyond that, there are good reasons for why the size of past defaults may not matter for the default premium, going forward. This is because investors in competitive bond markets only care about the expected future haircut. Actual past haircuts – which vary widely across countries and epochs4 – would matter if defrauded lenders consistently decide to punish the borrower by charging an extra premium to re-coup their losses without being undercut by new lenders. This is not an easy feat in competitive international markets with lots of lenders. Importantly, we also find that nearly 90% of the actual haircut is explained by variables in principle known to investors and already included in our regression model – notably debt-to-GDP and the length of the default. So, much of the actual haircut is already ‘expected’. Including an indicator of actual past haircuts thus adds little value to the estimation of the default premium. This is not necessarily inconsistent with evidence in Cruces and Trebesch (2013) that countries inflicting higher-than-average haircuts on investors face a higher and more persistent default premium. This could be the case if exceptionally high past haircuts convey information that is not contained in other indicators about the future probability of default and its size. Be that as it may, the main takeaway is that the default premium is sizeable and persistent even for defaults with average haircuts.
The interest rate premium on past default has been underestimated. This is partly due to narrower credit history indicators and, crucially, to the narrower data coverage of previous studies. Once we correct for these problems, a sizeable and persistent default premium emerges, and one which rises on the duration of the default. This means that the longer a country that stays in default the higher the premium it will pay once it resumes borrowing from private capital markets.
A main lesson is that international investors do not seem to forget easily their forgiveness of past debt so easily. The findings reported here also help rationalise why governments try hard not to default and, if they do, why an early renegotiation is sought.
Authors’ note: The views expressed here are those of the authors and do not necessarily represent those of the institutions to which they are affiliated.
Benczur, P and C Ilut (2015), “Evidence for Relational Contracts in Sovereign Bank Lending”, Journal of European Economic Association, forthcoming.
Borensztein, E and U Panizza (2009), “The Costs of Sovereign Default”, IMF Staff Papers 56(4): 683-741.
Bussière, M and M Fratzscher (2006), “Towards a New Early Warning System of Financial Crises”, Journal of International Money and Finance 25(6): 953-973.
Catão, L A V and G M Milesi-Ferretti (2014), “External Liabilities and Crises”, Journal of International Economics 94(1): 18-32.
Catão, L A V and R Mano (2015), “Default Premium”, IMF working paper WP/15/167.
Cruces, J and C Trebesch (2013), “The Price of Haircuts”, American Economic Journal: Macroeconomics 5(3): 85-117.
Eichengreen, B, and R Portes (1986), “Debt and Default in the 1930s: Causes and Consequences”, European Economic Review 30: 599-640.
Flandreau, M and F Zumer (2004), “The Making of Global Finance 1880-1913”, Paris: OECD.
Gourinchas, P-O, and M Obstfeld (2012), “Stories of the Twentieth Century for the Twenty-First”, American Economic Journal: Macroeconomics 4(1): 226-65.
Jorgensen, E, and J Sachs (1989), “Default and Renegotiation of Latin American Foreign Bonds in the Interwar Period”, in B Eichengreen and P H Lindert (eds.), The International Debt Crisis in Historical Perspective, Cambridge: 48-85.
Kaminsky, G and P Vega-Garcia (2014), “Varieties of Sovereign Crises: Latin America, 1820-1931”, NBER Working Paper 20042.
Lindert, P H, and P J Morton (1989), “How Sovereign Debt Has Worked”, in J Sachs (ed.) Developing Country Debt and Economic Performance, Cambridge: MA.
Ozler, S (1993), “Have Commercial Banks Ignored History”, American Economic Review 83: 608-20.
Reinhart, C and C Trebesch (2014), “A Distant Mirror of Debt, Default and Relief”, NBER Working Paper 2057.
Sturzenegger F, and J Zettlemeyer (2006), Debt Defaults and Lessons from a Decade of Crises, Cambridge, MA.
1 Eichengreen and Portes (1986), Lindert and Morton (1989), and Jorgenson and Sachs (1989) find no benefit. Ozler (1993) and Benczur and Ilut (2015) find a small reduction in spreads (about 25 basis points at mean).
2 For the reasons advocated in Bussière and Fratscher (2006), Gourinchas and Obstfeld (2012) and Catão and Milesi-Ferretti (2014), these 3000 data points eliminate the years when a country is in default. If one chooses to include those, our sample is even larger.
3 The definition of default used here is the same Standard and Poor’s criterion used in most studies, i.e. default is as either a unilateral interruption of repayment of contractual interest and/or principal, or an event where old debt is swapped by new debt with a net present value loss. A default is classified as ended when a settlement with lenders on (all or most) outstanding arrears is reached and no further near-term resolution of creditors' claims is likely.
4 See Sturzenegger and Zettelmeyer (2009), Reinhart and Trebesch (2015), and Kaminsky and Vega-Garcia (2015).