A 100-year perspective on sovereign debt composition in 13 advanced economies
S. M. Ali Abbas, Laura Blattner, Mark De Broeck, Asmaa El-Ganainy, Malin Hu 27 October 2014
There’s been renewed interest in sovereign debt because the Global Crisis, but relatively little attention has been paid to its composition. Sovereign debt may vary with regards to the currency it really is denominated in, its maturity, its marketability, and who holds it – and these characteristics matter for debt sustainability. This column presents evidence from a fresh dataset on the composition of sovereign debt in the last century in 13 advanced economies.
Why sovereign debt composition matters
Academic, policy, and market interest in sovereign debt has spiked because the 2008 Global Crisis. Researchers have sought to put the post-Crisis synchronised build-up in sovereign debt ratios in advanced economies within a longer-term/historical context, drawing comparisons with debt surges through the Great Depression, debt consolidations in the aftermath of World War II, and more. 1
However, this literature has largely abstracted from a discussion of sovereign debt ‘composition’, which, both theory and experience reveal, is central to questions of debt sustainability/management, optimal taxation, monetary policy, and even financial regulation. 2 For example, an inability to issue term debt in local currency is often referred to as ‘original sin’ (Eichengreen and Hausmann 2002, Eichengreen et al. 2003), which drives the high cost/riskiness of (higher yields on) emerging-market sovereign debt. Arslanalp and Tsuda (2012) have linked the holder profile of sovereign debt to countries’ capability to weather financial market stress. Reinhart and Sbrancia (2011) highlight the amenability of non-marketable debt to ‘liquidation’ during periods of financial repression. The recent wave of European sovereign debt crises has underscored the need for debt maturity – countries with longer debt duration registered lower sovereign risk premia than others, despite higher debt and deficit levels (Abbas et al. 2014a).
The largest obstacle to an effective historical treatment of sovereign debt composition, inside our view, is a insufficient data. Debt structure data are simply just not available within an easy to get at format across countries over long stretches of time. Our recent paper (Abbas et al. 2014c) seeks to handle this data gap. Using official sources for individual countries – in addition to some published cross-country datasets – we assemble a debt structure database spanning the time 1900-2011 and 13 advanced economies: Australia, Austria, Belgium, Canada, France, Germany, Ireland, Italy, holland, Spain, Sweden, the united kingdom, and the united states.
The dataset slices the sovereign debt pie along four dimensions: currency (foreign vs. local); maturity (local currency debt subdivided into short-term and medium-to-long-term); marketability; and holders (non-residents, national central bank, domestic commercial banks, and the others). Because that is an initial effort, we still report numerous gaps inside our data, which reflect, generally, having less coverage on certain types of debt (such as for example debt held by non-residents) or during certain stress periods (such as for example all over the world Wars). Notwithstanding the gaps, the dataset offers insights in to the evolution of debt structure in the last 11 decades, including around major sovereign stress events.
What the historical debt structure data reveal
A bird’s eye view of key debt composition ratios as time passes offers some intuitive patterns.
Around 90% of advanced economies’ debt is and was denominated in local currency. Still, six of the 13 countries saw the forex debt share go above 50% at some time – post-WWI France and Italy are notable cases in point.
The share of local currency medium-to-long-term debt altogether debt has averaged 68% over the sample (or three-fourths of local currency debt) and exhibits an intuitive pattern – governments issued longer-dated paper in memories and compensated for the bigger riskiness of their debts during bad times by shortening issuance maturities. Cross-country variation in the maturity structure of debt mirrors differences in, inter alia, a country’s vulnerability to fiscal/military crises, reserve currency status, and debt management preferences.
Before WWI, virtually all central government debt was issued by means of marketable securities. The share of such securities declined during post-WWI consolidations, before falling precipitously (to only around 55%) after and during WWII – a time characterised by financial repression and captive financial markets. The share recovered starting in the mid-1970s and today stands at about 80%.
The holder profile data reveals the average share with debt for national central banks around 10%, 3 and almost double that for domestic commercial banks. Moreover, both shares seem to be substitutes for a lot of the 1920-1970 period – central banks were clearly picking right up the tab from commercial banks (and other holders) around/after WWII. Both shares fell in tandem following the 1970s, as just non-resident participation in sovereign debt markets soared. The non-residents’ share of debt increased from 5% to 35% over 1970-2011, and reflected several underlying factors: financial innovation and globalisation; stronger sovereign debt management; independent central banks focused on low inflation; the introduction of the euro, which resulted in the de facto elimination of currency risk within the Eurozone; and the accumulation of ‘safe’ foreign sovereign securities by emerging Asian economies, especially China.
Figure 1 . Central government debt composition by instrument and currency
Figure 2 . Holder composition of central government debt
Source: Authors’ calculations.
We provide a far more ground-level view of how debt composition of individual sovereigns taken care of immediately major episodes of debt accumulation and consolidations during 1900-2011. We recover some broad common patterns.
Large debt accumulations – essentially, large increases with debt supply – were typically absorbed by increases in short-term, foreign currency-denominated and banking-system-held debt (Table 1 has an illustration from WWI).
Table 1 . World War I: Maturity and currency (% of GDP for debt/GDP; otherwise, % of central government debt)
This pattern, however, didn’t hold through the debt build-ups starting in the 1980s and 1990s, that have been compositionally skewed toward long-term local-currency marketable debt (see Table 2 below). We attribute this change to similar factors to the ones that drove higher non-resident demand for sovereign paper. Included in these are the emergence of a big (global) contractual savings sector with long investment horizons, and independent central banks focused on low inflation, thus providing confidence that governments wouldn’t normally inflate away the responsibility of long-term nominal debt burden.
Table 2 . THE FANTASTIC Accumulation: Maturity, holders, and marketability (% of GDP for debt/GDP; otherwise, % of central government debt)
On debt consolidations, we find support for the financial repression-cum-inflation channel through the post-WWII episodes, but can also show that channel didn’t operate with the same intensity everywhere. Table 3 shows two country groups, categorised according to whether debt maturities lengthened or shortened through the consolidations (debt maturity can be used here as a proxy for voluntary investor demand). In the first group, where maturities shortened, resort to central bank financing and inflation was higher, and your debt ratio fell by typically 152% of GDP. In the next group, where maturities lengthened, your debt reduction was only slightly more modest (145% of GDP), but inflation and central bank financing involvement were a lot more modest.
Table 3 . Post-World War II consolidation: Maturity, holders, and marketability (% of GDP for debt/GDP; otherwise, % of central government debt)
What this means for debt reduction ahead
The foregoing shows that governments have pursued two broad types of debt reduction strategies during the past: a normal approach focused on preserving long debt durations, accompanied by fiscal consolidation and, in some instances, moderate inflation (although moderate in earlier decades will be considered high today); and an unconventional strategy that involved reliance on debt holdings (often non-marketable) by fully- or semi-captive domestic investors, accompanied by high inflation, and at the price of shortening debt maturities. Given the existing environment of low growth and politically challenging fiscal adjustments, it bears asking if the unconventional strategy is feasible.
Table 4 . Sovereign debt composition, past and present (% of GDP for debt/GDP; otherwise, % of central government debt)
Inside our own view (never to be mistaken with the view of the IMF), the pre-requisites for pursuing such a technique – e.g. lack of capital controls, insufficient competing investment opportunities – aren’t met. With 36% of the advanced economy debt now held by non-residents (weighed against 3% at end-WWII), and only 21% held by the domestic bank operating system (weighed against 46% at end-WWII), sustaining involuntary demand for sovereign debt offering sub-market returns will be difficult (Table 4).
Similarly, recourse to surprise inflation (with or without financial repression) above conservative central bank inflation targets to dent the worthiness of already-issued medium-to-long-term securities will probably present challenges: (i) the existing ‘lowflation’ environment in lots of advanced economies renders engineering significant inflation surprises challenging; (ii) the globally-accepted regime of price stability (as the anchor for monetary policy) will not quite enable surprise inflation; (iii) the latter could entail significant economic costs, as governments would either need to accept permanently higher inflation with direct costs to efficiency and investment, or, if indeed they wished to go back to low inflation, accept the inevitably painful disinflation process (e.g. Fischer 1994, Bordo and Orphanides 2013); and (iv) higher inflation may likely entail a permanent hidden ‘cost’ when it comes to departure from a less risky (long-duration) debt structure, or a lesser degree of sustainable debt (Aguiar et al. 2014).
What this means for further analytical focus on sovereign debt
Further work is required to develop models that may cope with the observed ‘shifts’ in sovereign debt management objectives as time passes our debt composition data capture. For example, until WWI, sovereign debt management was targeted at debt issuance for specific and temporary purposes, such as for example to finance a war. After WWII, the focus shifted to reducing the debt-to-GDP ratio, including through financial repression and inflation policies. And in recent decades, debt management objectives have already been largely cast regarding minimising expected debt servicing costs, at the mercy of an acceptable degree of refinancing risk, which translated right into a shift towards longer-term debt denominated in local currency held by a diversified investor base.
More work can be needed on what and whether debt structure might help predict crises. A short analysis predicated on the paper’s data shows that changes in your debt composition typically thought to increase contact with crisis risk, such as for example maturity shortening, indeed could have these consequences. Similarly, governments’ willingness to default (outright) on the debt could be different if a lot of your debt were held by non-residents instead of domestic investors (especially banks).
Disclaimer: The views expressed herein are those of the authors and really should not be related to the IMF, its Executive Board, or its management.
Abbas, S M A, N Belhocine, A El-Ganainy, and M Horton (2011), “Historical Patterns and Dynamics of Public Debt – Evidence From a fresh Database”, IMF Economic Review 59(4): 717-742.
Abbas, S M A, N Arnold, M De Broeck, L Forni, and M Guerguil (2014a), “The Other Crisis: Sovereign Distress in the Euro Area”, in C Cottarelli, P Gerson, and A Senhadji (eds.), Post-Crisis Fiscal Policy, Cambridge, MA: MIT Press: 193-226.
Abbas, S M A, N Belhocine, A El-Ganainy, and A Weber (2014b), “Current Crisis in Historical Perspective”, in C Cottarelli, P Gerson, and A Senhadji (eds.), Post-Crisis Fiscal Policy, Cambridge, MA: MIT Press: 161-191.
Abbas, S M A, L Blattner, M De Broeck, A El-Ganainy, and M Hu (2014c), “Sovereign Debt Composition in Advanced Economies: A Historical Perspective”, IMF Working Paper 14/162.
Arslanalp, S and T Tsuda (2012), “Tracking Global Demand for Advanced Economy Sovereign Debt”, IMF Working Paper 12/284.
Aguiar, M, M Amador, E Farhi, and G Gopinath (2014), “Sovereign Debt Booms in Monetary Unions”, American Economic Review, forthcoming.
Bordo, M and A Orphanides (eds.) (2013), THE FANTASTIC Inflation: The Rebirth of Modern Central Banking, Chicago: University of Chicago Press.
Eichengreen, B and R Hausmann (eds.) (2002), Debt Denomination and Financial Instability in Emerging Market Economies, Chicago: University of Chicago Press.
Fischer, S (1994), “The expenses and great things about disinflation”, in J O de Beaufort Wijnholds, S Eijffinger, and L Hoogduin (eds.), A Framework for Monetary Stability, Dordrecht: Kluwer Academic Publishers.
Reinhart, C and K Rogoff (2009), THIS TIME AROUND differs: Eight Centuries of Financial Folly, Princeton: Princeton University Press.
Reinhart, C and K Rogoff (2013), “Financial and Sovereign Debt Crises”, IMF Working Paper 13/266.
Reinhart, C and B Sbrancia (2011), “The Liquidation of Government Debt”, NBER Working Paper 16893.
Sbrancia, B (2011), “Debt and Inflation throughout a Amount of Financial Repression”, Job Market Paper, Department of Economics, University of Maryland College Park.
1. See, for instance, Reinhart and Rogoff (2009, 2013), Sbrancia (2011), Reinhart and Sbrancia (2011), Abbas et al. (2011), and Abbas et al. (2014b).
2. Notable exceptions are Sbrancia (2011) and Reinhart and Sbrancia (2011), who look at debt maturity and marketability structure for 12 economies (a variety of emerging and advanced) over 1948-1980. However, this dataset isn’t yet public.
3. As a share of GDP, central bank holdings averaged about 5%, with a peak of 19% during WWII and a recently available post-Crisis uptick reflecting large-scale asset purchase programs (as in the united kingdom and the united states).