VoxEU Column International trade

Why sovereigns repay debts to external creditors and why it matters

With Greece on the edge and Argentina confronting adverse legal decisions, thinking about sovereign default is a front-burner issue. This column discusses how recent experience casts light on relative merits of the two main conceptual approaches to modelling sovereign lending and default – the reputation approach and the punishment approach. 

The ongoing Greek and Argentine debt negotiations underscore the crucial importance of understanding the relative bargaining positions of debtor countries, official institutions, creditors and the international community at large.

Yet, in academic research, the most fundamental question of all remains unresolved: what can foreign creditors really do to a sovereign country that does not repay its debts in full?

After all, the very nature of sovereignty is precisely that creditors cannot easily come in and seize assets inside a country, as they might do with a private company in bankruptcy. One might think that after centuries of experience with sovereign external defaults, and tens of dozens of examples, a clear and coherent answer would have emerged. But, perhaps surprisingly, there is still a wide range of views, and an even larger chasm between theory (see Aguiar and Amador 2014) and practice.

We aim to discuss how recent experience casts light on a debate that has been dormant since the late 1980s and early 1990s, and we hope to point the way forward to new research.1 In a related piece that acts as a companion to this column, we dissect the flow of net repayments by Greece to illuminate existing approaches as well as important gaps in our understanding.

Academic thinking on sovereign default

The literature can be divided into two rough branches:

  • The ‘reputation approach’ pioneered by Eaton and Gersovitz (1981) which builds on Hellwig (1977); and
  • The ‘direct punishments’ bargaining-theoretic approach of Bulow and Rogoff (1988b, 1989a) which in turn builds on Cohen and Sachs (1986).

For theorists, the institution-free Eaton and Gersovitz approach is much more popular (one measure being that has twice the number of Google Scholar citations).

The Eaton-Gersovitz reputation approach

The basic Eaton-Gersovitz premise is simple:

Countries value access to international capital markets because it allows them to smooth consumption in the face of volatile output and/or fluctuating investment opportunities. They can be trusted to repay loans in good times because they know some day bad times could return, and they don’t want to lose their reputation as reliable debtors.

In a reputational equilibrium, there is no need for international courts, or political pressures. One does not need to understand how International Monetary Fund programs are enforced. Conveniently, none of it really matters.

Instead, sovereign borrowing limits (on external debt) and the flow of repayments depend only on the basic calculus of consumption smoothing, a phenomenon that plays a central role in most modern macroeconomic models. If creditors extend too much debt to a country, there may come a point where the benefits to retaining capital market access, with its concomitant consumption smoothing benefits, simply do not justify sticking to the debt repayment program. Of course, this ‘incentive compatibility’ constraint is most likely to be hit in a period where, in order to maintain its reputation, the debtor is called upon to make an exceptionally large repayment.

In such a world, a country’s debt limit depends on factors such as the volatility of its output and its level of risk aversion. Volatility is observable and although risk aversion is not, at least it is a subject that has been widely studied.

There are myriad refinements of the Eaton and Gersovitz model, perhaps the most significant being the much less well-known work of Grossman and Van Huyck (1988), who extend the analysis to allow for partial default. This refinement is crucial, because in virtually every real-world case, default is always partial not complete, even if a payment comes years after the initial default. Importantly, in the Grossman-Van Huyck model, legal default is a formality and of no consequence in itself. An implicit contract governs the equilibrium, with the required payment being higher in good times. The debtor never cares about any external threats. The sole concern is maintaining reputation for repayment. It is willing to pay more when output is high simply because it has decreasing marginal utility of consumption, so the immediate pain of any given repayment is less than when output is low.

There is a large empirical literature on trying to estimate the Eaton and Gersovitz model, with a particularly influential branch stemming from the work of Aguiar and Gopinath (2006). As in all similar consumption smoothing models à la Lucas (1987), it is difficult to obtain large costs of capital market autarky (although introducing ‘rare disasters’ can help, as Barro (2009) shows). 

Thus most applied models add a direct punishment cost, very much in the Cohen and Sachs tradition. In fact, in the empirical applications, it is typically the direct punishment and not the reputation component that drives the debt limits. Aguiar and Gopinath (2006) appear to be the first to make this point is a detailed calibrated model2 (for a recent application see Schreger and Du 2015). Nevertheless, within this framework it may be theoretically possible to test whether fluctuations in debt capacity or changes in risk premiums demanded of borrowers can be explained by changes in a country’s reputation for repayment.

The reputational approach to sovereign default has the undeniably attractive feature of only requiring modelling skill rather than messy institutional knowledge to participate. But it suffers a number of rather fundamental empirical flaws, beyond the problem of rationalising realistically large costs. We shall discuss these later, but first turn to the leading alternative, the direct punishments approach.

The Bulow-Rogoff punishment approach

Bulow and Rogoff (1988b, 1989a) assume, as in Eaton and Gersovitz, that foreign creditors have no meaningful legal rights to repayment in debtor country courts.3 But in contrast to Eaton and Gersovitz, creditors are assumed to have rights in foreign creditor-country courts. This assumption accords well with the fact that the critical legal decision in sovereign debt contracts almost invariably take place in foreign courts, typically in New York or London, the most popular jurisdictions for sovereign debt contracts, and that contracts tend to be negotiated with specific waivers of sovereign immunity.4

Superficially, there are strong empirical parallels with the pure reputation approach, especially as one of the most direct and practical legal rights a creditor has is to threaten to cut off an uncooperative debtor from future lending. Instead of repayment being enforced by a ‘supergame’ equilibrium, it depends on legal rights. For example, if Argentina defaults on a loan to Citibank, and then proceeds to borrow from Bank of America, the later lender risks having any repayment claimed by the prior lender. Thus, countries that default will have an incentive to settle old debts as a prelude to borrowing again.

In principle, creditor legal rights can allow for other forms of commercial interference, ranging from making trade more difficult (through threats to seize shipments) to interfering with trade credits and trade insurance. Such actions, can, in principle, substantially interfere with a country’s gains from trade. There is a significant question as to whether creditors can really make threats to interfere with a country’s international commerce and trade credible, especially when the creditor must bear substantial costs to implement them. Importantly, this is precisely the kind of issue that a bargaining theoretic framework allows one to approach.

Differences between the approaches

Although the reputation and direct punishments approaches might seem to offer broadly similar predictions, in fact there are some very fundamental differences:

  • Practically speaking, perhaps the most important is that the direct punishment/bargaining approach lends itself very naturally to incorporating moral hazard;

Bulow and Rogoff (1988) show that because protracted debt renegotiations can harm third parties (in their model because of reduced gains from trade), the debtor country and its lenders can extract side-payments, a phenomenon that will be quite familiar to followers of IMF bailouts. The issue of moral hazard and the bargaining for side payments by third parties (e.g. the IMF or German taxpayers) are acutely illustrated by the recent Greek drama. Of course, in principle, anything can be embedded in a reputation model, including moral hazard. But economic models are only of use if they are falsifiable. And the standard reputation for the repayment model that is so popular in the literature fails on many counts.

  • As already noted, reputation models suggest that the governing law of the debt is irrelevant;

Yet, the experiences of both Greece and Argentina show that jurisdiction is extremely important. When the Troika of the IMF, ECB and European Commission decided to force a haircut on Greece’s private creditors it was able to impose a 53.5% write-down on debt governed by Greek law. But the Troika was unable to force a significant write-down on debt governed by British law. It is no surprise that private creditors demanded that any new Greek debt arising from the ‘private sector involvement’ (PSI) be created under European law so as to make unilateral default much more difficult.

  • In standard reputation for repayment models, write-downs are decided unilaterally –creditors’ particular concerns do not really matter;

This does not appear to coincide with experience.

  • The interests and welfare of unrelated third parties does not matter in standard reputation models;

But in reality, outside parties are invariably drawn in to debt renegotiations. While again it is possible to incorporate moral hazard into reputation models, the issue does not flow naturally and the whole approach is not fundamentally falsifiable.

  • Excessive write-downs reduce a country’s ability to borrow money because they hurt its reputation;

The Greeks and Argentines are hoping that write-downs will clear the way for them being able to borrow more money sooner. The bigger the write-down, the sooner they expect to become heavy borrowers of new money again.

  • Reputational debtors borrow in bad times and re-pay in good times, for purposes of income smoothing;

Defaults, if they are to take place, occur in good times since the country is due to receive money in bad times.5 In some models the country is unable to save, and so borrowing in bad times and repaying in good times is the only way to smooth.

In reality, many countries borrow as much as they can whenever they can. They can borrow more when their economy is stronger, and so lenders feel that the country has greater potential to be squeezed for repayments. Debt crises occur when countries do badly and creditors decide they want to reduce their loan exposure. To some extent, this issue can be addressed by assuming that income shocks are permanent and not transitory, but it remains difficult to rationalise country borrowing only on the threat of lost consumption smoothing.6

  • Creditor identity doesn’t matter;

Whether a country owes money to a hedge fund or to a neighbouring government, its reputation is determined simply by whether or not it repays. In fact, because private creditors care strictly about getting their money and official creditors have broader goals, countries are better off when official creditors hold their debts. In practice, official creditors may take a long time to write down debts, but in the interim they will require little, sometimes less than zero, in the way of repayments. This has largely been the case for Greece.7

  • In a consumption smoothing model the ability of a country to borrow will increase if its income is more volatile;

In one example, Barro (2009) points out that the cost of a rare disaster could considerably raise the cost of capital market autarky (and so the ability to borrow in a pure reputation model). Again, the prediction would be that countries would essentially invest in insurance contracts that would cause them to pay creditors in most states but to receive large inflows in crises. In a direct punishment model it is easier to borrow against stable income.

  • The reputation model does not lend itself well to analysing debt reduction schemes falling outside the usual debtor-lender relationship, and cannot really explain them;

This is to a large extent a corollary of not being able to seamlessly incorporate third parties and moral hazard.

A classic example is the analysis of debt buybacks in which a country goes to the open market and buys back its debt at discount, sometimes using funds donated by a third party, sometimes through related mechanisms such as debt for equity swaps. Bulow and Rogoff (1988b, 1991) show how one can analyse these issues within a direct punishments framework. They show that in contrast to the case where a company devotes excess cash to a debt repurchase, sovereign debtors typically lose – and their creditors as a group gain – from sovereign debt buybacks at market prices. Negotiated swaps such as the Greek PSI deal are easy to explain in the direct punishments framework, but require far more assumptions and contortions in a reputation model.

There are also more subtle viability issues with the reputation approach. In Bulow and Rogoff (1989b), we showed that if creditors truly have no legal recourse in the event of a default, then debtors always have the option of putting savings abroad rather than repaying creditors, as a form of self-insurance. Under surprisingly general assumptions, the existence of this option leads to the unravelling of any purely reputational equilibrium. The implication is that legal enforcement must ultimately lie at the heart of debt repayments, and any analytical framework needs to take this into account (there have been a great many further papers since our original one, for discussions see Aguiar and Amador 2014).

Broader considerations

Our description of both the reputation and direct punishment models is admittedly narrowly construed. Indeed, Bulow and Rogoff (1989b) show that if a debt repayments model is embedded in a broader domestic or international relations framework, then other enforcement mechanisms may be at work; their analysis has been significantly extended by Cole and Kehoe (1996).8 There are many other important nuances, such as the extent to which creditor country reputation also matters. Clearly northern European countries understand that whatever deal they cut for Greece could impact later negotiations with Ireland and Portugal.

One can particularly see the importance of other factors at play in the case of Greece, where the threat of expulsion from the EU is perhaps the greatest threat Greeks face. In principle, it is not possible to exit the single currency without also breaking the Treaty that underpins Greek membership in the EU with all its advantages in trade and economic integration. How negotiators would enforce the Treaty in practice is an unknown. Arguably, the fact that Greece so clearly faces broader costs of default than a typical emerging market debtor helped convince creditors to lend it far more than it otherwise might have been able to borrow. Relatedly, the collateral damage to other Eurozone nations of Grexit is also large, persuading creditors that moral hazard side payments might be exceptionally large as well, as indeed seems to be confirmed by the analysis in our companion piece.

Another important issue, and one that lies at the heart of the bargaining framework of Bulow and Rogoff (1988, 1989b), is that in practice, sovereign debt renegotiations focus very much on the flow of repayments, and much less on how the stock of debt evolves. This is precisely because all sides realise that any future promises can be renegotiated. In essence, their model is exactly a framework to understand the ‘extend and pretend’ approach officials have adopted in the case of Greece, which we treat in a companion piece. There are many practical subtleties we cannot explore in this short note, such as creditors being willing to take smaller repayments now in exchange for structural reform that makes the country more inclined to make larger payments in the future, for instance.

Testing the theories

Is there any way to decisively test what the right approach is to modelling sovereign debt? Probably not, as in truth, sovereign debt renegotiations surely involve more players and more dimensions than simple analytical models can easily express, not to mention factors like national pride (in more utilitarian terms, one can imagine that Greek citizens might be willing to trade off some cutting of pensions and government employment in return for having their bank accounts supervised by the ECB instead of the Greek government).

Perhaps the biggest breakthrough in the debate on why countries repay is the very recent case study of Hebert and Schreger (2015). They consider how Argentine debt and stock prices reacted to some quite surprising and arguably very unusual US court rulings on Argentine debt, which came out far more in favour of creditors than most observers had expected (this might have been because Argentine negotiators were exceptionally unwilling to adopt conventional compromises, leading the judge to become frustrated and rule against them). Using the risk-neutral probability of default inferred from credit default swap spreads, Hebert and Schreger show that expectations of default on sovereign external Argentine debt jump markedly in response to decisions that were unexpectedly and dramatically favourable to creditors. This is hardly what one would expect in a pure reputation world where legal rulings are irrelevant. Unfortunately, theoretical elegance and economic reality do not always coincide, and popular as the clean reputation for repayment model may be, the messy direct punishments approach just might be more relevant. For sure, the issues are more complex than any current approach can fully incorporate.


Aguiar, M, and G Gopinath (2006), “Defaultable debt, interest rates and the current account”, Journal of International Economics 69(1), 64–83.

Aguiar, M and M Amador (2014), “Sovereign Debt” in G Gopinath, E Helpman and K Rogoff (eds.), The Handbook of International Economics, Volume 4.

Arellano, C (2008), “Default Risk and Income Fluctuations in Emerging Economies”, The American Economic Review, American Economic Association 98(3): 690-713, June.

Barro, R J (2009), “Rare disasters, asset prices, and welfare costs”, The American Economic Review 99(1): 243–64.

Buchheit, L C (2013), “Sovereign Debt Restructurings: The Legal Context”, BIS Paper No. 72s, July.

Bulow, J, and K Rogoff (1989b), “Sovereign Debt: Is to Forgive to Forget?”, The American Economic Review: 43–50.

Bulow, J and K Rogoff (1989a), “A Constant Recontracting Model of Sovereign Debt”, The Journal of Political Economy 97: 155–178.

Bulow, J and K Rogoff (1988b), “Multilateral Negotiations for Rescheduling Developing Country Debt: A Bargaining-Theoretic Framework”, International Monetary Fund Staff Papers (35): 644–657, December.

Bulow, J and K Rogoff (1992), “Sovereign Debt Repurchases: No Cure for Overhang”, Quarterly Journal of Economics 106: 1219–35, November.

Bulow, J and K Rogoff (1988a), “The Buyback Boondoggle”, Brookings Papers on Economic Activity 2: 675–698.

Cohen, D and J Sachs (1986), “Growth and External Debt under Risk of Debt Repudiation”, European Economic Review 30: 529-60.

Cole, H L and P J Kehoe (1996), “Reputation Spillover Across Relationships with Enduring and Transient Beliefs: Reviving reputation Models of Debt”, NBER Working Papers 5486, National Bureau of Economic Research, Inc.

Eaton, J, and M Gersovitz (1981), “Debt with potential repudiation: Theoretical and empirical analysis”, Review of Economic Studies 48(2): 289–309.

Grossman, H and J Van Huyck (1988), “Sovereign Default as a Contingent Claim: Excusable Default, Repudiation, and Reputation”, The American Economic Review 78: 1088-1097, December.

Hebert, B and J Schreger (2015), “The Costs of Sovereign Default: Evidence from Argentina” mimeo, Harvard University.

Hellwig, M F (1977), “A Model of Borrowing and Lending with Bankruptcy”, Econometrica 45(8): 1879-1906, November.

Lucas, R E (1987), Models of business cycles, New York: Basil Blackwell.

Schreger, J and W Du (2015), “Sovereign Risk, Currency Risk, and Corporate Balance Sheets”, mimeo, Harvard University.

Weidemaier, W M C, and M Gulati (2013), “Sovereign Debt and the ‘Contracts Matter’ Hypothesis” in L C Buchheit (ed.), Sovereign Debt Restructurings: The Legal Context, BIS Paper No. 72s, July and also in Oxford Handbook of Law and Economics (forthcoming).


1 The authors are grateful to Jesse Schreger for helpful comments on an earlier draft.

2 See also Arenello (2008) where to get the model to sustain realistic levels of debt, it is necessary to assume that the costs of default are much higher in good states (when output is high) than in bad states (when output is low), much as the direct punishments literature.

3 In these papers we simplify the analysis by assuming that only the sovereign borrows and that all of its debt is external.

4 See Weidemaier and Gulati (2013).

5 As in Grossman and Van Huyck (1988), authors will sometimes divide up the net transfer between debtor and creditor into ‘repayment of existing debt’ and ‘new loans’, and say that the implicit contract allows for ‘defaults’ in bad states when the country is a net receiver of money (i.e. required repayments are less than instantaneous new loans). But it is only in the good states that the borrower has any incentive not to honor the implicit contract. Even if a country has a high discount rate a sustainable reputational equilibrium requires it to receive money in some states, and with consumption smoothing the country will be receiving money (and so not even have the opportunity to default) in bad states.

6 For example, Mexico’s foreign debt increased by over 1000% in the 1970s and early 1980s even as the country experienced an unprecedented boom in current and permanent income due to oil discoveries and price increases. Theoretically the borrowing increase could be explained by a surge in wise public sector investment opportunities that outstripped the growth in current income, justifying the additional borrowing, but the Echeverria and López Portillo years are not generally considered a model of fiscal probity.

7 As another example, the private investors in Ukrainian debt are looking to be paid in full through a bailout by official lenders. Private creditors insist on pari pasu clauses to prevent countries from giving new private creditors seniority but they welcome new official lenders.

8 For example, the US might lose from defaulting on its sovereign debt because of both domestic and international considerations even though much of it is held by other countries.

8,714 Reads