Debt with Potential Repudiation: Theoretical and Empirical Analysis

Jonathan Eaton and Mark Gersovitz, Review of Economic Studies 1981. Source note: no freely accessible copy of the article exists (it is gated at JSTOR, and no working-paper version could be legitimately obtained), so unlike the other entries in this reading list this piece is written from the model as it is standardly presented in the sovereign-debt literature — the treatments in Uribe and Schmitt-Grohé’s textbook and the Aguiar–Amador survey tradition — not from the original text. No numbers from the 1981 paper are quoted. A public lecture by Manuel Amador and Mark Wright covering this framework’s logic accompanies the Tomz–Wright entry in this list.

Every loan contract on earth rests on something the lender can do to the borrower if the borrower walks away. A mortgage has the house; a corporate bond has the bankruptcy court. A loan to a sovereign government has — what, exactly? The sovereign writes its own laws, its assets sit largely inside its own borders, and no court can attach next year’s GDP. Taken seriously, this observation says international lending to governments shouldn’t exist. It does exist, in size, and has for centuries, punctuated by defaults that are painful but weirdly non-fatal. Eaton and Gersovitz’s 1981 paper is where economics first built a coherent equilibrium around this fact, and nearly everything the field has done on sovereign debt since — including three later papers in this block — is a footnote, a rebuttal, or a computer implementation of it.

The mechanism: reputation as collateral

The idea is that the collateral is the future relationship. A country that repudiates its debt is punished not by seizure but by exclusion: it loses access to international credit markets. For an economy hit by fluctuating income, that access is valuable — it is the ability to smooth consumption, borrowing in bad harvests and repaying in good ones. So each period the sovereign weighs the cost of repaying today against the value of being able to borrow tomorrow, and repays exactly when

where the default value reflects life in financial autarky. Three properties fall out of this comparison, and they organize the whole literature. First, default is more tempting when debt b is large — obviously. Second, and this is the model’s signature, default is more tempting when income y is high enough that insurance is less needed or, in the standard quantitative implementations, when income is low and repayment most painful — which of these dominates depends on the income process, a subtlety that becomes the entire subject of Tomz and Wright’s paper two entries ahead. Third, lenders understand all of this in advance.

That third point is where the paper earns the “theoretical” in its title. Competitive lenders will never knowingly lend into certain default; instead they impose a credit ceiling — a maximum debt level at which the sovereign’s incentive to repay still binds — and price loans to break even given the default risk below that ceiling. Debt limits, in other words, are not an exogenous imperfection bolted onto the market (like the ad hoc borrowing constraints of the small-open-economy models in the first block of this syllabus); they are the equilibrium expression of the enforcement problem. The market for sovereign debt exists exactly up to the point where the borrower’s own self-interest can be trusted, and no further. When the closing-devices paper by Schmitt-Grohé and Uribe (entry three) treated the debt-elastic interest premium as a technicality with a coefficient of 0.0007, this is the economics that the later literature would load onto that same object at full strength — and Restrepo-Echavarría’s endogenous-borrowing-constraints paper later in this block is precisely the project of replacing the technical device with the Eaton–Gersovitz-style limit and seeing what changes.

Why the paper mattered, and what it left open

Before 1981, international lending was analyzed as if to a firm; the sovereign’s option to say no was an embarrassment handled by assumption. Eaton and Gersovitz made willingness-to-pay, rather than ability-to-pay, the binding constraint, and they did it in a full rational-expectations equilibrium where lenders’ beliefs about default discipline the contract set. The empirical half of the original paper — an econometric attempt to detect credit ceilings in 1970s developing-country borrowing — is, candidly, the less-cited half; the model is the legacy.

Two open flanks defined the next four decades. The first: is market exclusion actually a credible and sufficient punishment? A defaulter might be able to self-insure — save in foreign assets rather than borrow — and replicate much of what credit access provided. This is the precise crack that Bulow and Rogoff’s 1989 paper, next in this list, drives a wedge into: they prove that under certain conditions reputation alone cannot support any debt, so something else — direct sanctions, seizure of trade, legal harassment — must be doing the enforcement. The reputational and the sanctions views of sovereign debt have been contesting the field ever since. The second flank: quantification. The modern default literature — Aguiar–Gopinath and Arellano in the mid-2000s, and the Hamann–Méndez-Vizcaíno–Mendoza–Restrepo-Echavarría resource-curse paper in this very block — is exactly the Eaton–Gersovitz model solved numerically with persistent income shocks, calibrated default costs, and equilibrium bond prices q(b′, y) that make the credit ceiling a smooth schedule rather than a cliff. When today’s papers say “an Eaton–Gersovitz-style model,” they mean: discrete default option, value-function comparison, exclusion-plus-output-cost punishment, and lenders pricing the whole thing in.

It is a mark of the paper’s depth that its two headline descendants pull in opposite directions — the quantitative literature took the reputation mechanism and made it the workhorse, while the theory literature took Bulow–Rogoff’s critique and spent decades rebuilding reputation on firmer ground (partial exclusion, saving frictions, information, reentry). The 1981 paper survives both, because what it really established is prior to either: sovereign debt is a market whose supply curve is made of incentives, and every quantity and price in it must be read as an answer to the question “why would they ever pay this back?”