odious debt

8/6/2019 Odious Debt

1/11

Odious Debt

By SEEMA JAYACHANDRAN AND MICHAEL KREMER*

Trade sanctions are often criticized as ineffective because they create incentives forevasion or as harmful to the target countrys population. Loan sanctions, incontrast, could be self-enforcing and could protect the population from beingsaddled with odious debt run up by looting or repressive dictators. Governmentscould impose loan sanctions by instituting legal changes that prevent seizure ofcountries assets for nonrepayment of debt incurred after sanctions were imposed.This would reduce creditors incentives to lend to sanctioned regimes. Restrictingsanctions to cover only loans made after the sanction was imposed would help avoidtime-consistency problems. (JEL F34, O19, K33, P16)

Trade sanctions against dictators are oftenineffective because third parties have incentivesto break them. Even when trade sanctions bite,they can hurt not only the sanctioned regime,but also the people subject to the regime.

We argue that loan sanctionslimiting sanc-tioned governments ability to borrowwouldbe less prone to these problems. Suppose, forexample, that the United Nations Security

Council unanimously declared that any futuredebt incurred by a particular dictator would beconsidered illegitimate and nontransferable tosuccessor regimes. Suppose also that the UnitedStates and the European Union implementedlegal changes to prevent assets of the successorregime from being seized to enforce repaymentof the dictators debts. We argue that thiswould create incentives for lenders in thirdcountries to avoid lending to the dictator, and

could potentially eliminate equilibria with ille-gitimate lending.

Loan sanctions are likely to help the popula-tion of the sanctioned country. Both trade andloan sanctions may hurt the population in theshort run. Loan sanctions, however, create along-run benefit for the population by prevent-ing it from being saddled with debts run up bythe dictator to finance looting or repression.

Our analysis is related to the legal doctrine ofodious debt, which holds that debt should not betransferable to successor regimes if (a) it wasincurred without the consent of the people and(b) was not for their benefit (Alexander N. Sack,1927; Ernst Feilchenfeld, 1931).1 The underly-

* Jayachandran: University of California, Berkeley, 140Warren Hall, MC 7360, Berkeley, CA 94720, and Depart-ment of Economics, UCLA (e-mail: [email protected]); Kre-mer: Department of Economics, Harvard University, 200Littauer Center, Cambridge, MA 02138, The BrookingsInstitution, Center for Global Development, and NationalBureau of Economic Research (e-mail: [email protected]). We thank Charles Cohen, Jishnu Das,

Raquel Fernandez, Rachel Glennerster, Robert Litan, BenOlken, Anne-Marie Slaughter, Detlev Vagts, Peyton Young,several seminar participants, and three anonymous refereesfor helpful comments. We also thank David Clingingsmith,Jessica Leino, and Dan Wood for excellent researchassistance.

1 Other work on odious debt includes Patricia Adams

(1991) and Joseph Hanlon (2002), who argue the case thatmuch developing-country debt is illegitimate, and AshfaqKhalfan et al. (2003), who discuss related legal issues.Thomas Pogge (2001) proposes that a panel assess thedemocratic status of governments in order to deter lendingto autocratic regimes. James M. Buchanan (1987) arguesthat even in democratic polities, future generations may nothave an obligation to repay debts contracted by earliergenerations if the debts were intended to support the earliergenerations consumption rather than to provide lastingbenefits, unless such debts might have been agreed to in ahypothetical contract entered into behind a veil of igno-

rance. Geoffrey Brennan and Giuseppe Eusepi (2002) argueagainst Buchanan. We differ from previous work in discuss-ing the multiple equilibria of the debt market, the deterrenceof lending to dictators through elimination of creditorsincentives to issue these loans, and the tradeoffs between exante and ex post rulings.

82


2/11

ing principle is that just as an individual doesnot have to repay money that someone fraudu-lently borrows in her name, and a corporation isnot liable for contracts that its chief executiveofficer enters into without authority to bind thefirm, a country should not be responsible for

debt that was incurred without the peoples con-sent and was not used for their benefit. Thedoctrine arose after the Spanish-American Warwhen the United States contended that neitherthe United States nor Cuba should be responsi-ble for debt that Cubas colonial rulers had runup in Cubas name. The concept attracted con-siderable attention in 2003 when the Secretaryof the Treasury and other senior U.S. officialssuggested that debts incurred by Saddam Hus-

sein should perhaps be considered odious andnot the new Iraqi governments obligation torepay.2

Yet this doctrine remains a minority viewamong legal scholars, and U.S. policymakerseventually backed away from the odious-debtrationale when arguing for debt relief for Iraq.This is largely out of concern that the concept ofodious debt could prove a slippery slope. Coun-tries could claim that previous debt was odious

as an excuse to renege on legitimate debt. Moregenerally, any adjudicating body that had thepower to declare debt void might nullify legit-imate debt if it placed a high value on thewelfare of the debtor country. If creditors antic-ipated being unable to collect on legitimateloans, the debt market would shut down. Weargue that this time-consistency problem couldbe addressed if loan sanctions applied only tofuture debt contracted by a country, not existing

debt.Loan sanctions would be effective only in

certain cases, since many potential target re-gimes run their country into the ground eco-nomically and cannot borrow in any case.Others, probably including Iraq during the timeof Saddam Husseins borrowing, would be pro-tected from sanctions by the major powers. Insome cases, however, loan sanctions could havebeen a potent addition to the international com-

munitys toolkit of sanctions. For example, in1985 the United Nations Security Council im-

posed trade sanctions on South Africas apart-heid regime, but the regime continued to borrowfrom private banks through the 1980s. In 1997,when Franjo Tudjman of Croatia instigated vi-olence against political opponents and lootedpublic funds, the major powers cut off Interna-

tional Monetary Fund (IMF) lending to Croatia.Commercial banks nonetheless lent an addi-tional $2 billion to the Tudjman governmentbefore his death in 1999. If the major powershad imposed the type of loan sanctions we de-scribe in this paper on apartheid South Africa orTudjmans regime, creditors might have ceasedgranting loans to those regimes, and the popu-lations would not bear the debt today.

The decision whether to impose loan sanc-

tions on a particular regime would inevitably besubjective, and a concern is that the interna-tional community or individual countries wouldimpose them unjustly. These are important is-sues, though not ones unique to loan sanctions;these issues also pertain to trade sanctions, mil-itary force, and the other measures in use today.This papers objective is to describe loan sanc-tions and to show why, when applied againstrepressive or looting regimes, they may be more

effective and beneficial to the people than ex-isting sanctions.

The remainder of this paper is organized asfollows. Section I presents the model and dis-cusses equilibria in the absence of either loan ortrade sanctions. Section II compares loan andtrade sanctions. Section III discusses extensionsof the model. Section IV argues that decisionsabout which loans are illegitimate and should beunenforceable are subject to time-consistency

problems and therefore should be taken ex ante.Section V concludes.

I. Model of Sovereign Debt and Odious Regimes

Modeling loan sanctions requires an underly-ing model of why sovereign debt is repaid inthe first place. In this section we embed theanalysis of odious debt in a standard reputa-tional model of borrowing. Later we discuss

loan sanctions in a broader class of reputationalmodels, and in models in which debt repaymentis enforced with the threat of trade sanctions,exclusion from foreign assistance, or other sim-ilar penalties.2 Wall Street Journal, April 30, 2003.

83VOL. 96 NO. 1 JAYACHANDRAN AND KREMER: ODIOUS DEBT


3/11

A. Setup of the Model

Production.Suppose there is a set of coun-tries each with a population normalized to 1. Ineach period t the government allocates laborLF

(t) to harvesting fruit and 1 LF

(t) to build-

ing palaces. Fruit production is A(t)LF(t) if har-vesters are paid at least w w ; if w w theharvesters are too sick to work. A(t), the pro-ductivity stream, is uneven. For some of thecountries, A(t) a if t is odd and A(t) A aif t is even. For the others, A(t) A if t is oddand A(t) a if t is even. This unevennessgenerates a reason for either borrowing or sav-ing, namely to smooth consumption. We focuson countries where first-period productivity is

low: as discussed below, such countries will preferto borrow in period 1 and continue with a patternof borrowing in odd periods and repaying in evenperiods. We assume a w to ensure that fruitproduction is sufficient to support harvesting, evenin bad years. Fruit is nonstorable.

Fruit can be bought and sold on internationalmarkets at price 1. A large number of countries,which do not face the uneven productivitystream, comprise the international markets.

These countries also competitively producemarble, issue loans, and offer savings accounts,as discussed below.

Palaces are produced in integer amounts us-ing imported marble with price P

Mand well-fed

workers who can maintain their concentra-tion. Marble is supplied by Bertrand competi-tors who can produce at cost P

M. Production is

int(min{(1 LF

(t))/, M/}) if builders arepaid at least wW w and 0 otherwise, where

M is the amount of marble imported, is thenumber of workers and amount of marbleneeded to build one palace, and the operatorint returns the largest integer less than orequal to the argument.

Credit and Savings Markets.Loan con-tracts are as follows. A creditor lends anamount d(t) 0 in period t, and the countryis expected to repay d(t)R in period t 1. A

country also can place assets in a foreignsavings (demand deposit) account that earnsinterest rate R.

We assume assets that a country holdsabroad can be seized by creditors to enforce

debts.3 As shown by Jeremy Bulow and Ken-neth Rogoff (1989), without this provision,reputation could not sustain sovereign bor-rowing. At some point the country would bebetter off reneging on its debt, saving thefunds that would have been used to repay the

debt, and using them to smooth its consump-tion in the future. Our assumption that fruitis nonstorable implies there are no domesticsavings (more realistically, domestic invest-ments may be less attractive than diversifyinginternationally).

Consumption.Citizens utility u is con-cave in fruit consumption and additively sepa-rable over time with discount rate . Citizens

receive no utility from palaces.In the beginning of the first period, a potential

dictator faces a utility cost of taking powerdistributed according to F(c) where F(0) 0, sothat given expected utility from being dictatorof V, a dictator takes power with probabilityF(V). Let G {odious, nonodious} be the gov-ernment type in period 1, where G odious ifthe dictator takes power. For simplicity we as-sume governments are always nonodious in

subsequent periods.A nonodious government maximizes the pop-

ulations discounted utility. Dictators maximize

3 In the United States, the Foreign Sovereign ImmunitiesAct of 1976 gives creditors this right by denying sovereignimmunity in claims related to commercial activity; a cred-itor from any country has the right to seize a foreigngovernments assets held in the United States for breach ofa lending contract. Similar laws are in place in most devel-

oped countries where debtor countries would consider plac-ing their savings.

The laws applicability to sovereign debt was upheld inthe 1992 Supreme Court case ofRepublic of Argentina et al.v. Weltover. Argentina issued bonds repayable in NewYork, among other places. When the bonds began to maturein 1986, Argentina extended the time for repayment. TwoPanamanian corporations and a Swiss bank brought thisbreach of contract action to the District Court in New York.The District Court, the Court of Appeals, and then theSupreme Court ruled in the plaintiffs favor.

Other countries have similar provisions. For example,

the European Convention on State Immunity, the StateImmunity Act 1978 of the United Kingdom, and Article 5 ofthe Draft Articles on Jurisdictional Immunities of States andTheir Property of the International Law Commission, whichrepresents the consensus view in international law, denystate immunity for commercial activity (acta jure gestionis).

84 THE AMERICAN ECONOMIC REVIEW MARCH 2006


4/11

their own utility v(p, f), which is increasing inpalace ownership and fruit consumption. Weassume preferences for palaces are satiated atone palace so that v(p, f) ffor p 1 and v(p,f) f for p 1. We also assume (P

M W a w ), or that an odious gov-

ernments utility from palace ownership out-weighs the input costs and forgone fruitproduction, implying that it constructs a palace,if possible.

These assumptions about v(p, f) generate abenefit to the population from an odious gov-ernments trading and possibly borrowing. Pre-venting an odious regime from importingmarble deprives part of the population of thewage premium W w associated with palace

construction, as does a loan sanction if an odi-ous government needs to borrow to pay for apalace (which is the case if a (P

M W)/

(1 ) w ).

Timing.At the beginning of period 1, thedictators takeover cost is realized, and the dic-tator decides whether to take over, determiningG. Governments may enter loan contracts. Next,the government chooses a wage for the popula-

tion and allocates the populations labor. Thensimultaneously production occurs, the popula-tion receives its wage, the government receivesany surplus production, and marble and fruitare bought and sold. Fruit is consumed. Thegovernment may make a debt repayment. Sub-sequent periods are identical except the govern-ment is always nonodious.

B. Equilibria and the Status Quo of the

Sovereign Debt Market

This subsection discusses the multiple equi-libria of the model with a focus on trigger-strategy equilibria that can support lending. Wethen characterize what we call the status quoequilibrium, that is, the equilibrium in themodel that best describes the actual sovereigndebt market today. We argue that in the statusquo, creditors lend to odious governments just

as they lend to nonodious governments, andsuccessor governments repay debt they inherit,be it odious or not.

We consider subgame perfect Nash equilibriaof the model. The folk theorem implies there are

multiple equilibria in this type of repeated lend-ing game. In one equilibrium, creditors neverlend: if they did, the governments would neverrepay their loans. In other equilibria, creditorslend to countries that have never defaulted, andcountries repay since failure to do so would

exclude them from future borrowing. Nonodi-ous governments are able to smooth consump-tion by borrowing in odd-numbered periods andrepaying in even periods. They maximize u(a d(t)) u(A d(t)/) for odd t, trading offincreased consumption at present with de-creased consumption in the next period. Sincecredit markets are competitive, the interest ratewill be R 1/. The solution to the first-ordercondition, u(a d(t)) u(A d(t)/), de-

fines their optimal odd-period loan amount D (A a)/(1 ), which will enable the coun-try to consume (aA)/(1 ) in each period.Odious governments always want to borrow asmuch as possible since they do not care aboutfuture repayment.

We assume that nonodious governments areable to borrow their optimal amount. That is, Dis less than the maximum loan that can besupported through reputation, denoted D* and

defined implicitly by u(A D*/) u(a D*) u(A) u(a). The term on the left is thediscounted utility for an even period and thenext period if the country repays, and the termon the right is if the country defaults and lives inautarky. We also assume that (P

M W)

(1 )(a w ) D, which ensures that gov-ernments can borrow enough to satisfy odiousgovernments desire for palaces. In the inequal-ity, (P

M W) is the input costs for a palace,

(1 )(a w ) is one of the governmentssources of money to pay those input costs,namely profits from the fruit sector, and D is itssecond source of money, a loan.

Define a reputational equilibrium as a sub-game perfect Nash equilibrium in which on theequilibrium path creditors lend in even periodsan amount D

Oif the government is odious and

DN

if the government is nonodious where DO

,DN D*, and are repaid 1/ times that amount

from the successor government the followingperiod. If any player has ever deviated from thisbehavior, creditors refuse to lend and govern-ments borrow if possible but default on all loansincurred. Both types of governments always



5/11

choose LF

, M, and w to maximize that govern-ments utility subject to the budget constraint ofborrowing or debt repayment and unit popula-tion size. For simplicity, we will focus on repu-tational equilibria with D

O D

N D, though

the results do not rely on this restriction.

PROPOSITION 1: Under the preceding as-sumptions about preferences, players, and tech-nologies, there is a reputational equilibrium inwhich both odious and nonodious governmentsborrow D in odd periods and successor govern-ments repay D/ in even periods.

PROOF:Creditors behavior is such that if a country

did not repay D/ it would never be issued loansagain. Since by assumption D D*, the max-imum loan incentive compatible for countries,continued borrowing is preferable to autarky, sothe country would repay the loan. The loans arezero profit and hence incentive compatible forcreditors. Given the behavior of creditors (gov-ernments) off the equilibrium path, govern-ments (creditors) prefer to borrow if possiblebut default (never lend) in response to nonequi-

librium path behavior by creditors (governments).

The status quo of the sovereign debt marketindeed seems to be that successor governments,concerned about their reputation, typically ac-cept responsibility for debt, independent of thenature of the preceding regime. For example,Anastasio Somoza was reported to have lootedbetween $100 million and $500 million fromNicaragua by the time he was overthrown in

1979. Daniel Ortega, leader of the Sandinistagovernment that succeeded Somoza, told theUnited Nations General Assembly that his gov-ernment would repudiate Somozas debt, but hereconsidered when his Cuban allies advised himthat doing so would unwisely alienate Nicara-gua from Western capitalist countries. Simi-larly, the South African government, in order toremain in the good graces of investors, hasdistanced itself from the popular movement to

nullify its apartheid-era debts.4

The equilibrium in the model that best char-acterizes this status quo is the one described inProposition 1 in which both types of govern-ments borrow D in odd periods.5 If the govern-ment is odious, workers are employed aspalace builders and are paid W, and the remain-

der harvest fruit and are paid w . The govern-ment consumes fruit valued at D (1 )(aw ) (W P

M) plus a palace. If the govern-

ment is nonodious, all workers harvest fruit andare paid a D in odd periods. In all evenperiods, the nonodious government repays ex-isting loans and A d(t)/ is passed along tothe people, all of whom harvest fruit.

Of course, in addition to the equilibria dis-cussed above with no lending or with lending in

odd periods to all governments, the folk theo-rem supports a plethora of other equilibria. Forexample, there are equilibria in which loansmade in time periods ending in the number 1would not be repaid, and hence are not ex-tended. There are also equilibria in which loansare issued only to nonodious governments.

PROPOSITION 2: There exists a reputationalequilibrium in which nonodious governments

borrow D in odd periods and successor govern-ments repay D/ in the successive even period,but odious governments receive no loans.

PROOF:Consider strategies in which if a country did

not repay loans issued to a nonodious regime itwould never be issued loans again. Since u(AD/) u(aD) u(A) u(a), the countrywould repay the loan. The loans are zero profit

and hence incentive compatible for creditors.Suppose if a successor did not repay loans is-sued to odious regimes the country would con-tinue to receive loans in the future. Compared torepayment, nonrepayment would imply a lumpsum gain of D/ in period 2, raising the popu-lations consumption by D(1 )/ in every

4 Somoza Legacy: Plundered Economy (WashingtonPost, November 30, 1979); Cubas Debt Mistakes: A Les-

son for Nicaragua (Washington Post, October 5, 1980); S.

Africa Shuns Apartheid Lawsuits (Guardian, November27, 2002).

5 The status quo equilibrium could also be one in whichodious governments borrow a different amount than nono-dious governments (any D

Osatisfying 0 D

O D*). This

would not alter our results substantively.



6/11

period. The successor would not repay, sincenonrepayment maximizes the populations util-ity. Anticipating this, creditors would not issueloans to odious regimes. The remainder of theproof follows as in Proposition 1.

One reason why we are not in such an equi-librium may be that creditors and governmentshave coordinated on the simpler equilibrium inwhich loans do not depend on government type.This coordination may be a reflection of history:in the past, more governments were undemo-cratic and may have had little interest in inter-national norms that cast doubt on the legitimacyof sovereign governments.

II. Comparison of Trade and Loan Sanctions

In this section we first argue that cutting offdictators from either trade or lending could de-ter at least some dictators from taking power.We then show that firms have an incentive tobreak trade sanctions and that, if trade sanctionsare enforced, they make the population worseoff in the period in which they are enforced.Finally, we argue that loan sanctions have more

attractive properties.

A. Deterrent Effects of Trade or LoanRestrictions

The probability of a dictators arising in pe-riod 1 is endogenous to the spoils from office,and a dictators utility from being in office willbe lower if he cannot trade or borrow. If odiousgovernments can borrow up to an amount D and

there are no trade restrictions, the utility fromoffice is D (1 )(a w ) (P

M

W)). If a dictator is able to borrow D but cannottrade, his utility is reduced to D a w sincehe cannot import marble to build a palace andinstead spends all his resources on fruit. If adictator is able to trade but cannot borrow, hisutility is reduced to (1 )(a w ) (P

M W)) if he is able to pay for a palace by

selling fruit; if he is unable to pay for marble

and build a palace without borrowing, his utilityis the lower amount a w obtained from allo-cating all labor to harvesting and consumingonly fruit. If a dictator can neither borrow nortrade, his utility from office is a w . In each of

these cases, restrictions on trade or borrowingreduce the utility a dictator would obtain frombeing in power and therefore lower the proba-bility that he takes over the country.6

B. Trade Sanctions

In the context of our model, trade sanctionsagainst a country consist of an agreementamong all other countries not to allow theirdomestic marble suppliers to furnish marble tothe sanctioned country.

Trade sanctions are fragile. Consider the casein which marble suppliers are Bertrand compet-itors. If one country with two or more marblesuppliers does not agree to the sanctions, the

sanctions do not affect the payoffs for the sanc-tioned government. Furthermore, there arestrong incentives for marble suppliers to breakthe sanctions.

A second weakness is that when trade sanc-tions bind, the population is made worse off.The government can no longer build palaces,which deprives a portion of the population ofthe efficiency wage premium W w .

C. Loan Sanctions

Under the model, loan sanctions consist oflegal changes made by countries to prevent as-sets held there from being seized to repay debtincurred by a particular regime.

PROPOSITION 3: The imposition of loansanctions against a particular regime elimi-nates equilibria with lending to that regime.

PROOF:Suppose a creditor issued a loan of size d

0 to a sanctioned government. The sanctionimplies that even successor governments thatdefault on the loan to the sanctioned govern-

6 Another rationale for sanctions is that the threat ofthem improves the behavior of a government in power. This

effect is captured by our model if the production of a newgovernment at the beginning of the period is instead con-strued as a change in the behavior of an existing govern-ment. A further argument for sanctions is that they mayhasten the fall of the government. We do not model thiseffect.



7/11

ment can save overseas at R 1/ withoutassets being seized. If in period 2 the successorgovernment does not repay the loan, the coun-trys income stream for t 2 onward is A, a, A,a ... , and its discounted utility is maximized byconsuming (A a)/(1 ) in each period. It

achieves this by saving (A a)/(1 ) ineach even period and withdrawing (A a)/(1 ) in each odd period. Borrowing does nothelp with consumption smoothing, so with orwithout the ability to borrow, it chooses not toborrow. If the successor government insteadrepays the loan, the countrys income stream netof debt repayments for t 2 on would be A d/, a, A, a ... which gives lower discountedutility than the income stream A, a, A, a ... ,

regardless of whether the country is able toborrow. The successor does not repay loansissued to the sanctioned regime, so creditorswill not issue loans to a sanctioned regime.

The loan sanction eliminates the penalty acountry faces for repudiating debt incurred bythe sanctioned regime and, anticipating this,creditors would not issue loans to a sanctionedregime in the first place. This is a straightfor-

ward application of Bulow and Rogoff inwhich, if a country can use savings to smoothconsumption in lieu of borrowing, a reputa-tional equilibrium with debt cannot be sus-tained. With a system of loan sanctions, itremains the case that governments that renegeon loans made to unsanctioned regimes wouldhave their savings seized; countries would haveincentives to repay these loans, so reputationalequilibria with loans to unsanctioned govern-

ments would continue to exist. Governmentsthat inherit loans made to sanctioned regimes,however, would not value being able to borrowin the future since they could save abroad in-stead. The prospect of future loans cannot en-force repayment of loans issued in violation ofa loan sanction. More generally, even if gov-ernments face more complicated incomestreams where borrowing might still be valu-able, loan sanctions may shift creditors behav-

ior toward not viewing default on sanctionedloans as cause for curtailing future lending (asin the equilibrium of Proposition 2).

While trade sanctions are fragile since thirdparties have incentives to break them, loan

sanctions are self-enforcing. Potential creditorshave incentives to abide by a loan sanction,since any credit issued will not be repaid underthe model.

In addition, loan sanctions have better wel-fare implications than trade sanctions for the

population living under an odious regime. If adictator needs to borrow to pay for palace build-ing, loan sanctions, like trade sanctions, reduce thepopulations consumption in the short run becauseworkers lose the wage premium W w . If adictator can pay for a palace without borrowing,the population does not suffer this short-term costof lower wages. But in either case, loan sanctionsmake the population better off in future periodssince it has no debt repayment to make. If the gain

from not inheriting debt outweighs the short-termeconomic cost of a sanction, which seems plausi-ble, then ex post a loan sanction is welfare-improving for populations ruled by an odiousregime, even setting aside its effect on the proba-bility that dictators come to power.7 Proposition 4formalizes this argument.

PROPOSITION 4: If 1/(1 ){u((A a)/(1 )) u((A a d(1 )/)/(1

))} (u(W) u(w ))/, a loan sanctionimposed on an odious regime is welfare-improving for the population of that countryrelative to the populations welfare under anequilibrium in which the odious government isnot sanctioned and borrows d.

PROOF:By not repaying debt, the country gains d/

in period 2; it smooths this windfall by consum-

ing an extra d(1 )/ in t 2, 3, ... Itsdiscounted utility increases from 1/(1 )u((A a d(1 )/)/(1 )) to 1/(1 )u((A a)/(1 )). The short-term cost offoregone efficiency wages is a utility loss of(u(W) u(w )) in period 1.

Proposition 4 implies that loan sanctions maybe particularly attractive instruments in cases inwhich the sanctioning government or interna-

7 If a trade sanction prevents a dictator from depleting anonrenewable natural resource with no domestic market, thetrade sanction could confer a similar benefit on the popula-tion. They would inherit a larger stock of natural resources.



8/11

tional body does not want to make the popula-tion of the target country worse off, but simplyits leadership.

Another important point is that the loan sanc-tions attractive welfare properties for the peo-ple pertain only when the sanction is applied

against regimes that are borrowing against thepeoples interests, i.e., governments defined asodious in the model.

PROPOSITION 5: A loan sanction imposed ona nonodious regime during a low-income pe-riod (odd t) is welfare-decreasing for the pop-ulation relative to their welfare under anequilibrium in which the nonodious governmentis not sanctioned and borrows.

PROOF:This follows from revealed preference since a

nonodious government maximizes the popula-tions utility. Because u is concave, the pop-ulations indirect utility is increasing in theamount borrowed in odd periods under a nono-dious regime.

Proposition 5 highlights an important policy

issue: a potential peril with loan sanctionsaswith other sanctionsis that they could be ap-plied at will by countries or international bod-ies. If in practice loan sanctions are imposed ona government that acts in the populations in-terest, for example because the major powersdisfavor the government for foreign policy rea-sons and want to weaken it, the loan sanction hurtsnot only the government but also the people.

Stepping outside the model, it is worth noting

that a loan sanction is also potentially welfare-improving for the population if the governmentis not malevolent and its intent is to maximizethe populations utility, but it is incompetent atdoing so. A government that neither loots norrepresses might simply use loans to pursue badinvestments, and the population would potentiallybe better off if the government could not borrow.

III. Extensions of the Model

A. Relaxing Assumptions

The model assumes that odious regimes leaveoffice after a single period. If odious regimes

are allowed to stay in power longer, loan sanc-tions retain their attractive features. Supposethat in each period there is some fixed proba-bility that the regime survives until the nextperiod, and that a loan sanction continues toimply that successor regimes suffer no penalty

from refusing to repay debts incurred by thesanctioned regime. In the model as it stands,dictators have no incentive to repay loans, andin fact would have no need to return to the loanmarket after period 1 since they will be satiatedwith palaces after building just one. One canimagine, however, an extension of the model inwhich dictators have incentives to repay loans(for example, out of Ceausescu-like pride orbecause long-lasting dictators want to be able to

borrow in the future, perhaps because palacesdepreciate). In this case, creditors might lend toan odious regime despite the loan sanction.Since a loan will be repaid only if the dictatorsurvives to repay it, the risk that he will losepower increases the interest rate he faces.

The larger the probability of regime change,the higher the interest rate that lenders willrequire since, in expectation, an odious regimemust repay its own loans. Once the probability

of regime change becomes large enough, it willno longer be incentive compatible for a dictatorto repay since the required repayment in thestate of the world in which he survives will betoo high. For a high enough probability of re-gime change, creditors will therefore not lend,and the results will be the same as in the casemodeled where there is zero probability of sur-viving and no lending to sanctioned regimes.

Even in cases where an odious regime con-

tinues to receive loans, sanctions have benefi-cial effects. An odious regime is worse off withthe loan sanction than without since it faces ahigher interest rate. The prospect of loan sanc-tions would continue to deter odious regimes.Most importantly, the population has a smallerdebt burden when the sanction is in place, be-cause it would not have to repay odious debtthat is outstanding when the dictator is toppled.

It is also possible to relax the assumption that

dictators take power only in the first period. Inthis generalized case, odious regimes may in-herit debt from the previous government andwill always choose to default on it, regardless ofwhether loan sanctions are in place. Lending



9/11

that adjusts for this risk can continue to takeplace. Loan sanctions against dictators still limitthe debt burden that dictators can bequeath tothe country.

B. Loan Sanctions More Broadly Construed

Previous sections examined loan sanctions inthe context of a specific reputational model ofdebt. It is worth considering the impact of loansanctions in another class of reputational mod-els of debt, as well as in models in which debtrepayments are enforced with the threat of tradesanctions or denial of foreign assistance.

In Harold L. Cole and Patrick J. Kehoe(1996), failure to repay debt hurts a countrys

generalized reputation, and a country values agood reputation for reasons that go beyond ac-cess to credit. Thus, even with a means ofstoring assets, a country might still have incen-tives to repay loans. In a previous version of thispaper (Kremer and Jayachandran, 2002), wework with a version of Cole and Kehoes modeland show that there will be a multiplicity ofequilibria, including equilibria in which there isno lending to odious regimes. In such a model,

eliminating creditors ability to seize assetswould not necessarily eliminate equilibria withlending to the sanctioned country but could stillpotentially coordinate players on an equilibriumin which this lending does not take place. Forexample, if the United States, the EuropeanUnion, and the United Nations Security Councilall declared that they would regard any futureloans to a particular dictator as odious andwould not expect a successor government to

repay them, it seems plausible that this couldhelp coordinate creditors and debtors on anequilibrium in which loans to the dictator wouldnot be repaid, and hence would not be extended.

States that give foreign aid or have substan-tial influence over multilateral donors might beable to eliminate equilibria with odious debt byannouncing that they regard any future loans toa dictator as odious and would not provide aidto countries that are repaying this illegitimate

debt. In other words, donors could refuse to giveaid to a successor government if the successorwill, in effect, hand over this aid to creditorswho issued odious loans. If the foreign aid isvaluable enough, the country would have incen-

tives to repudiate odious debt, and creditors,foreseeing this, would not originate such loans.

In another class of models of internationalborrowing, debt repayments are enforced notthrough reputation but through sanctions im-posed by creditor countries, such as trade sanc-

tions or denial of foreign aid. In these modelsloan sanctions would take the form of countriesannouncing that they would not impose penal-ties on successor governments that refuse to paycertain debt. The extent to which internationalcooperation would be needed to make loansanctions effective depends on the nature of theunderlying penalty for failure to repay debt. Forexample, if countries failing to repay debt aresubject to retaliation through trade sanctions,

but competition is Bertrand, then even if a fewcountries announce they will not impose tradesanctions, successor governments will not haveincentives to repay, and hence creditors will notlend.

IV. Time Consistency

Preventing enforcement of debts is subject totime consistency problems, so there is a case for

allowing this to be done ex ante, but not ex post.To see this, suppose a body that has the powerto prevent seizure of assets to enforce repay-ment of certain loans receives utility

Bfrom

transferring a dollar to creditors, and P

fromtransferring a dollar to the population, where

B

and P

are nonnegative random variables real-ized at the beginning of each period. We con-sider two possible timings of the decisionmakers actions: (a) before creditors make lend-

ing decisions and (b) at the end of the period inwhich the lending decision was made.

If the body acts ex post, then ifP

B, it

will prevent creditors owed payment for out-standing nonodious loans from seizing assets. Itdoes this as a way to redistribute resources fromcreditors to the debtor country. Anticipatingthis, creditors will not lend. (Or, if the decidingbodys preferences are uncertain instead ofknown at the time creditors make lending deci-

sions, lenders will charge higher interest rates.)If

B

P, an ex post decision maker might

refuse to block seizure of assets to enforceloans, regardless of the legitimacy of the regimethat incurred the debt.



10/11

In contrast, if the decision-making body actsex ante, these preferences do not pose the sameproblem, because they create an incentive toallow lending to occur when it benefits thepopulation, that is, when the government is notodious. If the ruling is ex ante, creditors are, to

a first approximation, indifferent to lending tothe country or to another. Hence, as long as

P

is nonnegligible, the decision maker will notallow odious loans, but will allow legitimateloans.

While making decisions regarding the legiti-macy of loans ex ante rather than ex post willhelp address potential time consistency prob-lems that arise if the decision maker asymmetri-cally values the population and their creditors, it

cannot address the problem of political bias foror against particular governments by the deci-sion maker. If the policy goal is to do no worsethan the status quo of indiscriminate lending,safeguards could be put in place with ex antedecisions by requiring a supermajority of thedeciding bodys members to prevent seizure ofassets. With a supermajoritarian voting rule, ifmembers biases are not completely correlated,the decisive voter would be less biased against

the government than under a simple majorityrule.

V. Conclusion

The debt relief movement rests on two mainarguments: debt further impoverishes poorcountries, and loans were often illegitimate inthe first place. Economic models of debt over-hang formalize the first argument, suggesting

that debt relief may enhance efficiency if acountrys debt is large relative to its income.Partly in response, donors have granted debtrelief to several debtor countries under theHeavily Indebted Poor Countries (HIPC) policyinitiative. However, other countries with argu-ably illegitimate debt, including post-apartheidSouth Africa, are not eligible for HIPC debtrelief. Given the extent of looting and repres-sion by many dictators, it seems plausible that

the efficiency gains from preventing odiousdebt are much larger than the efficiency gainsfrom solving debt overhang. Loan sanctionsagainst such dictators could potentially preventsome of this borrowing.

Loan sanctions are potentially self-enforcing,unlike trade sanctions which are fragile. In ourmodel, equilibria with lending to governmentsjudged to be illegitimate could be eliminated bylegal changes to prevent creditors from seizingassets for nonrepayment of loans issued to such

governments. More broadly, public ex ante an-nouncements about the legitimacy of loansmight deter lending by leading to coordinationin the debt market on an equilibrium with lend-ing only to legitimate governments. If in addi-tion donors tied their foreign aid to pastannouncements and withheld foreign aid fromcountries that were repaying predecessors ille-gitimate debt, equilibria with odious debt couldbe eliminated.

Since more countries engage in foreign tradethan in sovereign borrowing, a loan sanctioncould be applied only in certain cases, but inthese cases (e.g., Croatia under Tudjman orapartheid South Africa), it could have a signif-icant impact.

Creditors would potentially be better off un-der a system in which the rules of the gameare known in advance. Currently, there is amovement to nullify some debt on the grounds

of odiousness, but it is hard for creditors toanticipate which loans will be considered odi-ous in the future. If odiousness were declared inadvance, creditors would avoid lending in thefirst place and would have to find alternativeborrowers, but they would not risk large lossesfrom a successful ex post campaign that nulli-fied some of their outstanding loans. Accord-ingly, interest rates could fall for legitimategovernments.

Could loan sanctions be put into practice?Rather than being adopted overnight as a gen-eral policy, it seems more likely that loans sanc-tions might be imposed against some particularfuture dictator, and then a general policy wouldevolve. For example, if there were a coup inNigeria and the international communitywanted to respond, it seems plausible that somecountries would choose to impose loan sanc-tions alongside trade sanctions.

Many other important issues would need tobe addressed before adopting loan sanctionswhat standards to use for odiousness, whetherhumanitarian loans should be blocked, etc. Ar-guably the single most important issue is who



11/11

should impose loan sanctions. One possibility isthat courts could apply loan sanctions, adjudi-cating in response to lawsuits. Either domesticcourts or an international court could play thisrole. While our model suggests loan sanctionsmight be effective if imposed unilaterally by a

single large country, it might be preferable tohave a system that required international con-sensus to impose them. A possibility is that theU.N. Security Council could decide on loansanctions, a natural extension of its role in im-posing trade sanctions. The United States andother major powers would have veto rights, andwhile their ability to act unilaterally would beconstrained, they would gain veto power overother countries actions.

REFERENCES

Adams, Patricia. Odious debts: Loose lending,corruption, and the Third Worlds environ-mental legacy. London and Toronto: Earth-scan, 1991.

Brennan, Geoffrey and Eusepi, Giuseppe. TheDubious Ethics of Debt Default. Public Fi-nance Review, 2002, 30(6), pp. 54661.

Buchanan, James M. The Ethics of Debt De-fault, in James M. Buchanan, Charles K.Rowley and Robert D. Tollison, eds., Defi-cits. New York: Basil Blackwell, 1987.

Bulow, Jeremy and Rogoff, Kenneth. SovereignDebt: Is to Forgive to Forget? AmericanEconomic Review, 1989, 79(1), pp. 4350.

Cole, Harold L. and Kehoe, Patrick J. Reputationspillover across relationships: Reviving rep-utation models of debt. Research Department

Staff Report No. 209. Minneapolis: FederalReserve Bank of Minneapolis, 1996.

Feilchenfeld, Ernst. Public debts and state suc-cession. New York: Macmillan, 1931.

Hanlon, Joseph. Defining Illegitimate Debt andLinking Its Cancellation to Economic Jus-tice. Unpublished Paper, Norwegian ChurchAid, 2002.

Khalfan, Ashfaq; King, Jeff and Thomas, Bryan.

Advancing the Odious Debt Doctrine.

McGill University, Center for InternationalSustainable Development Law Working Pa-per: No. 2003-0311, 2003.

Kremer, Michael and Jayachandran, Seema.

Odious Debt. National Bureau of Eco-nomic Research, Inc., NBER Working Pa-pers: No. 8953, 2002.

Pogge, Thomas W. Achieving Democracy.Ethics and International Affairs, 2001, 15(1),pp. 323.

Sack, Alexander N. Les effets des transforma-tions des etats sur leurs dettes publiques etautres obligations financieres. Paris: RecueilSirey, 1927.

http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]http://www.ingentaconnect.com/content/external-references?article=0002-8282(1989)79:1L.43[aid=3904846]

odious debt

Documents