Agency law and odious debts

ABSTRACT Because of the way that the international lending system works, poor nations have been forced to repay sovereign debts without having a moral obligation to do so. Suppose a corrupt public official borrows money from an international agency, or from private investors, and later on embezzles this money, or uses it to oppress the population. Suppose, further, that the lender is aware of the potential of this situation and still lends. Typically, the international community considers that successor governments have the obligation to repay the funds and the interests associated to them. In fact, this is what they usually end up doing. Public officials are all aware that if they do not honour sovereign debts, they will face all kinds of negative consequences, including exclusion from future markets, loss of reputation and legal sanctions. Owing to this mechanism, entire generations have been burdened with debts fraudulently incurred in their name by governments in the past. These kinds of debts have been known in the legal literature as ‘odious’. In this article, I discuss the conditions defining the bindingness of a debt. I suggest that they can be made explicit by looking at the rules under which the lending system works at the domestic level, and by then extending these rules to the international domain. I argue that, because of their plausibility, these are the rules that should govern international lending from now on. I also discuss the feasibility of extending these rules globally, and consider potential objections to my proposal.

population. Suppose, further, that the lender is aware of the potential of this situation and still lends. Typically, the international community considers that successor governments have the obligation to repay the funds and the interests associated to them. In fact, this is what they usually end up doing. Public officials are all aware that if they do not honour sovereign debts, they will face all kinds of negative consequences, including exclusion from future markets, loss of reputation, and legal sanctions. Owing to this mechanism, entire generations have been burdened with debts fraudulently incurred in their name by governments in the past. Debts that states do not have the obligation to repay have been labelled 'odious' in legal literature (Sack 1927). 1 This process is endorsed by international positive law (Thompson 2002, 4;Crawford 2002), 2 which holds as a central tenet that, whenever public officials make decisions in the name of the state, the state is held liable for these decisions. So if public officials sign trade or environmental agreements, or decide to become members of an international organization, citizens are bound by these agreements. Also, if an official asks for a loan in the name of the state, the state will be holistically considered as having the obligation to repay it. Through taxes, each citizen will, therefore, be burdened with this debt (Howse 2007;Murphy 2010, 303). 3 It is clear, however, that, under certain conditions, political decisions do not morally bind citizens.
In this article, I discuss the conditions defining the non-bindingness of a debt. The existing literature on the topic is not very clear on exactly what these conditions are. To fill this gap, I develop a framework to analyse the non-bindingness of debts. Specifically, I suggest that the conditions for a debt to be considered as non-binding can be made explicit by looking at the rules under which the lending system works at the domestic level (as specified by agency law) and by subsequently extending these rules to the international domain. Agency law is based on the very compelling idea that nobody can be held responsible for decisions made by agents on their behalf, unless these agents are authorized to do so. Extending this idea to the international domain will require a philosophical analysis of the conditions under which actions of public officials count as non-authorized. Ultimately, I argue that, because of their plausibility, these rules should govern international lending from now on. One of the remarkable benefits of this discussion is that it makes evident that some debts are clearly odious; even if we accept a minimal, non-demanding threshold of non-bindingness. The article is structured as follows. In 'Scholarship on odious debt', I discuss the relevant literature on the topic and situate my contribution in this debate. In 'Private law principles that regulate domestic lending', I discuss the private law principles that regulate domestic lending, especially as described by agency law, and argue that the crucial principles for debts are those of 'authorization' and 'good faith'. In 'Extending private law principles to sovereign lending', I extend these principles to international lending. In 'Principle of 1 The term 'odious debt' was used for the first time by the Russian legal scholar Nahum Sack (Sack 1972). 2 As the Vienna Convention on the Law of Treaties stipulates with reference to international agreements, 'every treaty in force is binding upon the parties to it and must be performed by them in good faith'. http://legal.un.org/ilc/texts/ instruments/english/conventions/1_1_1969.pdf . A discussion of the idea that states are liable can be found in Thompson (2002, 4) and in Crawford (2002).
authorization', I discuss what the authorization principle would look like when applied in the international domain. This section discusses possible criteria to classify the actions of public officials as 'non-authorized'. In 'Good faith principle', I discuss what the good faith principle would look like when applied at the international level.

Scholarship on odious debt
The literature on non-binding debts (i.e. odious debts) is vast. The topic has mainly been approached by legal scholarship, by global justice literature, and by social movements and NGOs (Khalfan, King, and Thomas 2003;Jayachandran, Kremer, and Shafter 2006;Buchheit, Gulati, and Thompson 2006;King 2006;Howse 2007;Adams 1991;Raffer 2007;Toussaint 2016;Acosta and Ugarteche 2008). 4 These contributions, however, do not allow us to reach a definite conclusion on what the conditions for nonbindingness are and, consequently, they do not indicate what a feasible mechanism for solving the issue would look like. We can consider each of them in turn. Legal scholarship has discussed the original definition of odious debt, possible different interpretations of it, as well of the possibility of implementing them in a real world context.
The first definition of odious debt was proposed by Sack (1927), who stipulated that the conditions for the odiousness of a debt are that: (1) The debt is contracted by a despotic power, (2) for a purpose that is not in the general interests and needs of the state, and (3) the lender knows that the proceeds will not benefit the nation as a whole. 5 This account has undergone several revisions and interpretations. Toussaint (2016), for example, has argued that Sack's doctrine should be interpreted as claiming that a 'debt is odious if it has been incurred against the interests of the population and the creditors were aware of this at the time', 6 meaning by this that the nature of the regime that borrows (whether democratic or autocratic) is irrelevant, and highlighting the fact that the purpose of the loan is what matters. Toussaint (2016) proposes to improve on Sack's definition by adding that we should also take into account the liability of the creditors, who 'regularly violate the established treaties and other international instruments for the protection of rights '. 7 Alternatively, Lienau (2014, 8) 8 interpreted Sack's conditions as suggesting that his definition requires some sort of popular consent in order to generate binding obligations; and that binding sovereign obligations must be entered into for the purpose of 'benefitting the underlying people'. 9 4 See for example Jayachandran, Kremer, and Shafter (2006), Buchheit, Gulati, and Thompson (2006), King (2006), Howse (2007), Adams (1991), Raffer (2007), Toussaint (2016), and Acosta and Ugarteche (2008). Several NGOs and political movements have also campaigned for odious debts and debt cancellation. Some of them are CATDM (http://www. cadtm.org/); Debt Watch (http://www.odg.cat/), Jubilee (http://advocacyinternational.co.uk/featured-project/jubilee-2000); and OXFAM (https://www.oxfam.org/en/tags/debt-relief). 5 See Sack (1927, 1). 6 See Toussaint (2016). 7 Toussaint (2016) mentions as examples of creditors who act in such a way, referring to the IMF and the World Bank 'who have continuously and deliberately imposed policies on debtor countries that violate human rights; and the Troika, which imposes brutal austerity policies on Greece'. 8 Lienau (2014). 9 Lienau (2014, 8).
On the other hand, Jayachandran, Kremer, and Shafter (2006) 10 hold that odious debts should be understood as debts incurred by the government of a nation without either popular consent or a legitimate public purpose, shifting the attention from 'lack of benefit' as a criteria for odiousness to 'lack of legitimate public purpose'.
Finally, Howse (2007, 2) has interpreted Sack as claiming that a debt is odious when it is 'contracted and spent against the interests of the population of a State, without its consent, and with full awareness of the creditor'. 11 A quick revision of these definitions show that there are many different candidate conditions for odious debts. However, each of them have specific problems. Toussaint's suggestion that the most relevant condition is the purpose for which the money was used seems to be on the right track, but he is not very specific about what counts as an odious-generating purpose or how exactly we should define it. Lienau, on the other hand, mentions 'consent' as the relevant standard, but what exactly consent should apply to (e.g. to the nature of the regime, the specific loan, etc.) and what counts as valid consent remains unclear. Kremer et al.'s claim that we should focus on 'legitimate public purposes' seems to make sense, but they do not provide a detailed account of what a legitimate public purpose might be. Finally, other definitions, such as Howse's and Lienau's, mention as the relevant standards the 'absence of benefit', or the loan being 'against the interest of the population'. However, as I will argue, these conditions seem neither necessary nor sufficient for the definition of what makes a debt 'odious '. 12 Here I will show that the central condition for non-bindingness is how money is spent, and specifically when it is spent in ways that are incompatible with the role of public officials.
Rather than producing a thorough examination of these conditions, legal scholars have also focused their attention on the legal status of odious debts and on issues of implementation. That is, they have been concerned with whether or not institutions should be modified so that odious debts are eliminated or prevented, and how exactly institutions should be changed. In this vein, Cheng (2007) has been quite skeptical about the relevance of odious debts in international law, as they do not exist under any treaties, nor do they exist in practice. 13 States are bound by their obligations, so they should honour their debts, he says. However, other scholars have proposed concrete reforms. Toussaint (2016), for instance, has argued that 'a sovereign state that discovers that it has an odious debt can and should repudiate it unilaterally. The first steps towards this goal would be to suspend payments and to conduct an audit with the participation of the citizens'. 14 . Also, Acosta and Ugarteche (2008) have proposed the creation of a permanent independent arbitration tribunal, possibly under the auspices of the United Nations, to hear cases of generalized repayment difficulties or disputes. 15 A key feature of both models is that independent arbiters would be empowered to judge 10 See Jayachandran, Kremer, and Shafter (2006). 11 See Howse (2007, p2). 12 Other possible variations or interpretations are that odious debts are 'debts incurred by the government of a nation without either popular consent or a legitimate public purpose' (see Jayachandran, Kremer, and Shafter 2006), or that odious debts imply 'Absence of consent, absence of benefit, creditor awareness' (Khalfan et al. 2003, 2). Khalfan also defends this definition in http://www.choike.org/documentos/odious_debt_presentation.pdf, page 2 onwards. 13 Cheng (2007). 14 See Toussaint (2016). 15 See Acosta and Ugarteche (2008).
instances of illegitimate debt and to declare those debts null and void. On the other hand, Raffer (2007) has put forward several objections against the odious debt doctrine, as defined so far; 16 but he still believes that some sovereign debts can be considered illegal on the grounds that they: violat[e] the law, basic legal principles or that are legally null and void: debts incurred in violation of national laws, of international law, such as in breach of IFI-statutes and general universally accepted legal principles, especially debts, whose servicing violates human rights (p. 7). 17 Also, in order to deal with these illegal debts, he proposes that a neutral court or panel decide whether international financial institutions have violated their own statuses by, for example, providing criminal loans (i.e. loans to government who have committed gross human rights violations) (Raffer 2007, 14). 18 Finally, Jayachandran, Kremer, and Shafter (2006) have made an interesting proposal for implementation, which basically suggests that lending to autocratic regimes should be blocked.
These proposals for implementation are all worth exploring. However, they should be considered ex-post analysis, and not ex-ante, in the sense that they propose what to do with a debt once it has been established that it is odious, instead of proposing how to implement institutional reform aimed at preventing odious debts from now on. The latter approach seems to be the most useful for investors and financial institutions to guide their decisions in the future. The principles that I propose in my article should be useful to structure principles for lending from now on. Thus, in contrast with most of the literature on implementation, I propose an ex-ante approach that remedies this shortcoming in existing literature. The only proposal which seems to be made from the ex-ante perspective is the one that Jayachandran et al. have put forward. However, their account is based on the assumption that only dictators can incur odious debts. As I will later show, this assumption is misleading.
The topic of non-binding (or odious) debts has also been approached by global justice literature. Pogge (2001Pogge ( , 2008, for instance, approaches the topic, but he does so from a wrong perspective. In his view, we should discuss the issue of debts in terms of the effects they cause on the quality of government and world poverty. 19 Thus, for Pogge, the issue can be settled by relying on empirical discussions: whenever loans aggravate poverty, they should be excluded. 20 The borrowing privilege (that is, the fact that the international community grants autocratic governments the right to borrow) is a crucial example of this. Because of the existence of the borrowing privilege, Pogge says, millions of people have become impoverished or their situation has worsened, and that is what makes it unjust. Had this privilege not existed, or if we implement some (minor) modifications to the rules that govern lending, the poor would be better off. Thus, Pogge says that '[. . .] the existing world order is itself a crucial causal factor in the prevalence of corruption and oppression in the poor countries'.
The borrowing privilege we confer upon a group in power includes the power to impose internationally valid legal obligations upon the country at large. Any successor 16 See Raffer (2007, 234-235 Pogge (2001) or Pogge (2008). 20 For a discussion about this point, see (Cohen 2010). government that refuses to honour debts incurred by an ever so corrupt, brutal, undemocratic, unconstitutional, repressive, unpopular predecessor will be severely punished by the banks and governments of other countries. At minimum it will lose its own borrowing privilege by being excluded from the international financial markets. Such refusals are, therefore, quite rare, as governments, even when newly-elected after a dramatic break with the past, are compelled to pay the debts of their ever so awful predecessors (Pogge 2003, 10). 21 Moreover, he mentions three specific effects of the borrowing privilege: that it facilitates cheap borrowing by destructive rulers, that it undermines the capacity of successor governments to implement necessary reforms, and that it strengthens incentives towards coup attempts. In a different text, he says that the borrowing privilege 'provide strong incentives to potential predators (military officers, most frequently) to take power by force' and to oppress their people and divert state revenues into their own pockets' (Pogge 2005, 49); 22 and in another one that the borrowing privilege is 'an unmitigated disaster for the global poor who are being dispossessed through loan and resource agreements over which they have no say and from which they do not benefit.' (Pogge 2007 ). 23 As we can see, Pogge clearly connects the borrowing privilege with its effect, in order to condemn it. 24,25 Here, I propose a different strategy. Instead of discussing the issue of debts in terms of their impact on poverty, I will discuss the conditions under which they are not binding for states, regardless of such distributional consequences. In other words, my account will not ground the claim that a debt is not binding on the fact that it causes poverty, but will rather base the non-bindingness of a debt on other considerations, which I shall discuss later on. Incidentally, my account will also assume that the relative wealth of the parties is irrelevant to establish the non-bindingness of a debt. That is, the fact that the borrowing state is poor or wealthy, or that the lender is small or big, will have no bearing on the issue of odious debts. As I will show later, any kind of state and lender can generate odious debts; in the same way that any kind of citizen, whether poor or rich, can be defrauded by someone who borrows in her name.
Another global justice scholar who has addressed debts from a normative standpoint is Barry and Herman (2007, 60). 26 Barry argues that a central issue regarding justice in debts is that the individual agents who are empowered to agree to the contract and those who benefit from it are often different; and that those who are bound by the contract are not always given adequate consideration. However, in this discussion, it is still quite unclear whether the immorality of debts resides in the fact that the agents who borrow and the agents who are bound are different, or that those who are bound 21 See Pogge (2003), 'Assisting the Global Poor', in http://www.princeton.edu/rpds/seminars/pdfs/pogge_assistingpoor. pdf, page 10. 22 See Pogge (2005, 49). 23 See Pogge (2005). 24 In most of his writings, Pogge discusses the ethics of lending in terms of the consequences it generates. However, he occasionally deviates from this perspective and seems to argue that the borrowing privilege is unjust, on the grounds that the government is a de facto one, and, therefore, not authorized to borrow. Something along those lines is suggested in the warehouse analogy that he makes. See Pogge (2005, 737). In this article I only address his outcomebased approach. 25 Pogge´s view is also represented in Pogge (2005). For the view that Pogge connects debts with consequences, see Steinhoff (2012). 26 See Barry and Herman (2007, 60). by it do not really benefit from the loan, or that the loan was incurred by 'oppressive elites' or 'dictators', or that future citizens do not consent, or that the loan 'harms' the country, or a combination of these. 27 This indeterminacy with respect to the conditions of fairness and unfairness points to an urgent need for clarification. Gosseries (2007) has also briefly addressed the topic of odious debts in a chapter, 28 but, as he says, his essay is not concerned 'specifically with odious debt', (p 101), 29 but rather mentions them as a starting point for another discussion.

Private law principles that regulate domestic lending
In order to understand the injustice of odious debts, we should identify the rules under which lending should operate. Given the considerable stakesbasically, the viability of the whole international lending systemthese rules should be both compelling and plausible. Otherwise, it will be difficult to convince relevant agents to implement them. These rules already exist, as they can be found in private agency law, and they regulate lending between agents domestically. What we should do is extend those rules globally, and give an account of how they would work. 30 We can make these rules explicit through the analysis of a specific example. A C.E.O. of a corporation borrows money from a bank in the name of the company for which he works. The C.E.O. uses the money he obtains from the bank to decorate her or his daughter's office. That is, he uses the money for purposes which are clearly incompatible with her or his role as C.E.O. Moreover, the C.E.O. acts in ways that are visibly suspicious, suggesting in many different forms, before borrowing, that he will spend those funds to decorate her or his daughter's office. After a while, the bank attempts to recover the loan and interest payments, but the corporation in whose name the loan was incurred is the party that the bank considers liable, not the C.E.O. who borrowed the money. The injustice of this situation is obvious. The C.E.O. was acting on her or his own initiative, and was not allowed by the corporation to borrow for the purposes for which he borrowed. It seems clear that, in this case, the debt should not be considered binding on the corporation, and that the C.E.O. who actually incurred the loan is the one who should be liable for it. This would be so, even if the rogue has dropped out of sight or has declared personal bankruptcy so that nothing can be recovered from them.
Cases like these have a clear resolution in agency law. Agency law can be defined as a consensual relation created by contract or by law where one party, the principal, grants authority to another party, the agent, to act on behalf of and under the control of the 27 The benefit standard is suggested by Barry. See Barry and Herman (2007, 60). 28 See Gosseries (2007). 29 See Gosseries (2007, 101). In his brief discussion on odious debts, Gosseries states that since 'only a democratic government can be properly regarded as having a mandate from the people, only a democratic government can be said to validly bind the people it represents'. He, thus, mentions in passing an important condition under which a debt is odious (the autocratic government condition). Later in this article, I show that it is a mistake to defend such a condition. 30 In this article I will not discuss how private law deals with the problem of lending to poor people. This discussion is very interesting and can be related to debt cancellation debates. However, in my view, poverty does not necessarily generate a strong reason for non-repayment, as the fact that someone is poor is not a sufficient condition to dissolve an obligation to repay a loan, and does not automatically make the debt immoral either. The strategy I pursue, in contrast, seems a straightforward and compelling way of making the case for non-repayment.
principal to deal with a third party. An agency relationship is fiduciary in nature, and the actions and words of an agent exchanged with a third party bind the principal. 31 In the case of the C.E.O., the principal would be the corporation, and the agent would be the C.E.O. The C.E.O. is authorized by the corporation to act on its behalf, and the decisions of the C.E.O. bind the corporation. Now, according to agency law, the agent will exceed her or his authority if s/he makes decisions that are incompatible with her or his role as agent. This is because the authority of the agent is not unrestricted. According to agency law, the scope of an agent's authority, whether apparent or actual, is considered in determining an agent's liability for her or his actions. An agent is not personally liable to a third party for a contract the agent has entered into as a representative of the principal so long as the agent acted within the scope of her or his authority and signed the contract as agent for the principal. If the agent exceeded her or his authority by entering into the contract, however, the agent is financially responsible to the principal for violating her or his fiduciary duty. 32 In the case of the corrupt C.E.O., the C.E.O. was authorized by the corporation to act on its behalf (so the C.E.O. was an authorized agent), but s/he exceeded her or his authority (or failed to act within her or his authority) by using borrowed money to decorate her or his daughter's office. Moreover, in agency law, if a lender ignores visible indications that an authorized agent will exceed their mandate, and still lends, they will not be entitled to claim repayment of the debt from the principal (provided, of course, that the borrower ends up in fact using these funds for purposes that exceeded their mandate). 33 The lender, after all, had the opportunity to check whether or not the agent was exceeding her or his authority, and decided to ignore indications that the agent was going to act corruptly. Also, even if the lender did check diligently whether the agent was exceeding her or his authority, and still lent, he would still not be entitled to claim repayment from the third party either. Since the borrower was the one who defrauded the lender, the person in whose name the loan was taken out is not a party to the transaction. The loan, in other words, was incurred at the lender's risk.
So far, I have described a case where the agent is authorized to act as an agent but exceeds her or his mandate. The picture is even clearer when the C.E.O. is not even an authorized agent in the first place (that is, the principal did not authorize him to act on her or his behalf). Suppose that the C.E.O. claims to be the corporation's agent, but that the corporation does not even know the C.E.O., or never authorized her or him to be the agent. The fake C.E.O., then, decides to borrow money in the name of the corporation, and subsequently decides to use the money to buy a birthday present for their daughter. The corporation, clearly, cannot and should not be held liable for the loan. In this case, someone acted fraudulently in its name. 31 A full definition of the concept of agency law can be found here: http://legal-dictionary.thefreedictionary.com/agency. See also the Restatement (Third) of Agency in the US. 32 http://legal-dictionary.thefreedictionary.com/agency. 33  From these two cases we can see that private law defines two conditions under which an agent does not bind the party s/he represents (the 'principal'). 34 Under these conditions, one might say that the principal is not 'on the hook'. We can call the first one the authorization condition and the second one the good faith condition. Under the authorization condition, the principal is not legally bound by actions of the agent when the agent is either not authorized to act in their name; or when the agent is authorized to act in their name but exceeds their authority (that is, he borrows for purposes for which he was not authorized). In agency law, however, if the agent is authorized and exceeds their authority, the lender can still hold the principal liable for the debt. This happens when the agent acted with apparent authority (that is, when it was not clear, and could not have been clear for the lender, that the agent was overstepping their authority). Thus, in order for the principal to be 'off the hook', a second condition should be introduced: the loan could not have been made in good faith. That is, lenders cannot plausibly argue that the agent had apparent authority if they knew, or could have known, that the agent was not authorized.
A further consideration needs to be introduced. In agency law, even if the agent has no actual authority, the principal may still be liable if it ratifies the agent's acts in her or his name. 35 By 'ratifying', I mean here a voluntary act by which the principal may explicitly, or by remaining silent, manifest that s/he is willing to accept the decisions made by the 'agent' in his name. If the agent does something that the principal is aware of, and the principal does not say anything, or explicitly agrees, the principal will be bound. As I will show later on, the notion of ratification will be important in my account of odious debts.
The conditions of agency law described so far coincide with the basic intuitions shared by most of us. The idea that we cannot be made responsible for a decision made by others in our name (unless we have authorized it) is so intuitively plausible and reasonable that people hardly reject it. However, and despite its plausibility, these conditions are absolutely absent at the international level. As stated earlier, international law considers states (and not individual members within it) to be liable for their debts. 36 Thus, international law ends up burdening states, even when their public officials exceed their mandate. As a consequence, states have been burdened for generations with debts fraudulently incurred in their name by all kinds of corrupt and brutal dictators and rulers. What we should do is apply to international lending the same set of principles that govern private law. The authorization principle (along with an account of what counts as exceeding the authority of public officials) and the good faith principles are the ones that most clearly and persuasively identify the reasons why sovereign debts are non-binding. This distinction highlights the fact that what crucially underlies the problem of odious debt is that someone who is entitled to make decisions on behalf of a third party exceeds her or his authority. I suggest that these rules should be used as minimal standards. All lending that fails to meet these standards will not be binding for states; rather, it would be classified as personal debts incurred by rulers. 34 These two conditions can be found here: http://legal-dictionary.thefreedictionary.com/agency. 35 See RESTATEMENT (THIRD) OF AGENCY 4. 01-4.03 (2006). 36 For this point, see the definition of Pacta sunt servanda above and Crawford (2002).
The idea of drawing a parallel between agency law and odious debts has been briefly discussed by a few legal scholars. However, these discussions have a conceptual deficit: they assume that the criteria to consider that a public official had exceeded their authority (or acted in a non-authorized way) was already established, and proceed with the analysis without putting this assumption into question. 37 Extending private law principles to sovereign lending To extend the principles of agency law to the international domain, it is necessary to specify who the relevant actors are. 38 In my account, public officials are the agents of the state, as they are authorized to act on behalf of the state. The 'state', then, would be the principal, as public officials act on its behalf. And the third party would be 'the lenders', who in the case of sovereign debts would basically be other states, private investors (such as bondholders), and international financial institutions, such as the IMF and the World Bank.
Once we have stipulated this parallel, and consistently with the categories defined by private law; we should extend the two main principles that govern agency law in private law: the authorization principle, and the good faith principle. In accordance with the first one, a public official exceeds her or his authority when s/he uses borrowed funds for purposes which are clearly incompatible with her or his role as a public official.
In accordance with the second principle, if lenders knew, or should have known, that the first condition was not satisfied (that is, if there was something visibly spurious about the loan), and still decided to lend, they will not be entitled to recover funds from the state to which they are lending, if the money is in fact misused. The loan, in other words, will be incurred at the lenders' risk. When both of these principles are violated, a sovereign debt should be considered odious.
It can be argued here that violating the first principle only would suffice to generate a non-binding debt. If a public official embezzles all the money s/he borrows, arguably the state should not be burdened with the debt. This would be so, even if the lender lent in good faith. Thus, the second condition (the good faith condition) does not seem to be necessary. This might be true, and I am sympathetic to that point. However, in order to make the case for odious debts as convincing and clear as possible, I will set aside the issue of whether the second condition is also necessary, and argue that, when both conditions are violated, the debt will definitely be odious. Given that each of these principles are so important, and that they can potentially be challenged by corrupt 37 See, Buchheit, et al. (2006) for example, supposing to be the case that when a public official, acting as an agent, fails to benefit the population, it is overstepping its authority. This claim, as I will show later, is misleading, as failing to benefit the population does not necessarily mean that the agent is acting beyond the scope of authority, in the same way that it would not mean for a C. officials, creditors and other relevant agents, we need to provide compelling versions of each of them. That is what I will do next.

Principle of authorization
This principle requires that we lay out the conditions under which spending by public officials cannot possibly count as being authorized by the population. I will argue here that a public official's action cannot possibly be authorized to be carried out in the name of the state if (a) public officials do not even have authority to rule in the first place, and, additionally, act in ways that are clearly incompatible with their role as public officialsin private law, this would be the parallel to a C.E.O. who takes over a company through unclear procedures, and who subsequently acts in ways that are clearly incompatible with their role as C.E.O.; or if (b) the government does have authority to rulesay, because it was democratically elected in free and fair elections but exceeds that authority.
Cases where public officials act as agents of a country to which they do not even belongparallel to the case of a rogue customer who, out of the blue, asks for a loan in the name of someone who does not even know themwill not be discussed here, as they are exceptional or non-existing in real life. Cases of benevolent autocratic officials will not be discussed either. That is, cases of public officials who are ruling de facto, but who use borrowed funds for purposes that are clearly compatible with their role as public officials (for example, they build good functioning schools and hospitals) will also be excluded from the discussion. Such cases are not clearly instances of odious debts, and international law has, in fact, considered de facto rulers as capable of binding the state. 39 That autocratic rulers can in principle bind the state does not seem unreasonable. If we had no such rule, then every court or tribunal could question the constitutionality of some ruler's ascent to power and refuse to enforce their treaties and transactions, no non-democratic country could contract a binding debt, and much of the world would be locked out of development-related lending and presumably investment and treaties as well. The implications would be enormous. Some might argue here that the concept of 'benevolent autocracy' is contradictory in itself, because autocracies are by definition corrupt and, therefore, loans attached to them will always be odious. However, we should keep in mind that even autocratic governments can eventually make good decisions about how to use public funds. This might be rare, but it is not impossible. The crucial point is that an autocracy should be defined as such on the grounds that it came to power through force, or that it is constituted by an authoritarian institutional structure. Once this autocratic government is in power, it can eventually make good decisions. So we should not conclude that all lending to autocratic governments necessarily generates odious debt.
Cases like (a) correspond to cases where the public official of a state rules de facto. Additionally, such public official acts in ways that are visibly suspicious. The fact that the official rules de facto and that they act in ways that are visibly suspicious are usually related. It is a well-known fact that autocratic governments are prone to illicit behaviour, and so lenders are usually aware of the fact that, if they lend to such governments, funds will very likely be misused. Loans to many heads of state of African countries, such as Zimbabwe, Eritrea, and Equatorial Guinea, satisfy this condition, as it is publicly available information that these heads of state are corrupt and, 40 additionally, that these states rank poorly in terms of the personal freedoms that their citizens enjoy. 41 It is possible that, occasionally, autocratic rulers act in accordance with perfectly acceptable purposes, that lenders know about these purposes, and that the population ruled by the state supports these rulers' actions. Rulers, for instance, can build dams when needed, or improve the traffic system. In these cases we can arguably claim that a loan can be binding. 42 In order for a debt to be odious, the money needs to actually be spent towards non-acceptable purposes.
Cases like (b) apply to democratic countries; that is, to countries whose governments are authorized to rule, having been democratically elected by the population, and, additionally, who will clearly and visibly exceed their authority.
As we can see, under both (a) and (b), a debt becomes odious when the public official oversteps their authority. Now, in order to expand the applicability of agency law to sovereign lending, we need to determine exactly what 'exceeding' or 'overstepping' authority means in the international context. This account will lead us to a philosophical discussion: a discussion of the conditions under which the actions of public officials would be non-authorized.
What are the things that public officials are not authorized to do? In other words, what are the actions that count as being clearly incompatible with their role as public officials? We can consider three possible candidate answers, two of which will not be valid: (a) the benefit standard, (b) the disagreement standard, and (c) the role of public officials standard.
The 'benefit' standard A possible candidate standard is that public officials are not authorized to make decisions that do not generate a benefit for the population. According to this standard, whenever public officials spend money on X, and X does not benefit the population in any clear way, the actions of officials are not binding for the population. Suppose that an official decides to spend all of the available public funds to build roads that go nowhere. Clearly, this would not benefit the population. So, in this view, borrowing for the purposes of building these roads would not bind the state. A version of the view that lack of benefit is what makes a debt odious has been defended by Buchheit, et al. (2006), who claim for instance that 'a debt becomes odious in the eyes of the citizens of a country, however, in part because the proceeds of a borrowing do not benefit those 40 Transparency International classifies these countries as three of the most corrupt countries in the world. See more details in the Corruption Perceptions Index; http://www.transparency.org/cpi2015. 41 See, for example, the Freedom House Report as evidence of this, here: https://freedomhouse.org/report-types/ freedom-world. Although these reports are controversial, the point I am simply trying to make is that, given their existence, and the fact that they are publicly available, lenders cannot plausibly argue that they 'did not know' that heads of state were arguably corrupt. Interestingly enough, the legitimacy of the debt of Zimbabwe has been challenged precisely on the grounds that it is odious. http://cadtm.org/Zimbabwe-the-case-for-a-debt-audit. 42 Suppose, for instance, that an autocratic government borrows funds for the purposes of building a dam, that it is clear about the purposes of the loan, and actually builds the damn. Since people really benefit from it, and lenders lent in good faith, the lenders' claim of repayment seems compelling.
people; the benefits flow to the governing regime that incurred the debt;' 43 and, earlier in the text, that in order for a debt to be odious 'the debt must be incurred by a despot . . . and it must not benefit the state as a whole . . . '. 44 However, benefit is neither a sufficient nor a necessary condition for an action to count as non-authorized. There are many possible political actions that count as non-beneficial but that, nonetheless, are still binding for the population; and, conversely, there are many possible political actions that are beneficial but not binding. Consider the case of a failed but perfectly legitimate diplomatic mission to a foreign nation, or the case of a just defensive war. These actions do not bring about any benefit to the population. However, there is nothing wrong with public officials deciding to carry out these actions. In principle, they are authorized to go on diplomatic missions, even if these eventually fail, and they are authorized to go to war under certain conditions, even if they do not bring about any tangible benefit. Consider, on the other hand, the case of a public official who receives illicit funds from a foreign corporation, and uses a portion of these funds to build roads that people do need. This official's actions are beneficial for the population, but it is highly dubious that her or his decision to build this road can be considered to be an authorized one. What we can see from these examples is that 'benefit' or 'lack of benefit' is not an adequate standard to determine whether a public official acted in accordance with authorized purposes. 45 Needless to say, a government that always fails to benefit the population can hardly be acting in a legitimate way. In fact, the idea that people are better off with a government than without one has been at the heart of social contract theories since Hobbes (1996), and it is the main reason why people voluntarily decide to delegate authority in the first place. However, the fact that people are in general better off under a government does not mean that public officials are required to generate a benefit for the population with every single decision they make, or that, every time they fail to benefit the population, they no longer bind it. Thus, the statement that a specific loan is odious when the government spends the money from that loan for purposes that are not beneficial for the people is misleading. The practical implications of this conclusion are quite radical. Many political campaigns, legal articles, and popular movements demand debt cancellation on the grounds that debts incurred by governments in the past did not benefit the people (for example, in Greece or Spain, during the 2009 financial crisis). However, since 'benefit' cannot be used as a valid benchmark, these demands are groundless for our current purposes.
The 'disagreement' standard A possible alternative standard we can use to determine whether an official has acted in accordance with non-authorized purposes is that the population, or a portion of it, 43 See Buchheit, et al. (2006Buchheit, et al. ( , 1244. 44 See Buchheit, et al. (2006, 1218. Gulati is just an example of this view. Most of the legal literature on odious debts has been based on the assumption that 'benefit' is a necessary component of the definition of odious debts. 45 An alternative version of the benefit standard would be that, when public officials spend money on X, and X does not generate an expected benefit, the actions of public officials are not binding for the population. However, we encounter a similar problem with this version of the standard. Often, officials do not know what exactly will happen as a consequence of their action (that is, whether or not it will generate a net benefit), but they are still authorized to carry it out; and often officials expect a benefit out of their actions, but the action is not an authorized one. A defensive war can be an example of the former, and building a statue of the political leader of the country in the private garden of one of the officials can be an example of the latter (in this case, the benefit would be that jobs would be created, and the community would receive an additional income).
disagrees with the legitimacy of the purposes for which the loan is used. Suppose a public official decides that the most convenient policy for a society in times of crisis is to bail out private banks through transfers of public funds. Undoubtedly, such a decision will be met with much resistance. According to the 'disagreement' standard, the decision's unpopularity makes it not binding. The problem with using this standard is that it would basically imply that all decisions made by a public official would be nonbinding, as it is clearly the case that all public decisions are always contested by at least a portion of the population in democratic and pluralist societies. This standard, in other words, would not be demanding enough, as all political decisions would count as nonauthorized under it, and this is implausible. The ineffectiveness of this standard also has implications on the debate on debts as it has been structured so far. Many people claim that officials have overstepped their authority by making 'unpopular' decisions such as borrowing money from banks at high interest rates, or by benefitting a specific social class while disadvantaging another. The fact that these decisions are unpopular, however, does not seem to be a sufficient condition for them to count as non-authorized. In principle, a decision can be unpopular but authorized. Reducing the salaries of public employees will certainly be unpopular, but a government is authorized to make this rather harsh decision in order to avoid a deeper financial crisis. So another condition seems to be required for such decisions to count as non-authorized.

The role of public officials' standard
Here I will argue that, in order for a decision made by a public official to count as nonauthorized, that decision should be clearly incompatible with the role that public officials are supposed to have. By 'incompatible', I do not simply mean here that the decision lacks effective support from the population, or that it does not generate any clear immediate benefit, but rather that we cannot reasonably argue that the population would delegate authority for those specific purposes in the first place. There is a crucial difference between bailing out companies, or waging wars for the purpose of defending the population; and using public funds for private purposes or to oppress the population. The former can always be supported with reasonable arguments, such as the need to preserve economic stability, security, or other such priorities. For the latter, however, we cannot possibly and plausibly find a reasonable argument that can support delegating authority for those purposes. I will further discuss the clearest cases (i.e. human rights violations and corruption) in the next section. Thus, the role of consent is important in this standard. A debt is not odious when people decide to express their lack of consent by removing their government, as De Mott (2007) seems to suggest. 46 This, indeed, might be impractical in many cases. The crucial point is that the debt is odious when people could not have consented to the specific use of the loan, under any reasonable interpretation of the role of public authority. One might argue here that the view that public officials are agents who can eventually overstep their authority applies only to some societies (particularly liberal ones). However, as Buchheit, Gulati, and Thompson (2006) See Buchheit, Gulati, and Thompson (2006).
Kenya is illustrative. 48 In this case, a businessman bribed President Moi to obtain a contract with the Republic of Kenya. The businessman, in his defense, argued that this bribe had not been paid to the agent of the state, but to the President, who was the state. In his words, he had paid the bribe to the 'remaining "Big Men" of Africa, who, under the one-party State Constitution was entitled to say, like Louis XIV, that he was the state.' 49 The defender's strategy, thus, appealed implicitly to the idea that the payment was not really a bribe, because the president was the state and, consequently, he was entitled to decide what to do with the money. The arbitral panel did not accept the plaintiff's argument and argued that President Moi was 'no more than an agent for the state, no matter what his self-conception might have been' (Buchheit, Gulati, and Thompson 2006, 185). 50 The underlying idea in this statement was also that, as in agency law, a ruler is not the state and cannot use public funds as s/he pleases. The only exception that we can find to the idea that authority has limits has been defended by De Vitoria (1991), who has claimed both that there was an absolute obligation to obey superiors because they were superiors, but also that citizens were fully responsible for the deeds of their rulers. So, if a ruler foolishly went to war, in his words, 'the whole commonwealth may be punished for the sins of its monarch' (De Vitoria 1991, Section 12, question 1, article 9). 51 However De Vitoria, of course, is a pre-modern thinker; thus, a person whom nobody is likely to defend these days.
As stated earlier, the case for odious debts is even more compelling when a second condition is introduced: the good faith condition. Next, I will discuss this principle in detail.

Good faith principle
As in private law, if loans are made in good faithi.e. in a situation in which lenders are completely unaware of the purposes of the loan, and could not have possibly known about themto public officials who are authorized to borrow, and who showed no indication that funds were going to be misdirected; and the funds are subsequently stolen or used for corrupt purposes, creditors will still plausibly have a claim of restitution against the state. In fact, since it was impossible for them to foresee the use of those funds, it would be unfair to make creditors responsible for illegitimate uses of funds. This point is sound; after all, why would lenders be responsible for something they could not have avoided, or even predicted? It has to be possible for lenders to exercise due diligence, and to determine ex-ante whether a government has a fraudulent purpose in mind. 52 The claim that some loans are odious when funds are used for non-authorized purposes is, thus, much more convincing when a second condition is introduced: debts are odious when loans are also not made in good faith. That they are not made in 'good faith' basically means that the lender knew, or should have known, that funds 48 See World Duty Free Co. v. Republic of Kenya (ICSID Case No. Arb/00/7, 4 October 2006). This case is explained in more detail in Buchheit, Gulati, andThompson (2006, 1239). 49 Buchheit, Gulati, andThompson (2006, 1239). 50 Buchheit, Gulati, andThompson (2006, 1239). 51 See De Vitoria (1991, Section 12, question 1, article 9). 52 The issue of creditors' awareness has been included in the discussion by many legal scholars, including Sack. were being or could have been plausibly misdirected away from public purposes. This seems reasonable: it is hard to argue that a successor government should be liable for the dishonest or criminal act of two others. So, when a lender lends to a notoriously corrupt government, say Mozambique, even if there is nothing particularly suspicious about that specific transaction, the loan will count as not having been made in good faith. It is clearly the case that lending to these governments should put the lender on notice of possible subsequent misuses of funds. Lending to these governments is like buying a stolen watch from a very suspicious person in the streetthe purchase could hardly count as a good faith one.
There are many possible cases of public officials lacking actual and apparent authority. Here I will mention just two. These are significant enough to make the case of odious debts compelling.
These two possible cases are human rights violations and straightforward corruption. There might be other cases as well or even grey areas (i.e. cases that are hard to settle). 53 Because of their plausibility, they will have the advantage of being widely accepted by many people and, consequently, proposals for reform of international law and institutions would be easier to implement. I will now discuss each of them in some detail, focusing on how, in these situations, the simultaneous consideration of the principles of authorization and good faith can allow for specific debts to be qualified as non-binding, and therefore, odious.

Human rights
Public officials act in non-authorized ways when they systematically violate a broad group of rights of a large proportion of citizens. What exactly the population's rights are is, of course, up for debate. However, there is strong consensus about what the core human rights are among different theories of justice, in human rights conventions of the existing system of international law, and for the United Nations. These include the right to life (that is, the right not to be unjustly killed), the right to physical security (which includes the right to bodily integrity, not to be tortured, the right not to be subject to arbitrary arrest, detention or imprisonment), the right against enslavement and involuntary servitude, and the right of association. 54 Borrowed funds that directly contribute to massively violating these rights cannot be binding for the state. If, say, a government uses borrowed funds to buy chemical weapons to tyrannize some of its citizens, the authorization principle will have been violated.
A potential objection we face here is that the population might approve of these violations, on the grounds that, although unpleasant, they are necessary to improve the economic situation of the country. 55 A possible response here can appeal to the notion of accountability. If the population knows about the purposes for which public officials 53 If a public official decides to increase the defense budget and reduce the education budget, certainly a big portion of the population will complain. However, this decision will not count as one that cannot possibly be authorized by the population (i.e. it will not be a violation of (i) above). Something similar can be said about disputes regarding whether it is acceptable to use public funds to support private companies during times of recession (e.g. bailouts). Certainly a big portion of the population would oppose these policies. However, they do not count as policies that cannot possibly be authorized. Cases that would count as not possibly being authorized by the population are cases in which public officials clearly exceed their mandate by acting in ways that, if cognizant of the facts, the population would not accept under any circumstance. 54 For a detailed discussion of the basic human rights that international law should recognize, see Buchanan (2007). 55 This argument has been typically used, for example, to defend the Chilean dictatorship in the 1970s and 1980s. are borrowing, is able to impose sanctions or to openly show their disagreement, and decides not to take action, the objection might be a plausible one. In private law terminology, we might say that the principal is ratifying the action of its agent. However, if the population knows about these purposes, and shows clear signs of disapproval, or cannot possibly even show disapproval of them (say, because they are too oppressed by their own government, or because the government is acting secretly), the argument that human rights violations have support among the populationespecially among those who are victimswill no longer work. It is virtually impossible to demonstrate that people who are tortured by their own government might have authorized the government to act for such purposes. A different issue consists in establishing whether or not the human rights' violations actually occurred, and who was responsible for them. 56 Corruption A second related way in which lenders fail to satisfy the 'good faith' condition is when they lend to corrupt governments. By 'corrupt', I mean here that the government engages in a known pattern of 'abuse of public office for private gain'. 57 Public officials can receive salaries and benefits in return for their service, but they cannot receive any other benefit on top of those benefits, such as bribes and gifts in return for favours. Neither can they transfer public funds to their personal accounts, or use their power or influence to favour friends and allies. As in the case of human rights, there are grey areas here. Public officials can eventually benefit themselves but obtain support from a large portion of the population, they can assign their family members key positions of power in efficiently run companies, or they can spy on political opponents for security purposes. However, the fact that there are secondary benefits or potential justifications for these actions does not imply that they are not corrupt in the first place. We can, thus, state that, when a public official uses public office to benefit themselves, or to benefit some particular person (in addition to the benefit or treatment that any of them are already supposed to receive, for publicly known and agreed upon reasons), that public official is overstepping their authority. Cases of corruption include bribing, nepotism, and embezzling public funds, money laundering, illicit funds and others.
As in the case of human rights violations, someone might argue that if people ratify corrupt uses of funds (that is, they do not say anything against them, or they explicitly support them) the corrupt action will count as an authorized one. However, it is virtually impossible to show that people can ratify corrupt uses of funds, as this would imply that people have authorized public officials to become richer at their expense, which is implausible.
Notions of human rights and corruption can be derived from theories of social contract, such as those proposed by Hobbes,Locke,and Kant. 58 From the point of view of these theories, what makes us liable for decisions made by our governments is that 56 We should note here that the intentions that public officials allegedly have when making decisions, or the arguments they use to justify their behaviour, are irrelevant to determine the actual impact on citizens' rights. Public officials are usually deft at vindicating their policies, or at finding excuses when they act unjustly. Even the most egregious violation of human rights has historically been defended by corrupt governments with some sort of imaginative argument. 57 For a complete discussion of this definition of corruption see Kolstad (2012). 58 See, for example, Hobbes (1996) or Immanuel Kant: Practical Philosophy (1996 they are authorized by us to interpret and defend our rights, as long as they do so impartially (that is, as long as they do not obviously favour a group or one of the rightholders). Persons have basic rights, and governments are in a better position to interpret them and enforce them than wethe peopleare. This is because our own judgement is partial, biased, or limited, and the understanding that we have of our rights normally conflicts with or is different from the understanding that others have. We, therefore, need an arbitrator's judgement that can provide a unitary interpretation of our rights. By endorsing a central authority, we obtain a unitary interpretation of our basic rights.
In these accounts, however, we are not liable for the private deeds of public officials. State authority has limits, and those limits are set by foundations of public authority. These accounts might not be compelling for everybody. However, they are a familiar line of thought in political philosophy, and they offer a possible way of justifying limits to state authority. 59 The human rights and corruption standards apply to both autocratic and democratic governments. Lending without restrictions to autocratic regimes with a bad record of corruption and of human rights' violations cannot be considered to be a 'good faith' loan. Lenders can, or should be 'on notice' that these governments might be exceeding their authority. Lending funds to those kinds of governments is parallel to the domestic example of lending funds to a well-known crook who borrows in the name of a neighbour for purposes that are totally unclear. It is important to note here that lending money to these governments would not automatically make the debt odious, as autocratic governments, although not authorized to rule, can in principle use funds for perfectly acceptable public purposes. However, lenders should be aware that, since these kinds of governments are prone to corrupt uses of funds, they will likely use the funds incurred in that specific transaction for illegitimate purposes.
Something similar could be said about corrupt democratic governments. A public official of a democratic country might ask for funds for purposes such as building dams or bridges for a suspiciously high price, or for building a private airport for her or his family. If lenders are aware of this situation, or if there are clear indications that this could happen, and they still lend, their entitlement to recover funds from successor governments ought to become much weaker. This is because lending money to an organization or state when it is not clear whether or not its representatives are acting within their mandate is something that lenders do at their own risk. The risks involved when lending are two-fold. First, there is a risk of default; that is, that the borrower will not repay the debt. Secondand this is the kind of risk that is most relevant for the topic of odious debtsthere is the risk of corruption; that is, the risk that the borrower is overstepping its authority as an authorized agent. The risk involved here is that the lender might not be able (and will not be entitled) to recover the funds from the party incurring the loan. In order to avoid this, lenders should verify the purposes for which the client is borrowing. If they fail to do this, and it turns out that the client steals the money, or uses it for purposes that exceed her or his authority, a third party cannot plausibly be held liable for the debt, for s/he was not even a party in the transaction.
In sum, loans fail to pass the 'good faith' test when the purposes for which the loans are incurred are, or should be, suspicious to lenders. This condition applies to both democratic and autocratic regimes. Lenders cannot claim innocence when officials borrow for purposes that suggest that funds might be misused. Although these officials might generally be authorized to borrow in the name of the state, they cannot overstep their authority by using the funds for corrupt purposes. Therefore, such loans should not be binding on the state. 60 So now we know why loans incurred by Dictator Marcos in Philippines, among others, can be considered odious. The two basic principles I have proposed have both been clearly violated. Officials have violated the principle of authority by not being authorized to rule, and by using funds from loans for corrupt purposes (i.e. basically, to enrich themselves). On the other hand, loans cannot have been made in good faith, as it was publicly known that Marcos was an autocratic ruler and that the circumstances surrounding the loans were spurious. A consistent application of these two principles would yield, I suspect, the impressive result that massive amounts of debts in the world are, in fact, not binding.

Conclusion
There is a clear injustice in the current international financial system. The injustice is that most debtor countries are being forced to repay national debts that they do not have the moral obligation to repay. Following an old legal concept, we can call these debts 'odious'. I have suggested in this article that the nature of this injustice can be clarified by applying on a global scale some of the standards that are normally applied for domestic lending. An important benefit of doing this is that it clearly shows how a large part of the world´s population has been saddled with debts that were fraudulently incurred in their name. Here I have not suggested a possible solution to the problem of odious debts. The main institutional aspects of this solution, I believe, would be to organize international lending around the principles that we consider valid domestically. This would probably entail creating institutions that would put lenders on notice of possible misuses of funds, so that they cannot claim to have made the loan innocently; and that could declare debts as non-binding after loans are in fact misused. The injustice of odious debts is massive and should not be ignored. Surprisingly, however, philosophers have not said much about this issue. In this article, I have attempted to fill this gap. 60 A possible objection to the claim that debts are odious when lenders intended them for private purposes is that proving 'knowledge' of such intentions is difficult. This objection fails, however, in all cases. Some governments or policies are notably and obviously corrupt, and to claim ignorance of such purposes is simply implausible. What follows, in any case, is that international bodies have an interest in creating institutions that make such abuses transparent to lenders so that they have no excuse; if they continued to lend, they would become, in effect, coconspirators in the embezzlement. Such institutions should not block all loans to autocratic regimes for the mere reason that they are autocratic. Even if autocratic governments occasionally spend funds for legitimate public purposes, the standards of liability should certainly be set very high. Such institutions are currently non-existent, and should be created.