The Tuna Bond Scandal: The Continued Lack of Transparency in Bank-to-State Credit Facilities Agreements

The United Nations Conference on Trade and Development Principles on Promoting Responsible Sovereign Lending and Borrowing provide inter alia that creditors should ensure debt contracts are duly authorized; yet, the Mozambique Tuna Bond Scandal suggests that in the market for private bank-to-state credit facilities agreements, there is still work to do. This article presents empirical evidence suggesting growth in the amount of private bank-to-state loans and argues that this growth, combined with a lack of transparency, will lead to increasing numbers of legal disputes over want and abuse of official authority. The English law doctrines of ultra vires contracting in private international law and abuse of agent authority, usually applied in the context of corruption, are considered, and various reforms are proposed, including a transparency register and a legislative reversal of presumption of agent authority.


I. IN TRODUCTION
In 2013 and 2014, 1 three special purpose vehicles (SPVs) controlled by the Republic of Mozambique entered into contracts with three companies ultimately controlled by Mr Iskandar Safa (the Privinvest Companies) for the supply of fishing boats, security, and facilities linked to the exploitation of the tuna fisheries of the Mozambique's Exclusive Economic Zone in the Indian Ocean. These purchases were financed by two credit facilities agreements (CFAs) in the London form and a third private placement loan note style agreement also fairly standard in London financial practice (together the finance agreements). The leading lenders under the finance agreements were various entities of the Swiss bank, Credit Suisse, and the deals were arranged out of Credit Suisse's London branch: Credit Suisse Europe. The liabilities of the three SPVs under the finance agreements-including USD1.762bn of principal-were backed by sovereign guarantees under which the Republic was the primary guarantor. (ii) Building on the Meron's analysis of state responsibility, Buchheit et al. 's account of contractual responsibility under New York law, 6 and Jeff King's related analysis of the enforceability of loan contracts having unlawful purposes, 7 what is the current English law of powers and agency as it relates to sovereign states entering into finance contracts and on which party does the risk of agent abuse of power fall (section III)? (iii) Given the legal deficiencies and empirical evidence, what possible emendations to the English law of agency in these cases might counter the mischief of official agents and others conspiring to abuse their power when negotiating bank-to-state CFAs (section IV)?
This article builds both on modern legal histories of the transactions and on doctrine underpinning loans made to states and, with its Mozambican-focus, country-specific analyses, which combine several legal subdisciplines: the private (international) law of loan agreements, 8 work on the G20 and Paris Club 2020 Debt-Service Suspension Initiative, 9 country-specific analyses of sovereign financing, 10 and work on the Highly-Indebted Poor Country debt forgiveness programme that foreshadows the move to bank-to-state debt. 11 This article adds to this literature by focusing on Mozambique, which to date has rather been the subject of studies on internal corruption. 12 One should also note an innovative study 13 of creditor-China and its banks-rather than debtor state legal practices.
The second limb of this article analyses a specific doctrinal issue relating to lending to sovereigns: where a CFA is signed in abuse of authority by a corrupt official. This particular wrong would therefore fall foul of the 2012 restatement of United Nations Conference on Trade and Development (UNCTAD) Principles on Responsible Sovereign Lending and Borrowing, 14 Principle 3 of which provides that: Lenders have a responsibility to determine, to the best of their ability, whether the financing has been appropriately authorized and whether the resulting credit agreements are valid and enforceable under relevant jurisdiction/s. There Michael Waibel argues that the Principles 'reflect, to a considerable degree, how states and creditors of states act', 15 and while Waibel's focus is primarily on the restructuring aspects of these Principles, there is a need for an empirical analysis of actual loan agreements to assess whether compliance with Principle 3 is indeed the norm. Anna Gelpern is more circumspect, 16 writing '[a]gency problems, time inconsistency, information asymmetries, and moral hazard are perennial risks in sovereign lending and borrowing. Public officials borrow in the name of the people, but not in their interest ' . 17 She argues that there is a need for 'substantive and technical elaboration over time' and that the Principles could support best practices, due diligence, and domestic legislative efforts. 18 One attempt at this augmentation has been the Institute of International Finance's 2017 Voluntary Principles for Debt Transparency, 19 but while expanding on the notion of full-disclosure, they too are voluntary. In analysing one such substantive and technical aspect and in proposing the binding legislative reform, this article also builds on Gelpern's call to action in concrete ways.

II. E M PIR IC A L E VIDENCE OF GROW TH IN BA NK-TO -STATE LOA NS
A. The evidence 1. Typology As stated above, we are concerned with a very specific subset of what the International Monetary Fund (IMF) generally classifies as 'loans' . 20 Following legal practice, five categories can be distinguished: (i) CFA means (a) any kind of loan, interest on any loan, guarantee or indemnity, and any other kind of accommodation or facility in the nature of credit and (b) any fee relating to a matter specified in (a), structured to be bilateral or (part-)syndicated. (ii) Multilateral agency (MLA) means international institutions with governmental membership, which conduct all or a significant part of their activities in favour of development and aid recipient countries and which include multilateral development banks. 21 (iii) Private finance party (PFP) means any person acting as lender, arranger, or agent, which is neither a state nor MLA. (iv) Bank-to-state CFA means any CFA whereby one or more PFP agrees to provide credit facilities to a state, or organ of a state, for consideration (usually fees and margin). (v) Bank-to-state debt means debt incurred under a bank-to-state CFA.
These definitions are terse and do require some familiarity with the jargon. The definition of bank-to-state CFA is designed to capture bilateral and multilateral credit agreements under which banks, normally, agree to make credit available to a state. The state may then choose to draw down this credit, usually in a prescribed manner. In our case, the majority of such facilities for which there is information are stated to be on a 'term' basis, that is, the borrower is expected to repay the loan according to a fixed amortization schedule. Such agreements have a relatively standard form, even if the governing law differs. This standard is the so-called London form of CFA as promulgated by the Loan Market Association.
The definition of bank-to-state CFA is non-exclusive in that it captures any CFA to which a PFP is a party even if a non-PFP-a state or MLA-is also a party. when states and MLAs do enter into agreements as part of a syndicate with PFPs, it is usual that CFAs still reflect the legal substance of the CFAs that interest us. It is also common, however, for PFPs and MLAs to provide funding under separate CFAs, their respective rights being dealt with by an intercreditor agreement.

Method and empirical methodology
The data concerning bank-to-state CFAs were collated on 11 October 2018 (the first reference date or '1RD') and rerun on 10 November 2019 (the second reference date or '2RD'), and all data stated to be my own are drawn from this one data set, cross-checking with available data from Jubilee Debt Campaign. 22 The data were sourced from a standard Bloomberg terminal and are not compared with data sourced from Bloomberg's chief competitor Thomson Reuters Datastream. The principal reason for this is that Datastream does not provide the granularity required to isolate bank-to-state CFAs, being rather focused on fixed income debt (commonly known as bonds).
As to Bloomberg, a LSRC loan search was applied to all active loans, ranging over deals and tranches. These were filtered using the Bloomberg Industry Classification for Sovereigns, within this selecting the subsets: governments, local government, regions, and government agencies. This excluded lending to central banks and other public financial institutions. The African data were filtered by country of risk 'all Africa', 23 which limited the obligors, including guarantors, to that continent. They were further filtered: (i) restricting to date of loan signing: 1 September 2008 to 31 August 2018; (ii) restricting to structure type: bilateral or (part-) syndicated; and (iii) excluding public finance parties-only deals, which was done by manually filtering states and each relevant MLA.
The search garnered 234 results, which provided the following information where known: (i) parties, (ii) dates of signing and maturity, (iii) total commitments (amounts capable of being borrowed) and total outstanding, (iv) guarantor, and (v) loan type.
The data were restricted in time and space for the following reasons: (i) The credit crunch of 2008 is a marked macroeconomic event, particularly in the financial sphere. In the short to medium term, it offers, if not a 'year zero', at least a point in time when finance retreated, and policymakers began their attempts to reconfigure financial regulation. (ii) All Africa has been chosen as the initial lead on this research arose from a deal between European banks and an African state. The analysis could be readily extended to all emerging markets for which data are available.
The data are subject to the following caveats: as arranger, but the lenders were PFPs. This should have not been excluded by the search, but the lenders may have been ignored by the classifier or not disclosed. (iii) The data indicated a handful of deals where PFP lenders were sub-participants. Legally speaking, they are not parties to a bank-to-state CFA but to a back-to-back arrangement with a nominal lender under the bank-to-state CFA and indirectly exercise many rights under the bank-to-state CFA. It may be the case that synthetic structures such as this were included. (iv) Both active and matured deals were searched, although the results did not differ. This seems improbable, and perhaps matured deals were not picked up by the search.
The data were then subjected to a largely first-order statistical analysis, which revealed the following evidence of note. This evidence will be used to bolster the wider critical doctrinal methodology applied to the case for reform around bank-to-state CFAs.

Growth in bank-to-state debt
If we look at Figure 1, we see a growth trend from Q3 2008 to date, this being roughly approximated by the linear trendline shown.
We observe that these deals are somewhat cyclical, tending to occur in bunches biannually. The 12 largest deals, constituting commitments greater than or equal to USD20bn, severally involved Egypt, Mali, and Cameroon. The respective lenders are set out inTable 1.
As we are primarily interested in the behaviour of the parties to bank-to-state CFAs, it is most interesting to see who amongst those reporting these deals is most involved in the subsector. Seventy-one different banking groups were found to have signed bank-to-state CFAs during the reference period. By this is meant that financial institutions may have signed agreements in various capacities and that it is not uncommon for different members of a banking corporate group to play these roles. For example, a secured loan will require a security agent, and where the security is subject to German law, for example, it is not uncommon for the security agent to be a    separate legal person incorporated under German or Luxembourg law. Deutsche Bank, London Branch, may act as a lender and facility agent, while Deutsche Bank Luxembourg S.A. will act as a security agent. These group entities have been counted as a single economic unit for the purposes of this analysis. Many such groups engaged in one or two deals during the reference period. Table 2 shows those 21 groups that signed three or more deals during the reference period. As one can see, even where data have been collected, a number of deals took the option of not disclosing some or all of the lenders. In the instances considered, there was no pattern to non-disclosure: sometimes the disclosed arranger was an MLA; sometimes some private banks were not disclosed.
Unsurprisingly, we find the UK's Standard Chartered as the leading 'pure' private bank in the sector. Standard Chartered has a long history of lending in territories of the former British Empire. In a similar vein, we find French banks lending in Francophone and North Africa, while Germany has developed links with Cameroon. Middle Eastern and North African banks feature heavily, although their lending is concentrated in Egypt in the sample. Chinese and Indian finance also predominates, although interestingly Chinese banks are the primary vehicle for credit, whereas Indian finance comes largely from one source: EXIM Bank of India.
A related factor for consideration is not the headquarters of a given bank but the branch at which a bank-to-state CFA was signed. This information was not provided in the majority of cases, but where it was we found that many banks operated through their Johannesburg branch or through London. The lack of data makes it difficult to draw conclusions on a preference for branch accounting purposes. One should note that where bank-to-state CFAs are signed in these branch jurisdictions, one would expect the choice of governing law to be English law. This is because an investment-grade CFA is expected to be English law-governed, whereas socalled 'South African Agreements' are a deviation from the norm and are governed by local law. 24 If we analyse the data by number of deals in a given year irrespective of value, a different picture emerges: the number of deals tends to fall over the decade even as total amounts increase, as shown in Figure 2.
What we see is an inverse trend: the number of discrete deals falls even as the value of these deals increases. Of particular note is the fall in deals and deal value in 2013. The reasons for this are not immediately obvious, but there are at least three macroeconomic factors that may have played a part. Firstly, on the supply side, Spring 2012 broadly marks the end of the deepest point of the Eurozone crisis, when various measures had been put in place to shore up Southern European banks and restructure Greek debt. Plausibly, this hangover from the 2008 crisis reduced the willingness of European banks to lend, especially in emerging markets. Secondly, 2013 marked a major tour of Africa by President of the USA Barack Obama at which he announced a large aid programme to fund infrastructure. These funds may well have crowded out private debt that could have funded those projects. Thirdly, 2013 also marks the end of a period of African debt relief under the Heavily Indebted Poor Countries scheme, where mainly Paris Club creditors, but also some private creditors, provided a degree of debt relief. This scheme, which was largely focused on Africa, may have instigated a period of fiscal retrenchment under which financial indebtedness was constrained through to the end of 2013 before new rounds of borrowing, based on an improved debt-to-gross domestic product (GDP) ratio, began in 2014. Yet as we shall see when comparing bank-to-state CFAs with all kinds of debt, all debt continued to increase in 2013. This is perhaps due to alternative forms of sovereign borrowing, such as bank-to-state CFAs, bearing the brunt of reduced activity.

Lending concentration by state
Which states are contracting for bank-to-state debt? Using country of risk as the determinant, the data can be mapped as an intensity chart in Figure 3 (crosshatched states lack data).
One notices three zones of concentration: (i) Francophone Africa, (ii) Tanzania and Uganda, and (iii) variously South Africa and Egypt. South Africa and Egypt can be separately explained by the nature of their economies and the involvement of state and credit agencies in large projects in these states. High levels of bank-to-state debt should not be surprising for these states. The case of Francophone Africa is explained by the continued role of France in the affairs of these countries combined with France's dirigiste model of corporate governance. That is, French banks, like other major French corporates, are more likely to be involved in lending that would be undertaken by public bodies of other states such as the UK. However, while the data show the significant presence of French banks in Mali and Côte d'Ivoire, bank-to-state debt is also provided to a large extent by Middle Eastern and North African banks to Niger, and German state-backed finance houses to Cameroon.  What about Tanzania and Uganda? Surprisingly, perhaps, this concentration of bank-to-state debt is due to Japanese banks. TanzaniaInvest reports 25 strong ties between these countries. Exports from Japan to Tanzania have grown by 275% in 10 years , amounting to approximately USD266.7m in 2014. Similarly, exports from Tanzania to Japan attained growth of 193% in the same period, amounting to approximately USD210m in 2014. The Ugandan investments appear to be linked to the use of Japanese banking corporations (as well as state bodies) to fund oil and gas projects there and in East Africa generally. This is driven by an internal assessment by Japan of its energy needs following the Fukushima disaster. 26

Sources of lending -London and London branches
Finally, the data hint at the role of London as the locus for bank branches involved with bank-tostate CFAs. To emphasize, our concern is with the current state of English law in its treatment of ultra vires state contracts, and questions of agency are governed by the law of the contract (English law), with London also likely to be the exclusive forum. Although data were sought on the governing law of bank-to-state CFAs, insufficient data were forthcoming. This article therefore assumes a correlation between English law-governed bank-to-state CFAs and bank-to-state CFAs to which UK banks or non-UK banks acting through London branches are a party. Indeed, as with the case of Johannesburg branches, one expects many more bank-to-state CFAs to be governed by English law than just these London-originated deals.
Of the 235 deals analysed, 13 separate deals were found involving UK banks without a qualification as to a non-UK branch (Barclays, HSBC, and Standard Chartered) and 14 more which stated that the banks (Table 3) operated from their London branches. The data on this issue suffer from three problems. Firstly, the data were gathered pre-Brexit, and so, the European Union (EU) law on company branches was in force. This allows a company incorporated in an EU Member State to establish a branch in any other Member State for accounting purposes. 27 The data do not capture whether a given non-UK bank is using a separate legal entity in the UK to lend-although FBN Bank UK Ltd may be such a case. Secondly, the data are highly reliant on the correct recording of branch information, if this has occurred at all. Thirdly, the data include 14 deals involving undisclosed lists of lenders who may be UK-based. 25 https://www.tanzaniainvest.com/japan (visited 9 July 2021). 26 The World Folio, 'Japan looks to grow its presence across Africa' (2016), http://www.theworldfolio.com/news/japan-looksto-grow-its-presence-across-africa/4264/ (visited 9 July 2021).

B. Comparison with wider movements in sovereign debt
Returning to Figure 1, overall, the aggregate value of these deals expands over the decade to date. What does this expansion mean in the context of all African debt? Drawing on statistics from the Bank for International Settlements, Figure 4 charts the total debt lent to all emerging markets (Africa not being specifically differentiated in the Bank for International Settlements data) for the period. The number of states and level and instruments of indebtedness are significantly greater in this instance, providing robustness to the positive growth indicated by the trendline. Given that the figure is cumulative, of most interest is the increasing trend for all debt which manifests growth (some USD2.2bn/year) of all debt over the period. The much looser fitting trend for bank-tostate CFAs shows a smaller increase over the decade, being of a different scale (approximately USD214m/year). Thus, we only draw the simplest conclusion: that the growth in bank-to-state CFAs into Africa increased at the same time that all debt increased into emerging markets, although the growth in bank-to-state CFAs was intermittent, with bigger biannual leaps compensating for quieter intermediate periods. On this basis, bank-to-state CFAs do not seem to replace more traditional sovereign debt finance (bonds and MLA debt). Rather, it appears that this form of financing, characteristic of corporates, is expanding to meet the 'demand' for alternative forms of credit and potentially alternative 'corporate scale' credit purposes as markets are developed. In other words, there are only so many 'big ticket' infrastructure projects that can be built.
That is the general trend for all external debt, but can we find a more specific comparator? The International Debt Statistics of the World Bank provide such finely categorized data, although not fine enough to identify Bank-to-State Debt. Here we identified the debt disbursed and outstanding made available by external private creditors (principally financial institutions such as banks) to governments, aggregated on an annualized basis. 28 In other words, we are observing at each data point the aggregate amounts of debt drawn down by sovereigns and the amounts to be repaid, noting that if an amount is outstanding for a period greater than 12 months then it will be counted twice, once on each reference date falling within that period. The states chosen were those of sub-Saharan Africa. Data were available for 44 such states, the most notable omission being the Democratic Republic of Congo. Figure 5 shows faster growth than that for the bank-to-state debt (see Figure 2). A similar visualization of the composition of these values by state over time is unhelpful, but the data do provide insights. South Africa is the leading debtor of the 44 states, being subject to private debt of almost six times that of the next most indebted sub-Saharan state, with USD73bn in 2017 falling to USD68bn in 2018-some 49% of sub-Saharan Africa's private creditor debt burden. Amongst the remaining sub-Saharan states with data for the time period, there are nine such states with disbursements and outstandings in excess of USD2bn as on 31 December 2018, accounting for roughly 42% of the total for sub-Saharan Africa. These states manifest the greatest movements in indebtedness, as Figure 6 shows.
These absolute trends must also be relativized to the GDP of the individual states if we are to obtain a full picture. Figure 7 is a heat map of private creditor disbursements and outstandings in 2018 per state, as a percentage of GDP. 29 While Cabo Verde is an outlier with private debt-to-GDP of 25%, six African states have the dubious honour of owing between 10% and 20% of their GDP to private creditors: in descending order Zambia (19.8%), South Africa, Cote d'Ivoire, Gabon, Ghana, Sudan, and Angola (10.6%).
Mozambique (the beginning point of this research), although its borrowings from private creditors are in the hundreds of millions, not billions, has not bucked this trend. While its private debt to GDP ratio is some 5.0%, 30 the trend of increasing private debt in absolute terms has followed that of the major sub-Saharan debtor states. In 2008, government debt (disbursements and outstandings) was USD7.  C. Drawing conclusions from the data These data establish a general trend of increasing sub-Saharan sovereign indebtedness to private creditors, providing context for the underlying growth in bank-to-state CFAs established. While the identified bank-to-state CFAs constitute only 2.8% of the total private creditor debt to governments, when we consider these amounts on a state-by-state basis, we observe cases for concern, such as those of Mozambique, Angola, and Cabo Verde. The growth in bank-to-state CFAs forms part of this generalized growth in private debt, but existing World Bank statistics do not have the granularity permitting us to identify the legal nature of the private creditororiginated indebtedness disbursed and outstanding. It has therefore proved necessary to glean such data as we can from other financial sources. These data are inherently incomplete because of the voluntary nature of reporting and the general confidentiality of bank-to-state CFAs-a confidentiality that this article claims aids fraudulent loan-making.
This article argues that the empirical data set against just this lack of transparency motivate consideration of whether the English law of state liability for bank-to-state CFAs should be reformed to increase transparency. Our particular focus will be the legal principles encapsulated in UNCTAD Principle 3: the law relating to agency and authorization of state officials. We have already noted Anna Gelpern's counsel that these UNCTAD Principles should support but also be fleshed out in particular initiatives, including due diligence and legislative reforms in individual jurisdictions. If the numbers of bank-to-state CFAs from banks' London branches continue to increase and increasingly burden states, then the opportunities for fraud-fraud relying on what this article argues is the currently antiquated stance of English law-will only increase. When we consider that the cases of Mozambique (and as we shall see Ukraine) demonstrate the stresses placed on the law of want and abuse of official authority and creditor knowledge, one would expect that we will also see an increasing number of cases before the High Court or arbitral tribunals in which the spotlight is thrown on City of London banking practices operating in the creditor-friendly shadow of English law. Therefore, in section III, this article asks whether English law unwittingly favours corrupt transactions involving bank-to-state CFAs by placing the greater part of the risk of corruption on the side of states.

III. THE CUR R EN T STATE OF THE L AW
As LJ Garay Salamanca has shown, 32 bank-to-state CFAs are hardly new, having developed in the Eurodollar market in the late 1960s, before taking off in the late 1970s, particularly as more actors entered the market. Thus, Gradually, the criteria for credit extension as well as the borrowers' objectives deemed eligible for financing became more flexible, and more financing was granted to the public sector. … In the Latin American case, the public sector share in the stock of medium-and long-term external commercial bank debt increased from 53 percent to 58 percent between 1975 and 1981, while the share of commercial bank debt increased from 49 percent to 65 percent of the total stock of medium-and long-term external government debt. Meanwhile, the share of short-term debt in the total commercial debt balance increased from 21 percent to 28 percent during this period. (IDB, 1984) 33 The consequences of defaulting on these loans are well-known. States languish in a preinsolvency law realm in which sovereign pseudo-bankruptcies descend into a creditor free-forall. Nor need this article rehearse the many instances of (un)successful workout, the calls for some kind of international insolvency regime perhaps modelled on the US Bankruptcy Code, Chapter 9, 34 or the ad hoc efforts of stakeholders to ameliorate this situation, for example, through recommendations on the adoption of Collective Action Clauses, 35 or the Highly-Indebted Poor Countries (HIPC) programme. 36 Our concern, however, is not the endgame of indebtedness but rather the moment of creation of the indebtedness-a moment in which debts may be incurred in a questionable manner.
Our particular concern is just those situations where one or more banks lends to a state on the basis of a CFA governed, for the sake of this article's argument, by English law (although the discussion is generalizable to other common law systems that follow English agency law, such as Singapore and Malaysia). As noted above, the UNCTAD Principle 3 promotes the idea creditors have responsibility to determine whether a creditor has been authorized, but this responsibility is not reflected in English law. First, the state must act through its officials, and, as in the case of the Mozambique loans scandal, those officials may be acting for some alleged ulterior motive in binding the state to a CFA. As a matter of applicable law, will the state be bound to such a contract? Excluded from this analysis are cases identified by Jeff King, 37 where the contract is contrary to public policy or otherwise for an unlawful purpose-there may well be fraud in cases such as the Tuna Bond Scandal, but the fraud in such cases uses the lawfulness of a contract as a façade.
The basic principle, from which our analysis follows, is that as a matter of English law a foreign state that enters into a private English law agreement with a private entity is regarded as acting analogously to a corporation. 38 Indeed, 'the case of a foreign government is the same as that of a corporation, and the ordinary practice applicable to a corporation must be adapted, as well as it would admit…. ' 39 In what remains the leading international law analysis of state's liability for contracts and their right of repudiation, Theodor Meron 40 finds that in the municipal laws that he has examined, states are either explicitly regarded as corporations or implicitly so in that the analysis of authority that follows assumes an equivalency. English law appears to adopt the implicit approach, which the Court of Appeal quoted in Law Debenture Trust Corporation v Ukraine (2018) 41 : In conclusion, it appears that although analogies with corporations might usefully be drawn in particular respects, a foreign State in English law is not a foreign corporation, or an English corporation, either sole or aggregate, or a "quasi-corporation", or even an English legal person of whatever kind.
The Court of Appeal qualified this formulation 42 : Ukraine was present for the purposes of private rights in England because of recognition by the Crown, acting through Her Majesty's Government. The recognition conferred legal personality (although not a personhood constituted ex nihilo by English law or foreign law), and the cases were clear that this personality was its own and not the same as that of individual state officials. Further, some analogy could be drawn with international organizations that are explicitly regarded as corporations. 43 In any event, it was enough to observe that having legal personality, a foreign state had the power to make contracts and to authorize agents to make them on its behalf.
The finding that a state has a capacity analogous to a corporation leads us to the following analytical questions: (i) Can a state only act in furtherance of its objects? (ii) May a state only act by deploying its powers? (iii) How does a state, being a legal person, act through its duly authorized agents (directly or indirectly, natural persons)? Questions (i) and (ii) are usually regarded as part of the ultra vires rule and should be considered first in any legal analysis. Question (iii) initiates a secondary consideration of the rules of agency, which tie directly back to UNCTAD Principle 3.

A. Ultra vires
On the above assumptions, the Court of Appeal in Law Debenture Trust Corporation (LDTC) v Ukraine 44 regarded the application of the doctrine of ultra vires as fairly straightforward: English law imposes no restrictions on the capacity of those who have legal personality under English law, so as to render any act by them ultra vires and void…. There is no prerogative or statutory authority for limiting the capacity and powers enjoyed by a foreign state, which has legal personality in English law by virtue of its recognition as such by HMG. It follows from the absence of any limits at common law that as a matter of English law the foreign state enjoys unlimited capacity. Where it enters into a contract governed by English law, a foreign state will not therefore lack capacity to make and perform the contract, irrespective of the provisions of its own domestic constitution and laws. 45 Thus, 'recognised foreign states have unlimited capacity (objects and powers) to contract and dispose of their assets notwithstanding any restrictions there might be in their constitutions' . 46 That a local law or constitution purports to limit a state's capacity and so render an ultra vires contract void (as in LDTC v Ukraine) was beside the point-English common law followed public international law in regarding a state as labouring under no such restriction and so the contract was valid.

B. Authority
We therefore turn to the third question: whether the purported agent of the state has due authority to bind the state to the CFA in question. As far as English courts are concerned, 47 this is a matter of the English law of agency as it pertains to contracts.
The first principle is that it is the state that has the power to do some act and that an agent can only do that act if that power is donated to it. 48 The second principle, to which this article shall return, is that where an agent is the donee of a power, they must use that power to pursue the objects of the donor and not for an improper purpose. Upon these two pillars is erected the following structure: an agent may have actual express, actual implied, or ostensible authority to bind the state. In short, actual authority is where the agent has expressly been donated a power to some specific act or more commonly has been appointed to some office or other role, which implies that they have authority to carry out a range of acts. 49 Thus, as the law stands (and local legal opinions will confirm), a finance minister implicitly has authority to sign off on a bond issuance just by virtue of being finance minister, but this is less likely to be the case for a junior undersecretary for sports affairs. 50 Up to 2006, ostensible authority arose fairly often in the domestic courts because they dealt with cases where a third party was led to believe that a person was an agent of a corporation. Lord Diplock settled the rules for determining whether the corporation was bound, namely that (i) there be a representation to a third party by a principal that the agent has authority; (ii) the principal itself has the power to make such representation; (iii) the third party be induced to contract, and (iv) the corporation's constitution permits the contract. It is enough that those with actual authority acquiesce where the agent purports to act. 51 Nevertheless, arguments as to want of authority have proved to be a battleground in cases involving states. 52 We need to look no further than LDTC v Ukraine, 53 where Ukraine's argument was that the Finance Minister who executed a trust deed constituting notes to be issued to the Russian Federation lacked authority due to internal borrowing limits. The Court of Appeal held in 2018 that either the Finance Minister had usual (sc. actual implied) authority or failing that that he had ostensible authority unless LDTC knew or ought to have known that the Finance Minister was exceeding these internal borrowing limits. The Court of Appeal determined that LDTC did not and could not be expected to have such knowledge (even Ukraine was initially unaware of the limit breach). 54 If the Supreme Court confirms the existing view that state finance officials possess actual implied authority to bind the state to CFAs (unless the third party has notice of an actual internal restriction on that authority), then-in cases such as the Tuna Bond Scandal where deceit is in issue-much of the weight of argument will fall not on whether an agent had authority to exercise a state power but on whether they exercised it for an improper purpose. This appears to be a straightforward question, but because the improper purpose may be bound up with some fraud, the case law is difficult. This is because a third party dealing with a 'corporation' can be expected to know the powers of that 'corporation' , given the public nature of its constitution, but it can hardly be required to investigate whether the internal procedures of the 'corporation' have been followed such that an agent, actually having the power to do some act, is acting for its principal's benefit and not for an improper purpose. Rather, the third party is entitled to assume, absent notice, that the internal management of the company is sound: the so-called rule in Turquand 55 extended to states in Donegal International Ltd v Zambia. 56 Following the courts' analogy, the leading English corporations law case is Rolled Steel Products Ltd v British Steel, 57 where the Rolled Steel Products Ltd (RSP) director (Mr Shenkman) signed a guarantee in favour of SSS Ltd, in which he was personally interested. RSP had the general power to guarantee, but this guarantee in no way benefited RSP; rather, it benefited Mr Shenkman personally via SSS Ltd. All parties knew of this abuse. It was held that (i) the act was intra vires; (ii) the power was not used to further the company's objects; (iii) SSS Ltd, being the 'creature' of Mr Shenkman, had actual knowledge of the fraud; and (iv) having such knowledge, SSS Ltd could not rely on the rule in Turquand. to explain the decision in Hopkins v TL Dallas Group Ltd, 58 in which a deputy managing director fraudulent signed highly disadvantageous letters in his favour and to the knowledge of the counterparty.
An initial reaction 59 to the Mozambique scandal followed Rolled Steel, or at the very least analogized from the basic contract law principles that a fraud between agent and third party does not bind a principal. The problem with this 'gut reactions' was that it ignored a rather central distinction between Rolled Steel and the Mozambique scenario, at least as initially reported. Remember, at the time it was not known that Credit Suisse employees had been involved in bribery, and these same employees continue to deny that they knew of any conspiracy directed at the Republic. In Rolled Steel, the third-party beneficiary, SSS Ltd, was a co-conspirator with the agent director. That is why the third party could not rely on Turquand: it knew or was deemed to know that the agent director was abusing his power. In the Mozambique scandal, on the facts as we understood them in late 2018, the banks did not know that the state officials were abusing their power. As such, Rolled Steel says that the banks could by analogy rely on the Turquand Rule and that they could assume that the agent, having actual implied authority, had been internally authorized to sign the finance agreements. The agreements were binding on Mozambique, and if the officials had breached fiduciary duties to their principal, Mozambique's claim was against these officials and not the banks.

C. Consequences and policy implications
Let us now summarize our results. Ultra vires is inapplicable to a foreign state; it has unlimited capacity to enter into a contract such as a bank-to-state CFA under English law. Where the validity of the contract is called into question due to want of authority of an actual or purported agent of the state, then on a finding of lack of authority the contract is voidable-the state may elect to affirm the contract or treat it as void. If an agent has authority but the third party knows that the agent is abusing that authority for an improper purpose, the contract will be voidable. If the third party has no notice of the abuse, they are entitled to assume the agent's actions are approved by the state and the contract is enforceable.
The voiding of a disbursed bank-to-state CFA can be disastrous for a bank: evidently the property (money) has gone somewhere and the bank may be out of pocket several million USD. If a contract procured by fraud against the principal is upheld and is enforceable, it is the state that bears this loss; it is of less concern to the bank who must repay the loan, for the bank has performed its side of the bargain and is entitled to delivery of the specified sums in repayment by the named obligor.
The consequences of the current state of the law for a country such as Mozambique appear to be this. Mozambique has unlimited objects and the capacity to enter into loan contracts. Depending on their status, an official of Mozambique has implied actual authority to bind Mozambique to bank-to-state CFAs and may have ostensible authority in certain circumstances. In the absence of actual notice that an official is abusing their power to bind Mozambique to a CFA, a foreign party is entitled (pace the UNCTAD Principles) to assume that the official is authorized to bind Mozambique. The bank-to-state CFA is therefore valid, and Mozambique has only an 'internal' claim against its official.
Buchheit The sovereign debt context raises a typical concern of agency law -that of faithless agents who purport to bind principals to an obligation to third parties when the fruits received in return for the obligation only accrue to the agents. Who, as between the principal (country and citizens) and the third party (the lender) should bear this risk of the faithless agent? 60 As we have seen, UNCTAD Principle 3 places responsibility on creditors to verify authority, but we have established that English law disagrees with that view. Likewise, Buchheit et al. affirm that as in England, the risk under New York law 'usually lies at the feet of the principal' for reasons that reflect 'both business expediency and some element of efficient risk bearing'factors that we have seen also informed the English law approach to corporations and agency since Turquand and indeed the international arbitral awards identified by Meron. Yet, whereas it seems that English law will only overturn the presumption that a third party may assume internal management is sound if the third party has actual knowledge of some fraud, Buchheit et al. inform us 61 that the New York cases are more forgiving. So, for example, if objectively a guaranty is of no corporate benefit to the guarantor corporation, then a third party is put on notice and should carry out further due diligence to establish why such an otherwise inherently ultra vires act is being undertaken. 62 Likewise, a third-party beneficiary of a guaranty cannot rely on a representation by a vice-president and treasurer of a corporate guarantor as to their authority to bind that corporation in the face of visibly suspicious circumstances or behaviour. 63 The line English law has drawn is hardly immutable; rather, it rests on a weighing of where the burden of due diligence and so risk should fall. Even if the view in Turquand is correct, it surely engages a balancing of interests materially distinct from our scenario in which state and private party are engaged. Indeed, this is the lesson of Meron's review of the cases: up to his time the trend was towards protecting foreign private parties from the sovereign power of state parties by placing the risk of official faithlessness on the state. The policy here is plain enough: the state should get its house in order, being deemed to be either corrupt or mismanaged such that corruption is prevalent. The whole discourse of 'corruption' is highly problematic, and this article does not seek to engage with it. At the other end of the spectrum, we have the 'underlying' English law rule that Turquand sought to mitigate, that a corporation registered under the Companies Acts was required by statute to publish its constitution and other information and that to give effect to regulation through publication a third party was deemed to have notice of constitutional limitations on authority irrespective of whether it was ignorant of them. In our case of state and foreign third party, it seems that English law has tended with the international cases to move towards protecting the third party. Only if there is actual notice of fraud will the third party bear the risk.
English law's implicit policy assumptions seem to be that (i) foreign private parties are not in a position to monitor the internal bureaucracies of states (cost/benefit calculation), (ii) the power imbalance of the relationship means that foreign private parties feel unable to carry out due diligence (a pseudo-duress argument), and (iii) the law ought to encourage states to improve their internal management to a standard the English court expects (the moralizing argument). The Court of Appeal in LDTC v Ukraine tacked towards a doctrinal discussion, and there are only hints that the cost/benefit calculation may be playing a part. It was tellingly silent on the UNCTAD Principles. The policy question must be asked however, and this article claims that the above bases are at least subject to some qualification, as the Mozambique scandal shows. Firstly, as to cost/benefit, Mozambique is one of the poorest states whereas Credit Suisse is a 60 Buchheit, above n 48, at 1240. Global Systemically Important Bank 64 whose business is assessing the creditworthiness of borrowers. Mozambique's GDP was USD14.396bn in 2018; Credit Suisse had annual revenues of USD23.66bn and profits of USD4.94bn in 2019. 65 The bank is more skilled and economically more capable in carrying out due diligence in the area. Nor would requiring due diligence impose an additional burden, for a cursory review of any investment-grade loan agreement shows the lengths banks go to, through conditions precedent and representations, to establish the capacity of obligors and their agents. To hold banks to a lower standard of diligence is to hold them to a standard they themselves do not accept. While Credit Suisse's relative economic size and skill support placing it under the greater risk burden, the UNCTAD Principles underscore that all creditors should be engaging in responsible lending and in particular should be checking that appropriate authorization is obtained.
For its part, an argument from some form of duress remains highly specific to the facts, but the general trend of English courts has been to look to actual wrongdoing (such as a deceit or threat of breach of contract) as a basis for so drastic a finding. The advantage in this context of seeking an identifiable wrong is that this must be preferable to the vague prejudices about southern states that haunt the older international arbitral awards. If a state has prevented due diligence in some way, let this be pleaded out.
Finally, the moralizing argument is outmoded and arguably ineffectual. States such as Mozambique lack the resources to enforce strict internal oversight processes. As David Stasavage has shown, 66 where Mozambique has been able to reduce corruption, it has been because of increasing the wages of functionaries and reducing the gap between the black market and official exchange rates for the metical. Yet, Stasavage continues, this has not been enough: Mozambique cannot afford to employ sufficient officials, which results in individual officials having wideranging economic power and discretion. Furthermore, as Tvedten and Picardo confirm, while initial anti-corruption efforts made headway in the 1980s, the transition to a market economy led to increasingly complicated forms of corruption, which the state was simply ill-equipped to investigate despite its best efforts. 67 Finally, there has been a critical shortage of people skilled to take on official roles, either as civil servants or as prosecutors. The most capable are offered much better wages in the private sector or outside Mozambique. 68 In short, judicial attempts to chastise states such as Mozambique into stamping out corruption are misguided and patronizing. English law should work to facilitate the diligent work of the vast majority of skilled and committed Mozambican officials working to combat corruption.
Such is the state of the law at present: banks know that the English court places the risk of a corrupt state agent upon the state-principal, and that irrespective of the ultimate destination of disbursed funds, the bank's balance sheet will continue to note the liability of a sovereign state as an asset. This article has posited policy arguments as to why such a legal settlement is unacceptable, motivated by the empirical evidence showing why this apparently niche issue may well become increasingly important, and a source of further injustice orchestrated from London's financial and legal centres of gravity. What though, is to be done?

IV. DE VELOPIN G R EFOR M PROPOS A LS
A. Introduction

Reporting requirement
In keeping with recent trends to regulate corporate behaviour (for example, the Modern Slavery Act 2015 (the MSA2015)), an institution would be required to engage in internal due diligence regarding bank-to-state CFAs and to report these processes in its annual report, making a statement in its annual report that no bank-to-state CFAs have been made in breach of the UNCTAD Principles (or if so, explaining how). 71 This kind of reporting requirement tends to be enforced by means of summary criminal sanction (a fine or in the most serious cases imprisonment) for institution officers. This method of regulation does not attack the bank-to-state CFAs themselves, and states will still be liable under the relevant agreements.
The principal inspiration is the MSA2015 but with one substantial change: the MSA2015 is founded on specified criminal offences linked to slavery, whereas no such basis exists for CFAs. Therefore, this article proposes integrating the approach of the Money Laundering Regulations 2007 (the MLR2007), which focuses on improving institutional behaviour with respect to preventing and monitoring what institutions suspect to be unauthorized bank-to-state CFAs (cases where authority cannot be completely verified). This suspicion is defined analogously with MLR2007 and avoids requiring non-legally qualified actors to make an essentially legal judgement as to whether corruption linked to a bank-to-state CFA is taking place, shifting the focus onto a more subjective risk assessment of the satisfaction of conditions precedent-such as written proofs of authorization and ratification by state legislatures-already part of good banking practice.
Analogous norms, such as the MSA2015 or MLR2007, all have a basis in a general legislative provision, and so, one would expect this proposal likewise to derive from legislative change.

Legislative reform on voidability of CFAs
This option makes a small but effective statutory or legislative change to the existing common law to create a rebuttable presumption that institutions have notice of any lack of authority of a state's agent who attempts to bind the state to a bank-to-state CFA. In other words, it is motivated by and builds on the legal analysis undertaken in in section III above.
As we saw, the common law of state's agent authority is out of sync with UNCTAD Principle 3 in that a bank need not enquire terribly deeply as to the lack of authority of an agent, and a state may be bound even where an agent lacked actual authority or abused it.
This article claims that this prevailing presumption is inappropriate in the context of bankto-state CFAs. Rather, the presumption, to be set out in statute, should be that institutions are deemed to have notice of any want of authority of an agent unless the institution has met certain 'safe harbour' conditions of due diligence. The best such safe harbour would be to register the bank-to-state CFA on closing (section 1), giving the world notice of the transaction before drawdown.
The effect of a finding of (deemed) notice of non-authorization is that the state may not be bound by the bank-to-state CFA. The bank-to-state CFA is declared voidable, which means that the state may unilaterally opt to treat the bank-to-state CFA as void. If the state does decide to treat the bank-to-state CFA as void, then it is as if the bank-to-state CFA had never existed; the state is under no further obligations because the obligations never arose. If an institution has disbursed funds, the funds have been transferred unilaterally and in error and the onus is on the institution to recover them from whichever actual recipient of the funds, by means of certain technical procedures available for this purpose. 71 I would like to thank Dr Marie Pillon for her assistance in understanding the MSA2015. Unlike the reporting requirement approach in section 2 above, the voidability approach catches all English law-governed bank-to-state CFAs, and it is for the institution to ensure that appropriate due diligence is undertaken to overturn the presumption of lack of authority. This ought to be no difficulty for institutions, given that CFAs normally require proofs both as part of their representations and conditions precedent, and this is in line with UNCTAD Principle 3. Yet, such a fundamental change to private law would require primary legislation.

C. Conclusions on reform
A combination of the registration and voiding options would provide the neatest 'surgical' reform with teeth sufficient to emend bank (and rogue official's) behaviour. The registration or voidability option has the benefit of travelling a legal furrow well-known to English banks: a failure to register a floating charge already leads to the unenforceability of that security against third-party creditors. Furthermore, inverting the onus on due diligence to banks from states allocates burden and risk to the party most able to bear it in line with UNCTAD Principle 3. A registration requirement alone would still benefit from public oversight by civil society, and one would expect such transparency to be as effective as that provided for under the MSA2015. Yet, no one can wait months for an annual report to disclose a problematic loan; states and civil society need to be able to respond in the crucial period between signing of the contract and drawdown. Furthermore, there are surely benefits in an ex ante voidability option mirroring the existing floating charge regime that places the burden on banks to undertake the due diligence we should surely expect of them and their professional advisors.

V. CONCLUSION
In late June 2019, following representations by Jubilee Debt Campaign, the UK Labour Party adopted a proposal regarding the registration versus voidability law for CFAs. This proposal subsequently formed part of the Labour Party Manifesto in the December 2019 UK General Election. In January 2020, Credit Suisse issued a counterclaim against Mozambique for recovery of the USD622m it claims is owing, Mozambique having attempted to obtain a legal declaration that the CFA in issue be cancelled (along with its guarantee). 72 Once again, English law will be the crucible for adjudicating claims between international capital and states, and once again, a light will be shone on certain banking practices occurring in the City of London and involving branches of banks from around the world, practices benefitting from the current approach of English law.
The empirical data sketch a context of increasing borrowing by sub-Saharan states from private creditors and borrowings that in some cases exceed 10% and even 24% of GDP. Further, the amounts owing under bank-to-state CFAs, the legal instruments that are the focus of this article, have increased against this wider backdrop. This suggests that if English law continues to place the risk of an agent's abuse of power on low-income country principals, rather than on the skilled banks and their advisors, we are likely to see an increasing number of cases in which English law and the City of London are brought into disrepute. It is hoped that this article can provide further momentum towards much-needed reform. 72 Reuters, above n 32.