MACROECONOMICS, TRADE AND INVESTMENT MACROECONOMICS, TRADE AND INVESTMENT EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT Resource-Backed Loans in Sub-Saharan Africa David Mihalyi Jyhjong Hwang Diego Rivetti James Cust Resource-Backed Loans in Sub-Saharan Africa* David Mihalyi Jyhjong Hwang Ohio Diego Rivetti James Cust NRGI State University World Bank World Bank * The views expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/ World Bank, the Natural Resource Governance Institute (NRGI) or any affiliated organizations. We would like to thank Balint Parragi for excellent research assistance and Susan Maslen, Amir Shafaie for their contributions to the draft. We thank NRGI for their support to the project. © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved. This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO), http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Rivetti D. 2021. Resource-Backed Loans in Sub- Saharan Africa. EFI Insight-MTI. Washington, DC: World Bank. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third- party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Cover design and layout: Maria Lopez / lopez.ten@gmail.com >>> Abstract This paper investigates the characteristics of resource-backed lending across sub-Saharan Africa. To shed light on this type of lending, the paper presents new information on thirty resource- backed loans identified through publicly available information between 2004-2018. These loans are concentrated in a few countries, where they represent a sizable fraction of all borrowing, they are typically taken by central governments and state-owned enterprises. While loan terms are mostly opaque, where data is available, we find that such loans are not cheaper than regular loans. We highlight opportunities to improve transparency and offer some suggestions for improving the governance of collateralized borrowings across developing countries. >>> Contents 1. Introduction 1 2. Literature Review 3 3. Data 6 4. Results: Key Characteristics of RBLs 8 a. An Overview of the RBLs Analysed 8 b. RBLs by the Borrowing Entity 9 c. RBLs by Type of Lender 10 d. RBLs by Type of Collateral Arrangement Observed 11 e. RBLs by Maturity 12 f. RBLs by Interest Rates and Fees 13 g. Repayment Schedules 14 h. Allocation of Proceeds 15 5. Policy Considerations 16 a. RBL Importance as a Source of Finance 16 b. Transparency 21 c. Cost of Financing 22 d. Spending Considerations 25 e. Negative Pledge Undertakings 26 f. RBL Renegotiations 27 6. Conclusions 29 References 31 Appendix 33 1. >>> Introduction Developing countries often face difficulties accessing large-scale financing to meet their development needs. Major constraints are the lack of sufficiently deep domestic markets and the limited or costly access to international capital markets. In response to this challenge, a new resource-linked financing model has become popular that provides new opportunities for countries to access finance in exchange for, or collateralized by, future streams of income from their commodities: resource-backed lending. Resource-backed loans (RBL) describe the practice of using a country’s natural resources to serve as either a direct source of repayment or as an underlying guarantee of repayment in respect of the loans. This kind of borrowing is typically undertaken by a central government or by a state-owned company with lending from another government, a state-owned company, the private sector and/or international financial institutions. Such borrowing practices go back at least a century (e.g., Peru’s guano-backed borrowing in mid-19th century, Vizcarra, 2009), but have become widely used across resource-rich developing countries during the recent commodity boom (NRGI, 2020). The COVID-19 pandemic has generated considerable focus on these types of transactions in light of widespread debt sustainability challenges in developing countries combined with volatility in commodity prices. 1 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA RBL proceeds are often used to finance infrastructure projects. were identified through publicly available information. Based In theory, if the borrowing process is carried out transparently on this review, we highlight key lessons and considerations and effectively, if the loan has favorable terms and if the for improving governance of these loans and other forms of infrastructure projects are well selected and well executed, the collateralized borrowing across developing countries. transaction would be expected to generate positive returns for the borrowing country’s economy. Such increased economic The paper is structured as follows. Section 2 presents the activity helps to generate the tax base available to the literature relating to resource-backed and resource-linked sovereign for repaying the loans, and also helps the country lending. Section 3 provides an overview of the data set of deliver on its development strategies. However, in the event resource-backed loans that is analyzed in the paper, the of weak governance or weak capacity in debt management, data collection process and its limitations are also discussed. or if the relevant infrastructure projects are poorly selected Section 4 presents the loans and main findings relating or executed, large RBLs may become overly burdensome to them, and proposes different typologies for classifying for borrowers. RBLs can also exacerbate debt sustainability key features. Section 5 reflects on the main findings and problems, especially when they are large or if they set off a implications for policy making; for example identifying actions cycle of increased dependence on this type of borrowing. that might help improve the performance and governance of these instruments. Section 6 concludes. A set of RBLs undertaken by countries in Sub-Saharan Africa over the 2004-2018 period was analyzed, the loans 2 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 2. >>> Literature Review The history of collateralized lending and resource-backed loans goes back at least to the mid- nineteenth century when Peru borrowed using sizable guano deposits as collateral (Vizcarra, 2009). In Sub-Saharan Africa, this practice was first well documented in Angola in the mid- 90s using future petroleum revenues as guarantees (Brautigam, 2011). In the 2000s, the state-owned Export-Import Bank of China (Eximbank) started offering similar loans to Angola (Brautigam, 2011). Since then, the practice has become relatively commonplace in resource- rich countries in sub-Saharan Africa, Latin America and beyond through loans extended mainly by state-owned development banks from China and commercial players such as commodity traders (NRGI, 2020). Resource-backed loans are often labelled as or seen as a subcategory of collateralized loans. Collateralized loans are defined by the International Monetary Fund (IMF) and the World Bank (WB) (IMF-WB 2020) as loans where “the creditor has rights over an asset or revenue stream that would allow it, if the borrower defaults on its payment obligations, to rely on the asset or revenue stream to secure repayment of the debt. In a legal sense, it entails a borrower granting liens over specific existing assets or future receivables to a lender as security against repayment of the loan. More broadly, it also includes arrangements that do not constitute granting of a security interest, but that have an equivalent effect.” 3 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA As used in this chapter, and following on NRGI (2020), Others have emphasized the risks associated with these loans. resource-backed loans refer to loans provided to a government Manzano & Rigobon (2007) argue that the collateralization or state-owned enterprise (SOE) where: (i) the repayment of future resource revenues in the 1970s led to a large debt is either made directly in natural resources (i.e., in kind) or overhang in the 1980s, which was a major reason for the weak from a natural-resource-related future income stream; (ii) the economic performance of resource exporters throughout repayment is guaranteed by a natural-resource-related income this period. Venables (2016) opines that the absence of stream; or (iii) a natural resource asset serves as collateral.2 transparency in securing RBLs has, in most cases resulted in host countries being offered poor contract terms in exchange Non-commodity-related types of collateralized loans also for their natural resource wealth. Horn, Reinhart and Trebesch exist, notably those linked to infrastructure (e.g., ports, toll (2021) warn that Chinese loans to developing countries roads), and governments have also collateralized a wide collateralized against natural resource revenues have the range of asset classes, including proceeds from lottery ticket potential to become a threat to debt sustainability due to the sales, electricity generation, value-added tax proceeds (IMF- size of such deals and their lack of transparency. The lack of WB, 2020).3 These are not considered in this paper. transparency is highlighted by the authors’ estimate that 50% percent of Chinese lending (overall, not just RBLs) is missing In considering RBLs in this paper, we focus only on those from official statistics. loans whose proceeds are used for purposes external to the revenue stream that is used for repayment/as collateral. In the Some studies have focused on the role of China. Brautigam typology of IMF-WB (2020), these are labelled as “unrelated and Gallagher (2014) provides a systematic empirical collateral” transactions. As such we exclude project financing overview of Chinese commodity-backed loans (with data on or acquisition financing and carry arrangements (prevalent in loans until 2011) and discuss some key features of these the oil sector) and instead focus on financing with purposes loans. Sanderson and Forsythe (2012) analyze these deals beyond the resource sector. As a result, we excluded a large from the perspective of China’s development banks. Meidan section of conventional financing arrangements by state- (2016) analyzes their importance in terms of China’s access owned enterprises from this review; this is in order to focus to oil resources. on loans which can be considered substitutes to regular government borrowing. Another group of relevant studies focus on the resolution of debt repayment problems in cases where there is The academic literature on the use of resource-backed loans collateralized lending; and two papers look at the outcome or other forms of collateralized borrowing in developing of renegotiating collateralized debt extended by China. In the countries is relatively short. case of Gardner et al. (2020) the borrower is the Republic of Congo, while Acker et al. (2020) look at loans taken by Angola A handful of studies have put forward a strong economic and Chad. Both papers contrast collateralized loans with argument in favor of countries borrowing against their resource uncollateralized loans. wealth. A book by Halland et al. (2014) discusses the potential for resources to be used as collateral for infrastructure deals, There has been very little work on the socio-economic impact which we consider as a subset of RBLs. Collier and Cust of projects financed via RBLs. Fieldwork by Tang and Shen (2015) and Songwe (2013) discuss the important opportunity in Ghana found that the Bui Hydropower Project in Ghana, that natural resources represent for African economies to financed via an RBL backed by cocoa proceeds, “has improved borrow for development. Xu et al (2021) provide a model to local urban households’ access to electricity and increased justify how collateralized lending can alleviate credit rationing their ownership of electric appliances” (Tang & Shen, 2020). in poorly governed settings. All four studies provide some strong theoretical arguments in favor of such deals while The World Bank and IMF, among other international financial cautioning on some of the risks involved and providing limited institutions, have analyzed collateralized loans in order to empirical validation. formulate their own recommendations. The IMF-WB (2020) 2. Note that without access to the contract text it is not possible to establish definitively if a collateral is legally enforceable. 3. A broad definition of collateralization would also include arrangements that do not constitute granting a security interest, but that have a similar economic effect (i.e., structures that do not give lenders the legal right to seize/liquidate an asset in an event of default, but give them a “first mover advantage” ahead of other legally unse- cured lenders, for instance to withdraw from an escrow account). 4 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA concludes that collateralized transactions “can be beneficial to a There are also multiple case studies focused on a single developing country borrower under a range of circumstances”, or small number of RBLs. These include Alves (2013) who but also points to risks and discourages this practice when compares the experience of Angola and Brazil; Gillies and the proceeds are not spent on assets which can be used to Quaghe (2018) who discuss a proposed deal in Nigeria; repay the loans, when loans are excessively large and when and Landry (2018) who looks at the Democratic Republic of the details of such borrowings are opaque. The policy paper Congo’s (DRC) Sino-Congolaise des Mines (Sicomines) case. also touches on a number of considerations, many of which are also stated in this paper, including the fact that such loans The current paper adds to the literature by providing a more are often over-collateralized (they use excessive amounts detailed large-scale empirical review of existing RBLs based of collateral) to enhance the borrower’s creditworthiness. on data that we reviewed. Our work builds on earlier findings They also highlight how the secured nature of certain RBLs published by the Natural Resource Governance Institute might run afoul of negative pledge clauses in loan contracts, (NRGI, 2020), and extends it significantly using novel data including loans made by multilateral development banks. and further analysis. 5 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 3. >>> Data The analysis in this paper is based on an extended database of thirty major resource-backed loans. These loans cover the period 2004-2018 across sub-Saharan Africa, covering eleven countries. This database builds on a dataset of RBLs first published by NRGI (2020) which itself built primarily on the Johns Hopkins SAIS China-Africa Research Initiative’s (CARI) dataset on Chinese lending to Africa.4 This new extended dataset includes information on the lending entity, the borrowing country, the amount of the loan, the year of loan agreement signing, the type of resources involved, the nature of collateralization, when available, projects to be financed by the loan and repayment terms. The information collected on each loan is compiled from a variety of sources, including African government sources, lending entities, Extractive Industries Transparency Initiative (EITI) reports, AidData’s Global Chinese Development Finance Dataset (Dreher et al. 2021), investment reports from multilaterals, as well as fieldwork and interviews with experts. Primary source information is then complemented with a review of the financial press. 4. CARI-BU (2021) https://chinaafricaloandata.bu.edu/ 6 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA The data has been checked against the World Bank’s Debtor Our research identified 30 RBLs made in the 2004 to 2018 Reporting System (DRS). The DRS collects detailed loan- period in Sub-Saharan Africa that match the focal area of by-loan information on new public and publicly guaranteed our review and where sufficient minimum information was external debt. The DRS does not include information on available to incorporate into our dataset. These minimum any collateral features of loans, however. We therefore also criteria consisted of having details on both the lending and contacted relevant WB country economists or national debt borrowing entity, the size of the loan, the year the loan was offices to verify the information where possible. agreed and confirmation that the loan had a repayment period beyond a single year. While significant effort has been spent on ensuring the accuracy of the data, the reliance on mixed sources and Our dataset is by no means comprehensive, and there are methods may justify a level of caution and verification in the certainly more RBLs which match our criteria but for which use of facts and figures from this research in further analysis, insufficient information was readily accessible. The dataset, including referring directly to the underlying datasets and and therefore the analysis, is limited by the information that is sources for specific details. publicly available. As described further in Section 5b, the RBL landscape remains largely opaque, with limited information RBL contracts are rarely publicly available. Nevertheless, publicly available about the terms, and at times even the whenever publicly available, we analyzed contract text. These existence, of RBLs. were identified through NRGI’s oil, gas and mining contracts5 and Gelpern (2021). 5. http://ResourceContracts.org 7 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 4. >>> Results: Key Characteristics of RBLs a. An Overview of the RBLs Analysed Our research identified 30 RBLs in Africa signed between 2004 and 2018. Included in our review were 11 African countries which borrowed RBLs during the period: Angola, Chad, the Democratic Republic of Congo (DRC), Ghana, Guinea, Niger, the Republic of Congo, São Tomé and Príncipe, South Sudan, Sudan, and Zimbabwe. The relevant RBLs totalled USD$ 46.5 billion agreed in this period, of which Angola accounted for nearly half the total.6 7 The manner in which loans are discussed and aggregated in this chapter varies depending on the focus of the analysis. For example, credit lines can accommodate several independent projects, and thus entail multiple disbursements.8 In addition, we split up single ‘deals’ into separate loans where they had somewhat different terms (for example Ghana’s loans from the China Development Bank (CDB) [#13] & [#14] had two tranches with different interest rates and maturity, and thus are treated as two separate RBLs in our dataset). All the RBLs used in this paper are identified in the Appendix.9 6. For the purpose of this paper, commitment is defined as amounts committed under signed loan agreements. In the cases of large credit lines that support multiple projects, disbursed amounts are used when available. 7. All dollar signs ($) refer to United States Dollars (USD) in this chapter unless otherwise specified. 8. All of these grouped disbursements would share the same financing arrangement, from repayment terms to interest rates, and together be consider a single RBL in our dataset. Credit lines tend to have a limited drawdown or availability period, such as 8 years for Angola to drawn down a $2bn Eximbank RBL [#1] signed in 2007 Amounts that remain undrawn at the end of the availability or drawdown period is usually cancelled by the lender. Given that some credit lines may remain undrawn, or not fully drawn, by the end of the availability period, this paper reports both the total credit line amount made available at signing, as well as the amount actually disbursed (where such information is publicly available). 9. Each RBL cited in the paper is identified using [#ID] referring to numbering in the Appendix table A.2.. 8 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA The rest of the results section summarizes key characteristics leading to a protracted renegotiation with the oil trader that of the RBLs by category. First, we review the nature of concluded in 2018 (a successful renegotiation was also borrowing entities, second the types of lender, third the form a precondition for IMF support to Chad). In this case, the of collateral arrangements observed and fourth, the maturity sovereign’s provision of a guarantee meant that the RBL conditions. became a liability for the central government. Occasionally, the RBL borrowing entity is a joint-venture b. RBLs by the Borrowing Entity between the SOE of the borrowing country and a commercial company from the lender side or a special purpose vehicle (SPV) that has been formed. In 2008, a consortium of Chinese companies led by Sinohydro and China Railway Engineering The central government and state-owned natural resource Corporation (CREC) formed the joint-venture company enterprises are the primary borrowers of RBLs. We identified Sino Congolaise des Mines (Sicomines) with DRC’s state- 9 loans out of 30 where the borrowing entity is a state-owned owned mining company La Générale des Carrières et des natural resource enterprise. However, the distinction between Mines (Gécamines). The Chinese companies owned 83% of SOE and central government borrowing is sometimes blurry. the copper/cobalt venture. This newly formed joint venture company, Sicomines, then borrowed from the Eximbank [#9]. A central government’s control over SOE borrowing This loan also has an explicit financial guarantee from the decisions will vary depending on: the rules of the country, the government to repay the loan in case the designated revenues establishment or constitutive arrangements of the specific were not sufficient to fully repay the loan.10 As the modified SOE, or the sector in which the SOE operates. Such levels loan agreement states11: of control may range from very close to very tenuous. For example, in some countries where oil is a major export, the Article 10.3. « The Democratic Republic of Congo has sovereign/ central government might both fully own and guaranteed repayment of the debt, in line with the current control the oil-related SOE, and national legal frameworks Convention on Collaboration, through the granting of may prescribe that the sovereign has exclusive purview and mining concessions or concessions for other natural ownership over exploration and assets related to the oil and resources to SINOHYDRO if repayment of the loan due to hydrocarbon sector. Similarly, the central government may JV Minière falls short of the repayment needed to cover its be required under relevant legal frameworks to specifically investment in the Infrastructure Projects. » approve a relevant SOE’s borrowing decisions, and in some cases may guarantee the SOE’s repayment of a loan. In some Article 13.3.4. « In the case that JV Minière has not been cases, there may not be any explicit government guarantee, reimbursed for the investments and interests from the but parties to the loan may expect the government to provide Infrastructure Projects within 25 years of the start of the support in case the SOE runs into repayment difficulties (implicit loan, the Government of the Congo is committed to paying guarantee). In other cases, for example where an SOE is only the outstanding balance by any means necessary. » partially owned and not managed by the government, or where the SOE functions autonomously as a commercial enterprise Loans taken by Republic of Congo’s Société Nationale des separate from the central government, then the sovereign Pétroles du Congo (SNPC) from commodity traders were may not be aware of such SOE’s borrowing activities. kept off the Ministry of Finance’s books. The IMF and other observers became aware of these loans in June 2017 as For example, in 2013 and 2014 Société des Hydrocarbures du soon as SNPC experienced difficulties servicing them.12 [#19] Tchad (SHT, Chad’s national oil company) borrowed about $2 [#21] [#22] billion in an oil-backed loan [#7] & [#8] from the commodities trader Glencore to purchase back certain interests in Chad’s Publicly available data on sovereign guarantees that support Doba oil field from Chevron, and for budget financing. These SOE debt is sparse (see section 5/b). It was therefore difficult RBLs were guaranteed by the state. The oil price collapse in to ascertain how commonplace such arrangements were. 2014 put the Chadian government in a difficult fiscal position, 10 The original 2008 loan agreement guaranteed both the infrastructure loan and a loan for the mine. http://congomines.org/system/attachments/assets/000/000/276/ original/B5bis-Sicomines-Convention-Incl-Anx-2008-Consortium-Entreprises-Chinoises-RDC.pdf?1430928308 11. A 2009 amendment narrows the guarantee to the infrastructure loan only. http://www.congomines.org/system/attachments/assets/000/000/502/original/Sicomines-2009- Avenant3ConsortiumEntreprisesChinoises-RDC.pdf?1430928992 12. IMF, Country Report No. 19/244 on Congo: Staff Report (2019), www.imf.org/~/media/Files/Publications/CR/2019/1COGEA2019001.pdf 9 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA c. RBLs by Type of Lender From our review of 30 RBLs, we identified 13 different creditors of investment assurance. These entities consist of companies that extended RBLs to African countries during the period operating in sectors outside of the concerned natural studied. These can be grouped based on their economic resources, and who face high risk due to the investment functions or based on their classification as either an official environment. The China National Aero-Technology Import & bilateral creditor versus private/commercial creditor. Export Corporation (CATIC) provided $110mn to Zimbabwe for rural electrification equipment [#28]. The loan was to be repaid Based on functional grouping, the primary sources of RBLs with tobacco sales receivables, even though CATIC does not are state policy banks/development banks. The two policy appear to have a tobacco exporting business in Zimbabwe. banks of China, China Development Bank (CDB) and China CATIC is not interested in accessing Zimbabwe’s tobacco Export-Import Bank (Eximbank) are the top RBL lenders production, but seeks the certainty provided by tobacco by volume. These two policy banks were initially set up by revenues. [#28] Similarly, The Industrial and Commercial the State Council of China to carry out economic policies.13 Bank of China (ICBC) offered a $2.5bn RBL to Angola [#5] for Within our dataset, Chinese policy banks are the source of the construction of Kilamba Kiaxi New Town, to be repaid with 76% ($36bn) of the RBLs to Africa, with Eximbank and CDB proceeds from Angola’s oil sales to the Chinese SOE Sinopec. contributing $17bn and $18bn each. [#5] ICBC is not interested in oil access per se, but seeks the certainty of oil revenue proceeds to support the borrower’s Besides China, the Republic of Korea has also offered RBLs. repayment capacity and thereby protect ICBC’s investment. In 2011, the Export-Import Bank of Korea (KEXIM) announced Such lenders usually utilize resources that already have a that they signed an “Economic Development Cooperation record of generating a stable revenue stream, like tobacco in Fund” (EDCF) loan with the DRC government. “Under the Zimbabwe and oil in Angola. agreement, Korea would construct a network of water supply pipelines in the capital of Kinshasa with the condition that As noted above, creditors can also be grouped according Korea would acquire development rights to Musoshi copper to official bilateral versus private/commercial creditor mine.” [#10] classification. The Chinese policy bank and Korean policy bank loans are categorized as official bilateral loans according A second source of RBLs is entities interested in natural to information in the DRS, while the loans extended by private resource access, which include state-owned and private commodity trading companies fall under private creditors natural resource companies. The state-owned oil company (supplier and commercial credit). However, this distinction is China National Petroleum Corporation (CNPC) lent $1bn in not always clear and the classification, for example, of a state- RBL to South Sudan in 2015 [#25], and Russia’s state-owned owned policy bank may be viewed differently among lender oil company Rosneft has also provided oil-backed loans in countries. Many loans extended by Chinese SOEs, such as the past, though (based on publicly available information) we ICBC or CNPC, tend to be commercial in nature (and are found no evidence of it doing so in Africa.14 Besides these classified as such by CARI) and their treatment in the DRS state-owned oil companies, private commodities traders such (as reported by authorities) is mixed.15 This distinction matters as Trafigura and Glencore also extend RBLs, usually for as official bilateral and private loans may be treated differently shorter terms, in order to secure shipments of oil. in cases of restructuring. A third source of RBLs is entities that want to protect their investments and accept natural resource security as a form 13 CDB website: http://www.cdb.com.cn/English/gykh_512/khjj/ Eximbank website: http://english.eximbank.gov.cn/Profile/AboutTB/Introduction/. The Agricultural Development Bank of China is the third Chinese policy bank, but has no known record of providing RBLs. 14. Rosneft offered RBLs to Venezuela in 2006. 15. CARI methodology only considers loans from the People’s Bank of China (China’s central bank, distinct from the commercial Bank of China), the Ministry of Commerce, or the three policy banks (Eximbank, CDB, Agricultural Development Bank of China) as official bilateral loans. 10 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA d. RBLs by Type of Collateral Arrangement Observed Collateral arrangements can be grouped into four categories: the loans. Another example is Zimbabwe’s $110mn loan with resource sales receivables, resource sales pre-payment, China CATIC for the supply of rural electrification equipment resource development access and direct resource collateral. secured by tobacco sales revenues, as discussed above [#28]. Resource sales receivables are the most common collateral Resource sales pre-payments (SPP) are advances made in arrangements. Such arrangements require (agreed) specific respect of purchases of resources, most commonly used by amounts of natural resources (barrels per day, tons, etc.) to commodities traders who want to purchase natural resources. be sold to designated buyers. The amount that the designated For example, the multinational commodities trader Trafigura buyers are obligated to pay for the resources is paid for the advanced $550mn to the Republic of Congo in 2015 for 9-10 benefit of the lender, whether directly or into a deposit account cargos of oil. [#21] SPPs tend to be very short term, ranging owned by the borrower but over which the lender has rights to from a few months to a few years, we only consider those that deduct funds in the event of a loan repayment failure. go beyond one year. However, these tenors may have to be renegotiated and extended should the borrower fail to deliver For example, in 2004, Angola signed a $2bn loan with China the resources on time. Glencore advanced $500mn out of a Eximbank to be used for infrastructure development [#1]. $700mn purchase of oil from the Republic of Congo in late Monthly proceeds of the sales of 10,000 barrels per day (bpd) 2015, and experienced difficulties in receiving the expected from Chinese state-owned oil company Sinopec’s 4 oil blocks cargo shipments in a timely manner [#22]. In December in Angola were to be deposited in an offshore account to repay 2016, Glencore and Trafigura offered to extend the delivery 11 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA of oil to 2019, yet the Congolese government had made “The Ministry of Finance is required to ensure that such restructuring conditional on an additional combined the proceed generated by the Government of Ghana advancement of $380mn, which the two companies rejected from selling bauxite will be deposited into an Escrow [#22] & [#23]. The renegotiation between these two companies Account which shall be opened and maintained in a and the government has still not concluded.16 reputable offshore Bank acceptable to Sinohydro and the financiers.”20 Resource development access ties the expected returns from granting mineral development rights to an investor to the Occasionally, the borrower may be required to maintain a repayment of a loan. They are sometimes labelled “resource certain minimum amount of cash in the escrow account. for infrastructure” deals or “barter deals.” The KEXIM’s The Republic of Congo contracted a $1.6bn loan with China 2011 loan for the Musoshi Copper Mine development rights Eximbank in 2006 for infrastructure development [#18]. The mentioned in the previous section is one such example, IMF describes the arrangement as: with Taejoo Synthesis Steel given the rights to develop the mines [#10].17 The 2008 Sicomines deal follows similar logic, “as a guarantee for the loans, the Congolese authorities where the profit (100% for an initial period, only 85% later are required to keep a minimum deposit balance equivalent on) of a newly developed copper-cobalt mine was earmarked to about twenty percent of total outstanding loans in an for the repayment of two loans for infrastructure and the escrow account in China’s EXIM Bank from the proceeds mine itself [#9].18 These two loans also had an additional of their oil sales to China.”21 government guarantee. A variety of natural resources besides oil have been used as Finally, direct resource collateral involves collateralizing loan collateral, ranging from base metals such as bauxite, undeveloped resources that are still in the ground, usually copper and cobalt, to precious minerals such as diamonds and mineral or oil deposits. In 2006, Zimbabwe contracted a platinum, and agricultural outputs such as cocoa and tobacco. $200mn loan from China Eximbank for the purchase of agricultural equipment, the platinum deposits in Selous and Northfields reserves were given as collateral [#29]. There is no e. RBLs by Maturity information on whether platinum production was to be carried out by Zimbabwean or Chinese mining companies, only that China reportedly asked that the rights to 50% of the potential Our data reveals that the loans provided by commercial platinum reserves be put up as collateral for the loan, which commodity traders typically have short maturities, between is freed once the loan is fully repaid in cash.19 We have rarely 2 and 5 years. Note that our data excludes all advances by observed still-in-the-ground resource assets (rather than commodity traders with a maturity of less than one year. Some flows) being used as collateral. This is in line with Gelpern of these trader loans have seen their maturity extended, after et al.’s 2021 review of Chinese loan contracts more broadly, renegotiation, to up to 14 years. where they found limited evidence for using still-in-the-ground resource assets as collateral. The credits extended by Chinese SOEs tend to have long maturities ranging from 8 to 25 years. The longest maturity we As discussed above, lenders may be interested in resource observed in our review of 30 RBLs is 40 years for the Korea access as well as repayment assurance. Exim loan. The grace periods, where they exist, are typically three to five years, though eight-year grace periods were also Collateralization may be facilitated through the requirement observed. (See figure 1). to convert all earmarked payments into escrow-type accounts held offshore, such as in the case of Ghana’s Sinohydro loan [#15], where a loan summary states: 16. https://www.imf.org/~/media/Files/Publications/CR/2020/English/1COGEA2020001.ashx . Box 1. page 9. 17. Korea-Africa Center, 2017. 18. In our definition, we only consider the latter to be an RBL, although clearly the two deals are interlinked and part of the same resource-for-infrastructure deal. 19. https://www.theindependent.co.zw/2011/02/03/chinas-us3-billion-offer-for-platinum-raises-tension/ 20. Report of the Finance Committee of Parliament of Ghana on the Ghana-Sinohydro loan. 21. IMF CR 14/272 Rep of Congo (see page 9 in Box 1) https://www.imf.org/external/pubs/ft/scr/2014/cr14272.pdf 12 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA > > > F I G U R E 1 - Distribution of (Original) Maturities of RBLs 10 8 6 6 6 Frequency 5 4 3 2 1 1 1 0 0 5 10 15 20 25 30 35 40 Maturity (years) Lending countries China International Korea Due to fluctuations in commodity prices, the targeted quantity of sales necessary to generate cashflows to an amount sufficient On the other hand, the Master Project Support Agreement for the borrower to repay the RBL may be a variable. Such for Ghana’s Sinohydro loan [#15] document23 states that variables may be addressed in a range of ways depending any shortfall in the earmarked revenue stream should be on the commodity, market practice and the parties. When compensated for by the government: the repayment modality is defined as, or linked to, the sale of a certain volume of resources, the sale of the pre-specified “2.3 (b) GoG shall make payments for the Sinohydro quantity may nonetheless fall short relative to the necessary Arranged Project Financing out of receipts from the RBL repayment amounts over the prescribed timeframe, due transfer of refined Bauxite (Alumina or Aluminium) to its to changes in the price of that commodity. As a result, the strategic partner (Offtaker), and where receipts from the parties may have to size and calculate the repayment duration refined Bauxite are not sufficient for the repayment of the for an RBL based on prevailing market assumptions about the Sinohydro Arranged Project Financing, GoG shall use commodity’s anticipated price levels (e.g., oil prices), whilst other sources for the repayment of Sinohydro.” also building in loan features that account for the prospect of price volatility. For example, the CDB-Ghana loan [#13] summary document22 states: f. RBLs by Interest Rates and Fees “The period of the (Offtaker) Agreement will be 15 years and 6 months, as it was agreed under the (Master Facility We could only identify information on the interest rate for 17 Agreement) MFA that the (Offtaker) Agreement would out of 30 cases surveyed. 7 of these have fixed rates, while 10 extend beyond the repayment period of the loan to allow have floating interest rates. CDB to be fully paid if necessary”. 22. https://www.mofep.gov.gh/sites/default/files/reports/economic/CDB_Loan_Summary_050112.pdf 23. http://ir.parliament.gh/bitstream/handle/123456789/980/321122114725_0001.pdf 13 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA To the extent that interest rate information was available, annual repayments change, with less to be repaid in earlier Chinese policy banks offered fixed interest rates as low as years. Some set a range of volumes (e.g., 150,000 to 200,000 0.25 percent on RBLs, as in the case of a 2009 loan made by barrels per day or up to up to 40,000 metric tons of cocoa). China Eximbank to the Republic of Congo [#18]. Interest rates Some of these RBLs may require higher volumes within the were as high as 2 percent in Niger and Zimbabwe. Among the set range to be shipped when prices are low, but still effectively RBLs reviewed, all fixed interest rate RBLs made by Chinese cap the repayment value. lenders were made by China Eximbank. Floating interest rates generally use the London Inter-bank Offered Rate (LIBOR) as In some cases (especially RDA deals) the loan is to be repaid the reference rate, with rates ranging from Libor+ 1.0 percent from a designated commodity income stream (or a percentage [#9] to Libor+ 2.95 percent [#14]. The Korean loan to the of it) and repayment speed depends on how much revenue is DRC has the lowest weighted interest rate in our dataset with generated by that income stream. Examples of such earmarks 0.01% [#10]. include the Sicomines copper and cobalt mine in the DRC (85% of mine profits), the Agadem revenues in Niger (the We only have interest rate data for commodity trader loans in the government’s share of oil revenues) and the Badoit project in case of the two Glencore loans to Chad which had an interest Chad (a fraction of profit oil and royalty oil). rate initially set at Libor+6.6% and Libor+6.625% [#7] & [#8]. These loans were first renegotiated in 2015, when they were Chad’s Badoit deal with Glencore, renegotiated in June 2018 combined, their maturities were extended and their interest has repayment terms that depend specifically on the oil price. rate was adjusted to Libor+6.75%. The second renegotiation When the price of the locally produced oil blends is over a happened in 2018, when the base interest spread was set at certain threshold (about $55/barrel) in a given year, both the 2% until 2021, and 3% thereafter (+ additional 2% of deferred amount of principal and the interest rate due increases [#7] interest payments).24 & [#8]. We have little information on fees. Ghana’s 2018 RBL from São Tomé and Príncipe’s $30 million RBL from Nigeria for Sinohydro [#15] reveals that on top of an annual interest of administering the Joint Development Zone in 2009 also LIBOR + 2.8 percent, there is also a requirement to pay a presents an interesting case [#23]. Though São Tomé and flat management fee of 0.7 percent, a commitment fee of 0.5 Príncipe received and spent the loan funds, it is not servicing percent per annum and a one-time China Export and Credit the loan because the RBL terms specified that repayments Insurance Corporation (Sinosure) premium of 7 percent. The would be made from the revenue streams derived from loan in Guinea [#16] attracts a management fee of 0.5% while the project. However, no oil project, and consequently no that in Niger [#17] of 0.25%. The Glencore-Chad loan involved oil revenues, materialized. The loan (currently in dispute) fees for a total of USD37 million plus 0.75 % annual fee on provides a cautionary tale on lending against highly any outstanding amount, later lowered to USD 0.6 million uncertain revenues. [#7] & [#8]. The repayment structures presented above lead to lower We analyze how the rates of these loans compare to traditional repayment values when commodity prices are low, and higher loans in section 5/c. Given that data on fees is not systematic, when prices are up, and therefore have strong similarities with these are not analyzed. state-contingent debt instruments. Most of the loans discussed appear to have net present value (NPV) neutral features (i.e., lower repayment now attracts additional payment with interest g. Repayment Schedules later), although their merits would necessarily depend on key assumptions such as discount factor or collateral reference price. The renegotiated Chad-Glencore loan [#8] stands out as a case where the benefit of the interest reduction and maturity In seven cases we observe RBLs with repayments set to extension have been almost entirely cancelled by complex yearly, quarterly or a daily volume of commodities used to state-contingent clauses i.e., a “cash-sweep mechanism” service the loan. These repayment volumes are converted into (CSW). A CSW can be a useful instrument to make debt service their monetary value based on market prices and deducted proportional to oil sales. However, in the case of Chad, the from the total outstanding loan amounts. In some cases, the CWS parameters were designed in a way to significantly favor 24. See EITI 2018 report for further details. http://itie-tchad.org/wp-content/uploads/2020/12/Rapport-ITIE-Tchad-2018-sign%C3%A9.pdf 14 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA Glencore in case of upside scenarios, without significantly with reduced creditworthiness, capacity challenges or poor reducing Chad’s debt burden in case of downside scenarios. project implementation performance history may struggle to In fact, at the time of the agreement, the oil price was close to otherwise finance. 60 USD per barrel. The downside trigger was set at a very low level (42 USD) and this scenario, which materialized in 2020, Though infrastructure projects constitute the most common involves very limited debt service reduction (1% principal sectors financed, they do not constitute the bulk of the deferral). Conversely, the upside triggers were below the spot RBLs in terms of value. Instead, the bulk of the RBL in value oil price (between 53 and 57 USD), thus generating significant went to resource sector related projects, including refinery principal payment acceleration (more than 5%) and a rise in construction or gas infrastructure development. For example, interest payment (+2%) in 2018-2019. CDB contracted a $15bn RBL with the Angolan government in 2015, of which $10bn was then on-lent to the state-owned oil company Sonangol as long-term financing for Sonangol’s h. Allocation of Proceeds planned 200,000 bpd Sonaref refinery in Lobito [#6]. Similarly, in Ghana, a $1.5bn oil-backed loan from China Eximbank in 2011 went towards the development of gas processing facilities and an accompanying information and communications Infrastructure projects are the most common sectors financed technology surveillance platform [#12]. through RBL. Of the 30 resource-backed financings in Sub- Saharan Africa that were reviewed, 18 are for infrastructure There are a number of RBLs where natural resource projects including roads, energy, water supply, etc. Some of the development projects are complemented by infrastructure very large credit lines are intended for multiple infrastructure projects. Ghana’s $1.5bn gas processing development projects. Angola’s first RBL in 2004 was a China Eximbank RBL is complemented by another $1.5bn loan, designated RBL that totalled $2bn and was disbursed for at least 50 for infrastructure development such as railway and port different projects, averaging about $40 million each. The construction [#11]. Similarly, in DRC, the Sicomines deal projects ranged from construction projects for schools and had two loans, one for unrelated infrastructure and one for hospitals, power lines, water supply projects, and procurement developing the mine itself [#9].26 This kind of two-pronged of equipment and vehicles. These credit lines streamline the financing simultaneously ensures the production of the disbursement process such that both sides do not need to collateral, while supporting the development of other essential renegotiate every time a new project needs financing. This infrastructure projects that may be hard to finance through also means that credit line RBLs are often earmarked ahead project-specific funding structures. of time. Beyond sector-specific RBLs, RBLs can also take the form Some RBLs are designated for specific projects, often large of pre-payments from commodity traders who are interested scale. Ghana contracted close to $600mn in a cocoa-backed in securing shipments of natural resources. Six out of the 30 RBL from China Eximbank for the construction of the Bui RBLs were advanced payments of shipments of oil by private Hydropower Project in 2007 [#11] [#12]. A similar big-ticket commodity traders (plus one by CNPC) in Chad, Rep Congo item was Angola’s Kilamba Kiaxi New Town, a large-scale and South Sudan. Most private trader loans do not appear housing project in Angola, financed with a $2.5bn oil-backed to be earmarked for any project and the proceeds appear to loan from ICBC. What counts as “big-ticket items” varies by be used at the discretion of the borrowing government; where country [#5]. South Sudan’s Nadapal-Torit-Juba Road cost any detail is provided, they are described as being used to around $169mn and was financed through oil-backed China cover budget shortfalls. Occasionally, commodities traders Eximbank loans [#27]. This amount may sound relatively small may lend to sovereign borrowers in a bid to purchase mineral compared to those in Ghana and Angola, but is nonetheless assets. For example, a syndicate of banks led by Glencore the largest loan that South Sudan had contracted with China provided a total of $1.5bn to Chad State National Oil Company as of 2020.25 There are several possible drivers as to why in 2014 to be used to help Chad purchase back Chevron’s oil RBLs tend to flow into infrastructure projects. These are assets in Chad [#8]. often low internal return rate projects that would not succeed in raising capital independently in the commercial market. Such projects tend to have large upfront costs that countries 25. CARI dataset. The second largest is the New Juba Airport loan, $158mn from Eximbank in 2013. 26. We do not consider the latter loan an RBL under our definition, given that it is repaid via the project it finances. 15 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 5. >>> Policy Considerations In this section we discuss the insights derived from our analysis. Specific features of the RBLs reviewed suggest there may be several opportunities to improve the governance of the RBLs. a. RBL Importance as a Source of Finance The RBLs we identified represent a small share of total lending to Sub-Saharan Africa. The 30 loans that we identified as RBLs account for $46.5 bn (with a large share of some loans only partially committed or disbursed). The total new external debt over the 2004-2018 period is recorded by the World Bank’s International Debt Statistics unit at $600 bn, of which $450 bn was disbursed. While our recordings on RBL commitments and disbursements don’t always match the DRS, we can provide a ballpark estimate of their economic significance. RBLs represent about 8% of total borrowing in Sub-Saharan Africa. RBLs are heavily concentrated among a few countries. Figure 2 shows new external debt disbursements relative to RBL agreement dates, and while disbursements may experience a time lag following the signing of RBLs deals and some RBLs went unrecorded (see next section), the RBLs represent a major source of new financing in all of the countries listed below. DRC and Sudan stand out for having much higher new RBL deals signed than what was disbursed. This is consistent with our own data, which shows that the share of loan which was disbursed (by end- 2018) in DRC was about 25% (across two loans) and 50% in the case of Sudan - more than 10 years after the loan was agreed. 16 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA > > > F I G U R E 2 - New Disbursements of External Public and Publicly Guaranteed Debt vs. RBL Committed Amounts ANGOLA CHAD 15,000M$ 1,500M$ 10,000M$ 1,000M$ 5,000M$ 500M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 C O N G O , D E M O C R AT I C R E P U B L I C CONGO, REPUBLIC 3,000M$ 2,500M$ 2,000M$ 2,000M$ 1,500M$ 1,000M$ 1,000M$ 500M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 SUDAN ZIMBABWE 3,000M$ 600M$ 2,000M$ 400M$ 1,000M$ 200M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 Public debt disbursement RBL agreement 17 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA NIGER SAO TOME & PRINCIPE 1,000M$ 40M$ 750M$ 30M$ 500M$ 20M$ 250M$ 10M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 GHANA GUINEA 3,000M$ 750M$ 2,000M$ 500M$ 1,000M$ 250M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 Public debt disbursement RBL agreement Source: WB International Debt Statistics, NRGI.I 18 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA > > > F I G U R E 3 - External Public and Publicly Guaranteed Debt Stock vs RBL Committed Amounts ANGOLA CHAD 30,000M$ 30,000M$ 20,000M$ 20,000M$ 10,000M$ 10,000M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2025 2000 2005 2010 2015 2020 2025 C O N G O , D E M O C R AT I C R E P U B L I C CONGO, REPUBLIC 6,000M$ 9,000M$ 4,000M$ 6,000M$ 2,000M$ 10,000M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2025 2000 2005 2010 2015 2020 2025 GHANA GUINEA 20,000M$ 15,000M$ 3,000M$ 10,000M$ 2,000M$ 5,000M$ 1,000M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2025 2000 2005 2010 2015 2020 2025 Public debt stock RBL agreement 19 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA NIGER SAO TOME & PRINCIPE 3,000M$ 300M$ 2,000M$ 200M$ 1,000M$ 100M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2025 2000 2005 2010 2015 2020 2025 SUDAN ZIMBABWE 4,000M$ 15,000M$ 3,000M$ 10,000M$ 2,000M$ 5,000M$ 1,000M$ 0M$ 0M$ 2000 2005 2010 2015 2020 2025 2000 2005 2010 2015 2020 2025 Public debt stock RBL agreement Source: WB International Debt Statistics, NRGI. Note: Year of maturity indicated by arrows Figure 3 shows RBLs as a share of total debt stock rather is critical to monitor debt sustainability. This is evidenced than just new debt issued. RBLs represent a substantial share by cases in Angola, Chad, Republic of Congo, where the of these countries’ borrowings by this metric, with the total renegotiation of RBLs was presented as a key factor in amount of RBLs representing over 10% minimum and over restoring debt sustainability by the IMF.27 30% of the median country’s total external public debt stock in the periods after the RBLs were signed. The fact that multiple countries are repeat borrowers of resource-backed loans also raises the question on whether This suggests that countries either chose not to take on RBLs these loans may be displacing or discouraging other forms or when they do, they decide to take them on in large volumes. of more conventional borrowing, and if they carry the risk of Given the magnitudes observed, understanding their terms increasing resource dependency. 27. IMF Country Report No.20/26 on Rep Congo, IMF Country Report No.18/108 on Chad. IMF Country Report No.20/281 on Angola. 20 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA b. Transparency Data indicates that countries which borrow using resource- debt statistics and debt management documents posted on backed loans have weaker debt disclosure practices.28 We national authorities’ websites among the 76 International looked at how 10 RBL borrower countries performed on the Development Association (IDA) eligible countries.29 The World Bank’s Debt Transparency Heatmap (1st April 2020 results are presented in Figure 4, where the box represents evaluation). The heatmap presents an assessment based the range of scores between the top and bottom quartiles on the availability, completeness and timeliness of public across IDA countries. > > > F I G U R E 4 - Debt Transparency in RBL Borrowing Countries30 25 Ghana Congo, Democratic Republic 20 Guinea Debt transparency heat map total score 15 Median score all IDA countries Sao Tome & Principe Congo, Republic 10 5 Niger Sudan 0 Zimbabwe South Sudan Chad Countries RBL borrowed in SSA Other IDA countries Source: WB Debt Transparency Heat Map data Note: Each heat map criteria is assigned between 0-3 points based on color code. Each country’s score is the aggregate number of points it received across the criteria on the heat map. 28. NRGI (2020) highlights how RBL borrowers also have a weaker governance of their natural resource sector. 29. Angola is excluded from the Heatmap, as Angola is not an IDA country. 30. The scores shown are calculated taking the sum of the performance (on a 0-3 scale) across all 9 criteria measured on the World Bank’s Debt Reporting Heat Map. (Rivetti, 2021). 21 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA As Figure 4 shows, five RBL borrowing countries were in the could be clearly matched and 4 could be partially matched bottom range in terms of transparency (where timely public with the reporting provided in the DRS.35 There are several debt reports are available); two were just below the median possible reasons for why these debts were missing from (the thick line); and three countries were among the stronger the system. Reporting coverages and definitions vary by performers. Among the criteria used for scoring the heatmap, country, and as discussed above, loans borrowed by SOEs two are especially relevant for the type of debt we analyze: (or joint ventures that involve state participation) may not be sectoral coverage of debt reporting and debt instrument centralized and reported to DRS. In addition, some types of coverage. Sectoral coverage of debt reporting is important as financing may have not been treated by the reporting country SOEs, and not central government, are often the borrowing as a debt instrument, but rather as an advance payment to a party. Debt instrument coverage is also important as it reveals supplier. The absence of information on collateral features in whether the borrower has received a government guarantee the DRS may have also led to mis-reconciliation, if there were in support of its loan repayment obligations. Across the RBL discrepancies in reporting across sources. borrower countries scored on the heatmap, only the DRC has full coverage across both sectors and instruments. Among the loans not found in the DRS, many could be Guinea does not report SOE debt. Ghana reports on directly verified in government reporting. For example, DRC’s government-guaranteed SOE debt, but Ghana’s reporting public debt report provides details of Sicomines loan under on non-guaranteed SOE debt is missing. None of the other the heading of external loans with government guarantees. RBL countries report on either of these aspects according to South Sudan and Republic of Congo report intermittently on the heatmap. the debt that they have outstanding with commodities traders. Angola’s 2018 sovereign bond prospectus provides details of The transparency of SOEs is critical given that they were Sonangol’s CDB loans. the borrowers in nine of the 30 loans. These borrowers are Ghana Integrated Aluminum Development Corporation The Sicomines loan stands out as the sole RBL transaction for (GIADEC), Gécamines (DRC), SNPC (Rep. Congo), SHT which the loan contract is publicly available.36 (Chad), Sonangol (Angola’s national oil company), and Zimbabwe Electricity Supply Authority (ZESA)). Of these The lack of transparency on RBLs has a number of SOEs, Sonangol publishes audited financial statements implications. First, these deals are often highly complex and yearly (latest available 2018). SHT recently released audited are therefore susceptible to receiving a poor evaluation of financial statements for the first time.31 SNPC recently the risks that they entail. Second, given limited competition released its audited financial statements for 2012-2018 for the in providing such loans (few lenders offer them), there is also first time, although consolidated financial statements for the a risk of poor negotiation outcomes. In both cases, the risk is SNPC Group have yet to be disclosed.32 ZESA does not have higher when the outcome is subject to limited scrutiny. Finally, a functioning website. Gécamines produces audited financial RBLs may utilize transaction structures that are not reflected accounts which it files to the public registry and shares with in standard debt statistics or may confer undisclosed seniority the EITI secretariat but they are not made public.33 GIADEC or payment advantages to their lenders, which ultimately leads does not publish audited financial statements, which may be to a mispricing of the country’s risk by other creditors. because it was only recently established.34 In many additional instances SOEs played an important intermediary role (for example through an additional offtake contract offering to c. Cost of Financing supply resources which are then deployed as collateral for repayment of the sovereign’s loan). We compared the interest rates on RBLs to the rates on regular Only 15 of the 30 loans analyzed could be identified on the loans. We identified 19 RBLs for which we had the necessary World Bank’s Debtor Reporting System. Of these 15, 11 loans data on interest rate and maturity needed for evaluation. We 31. https://eiti.org/chad 32. https://eiti.org/news/republic-of-congo-from-disclosure-to-diagnostics 33. IMF Country Report No.19/286, statement here: https://www.africanews.com/2018/11/28/gecamines-publishes-a-comprehensive-report-in-response-to-outrageous-and- systematic-allegations-of-certain-ngos/ and this from 2016 annex 4 more authoritative, though unclear on audits. https://drive.google.com/file/d/1JVPyPgiGzieGjRe- N8ux-2I3s6Ix92zp/view 34. https://www.mofep.gov.gh/sites/default/files/reports/economic/2018_State_Ownership_Report_v3.pdf 35. We label “partial match” when a number of distinct features of RBLs in the public domain match the DRS, but not all. 36 The contract documents are available on the Ministry of Mines website, mines-rdc.cd, and on congomines.org 22 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA compared each of these loans to an individual control group analysis as low or medium risk vs high risk and in debt distress with similar features (other than collateral) constructed based vs not scored), same currency, same interest type (floating vs on the WB Debt Reporting System, which contains over fixed), similar maturity, similar prevailing global lending rates 25,000 loans. We filtered the data to include only those loans at time of signing (Libor within +/- 100 bps range). Each control extended by private sector and bilateral creditors to the public group contains at least 5 loans, some as many 500 loans. We sector. The loan features used for matching were for borrower compared loan terms based on their weighted interest rate, a countries that have similar risk ratings (splitting countries metric used by WB to denote average interest rate for complex between those classified by the WB/IMF debt sustainability loans. Figure 5 below denotes key results. > > > F I G U R E 5 - Interest Rate of RBLs vs Other Similar Loans 420 76 114 657 520 5 11 8 194 35 84 115 19 325 325 29 368 51 39 9% Interest rate 6% 3% 0% #1 #2 #3 #5 #6 #7 #8 #9 #10 #11 #12 #13 #15 #16 #17 #18 #20 #29 #30 RBL lender type Bilateral Private Size of control group RBL interest rate Distribution of interest rate of the control group Note: The contorl groups are selected according to similarity with each RBL. Similarity is based on: debtor type, currency, interest type, DSA credit risk measure, similar debt maturity (difference within +/-60%) and LIBOR rate at the time of agreement (+/-1%) 23 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA Our results suggest RBL borrowing rates are, on average We also evaluated the rates using regression analysis. We ran somewhat higher than other sources of financing with a linear regression on the same sample of loans and provide comparable terms (other than their collateral features). Six an estimated interest rate for each of the 19 RBLs based on RBLs have interest rates within the 50 percent range of their a list of explanatory variables identical to the above matching control groups (denoted by the box), but most loans have approach (risk ratings in DSA, currency, interest type floating rates within the 90 percent range (denoted by the line). There vs fixed, years of maturity, prevailing global lending rates are more RBLs with rates higher than their control group at time of signing. All explanatory variables were significant than with lower rates. For the two commercial RBLs issued (results not shown here). Figure 6 provides the estimated rates by commodity traders for which we have data, interest rates based on regression analysis of key characteristics (excluding are higher (prior to renegotiation) than their control groups, collateral features) vs the actual observed interest rates. although the control groups are also very small. > > > F I G U R E 6 - Interest Rate of RBLs vs Estimated Rate Based on Other Loan Characteristics 10% 8% 6% 4% Interest rate 2% 0% -2% -4% #1 #2 #3 #5 #6 #7 #8 #9 #10 #11 #12 #13 #15 #16 #17 #18 #20 #29 #30 RBL lender type Private Actual interest rate Bilateral loans Bilateral loans Note: The following variables were used to estimate the interest rate using a linear regression model: LIBOR rate at the time of agreement, interest type, debt maturity in years, currency, DSA credit risk measure. 24 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA Our results are quantitatively similar to the approach used for in government revenue. It may also not fix the underlying constructing a control group.37 The RBL rates are somewhat causes of low growth or low tax revenue, which may require higher overall than their other features would predict from the different types of spending or structural reform. regression. For the two commercial RBLs issued by commodity traders for which we have data, interest rates are significantly As discussed in section 3/F most RBLs were spent on higher (prior to renegotiation) than the regression estimates. infrastructure (esp. petroleum sector, road, energy). In theory, these can yield returns which outweigh their costs, but there Overall, we found large variations in how favorable RBL are a number of loans which went either for budget support or borrowing rates were for the small sample for which we had for unknown goals. Pre-payments are especially risky in this data. Some RBLs extended by bilateral lenders (South Korea, respect, as they are often not earmarked for specific projects, China) have outstandingly low rates, but many come with and are used at the borrowing governments’ or SOE’s rates well above the median, including those extended by discretion. The lack of transparency on the use of proceeds commodity traders. might increase the risk of payments being used inefficiently. There are some important caveats to the method employed. There are also mixed cases. In 2006, Zimbabwe contracted a They are based on a limited number of RBLs for which we $200mn loan from Eximbank to purchase agricultural inputs have interest rate information and we can only control for a such as tractors, fertilizer, and pesticides. Clearly, the latter limited number of parameters in order to get to reasonable two items are short-term consumables, with no guarantee that sized comparators. a one-time increase in agricultural production that year would somehow reduce the cost for future demand of fertilizers The analysis above suggests that RBLs do not provide a and pesticides [#29]. While this may appear intuitive at the uniquely cheap mode of financing. Nevertheless, RBLs from outset, the pathway from a one-time injection of fertilizers and bilateral lenders have interest rates within a broad range of pesticides, to an increase in production, and then to higher other loans so their use should not be ruled out on this basis. government revenue is rife with other uncertainties. It may also be possible that some of the loans would have not materialized at all without some collateral. The contracting process is another consideration for borrowing countries. While there are no hard rules dictating This suggests that borrowers should seriously consider that RBLs have to involve tied contracting, e.g., Chinese whether giving resource-based collateral in respect of their loans often involve construction or procurement to be carried borrowings provides tangible additional benefits. This could out by Chinese companies (at least in respect of a portion be done by seeking alternatives, and encouraging more of the loan), as a result of their ability to best leverage the competition amongst lenders. More transparency in lending preferential financial support that the state policy banks bring. terms is also needed to increase scrutiny and to ensure that This leads to an open question on whether it may be advisable only competitively priced loans are taken up. to trade potentially cheaper credit for a more competitive bidding process. d. Spending Considerations There is no guarantee that governments will use all loans for their intended purposes. In this study, we did not attempt to verify RBL loan use comprehensively. However, our review of press reports and articles on RBLs confirms that the RBLs, like any other debt represent a financial burden, so designated projects are generally the ones that receive the covering debt repayments must be sourced from future funding. One counterexample exists in an RBL that South forms of revenue. Typically, borrowing can be used to make Sudan took from a commodity trader to finance its Green productive investments such as infrastructure that might Horizon farming project and provide general budget support. generate revenues for government directly, or generate future In an OCCRP investigation, it has been alleged that some revenue via increased economic growth and tax payments. of these RBL funds were instead used for military spending Generally speaking, using RBL receipts to plug budget (although this has been disputed).38 shortfalls, rather than fund capital expenditure may be risky, as a one-time cash injection might not prevent future shortfalls 37. There is 0.84 correlation in the size of the gap between actual RBL interest rate minus the average control group or estimated rates. 38. Sam Mednick, Sprouting Weapons of War, OCCRP, 17 July 2019, www.occrp.org/en/investigations/sprouting-weapons-of-war. 25 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA To summarize, the majority of RBLs are earmarked for Given the possible variety in scope of negative pledge financing particular infrastructure projects. Since the undertakings, the question of whether a particular transaction infrastructure gap in developing countries tends to be large, violates a negative pledge undertaking will depend both on the it may economically justify the borrowing if the social returns language used in the negative pledge clause itself, as well as from the infrastructure projects are higher than the interest the structure of the transaction and the language of the legal charges and risks associated with the loan. However, there documentation setting out the deal. It is clear that a negative are also risks associated with project selection and execution, pledge clause will not catch all types of resource-backed loan especially when there is a lack of transparency, accountability structures, and indeed it may not be desirable or practical to and competition. It is therefore critical for lenders and loan do so. Irrespective of whether legal security or quasi-security recipients to create safeguards to ensure that RBLs are is used, sovereign and SOE borrowers in resource-backed invested productively. loans need to consider the extent to which a collateralized transaction structure effectively ties up the borrower’s assets, for example with resource sale proceeds being directed e. Negative Pledge Undertakings39 and held in specific accounts where the borrower’s ability to withdraw funds is subject to contractual limitations tied to the loan. These limitations may significantly counter-balance the charm and allure of any low interest rate offered in connection In addition to considering the merits and terms of a RBL with such loan. transaction, there are legal factors that a sovereign or SOE ought to consider prior to finalizing arrangements. The breadth of a negative pledge undertaking turns on more than just the definitions of concepts such as “lien” or “security”. One such consideration is whether the structure or terms of a Exceptions are often contemplated within the negative pledge proposed loan would cause non-compliance with contractual clause itself. Examples include exceptions given to: (i) the undertakings that the sovereign or SOE borrower has already giving of liens or security by the borrower in connection given in favour of other lenders. A relevant example of such an with debt that has a maturity of 12 months or less; and (ii) undertaking is the negative pledge clause. transactions that entail a lien or security created on property at the time of purchase of such property, where the lien is While the precise application of a negative pledge clause will created solely to secure payment of the purchase price or vary depending on its text, the key aim of a negative pledge to secure the repayment of debt incurred for the purpose of provision is to preclude or limit the ability of the borrower to purchasing the property. In addition, and depending on the grant collateral through liens and similar forms of security context of the debt in question, a lender may limit the scope in favor of other lenders, unless the borrower equally and of the negative pledge undertaking to external debt incurred rateably secures the lender in question, or provides equivalent by the borrower. External debt is usually defined either as security or otherwise obtains its consent. Negative pledge debt that is denominated in a foreign currency relative to the clauses are important to a lender: if, for example a borrower borrower; or as debt incurred by the borrower to a lender that provides collateral against an asset by giving a lien or other resides overseas. legal security40 in favour of a specific lender, then in the event of non-payment or default by the borrower, the lender will Negative pledge undertakings are used in loans made by have a legally enforceable right and claim against the secured a wide range of lenders, including multilateral development asset, and be able to liquidate the asset in order to repay itself banks, private or commercial lenders, and state-owned the outstanding loan amounts. When specific assets of the policy banks. They are also usually included in the terms and borrower are secured in favour of specific lenders, it leaves conditions of bonds issued by sovereigns and commercial fewer borrower assets out of which unsecured lenders can be entities alike. Particularly in commercial contexts, negative repaid in the event of a default.41 39. Susan Maslen (WB-LEG) has contributed to this chapter. 40. The concepts of “lien” or “security” are usually defined terms in the loan agreement and so the precise definition and contents will vary by lender, market etc. However, in general, the concepts used in negative pledge clauses typically include liens, charges, security interests, pledges, mortgages and other encumbrances. 41. Some negative pledge clauses have an expanded scope in that they apply not only to transactions involving the creation of legal security or a lien which is legally enforceable, but also to transactions with collateral arrangements that offer lenders a kind of commercial security and that do not amount to true, legally enforceable security in an insolvency or similar scenario. Such transactions may be considered to have an equivalent (economic) effect to legal security, albeit not the same legal effect (or remedies). Under such arrangements, contractual rights and transaction structuring techniques may be used to give the relevant lenders advantages vis-à-vis other lenders, despite the absence of a legally enforceable claim against a specific asset. These contractual features are often referred to as “quasi-collateral” or “quasi- security”. In negative pledge clauses that seek to also target quasi-security, the definition of “lien” or “security” may, for example, be extended to include (language like) “ …and any arrangement having an equivalent effect”. 26 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA pledge texts may be accompanied by a list of permitted liens an ally to sovereign and SOE borrowers in resource-rich Sub- or similar limitations to the application of the negative pledge Saharan African countries by giving the borrower something clause itself. to point to in seeking to minimize any legal security or quasi- security proposed by lenders. It is not the responsibility of lenders to monitor or check compliance with negative pledges Permitted Lien Provisions and permitted lien provisions. However, if information about a Under “permitted lien” provisions the lender will upfront, sovereign’s or SOE’s borrowing activities is available, a lender through the language of the clause, permit the borrower to may decide to conduct its own review. The difficulties faced by enter into collateralized transactions of a certain nature and lenders in assessing the borrowing activities of sovereigns and these transaction types are usually listed and described with SOEs (whether for the purpose of reviewing compliance with some specificity and parameters. For example, in a resource- negative pledge clauses or for the purpose of assessing debt backed loan context, a lender may agree under permitted lien sustainability and creditworthiness) are often complicated by language to permit a borrower engaged in the oil sector to the lack of public disclosure of such information. undertake limited recourse project finance transactions, where security is given to the relevant lenders over oil production equipment or facilities and where the aim of the relevant borrowing is to increase or improve the overall capacity or f. RBL Renegotiations efficiency of the borrower in oil production and sales. Similarly, a lender may permit a borrower, upfront, to provide security to other lenders in respect of trade finance facilities. A number of key differences emerge when comparing how RBLs can be renegotiated compared to traditional In permitting certain types of liens or secured transactions lending instruments. upfront, a lender may stipulate maximum monetary values of a permitted loan; they may limit permitted liens to loans RBLs are likely to involve fewer creditors than other forms of with a specific purpose; or specify other limitations beyond sovereign debt, especially sovereign bonds. This may make it which the borrower is required to obtain the lender’s consent easier to reach an agreement. The renegotiation of bonds is or otherwise to equally and rateably secure such lender. The often complicated by differing interests among bondholders lender’s motivation for permitting, upfront, the borrower to and holdout creditors that are less amenable to bargaining.42 give such liens or security in favor of other lenders over the life of the loan will vary, but may include any of the following: The parties to an RBL are also often more heavily (i) recognizing that such liens are usual in the borrower’s interdependent. The interdependence may stem from business context or sector; (ii) recognizing that such liens are having agreed on multiple loans, tied contracting practices likely to be given in connection with specific investment types with regards to spending loan proceeds (referenced above) that enhance the borrower’s overall business and value; and or additional contracts related to commodity production or (iii) a desire on the part of the parties to limit the amount of marketing. Therefore, lenders may have additional incentives time and transaction costs expended over the life of a loan to reach an agreement to continue their business operations. in obtaining the lender’s consent. In some instances, the number and scope of permitted liens or transaction types is While the collateralized nature of RBL transactions may seem so extensive that one might wonder if the extensive number like a strong guarantee for repayment, in practice it is often of carve-outs impairs the purpose of the permitted lien (and difficult to enforce such collateral. This is especially the case negative pledge) language in the first place. when the collateral involves assets located in the borrowing country. Assets such as the bauxite reserves supposedly Compliance with Negative Pledge and serving as collateral for the Selous and Northfields platinum reserves in Zimbabwe are highly sensitive and their loss Permitted Lien Requirements is the would face great resistance.43 If lenders tried to seize them Obligation of the Borrower rather than negotiate, they would be heavily dependent Compliance with a negative pledge clause and permitted lien on local cooperation by state and citizens for using them provisions is the obligation of the borrower. The limitations of productively, which they would likely find difficult. This may be these clauses, whilst often viewed as a burden, can function as 42. Innovations in collective action clauses have recently been implemented to ease this problem. 43. “Storm Brews Over China Deal,” Zimbabwe Independent, 9 July 2009, www.theindependent.co.zw/2009/07/09/storm-brews-over-china-deal/. 27 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA the key reason that no claims on such subsoil collateral have restructuring with Glencore, which lengthened the maturities emerged publicly from the various RBLs facing repayment on the debt, lowered interest rates and made the debt more difficulties.44 The same cannot be said about foreign assets counter-cyclical by including contingencies which adjust used as collateral.45 repayment depending on oil price. The use of proceeds as collateral also has strong limitations. The pandemic has accentuated the already sharp debt In Chad, in order to service the RBL following the oil price sustainability challenges in a number of developing countries. crash, Glencore was allowed to keep a large share of In response, the G-20 has agreed to an initiative which enables government’s oil revenues (100 percent of equity oil and 70 the temporary suspension of debt servicing (DSSI) for the percent of royalty oil). But the government found ways to world’s poorest countries. However, most RBLs were outside reduce its repayment. SHT started diverting some of the oil its scope, mostly because only official lenders participated to the domestic refinery rather than paying Glencore, showing and not commercial lenders or commodity traders.48 Further that designated RBL revenue streams can be altered by consideration should be given to how international financial governments looking to force a renegotiation.46 The country’s institutions may be able to assist RBL borrowers to help president also threatened to give Glencore’s monopoly trading restructure their debt effectively (in a way that also ensures rights to ExxonMobil. Glencore’s upstream interests may have inter-lender equity) in the eventuality of debt burdens also made the company more amenable to renegotiate.47 In becoming unsustainable. 2018, as part of its IMF program, Chad completed its debt 44. Acker, et al.’s 2020 review of disputes involving Chinese loans also hasn’t found evidence for “asset seizures”. 45. For example, the dispute over the ownership of CITGO pledged by PDVSA as security. CITGO’s refinery assets are above ground in the US, rather than underground as in Venezuela. “New York court decision tightens bondholders’ grip on Citgo Petroleum, Venezuela’s most valuable foreign oil asset” (https://www.wsj.com/articles/ citgo-backed-bondholders-win-court-ruling-against-venezuela-11602863428) 46. Madjiasra Nako and Julia Payne, “Exclusive: Chad wants to cut off Glencore’s oil supplies in debt row,” Reuters, 30 October 2017, www.reuters.com/article/us-glencore- chad-oil-exclusive/exclusive-chad-wants-to-cut-off-glencores-oil-supplies-in-debt-row-idUSKBN1CZ1TK 47. Ian Lewis, “Oil price recovery eases Chad’s gloom,” Petroleum Economist, 25 June 2019, www.petroleum-economist.com/articles/politics-economics/africa/2019/oil- price-recovery-eases-chads-gloom 48. Chad did request suspension from Glencore in September 2020. 28 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 6. >>> Conclusions In this paper we analyze the main features and the impact of resource-backed loans on borrowing countries. While our review was restricted to publicly disclosed information about borrowings by central governments or SOEs in Sub-Saharan Africa during the 2004-2018 period, our conclusions are relevant to developing countries in general and may also be applicable to other forms of collateralized borrowing. We found that while RBLs have limited use in a handful of countries, they represent a sizable proportion of debt in many of them. The countries that rely most on this form of borrowing are generally less transparent in their debt reporting, especially with regards to SOE debt, and collateralization is not disclosed in public debt records. Within our analysis sample, we found wide variations in how favourable RBL borrowing rates were. While a few RBLs extended by bilateral lenders (South Korea, China) have outstandingly low rates, most come with rates well above the median. This is markedly the case for the few RBLs extended by commercial commodity traders. Some RBLs have resource price contingent repayment features, which allow for faster repayment in commodity booms and less in bust times, which could be attractive for resource-dependent borrowers. Most RBLs are earmarked for infrastructure investment, which in theory can yield returns that offset borrowing costs, however, non-earmarked loans (more prevalent among pre-payment deals) may be hard to justify economically, especially given their often-higher rates. We highlighted the nature of negative pledge undertakings and permitted lien clauses, and their relevance to borrower decisions related to the giving of collateral in the context of individual RBLs, and finance transactions more broadly. Finally, we show how collateralization can threaten inter-lender equity and undermine the process of orderly debt dispute resolutions. RBLs have been so far renegotiated on an ad-hoc basis, which is likely driven by the strong interdependence between borrower and lender and by the practical limitations to enforcing any collateral granted through these loans. 29 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA When transparently conducted, borrowing operations that as they face narrow fiscal space and reduced control over leverage natural resources as credit enhancement may scarce resources that are tied up in highly structured loan represent an opportunity. However, their net benefit depends arrangements. The complexity and critical nature of RBLs on a country’s debt sustainability, the use of proceeds and presented in this paper underscores the importance of the the improvement in financial terms. On the downside, these international community’s continued efforts in developing contractual arrangements inhibit the ability of a sovereign or technical capacity in Sub-Saharan African countries, it SOE borrower to freely direct their own cashflows, potentially also highlights the key role of financial and legal technical impairing the borrower’s ability to respond to economic assistance in promoting borrowers’ interests in negotiations shocks and source price volatility. The COVID-19 pandemic with lenders. has exposed some Sub-Saharan countries to these risks, 30 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA >>> References Acker, Kevin, Deborah Brautigam, and Yufan Huang. “Debt Relief with Chinese Characteristics.” SAIS-CARI Policy Brief. 2020. Alves, Ana Cristina. “Chinese economic statecraft: A comparative study of China’s oil-backed loans in Angola and Brazil.” Journal of Current Chinese Affairs 42, no. 1 (2013): 99-130. Brautigam, Deborah. “The dragon’s gift: the real story of China in Africa.” OUP Oxford. 2011. Brautigam, Deborah and Kevin P. Gallagher. “Bartering Globalization: China’s Commodity- Backed Finance in Africa and Latin America.” Global Policy. 2014. Collier, Paul, and James Cust. “Investing in Africa’s infrastructure: Financing and policy options.” Annu. Rev. Resour. Econ. 7, no. 1 (2015): 473-493. China Africa Research Initiative and Boston University Global Development Policy Center. “Chinese Loans to Africa Database”, 2021. Dreher, A., Fuchs, A., Parks, B. C., Strange, A., & Tierney, M.J. “Aid, China, and Growth: Evidence from a New Global Development Finance Dataset”. American Economic Journal: Economic Policy 13(2), (2021): 135-74. Gardner, Alyssa, Joyce Lin, Scott Morris, Bradley Parks. “Bargaining with Beijing: A Tale of Two Borrowers.” AidData – CGD. 2020. Gelpern, Anna, Sebastian Horn, Bradley C. Parks. Scott Morris, and Christoph Trebesch. “How China Lends: A Rare Look at 100 Debt Contracts with Foreign Governments.” AidData. 2021. Gillies, Alexandra, and Zira John Quaghe. “Ensuring a Fair Deal for Nigerians with Oil-Backed Loans.” Natural Resource Governance Institute. 2018. Halland, Håvard, John Beardsworth, Bryan Land, and James Schmidt. “Resource Financed Infrastructure: A Discussion on a New Form of Infrastructure Financing.” World Bank. 2014. Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. “China’s overseas lending.” Journal of International Economics. 2021. IMF. “Assessing Public Sector Borrowing Collateralized on Future Flow Receivables”. 2003. 31 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA IMF/WB: “Collateralized Transactions: Key Considerations for Public Lenders and Borrowers.” 2020. Landry, David. “The risks and rewards of resource-for-infrastructure deals: Lessons from the Congo’s Sicomines agreement.” Resources Policy 58 (2018): 165-174. Manzano, Osmel, and Roberto Rigobon. “Resource curse or debt overhang?” No. w8390. National Bureau of Economic Research. 2001. Meidan, Michal. “China’s loans for oil: asset or liability?” 2016. Mihalyi, David, Aisha Adam and Jyhjong Hwang “Resource-Backed Loans: Pitfalls and Potential.” Natural Resource Governance Institute. 2020. Rivetti, D., 2021. “Debt Reporting in Developing Countries.” World Bank Paper, forthcoming. Sanderson, Henry, and Michael Forsythe. “China’s superbank: debt, oil and influence-how China Development Bank is rewriting the rules of finance.” John Wiley & Sons. 2012. Songwe, Vera. “From Bottom Billion to Top Trillion: Using Commodity-Backed Securities to Support the Future of Africa’s Resource Economies.” Brookings. 2013. Tang, Keyi, and Yingjiao Shen. “Do China-financed dams in Sub-Saharan Africa improve the region’s social welfare? A case study of the impacts of Ghana’s Bui Dam.” Energy Policy 136 (2020): 111062. Venables, Anthony J. “Using natural resources for development: why has it proven so difficult?” Journal of Economic Perspectives 30, no. 1. 2016. Vizcarra, Catalina. “Guano, Credible Commitments and State Finances in Nineteenth Century Peru.” The Journal of Economic History. Vol. 69, no. 2. 2009. Xu, Jiajun, Xinshun Ru, and Pengcheng Song. “Can a new model of infrastructure financing mitigate credit rationing in poorly governed countries?.” Economic Modelling 95. 2021. 32 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA >>> Appendix a.1. Acronym List Bpd barrels per day CARI China-Africa Research Initiative CATIC China National Aero-Technology Import & Export Corporation CDB China Development Bank Chalco Aluminum Corporation of China CHICO China Henan International Cooperation Group CNPC China National Petroleum Corporation CPI China Power Investment Corporation CREC China Railway Engineering Corporation CSW cash-sweep mechanism DRC Democratic Republic of Congo DRS World Bank Debtor Reporting System DSSI temporary suspension of debt servicing EDCF Economic Development Cooperation Fund EITI Extractive Industries Transparency Initiative Eximbank China Export-Import Bank Gécamines La Générale des Carrières et des Mines (DRC) GIADEC Ghana Integrated Aluminum Development Corporation ICBC Industrial and Commercial Bank of China IDA International Development Association IMF International Monetary Fund KEXIM Export-Import Bank of Korea (South Korea) LIBOR London Inter-bank Offered Rate MFA master facility loan NPV Net present value NRGI Natural Resource Governance Institute OCCRP Organized Crime and Corruption Reporting Project RBL resource-backed loan 34 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA SHT Société des Hydrocarbures du Tchad (Chad) Sicomines Sino-Congolaise des Mines (DRC) SNPC Société Nationale des Pétroles du Congo SOE state-owned enterprise SPV special purpose vehicle WB World Bank ZESA Zimbabwe Electricity Supply Authority (Zimbabwe) 35 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA a.2. List of RBLs BORROWING MAIN LENDING AGREEMENT ID BORROWER ENTITY RESOURCE COUNTRY ENTITY YEAR 1 Angola Government Eximbank 2004 Oil 2 Angola Government Eximbank 2007 Oil 3 Angola Government Eximbank 2007 Oil 4 Angola Government Eximbank 2009 Oil 5 Angola Sonangol ICBC 2010 Oil 6 Angola Government CDB 2015 Oil 7 Chad SHT Glencore 2013 Oil 8 Chad SHT Glencore 2014 Oil 9 DRC Sicomines Eximbank 2008 Copper & Cobalt 10 DRC Government Korea Exim 2011 Copper 11 Ghana Government Eximbank 2007 Cocoa 12 Ghana Government Eximbank 2007 Cocoa 13 Ghana Government CDB 2011 Oil 14 Ghana Government CDB 2011 Oil 15 Ghana GIADEC Sinohydro 2018 Bauxite 16 Guinea Government ICBC 2017 Bauxite 17 Niger Government Eximbank 2013 Oil 18 Republic of Congo Government Eximbank 2006 Oil 19 Republic of Congo SNPC Gunvor 2010 Oil 36 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA 20 Republic of Congo Government Eximbank 2012 Oil 21 Republic of Congo SNPC Trafigura 2015 Oil 22 Republic of Congo SNPC Glencore 2015 Oil 23 São Tomé and Príncipe Government Government 2010 Oil 24 South Sudan Government Trafigura 2015 Oil 25 South Sudan Government CNPC 2015 Oil 26 South Sudan Government Eximbank 2016 Oil 27 Sudan Government Eximbank 2007 Oil 28 Zimbabwe ZESA CATIC 2004 Tobacco 29 Zimbabwe Government Eximbank 2006 Platinum 30 Zimbabwe Government Eximbank 2011 Diamond 37 >>> RESOURCE-BACKED LOANS IN SUB-SAHARAN AFRICA