GETTING REPAID IN ASSET FINANCE A Guide to Managing Credit Risk Nov. 2019 Matthew Soursourian and Ariadne Plaitakis May 2021 June 2021 Daniel Waldron, Holger Siek, Max Mattern, and Walter Tukahiirwa Acknowledgements The authors wish to thank the many experts and organizations without whose invaluable help this technical guide would not have been possible. Dedicated professionals at asset finance companies around the world shared information, insights, and took the time to walk us through complex, sensitive areas of their businesses. Specifically we would like to thank Alison Boess and Andreas Thiele (Engie Energy Access); Simone Vaccari (PEG Africa); Jon Saunders and Henry Clarke (SunCulture); Gaurav Mehta, Rajeev Ranjan, and Sarah Keller (Dharma Life); Michael Wilkerson, Marisa Fee, and Rodney Schuster (Tugende), Alan Mathers (EFTA); Abderahmane Sow, Loïc Vervoort, and Thebean Gilfillian (Oolu); Victoria Arch, Jennifer Boughton, Priscilla Choi, and Peter Kunhardt (Angaza); Esther Altorfer and Graham Day (Sistema. bio); Loïc Deschamps and Caroline Frontigny (uPowa); Laura Palgrade and Thomas Samuel (Moon); George Mugenyi (Cosma Sustainable Rural Development); Kany Keita (Energy+ Mali); Russell Lyseight (Vitalite); Kate Steel and Madeleine Gleave (Nithio); Alexandre Galicier (d.light); Sarah Mijabi (Greenlight Planet); Dustin Kahler (M-KOPA); and Anthony Mabonga (Bboxx). Dedicated risk management experts opened themselves up to new learnings and ideas in order to make this work happen. From Frankfurt School we would like to thank Dr. Joachim Bald, Basel Chaban, Natalia Carranza, and Christian Kuye. From IPC we must thank Isaac Williams, Clementine Hedan, Sathya Krithivasan, Rui Chikande, and Alaa Nasser. In addition, energy access and asset finance experts contributed key insights at various stages of the report, including Kevin Kennedy, Tamara Cook, Adrian Bock, Leslie Labruto, Geoff Manley, Leo Blyth, Buddy Buruku, Jamie Anderson, Michael Franz, Peter Storey, Simon Brossard, and Kat Harrison. Our own colleagues provided invaluable contributions: Nicky Khaki, Sai Kumaraswamy, Xavier Faz, Alexander Sotiriou, Juan Carlos Izaguirre, and Estelle Lahaye. Lastly, this research would never have been possible without the work and cooperation of the Global Off-Grid Lighting Association (GOGLA) and the Global Distributors Collective (GDC). We would like to especially thank Drew Corbyn, Rebecca Rhodes, and Roan Borst at GOGLA; and Emma Colenbrander and Charlotte Taylor at the GDC.  Consultative Group to Assist the Poor 1818 H Street NW, MSN F3K-306 Washington DC 20433 Internet: www.cgap.org Email: cgap@worldbank.org Telephone: +1 202 473 9594 Cover photo: Temilade Adelaja via Communication for Development Ltd. © CGAP/World Bank, 2021. GE T TING REPA ID IN A S SE T FIN A NCE RIGHTS AND PERMISSIONS This work is available under the Creative Commons Attribution 4.0 International Public License (https://creativecommons.org/ licenses/by/4.0/). Under the Creative Commons Attribution license, you are free to copy, dis- tribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Cite the work as follows: Waldron, Daniel, Holger Siek, Max Mattern, and Walter Tukahiirwa. 2021. “Getting Repaid in Asset Finance: A Guide to Managing Credit Risk.” Technical Guide. Washington, D.C.: CGAP. Translations—If you create a translation of this work, add the following dis- claimer along with the attribution: “This translation was not created by CGAP/ World Bank and should not be considered an official translation. CGAP/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 CGAP/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 CGAP/World Bank.” All queries on rights and licenses should be addressed to CGAP Publications, 1818 H Street, NW, MSN F3K-306, Washington, DC 20433 USA; e-mail: cgap@worldbank.org GE T TING REPA ID IN A S SE T FIN A NCE ACRONYMS AFC Asset Finance Company CDU Consecutive Days Unpaid CR Collection Rate EaD Exposure at Default EL Expected Loss KRI Key Risk Indicator KYC Know Your Customer LGD Loss Given Default MFI Microfinance Institution MIS Management Information System NPV Net Present Value PAR Portfolio at Risk PAYGo Pay As You Go RAR Receivables at Risk RAR(CDU) Receivables at Risk (Consecutive Days Unpaid) RAR(CR) Receivables at Risk (Collection Rate) PD Probability of Default RMC Risk Management Committee SHS Solar Home System TA Technical Assistance GE T TING REPA ID IN A S SE T FIN A NCE CONTENTS Introduction 1 Chapter 1: Asset Finance and Credit Risk Management for Low-Income Consumers 2 1.1 Credit Risk in Asset Finance 3 1.2 Key Challenges to Managing Credit Risk in Asset Finance 5 Chapter 2: The Organization 8 2.1 Risk Management Strategy and Risk Appetite 9 2.2 Risk Culture 10 2.3 Governance Structures 10 2.4 Management Information Systems in Asset Finance 17 2.5 Policies 18 2.6 Implementing Risk Management 20 2.7 Timeline for Adopting Risk Governance 22 Chapter 3: Product Design 24 3.1 Physical Product Design 24 3.2 Financial Product Design 27 Chapter 4: Credit Transaction Risk 33 4.1 Marketing, Origination, and Sales 35 4.2 Credit Risk Assessment 37 4.3 Decision-Making and Disbursement 42 4.4 Monitoring and Repayment 43 4.5 Collections 44 Chapter 5: Portfolio Management 49 5.1 Key Terms and Concepts 50 5.2 Measuring Portfolio Performance 53 5.3 Portfolio Analysis 57 5.4 Concentration and Diversification 61 5.5 Expected Loss and Unexpected Loss in Risk Management 63 5.6 Monitoring and Dashboards 65 Conclusion 67 References 68 Annexes 70 GE T TING REPA ID IN A S SE T FIN A NCE INTRODUCTION A SSET FINA NCE A LLOWS PEOPLE TO and assisting in implementing portfolio analytics such as obtain the physical items they need to generate vintage analysis and transition matrices. money, save time, reduce drudgery, and improve This guide summarizes the lessons learned from those their lives. From Sub-Saharan Africa to the Indian engagements and offers executives and managers at AFCs Subcontinent, asset finance and leasing companies are suggestions on how they can manage credit risk while doing invaluable, innovative work to finance critical assets growing their operations. Investors and other sector for low-income and informal borrowers.1 But unlike banks stakeholders may also find this guide useful to inform and microfinance institutions (MFIs), many of these their own due diligence and TA. companies do not have deep experience in organizing a credit operation, mitigating risk throughout a credit Chapter 1 offers a definition of the types of asset finance transaction or managing a portfolio of loans or leases. This business models addressed by this guide and discusses has important implications for the ability of asset finance the importance of credit risk management to asset companies (AFCs) to achieve financial sustainability: poor finance and the barriers to credit risk management in the credit risk management will prevent them from turning sector. Chapter 2 describes the necessary infrastructure receivables into cash, inhibiting their potential scale.2 for credit risk management within an organization. Chapter 3 covers the design of an asset finance product. Recognizing the need for more professional credit risk Chapter 4 goes through the steps of a credit transaction management in asset finance, CGAP provided technical and discusses how risk can be mitigated throughout assistance (TA) on credit risk management to more than the process. Chapter 5 dives into the management of a 15 AFCs in Sub-Saharan Africa and Asia between 2018 credit portfolio, including the analytical frameworks that and 2020. This TA, provided in collaboration with the managers will need in order to identify and mitigate risk. Frankfurt School of Finance and Management and IPC GmbH, came in two parts. First was a diagnostic of the companies’ credit risk management governance, policies, and practices: what was effective, ineffective, or absent. We laid out a set of recommendations for each company. Second was assistance in implementing the most pressing recommendations. These included training senior management on how to build risk management and audit departments, teaching credit scoring methodology, developing credit scorecards, 1 For more on CGAP’s global review of asset finance business models, see https://www.cgap.org/research/slide-deck/innovations-asset-finance 2 Throughout the paper, ‘AFC’ refers to an asset finance company operating in emerging markets in Sub-Saharan Africa or Asia, with at least a partial focus on serving low-income customers. GE T TING REPA ID IN A S SE T FIN A NCE 1 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management CH A P TER 1 ASSET FINANCE AND CREDIT RISK MANAGEMENT FOR LOW-INCOME CONSUMERS P SUMMARY HYSICA L ASSETS —W HICH IN THIS guide refer to tangible items such as vehicles, smartphones, productive equipment, and more— • ‘Asset finance’ refers to loans or leases that have been shown to help low-income customers improve allow a borrower to use a physical asset while their productivity, increase their incomes and enhance the paying for it over time. quality of their lives (Kumaraswamy et al. 2020). • ‘Credit risk’ is the potential loss that may But how do poor households acquire assets? In a country occur if one or more borrowers fails to make like Uganda, where 70 percent of the people live on less agreed-upon payments to their lender. In than $3.20 a day (World Bank 2020), few households can asset finance, this risk is increasingly being afford the upfront cost of a smartphone in cash, let alone borne by the makers or distributors of the something more expensive like a motorcycle. To acquire physical assets. these assets, they need to be able to spread the payments • Managing credit risk is essential for out over time. Practically, this means either saving up the (a) achieving financial sustainability for money to buy something or acquiring it and paying for it providers, (b) maintaining affordability for over time, plus interest (Rutherford 1999). customers, and (c) ensuring economic development. In this guide, ‘asset finance’ refers to the latter: loans or leases that allow a borrower to use a physical asset while they pay • Asset finance faces unique challenges in for it. In this type of finance borrowers usually do not pledge effectively managing risk: (i) it caters to additional collateral; the asset itself can be repossessed in the low-income, informal customers, (ii) it needs event of default. Ownership is typically transferred to the incentives to grow sales at the investor, borrower/lessor at some point in the transaction.3 provider and agent level, and (iii) it lacks traditional risk management structures. But Importantly, this guide specifically covers asset finance these can be addressed! businesses that target low-income customers. Innovations 3 Asset-backed loans and financial leases transfer legal ownership of the asset at different points, but they are similar enough to be covered together in this guide. Companies using operating leases may also find this guide useful. GE T TING REPA ID IN A S SE T FIN A NCE  2 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management TABLE 1. Main business models encountereda Pay As You Go (PAYGo) Leaseb Asset Types Smartphones, Cookstoves, Solar Home Motorcycles, Three-Wheelers, Solar Water Pumps, Systems (SHS), Televisions, Refrigerators Biodigesters, Larger SHS Price Range $50 – $500 $400 – $2,500 Loan Tenors 3 – 36 months 12 – 42+ months Repayment Flexible, although unit is deactivated Fixed, payments required on-time monthly Schedule for missed payments Average Risk Relatively higher default levels Relatively lower default levels Profile Credit Upfront deposit required. Data often limited References and guarantors required, payment capacity Assessment to Know Your Customer (KYC) information assessed, detailed questionnaire Credit Risk Lockout technology, call center follow-up Thorough assessment, stricter collections, relatively Mitigation on delinquents, attempt to price in expected faster repossession, ability to recover value from credit losses repossessed assets  he choice of lending model comes down to the nature of the asset and client type. Companies financing smaller assets or serving more remote a T customers often adopted PAYGo approaches; companies financing larger assets within tighter value chains (e.g., solar water pumps for cash crop farmers with a dedicated offtaker) or denser areas often opted for hire purchase. The term ‘lease’ represents two main business models here (hire purchase and financial leasing). In both of them, ownership remains with the lessor/ b vendor of the asset until all repayments are made, at which point ownership transfers to the payor. in business models and new technologies have enabled In Sub-Saharan Africa, for example, just one in three asset finance to expand beyond its traditional focus on households owns a television, one in six owns a refrigerator more affluent consumers and firms. And while there are (Efficiency for Access Coalition 2019), and one in nine many similarities to both more established asset finance owns a computer (ITU 2020). and retail finance, financing assets for low-income Asset finance for poor households entails a number of customers entails unfamiliar risks which require a tailored serious risks for companies (which is one reason why approach to manage. the sector remains underdeveloped). Assets may break, Overall, the business models covered by this guide tend to currency values may fluctuate, rains may fail, capital fall on a spectrum between PAYGo consumer financing may dry up, and regulations may change. Some of these for small assets and finance leasing (or hire purchase) for risks can be anticipated and planned for. Others, as the larger asset types. These types are compared in Table 1, COVID-19 pandemic made abundantly clear, cannot. although many businesses fall somewhere between the This guide focuses on one type of risk in asset finance: two. credit risk.4 Whenever a customer is given an asset now and promises to pay for it later, it is possible they will not fulfill that promise. That possibility, multiplied thousands 1.1 Credit Risk in Asset Finance or millions of times across a portfolio, is credit risk: the potential losses that may occur if one or more borrowers Because reliable, high-quality assets are expensive, fails to make agreed-upon payments to their lender. Not the demand for asset financing is high. Yet in low- all AFCs consider themselves creditors; many are not income countries that demand remains largely unmet. 4 It will become clear that credit risk is inextricably linked to other risk factors, such as operational risk, reputational risk, interest rate risk, asset liability management, and others. None of these can be looked at in a silo. Our main focus throughout is on credit risk, but we will touch on others when appropriate. GE T TING REPA ID IN A S SE T FIN A NCE 1.1 Credi t risk in a s se t fin a nce 3 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management regulated financial institutions and may operate more like needs like working capital, school fees and more. It is retail companies that allow customers to pay over time. AFCs who are assessing potential customers, making However, the basic principle of credit risk (an expected credit decisions, and financing sales on their own balance payment that may or may not be received) is a feature of sheets. However, many of these companies are new to any business putting products in the hands of customers the business of lending and do not have experience in in exchange for a promise of future payment.5 managing credit risk. L E N D I N G T O P O O R H O U S E H O L D S: WHY IS CREDIT RISK W H O TA K E S T H E R I S K? M A N A G E M E N T I M P O R TA N T ? For asset ownership to grow in lower-income countries— Credit risk management is at the heart of successful asset where many customers cannot afford to purchase finance for three reasons: important assets upfront—someone must take on the 1. Provider financial sustainability. If an organization credit risk of lending to poor borrowers working in the is unable to manage its credit risk (i.e., it is unable informal economy. But who? to turn contractual receivables into actual cash at its Partnerships between asset retailers and financial expected rate), then it will underperform financially institutions, for whom credit is a core competency, have and struggle to raise debt, eventually becoming achieved modest success at best. Commercial banks illiquid and then insolvent. If this happens across in many low-income countries do not directly lend to enough companies then the sector’s risk profile will households engaged in the informal economy (Srinivas rise, making capital more expensive. 2016; Bruhn and Love 2009), which excludes much, if 2. Customer affordability. If a company is pricing-in not most, of the population. MFIs lend to exactly these expected credit losses (as it should be, see Chapter 3 types of borrowers, but their procedures and skillsets are and Chapter 5), then every customer that defaults on often ill-suited to financing physical assets (Waldron et al. an asset loan/lease beyond the expected level makes 2019). Relying on financial institutions to provide financing that same asset more expensive for the next group therefore risks excluding many potential customers, with of customers. If companies can minimize defaults negative implications for impact and revenue. through better assessment, monitoring or collections, This has led some companies targeting low-income they will be making their assets more affordable (see customers—manufacturers and distributors of vehicles, Box 1 for an example). solar home systems (SHS), smartphones, and other 3. Market development. Low-income households need assets—to finance their sales on their own balance sheet, access to assets. But CGAP’s experience working with taking the credit risk on themselves.6 This has broadly AFCs shows that a failure to manage credit risk can been a positive development: credit risk creates incentives push companies to consider shifting more of their to distribute high-quality assets and provide quality after- lending toward higher-income customers and higher- sales service. In doing so, AFCs are often building the first value assets. Therefore, every time a lender tries and formal financial relationships that their customers have fails to serve low-income clients, they risk making ever had, contributing to their financial inclusion and it harder for those clients to access the things they enabling their access to further financing for important need. It may be that some people are not financially 5 Note that credit risk is the possibility of nonpayment. Once a customer has been determined (contractually or otherwise) to have failed to pay, the risk has materialized and is now a credit event (usually referred to as a default). The determination of default is subjective; different companies can and do have very different policies towards determining default. 6 This type of credit has a long history; manufacturers and retailers have formed an important part of the formal financial ecosystem since at least the 19th century, and the informal one long before that (Calder 1999; Fleming 2018). Retailers with in-store banks in Latin America (financing con- sumer durables), mobile network operators (financing smartphones), and manufacturer-owed auto finance companies are some of the largest lenders to lower-income households. GE T TING REPA ID IN A S SE T FIN A NCE 1.1 Credi t risk in a s se t fin a nce 4 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management viable customers for some assets.7 But if the failure AFCS ARE FINANCING ASSETS FOR A RISK Y were addressable, for example if it were the predictable G R O U P O F C U S T O M E R S: P O O R H O U S E H O L D S result of poor risk management, then this would WORKING IN THE INFORMAL ECONOMY represent a lost opportunity for companies, and for About 80 percent of Sub-Saharan Africa’s economic households that need every opportunity they can get. output comes from the informal sector. So does half of India’s economy, where the informal sector also employs 83 percent of its workforce (Murthy 2019). These are hundreds of millions of people who work desperately hard to provide for themselves and their families. But the  llustrating BOX 1. I the importance of credit risk managementa informality of that work is a significant barrier for formal lenders. Without proof of formal employment or cashflow A hypothetical case: A fictional PAYGo solar records for a microenterprise, lenders cannot be reasonably company in Kenya (we’ll call it SolarCredit) disburses assured that borrowers have the capacity to repay a loan. 12-months loans, each worth 1,000 Kenyan Shillings To complicate matters further, borrowers in the informal (KSH). Let the annual interest rate be 10%, so the sector face a greater set of risks: their incomes are often expected payment after 12 months would be KSH less regular and their employment less reliable than 1,100 per loan/lease contract. If one of the loans contracted employees. And security for lenders is hard to goes bad immediately (i.e., the borrower does not come by: collateral in the form of small household assets is make a single payment), how many “good” loans costly and difficult to repossess. would SolarCredit need to make to cover that loss? MFIs have sought to bridge these gaps using tools such In this example, the company would need to earn the as group liability, references and guarantors, in addition interest on ten good contracts each of them earning to detailed in-person cashflow assessments.8 But in KSH 100, a total of KSH 1,000, to compensate for many cases AFCs are trying to serve an even poorer the loss. In other words, avoiding one bad contract customer base, requiring them to adopt some tools from is as valuable as disbursing ten good ones. This microfinance, adapt others and even add a few of their reality underscores the importance of effective credit own. These include: assessment and scoring. • Remote lockout technology that reduces the risk Look for ongoing mention of fictional company SolarCredit a  of theft and expands the realm of possibilities in dark grey boxes throughout this guide. for nonpayment contract clauses beyond simple repossession. Remote lockout can also enable flexible financing terms, which adhere more closely to the lumpy income patterns of many informal sector jobs. 1.2 Key Challenges to Managing • Remote-sensing and location devices that facilitate Credit Risk in Asset Finance repossession and mitigate against unauthorized use of an asset. Despite the clear importance of credit risk management and mitigation, structural aspects of asset finance make • Digital payments, automated loan-management it harder for AFCs to prioritize and operationalize risk software, and well-trained call centers that keep costs management, particularly in their early growth stages. low by allowing large portfolios to be serviced by relatively few staff. 7 Indeed, subsidies have an important role to play in reducing the cost of critical assets for low-income households. 8 Unlike AFCs, MFIs have historically lent exclusively to micro and small enterprises or financed productive ventures. They have not traditionally financed household or consumer assets, although companies such as Inthree Access in India are beginning to change that. GE T TING REPA ID IN A S SE T FIN A NCE 1.2 K e y ch a l l enges t o m a n a ging credi t risk in a s se t fin a nce 5 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management • Use of the asset itself as collateral, eliminating the need impact. And in most cases, incentive structures reward for borrowers to pledge an existing asset, such as their agents or staff responsible for sales with a commission only house or land, which many customers may not own or when a sale is made, even though there may be higher have a formal title to. rewards for ‘quality’ sales. Taken individually, all of these incentives are By combining these and other attributes, AFCs have understandable. Taken collectively, they form a powerful been able to significantly reduce the risk of lending to push toward growth, at multiple levels of the business. low-income households. Although greater investment in James Grant, a financial journalist, wrote in Money of the formal risk management is still needed (see below), these Mind that “growth at an exceptional rate is a red flag in concerns are only relevant because of innovation on the banking…If loans are expanding too quickly, the lending part of AFCs that has overcome some traditional barriers officers have probably been saying ‘yes’ too frequently.” to financing-excluded customers. Unless the appropriate risk management structures are in place, the combination of social and financial incentives I N C E N T I V E S T H R O U G H O U T T H E VA L U E can lead toward growth that is unsustainable. C H A I N R E M A I N O V E R LY S K E W E D T O WA R D SALES GROW TH What does unsustainable growth look like? (See Box 2 Any company with assets wants to sell them. Social for examples.) For lenders to grow, they need to originate enterprises, whose assets can change people’s lives, really more loans or leases. This growth can come from (a) want to sell them. Social impact investors, who have increasing sales at existing branches and agents, (b) by placed equity or debt investments in these companies, also adding new locations and personnel, or (c) both. When want them to make sales because sales are what create the growth is expected from existing areas, loans are often BOX 2. Growth and credit risk in microfinance In the mid-2000s microfinance was viewed as a for-profit discipline to reach volume and increase credit risk.” In mechanism for eradicating poverty, and MFIs began the same paper, one MFI investor was quoted as saying attracting significant investment from development finance “These MFIs were growing so rapidly, they didn’t have institutions and private equity firms. These investors time to put proper risk management in place, and they were aware of the potential for this growth to cause a didn’t see the downturn coming.” decline in portfolio health but were able to point toward The 2008 financial recession caused or coincided with resilient asset quality as a sign that their investments were repayment crises in Bosnia and Herzegovina, Morocco, catalytic, not distortionary. An industry survey conducted Nicaragua, and Pakistan, among others. By 2009 ‘credit in 2007 ranked ‘credit risk’ as only the tenth highest risk to risk’ had soared to become the number one perceived the microfinance sector (Lascelles 2008). danger in the MFI sector (Lascelles and Mendelson Investment came with expectations of rapid growth, 2009). In 2010, a complex combination of market factors with the hope of bringing the benefits of microfinance and local politics converged in Andhra Pradesh, where to more and more people. But the institutions did not excessive lending and misaligned incentives within leading always have the systems or culture in place to manage MFIs (some of which had received significant external risk while doubling or quadrupling their loan volumes. funding) led to abusive collection practices, which in turn Reille and Forster (2008) wrote: “In many places, the debt led to an epidemic of farmer suicides. In the end, most market is overheated, and pricing does not reflect credit MFIs were forced to write off the majority of their Andhra and country risks.” Chen et al. (2010) noted that “target- Pradesh portfolios; SKS Microfinance alone wrote off driven, high growth can tempt MFIs to relax their lending loans totaling almost $200 million (Menon 2016). GE T TING REPA ID IN A S SE T FIN A NCE 1.2 K e y ch a l l enges t o m a n a ging credi t risk in a s se t fin a nce 6 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management given to more marginal borrowers, whose ability or established processes, and independent risk control). Yet willingness to pay may be sub-optimal. When growth these are the foundations of risk management and are is expected from new areas, the organization must absolutely required to sustain this kind of lending. onboard new personnel while maintaining its culture and The results are predictable: repayment rates are below standards. If the rate of growth is too fast it can lead to expectations for many companies. Not all companies poor underwriting, a deteriorating risk culture and a lack provision their expected losses (EL) fully or appropriately. of supervision. Data is gathered haphazardly and stored in ways that make This is the social impact credit trap (Waldron 2021), it difficult to analyze. EL are not always properly calculated, and it is a familiar one. Similar waves of social and are often not sufficiently priced in. investment have led to crises of repayment and consumer As described above, many of these firms began life as intimidation in both microfinance and digital credit retailers, distributors or manufacturers, and so do not (Izaguirre et al. 2018; Faux 2020). Whether the same have the requisite expertise in managing credit operations. pattern will be repeated in asset finance remains to be These companies have not compensated by investing in seen, but recent parallels bear watching. their risk management, and their cost of risk is likely higher than necessary (see Figure 1). That excess risk is A S S E T F I N A N C E C O M PA N I E S W I T H O U T C R E D I T either passed on to their customers (making their products EXPERIENCE OFTEN L ACK THE STRUCTURES more expensive) or to their shareholders (making the AND PROCESSES TO MANAGE CREDIT RISK company and sector less viable). Yet this is correctable, There is no one way to manage risk. Some lenders may if companies take reasonable steps to effectively manage prefer to lend only to high-quality borrowers. Others may credit risk. wish to be more aggressive and price-in higher expected loan losses. Neither of these approaches is inherently FIGURE 1. B  alancing the costs of risk wrong or right. But more aggressive approaches toward and risk mitigation risk-taking require more robust systems in place to manage that risk. “A bank driven to achieve aggressive $ Cost of growth targets may require more detailed credit policies Mitigation and more controlling administrative and monitoring systems to manage credit risk properly. Consistently successful banks achieve a balance between asset quality, growth, and earnings. They have cultural values, credit policies, and processes that reinforce each other and that are clearly communicated, well understood, and carefully Current followed.” (OCC 1998) Approach That is not what we see in the emerging market asset financing sector, where risk management is often treated as a technical exercise, as opposed to an issue of culture Cost of Risk and governance. Many companies in this sector prioritize Less Risk More Risk reaching scale—particularly those financing high-volume, low-cost consumer assets—and are naturally aggressive in their lending practices. Yet many are also relying on a mechanical approach to risk management, focusing on the ‘tools’ (lockout technology, pricing adjustments, agent commissions, etc.) as opposed to the ‘toolbox’ (articulated risk appetites and loan limits, detailed credit policies, GE T TING REPA ID IN A S SE T FIN A NCE 1.2 K e y ch a l l enges t o m a n a ging credi t risk in a s se t fin a nce 7 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management CH A P TER 2 THE ORGANIZ ATION I SUMMARY N T R A DIT IONA L F I NA NCI A L institutions—where risk-taking is a fundamental part of the business—corporate governance, independent • AFCs need to have a well-articulated risk risk management, and internal control work together to management strategy and risk appetite. This make sure that risks are managed, mitigated, or avoided will set the boundaries for how much risk the in line with the institution’s strategy.9 Although AFCs do credit operation can incur. not face the same regulatory scrutiny as these financial • A ‘risk culture’ is set by a company’s institutions, the complexity of some of their business executives. It should demonstrate that risk models, the challenging environments they operate in, management is a priority at every level of the and the volatility of their customers’ incomes makes business. risk management just as, if not more, important to their • The Three Lines of Defense, together with the success and survival.10 Board Risk Management Committee (RMC), As mentioned in Chapter 1, CGAP’s work with providers ensure that the risk management strategy reveals that while various risk management tools can be and policies are abided by. This management powerful and useful, they need to operate within a strong information system (MIS) enables monitoring culture and sturdy framework for managing risk (the and oversight of credit transactions and toolbox) to be effective. portfolio. This chapter explores in turn the elements present in • A policy framework is important to Figure 2: strategy, culture, governance, information formalize the company’s risk management systems, policies, and implementation that enable effective procedures, but companies need to conduct risk management. Not all of these frameworks will be interactive, participatory training to achieve feasible (or even reasonable) for early-stage companies understanding. to adopt, so the end of the chapter also provides rough • Companies can build out their risk guidelines on what the timeline of adoption ought to management infrastructure as they grow, but look like for an AFC. That said, all AFCs should have certain key elements should be present from the fundamentals of credit risk management: a clearly the beginning. expressed strategy for managing risk that is understood and embedded at every level of the organization. 9 Although originally designed for banks, Principles 14 and 15 in the 2012 BCBS “Core Principles for Effective Banking Supervision” as well as the 2015 BCBS “Corporate Governance Principles for Banks” describe current best practices in the governance of risk in financial services more generally. 10 See Sotiriou et al. (2018) GE T TING REPA ID IN A S SE T FIN A NCE 1.2 K e y ch a l l enges t o m a n a ging credi t risk in a s se t fin a nce 8 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 2: Organizational elements of risk management Risk Strategy and Culture Board of Directors Governance Information Systems and Policies 3 Internal audit 2 Risk management unit 1 Front-line staff and management Lines of Defense 2.1 Risk Management Strategy a high appetite for credit risk. The statement should and Risk Appetite consider all risk dimensions that may significantly impact operations, as well as potential reputational damage to the The first step in risk management is developing a plan company and its stakeholders, and be consistent with for identifying new risks, mitigating known risks, and the company’s capacity to manage risk. It should be regularly reviewing the organization’s approach. This plan approved by the board of directors and well disseminated is the organization’s risk management strategy. The throughout the company, especially for frontline staff and core components of that strategy are (a) the policies and agents. (A sample risk appetite can be found in Annex 1, processes around risk-taking activities (addressed later in as well as in IRM (2017), Deloitte (2014), and Batty et al. this chapter), and (b) the risk appetite statement, which (n.d.), among others.) is then translated into quantitative targets through risk The risk appetite should lay out each of the following capacity, tolerance, targets, and limits. (indicative examples using fictional company SolarCredit in italics). RISK APPETITE A company’s risk appetite is the nature and amount of 1. Risk capacity is the maximum risk that an risk that it is prepared to accept, tolerate or be exposed to organization is able to take regarding a certain risk. at any point in time while pursuing its business model. SolarCredit can only continue functioning if early - The initial choice of business strategy implies a certain defaults (within the first 90 days) stay below 5 percent. risk appetite, but the risk appetite statement formalizes 2. Risk tolerance is the maximum risk that an it by defining the types and magnitude of risk that the organization is willing to take regarding a certain company is willing to accept, or chooses to avoid, in risk. This tolerance is defined through quantified order to achieve its objectives. The risk appetite should risk measures of risk exposure. align with the company’s mission and strategy: we might expect a company whose mission is to finance small- SolarCredit is willing to tolerate 2 percent early defaults - scale assets for low-income farmers, for example, to have for each monthly cohort. GE T TING REPA ID IN A S SE T FIN A NCE 2.1 Risk m a n a gemen t s t r at eg y a nd risk a ppe t i t e 9 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management 3. Risk targets are the optimal levels of risk exposure This is not to say that AFCs should avoid credit risk; that a company wants to take. the only way to do that is to stop lending altogether. But it is the job of a credit operation to evaluate potential -SolarCredit is targeting early defaults of 1 percent. risks and determine whether they can be taken on and 4. Risk limits are thresholds set by an organization mitigated, or should be avoided altogether. If the sales on actual risk exposure. These are based on risk department is the gas pedal on a car, then the credit targets, and trigger automatic warnings and reviews department is the speedometer, and (if needed) the brake. if they are breached. A good risk culture involves acknowledging that the - Based on its risk target, SolarCredit sets a limit on early company has an appropriate ‘speed’ to run at, and that defaults of 1.25 percent. If breached, this requires an risk management is crucial to maintaining that speed. immediate meeting of the executive credit committee, and Practically speaking, a risk culture is determined by the the board RMC is to be notified. actions of leadership, emulated by managers and regularly In practice, many AFCs do not have a clearly articulated communicated and demonstrated to staff and agents at all credit risk appetite. Risk is often managed through levels. It includes: assumptions on losses, which are fed into pricing models • Open communication throughout the organization and revenue projections. These targets are not always (or in • Sharing knowledge some companies, even often) hit. • Continuous improvement One of the clearest takeaways from our work is the need • Dedicating sufficient resources to risk management for AFCs to better articulate both the amount and nature of risk they wish to take on, and how that risk will be • Use of information and reporting systems managed. They can then structure their operations, • Commitment to ethical and responsible business reporting and culture to consistently meet that statement. practices • A code of conduct that staff are required to sign onto 2.2 Risk Culture Other stakeholders can do their part to set or maintain a risk culture. Investors, in particular, should be wary The blunt truth is that none of the frameworks, principles of imposing aggressive growth targets that could lead a or tools described in this guide will make a difference company’s risk management to falter. They should also unless everyone in the company, from the leadership down, nominate directors who will help establish a sound risk believes that managing risk is important and should be strategy and appetite, set limits, monitor adherence to these, taken seriously. To quote one risk manager: “Effective and take action if the company’s performance deviates. risk management doesn’t function in a vacuum and rarely survives a leadership failure” (DeLoach 2015). This awareness of risk and accountability for managing it across the entire organization is often referred to as a ‘Risk Culture’. 2.3 Governance Structures In our experience, AFCs in low-income countries tend to Once an AFC has set its risk management strategy and be founded and staffed by people who embrace risk. This expressed its risk appetite, it will require a formal (i.e., ethos of risk-taking, experimentation and rapid iteration is explicit and documented) set of structures, reporting lines, critical in building businesses that can operate in remote policies, and procedures to make sure that the strategy is settings with limited infrastructure, serving low-income followed and that risk-taking stays within the set appetite. households. However, this same spirit can create problems Collectively, these structures are what we refer to as for a finance company that must place some limits on its ‘governance’ and are depicted in Figure 3. risk-taking. GE T TING REPA ID IN A S SE T FIN A NCE 2.2 Risk cult ure 10 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 3. The structure of risk governance Board of Directors Risk management Operations & Finance committee & audit committee administration committee Lines of Defense 3 Internal audit 2 Risk Manager (CRO) Compliance officer Risk Specialist & Compliance Staff Credit Risk Analysts Field Inspectors 1 Chief Executive Officer CFO COO Accountants Treasury Analysts Head Office, Branch Staff, & Agents Legend Both Risk Risk Management Operational Members of Management and Functions Functions Executive RMC Operational Functions This section discusses the four major levels of risk BOARD OF DIRECTORS governance in an organization: Effective risk management starts at the top of an • Board of directors and senior management organization. For most companies, ‘the top’ means the board of directors or supervisory board (for countries with • Front-line staff and management a two-board requirement). The board is the key interface • Risk management unit between owners of a company and its executive managers. • Internal audit Thus, this is where risk management strategy and risk appetite must be articulated and where responsibility for The latter three are referred to as the ‘Three Lines of overseeing risk management must ultimately fall. Defense.’ This section also covers the importance of a A foundational piece of sound risk governance is a board strong MIS to risk management. Policies and procedures, of directors that includes independent, non-executive although they fall within the umbrella of risk governance, members. Appropriate board members must be suitably are covered in the following section. qualified, effective in their work, loyal to the interests of the institution and its stakeholders, and care about properly discharging their obligations. The board should: GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 11 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management 1. Authorize and oversee implementation of the RMC is typically chaired by the chief executive officer institution’s mission, risk appetite, risk management and includes the CFO and COO plus the Treasurer and strategy, and related policies. Risk Manager as ex-officio members. 2. Establish and communicate corporate values (e.g., Alternatively, if the board is small and its members have by means of a code of conduct) which ensure that the technical expertise, the full board could serve as the consumer protection principles are understood and risk counterpart. Often, the Audit Committee may play followed, among other items. the role of the RMC as well. 3. Install a strong control environment wherein conflicts of interest are mitigated through appropriate policies. T HE T HREE LINE S OF DEFENSE: ORG A NIZ AT ION A L LE V EL S OF RISK M A N AGEMEN T 4. Actively oversee the institution’s compensation system Modern risk management in financial services is built and ensure that incentives are aligned with prudent around a ‘Three Lines of Defense’ paradigm, which risk-taking. requires companies to achieve the following: More specifically to credit risk management, the board • Everyone, from sales staff to call centers to service must ensure that: mechanics, does their part in controlling and 1. A sound risk management culture is established mitigating risk (First Line of Defense). throughout the institution and communicated to • Decisions taken at the first line of defense are third parties. analyzed by someone who can provide an independent 2. Policies and processes are developed for risk-taking perspective and communicates directly to executives that are consistent with the risk management strategy and the board (Second Line of Defense). and the established risk appetite, then reviewed • Owners and directors keep apprised of the functioning periodically and adjusted as the business evolves. of the organization by way of an internal audit function that reports findings directly to the board 3. Uncertainties attached to risk measurement are and has only limited reporting to senior management recognized. (Third Line of Defense). 4. Appropriate credit limits are established that are consistent with the institution’s risk appetite, risk While the Three Lines of Defense framework emerged profile and capital strength, and that are understood from the banking industry, CGAP’s global experience by, and regularly communicated to, relevant staff. indicates that it also translates well to risk management 5. An adequate MIS exists that allows them to obtain for non-bank lenders, including MFIs and AFCs. And timely and appropriate information (e.g., the health of although the budgets and number of staff engaged the loan portfolio). in internal control should be scaled to the size and complexity of the business (see Timeline section below), 6. Senior management take the steps necessary to we also find that the first two lines of defense can be monitor and control all material risks consistent with implemented by companies regardless of maturity, with the approved strategies and risk appetite. the third (internal audit) added as a company grows. In a typical microfinance bank, the independent risk and control functions include a RMC, the executive- First line of defense — front-line staff and management level forum where risk management happens. The RMC Credit risk management happens every time an agent covers both the asset liability management scope of presents a solar water pump at a district fair or someone interest rate risk, foreign exchange rate risk and liquidity calls the number on an ad for financed motorcycles. In an as well as the all-important credit risk and the related organization that distributes, sells and finances assets, most operational exposures in the credit process. The executive staff members have a role to play in managing credit risk. GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 12 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management BOX 3. First line of defense SunCulture, a company takes into account when reviewing the application. This headquartered in Kenya, assessment is intended to mitigate overindebtedness sells and finances solar- of the customer, as well as any fraud. The Relationship powered water pumps and Manager approves or denies the application, together other assets, and also provides agronomic support to with the Head of Credit. their farmer clients. SunCulture’s products range in price Once a deposit is received, the Engineer conducts a from $500–$1,500 and are financed from 24–36 technical assessment of the client’s location. They flag months, as well as being sold on a cash basis. With the any potential technical issues with the site or product’s size of the asset and the tenor of the exposure, intended usage. Engineers also review and update/ SunCulture has prioritized building a multi-step process provide feedback to the credit team (i.e., if a customer for bringing in new clients: clearly doesn’t have a use case for the pump, then they SunCulture sales agents make contact, describe the won’t install the unit and report back to the credit team). attributes and terms of the product. If the potential client The Relationship Manager calls the client is interested and meets minimum KYC requirements post-installation to confirm understanding of credit (national ID, age, credit bureau check, and can make the requisite deposit) they are forwarded to a terms and conditions by reviewing the contract Relationship Manager. (again), as well as the customer’s obligations (i.e., confirming they know when and how to pay). Relationship Manager assesses the client’s intended use of assets (business or personal) as well as Automated payment reminders are sent to customers their ability and willingness to pay. Additionally the 3 days before their scheduled payment day. Relationship Manager takes the customer through In the event of nonpayment, the Relationship Manager SunCulture’s credit survey process, which is a follows up with the client to understand the reasons for continuation of updating the profile that the Agent has delay (to help assess if the pump should be turned off started. All survey questions are weighted and tabulated or repossessed), and if needed to support in collections into a final ‘credit score’ which the Relationship Manager and eventually repossession. Making sure that they are informed and able to do so in role in avoiding credit risk through customer assessment. their operational capacity is the first line of defense for any Service staff such as those in call centers or maintenance AFC. This means that executives need to make sure that shops also must understand the importance of their role all operations are planned, and appropriate policies and to managing risk. This will be explored in more depth in procedures are drafted, with risk management in mind. Chapter 4. The first level of controls is always directly embedded in Second line of defense — operations through separation of duties: a branch manager risk management function will approve the loan applications prepared by his field The second line of defense is the risk management unit staff, the head teller in the branch controls the junior covering all material risks to the company, including credit. tellers’ daily ledger, etc. See Box 3 for an example. The Risk Manager heads up the unit. In a larger institution, In AFCs, the most direct way this happens is through the the Risk Manager may be elevated to executive rank and agents or staff whose responsibility is to sell the physical carry the title of Chief Risk Officer. The Risk Manager asset. Because they are the first (and sometimes only) reports directly to the highest executive officer and has point of contact with the customer, they play an outsized frequent interaction with the board-level risk committee. GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 13 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Key qualities of the Risk Management unit include: compliance with risk policies. This approach weaves a risk perspective directly into the fabric of day-to-day business, • Provides a comprehensive company-wide view of risk where the risk-taking happens (see Box 4 for an example). across all material risk types. • Captures macro risks arising from the markets in which the company operates, and incorporates such assessments into the company’s risk management process. BOX 4. Building a risk management unit • Invested with sufficient resources, independence, Tugende is an AFC authority, and access to the Board to perform its duties operating in Uganda effectively. and Kenya. Its primary • Supported by information systems that are adequate loan product for many (both under normal circumstances and in periods of years was a lease-to-own motorcycle for established stress) for measuring, assessing and reporting on the motorcycle taxi drivers (known in Uganda as ‘bodas’). The motorcycle taxi business in Uganda is size, composition and quality of exposures across all well-structured, with drivers operating as risk types, products and borrower segments. independent members of local ‘stages,’ where they • Subject to regular review by the internal audit function. operate and are known. This structure enables Tugende to assess The Risk Manager and the small expert team serve first risk locally, through in-person interviews and and foremost as a competence center and as internal discussions with other drivers in a stage. It also consultants in risk management methodology. Their means that risk management at Tugende is mostly primary responsibility is to support the business units decentralized. While the company was in its in maintaining and complying with policies and infancy, credit procedures were built up over time procedures to ensure that all risks are identified, measured at a limited number of branches. Eventually these and managed. As part of this they are responsible for were codified into credit policies, and are now developing, owning and upgrading risk policies. Risk executed at a branch level, with a second level of Management also takes independent risk measurements risk management for oversight. as necessary (e.g., on a key risk metric such as early To ensure that branch credit officers are repayment) and reports those directly to the board and knowledgeable about Tugende’s policies and able executive management. to execute them, Tugende created a Head of Risk position. This person is responsible for overseeing Embedding the Risk Management Unit compliance with Tugende's credit policies. in Everyday Credit Activities Even with a well-established and independent risk • Monitoring portfolio health. management unit, it would be unrealistic to believe that • Updating policies on a regular basis. self-control by sales agents or credit officers is always good • Helping HR to develop risk-based elements of enough. Meeting disbursement targets today almost onboarding training. always beats portfolio quality in six months’ time, and catching deviations from lending policies or outright fraud The Head of Risk reports directly to Tugende’s CEO a year later at the next internal audit inspection will often and chairs the RMC. The RMC is also responsible for be too late. So Risk Management may carry out a second approving all extraordinary exposures, such as those level of risk control using field inspectors and/or risk resulting from the introduction of new products. analysts, who are deployed throughout the organization and oversee staff engaged in credit activities. These risk staff are not generally taking actions on a specific loan/lease but rather are monitoring these activities for GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 14 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management The defining feature of a risk management unit should Third line of defense — internal audit be its independence. This means that risk management The Internal Audit function is the control level of staff should not be incentivized on commercial risk- controls—and the last line of defense. Its remit includes taking, should not be directly involved in credit the full scope of the company’s activities, but specifically transactions and should not be subordinate to the business also activities and reports of the executive managers managers whom they control. including the CEO, finance and accounting, as well as the risk management unit. Internal Audit assesses In practice, when Risk Managers participate in the effectiveness of the company’s risk management commercial decisions their vote should not necessarily frameworks. It reports directly to the board level, typically dictate the final outcome. Otherwise, they might be the Audit Committee. Importantly, the Head of Internal tempted to say no to everything. Every business failure Audit is hired, fired and evaluated by the board Audit would be blamed on risk staff, who prevent line managers Committee, not by executive management. This is to from building a profitable business. Instead, the vote of ensure their independence and that the board receives an the Risk Manager should be positioned as a technical unbiased view of the company’s risk management. opinion on the compliance of the proposed decision, activity or product within the established risk policies and Internal Audit serves as the ultimate control instance, tolerances. The business managers may overrule the vote typically carrying out its analysis on an ex-post and of the Risk Manager against a certain transaction in what sample basis. In a three-tiered control structure, Internal becomes a documented exception or override. Internal Audit can focus on assessing the effectiveness of the prior Audit will take a special interest in the nature, frequency controls rather than being relied upon to actually catch and ultimate outcome of such overrides and report those and rectify individual errors and omissions. to the Board. BOX 5. Risk management self-evaluation Fenix International is a School helped Fenix develop a risk management next-generation energy self-evaluation, to be conducted annually in each company that operates in six country by a risk manager from another country. African countries. Acquired by Given that different Fenix country units are run with a Engie in 2017, Fenix’s Credit Department manages risk significant degree of autonomy, this approach offered through its sales agent network, call center, and device the right mix of pragmatism and independence to be lockout technology. implemented quickly. Given its state of growth, Fenix worked with CGAP to The actual evaluation helps evaluate the status of understand how it could incorporate the functions of various risk management components, which are an Internal Audit department without building out an weighted by their importance. The tool produces a risk entire independent unit. In the end, CGAP and Frankfurt matrix, as in the example seen below. Critical 4 — — — — — Critical 12—16 — Importance Significant 3 — — — — — Major 8—11 — Relevant 2 — — — — — Moderate 4—7 — Minor 1 — — — — — Minor 1—3 — Irrelevant 0 — — — — — Okay 0—0 — 0 1 2 3 4 Good Adhered Partially Weak Missing practice adhered Status GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 15 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management BOX 5. Risk management self-evaluation (continued) This matrix can then be translated into more detailed findings, such as this example here: Findings Category Sub Category Issue Finding Comment 1 Concentration Limits Limit Reporting Responsibilities Critical No policies/procedures in place to minimize portfolio risk. 2 Credit Portfolio — UL Calculation/Stress Critical No calculation of unexpacted Risk Identification & Testing losses and/or stress testing. Measurement 3 Credit Risk Policy — Process Description Critical No policies in place to minimize & Procedures — portfolio risk. Portfolio Risk 16 Credit Risk Policy — Diversification Strategy Major Diversification driven by strategy/ & Procedures — business focus as opposed to Portfolio Risk risk management focus. 17 Internal Control — Regular Control Field Major Control units set-up in the past year. Visits No resident Internal Audit function. Currently internal audit team is from holding company and may not fully appreciate business model. 18 Customer Recovery & Adequate Handover of Major No formal process for handover of Assessment Collection Problematic Cases defaulting/problematic cases. PHAs face difficulty in tracing “cancelled” clients. 37 Credit Risk Policy — Risk Measurement & Moderate Portfolio reporting functionality & Procedures — Accounting in place. However, reports on Portfolio Risk concentration, limits, etc. not available. 38 Credit Risk Policy — Job Descriptions Moderate Portfolio reporting available, & Procedures — however, there is no clear Portfolio Risk assignment of responsibilities for portfolio monitoring. 39 Internal Control — Independent Internal Moderate No resident Internal Audit function. Audit Function Currently internal audit team is from holding company and may not fully appreciate business model. 63 Risk Limits Limit Monitoring Time Minor none & Reporting 64 Staff Incentive System — Alignment to Fenix Minor none International’s Social Mission 65 Internal Control — Client Satisfaction Surveys Minor none & Complaint Procedures 76 Key Risk Indicators — Receivables at Risk (RAR) Okay — (PAYGo PERFORM) 77 Credit Portfolio — Vintage Curves Okay none Risk Identification & Measurement 78 Credit Portfolio — Transition Matrix Okay none Risk Identification & Measurement GE T TING REPA ID IN A S SE T FIN A NCE 2.3 Go v ern a nce s t ruc t ures 16 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management In an asset finance context, auditors might check the However, in regard to credit risk in asset finance, a documentation received for a sampling of borrowers or properly functioning MIS is what connects the Three observe a credit assessment to ensure that procedures Lines of Defense. It is critical for accomplishing the are being followed. They would certainly want to ensure following tasks: that inventory is being managed appropriately, and that • Recording customer information for KYC and credit recoveries and repossessions are proceeding as intended. assessment Internal audit is the least common line of defense in the • Approving/processing individual credit transactions AFCs that CGAP has worked with. However, Box 5 • Monitoring portfolio health and key risk indicators illustrates how a growing AFC can accomplish some of the (KRI) (see Chapter 5) goals of an internal audit department before it is able to dedicate the full resources necessary. • Monitoring agent activity and tracking commissions • Analyzing repayment patterns by cohorts/segments • Auditing credit transactions and risk monitoring 2.4 Management Information • Managing and communicating with a portfolio of Systems in Asset Finance financed assets • Supporting Board and senior management oversight Risk management is a data-driven exercise. That data must be recorded and stored on a software platform that: • Enables staff to access the information they need to do BOX 6. A single source of truth their job In four of our seven projects, we encountered serious • Allows for the analysis and visualization of information issues when we tried to analyze loan repayment data, or replicate company analysis. While the • Generates relevant reports/dashboards in an accurate causes varied, in all cases executives were forced to and timely manner rely on multiple inadequate data sources to inform • Compels adherence to defined procedure their decisions. • Provides a clear audit trail The single most important function that an MIS serves for a credit manager is as a ‘single source For the purposes of this guide, that software platform is of truth’ that consolidates real-time transaction and called a MIS, although some organizations may prefer portfolio-level information. Without an effective and ‘customer relationship management.11 reliable MIS in place, credit risk management is almost impossible. The scope of an MIS goes beyond the management of credit risk. Such a system may also be used to track inventory, oversee operations, manage budgets, and for other critical tasks. Some or all of these functions are likely to be carried out by distributed applications (one This last function is the most prominent for many of our for inventory, another for accounting, etc.) running on a partner AFCs (see Box 6). Particularly in the PAYGo single network, in which case the MIS may include a data sector, customers’ payments need to be rapidly and warehouse and middleware to facilitate the movement of automatedly processed to unlock their devices. There information between different systems. are several specialized software providers that offer this service, while some companies have developed their own proprietary software. 11 In banking, an MIS refers specifically to a system used for reporting, distinct from a core banking system. However, in microfinance, ‘MIS’ is often used to refer to the entire back-end system, which is how it is employed here. For more detail see Braniff and Faz (2012). GE T TING REPA ID IN A S SE T FIN A NCE 2.4 M a n a gemen t Inform at ion S y s t ems in a s se t fin a nce 17 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Selecting or developing an MIS is an organizational create trust for third parties such as investors or regulators. process beyond the scope of this guide. Risk managers They inform workflows, onboarding and periodic training should be keenly aware of their requirements for an MIS, (see Box 7), although they may need to be translated into and ensure that these are met. The MIS also introduces other media to be most effective in staff training. an element of credit risk: for example, a failure to unlock Policies should be directly aligned with a company’s assets will quickly cause a payment crisis. Regularly risk management strategy. In asset finance, that could updating and testing risk-adjacent software is one of the mean that management of ‘problem loans’ may focus risk management unit’s responsibilities. more strongly on repossession (including refurbishment and resale) and remote lockout (if applicable) than on rescheduling or denying access to future loans. 2.5 Policies As an institution grows, so too will the number of policies Having set up the structures and systems for it establishes and the level of detail. For example, a implementing a risk management framework, the next company may start with general risk management policy, step is to give them something to do. Risk management then add another specific to its more important risk, policies take general concepts, processes, responsibilities, such as a credit risk management policy. The details can and actions and formalize them into written documents. later be described in supporting documents such as work A formal risk policy helps to ensure consistent procedures or process charts. implementation and company-wide understanding of the Risk policy documents should be developed, “owned” and various procedures, responsibilities and processes around regularly updated (see Box 7)—typically at least on an risk management. Written policies are also essential to annual basis—by the Risk Management Department (or BOX 7. Translating policies into training materials Tugende has a which involves gaining expertise and exchanging comprehensive in-house experience by rotating between different branches and training system managed departments. Tugende has a team of nine in-house by a Deputy in its HR trainers (seven of them from operations) and provides department who is responsible for training at the classroom training and online training done through company. Training for newcomers includes one-week an educational platform, which includes testing. The onboarding training that includes an introduction to all company also uses external training providers and departments and functions, a one-week induction partners with a local banking institute. training received from the functional supervisor and However, in some cases evaluators found that then an on-the-job training and probation period of employees had a different understanding and three-to-six months. interpretation of the procedures, which affected Tugende offers corporate and functional courses for its processes standardization and thus could increase employees and provides refresher functional training, the risk. This mismatch between procedures and communication skills, critical thinking, typing skills practice also affected training, which is based on trainings as well as Training of Trainers. Permanent written procedures. This was a clear sign that policies staff develop a personal training plan aligning it with and training need to be regularly updated, then their own chosen career path. Ambitious employees monitored for their compliance, especially in high may apply for a leadership training program, growth companies. GE T TING REPA ID IN A S SE T FIN A NCE 2.5 Pol icies 18 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management similar body). It is recommended that their approval be • Introduction (vision, mission, risk management executed by the senior management and/or board. strategy, risk appetite) • Organization of credit (organizational structure, CREDIT RISK POLICY staffing, job descriptions, roles/responsibilities) For AFCs, having a solid credit risk policy framework is • Loan/lease products just as important as it is for banks and MFIs. The same • Processes (assessment, onboarding, monitoring, cross- goes for the major topics that must covered by such selling, collections, recovery) policies (as well as accompanying manuals and procedures) and which can be broadly divided into transaction and • Standards and compliance (consumer protection portfolio risk management actions. For small providers principles, internal code of conduct) it may be appropriate to integrate all this into one credit • Portfolio management (segmentation, limits) (risk management) policy. The following outlines the core • Reporting (KRIs, reporting frequency) ingredients of a possible credit policy, although companies may wish to add additional elements (see Annex 2 for As an organization grows, their credit risk management more detail). policy framework may look like Figure 4. FIGURE 4. Credit risk policy framework template Credit Risk Management Policy Customer Collateral Restructuring & Loan Loss Credit Risk Portfolio Evaluation & Problem Loan Provisioning Assessment Management Management Management & Write-off & Monitoring Manual Manual Manual Manual Manual Risk Classification/ KYC Procedure Scoring Procedure Influences Job Descriptions and Risk Management Responsibilities GE T TING REPA ID IN A S SE T FIN A NCE 2.5 Pol icies 19 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management JOB DESCRIPTIONS S U M M A R Y: P U T T I N G P O L I C Y I N T O P R A C T I C E AND TRAINING REQUIREMENTS Experience shows that in companies with more than Risk management policies typically include short job a handful of people, formalization is necessary but descriptions for all relevant positions, as well as the key not sufficient to ensure that behavior on the ground risk management tasks that they must be trained and able complies with the organization’s policies. For many to carry out. For AFCs, this is an opportunity to map out AFCs a formal policy framework will have little direct how risk should be managed throughout an organization, impact on fieldwork, as few agents or even junior staff even for employees whose tasks may not appear to be are likely to read it or commit it to memory. Instead, the directly related to credit. The goal is to convey that policy framework should inform standardized trainings, ‘everyone is a risk manager’ in their given role. ongoing staff capacity building, consistent leadership, Job descriptions should include key activities, critical a system of checks and balances, and well-designed company risks related to their job, detailed explanations incentive structures. These are the components of a well- of important risk management actions, and an example of implemented institutional risk management system, of how any risk-related compensation is rewarded. which credit risk management is a major part. Initial and periodic training is the main way that staff will be introduced to the company’s risk management strategy, risk appetite and risk culture. These trainings should 2.6 Implementing be put together with great care. Training of trainers is Risk Management therefore of fundamental importance and an area where the company should dedicate resources. Case studies I N T E R N A L T R A I N I N G A N D C O M M U N I C AT I O N should include real-life instances where the company fell How often, and in what way, organizations choose to short in managing risk, and how it could have done better. communicate their expectations around risk management All of the above (policies, job descriptions and training) has a significant impact on its implementation. There are become more important, not less, for third parties such as several key stages in the communication and training of sales agents. In asset finance these parties typically carry risk management content: out critical tasks but are more difficult for the organization 1. Staff sensitization. Management must ensure that to manage as they are not employees. Policies need to every staff member understands the importance factor in the particular role they play, and the challenges and principles of risk management. At a minimum, that creates for risk management. everyone involved in credit operations must Lastly, we strongly recommend formalizing every staff understand that any scheduled payment that is not member’s commitment to respect the company’s code of received, or arrives later than expected, has a negative conduct, for example when it comes to abusive collection effect on the company’s cash flow, profitability and measures. A code of conduct is a written collection of the impact. Staff need to know that if non-payments rules, principles, values, employee expectations, behavior, reach a critical mass, the long-term survival of the and relationships that an organization considers significant company may be at risk. and believes are fundamental to their successful operation. 2. General risk training. Risk Management staff It should be available on a public folder, read and signed should provide regular classroom sessions for all new by every staff member. The code contains guidelines staff, which include participatory exercises and group for appropriate behavior toward colleagues, clients, and work. Infrequent but regular refresher courses for all other stakeholders. It clearly states what is considered experienced staff are also important. For staff involved misbehavior and lists possible sanctions. It has a reference in credit risk management, trainings should start with to general customer protection rules (e.g., GOGLA, a general risk awareness training, and later focus on SMART) and addresses issues like sexual harassment, the particular work each staff member is expected physical and psychological violence, racism, etc. GE T TING REPA ID IN A S SE T FIN A NCE 2.6 Impl emen t ing risk m a n a gemen t 20 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management to do, and how their role helps to manage risk for M A N A G I N G S TA F F A N D I N T E R A C T I O N S the entire company. Back-office staff responsible for While some aspects of good risk staff management are provisioning and loss calculation require specific similar to those of other business units (clear reporting training in those areas, while Loan Officers require lines, well-articulated goals and priorities, open more training in customer analysis, questioning communication channels), risk management relies heavily techniques, etc. on the concept of checks and balances. A company should 3. Interactive learning. Very few, if any, non-risk staff ensure that Dual Control and the Four Eyes Principle, as will be interested in memorizing technical details. examples, are widely implemented in practice. Most people tend to learn better through interactive Four Eyes Principle. This requires that any activity by learning formats, including group work, collective an individual within the organization with a sufficiently discussions, and role plays. Here is an example (Figure 5). serious risk profile (as determined by company policy) must be reviewed or double checked by a second FIGURE 5. A fun example of how to communicate  competent and independent individual. This is done to risk concepts mitigate the risk of poor execution through mistake, oversight or fraud. Dual Control. This is similar to the Four Eyes Principle, but more stringent. It requires that a sensitive activity may only be undertaken when two people are Risk Perception Risk Analysis Risk Management Identify and Analyze the Take action simultaneously present. acknowledge probability and to mimize the the risk impact, consider likelihood of the risk These can be especially relevant for assets that require both mitigation options a technical and financial appraisal of the customer. But the basic principle of involving another staff member in a credit decision is a key tactic for ensuring that practices 4. Cross-fertilization of ideas. Risk staff ought to are complying with policies. regularly present their work to other departments of the company, as this is useful to create mutual Additional checks and balances can be operationalized in understanding. This is not meant to completely several ways (See Table 2). eliminate the natural tension between business and risk staff— a friction which is healthy and needed TABLE 2. Technical checks on risk taking to a certain extent. However, having an idea of how Dedicated IT Approved rights to read, write or change the “other side” works and why it assesses things as it user rights certain information in the system. does can help to facilitate constructive discussions and Mandatory For critical issues (e.g., a credit vote), the a search for problem-oriented solutions. By the same second vote system would require a second review, token, ‘non-risk’ staff should also be encouraged to recorded in to be confirmed by ticking a box and the system recorded in the system. provide risk staff with feedback on policies and any Formally Similar to IT user rights. Credit decisions emerging issues they may observe. assigned can only be made and recorded by 5. Updates. When policies change, all affected staff approval rights persons who have the approved right to do so. should be informed in an interactive format (typically Automated A system warning in case of limit a classroom presentation), highlighting the key flag of limit breaches – (e.g., e-mail notification for changes and the impetus behind them, demonstrating breaches management and controllers). their implementation through examples or role play Built-in Recording login and logout times for all and offering plenty of time for questions. log files system users. Helpful for fraud detection (e.g., repeated night access or stolen passwords). GE T TING REPA ID IN A S SE T FIN A NCE 2.6 Impl emen t ing risk m a n a gemen t 21 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management The internal audit function, as the third line of defense, 2.7 Timeline for Adopting would be the body to ultimately check on-the-job Risk Governance compliance with what is fixed in the policies. Each core function should be checked about once a year by an No institution is immune to borrower default, employee internal auditor, who would record policy violations and fraud, theft, reputation risk, foreign currency exposure, report those to the board. Such checks can be sample or other particular risk dimensions just because they based—not necessarily random but representative, perhaps are small. Quite the opposite: smaller and less mature with a stronger emphasis on riskier exposures (e.g., by organizations are more vulnerable to unexpected losses checking not only files from “good” loans but also those from risks that they do not understand or lack the capacity from defaulters and restructured cases). While generally to mitigate. Risks do not wait, and therefore it is never pre-announced, a couple of audit visits may even be too early to begin managing them. As many companies conducted as surprise visits. turn to on-balance sheet financing in absence of a willing financing partner, long-term viability will depend on S TA F F M O T I VAT I O N established structures for managing risk. Incentive systems play an important role in aligning At the same time, there is not a one-size-fits-all approach quantity and quality targets, and should match with to managing risk. We do not advocate for companies or both the company’s mission and its risk appetite. These their investors to impose an oversized internal control could include sales commissions, good-portfolio bonuses, monster upon early-stage asset finance businesses. Many recovery commissions, and others. innovative AFCs are comparably young, remain on the Typically, customer care is split into different activities, smaller side when it comes to revenue and portfolio size, such as customer acquisition and analysis, troubleshooting, and cannot afford to hire full-time internal auditors or maintenance, arrears management, and collections. This risk managers. makes it easier to measure the performance of each team Our approach in this guide is rather to set out widely and to avoid “perverse incentives.”12 For a relationship accepted and effective practices that are easily scalable to manager who intervenes in cases of early arrears, the fit the complexity and budget of a particular company. As incentive might be negative: a deduction from their good- companies grow, we would expect their internal control portfolio-bonus based on any balances that exceed their units to increasingly resemble those of larger financial set arrears threshold. institutions. But smaller organizations can still find ways We explore some of these incentives in more detail in to ensure that these responsibilities are carried out. As an Chapter 4. For risk managers, their job is to regularly example, an early-stage start-up may have the second line review whether these schemes are having the desired effect, of defense covered by the CFO, while internal audit can whether they are creating perverse incentives, or even be outsourced to external auditors as a separate task from leading to outright fraud. Internal audit will wish to check their traditional financial audit. a sample of these bonuses or commissions as well. 12 Perverse incentives could arise, for example, if a relationship manager or loan officer would be entitled to the same bonus on collections as a mem- ber of the escalation team. There have been cases where a loan officer actively encouraged clients who were in early arrears to not pay right away, but to wait until the next month to make the installment. This would mean an installment collected from a client in 91–120 days rather than in 61–90 days and thus a higher incentive on the recovered installment. GE T TING REPA ID IN A S SE T FIN A NCE 2.7 T imel ine for a dop t ing risk go v ern a nce 22 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Table 3, as well as the section below, presents several key 3. Ex-post controls. Internal audit controls verify the pillars of risk governance and how they might evolve over risk management framework (including but not time in a growing asset-finance startup: limited to credit risk) regarding both formalization and practice. In a small institution (lease portfolio 1. Senior management/separation of functions. below $25 million or equivalent), the internal audit Existence of at least a full-time CEO and CFO. For may also be outsourced to an external auditor or companies with >$10 million in assets, a full-time consulting firm. Risk Manager. For companies with >$25 million in assets, a risk management function and an internal audit function; fully independent up to the management level. 2. Formalization and dissemination of policies. Existence of at least a Risk Management Policy. For institutions with lease portfolios >$25 million (or equivalent) supplemented by up-to-date credit manuals, credit risk management procedure, financial management manuals, and internal control manuals, at a minimum. TABLE 3. Risk management needs, by stage, for growing AFCs Early-stage Growth Maturing ($0-$10M portfolio) ($10-25M portfolio) ($25M+ portfolio) Vision and Mission Statement Risk Appetite Statement Board of Directors RMC Risk Management Policy (incl. Credit) Risk Culture MIS Chief Risk Officer Internal Risk Management Dept. Internal Audit Dept. Detailed Process Manuals GE T TING REPA ID IN A S SE T FIN A NCE 2.7 T imel ine for a dop t ing risk go v ern a nce 23 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management CH A P TER 3 PRODUCT DESIGN B SUMMARY EFOR E A CLIENT BU YS A PRODUCT on credit there needs to be a product to sell. In the case of asset finance, a ‘product’ consists of both a • Asset finance entails the sale of both a physical, tangible asset, as well as a financial product that physical asset and a financial product. The allows the customer to pay for the asset over time. The design of each can play a major role in features of each of these products, and the relationship mitigating or exacerbating credit risk. between them, do much to mitigate or increase the risk of • The quality of the physical asset is the single individual client transactions. greatest determinant of credit risk for AFCs. If customers are not satisfied with the physical asset, they are much more likely to stop 3.1 Physical Product Design paying. • Determining effective interest rates is The quality of the physical asset being financed is the a challenge in asset finance due to the single greatest variable in managing credit risk for subjectivity of retail vs. finance costs. But asset finance providers. Meeting customers’ intended it is crucial for allowing customers to make purposes is a key outcome of responsible finance. It is also comparisons and setting a benchmark rate important for mitigating credit risk. Put simply: if the for financial decisions. asset does not work as expected then people will not pay for it. Low-income customers of AFCs are unlikely to be • Pricing is a key tool for many companies concerned about future credit from those companies, and in their risk management, but it cannot even less likely to be worried about their credit score— wholly replace risk management. Using although companies are increasingly introducing new longer tenors to make regular installments products available to customers in good standing and affordable can create credit issues down that may incentivize credit discipline. In the absence of the line, as fatigue sets in and newer assets these incentives, physical asset performance is a necessary become available. (though not sufficient) condition for managing credit risk. A few of the major risk factors in physical product design are captured in Table 4. on physical product attributes that drive risk, to propose Executives, product designers and procurement leads solutions for mitigation or to advocate for less risky asset should have these features in mind when deciding features, types or brands (see Box 8). which products get offered to customers. But credit risk Despite the importance of the physical product, its quality managers are also well within their rights to weigh in and value will not eliminate credit risk. Clients may have the GE T TING REPA ID IN A S SE T FIN A NCE 3.1 Ph y sic a l produc t design 24 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management TABLE 4. Risk factors in physical asset design BOX 8. Returns to scale in physical assets Value for Money Customers will intuitively compare the value they get from an asset with what One simple variable—the they are paying. If there is a value deficit, quantity of your assets the risk of default increases. that the manufacturer Durability How often does the product break? produces—drives key Product malfunctions lead to delayed features such as price, warranty and ease of service. or foregone repayments and additional costs of collecting products and Assets produced in high quantity will, all else being swapping them out for customers, not equal, generally have lower unit costs and be more all of which are covered by warranty. readily replaceable. This enables shorter financing Longevity How long does your product function terms and greater customer value. as expected? For borrowers, this The practice of CGAP partner Greenlight Planet determines how soon they will need to purchase a replacement.a For borrowers demonstrates the benefits of scale. Before entering and lenders, it determines what potential into consumer finance, it had already sold millions resale value it may have, either at of solar lanterns and home systems. When the completion of the loan/lease, or at time organization made the decision to develop its own of repossession. consumer financing in 2015, it had already built Dependency What infrastructure does your product “a relentless focus on providing well-designed, need to work? Lack of access to a high-quality, yet low-cost, products for off-grid necessary component (e.g., fuel, power, consumers globally.” This enabled it to keep mobile network) can have an abrupt and negative effect on repayment. financing tenors between 9 and 12 months while still providing affordable prices to customers. Interoperability Does your product link with other products? One trend is for asset companies to create proprietary product ecosystems, ‘walled gardens’ of customers to easily pay for their assets, while also products that only speak to each other. This has advantages: it regulates quality providing a digital trigger for device unlock and clear and creates cross-selling opportunities. data trail for investors and auditors. But it also limits the resale value and • Remote sensing technology like GPS tracking lowers may trap the customer into purchasing higher-priced assets. the risk of asset finance by allowing companies to easily repossess their collateral in the case of default.  ssentially this describes a depreciation schedule, which most a E customers will grasp intuitively. Each of these has the potential, on net, to be beneficial for customers, as without them financing might not be available at all. However, they do raise a raft of consumer willingness to pay for a high-quality product, but that will protection concerns. At a bare minimum, customers need not make up for a lack of repayment capacity (see Box 8). to be fully aware of any technology that could affect their future use of an asset, as well as anything that is gathering CREDIT TECHNOLOGY and sharing data on their use patterns and movements. Technological innovation is one of the key factors enabling More mature industries should consider having an AFCs to reach low-income customers: independent evaluator of products that can conduct • Remote lockout enables flexible PAYGo financing. independent reviews and flag practices that have the potential to disadvantage customers. • Machine-to-machine communication creates huge amounts of analyzable data on consumers. Every piece of credit technology also adds to the overall • Digital payments allow remote and distributed cost of an asset. At regular intervals companies can evaluate what value (such as lower defaults or higher GE T TING REPA ID IN A S SE T FIN A NCE 3.1 Ph y sic a l produc t design 25 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management recovery rates) is brought by individual pieces of ‘credit- 2. Marginal operating expenses (marketing and tech’ and compare that to their cost. A/B testing of key customer acquisition costs) features can help to isolate their impact, for example by 3. Fixed operating expenses (staff, working capital expense) selling a subset of units without the lockout technology enabled to better understand its effect on the probability 4. Expected return (i.e., profit) of default (see Box 10 for details). We would not expect the cash price to cover consumer Credit technology is an important tool. But there is no finance costs, such as risk assessments, credit losses, or the mechanical substitute for properly structuring a credit cost of funding a loan portfolio. However, there are some organization, knowing one’s customer, setting limits on risk-taking, and operating within those limits. Relying solely on technology to manage risk is unlikely to succeed. BOX 10. Lockout effects In 2016 and 2017, CGAP CASH PRICE partnered with Fenix Any asset that can be purchased with a loan or lease International to help scale a should also be available to purchase up-front. The cash digital loan product for price of an asset is also the ‘principal’ (the amount being established PAYGo solar customers. The loan was financed) of an asset-backed loan or lease. The cash price offered to customers who had paid off a solar loan or should be determined by: were in good standing on a current loan and was meant to help them in paying school fees, which are 1. Landed cost of goods sold (COGS) a significant expense in Uganda (see Waldron and Emmott (2018) for details). Crucially, the loan was collateralized using the PAYGo lockout technology: if a customer missed a loan payment, their SHS was BOX 9. Quality without repayment automatically locked. One of CGAP’s asset finance partners carries out In 2018, researchers from UC Berkeley ran an regular customer surveys. As of June 2020, its Net experiment using that loan product. They offered Promoter Score (a standard measure of customer school fees loans to two groups of people: one satisfaction) stands at 75, which is very good. Clients group was offered the loan with lockout, while value the quality and after-sales services of the the other was offered the same loan, without company, according to the customer service manager. lockout. But within the lockout group, a subgroup However, portfolio quality shows a different story, of borrowers were not actually locked out, even for with the share of loans more than 30 days overdue nonpayment. The researchers then monitored the significantly exceeding organizational targets. The repayment patterns of these three groups, as well as vulnerability of low-income clients remains a major risk, a control. with lack of money and logistical issues with payment The results, presented in Wolfram et al. (2021 given as the main reasons for not paying. It is rare forthcoming), show that lockout technology reduced to find clients who do not repay because they are the default rate by 13 percentage points. One third of dissatisfied with the product. The logical conclusion is this effect was due to self-selection: people who knew that these clients have been oversold a product that they would be locked out of their SHS if they failed to they cannot afford. The role of the credit risk manager pay chose not to take the loan. The other two thirds is to ensure that the company does not solely look for of the effect was due to a reduction in moral hazard: scale in credit, which might lead to sales to clients who customers who knew they would be locked out made have the willingness, but not the capacity, to repay more payments, whereas those who could use the consistently for the asset over the pre-determined unit regardless did not pay as much. period, hence raising repayment risks. GE T TING REPA ID IN A S SE T FIN A NCE 3.1 Ph y sic a l produc t design 26 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management costs which could reasonably be allocated to either or both I N T E R E S T R AT E S ‘sides’ (sales or finance) of an AFC. For example, if a sales Interest, or the price paid to use an asset (whether it be agent is responsible for filling out a credit questionnaire, physical or financial), is meant to reimburse a lender for some portion of their commission may be accounted as a the time value of the money lent, compensate them for the finance cost, while the rest would be a sales cost. Other risk they took in lending it and cover any of their expenses examples include but are not limited to: incurred in funding, origination, servicing, or collecting • Fixed operating expenses. Business expenses such as the loan (Chen and Elliehausen 2020). rent or utilities may be incurred to support both sides Every company that charges more for a financed asset of the business. than it does to sell the same asset in cash has an effective • Cost of customer acquisition. Both the finance and interest rate (to learn how to calculate the effective interest sales sides of the business may incur costs to serve rate, see Annex 3). In our experience, this rate is not customers brought in by community demonstrations, always empirically determined by AFCs, and is even more such as fuel and marketing collateral. rarely communicated to customers. This lack of clarity can lead to pricing failures for the provider, where customers • Cost of credit technology. The purpose of lockout end up paying higher effective rates if they repay early (see hardware/firmware is strictly credit-related and should Box 11). At the same time, a lack of transparency on the be allocated to the finance costs. true cost of the loan/lease hurts customers. The importance of this allocation decision is addressed in the discussion of interest rates in the next section. BOX 11. Effective interest rates in practice The following is based on pricing from one of our asset finance partners. Note that although the 3.2 Financial Product Design total cost of credit is lowest for customers taking a 3-month lease, they are paying (by far) the highest Designing an asset-backed loan/lease product is a effective interest rate. In reality, it should be the complicated endeavor. It involves a host of variables, opposite, as the shorter-term lease is both cheaper including cost of goods sold, customer acquisition costs, and less risky to serve. cost of funds, and elasticity of demand. While many of these are independently determined, finance departments 3 months 12 months 30 months can manipulate financing terms to reduce (or increase) Deposit 78.75 70.5 26.7 their risk exposure. This section covers the following (USD) terms, as well as their implications for risk exposure and Monthly 78.75 21.9 11.7 consumer protection: Installment Total Cost 315 333.3 377.7 • Interest rates • Tenor Effective 91.29% 39.43% 25.90% Interest • Flexibility Rate • Deposits GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 27 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Determining interest rates in asset finance Why is it important In MFIs, as in most financial institutions, interest rates to know your interest rate? are determined by a set of four basic factors (Rosenberg Having a transparent interest rate: et al. 2013): 1. Enables customers to understand what they are • Cost of funds paying to purchase an asset vs. what they are paying • Loan loss expense to finance it. This also allows them to compare asset finance with other financing options (e.g., a loan of • Operating expense comparable value from an MFI); • Profit 2. Establishes a benchmark rate of return, an important These factors are equally relevant to asset finance. A variable in evaluating options for financing company’s interest rate should be sufficient to cover the and refinancing a portfolio, as well as assessing costs of funding a loan portfolio, expected credit losses, performance in flexible payment regimes; operating expenses related to its origination, and a sufficient 3. The effective interest rate also serves as the discount profit margin. The EL may vary by market, product, and rate, which is crucial for net present value (NPV) customer segment, which is why risk-based interest pricing calculations, such as those required in recovery can be helpful: it sets an interest rate that matches the analysis (see Chapter 5). historical EL of a similar loan to a similar borrower. Once an interest rate is calculated, an organization can TENOR arrive at the total financed price to the customer. The only The tenor or term of the loan/lease contract determines variables needed are (1) the lease/loan term, and (2) the over how many months the customer is expected to repay cash price. (how long they can take to actually repay depends on the amount of flexibility they are given). The loan/lease tenor As shown above, determining the latter can be a challenge. has a negative relationship with the individual payment Most AFCs effectively run an asset business and a finance size, and a positive relationship with the overall cost of business. Allocating all of the shared ‘costs’ to the product the loan/lease (see Figure 6). This dynamic has important side will make the cash price more expensive and interest consequences for credit risk. rates appear lower. Doing the reverse (allocating all costs to the finance side) will make interest rates appear higher. Imagine an asset with a cash price of $100, financed by a company at a 48 percent annual interest rate that is In general, we do not observe many companies allocating compounded monthly, for which the customer makes costs between ‘sales’ and ‘finance’ to arrive at an effective, monthly payments. risk-reflective interest rate. In fact, there is often no internal division between these units, and no solid • If the tenor of that loan is 12 months, the customer justification for the relation between cash and lease prices. must pay $10.24 a month, or $112.14 in total. These problems are addressed in depth in Sotiriou et al. • If the tenor is increased to 24 months, now the (2018), as is a potential solution: providers can go through customer only pays $6.11 a month, but the overall cost a cost allocation exercise that yields a clearly defined cash has increased to $146.66, as their interest expense has price. This, in turn, will allow them to define appropriate, quadrupled. risk-based interest rates. On its face, this may seem a simple choice for companies working with low-income customers: stretch the loan out as far as possible, make the monthly/weekly/daily price as low as you can, and reach as many customers as possible. GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 28 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 6. Relationship of tenor, monthly installments, and overall cost of ownership $120.00 $107.00 $106.00 Overall cost of ownership $100.00 $105.00 Monthly Installment $104.00 $80.00 $103.00 $102.00 $60.00 $101.00 $100.00 $40.00 $99.00 $20.00 $98.00 $97.00 $0.00 $96.00 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 Principal Total Interest Monthly Installment But companies that do so are significantly increasing their FIGURE 7.  he pros and cons T credit risk, in ways they may not always appreciate or price of increasing loan tenor in (see Figure 7). These include: • Technical obsolescence. Devices such as smartphones A S LOA N TE R M S LE N GTH E N get better every year. Newer, cheaper assets could lead customers to stop paying for older models. PROS CONS • Asset installments • Overall costs to • Technical failure. Increasing tenors put ever more become more customer go up pressure on the devices themselves. If they break, affordable • Credit learning the costs of replacing or repairing the asset—even if • Sales increase cycles take longer covered by warranty—may make the loan unprofitable (Waldron 2020). • Greater risk of technical failure/ • Income reduction. Many low-income customers obsolescence have seasonal or irregular incomes and are vulnerable to financial shocks (e.g., a lost harvest or family injury) which may reduce their ability to pay. The probability of There is also a knowledge cost to lengthening tenors. Credit such events, and therefore of default, increases with time. risk management is an iterative exercise: companies need • Fatigue. If customers get bored with an asset or tired to go through multiple loan cycles and analyze the results of making regular payments, their discipline to before they can perfect their models. Shorter loan tenors sacrifice and keep paying may erode. allow for faster analysis and adaptation, which in turn leads to better performing portfolios and stronger companies. These combine to increase the probability of default and In general, loan tenors ought to be kept as short as therefore the cost of risk. This ‘risk premium’ should be possible, but as long as necessary. Setting a loan term is an fully priced in, but also must be accompanied by strong important decision that determines the installment price, credit processes. In our experience when risk is priced in and each company must make its own choice bearing in without controls there is no cap on the risk premium, and mind their risk appetite, asset attributes and target client. it can rapidly increase, forcing good clients to pay ever- higher prices on behalf of the bad. GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 29 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FLEXIBILIT Y One of the much-discussed innovations in asset financing BOX 12. How PAYGo solar works in recent years has been the emergence of flexible, PAYGo Fictional company SolarCredit sells a PAYGo SHS financing (see Box 12 for an example). This has added a that is priced at $115, which includes a $25 down new and complex tool to the asset finance toolbox, and payment and $90 to be repaid over 12 months (the another decision point when designing a financial product. cash price is $80). Over 365 days, the PAYGo price is In this section, we explore the implications of traditional roughly $0.25 per day. As in most PAYGo models the and flexible approaches to financing assets. $90 amount is fixed; interest does not compound, nor are late fees tacked on for slow repayment. Traditional Financing Now imagine a customer pays $2 on June 1, which Traditional microfinance providers do not generally offer is Day 1 of their loan. This amount equates to 8 days flexible loan terms. Rather, payments are expected on a ($2/$0.25 per day = 8 days), which is how long their given date (e.g., the first day of every month), and missed SHS will be unlocked. At the end of Day 8, their payments are taken seriously: purchased ‘energy days’ elapse, and the unit shuts 1. If a client misses a scheduled payment by a few days, off. This customer, for whatever reason, does not pay again until June 12 (Day 12), when they again then pays, they were delinquent for the period of make a payment for $2. nonpayment. With PAYGo, the three missed days (June 9-11) are 2. If payments are missed by more than 30 days, the loan not treated as ‘days in arrears’ and are not expected begins to show up in Portfolio at Risk (PAR) metrics to be ‘made up’. Rather, the expected end date of (discussed in Chapter 5), and the financial institution the contract is implicitly rescheduled (extended) with will begin provisioning for higher losses. The loan each missed payment. On June 11 in our example remains ‘at risk’ until all missed payments (arrears) are the customer has paid for 8 out of 11 days since the repaid, and the client is current on their repayments. inception of their contract, and they have 357 days remaining to purchase (365 minus 8), meaning their 3. At 90 days late, the lender may begin legal action and expected completion date is now June 3rd of the attempt to recover collateral. next year (three days later than initially scheduled). 4. At 180 days late, the loan may be written off and sold After a bit more time has elapsed, SolarCredit to a third party for collections. should be able to determine the repayment profile for this customer and calculate an expected Many AFCs, particularly those financing larger or more completion date and a probability of default based commercial assets, also employ this approach to financing. on historical experience. They expect regular payments on scheduled dates, and if these payments are not made, it triggers actions similar to those described above. When these companies use lockout technology, the asset is normally only deactivated after some time has passed from the missed payment (15–45 Flexible financing days). For assets that generate income (i.e., productive Companies financing smaller assets, such as solar home assets), this grace period allows clients to generate income systems and smartphones, are increasingly using more to use for repayment. flexible terms, such as PAYGo financing (see Box 12). By linking payment for an asset-backed loan to use of that asset, PAYGo allows a borrower to repay a flat loan amount in flexible installments over time. The flat amount includes the price of the asset and interest. Units are deactivated automatically for nonpayment, but arrears do not accumulate. This conforms better with GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 30 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management TABLE 5. Advantages and disadvantages of flexible payments and fixed payments Advantages Disadvantages Fixed Prioritizes discipline and confers clear penalties •  People working in the informal economy may struggle •  for non-payment. with fixed monthly payments. • Easier to raise liabilities to fund fixed-term loans. Fear of missing a payment may scare off the most •  Simpler to monitor on an individual and portfolio •  conscientious borrowers. basis. If payment is linked to use of an asset, then clearing •  Risk and return are aligned: customers who •  arrears becomes financially difficult, as the user would repay faster (i.e., lower risk and higher return) have to pay for usage they never actually get. pay less in interest. Compounded interest can make it confusing for •  borrowers to understand the total amount they will be paying. Flexible Conforms better to the ‘lumpy’ unpredictable •  • Does not require repayment discipline. incomes of borrowers working in the informal Difficult to know, until a large number of loans have •  economy (Collins et al. 2009). been repaid, how long it will take to recover capital. Easier for borrowers who have missed a few •  • Harder to identify risky customers. days to resume payment, without paying an • Ignores the time value of money: additional penalty. Repayers who finish their contract early often pay ·  Flat amount to repay is more intuitive for •  much higher effective interest rates, even though borrowers. their risk is lowest and their return for the company Pre-payment for use of an asset is a familiar •  is highest.a behavior (e.g., mobile airtime). Slower repayers who take longer than the nominal ·  Can allow poorer borrowers to lower the •  tenor create opportunity cost for the company, as effective cost of owning the product the outstanding capital cannot be redeployed to (see Box 12). finance new assets. t is standard practice to offer discounts to early repayers. But such discounts do not usually discount by the full time value of money (which in this case a I is the effective interest rate on the loan/lease). low-income households’ income patterns, but can create Risk pricing uncertainty around average repayment time, particularly One way for an AFC to mitigate credit risk is to price its for new companies. Table 5 compares flexible and fixed EL into the cost of a product, the interest rate, or both. repayment schedules. On a portfolio basis, EL is the product of the outstanding Flexible or fixed? obligations, default probabilities and final loss rates. These The implications of PAYGo financing on portfolio analysis can be derived and calculated using analyses discussed will be discussed in more detail in the next chapter. But in Chapter 5. Once the EL is known on a portfolio basis, no matter the amount of flexibility offered, the time value lenders can use this to estimate more granular EL on of money concept still applies, whether in calculating specific products or customer segments. These estimates lease installments or in pricing of recovery streams are used to set the price of the loan or lease, with interest (e.g., from re-sale of repossessed items). That concept rates or margins adjusted to cover EL (see Box 13 for a should be applied to flexible payment schemes, as well. simple example). Assuming that benchmark interest rates are positive, a Risk pricing is an industry norm for traditional lenders top-up payment made earlier is worth more than the like banks and MFIs. But CGAP’s experience indicates same payment made later—whether a later payment is that AFCs are sometimes averse to accurately pricing in contractually allowed or not. When payments are made default risk, given low-income customers’ sensitivity to matters, so it follows that earlier repayers should pay less. price. While keeping prices low may promote higher sales and faster growth in the short-term, if pricing does not adequately compensate for risk then inevitable write-offs GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 31 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management will eat into profit margins and threaten the long-term viability of companies.  ricing BOX 13. P in expected loss Although the overall EL does need to be priced in, Fictional company SolarCredit wants to sell a new this can be done in a nuanced, segmented manner. If entry-level PAYGo SHS for $100, which would cover a company has sufficient historical data and a robust the cash price, cost of funds, operational expenses assessment process, they may be able to put potential related to finance, and margin. But that price customers into ‘less risky’ and ‘more risky’ segments. Once assumes no losses to credit risk. Analyzing the most this is done, they can price in the EL for that segment recent data for similarly priced assets, SolarCredit through an increased deposit, higher interest rate, or arrives at an EL of 10 percent for the product. some combination thereof. Alternatively, companies could To maintain the same margin, SolarCredit increases keep the price constant across customers, but change the the price by 11.1 percent (or 1/(1–10 percent)) for all percentage that must be received upfront. customers. The product will now sell for $111. This is a Risk-based pricing, where some customers pay more than simplified example: increasing prices will also decrease others, was generally perceived by our AFC partners as demand, which needs to be taken into account. problematic, and likely to raise issues when customers discovered they had paid more than their neighbors. However, risk-based pricing is important in enabling Recognizing this, AFCs also have a responsibility to financial service providers to lend to poorer, higher- ensure that: risk communities (Klein and Okagaki 2018). Providers • They don’t over-indebt clients. should begin experimenting with socially acceptable ways • They minimize both the installment and the total price of pricing in higher risk, or they may find themselves that clients pay. gravitating toward less risky populations overall. • Clients understand the full pricing structure of what they pay for, including interest rates. AFFORDABILIT Y AND SOCIAL ENTERPRISE Much of the discussion in this section has focused on • They lend to clients that have the capacity to pay pricing, which is an important—albeit passive—way regular installments for the full term of the loan (this is that AFCs manage risk. Socially-minded AFCs have a particularly true of PAYGo, with its longer tenors). responsibility to keep prices as low as they can in order to In summary, there are always ways to make an asset make their products affordable to people who need them. cheaper. Yet some of these are more sustainable than This is one reason why some companies told us they did not others. Cutting costs on risk management or financing invest heavily in credit departments or conduct thorough assets for too long may lead to short-term boosts in loan credit assessments: these would cost too much, making the origination. But based on our experience, these loans are overall price too high for poor customers. Other companies far less likely to turn into actual cash. And in the case took an opposite approach, using longer tenors to spread of PAYGo models, this means more customers locked their high-cost structures over several years. out from their assets and less impact. Companies need Still, it is important to acknowledge that, at the end to consider and balance all of these considerations when of the day, affordability is the product of a long-term designing products for their target customer. emphasis on minimizing costs, including the cost of risk. Credit risk management can reduce the risk premiums that companies charge to compensate for high EL. More importantly, proper risk controls can reduce the volatility of an organization’s losses. And it is those unexpected losses that can overwhelm price buffers and damage future profitability. GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 32 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management CH A P TER 4 CREDIT TR ANSACTION RISK T SUMMARY HIS CH A PTER FOCUSES ON credit transaction risk, the credit risk arising from an individual customer’s repayment obligation. • Credit risk needs to be managed throughout Credit transaction risk is measured by (1) the standalone each step of a credit transaction, from the probability that the borrower will not be able to meet their moment a potential customer meets an agent obligation as contractually agreed (probability of default), until they have made their last payment. and (2) the ultimate loss in the case of a borrower default • Compensating agents on sales volume can after depreciation of collateral, repossession costs, and increase credit risk exposure. Taking steps other factors have been accounted for (loss given default, to involve other parties or limiting their or LGD). More experienced risk managers will recognize compensation for poor-quality sales can help the EL framework here in miniature. We explore this balance this. framework on a portfolio level in the next chapter. • Assessing the ability and willingness to To understand credit transaction risk requires going pay for low-income, informal customers through each step of the credit life cycle and identifying is the greatest challenge in asset finance actions that can avoid or mitigate this risk. The credit life today. Collecting large amounts of data over cycle contains the steps shown in Figure 8. multiple loan cycles is a proven method for scoring clients but may not always be This cycle changes only slightly in asset finance, for which practical for smaller assets. Companies need ‘disbursement’ requires the delivery of physical goods, to experiment with approaches that keep some level of technical instruction, and often a physical credit risk within stated appetites. installation. Also, the ‘collections’ process in asset finance can vary significantly from other types of credit. • Repossession (remote disconnection through lockout technology) is a unique tool in asset As mentioned in Chapter 2, lenders need to start by finance. Companies need to make their developing clear policy documents and/or working collection and repossession policies clear, procedures covering transaction risk at each step in the then follow through on them while treating credit life cycle. These policies and procedures should customers respectfully and with care. clearly outline staff roles, responsibilities and tasks, and should be well-implemented in practice. Likewise, rules for handover of responsibility from step to step should be GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 33 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 8. Credit life cycle In addition to each of these elements, this section also covers two important areas that fall outside of the formal credit cycle. The first is the most over-arching: consumer protection. This is broadly defined as the group of laws, procedures, and standards that safeguard buyers Marketing against unfair business practices. There are a number of Credit useful consumer protection frameworks that companies Assessment may commit to, including the SMART Campaign13 and GOGLA’s Consumer Protection Code.14 Yet for consumer protection to be effective, it must be internalized by a Credit Life company and operationalized at each step of a customer Collections Cycle Decision- interaction. Increasingly, financial institutions are making & explicitly integrating consumer protection as part of their Disbursement risk framework by setting up roles and responsibilities for conduct risk management. In that spirit, consumer Monitoring protection issues will be brought up throughout this & Repayment chapter. The second is operational risk management. This plays an important role in driving or mitigating credit risk for AFCs, but is a wholly separate process whose full consideration is beyond the scope of this guide. However, it formalized and expected timelines for each stage should be is referenced here in regard to its implications for credit risk. understood and benchmarks set. In summary, our full view of credit transaction risk in In short, it should be clearly stated and widely understood asset finance is illustrated in Figure 9. for each stage of the credit cycle: • What is the task? FIGURE 9.  redit life cycle plus consumer protection C and operational risk • Where is it happening? Consumer Protection • When is it happening? • Who does what task? Operational Risk • Why is it being done? • How is it being done? • How long is it expected to take? Marketing Credit Assessment • How does the customer transition from one step to the next? Credit Life • What information has to be given to the customer, and Collections Cycle Decision- making & when? Disbursement • Who is responsible for initiating and completing that Monitoring & Repayment transition? • Who has the authority to complete/approve that step? • What documents or records in the IT system should be created at this stage? 13 See https://www.smartcampaign.org/about/smart-microfinance-and-the-client-protection-principles 14 See https://www.gogla.org/consumer-protection GE T TING REPA ID IN A S SE T FIN A NCE 3.2 Fin a nci a l produc t design 34 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management This chapter is organized around this framework, with each 6. How the credit/application process will be conducted step along the journey covered in the following sections by the institution, to set clear expectations from the onset. (several of which have been condensed for readability): Contracts, as well as all of the marketing material above, • Marketing, Origination, and Sales should be written in easy-to-understand language (in the • Credit Risk Assessment customer’s native dialect) and gone through in detail with the client. Illiterate customers should be given special • Decision-making and Disbursement assistance, such as a video explaining the contract terms. • Monitoring & Repayment The SMART Campaign as well as GOGLA provide good • Collections reference frameworks for customer care. Transparency and clear communication are essential because customers who are sold an asset they cannot 4.1 Marketing, Origination, afford, or who do not fully understand the terms of a contract, are more likely to default. Ensuring and Sales transparency and understanding is part of the credit risk The first step in the credit cycle is client attraction, or management process. marketing. Defining a target clientele, who may be In the absence of clear communication, customers may segmented into different profiles, is essential. For example, get the impression that the transaction is similar to a an institution may target low-income salaried employees, cash purchase. recipients of remittances and/or small businesses. Therefore, the credit component, respective obligations Specific decisions could also be made to target women or and possible consequences in case of not meeting them customers in a certain age group. This exercise is closely (e.g., negative credit bureau entry, penalty interest, and linked to the setting of a company’s risk appetite, as target repossession), must be explained as transparently and clientele is also determined by a company’s sales channels, simply as this: “You are signing a commitment to pay X at mission and product offerings. (See Box 3 earlier for a Y intervals over Z period. If you do not pay, these are the description of one sales process). consequences.” In some cases, this frankness is avoided Regardless of the marketing procedure, each customer on purpose to not “scare the client.” This is misguided. must receive the following information prior to a purchase Transparency is even more important in the case of agreement: flexible PAYGo schemes, where customers may easily get 1. Their contractual responsibilities, the consequences the impression that they only need to pay when they want should they not meet them, as well as their rights and light or irrigation. mechanisms for lodging complaints.15 A G E N T D Y N A M I C S A N D C O M P E N S AT I O N 2. A general technical introduction, including necessary Agents are the first, and sometimes only, personal maintenance requirements. contact that a customer will have with an AFC. They 3. Information regarding their warranty and technical play a major role in mitigating credit risk, as do their support. incentives. AFCs often work with non-salaried agents 4. A payment schedule, highlighting the portion of who sell their products and originate loans in exchange principal, interest, and fees/commissions, including for a commission. any late payment fees. Many firms we worked with use bonus systems to 5. Clear communication of the effective interest rate. incentivize asset sales. These bonuses can be a key driver 15 Each institution has also to establish an efficient complaint resolution process, which should not only define responsibilities but also standard response times. GE T TING REPA ID IN A S SE T FIN A NCE 4.1 M a rk e t ing, origin at ion, a nd s a l es 35 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management or mitigant of credit risk. For example, incentives that 2. Portfolio quality. Good repayment by an agent’s reward agents only on the quantity of loans originated will clients should be rewarded, while poor repayment motivate them to simply ‘move product’ without regard should lead to a reduction in commission. This creates for whether a client can repay their obligation. But if an incentive for agents to follow up with delinquent agents are asked to assess the ability and willingness of a borrowers and encourage repayment. potential client to repay the loan/lease, a significant part of 3. Clear link between action and payment. The agent the incentive should be tied to that actual repayment. should have no doubt regarding which outcomes Most companies recognize this and structure their will lead to compensation, and which will result in a commissions to encourage high-quality sales. Others reduction. choose to limit agent responsibility to generating leads, 4. Frequent and timely payment. Companies will want leaving the responsibility for credit assessments to staff to make payments frequently and on time, so that who follow up with interested customers using either a call agents do not stop traveling or selling. center or in-person visit. Bonuses related to quantity and quality of loans originated usually make up a portion of 5. Fraud. When developing an incentive scheme, it is those staff’s compensation. important to actively think about adverse effects/ misuse. How would you play or trick this scheme if For companies that choose to incorporate agents into you were an agent? their credit assessment, there are various schemes that can help to incentivize quantity and quality of sales. Each 6. Involvement. Agents should be given the have their own strengths and weaknesses. A well-designed responsibility of interacting with the customer when incentive scheme should consider the following issues, payment delays occur up to a certain threshold (e.g., as well as the trade-offs involved in emphasizing certain 30 or 60 days). considerations over others: Box 14 describes two approaches we encountered in our 1. Portfolio growth. Early-stage companies will want work. There is no single ‘right’ commission scheme, but to grow the overall number of clients. This can be companies should always consider the principles. Regular done by compensating agents for each new loan/lease trainings for agents, combined with well-established originated. control structures that allow managers to identify and  ample BOX 14. S agent compensation schemes  ictional company SolarCredit agent receives a) F The overall quality of an agent’s past sales b)  50 percent of the commission when the sale is determines the amount of commission they receive made, and the rest follows when certain customer from their next sale. An agent with a collection rate milestones are met (25 percent after the first >90 percent for their ‘portfolio’ receives $20 for a installment, and 25 percent after the second new sale. Another agent with a collection rate <70 installment). This scheme is meant as an incentive percent receives only $15 for a new sale, and a third to focus on financially healthy and trustworthy <50 percent receives only $10. customers. However, it pre-supposes that an agent This scheme incentivizes portfolio quality and is harder has more potential customers in their pipeline than to manipulate. But it requires high levels of agent they can actually serve—something which we rarely training and support, as the link between sales activities observed. Otherwise, the agent may still try to and received/unreceived bonus payments is harder to acquire every customer they can, hoping that each is track. Agents with rapidly deteriorating portfolios are going to be a “good” client. also unlikely to see an advantage in reaching out to clients, which could accelerate deterioration. GE T TING REPA ID IN A S SE T FIN A NCE 4.1 M a rk e t ing, origin at ion, a nd s a l es 36 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management address any problematic agent behavior early on, are also The core assumption is that objects are being sold critical to protecting customers and managing credit risk. in-person, and that at least one interview of a potential borrower will be conducted. More complex assets may require additional interviews, home visits, technical 4.2. Credit Risk Assessment assessments, and/or data requests. Before any business sells a product to any client, it W I L L I N G N E S S T O PAY A N D A B I L I T Y T O PAY wants to know if it will be paid. AFCs are no different: In asset finance for low-income customers, a credit they want to sell assets to clients who will pay them. A assessment is meant to establish two things: credit risk assessment is meant to evaluate a potential borrower’s ability to repay the obligation, their character 1. Willingness to pay. Is the customer willing to meet and willingness to repay, and any risks that may endanger their contractual obligations? repayment. This assessment tells the company how likely 2. Ability to pay. Are they able to do that? they are to be paid, which should inform the decision to Where one or both of these elements are lacking, default is finance a product or not, as well as how they will price the likely to occur. Under the constraints faced by low-income risk on that loan/lease. or poor households, willingness and ability to pay are often In high-income countries, consumer credit assessments are often automated and rely heavily on credit reference bureaus. However, AFCs can rarely rely on these tools: even when reference bureaus are operational, few  ifferent BOX 15. D assessments customers will have a record to analyze. In microfinance, for different assets creditworthiness is usually assessed through in-person upOwa finances solar interviews, reference checks and other high-touch products for customers interactions. But this often conflicts with AFCs’ desire to in Cameroon and keep costs low and use staff/agent time efficiently. employs a tiered AFCs must strike their own balance between mitigating approach to customer assessment, offering three the risk of nonpayment and minimizing the cost of the different SHS products of different capacity and mitigation. Often, the approach they take is heavily pricing. Only the “premium” product requires a influenced by the size and type of asset being financed. detailed customer survey, the outcome of which is used to fill out an expert-based scorecard. upOwa The cost of more thorough assessments may be hard to sees smaller products as an entry point for new justify for small assets with a low profit margin, while customers who, if they repay, will have proven higher-value income generating assets and consumer goods willingness and capacity and thus qualify for may require different approaches to credit assessment (see additional financed products. At the moment, upOwa Box 15 for an example of a tiered system). just collects KYC information for these products; in This section focuses on the key principles for conducting the future, they plan to develop an automated credit risk assessments in asset finance. It covers: risk-based scoring system for smaller units. • Willingness and ability to pay This proportional approach is found in microfinance as well. Customers who have successfully repaid • Credit scoring, and the difference between expert and their first, second, and third loan typically show low statistical scoring defaults on subsequent loans. Such a graduated • Methods for streamlining assessments and verifying approach adds a behavioral component to the information assessment, replacing the credit bureau with the company’s own experience. • The importance of secure data gathering and storage GE T TING REPA ID IN A S SE T FIN A NCE 4.2 Credi t risk a s ses smen t 37 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management closely intertwined. There is rarely a month in the life of that it necessitates rejection. Staff and agents should be a low-income borrower where each and every financial clear on the policies and their importance, as well as the obligation has been met and significant money is left over. consequences of not following them. This is what being poor means. Even if one accumulates Successful lenders can also try to cultivate willingness actual cash savings, this is the result of making real to pay among their target clientele through financial sacrifices in order to build up an emergency buffer. education. This can serve multiple purposes, as If having the ability to pay means that clients reach each highlighted in Box 16. loan payment date with all other bills paid, all household needs met and a few hundred dollars emergency cash left over after making the loan payment, then most low- BOX 16. Tugende and driver education income borrowers would be excluded. Therefore, ability to pay almost always comes down to the discipline required Tugende finances to prioritize a loan installment over many other competing motorcycles for taxi needs and wants. In other words, willingness to pay. drivers in Uganda and Kenya. Their waiting Assessing willingness to pay list is long and understandably so — they enable For this reason, an important factor in assessing the drivers to own assets that they have rented all their credit risk in consumer loans is determining the moral life. One of the ways that Tugende assesses character of the client. Moral integrity serves as a proxy potential clients’ commitment and for the willingness to pay even in the face of hard personal creditworthiness is by asking them to attend choices. This is both the most important factor to assess, educational sessions at the beginning of the and the most difficult. application. These sessions serve multiple goals: There are many proxies for willingness to pay that an agent or credit officer can try to assess in the credit Educate the client about Tugende’s rules and 1.  decision process. These could be factors that indicate procedures. stable life circumstances and responsible behavior: being Help provide the financial education necessary 2.  married and caring for children, more life experience, for drivers to avoid overindebtedness. living in a rural area with deep community links, are all  elp to discipline the client by explaining 3. H good for stability. repayment rules and penalties. Having some assets or owning land or a home can also 4. Provide the guide to safe driving and assets. be predictive of responsible, disciplined behavior. The question: “Do you own or rent the home you live in?”, Serve as a filtering device. If potential clients do 5.  not have the discipline to attend and absorb an provides helpful indications. If a poor household at least education session, they likely will not have the owns the modest shelter they live in, that is an indicator of discipline to pay for a motorcycle for over a year. industriousness, discipline, responsibility, personal pride, and an aspiration to better one’s life. These are often good 6. Help to build a trustful relationship. predictors of honest and reliable borrower behavior. Given the subjective nature of this assessment, it is vital that an organization’s criteria should be as comprehensive and standardized as possible (and put in writing) to avoid the agent taking a decision/assessment based on their personal bias. Nor should any one factor be unilaterally disqualifying, unless the failure is so uniquely egregious GE T TING REPA ID IN A S SE T FIN A NCE 4.2 Credi t risk a s ses smen t 38 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Measuring ability to pay That does not mean that the ability to pay is not  onsumer BOX 17. C protection important when conducting credit risk assessments. Just in client assessment because the ex-ante ability to pay is typically already One of the main goals of all common financial marginal, this does not mean that it could not get worse. industry customer protection frameworks is to If ability to pay goes from marginal to impossible—if the not overburden clients with debt. And one way cash simply is not there—then even the most willing and that providers can protect against over-indebting disciplined borrower will default. customers is by conducting a customer affordability assessment. Collecting income-related data that can be used to assess ability to repay is inherently difficult. Informal worker The importance of credit assessments is discussed here from the institution’s point of view, income is often irregular and comes from a number of but the same is also relevant from the customer’s sources, and as a result many customers struggle to provide standpoint. Correctly assessing a client’s payment reliable estimates of their earnings. In the absence of other capacity does not only prevent the provider from methods for calculating income, self-reported estimates losses but also the customer from becoming should be subjected to automated verification rules that flag, over-indebted. Social enterprises would do little for for example, when reported income is less than reported their social goals by signing clients to obligations expenses. This can help agents or credit officers spot issues that they are unable to fulfill. and dig deeper before making a decision. The first customer contact should be always used Assessing ability to pay is not just about credit risk to explain the purpose of the survey—in other management. It is also about impact: this is the critical words that the information collected will be used step that companies must take to avoid over-indebting in the credit assessment process and to ensure clients (see Box 17). that the terms of the loan/lease work for both the customer and lender. But often an asset-based approach is easier to implement and more reliable than income figures. For example, home visits can be used to check on ‘grades’ the potential borrower on a number of pre-set living conditions, the existence of motorcycles, TVs, criteria (asset holdings, income, character, references, etc.). smartphones, etc. The Poverty Probability Index (PPI) is an example of such an approach, and offers some • Statistical models, where data is captured on various promise as a tool in credit scoring. indicators, then an overall score is calculated based on historical predictors of customer default. CREDIT SCORING AND EXPERT These two are not mutually exclusive. Both can be used in V S . S TAT I S T I C A L S C O R I N G M O D E L S credit scoring. For ease of decision-making, most companies deploy a scoring model or ‘scorecard’ that enables various factors Expert-based scoring to be standardized and tabulated. These scoring models For early-stage companies, an agent or credit officer (the do not determine disbursement, but usually act as a factor ‘expert’) will typically collect data and calculate a score in decision-making. However, the inputs and structure of based on their impressions and experience. This score that scorecard vary based on the maturity of the company, and the collected data should be checked by at least one the nature of the asset, and the target client. centralized credit analyst who is not financially motivated In general, there are two types of scoring models used in (an example of the Four Eyes Principle at work). credit risk assessment: As companies grow and collect more data, they can begin • Expert-based models, where an agent or loan officer to run regression analyses on customer data to identify patterns and relationships between variables. This can be GE T TING REPA ID IN A S SE T FIN A NCE 4.2 Credi t risk a s ses smen t 39 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management used as a plausibility check, revealing dependence between not an end state, but rather a process. When done right, variables like family income and asset ownership, which a good scoring model typically improves its performance in turn can flag applications where reported values deviate with each loan cycle observed. Therefore, having an from past experience. efficient, secure data collection process is the pre-condition to develop a statistically proven algorithm in due time. Statistical credit scoring In other words, both quantity and quality of data matter, Once sufficient data has been collected, companies can go as does patience. Lastly, while reliance on algorithms and further and run regressions comparing customer data to machine learning can increase with time, it is important portfolio data in order to isolate variables that are highly to first develop internal capacity and set up strong predictive of repayment behavior. This should, over time, management and governance frameworks. Without these enable them to build a statistical scoring model that can be structures in place, technical innovations will not deliver tailored to their specific context and continuously improved their full value. upon as more and more data is fed into the model. Statistical scoring requires the company to: D ATA 1. Securely collect a wide range of variables (quantitative, Credit is a data-driven activity, and so the importance of qualitative, and/or psychometric, although not a well-elaborated data strategy cannot be overemphasized. necessarily all three together). Establishing a thorough data collection strategy, requires answers to the following questions: 2. Store data in a well-maintained customer database. 1. Which data to collect? 3. Do so for a large enough sample of cases (rule of thumb >5,000, but the more the better) to be representative. 2. Which format – e.g., date, integer, string from dropdown, etc.?17 4. Gather data for long enough to observe outcomes (i.e., complete their loan cycles or default). 3. How can data entry be managed to minimize the need for future cleaning? For example, managers can 5. Hold key loan features more or less constant. add rules to date of birth fields that reject any entry Developing an in-house scoring model requires a falling between 18 and 75 years prior to the date of significant initial investment, including long interviews entry (i.e., a potential client must be between 18 and and in-person visits. However, this investment can 75 years old). have a significant long-term return. A good scoring 4. How to collect data and when? model can reasonably predict the likelihood of default, limit the number of data points collected to only the 5. How to verify data? most predictive, enable remote assessments (e.g., phone 6. How to protect personal and sensitive data? interviews, submission of pictures), and automatically flag When prioritizing the data to collect on customers, inconsistencies that require physical verification. Statistical assessors should first focus on data that are minimally scoring is also an important use case for machine learning. invasive, can be easily collected and are difficult to falsify. Though the developed algorithms may differ, the same Socio-demographic indicators are a good example of such methodology is generally applicable regardless of context.16 data, including age, gender, number of children, and time It is up to each institution whether to use their algorithm spent in an area. as the main decision-making tool or simply as a A provider can and should collect the same data points supplemental input (e.g., in addition to a basic financial for each customer, and they should do so directly into analysis), to come to the final credit decision. Scoring is a digital system with a forced set of responses for each 16 E.g., collection scoring, predicting follow-up loans, measuring the success of marketing campaigns, etc. 17 Free strings are not recommended, as they are hard to code into modalities later on, although they are sometimes necessary, e.g., name and surname. GE T TING REPA ID IN A S SE T FIN A NCE 4.2 Credi t risk a s ses smen t 40 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management question. Over time this will result in a comparably large These concerns are important, but they can be partially database, showing the same elements for most customers. addressed. The following strategies are employed by Such a database is a key resource for data analytics effective financial institutions in order to gather customer and data-based risk management tools, and should be data and verify its accuracy: accompanied by a data dictionary that describes each • Have someone besides the selling agent conduct the variable, its derivation and its intended use (see Annex 4). data gathering. Ideally this would be a credit officer Typically, data storage capacity is not a limiting factor, conducting an in-person visit; more likely it will be an especially not in IT-driven companies like many newer analyst on the phone. AFCs, who typically rely on cloud storage. Most of the • Ask borrowers for both references and guarantors. The assessed companies had a strong IT environment in place, former will provide insight into their moral integrity. enabling them to record and analyze complex datasets and • Review credit bureau records (if available), allowing visualize them in dashboards and trends. providers to check not only customers’ repayment history but also their truthfulness about their G AT H E R I N G A N D V E R I F Y I N G D ATA self-reported loan history. Many AFCs in our work were skeptical of the value of self- • Incorporate logic checks into the scorecard (see Box reported information from customers. This skepticism is 18) which compare different information points and understandable: income and expenses, details about one’s examine for plausibility (e.g., stated income minus family and lifestyles—these are sensitive topics for anyone. stated expenses should roughly match the stated saving Executives at companies were also worried about agents rate, which should match household assets, etc.). ‘coaching’ clients to give ‘correct’ responses to questions. BOX 18. Verifying credit information For example, during the initial interview customers are asked about the crops they grow, and in what acreage. PEG Africa is an AFC in West Africa. Using government data on average crop yields and CGAP’s work with PEG Africa prices, we were able to build plausible estimates revolved around designing a credit for expected income from an acre of a given crop. approval process for larger assets, Significant deviations from those estimates (e.g., a such as solar water pumps. This tomato farmer who is also a self-declared millionaire) included three stages of assessment: will raise follow-up questions. 1. Initial sale and interview Similarly, the on-site visit will enable PEG to assess 2. Technical assessment on-site the client’s assets visually. When combined with PPI 3. Follow-on credit interview over phone data, this can also hint at income levels. If the on-site visit reveals a different living standard, or different farm The multiple interviews, including on-site visit, not only conditions, this could affect the ultimate score. allow PEG to build up a firm view of the client’s living situation and asset holdings. They also provide an The use of available data, multiple interviews and opportunity to verify answers given to prior questions. interviewers, and on-site visits are all helpful. But the model will still require multiple iterations and loan cycles to get right. The best time to build it was five years ago. The second best time is now. GE T TING REPA ID IN A S SE T FIN A NCE 4.2 Credi t risk a s ses smen t 41 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Importantly, the algorithm should not be a black box. In CGAP’s experience, companies leasing larger assets Clear policies are needed for updating, evaluating, and (e.g., vehicles) manage to filter out many unqualified re-calibrating the algorithm, all of which consider the applicants through their process, and still reject a actual outcomes produced and their impact. However, significant portion of finished applications. On the other its internal logic should not be too widely revealed to hand, companies financing smaller assets tend to have frontline staff, nor every question equally weighted. A bit looser criteria: the majority of potential customers who of mystery goes a long way in reducing potential ‘gaming’ can pay the deposit are approved. of the algorithm. Credit decisions should be based directly on the risk One way of doing this is to include several psychometric appetite and tolerance of the company. However, questions (e.g., “how often do you think suppliers cheat companies should not overestimate their ability to control their buyers by overcharging or under-delivering?”) into repayment behavior post-decision. The most important the questionnaire. Metadata (e.g., how long did a customer place to manage credit risk is before it begins (in other take to answer a given question) should also be evaluated words, pre-disbursement). Practically, this means that not as part of the assessment, as it can be highly useful and every person who wants an asset should receive one on may have predictive power. credit. A large part of risk is knowing when to say ‘no.’ That said, companies who are loath to reject willing STREAMLINING THE CREDIT ASSESSMENT customers may want to consider risk-pricing their deposit. For smaller asset classes, a process such as that shown in Asking those clients who are assessed to be riskier to pay Box 18 may not be possible or even necessary. Indeed, 30 percent upfront as opposed to 20 percent, for example, PEG does not follow this same process for all clients - may serve three purposes: (1) decrease the probability of lower-value assets require fewer interactions. Providers default by filtering out higher-risk clients, and (2) reduce the should explore ways of conducting assessments and exposure at default (EaD) for those who still want the asset. gathering KYC data remotely and make use of local referees wherever possible. M E C H A N I C S O F D E C I S I O N -M A K I N G A given staff member will generally be empowered to make credit decisions up to a certain limit (the more 4.3 Decision-Making senior the staffer, the higher the limit). Significant and Disbursement exposures may require a vote of a credit risk manager, and exposures that could materially affect liquidity or solvency The credit assessment informs a credit decision that may even need board approval. is taken post-assessment by the person or algorithm For companies financing higher value assets, decisions empowered to do so. This section covers the possible are often taken in a committee that consists of 2–3 outcomes of that decision, as well as the management of voting members. One member could be the client risk in the organizational setup around credit decision. relationship manager or credit analyst, while another member could be from the risk function.18 Each voting A P P R O VA L O R R E J E C T I O N: member should be required to sign a decision and MOVING BEYOND A BINARY IN ASSET FINANCE briefly state the reasons for their vote. This ensures Once the assessment is complete, companies must take the commitment, transparency and traceability. most important step in asset finance: decide whether or not to give the customer an asset on credit. 18 If the risk management representative’s presence is leading to a significant number of rejections, companies may think about installing a process for business managers to escalate the decision to the top management or board level. Internal control and audit would then review the performance of overruled credit decisions. GE T TING REPA ID IN A S SE T FIN A NCE 4.3 Decision-m a k ing a nd disbursemen t 42 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management For companies financing smaller assets, it may be more If done properly, monitoring serves two objectives: practical to rely on a tested scoring algorithm, to hold (1) maintaining the borrower’s willingness to pay and quick online committees or to allow credit officers to (2) finding out early if the financial situation of the approve individual loans and use committee meetings to borrower has deteriorated and is threatening their review these approvals periodically. ability to pay. The first is the more important rationale: monitoring reminds the client that they still have If a scoring algorithm is used it should be clear and an obligation, that the company (or even better, the formalized that the risk department owns that algorithm. relationship manager) will be personally disappointed if This means that they regularly (preferably annually) back an installment is late, and that severe consequences and test it, update it and define then implement expert overrules additional costs will follow if a loan falls into arrears. when necessary (e.g., in crisis situation). Cases decided by a scoring algorithm should not be exempted from audit checks, In asset finance, monitoring should entail three main which may include a review of data collection as well as the components: execution of proper back testing and model calibration.19 1. Regular notifications (through calls or SMS) of upcoming or recent payments due. Such DISBURSEMENT notifications should be triggered automatically by the In asset finance “disbursement” means physically handing credit risk management system, and should highlight over the asset to the customer. Sometimes this may require consequences if the payment is not made.20 an installation in a remote place (e.g., solar pumps), 2. Physical monitoring of the device. Though which also offers an opportunity to conduct some final this essentially refers more to operational risk verifications as suggested above. This has an indirect management, it is highly recommended to establish operational risk: if it takes too long to deliver and/or mechanisms that allow lenders to track an asset’s install an asset, it is reasonable to assume that a customer location (e.g., GPS tracker), as well as its proper may be more likely to default. This would typically apply functioning. This can help to mitigate reputational more to productive assets, especially those tied to seasonal risk, as well as fraud and theft. use like water pumps purchased to irrigate crops. 3. Personal check-ins with the client. Ideally done in person, but over the phone is also an option if 4.4. Monitoring & Repayment necessary. These check-ins help identify emerging risk factors in the personal life or business. They are also Once an asset has been installed, the goal for a company is a great chance to inquire into the customer’s further to keep the customer paying on (or ahead of) schedule until needs and satisfaction with the offered service. the obligation has been completed. This is accomplished Scheduling and implementing this monitoring requires a through regular and pro-active monitoring of the client. MIS that is able to trigger notifications, identify clients for check-in, and flag cases that need attention. That MONITORING system must also be able to track the payment status in Credit monitoring is done to ensure that in the borrower’s (almost) real-time and calculate due dates and amounts. daily battle of balancing priorities under tight financial A well-integrated call center function is also crucial in constraints, the credit payment comes first. credit monitoring. 19 The systemic risk of a scorecard-based credit decision is that the algorithm performs poorly. At development stage that can be tested quite easily with the help of a test set, which was not used to train the algorithm. However, model performance can also decrease over time, as customer profiles/behavior or environmental factors may change. That is why regular back testing and model calibration is so important. The audit’s role here would not be to test the formulas themselves but rather to ensure the back testing is performed regularly (at least once per year). 20 Flexible PAYGo schemes are not immune from this requirement. Independent of how flexible the payment is (e.g., when working with what often is called “grace days”), each customer shall be notified when they are close to missing a contractual obligation. GE T TING REPA ID IN A S SE T FIN A NCE 4.4 Moni t oring & repay men t 43 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Every monitoring action, whether regular or ad hoc, Collections is arguably the area in which asset finance should also result in an update to the client’s rating/score. differs the most from traditional microcredit. The ‘asset’ This provides additional data to study and use when in asset finance is its own collateral. Instead of hounding developing early warning indicators. clients for pennies on the dollar or pursuing legal action to seize hypothecated collateral, AFCs can repossess R E PAY M E N T the leased item and sell it to another customer with For clients who are in repayment, we recommend comparable ease. When combined with the ability to establishing an easy-to-understand payment schedule remotely switch off an asset, this provides AFCs with a with equal installments over a contract’s lifetime. An powerful lever to enforce collections. exception may be highly seasonal activities such as However, knowing when to use a given lever is a challenge agriculture or tourism. for early-stage companies. This section covers: Whenever possible, cash collection should be discouraged • The process of escalation that can lead to an eventual (better forbidden) to mitigate operational risk associated repossession with theft or fraud. More secure options such as mobile • The mechanics, rationale and math behind repossessing money, internet banking, branch payments and/or bank an asset transfers are preferred. However, companies need to • Alternatives to physical repossession, such as lockout ensure that the method of payment is understandable technology and easily accessible. For example, one should not rely on mobile money when there are no nearby agents, as customers may struggle to pay even if they are willing and Escalations able. Likewise, ensuring that customers understand how Credit escalations are steps in the process of getting ever to make payments (for example, how to initiate a mobile more serious with a delinquent or defaulted client, with money payment) is important to mitigating credit risk. the goal of having them pay off their arrears and resume regular payment. These steps can escalate from automated SMS to voice calls to home visits to repossession to legal action, and may include many steps in between. 4.5 Collections For almost any AFC, the first steps will be to remind the No matter how high-quality an asset is, and despite all of client that a payment date is at hand or just passed, and the work outlined in this chapter and the ones preceding to find out why they have not paid. Early-stage actions it, some clients will fail to pay on-time for their product. may then vary depending on the client’s risk profile and At the end of that first day (Day 0), the client is officially repayment history. This is why AFCs need to regularly delinquent. This is not an emergency—in fact, for many update their data, and use that data to segment clients by companies it may be the norm. Most clients will typically the degree of risk they represent. For less-risky clients, it pay within a few days and require little in terms of follow-up. may be sufficient to get a ‘promise to pay’ or even think But as the days tick by and more payments are missed, the about rescheduling, while for riskier clients it may be likelihood of the client eventually paying begins to diminish necessary to schedule a home visit. and a write-off become an increasingly probable outcome. The challenge with escalations is that while more serious It is the job of ‘collections’ to avoid that outcome. This is actions (such as house visits or repossession) may be good accomplished through a progression of increasingly serious ways of collecting value from an outstanding asset, they actions that are explicitly defined by the organization’s are also expensive and time-consuming. This is why Collections Policy (part of the policy framework described segmentation is so important—it allows companies to in Chapter 2). Each of those actions has the same basic prioritize collections actions by risk segment. goal: getting the borrower to pay as much as they possibly can while avoiding or minimizing any loss for the lender. GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 44 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management As clients move through various stages of escalation, the person responsible for engaging with the client may  redit BOX 19. C reference bureaus change as well: Independent credit reference databases play 1. Early-stage arrears may be managed by the staff important roles in financial inclusion: member or agent who originated the loan and who • They allow providers to look-up new borrowers to has a financial interest in keeping the client in good see what loans they have outstanding and how standing. they have paid in the past. 2. Moderate risk clients could also be reached by the • They are a mechanism for clients to leverage company’s call center, who may want to obtain a good repayment behavior into additional credit promise to pay and explain future steps, including for home/business. repossession. • They are a means of enforcing discipline on 3. Higher risk clients can be handled by a specialized existing borrowers. collections team who may conduct additional calls or This last role can be a useful lever for AFCs, who visits to push for an amicable arrangement. This could have few other mechanisms to legally and ethically include rescheduling or even partial forgiveness of the nudge a borrower to repay. But providers will want to loan to get a borrower back on track. make sure that negative marks are eventually cleared from a credit file to avoid inadvertently contributing to 4. Repossessions may be handled by the same team financial exclusion. or by a different group of specialists. At this point, nothing short of a major payment should suffice to prevent repossession. reference bureau. They should also clearly communicate 5. Legal action Where repossession is not feasible, the the customer’s rights, as well as the mechanism to deal final stage is often outsourcing collections to a law with customer disputes, including requests for corrections office, collections agency or a specialized subsidiary of or updates to account information. the lender. This may also be the time to report a client Case studies of three different AFCs’ escalation strategies to the credit bureau (see Box 19). are seen in Box 20. Splitting out the tasks in arrears management and collections among different units as above also makes it Repossession easier to measure the performance of each team and avoid In theory, the decision whether or not to repossess a defaulted “perverse incentives.” asset should be a question of straightforward math: For reasons of consumer protection and reputation, all • What residual value does the asset have? of these tasks as well as the people responsible for them, • How much is a repossession likely to cost (staff time, the timeframes and the suitable code of conduct, must fuel, warehouse space, etc.)? be explicitly laid out in a Collections Policy. This is an • How much value has historically been recovered in area particularly vulnerable to abuse, so being clear about resale of similar repossessed assets? collection policy and auditing the actions of collections staff is vital. However, there are complicating factors to consider. First Contracts should clearly highlight the company’s right is the signaling effect. For many lower-value items, or to switch off or repossess an asset, as well as the precise assets in the later stages of repayment, repossession may terms under which they would do so (e.g., after 90 days appear uneconomical on its face, as the remaining value of a consecutive nonpayment). Contracts should also be used asset may be low and the asset resalable only at a steep explicit about the criteria for reporting clients to a credit discount, if at all. Yet no one should underestimate the signaling effect of collection actions in general, and highly GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 45 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management  ase BOX 20. C studies in repossession uPowa uses lockout Clients in 1–4 are to be handled by call center and 1.  technology to motivate sales agents repayment, but sought to Clients 5–6 are to visited by collections agents 2.  define a collections strategy beyond lockout that could simultaneously recover the  epossession is to be considered if viable for Clients 3. R most value from its outstanding portfolio, while also in group 7, and conducted for clients in 8 maximizing resources available (sales agents, call center This new strategy required 80 percent of clients to be and collections agents). With the help of CGAP, they covered by call center and sales agents, and for the arrived at a data strategy where clients were segmented last 20 percent of clients (candidates for repossession) from 1 (lowest risk) to 8 (highest risk) based on a to be handed off to collections agents. Implementing number of factors, including the number of days they this strategy also required regular data updates and had missed since origination and the number of days improving the call center functions. since their last payment. Collections activities were then prioritized according to that segmentation: Update risk Calls, SMS Repossession Calls and SMS Repossession segmentation and visits to if viable for Legal to low risk for High risk to most recent moderate risk defaulting recourse clients customers period clients customers Days since last payment 1–30 31–60 61–90 91–120 121–150 151–180 181+ Current 1 1 2 2 3 4 5 1–30 days 2 2 3 3 4 5 5 31–60 days 3 4 5 5 5 6 7 Days in arrears 61–90 days 4 4 5 5 6 6 7 (behind 91–120 days 4 4 5 6 7 7 7 contract) 121–150 days 5 6 6 7 8 8 8 151–180 days 5 6 6 7 8 8 8 Beyond 180 days 7 8 8 8 8 8 8 EFTA is a leasing company reviewed and signed off by their branch manager and the operating in Tanzania that COO. Upon approval, two officers will move ahead with finances larger business the repossession, ensuring that the requisite technicians assets. When an EFTA lease and transporters have been hired to take care of the has been written off in their operation. financial system, that customer’s asset will be The equipment is then transported back to the supplier’s repossessed. The customer will have by now been issued premises or to an EFTA warehouse for storage pending with a series of verbal and written warnings, and lastly, as resale. A repossession form is completed by the a final resort, a repossession warning letter. An Investment Team to document that all equipment was investment officer (loan officer) will then write an internal recovered and that it was done safely and following repossession recommendation form outlining the reasons protocol. This is reviewed and signed off by the COO. why they believe this step must be taken. The form is GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 46 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management  ase BOX 20. C studies in repossession (continued) Tugende uses partial and their challenges and are not making efforts to pay off complete asset their arrears. impounding. Partial If arrears go beyond 15 days without communication impounding (their term for and payment, the collections team considers lockout) is applied to the clients from 1 to repossession. Assets repossessed by Tugende are 7 days in arrears and serves to motivate the client to first assessed for quality / upkeep and, if possible, repay the lease while the outstanding balance is still made available to new Tugende clients as one of manageable. Complete temporary impounding is the range of used vehicles it offers for sale on the considered for clients who have not communicated secondary market through the partner companies. visible repossessions in particular. Assets are often financed repossessions should be done respectfully and with in marketing campaigns across a limited geographical area. tact. But various AFCs have also found ‘voluntary The contagion effect of not pursuing repossession among repossessions’ to be a useful tool. In these cases, clients clients in the same community is a real risk, and word are allowed to return an asset when they are struggling travels extremely fast. If companies do not stick to their to repay and to receive their initial deposit back, either communicated policies on repossession, clients will know. partially or in full. Such offers cannot extend too long past the date of origination. But they can provide relief to Second, the monetary flows related to repossession need customers, minimize costs and maintain reputation. to be valued on a net present basis, meaning that all expenses and inflows occurring after the default should A final point to consider is that secondary markets for be discounted back to that date using the effective valuable assets are historically an important channel interest rate as the discount rate (for information on for reaching poor customers. Every asset that can be how to calculate NPV flows, see Annex 5). The reason repossessed and redeployed has a chance to help families for this is to account for the true cost of default, which who otherwise could not afford it. The more that includes both the actual and opportunity costs. If the companies can invest in their ability to refurbish and asset were resold one month faster, that payment could redeploy assets (particularly income-generating ones), the be invested earlier, typically into the same asset class (i.e., more social impact they can unlock. another lease contract), hence earning the same interest. Repossessed items typically take some time to turn into Lockout technology Much has already been written in this guide about the cash (see Box 21 for a hypothetical example). utility of lockout technology. It is a potentially powerful In our experience, many defaults occur relatively early in tool for risk management, although it has limitations. a repayment cycle, while assets still hold significant value. Remotely locking a device is, in effect, a ‘soft’ repossession, Companies should prioritize these assets for repossession, and should occur within the same documented as this will decrease their LGD. Also, repayment issues frameworks that are clearly understood by agents and early in a contract are likely a sign that a customer has clearly communicated to potential customers. But once been oversold or does not value an asset. It may be best these are communicated, they must be followed through to act quickly, within the stated terms. Repossession on: it is equally important to enforce a lockout and other decisions should follow clear, documented policies. They steps as it is to stipulate them in the contract. should not be left to the ad-hoc discretion of customer service teams. For a defaulted customer whose asset has been switched off, or for a customer who is making regular payments That said, the negative concept of repossession does but at a rate below the acceptable threshold, companies not have to be the reality. Even when involuntary, GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 47 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management  olarCredit BOX 21. S repossessions Jon defaulted on his solar water pump from fictional was repossessed, which cost SolarCredit $50 in total. company SolarCredit six months ago, with an From experience SolarCredit knows the repossessed outstanding balance of $1,000. Three months ago, system’s value to be $500 and expects it to sell in one SolarCredit collected $300 in cash. Today the system month. The effective interest was 30 percent per annum. What is the expected NPV of the recovered value? Time of Default Discounted at effective rate of particular loan (IAS 39 / IFRS 9) NPV of Future Collection Flows Flow Default +1 Flow Default +2 Flow Default +3 ... Flow Default +24 Time value of recoveries: (+/-) Cash Flow 1 Cash Flow 2 (–/+) = NPV # Months since default # Months since default 12 12 (1 + Eff.APR) (1 + Eff.APR) 300 50 500 – + = $666.15 3 6 7 (1 + 30%) 12 (1 + 30%) 12 (1 + 30%) 12 must decide whether or not to physically repossess the Just as important is to study the efficacy of various asset. For clients who have defaulted in the first 50–65 interventions, including remote lockout: percent of their contracts, repossession is recommended, • Which customers start paying again after their first, as it maintains the signaling effect described above, second, or twentieth lockout? recovers some value and prevents local mechanics from • After how many lockouts does a defaulter typically ‘jailbreaking’ the asset. Knowledge diffuses quickly: give default? Can this be predictive? enough clients enough time, and they will figure out how to disable lockout technology. • What is the average repayment boost from an SMS vs. a call vs. a visit? How should those be deployed? T H E I M P O R TA N C E O F D ATA I N C O L L E C T I O N S • Do repossessions increase the overall portfolio quality As with other aspects of credit risk management, in a given area? collections can be vastly improved by data analysis. It is important to study the reasons why clients default, as this can allow providers to correlate them with successful workout strategies. This may also improve default prediction power over time, eventually becoming part of the credit scoring algorithm. GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 48 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management CH A P TER 5 PORTFOLIO MANAGEMENT W SUMMARY HILE THE PR EV IOUS SECTION discussed the management of credit transaction risk, this chapter focuses on • The management of a loan portfolio and its portfolio monitoring and management. As stated in collective credit risk depends greatly on how Chapter 2, every lending institution needs to accept a certain a company measures and analyzes portfolio amount of credit risk. This is articulated in its risk strategy, health. tolerance and/or appetite statement. The objective of portfolio • The EL framework is a valuable way to management is to monitor and ultimately ensure that measure past losses, estimate future ones operations stay within the accepted risk boundaries. and price those into the overall cost of an A lender can be almost certain that some of its borrowers asset. will fail to repay their obligations. It is (or should be) easy • As a reaction to more flexible payment for a company to state how many borrowers have defaulted schedules in asset finance, traditional PAR and what losses occurred as a result. But risk managers metrics are being re-imagined. Metrics such have a much harder task: they need to project, with as as collection rate and receivables at risk can little uncertainty as possible, how many borrowers are add critical nuance to portfolio analysis. likely to default in the future and how much the company • Fast-growing companies must be careful will lose if they do. not to rely on portfolio averages, as new These are questions of risk, and risk is always future-looking: loans tend to pay better. Rather, they once something bad has happened it is not a risk anymore should include vintage curves and transition but an actual, unfortunate event. The quantification of risk is matrices in their regular credit analyses to also future-related, with unknown events and developments understand how risk is evolving over time for making it impossible to predict with perfect accuracy. monthly cohorts. However, what is possible is to quantify the probability that a • Analyzing a portfolio by geography or given risk will materialize, as well as the likely impact should product type can yield valuable insights and it come to pass. This reduction in uncertainty is the role of help maintain diversification, so that the portfolio management. portfolio does not become concentrated in one area or sector. This chapter covers the following aspects of portfolio management: • Key terms and concepts • Concentration and diversification • Measuring portfolio performance • EL and unexpected loss in risk management • Portfolio analysis • Monitoring and dashboards GE T TING REPA ID IN A S SE T FIN A NCE 4.5 C ol l ec t ions 49 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management 5.1 Key Terms and Concepts  olarCredit BOX 22. S write-offs An agent from fictional company SolarCredit sells an ARREARS SHS on credit. They enter this sale into SolarCredit’s The term arrears refers to payments that were expected database along with proper documentation. and obliged but have not been received. Simply put, the SolarCredit’s accountant now initiates several payments that are late. Borrowers are classified as ‘current’ balance sheet actions:a if they have made all of their payments at a given point in time. If a payment has come due but not been made, they She credits (reduces) the Inventory account 1.  are said to be ‘in arrears’. If borrowers pay the amount due by the principal amount of the loan (the face (plus any penalties or fees that may have accrued), they are value of the asset), and debits (increases) Loan once again ‘current.’ They may then stay current for the Receivables by the same amount. remainder of their obligation or end up in arrears again. She debits (increases) the Expense account by 2.  Once an exposure has been in arrears beyond a certain the amount of EL, and credits (increases) Provision limit of days, it will be classified as a non-performing loan. for Bad Debts by the same amount. This is money This label indicates that a company is no longer expected meant to cover EL. (by most accounting standards) to accrue interest, and When the first few payments are received, the that only payments on interest already accrued shall be accountant credits (reduces) the Loan Receivable, recognized.21 When an exposure is further in arrears, it and debits (increases) Cash. But unfortunately, the should be fully written-off from the portfolio as will be borrower quickly defaults, and after multiple efforts discussed below. SolarCredit decides that they will never be able to In loans or leases with fixed monthly payment schedules, collect more cash or repossess the asset. arrears are the total amount that has come due but has At this point, the accountant credits (decreases) not been paid at the time of evaluation. In flexible PAYGo Loan Receivables, and debits (decreases) Provision approaches, the definition of arrears is subjective. PAYGo for Bad Debts. This act of decreasing the Loan arrears could be viewed as the difference between the Receivables account is the actual ‘write-down’, as nominal amount of payments that should have been the value of that receivable is now recognized to received on a date (e.g., $10 after 40 days of a solar lease be 0. Hopefully the Provision account is a sufficient that costs $0.25 a day) and the amount that has actually buffer for this and other defaults (see later in this been received. However, the PAYGo model is unique in chapter for more detail). that it never requires these arrears to be settled. This example uses the ‘allowance method’ of accounting for a  loan losses. Different companies may use different accounting W R I T E- O F F S methods. Write-offs can be used as a shorthand for actual credit losses, but the term refers to an accounting procedure. If at some point after disbursement it becomes clear that a RISK APPETITE borrower will never repay in full then the value of the asset As discussed in Chapter 2, a ‘risk appetite’ is a company’s (the loan receivable) has reduced. This reduction in value stated accepted level of risk exposure, articulated within must be accounted for, and that accounting step is called a their broader strategy.22 This statement covers all major ‘write-down’ or a ‘write-off’ (see Box 22 for example). risks, which for lenders will include credit risk, and quantifies both the measurement of that appetite and 21 Note the shift from an accrual basis of accounting to a cash basis once the loan is declared non-performing. 22 For an in-depth discussion, see FSB (2013). For practical examples, see IRM (2017). GE T TING REPA ID IN A S SE T FIN A NCE 5.1 K e y t erms a nd concep t s 50 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management specific limits. A company might state its acceptable level of credit losses, expressed as the percentage of the total  implified BOX 23. S EL model portfolio that will be ‘non-performing’ or ‘written off.’ On December 31, a credit analyst for fictional As an example, one of CGAP’s AFC partners has the company SolarCredit reviews their annual following credit ‘goal’: performance, and finds the following results for their “ To develop and meet sustainable targets for the only loan product: percentage of write-offs and recoveries to new loans • 10 percent of SolarCredit borrowers defaulted whereby the [Non-Performing Loan] ratio should be throughout the year. <10 percent … [COMPANY] may set these targets • Their average exposure at the time of default lower from time to time to improve performance was 100 KSH. in the portfolio against these minimums. Non- Performing Loans are defined as any loans at over • On a NPV basis, just 20 percent of this exposure PAR 90.” was (or is expected to be) recovered on average, so LGD was 80 percent of EaD (100 percent – A different company may choose a different goal or use the 20 percent). same metric with a higher or lower target. When setting these goals in asset finance, companies should always Based on this information, the analyst computed the keep in mind their typical recovery rates on defaulted EL to be: assets, and potentially even include specific targets for repossession and resale. EL = PD • LGD • EaD = 0.1 • 0.8 • 100 = 8 KSH She then recommended that 8 KSH be provisioned EXPECTED LOSS for each loan expected to be originated in the next EL is the average amount of credit loss that is expected to be year, and that this amount be incorporated into incurred over a particular time period. The loss is measured SolarCredit’s pricing as their cost of risk. as the present value of receivables that are not expected to be collected and therefore will be written off or otherwise expensed over the time period. minus the percentage of EaD that will be recovered through liquidation of collateral and/or other post- EL is the product of the probability of default (PD), the default collection actions. As highlighted before, for LGD and the EaD. For an example of how it is calculated, the purposes of establishing LGD, the post-default cash see Box 23. flows from recoveries must be discounted back to the EL = PD • LGD • EaD time of default at the original internal rate of return of the defaulted contract. The probability of default (PD) is the probability of a borrower to default within a specified period of time, The following sections of this chapter elaborate on typically one year.23 appropriate ways to quantify PDs, LGDs, and EaDs - and hence to calculate EL. But as EL is the product of The EaD is the total balance owed by the borrower to the three different variables, there is an infinite number of lender at time of default, and is expressed in monetary combinations to result in the same outcome. Using the units (e.g., US$). example of SolarCredit, another company could reach the LGD is the percentage of the EaD that is expected to same EL result with a PD of 20 percent and an LGD of 40 be considered lost, once it has been established that a percent; their EL would still be 8 KSH. default has occurred. The LGD is equal to 100 percent 23 The PD is most often stated for a future period beginning immediately, but can also be expressed as a forward default probability beginning in one year for one year, for example. GE T TING REPA ID IN A S SE T FIN A NCE 5.1 K e y t erms a nd concep t s 51 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 10. Probability curve of different loss levels Mean EL (priced in) Probability % Unexpected Losses (covered by capital) Catastrophic Loss Stress Loss (not priced in / not covered by capital) (at specified confidence level) Annual Loss Amount AFCs are typically willing to accept higher PDs than follow a right-skewed probability distribution, the area typical MFIs, as they are usually able to achieve higher below which sums up to 100 percent.26 The first point on recovery rates based on asset repossession, and hence lower the X-axis (mean), which divides the curve into two 50 LGD. That strategy must be agreed upon and articulated percent areas, is what we would expect the portfolio loss to in a risk appetite. be. Every point to the right of that would be unexpected (but possible). The farther to the right, the less likely the UNEXPECTED LOSS respective portfolio loss is going to occur, but the more EL are calculated based on historical performance and devastating the impact would be. should be priced in and accounted for. However, actual losses may be less than expected, in which case the firm SUMMARY has higher earnings, or more than expected, in which case Credit portfolio management can be summarized as a set the firm has an unexpected loss: a yearly loss that is not of policies, work processes, and tools to: really expected but generally possible to occur even under 1. Monitor compliance with a company’s expressed risk ordinary business conditions. appetite Unexpected losses can still be calculated, within a given 2. Predict future developments regarding the risk confidence interval.24 Common standards for confidence appetite, and identify (adverse) trends early intervals range between 99 percent and 99.9 percent.25 The sum of expected and unexpected loss can be used as 3. Suggest actions to mitigate credit risk in case the reference for the minimum amount of loss-absorbing tier-1 accepted risk is or will be exceeded capital. This is the ‘Stress Loss’ point. 4. Account for occurred or likely-to-occur credit risk The concept of Unexpected Loss is best illustrated in As may already be clear, portfolio management is an a chart like Figure 10. Portfolio credit losses typically exercise in analyzing large numbers. It deals with shifts 24 Assume you flip a fair coin 100 times. The likelihood that you get more than 60 tails is approximately 2 percent. Or, in other words, with 98 per- cent confidence one would not get more than 60 tails. 25 A confidence interval of 99 percent implies that the yearly loss would not exceed X in 99 of 100 cases (i.e., years) – all else equal. However, the converse is that you should expect such a development once in a hundred years. 26 A portfolio credit loss is the sum of all individual credit losses over a certain period (e.g., one year). GE T TING REPA ID IN A S SE T FIN A NCE 5.1 K e y t erms a nd concep t s 52 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 11. Portfolio growth over time 30,000 25,000 20,000 15,000 10,000 5,000 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 January 2020 February 2020 in key variables over time of a portfolio made up of briefly covering growth and recoveries. thousands (or millions) of loans/leases. How these shifts are measured is discussed in the next section. GROW TH The speed (or lack thereof) that a company is growing its loan/lease ‘book’ is an important contextual factor for 5.2 Measuring analyzing a portfolio. It is important to look for sudden Portfolio Performance increases in growth particularly for branches and officers (or agents) where fraud is possible, where there is growth There are four main dimensions of portfolio performance in average loan size or where there is a strange surge in that should be regularly tracked by a portfolio analyst and lending activities toward the end of each month (see risk manager: Figure 11). • Growth. How many loans/leases have been originated in the last period? MEASURING ARREARS • Arrears. What percentage of the outstanding portfolio Measuring the amount of exposures in arrears in relation is owed by clients who are in arrears? What percentage to the overall size of the portfolio (‘who has not paid us, of loans have been labeled as ‘in arrears’ for more than and how much are we owed?’) is one of the key activities 30 days, 60 days, etc.? needed to measure portfolio health. Interpreting the arrears to estimate the risk level of the portfolio (‘who is • Write-offs and recoveries. What percentage of all likely to never pay us again, and how much would we lose originated contracts have been written off? What if they did not?’), then keeping that risk level within the percentage of defaulted balances have been recovered? stated appetite, is the more complex, probabilistic aspect.27 • Rescheduling. What percentage of originated loans Figure 12 shows the journey of nonpayment from have been rescheduled? How quickly does the company delinquency, to default, to loss. Because portfolio reschedule loans? Is there an ‘evergreening’ effect? managers will be tracking this journey for thousands or All of these are critical for formulating a complete picture even millions of loans, they need numerical indicators, of portfolio performance. However, some are more or metrics, that quantify the percentage of their portfolio complex to calculate than others, so this section will focus that is at various stages of nonpayment. mainly on how to measure and monitor arrears, while also Metrics that track delinquency such as PAR ratio are 27 For an engaging and detailed review of portfolio quality metrics and their complexities, see Rosenberg (1999). GE T TING REPA ID IN A S SE T FIN A NCE 5.2 Me a suring Port fol io Perform a nce 53 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 12. Credit deterioration over time Delinquent Default No payment PAR PAR PAR PAR (PAR) Write-off 8–30 days 31–60 days 61–90 days over 90 days 1–7 days Minor Risk, but Moderate Risk Serious Risk Low Chance of Lost needs to be Repayment watched forward-looking risk indicators that risk managers use We shall discuss PAR, Receivables at Risk, and Collection to predict the probability of progression to default and Rate in more detail below. loss. A customer who has missed one payment is not guaranteed, or perhaps even likely, to miss a second Portfolio at Risk PAR is a ratio that reflects a company’s credit risk or eventually default. But they will still have a higher position at a specific point in time. The definition of ‘at probability of default than a customer who has missed risk’ depends on the number of days that are chosen no payments, just as a customer who has not paid for 30 for analysis, with PAR X defined as the total balance days is more likely to default than a customer who has not outstanding under any loan contract which is X number paid for 3. As previously mentioned, the most common of days (or more) in arrears. metric for tracking delinquency with traditional lenders is PAR. Collection Rate (with its various permutations) PAR is measured as: and Receivables at Risk are also important delinquency Principal outstanding > X days late indicators for PAYGo lenders.28 PAR_X(%) = Total gross portfolio outstanding A second set of metrics, such as ‘nonperforming loans (NPL) ratio,’ indicate defaults - that is credit events that Most commonly, X is set at both 30 days and 90 days have already come to pass. These tell a company what to get a general impression of the credit risk position. percentage of their portfolio is held by clients who have These numbers are based on expectations of monthly stopped paying for a sufficiently long time that the payment; a customer who has not paid in more than 30 company can no longer accrue interest, should switch to days has missed at least 2 monthly payments, while one cash-based accounting and consider initiating the recovery who has not paid in 90 has missed at least 4. Therefore, or write-off process. These loans are non-performing or on any given day that PAR30 is calculated, it shows the in default; the company considers it unlikely that the percentage of an outstanding portfolio that is owed by customer will resume payment of their own accord. clients who are 30 or more days late on a loan payment. Lastly, a third set of metrics, such as ‘write-off recovery PAR90 would tell us the same information, but for clients rate’ and ‘write-off ratio,’ quantify the results of actual who are 90 or more days late. PAR90 can never exceed defaults: what percentage of losses were recovered and what PAR30, as every exposure falling into PAR90 must fall percentage had to be written off as an accounting loss. into PAR30, as well. 28 These and other metrics are discussed in far more detail in Khaki et al. (2021, forthcoming) and in materials that can be found on the PAYGo PER- FORM website: https://www.findevgateway.org/paygo-perform/. GE T TING REPA ID IN A S SE T FIN A NCE 5.2 Me a suring Port fol io Perform a nce 54 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management Thirty or more days in arrears is a reasonable alarm threshold where an institution may be seriously  roblems BOX 24. P with PAR, or how not all concerned about the potential default of the borrower. If arrears are created equal a borrower moves to PAR90, the default has more or less Imagine a portfolio analyst for fictional company materialized, and the account should soon be moved over SolarCredit is looking at two borrowers on to the collections team.29 It is critical that the portfolio January 3: Aisha and Andrew. Both have missed risk reports not only include the status of PAR but also a contractual payment of 100 KSH on January of the development of write-offs. Some lenders might be 1. Both are in arrears. But Aisha has made every tempted to write-off loans aggressively to affect the PAR payment for the eight months since originating her ratio, and so sudden shifts in PAR ratio accompanied by loan, although two payments were made 7 days late, while Andrew has missed his contractual payments simultaneous surge in write-offs should be a red flag if not in November and December as well (and so has not instructed by the management. made a payment for 93 days, since October 1). Companies financing smaller assets or companies who Aisha is currently delinquent on her loan, and would expect more regular payments (e.g., daily or weekly) may be counted in a PAR1 metric since her exposure wish to track earlier metrics (PAR1, PAR7, or PAR15). is delinquent for more than 1 day. But SolarCredit Even for lenders with monthly schedules, PAR1 has its should be reasonably confident (though not certain) place as an early risk indicator if considered together with that it can collect her arrears and will not yet the trends in PAR30 and PAR90. One day in arrears consider her future payments to be at risk. However, (PAR1) does not constitute a default. But an increase in SolarCredit should now consider it highly unlikely the number of customers falling under PAR1 can alert risk that Andrew will make another payment. His account managers to investigate and address emerging problems is included in PAR90 (since his exposure lies in the that may preview further deterioration in the future. loans that are delinquent for more than 90 days in arrears), his entire outstanding balance should, at It is common practice to analyze PAR statistics in segments: minimum, be considered ‘at risk’ and should likely be by origination period, branch, loan product, economic processed through a stricter collection procedure. If sector, responsible loan officer, etc. It is also important to his asset can be repossessed and resold, some or analyze specific customer segments (e.g., male and female, all of that balance may be recovered. Whatever is not on-grid and off-grid, farmers and traders). This can be must be written off as a credit loss. helpful in detecting risk and default concentrations in certain parts of the portfolio. Segmentation can also help explain (to some extent) the underlying risk factors that actually received in a given period, excluding the deposit. are driving the arrears behavior. See Box 24 for practical The complement of the Collection Rate (1-CR) is nuance in analyzing individual clients. whatever was not received. Collection rate There are two main ways of tracking Collection Rate: PAYGo PERFORM defines ‘Collection Rate’ (‘CR’) as Overall and Ongoing (terms vary from company to the “ratio of all collected receivables payments over total company). Overall Collection Rate shows, on a given date, receivables payments due for a period (not including the percentage of all expected payments that a borrower deposits).”30 Put simply, Collection Rate tells us what has made as of that date. Ongoing Collection Rate, on the percentage of a customer’s expected payments were 29 90 days overdue is a common industry standard for default, widely applied by regulators and supervisory organs. However, different industries may have different benchmarks for when a receivable is deemed non-performing. 30 The PAYGo PERFORM initiative is an open, transparent industry process that seeks to develop a reporting framework and set of key performance indicators for the PAYGo solar industry. It comprises investors (private and debt investors, local and international banks, and development finance institutions), PAYGo executives, and experts in energy and financial inclusion from around the world, and is led by Lighting Global, GOGLA, and CGAP. GE T TING REPA ID IN A S SE T FIN A NCE 5.2 Me a suring Port fol io Perform a nce 55 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management but resume paying as soon as possible. A customer who  alculating BOX 25. C Collection Rate pays 7 or 9 days out of every 10 may not be considered ‘at Aisha acquired a PAYGo SHS from fictional company risk’ at all. But one who pays as seldomly as #2 is surely SolarCredit on September 1. The SHS has a daily unacceptable, even if he has made a recent payment. rate of $0.25. When she acquired the asset, she This is the dilemma of evaluating PAYGo portfolio made one payment for $7.50, unlocking it for 30 health: an analyst needs to measure both nonpayment and days. However in October she made two payments, slow payment. For nonpayment, PAYGO PERFORM each for $3, on the 1st and the 15th. recommends a metric called Receivables at Risk– On November 1, Aisha’s Overall Collection Rate Consecutive Days Unpaid, or RAR(CDU), which only stands at 88.53 percent ($13.50 / $15.25). counts clients as ‘at risk’ if they have made no payments at Her Ongoing Collection Rate for the month of all for a given amount of time. RARX(CDU)X captures the October was 77.42 percent ($6.00 / $7.75). percentage of the outstanding portfolio held by those clients who have not made a payment for X consecutive days. RAR(CDU) = other hand, shows the Collection Rate for a given period, Outstanding Receivables > X Consecutive Days Unpaid for example the last 30 days. A sample calculation of both Outstanding Receivables is in Box 25. As PAYGo providers do not have fixed repayment This metric accounts for the flexibility of PAYGo without schedules, a borrower’s Overall Collection Rate is one overstating the risk situation by capturing every client who of the clearest indicators of their risk profile. Higher misses a payment, many of whom may not actually be Collection Rates mean a steady source of cash and an ‘risky.’ If a PAYGo lender can only calculate and monitor on-time loan completion. Low Collection Rates mean one RAR metric, it should be this one. But RAR(CDU) the company may not recover its capital on schedule, may does have its weaknesses. For example, it does not capture not be receiving enough cash to cover maintenance and those clients who are making payments but at too-slow servicing costs, and could be indicators of future defaults. a pace. Look at the previous example. Client #2 would Each company should have a clear idea of what its target never be counted in a RAR(CDU)90 metric, but remains Collection Rate is for borrowers at various stages in their a poor client overall. He may eventually stop paying (slow repayment cycle. payment is a strong predictor of eventual nonpayment), but even if he continues at this pace, he will not complete Receivables at risk his obligation for years past the contractual end date. As long as payment schedules are contractually fixed, A second RAR metric helps measure the risk of these PAR can be calculated the same for asset finance as for infrequent but still-active payers. Called Receivables- any other type of lending. However, as soon as payments at-Risk Collections Rate (‘RAR(CR)’), the metric is become more flexible, as is the case with PAYGo calculated as follows: approaches, it becomes difficult to define what an arrear is. RAR(CR) = Observe the four payment patterns shown in Table 6 Outstanding Receivables by Overall Collection Rate < [X]% below, showing payment for 10-day periods over 100 days. Outstanding Receivables In a traditional definition of PAR30 (that is, a client with arrears that were due more than 30 days in the past), every customer would be considered ‘at risk’ by day 51. RAR(CR) allows an AFC to identify receivables that are owed by clients who are paying frequently enough to avoid However, more flexible payment schedules allow exactly being counted in RAR(CDU), but infrequently enough the sort of payment behavior demonstrated by Customers to represent a potential risk. By combining this metric #1 and #4. The customer can skip a day when necessary, with RAR(CDU) (without double-counting) a company GE T TING REPA ID IN A S SE T FIN A NCE 5.2 Me a suring Port fol io Perform a nce 56 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management TABLE 6. Comparing portfolio at risk across multiple customer profiles Collection 10 20 30 40 50 60 70 80 90 100 Rate Customer #1 $ $ $ $ $ $ $ 70% Customer #2 $ $ 20% Customer #3 $ $ $ $ 40% Customer #4 $ $ $ $ $ $ $ $ $ 90% can get a clear picture of how much of their portfolio is 5.3 Portfolio Analysis at risk—in other words, held by clients who have stopped paying or who pay very little. In this section, we focus on three important analytics (the computational analyses of data) that all lenders should The risks of ‘at risk’ indicators be conducting on a regular basis to inform their risk New leases usually perform well: even the worst borrowers management: usually manage to pay a few installments before they stop. • Vintage Curves As a result, recently disbursed leases will almost always display lower default rates, and higher collection rates, • Transition Matrices than older generations of leases. But that does not mean • Recovery Analyses that they are less risky. One fundamental problem with PAR and RAR is that V I N TA G E C U R V E S they are lagging indicators: they do not account for recent Vintage curves (also known as cohort analysis) help leases that have not been active for long enough to reach address the lagging issue of PAR and RAR. Vintage curves the default threshold. It is a law of credit physics that no show the cumulative default/write-off rate (the ‘bad rate’) lease that was disbursed two months ago can be in default as a function of time since creation of the receivable. now if the default definition requires a lease to be more There is no size bias in comparing defaults at certain age than three months in arrears. This can lead the PAR and across multiple years of operation, because the bad rate RAR metrics to seriously underestimate risk levels.31 is expressed as a percentage of the originally disbursed If an organization is rapidly growing, there will always amounts. Each month of disbursements is then plotted on be more new loans than older loans in the portfolio, and a separate curve, with the independent variable (x) being overall PAR will appear better than it otherwise would. the months after disbursement and the y axis showing the Conversely, if portfolio growth slows or turns negative, cumulative “bad” rate. PAR will go up.32 However, neither of these movements The ‘bad rate’ may refer to the PAR/RAR definitions necessarily mean that the risk exposure of the company has above, and vintage curves could be plotted for 30 days changed. The next section provides detail on how to analyze overdue, 90 days, etc. The “bad rate” could also be set a growing loan portfolio in the face of these limitations. as write-offs, in which case the analysis would just show losses. Or companies may choose to combine the two, in order to compare losses and likely losses over time. Because the bad rate is cumulative, we would generally 31 For a detailed discussion of the complexities in monitoring arrears, see Rosenberg (1999). 32 If you stopped new disbursements all together, you could have PAR converging to 100%, as the non-defaulted loans are paid-off and the remaining portfolio consists exclusively of loans in arrears. GE T TING REPA ID IN A S SE T FIN A NCE 5.3 Port fol io a n a ly sis 57 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management expect the curves to go up or stay flat over time. But if underwriting criteria. Credit managers can use this type of PAR or RAR are included, there is a chance they may analysis to monitor progress in new loan cohorts and even improve over time. set targets, for example an expected cumulative write-off rate at 90 days of 3 percent or lower. Regardless, it is important to make sure that the denominator always stays the same for each month, and this should be the amount disbursed/leased in that T R A N S I T I O N M AT R I C E S particular month. Figure 13 illustrates the idea, using A transition matrix allows lenders to track evolutions in a ‘bad rate’ that includes cumulative gross write-offs, repayment status over time. Presented as a table displaying PAR90, and voluntary repossessions. migrations from one category of arrears to another, transition matrices show how frequently one type of The closer that curves are to the X-axis, the better the customer (e.g., 31–60 days late) deteriorates (becomes repayment situation. If newer curves show a steeper slope 61–90 days late) or improves (becomes Current). A (i.e., the bad rate is climbing faster compared to previous transition matrix is useful in guiding and tracking the months), it could signal a deterioration in portfolio quality. effectiveness of monitoring, collections and recovery Conversely, a flattening of more recent curves could show activities. A transition matrix can be used to calculate improvement. A variation on the same analysis, and using the probability of default, which taken together with the the same data, is shown in Figure 14 below. The higher the LGD or recovery analysis, can be used to calculate the bad rate, the darker red the shading appears. EL in a portfolio. The transition matrix is based on the Looking at the chart, we can clearly see that some change premise that if borrowers miss an installment, it does not has occurred in June 2020: the write-offs for loans necessarily signify a future default, but it is an indication disbursed that month are noticeably lower than preceding that their economic situation has deteriorated and that the months. Of course, we do not know why—the company risk of default and loss is growing. may have cut its disbursements by half and only chosen A typical transition matrix, such as in Figure 15, is built the best potential clients, or they may have changed their around 30-day arrear ‘bands’ (Current, 1–30 days late, FIGURE 13. Example vintage curve of ‘bad rate’ over cohorts 25% 20% 15% 10% 5% 0% 30 60 90 120 150 180 210 240 270 300 330 360 Days since disbursement Jan-20 Feb-20 Mar-20 Apr-20 May-20 Jun-20 Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 GE T TING REPA ID IN A S SE T FIN A NCE 5.3 Port fol io a n a ly sis 58 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 14. Heat map vintage curve Month disbursed 2020 2021 v Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb 30 1% 1% 1% 1% 1% 0% 0% 1% 1% 1% 1% 1% 1% 1% 60 2% 3% 3% 3% 2% 1% 2% 1% 2% 1% 2% 1% 1% 90 4% 4% 5% 4% 3% 1% 2% 2% 3% 2% 2% 2% Days since disbursement 120 6% 6% 7% 5% 5% 2% 3% 3% 4% 4% 3% 150 8% 9% 8% 8% 7% 2% 4% 3% 5% 6% Cumulative % of disbursed 180 10% 11% 11% 10% 9% 3% 5% 4% 7% principal written 210 11% 13% 12% 11% 10% 4% 6% 5% off, repossessed, 240 13% 14% 14% 12% 10% 5% 6% or PAR90 270 15% 16% 16% 14% 11% 5% 300 16% 17% 19% 15% 12% 330 18% 19% 20% 16% 360 20% 21% 22% FIGURE 15. Sample transition matrix with arrears bands 12 months Settled Current 1–30 31–60 61–90 91–120 121–150 151–180 180+ w/o 1y PD Settled 100.0% - - - - - - - - - - Current 31.9% 56.1% 2.4% 1.2% 0.9% 0.8% 0.9% 0.7% 3.7% 1.3% 7.4% 1–30 34.6% 26.7% 1.2% 0.6% 0.5% 0.4% 0.5% 0.5% 24.4% 10.7% 36.6% 31–60 23.8% 7.1% 0.3% 0.2% 0.1% 0.1% 0.2% 0.3% 45.2% 22.6% 68.5% 61–90 8.8% 2.2% 0.1% 0.0% 0.0% 0.1% 0.1% 0.1% 56.5% 32.0% 88.8% 91–120 0.9% 0.6% 0.0% 0.0% 0.0% 0.0% 0.1% 0.1% 59.5% 38.6% 98.4% 121–150 0.2% 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 61.8% 37.7% 99.7% 151–180 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 64.4% 35.4% 100.0% 180+ 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 65.1% 34.7% 100.0% w/o - - - - - - - - - 100.0% - Current Overdue Default 31–60, etc.). Clients are grouped first on the Y-axis by their Figure 15 represents an outstanding balance that was in a past arrears status (December 31, 2019 in this example), certain status as of 2019; the percentage figures along the then on the X-axis by subsequent arrears status (in this row show what share of that balance are in which states case, December 31, 2020).33 Each row in the transition in 2020. But transition matrices can also display currency matrix represents the balance in a given arrears status as figures or show the number and percent of borrowers in of December 31, 2019, ranging from current (no arrears) each transition. to just before the write-off point. The transition matrix Transition matrices are used not only to look backwards, then provides a view of how receivables act when they but also to project forwards. Frequency leads to reach a certain stage of arrears—in other words, how probability. Beyond just showing what happened in the many borrowers miss an additional payment, how many transition during a particular period, an analyst can recover and become current again, and so on. Each row in 33 As in PAR, the oldest arrears amount determines the classification of the entire principal balance. A client may have missed payments in each of the last 3 months, but all those arrears are captured in the 91–120 days bucket. GE T TING REPA ID IN A S SE T FIN A NCE 5.3 Port fol io a n a ly sis 59 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management FIGURE 16. Transition matrix with collection rate Transition Amounts (USD) 7/1/2019 6/30/2020 Collection Rate Outstanding (1) <=30% (2) 31%–50% (3) 51%–70% (4) 71%–90% (5) 91%– (6) >100% Detached Estimated Amount (USD) 100% and Arrears Unlocks >180 days 59,526 72,576 89,541 96,282 65,936 1,122,822 (1) <=30% 100,255 40,648 5,555 923 320 44 37,913 14,852 (2) 31%–50% 189,738 59,123 55,135 25,557 472 31,055 18,385 (3) 51%–70% 408,771 10,123 99,135 79,238 13,172 5 70,134 136,964 (4) 71%–90% 337,134 4,194 39,775 87,637 68,715 3,052 57,071 76,690 (5) 91%–100% 519,307 907 8,008 24,000 39,470 26,603 885 16,789 402,645 (6) >100% 1,269,300 11,197 29,100 59,323 70,433 51,616 82,435 14,926 950,269 Transition Rates (%) 7/1/2019 6/30/2020 Collection Rate Outstanding (1) <=30% (2) 31%–50% (3) 51%–70% (4) 71%–90% (5) 91%– (6) >100% Detached Estimated Amount (USD) 100% and Arrears Unlocks >180 days (1) <=30% 100,255 41% 6% 1% 0% 0% 0% 38% 15% (2) 31%–50% 189,738 31% 29% 13% 0% 0% 0% 16% 10% (3) 51%–70% 408,771 2% 24% 19% 3% 0% 0% 17% 34% (4) 71%–90% 337,134 1% 12% 26% 20% 1% 0% 17% 23% (5) 91%–100% 519,307 0% 2% 5% 8% 5% 0% 3% 78% (6) >100% 1,269,300 1% 2% 5% 6% 4% 6% 1% 75% forecast on the assumption that next month’s transition shows how transition matrices can be multiplied with might be similar. This moves us from observed frequencies themselves to calculate the overall probability of default.35 to anticipated probabilities. In the case of flexible lease schedules the concept does Based on the example above, which was constructed on 12 not change, but the various ‘buckets’ may. One approach months, there is, for instance, a 1.2 percent likelihood that (shown in Figure 16) is to use Collection Rate. In this an obligation which is Current in Period 1, will end up in case, the row/column headings show customers who are the 31–60 days bucket in Period 2 a year later (box framed >100%+, 91–100%, 71–90%, etc. The top table shows how in blue in Figure 15). To calculate the probability of a current the amounts outstanding in July 2019 have transitioned obligation going into default over a period of 12 months, it over a subsequent year, the bottom table shows the same is also possible to add the probability for all columns starting transition, but expressed in percentage terms of the July from 91–120. For example, looking at an obligation that is 2019 outstanding balance. currently 31–60 days in arrears, the probability of default If a company has a well-functioning scoring model in place, would be 68.5 percent (boxes framed in orange). Building on the same logic can be used with credit scores. However, this the above example, transition matrices are useful for portfolio requires a regular re-calculation of scores. A final approach managers when seeking to predict the future.34 Annex 6 is to use internally defined client segments as the transition 34 As long as individual risk profiles do not change a lot over time. If either customer segments change (e.g., due to an adjusted business strategy) or a crisis occurs (e.g., a pandemic), the future may look less like the past. 35 With an increasing error margin, the longer this period becomes—in particular if there are seasonal default patterns, those should be included for at least one cycle. GE T TING REPA ID IN A S SE T FIN A NCE 5.3 Port fol io a n a ly sis 60 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management buckets. For example, a company may have Good, Average, Risky, and Bad clients, where a Risky customer is one  ictional BOX 26. F company SolarCredit’s LGD who has not made a payment for at least 30 days and has a For simplicity’s sake, assume that Solar Sun has had collection rate between 30 and 70 percent. only three contracts default in its history. R E C O V E R Y A N A LY S I S Receivable Outstanding at LGD rate # Default (KSH) (NPV based) Recoveries are money received after default, which may 1 1,000 75% come from different sources (e.g., the client themself, 2 400 60% guarantors, other finance providers, voluntary sale of assets, 3 1,200 50% confiscation and realization of collateral/lease items, etc.) and most commonly are a result of collections actions. In asset finance, where the asset serves as ready collateral, Weighted Portfolio LGD = recoveries are a major part of credit risk management. (1,000 • 75%) + (400 • 60%) + (1,200 • 50%) The Recovery Ratio is often expressed as a percentage 1,000 + 400 + 1,200 of the EaD, which is the outstanding principal at the = 61.15% moment of write-off. The complementary ratio expressed as (1 – Recovery Ratio) is then called the LGD. As mentioned above, it is crucial to apply a strict NPV approach when calculating the value of recoveries, as well 5.4 Concentration as the cost of confiscation, re-sale and other costs (e.g., and Diversification re-sale discounts). For a more thorough explanation of how to calculate NPV of recoveries, see Annex 3. Concentration risk in a portfolio refers to the risk inherent when risk-carrying assets (e.g., loans, leases) However, while these focus on the recovery of individual have a common exposure, for example when they are asset values, we also want to stress the forward-looking concentrated in a particular geography or economic sector. component with regard to portfolio management. Empirical Imagine a portfolio that is completely concentrated in recovery history can and should be used to predict future one city that is regularly hit by natural disasters, or has LGD, potentially broken down by distinct customer concentrated exposure to farmers of a rain-dependent segments and/or loan products. It is also important to crop. It would only take a fairly routine event (an recognize that recoveries tend to be higher when there earthquake or a drought) to result in catastrophic losses. is collateral in place that can be resold or redeployed. As By diversifying the geographies and sectors included in mentioned before, this is particularly true for lease contracts their portfolios, lenders can manage concentration risk, where the leasing company retains ownership of the asset thereby avoiding unexpected and stress losses. and can therefore repossess it with less difficulty. Concentration risk in credit covers several types A typical portfolio LGD estimation approach would be of concentration, including name concentration, weighting the historically observed individual LGD rates product concentration, geographic concentration, and using their respective exposures at default and dividing sector concentration. Concentration analysis, along them by the total exposure of default. A model of this is with diversification and defined limits on portfolio shown in Box 26. concentration, allow lenders to proactively identify and address these concentration risks, as explained below. Name concentration The outsized concentration of credit exposure with one or several borrowers, where a single default could GE T TING REPA ID IN A S SE T FIN A NCE 5.4 C oncen t r at ion a nd di v ersific at ion 61 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management threaten the viability of the institution. This tends to be These concentrations may be acknowledged and accepted, a less significant risk in asset finance, but larger leasing but that does not remove the inherent risk. Wherever companies may still want to limit the percent of equity or possible, companies ought to diversify their exposure. This assets that is tied up in a single exposure. could happen by adding other financed assets, expanding into new geographies or undertaking detailed research on Product concentration existing customers to understand their vulnerability to Offering multiple products is a good way to mitigate various shocks.36 risks arising from technical failures or supply chains. It is possible that defects may emerge in a product over Concentration analysis time, or that the supply of a product could be disrupted. Companies need to regularly assess the concentration By financing a range of products (ideally from a range of of their portfolio within the parameters described above manufacturers), a lender can isolate any losses that result (product, geography and sector). Concentration analysis from technical failures while continuing to have a healthy, is meant to be extremely forward-looking: what regions growing portfolio. This is also a place to manage the risk or segments is the company overly exposed to and would of obsolescence; a risk manager should be careful that a therefore cause particular harm if defaults spiked? company’s portfolio is not overly concentrated in older Basic concentration analysis can simply show the assets, as these customers may stop paying if a newer and/ percentage of outstanding portfolio that is concentrated or cheaper model becomes available. Offering a range in various regions, products or sectors (see Table of diverse assets (e.g., smartphones and solar pumps) is 7). More detailed analyses can be done using the also a way to avoid overexposure to one client segment or Herfindahl-Hirschman or Gini indices, among others. economic sector (see below). Bandyopadhyay (2010) provides a good technical Geographic concentration breakdown of the various measures of concentration risk, Particularly for low-income customers, many risks such and their advantages and disadvantages. as droughts, floods or civil unrest are local. Early stage companies may have a limited number of customers, but D I V E R S I F I C AT I O N A N D L I M I T S ensuring that those customers are spread across different The underlying idea behind diversification is always to regions of one country can at least help to limit the reduce positive default correlations between individual damage of any one event. In the long-term, expanding exposures. The higher the positive correlations are, the into different countries (although that carries its own higher the likelihood that the default of one exposure risks) can offer even more diversification. implies the default of the second.37 An AFC should not be overly exposed to one region or branch, nor should any one Sector concentration region have an over-concentration in one customer segment. Closely linked to geographic and product concentration, lending to a single economic sector or occupation means To achieve this, companies can develop a limit framework that a downturn in that sector could create widespread like Table 7. This framework uses regions and sectors to defaults. Some AFCs focus on a single product or value divide the portfolio, with the ‘Limit’ column indicating chain, which is an obvious and presumably accepted risk. the maximum share of overall exposure that ought to be Other AFCs may finance multiple products for a single in a given region-sector-combination, and the ‘Actual’ customer segment, such as people engaged in, or linked to, column showing current portfolio concentration. smallholder agriculture. 36 There are a number of comparably easy-to-implement machine learning algorithms for clustering exercises. One of the most common ones is called “k-Means”. The current pandemic may give valuable insights on the vulnerability of different customer profiles to such shock. 38 We do not necessarily refer to a causal connection. The mere fact that two customers live in the same region makes them both prone to the same natural disaster. GE T TING REPA ID IN A S SE T FIN A NCE 5.4 C oncen t r at ion a nd di v ersific at ion 62 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management TABLE 7. Concentration limits and actual distribution Regional Total MSME Assets Durables Agriculture Transportation Personal Electronics Limit Actual Limit Actual Limit Actual Limit Actual Limit Actual Limit Actual Region 1 23% 32% 7% 9% 2% 4% 3% 3% 5% 6% 6% 10% Region 2 22% 22% 10% 8% 2% 2% 0% 0% 5% 5% 5% 7% Region 3 10% 12% 3% 3% 5% 6% 0% 0% 1% 2% 1% 1% Region 4 19% 18% 4% 5% 2% 5% 6% 2% 3% 2% 4% 4% Region 5 26% 16% 5% 3% 4% 3% 13% 5% 2% 1% 2% 4% Sector Total 100% 100% 29% 28% 15% 20% 22% 10% 16% 16% 18% 26% It is easy to see, looking at Region 1 or Personal LGD comes from the recovery analysis. Both can be Electronics in the above figure, how small over- calculated on the total portfolio or distinct sub-portfolios concentrations at a branch or product level can leave (products, segments, geographies). In any case, the PD is a company dangerously over-exposed on aggregate. It essentially a function of the current arrears status;38 EL is of crucial importance to establish who is responsible on an individual exposure go up once the arrears status for observing and reporting on limit breaches, then to deteriorates. The EaD is the outstanding exposure at the establish adequate measures once those breaches are moment of default, averaged across all relevant defaults. detected in order to avoid significant overexposures. All of As the term EL already implies, that is the amount a these roles and responsibilities can be outlined in policy company can expect to lose on a given portfolio under documents and reinforced by risk managers. ordinary (i.e., similar to the past) business conditions.39 Therefore, that amount should always be priced in. If it is not, assuming the average EL is 10 percent for an average 5.5 Expected Loss exposure, a company should increase the lease price and Unexpected Loss accordingly (to do that, divide 1 by the expected return, in Risk Management which is just the inverse of the EL. So (1/.9) - 1 = 11.1 percent) to achieve the same expected return. The EL that is priced should depend on the type of product, whether it C A L C U L AT I N G A N D U S I N G E X P E C T E D L O S S is new or used, what country it is financed in, etc. As EL As mentioned at the top of this chapter, the EL on a given increase or decrease, so should the interest rate. loan is the product of the probability of default (PD), the LGD and the EaD. The expected portfolio loss is the sum Provisioning of all individual expected losses. Companies are expected to account for their EL in their books at the time the loan/lease is originated, as EL = PD • LGD • EaD shown in Box 22 at the beginning of this chapter. The The PD is typically standardized to one year forward procedure of recognizing this amount as an expense is and can be taken from the transition matrix. The called provisioning. It is mandatory for regulated financial 38 As stated earlier, collection rate or internal credit scores can also be used. 39 Assuming a portfolio of 120 loans (with equal risk profiles) where on average every sixth loan defaults and further assuming the LGD to be 100% (hence no expected recoveries after default), that would be nothing else than rolling a die, where every [6] would show a default. Rolling the die 120 times (once for each loan) one would expect 20 defaults, with a standard deviation of (sqrt(120• (1/6)*(5/6)) = 4.1. GE T TING REPA ID IN A S SE T FIN A NCE 5.5 E x pec t ed lo s s a nd une x pec t ed lo s s in risk m a n a gemen t 63 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management institutions, and the standard reference framework on (2x or higher) of PAR/RAR and defaults. For PAYGo provisions is called IFRS 9.40 companies, a 25 percent reduction in Collection Rate. However, even for unregulated lenders, it is best and 2. Massive requests for restructuring, moratoria, etc. recommended practice to implement accounting 3. Limited opportunities to generate (healthy) new procedures for provisioning. It forces lenders to business, i.e., a rapid slowdown in growth rate. transparently report the quality of their portfolio and its EL, which in turn is an incentive for better assessment. 4. Liquidity problems for financial institutions as a result Provisioning also makes it harder to gloss over poorly of the above. aging receivables with lots of shiny new loans. An institution should regularly undergo internal stress tests to highlight vulnerable areas, especially effects UNEXPECTED LOSS AND STRESS TESTING on capital and liquidity, with an emphasis on short- If the risk profile for many or all loans increases term liquidity. Realistic scenario assumptions could simultaneously due to unexpected developments, the be informed by studying the effects of recent crises, EL formula calibrated on “the old parameters” may (e.g., the East African locust swarms of 2020). The tests fail spectacularly. For example, assume a portfolio is are usually built using portfolio quality shifts that are concentrated in a particular area in Ghana that is hit by a multiple standard deviations from a base scenario and drought destroying a significant part of all harvests in that macroeconomic developments (such as FX rate shifts) that region. In this case, the underlying assumptions from the are in the 99th percentile (i.e., severe and unlikely). PD estimation are no longer valid, and losses are likely to Each institution should define criteria under which a significantly exceed calculations of EL. stress test is considered passed, for example maintaining Stress testing a capital ratio of at least X percent, having Y percent of Stress testing is an appropriate and common approach that assets be highly liquid at all times, etc. If an institution lenders use to plan for catastrophic events. These tests fails a test, mitigating measures (boosting liquid asset simulate unlikely, but nevertheless possible, scenarios that levels, reducing leverage) need to be taken by management could have a devastating effect on portfolio quality and before the onset of a crisis event. repayment rates. The coronavirus pandemic—ongoing as Mitigating unexpected loss this Guide is being drafted—is one recent example. Other A particular scenario for AFCs to focus on is abrupt examples include: technological obsolescence (e.g., if a competitor issues a 1. Natural disasters of all kinds higher-quality, cheaper replacement). This could lead to a wave of clients ceasing payment and/or returning their 2. Terrorist attacks/political unrest, etc. equipment, while at the same time leading to higher LGD 3. Economic/financial crises as the outdated equipment will be difficult to redeploy It is beyond the scope of this paper to suggest realistic profitably (see Box 27). stress scenarios, as they must be specific to each individual One key measure can help mitigate vulnerability to such company and its context. However, a typical stress test events: keeping maturities short. A natural maximum might assume the following circumstances: maturity is the useful life of the asset, but keeping the 1. At minimum a significant delay in receiving lease term lower will lead to more sustainable businesses. contractual payments; a potentially serious increase Of course, responsible lending to the poor requires 40 The basic framework is relatively simple for standard loan and lease products. It distinguishes between three stages, which loans are classified as de- pending on their current arrears status. For a stage 1 loan, a 12-month PD shall be used to calculate EL. For Stage 2 loans, this turn into a life-time probability: if the remaining contractual life is two years, then multiplying the 12-months transition matrix with itself gives the PD estimate for 24 months. Stage 3 would be considered as defaulted, hence the assumed PD would always be 100 percent. See BIS (n.d.) for more. GE T TING REPA ID IN A S SE T FIN A NCE 5.5 E x pec t ed lo s s a nd une x pec t ed lo s s in risk m a n a gemen t 64 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management keep the resale value (blue line) higher than the outstanding  orrelated BOX 27. C losses for SolarCredit principal (various orange lines). This can be achieved with Assume a competitor has entered fictional company higher down payments and shorter maturities. In this SolarCredit’s SHS market with a cheaper, more example, even increasing the down payment would not be powerful system. Within several weeks, SolarCredit enough to always keep the outstanding principal below the customers begin to stop payment and switch to the resale value (the shaded area between 4 and 21 months). But new system. SolarCredit is able to recover many of reducing maturity (i.e., increasing monthly payments) along the defaulted systems, but the value of these is close with increasing the down payment accomplishes this goal to zero, given the improved technology on the market. (dark-brown line). These correlated spikes in PD and LGD are not Most of the defaults in asset financing happen relatively uncommon in asset finance. In the US subprime soon after disbursement, with a leveling off followed by crisis, a spike in defaults resulted in numerous forced a slow decline as clients default for various reasons. The home sales, which led to a significant decrease in longer the average tenor, the longer each loan is exposed to housing prices. LGD rates were therefore much a potential credit event, whether individual or widespread. higher than provisioned (Zhang et al., 2010). longer-than-average maturities so as to not overburden 5.6 Monitoring and Dashboards customers with high installments. However, the risk of Finally, all of the various analyses and indices described in technological change, combined with increasing defaults this chapter must be tracked, stored, and monitored. This and higher cost of funds, may endanger the business should happen on a daily basis for key risk staff, but should model of lenders with overly long tenors. also be presented and discussed by the company’s senior Figure 17 shows the decline in an asset’s value relative to management on a more periodic basis (weekly or monthly). a customer’s repayment. A company should try to always FIGURE 17. Looking at depreciation curves with different loan tenors and down payments Increasing down payment decreasing maturity Increasing down payment decreasing maturity 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Months Amount by which remaining Principal without down payment Principal with down payment payments exceed asset Principal with down payment resale value. Also known as Asset resale value and shorter maturity being “underwater”. GE T TING REPA ID IN A S SE T FIN A NCE 5.6 Moni t oring a nd d a shbo a rd s 65 Credit Risk in Asset Finance The Organization Product Design Credit Transaction Risk Portfolio Management K E Y R I S K I N D I C AT O R S • Transition Matrix (both month-on-month and Beyond PAR, RAR, and Collection Rates, we encourage year-on-year) institutions to use a wide set of KRIs to analyze portfolio • EL development/provisioning performance from different angles and to serve as early- • Repossession/Recovery Rate (in NPV) warning system for negative developments. KRIs are a set of indicators that condense portfolio quality data into easily • Disbursement growth/decline obtained numbers, which can then be tracked over time. • Diversification indicators Indicator calculation should require a minimum of • Early Defaults (segmented by agent or loan officer) human input; ideally all the work is done automatically in the system. Indicators should be linked to warning For all lenders, but particularly for AFCs, credit KRIs and alert thresholds, which are in turn linked to action cannot exclude operational risk. The following list triggers. The diversification matrix shown above (Table provides several examples: 7) can be understood as an example of a KRI with a set • Number of service tickets originated (tracked by week of warnings and triggers. A recommended action in case and month) of limit breach could be a disbursement slowdown in the • Average time to resolve service ticket (by week and respective area, or at least a stronger sales focus on the month, as well as MoM change) remaining areas. • Number of fraud incidents and value lost to fraud for KRIs should be calculated and easily accessible from a previous week/month centralized viewing platform, or credit dashboard. The • Number of repossession tickets originated dashboard should be built to enable regular management • Average time to repossess unit (from ticket origination) reporting, (preferably in real-time web-based dashboards) as well as ad hoc reviews in case of limit breaches. • Average time to refurbish and redeploy a repossessed unit Sample credit KRIs could include: • Number of inbound customer calls received • Percentage of outbound customer calls answered • PAR 1, 30, 90, including vintage curves • RAR at various ‘risk levels’ (e.g., 30 percent, 50 percent The list is certainly not complete. It is also important to and 70 percent)41 highlight that the company should be able to slice or filter • Write-off Rate these indicators across a number of parameters, e.g., per • Restructuring Rate including ‘cure rate’ (migration branch, agent, and then product. With that, managers from ‘at risk’ to ‘good’ status) should be able to zoom into an indicator to identify where problems occur and stem from. 41 PAYGO PERFORM recommends tracking RAR(CDU)>30 days and RAR(CR)<50% GE T TING REPA ID IN A S SE T FIN A NCE 5.6 Moni t oring a nd d a shbo a rd s 66 CONCLUSION Achieving sustainability in the asset finance sector business models, but there are no shortcuts: growth in will require long-term investment in risk management asset finance needs to be accompanied by an increased structures that underpin scale. investment in risk management. This means establishing the culture, strategy, appetite, and limits that enable good In an assessment of the systemic issues created in risk management, while at the same time investing in microfinance by a period of unprecedented growth, people: the analysts, auditors, compliance monitors, and Chen et al. (2010) wrote: “In the next decade, the focus risk managers who can ensure that an organization grows should be on sustainable growth…The recent delinquency at a pace that is commensurate with its appetite. crises are a reminder that microfinance remains a risk management business.” Their comments proved Poor households around the world need responsible prophetic: over the next three years, the microfinance consumer asset financing, and meeting that need means industry performed a deep self-assessment resulting in a taking risks. Managing those risks will be the difference renewed focus on risk management, increased adoption between failure and success in asset finance. of consumer protection principles such as the SMART Campaign and the publication of the Universal Standards of Social Performance Management. Evidence from the ‘Microfinance Barometer 2019’ shows what ‘sustainable growth’ can look like: the number of microfinance borrowers grew by seven percent year-on- year from 2012 to 2019, compared to almost 20 percent annual growth in the previous decade, while loan portfolios grew at a solid 11 percent annual average. For all that, concerns of overheating and overindebtedness still crop up, and likely always will, because lending is a risky business. Operational expense ratios fell while return on assets increased, although there are valid concerns that this may have come by moving away from lending to the poorest households. In short, the microfinance sector may be past its hype cycle and far from perfect, but it is doing important work at scale: 140 million people borrowed $124 billion from MFIs in 2018. This is what the future can hold for asset finance—that is, if practitioners and investors can learn from the past. Widespread scale and impact are possible for credit-based GE T TING REPA ID IN A S SE T FIN A NCE 67 REFERENCES Bandyopadhyay, Arindam. 2010. “Understanding the Effect of Deloitte. 2014. “Risk appetite frameworks: How to spot the genuine Concentration Risk in the Banks’ Credit Portfolio: Indian Cases”, article”, https://www2.deloitte.com/content/dam/Deloitte/au/ National Institute of Bank Management (NIBM), MPRA Paper Documents/risk/deloitte-au-risk-appetite-frameworks-0614.pdf No. 24822. https://mpra.ub.uni-muenchen.de/24822/1/MPRA_ Efficiency for Access Coalition. 2019. “2019 State of the Off- paper_24822.Pdf Grid Appliance Market.” https://efficiencyforaccess.org/ Batty, Mike, Sylvie Hulin, and Tom Fineis. N.d. “A Leading Practice publications/2019-state-of-the-off-grid-appliance-market-report Approach to Formulate a Risk Appetite Statement”. https://www.ssc. Faux, Zeke. February 12, 2020. “Tech Startups Are Flooding Kenya wisc.edu/~mbatty/Risk_Appetite.pdf With Apps Offering High-Interest Loans” Bloomberg Businessweek. Braniff, Lauren, and Xavier Faz. 2012. “Information Systems: https://www.bloomberg.com/news/features/2020-02-12/ A Practical Guide to Implementing Microfinance Information tech-startups-are-flooding-kenya-with-apps-offering-high-interest-loans Systems”, Technical Guide. Washington D.C.: CGAP. https://www. Fleming, Anne. 2018. City of Debtors: A Century of Fringe Finance. cgap.org/sites/default/files/CGAP-Technical-Guide-Information- Cambridge: Harvard University Press. Systems-Jan-2012.pdf Financial Stability Board (FSB). 2013. “Principles for An Effective Bruhn, Miriam, and Inessa Love. 2009. “The Economic Impact of Risk Appetite Framework” https://www.fsb.org/wp-content/ Banking the Unbanked: Evidence from Mexico”, World Bank Policy uploads/r_131118.pdf Research Working Paper 4981. Washington, D.C.: World Bank. https:// papers.ssrn.com/sol3/papers.cfm?abstract_id=1428628 Institute of Risk Management. 2017. “Risk Appetite Statements”. IRM: London. https://www.theirm.org/resources/ Calder, Lendol.1999. Financing the American Dream: A Cultural find-a-resource/2017-risk-appetite-statements/ History of Consumer Credit. Princeton: Princeton University Press. ITU. 2020. “World Telecommunication/ICT Indicators Database Chen, Lisa and Gregory Elliehausen. 2020. “The Cost Structure online” https://www.itu.int/pub/D-IND-WTID.OL-2020 of Consumer Finance Companies and Its Implications for Interest Rates: Evidence from the Federal Reserve Board’s 2015 Survey of Izaguirre, Juan Carlos, Michelle Kaffenberger, and Rafe Mazer. Finance Companies”, FEDS Notes No. 2020-08-12. https://doi. 2018. “It’s Time to Slow Digital Credit’s Growth in East Africa.” org/10.17016/2380-7172.2610, Available at SSRN: https://ssrn.com/ CGAP blog post, 25 September. https://www.cgap.org/blog/ abstract=3675659 its-time-slow-digital-credits-growth-east-africa Chen, Greg, Stephen Rasmussen, and Xavier Reille. 2010. “Growth and Khaki et al. (2021, forthcoming) Vulnerabilities in Microfinance”, Focus Note No. 61, Washington D.C.: Klein, Joyce, and Alan Okagaki. 2019. “The Price of Access: How CGAP. https://www.cgap.org/sites/default/files/CGAP-Focus-Note- Scale-Focused Microlenders are Pricing for Growth”, FIELD at the Growth-and-Vulnerabilities-in-Microfinance-Feb-2010.pdf Aspen Institute. Washington D.C.: Aspen Institute. https://www. Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlanda aspeninstitute.org/wp-content/uploads/2018/03/The-Price-of-Access- Ruthven. 2010. Portfolios of the Poor: How the World’s Poor Live on $2 a How-Scale-Focused-Microlenders-are-Pricing-for-Growth.pdf Day. Princeton; Princeton University Press. Kumaraswamy, Sai Krishna, Max Mattern, and Emilio Hernandez. Comptroller of the Currency Administrator of National Banks (OCC). 2020. “Assets Matter to Poor People”, Working Paper. Washington 1998. “Loan Portfolio Management: Comptroller’s Handbook” https:// D.C.: CGAP. https://www.cgap.org/research/publication/ www.lexissecuritiesmosaic.com/gateway/occ/bulletin/comptrollers- assets-matter-poor-people handbook_lpm.pdf Lascelles, David. “Microfinance Banana Skins 2008: Risk in a booming DeLoach, Jim. 2015. “The Importance of Risk Culture”, industry”, Centre for the Study of Financial Innovation. https://www. Blog Post, Corporate Compliance Insights. https://www. cgap.org/sites/default/files/researches/documents/CGAP-Microfinance- corporatecomplianceinsights.com/the-importance-of-risk-culture/ Banana-Skins-Mar-2008.pdf GE T TING REPA ID IN A S SE T FIN A NCE 68 Lascelles, David, and Sam Mendelson. 2009. “Microfinance Banana Waldron, Daniel. 2021. "The Social Impact Credit Trap in Asset Skins 2009: Confronting crisis and change”, Centre for the Study Finance." CGAP blog post, 04 May. https://www.cgap.org/blog/ of Financial Innovation. http://www.csfi.org.uk/Microfinance%20 social-impact-credit-trap-asset-finance Banana%20Skins%202009.pdf Wolfram et al. (2021 forthcoming) Menon, Shailesh. April 19, 2016. “Microfinance industry is out of World Bank. 2020. “World Development Indicators” Washington an unprecedent crisis, thanks to regulations, diligent borrowers” The D.C.: World Bank. https://databank.worldbank.org/source/ Economic Times. https://economictimes.indiatimes.com/industry/ world-development-indicators banking/finance/microfinance-industry-is-out-of-an-unprecedent- crisis-thanks-to-regulations-diligent-borrowers/articleshow/51886097. cms?utm_source=contentofinterest&utm_medium=text&utm_ campaign=cppst Murthy, SV Ramana. 2019. “Measuring Informal Economy in India _ Indian Experience”, Conference paper at Seventh IMF Statistical Forum. https://www.imf.org/-/media/Files/ Conferences/2019/7th-statistics-forum/seventh-imf-statistical-forum- ramana-murthy-presentation.ashx Open Risk Manual. N.d. https://www.openriskmanual.org/ Reille, Xavier, and Sarah Forster. 2008. “Foreign Capital: Investment in Microfinance Balancing Social and Financial Returns”, CGAP Focus Note No. 44. Washington D.C.: CGAP. https://www.cgap.org/ sites/default/files/CGAP-Focus-Note-Foreign-Capital-Investment-in- Microfinance-Balancing-Social-and-Financial-Returns-Feb-2008.pdf Rosenberg, Richard, Scott Gaul, William Ford, and Olga Tomilova. 2013. “Microcredit Interest Rates and Their Determinants: 2004–2011”, Forum No. 13. Washington D.C.: CGAP, Kfw, and MiX. https://www. cgap.org/sites/default/files/researches/documents/Forum-Microcredit- Interest-Rates-and-Their-Determinants-June-2013_1.pdf Rutherford, Stuart. 1999. “Savings and the Poor: The Methods, Use And Impact Of Savings By The Poor of East Africa”, MicroSave. Sotiriou, Alexander, Pepukaye Bardouille, Daniel Waldron, and Gianmaria Vanzulli. 2018. “Strange Beasts: Making Sense of PAYGo” Forum No. 14. Washington D.C.: CGAP & IFC. https://www.cgap. org/sites/default/files/Forum-Strange-Beasts-Jan-2018.pdf Srinivas, Hari. 2016. “14 Reasons why the Informal Credit Market is used by the Poor: Policy Implications for Microcredit Programmes in Developing Countries” GDRC Reseaarch Output E-111. Kobe, Japan: Global Development Research Center. Retrieved from https://www. gdrc.org/icm/14reason.html on Friday, 5 March 2021 Waldron, Daniel, Alexander Sotiriou, and Jacob Winiecki. 2019. “A Tale of Two Sisters: Microfinance Institutions and PAYGo Solar Companies.” Focus Note. Washington D.C.: CGAP. https://www.cgap.org/sites/ default/files/publications/2019_11_Focus_Note_Paygo_Two_ Sisters_2.pdf Waldron, Daniel. 2020. “The Breaking Point: How Warranties Support Sustainable Asset Finance” CGAP Blog. https://www.cgap.org/blog/ breaking-point-how-warranties-support-sustainable-asset-finance Waldron, Daniel, and Chris Emmott. 2018. “Off-Grid Solar Company Helping Customers Pay School Fees” CGAP Blog. https://www.cgap. org/blog/grid-solar-company-helping-customers-pay-school-fees GE T TING REPA ID IN A S SE T FIN A NCE 69 A NNE X 1 SAMPLE RISK APPETITE: STATEMENT FROM BHAR AT FINANCIAL INCLUSION The following is taken from “Annual Report of Bharat Financial through mitigating actions on a continuing basis. These Inclusion Limited (Formerly known as ‘SKS Microfinance are discussed with both the Management and the RMC. Limited’ for FY 2016-17” pages 39-42.42 Some of the risks relate to competitive intensity and the changing legal and regulatory environment. The RMC  ISKS A ND M A NAGEMENT I V. R of the Board reviews the risk management policies, in STR ATEGIES relation to various risks and regulatory compliance issues. Risk is an integral part of the Company’s business, and The Company identifies the following as key risks: sound risk management is critical to the success of the organisation. As a financial intermediary, the Company POLITICAL RISK is exposed to risks that are particular to its lending and The Company recognises political risk as one of the the environment within which it operates. The Company major risks facing the industry and believes that political has identified and implemented comprehensive policies risk can be mitigated through responsible lending and and procedures to assess, monitor and manage risk fair pricing by way of: throughout the Company. The risk management process • Lowest cost lender – The Company charges the lowest is continuously improved and adapted to the changing interest rates among NBFC-MFIs to its Borrowers. risk scenario and the agility of the risk management • Voluntary Cap on RoA from core lending – The process is monitored and reviewed for its appropriateness Company has voluntarily capped its returns from core in the changing risk landscape. The process of continuous lending at 3 percent. evaluation of risks includes taking stock of the risk landscape on an event-driven basis. • Robust Customer grievance redressal (CGR) Mechanism with Independent Ombudsman. The Company has an elaborate process for risk management. This rests on the three pillars of Business • Calibrated Growth – The Company’s growth strategy Risk Assessment, Operational Controls Assessment and aims to meet the requirements of its Members and also Policy Compliance Processes. Major risks identified by address concerns of various stakeholders. the businesses and functions are systematically addressed 42 See https://www.bseindia.com/bseplus/AnnualReport/533228/5332280317.pdf GE T TING REPA ID IN A S SE T FIN A NCE 70 CONCENTR ATION R ISK at 15%. The share of borrowing from top 3 banks reduced significantly from 61% in March 2013 to 24% in March The Company aims to avoid unbalanced concentration 2017. in both its loan portfolio and borrowings. To mitigate the concentration risk, the Company has a well-defined OPER ATIONA L R ISK geographic and borrower dependence norms. The core business of the Company is to provide collateral free loans in rural areas, and consequently, requires Geographic concentration norms: In order to mitigate enhanced operational risk management. To mitigate the risk of external intervention, concentration in any the operational risk, the Company adopts the following particular state, district or branch, as well as to manage strategy. non-payment risk, the Company has implemented the following limits: Integrated cash management system: (A) Disbursement Related Caps: The disbursement • The Company’s transactions with borrowers are limits stipulate each state to entail less than 15% of the predominantly in cash, making cash management total disbursements for the Company (except states of an important element of the business. To reduce the Karnataka and Odisha which have a 20% limit); each potential risks of theft, fraud and mismanagement, district to entail less than 3% of the total disbursements the Company has been implementing an integrated for the Company (except districts in states of Karnataka cash management system since July 2009 which is and Odisha which have a 4% limit); each branch to entail operational in approximately 1,328 of its branches, as on less than 1% of the total disbursements for the Company March 31, 2017. The system utilises an Internet banking (except branches in states of Karnataka and Odisha which software platform that interfaces with various banks to have a 1.25% limit); no disbursements to be made by provide the Company with real-time cash information branches that have an NPA of more than 1% or collection for these branches and the loan activity therein. efficiency of less than 95%. • The Company believes this integrated system augments (B) Portfolio Outstanding Related Caps: its MISs and facilitates its bank reconciliations, audits – Gross Loan Portfolio: Each state to ensure that and cash flow management. The system also reduces its Gross Loan Portfolio will not exceed 75% of the errors. Company’s net worth (except states of Karnataka, • Product and process Design – The Company Odisha and Maharashtra which have a 100% limit); each adopts a standardised approach to product design district to maintain that its Gross Loan Portfolio will not and operational procedures at the branches and exceed 5% of the Company’s net worth (up to 5% of the centers to enable predictability of transactions, as a operating districts may go up to 10% of the Company’s risk mitigant. net worth); each Branch to maintain that its Gross Loan • ISO Certified Internal audit – The Company Portfolio will not exceed 1% of the Company’s net worth has adequate controls and processes in place (up to 5% of the operating branches may go up to 2% of with respect to its operations. The Internal Audit the Company’s net worth). department acts as the third line of defence by – Loan Portfolio Outstanding: Each state to ensure monitoring adherence to controls and processes and that its Gross Loan Portfolio will not exceed 15% of the provides inputs for strengthening risk management. Company’s total portfolio (except states of Karnataka and The Internal Audit function has been certified with Odisha which have a 20% limit). The caps are subject to ISO 9001:2008. tolerance of 10%. Borrowing dependence norms: In order to reduce dependence on a single Borrower, the Company has adopted a cap on borrowing from any single credit granter GE T TING REPA ID IN A S SE T FIN A NCE 71 LIQUIDIT Y R ISK The Company places significant importance on liquidity management and has a bias for liquidity, mainly to address the operational requirements and corporate commitments. Along with funding strategy and asset liability management, the Company has well defined liquidity metrics, including cash burn, optimal liquidity and liquidity cap test, to ensure sufficient liquidity in line with business requirements and aid risk mitigation. I NTER NA L CONTROL SYSTEMS V.  A ND THEIR A DEQUACY The Company has a well-established and strong internal controls with well-designed systems, policies and procedures to maintain financial discipline. The Company’s Internal Control Systems are commensurate with the nature of its business and the size and complexity of its operations. Based on the guidelines received on various issues of control from the Reserve Bank of India and the Government of India, the Company’s Board of Directors and the Audit Committee of the Board are a part of the Internal Control System for better compliance at all levels. The Internal Audit Department of the Company is an independent function which ensures, checks and evaluates operational risks, internal controls, internal financial controls, adherence of systems, policies and procedures by conducting inspection of branches/ offices. These are routinely tested and cover all Branches, Regional Offices and the Head Office. Significant audit observations and follow-up actions, thereon, are reported to the Audit Committee. The Audit Committee of the Board oversees the Internal Audit function of the Company. The Audit Committee reviews the adequacy and effectiveness of the Company’s Internal Control System, including Internal Financial Controls and monitors the implementation of audit recommendations, including those related to strengthening of the Company’s risk management policies and systems. The Audit Committee monitors compliance with inspection and audit reports of the Reserve Bank of India, other regulators and statutory auditors. GE T TING REPA ID IN A S SE T FIN A NCE 72 A NNE X 2 CREDIT RISK POLICY OUTLINE Preamble This policy is part of the institution’ Risk Management Framework. It has been approved by the •  Board of Directors. Credit risk is naturally the dominant risk dimension for the institution, due to … (e.g., refer to risk •  exposure, probability of occurrence and potential impact compared to other risks incurred). • The institution engages credit risk deliberately and with the proper controls. Target Markets • Description of institution’s vision and mission Allowable clients, markets, products that the institution generally seeks to interact with and that may •  carry credit risk • Overall lending philosophy and credit risk appetite Social Performance • Desired economic and social development impact of lending operations • High-level impact measurement framework, indicators and achievement targets • Commitment to responsible finance and client protection principles Lease Products • Define credit product term sheets: · Eligible clients, · Eligible industries, · Amounts, rates, and fees, · Documentation requirements Credit Assessment • Evaluation and analysis of loan application Statistical measures: statistical scoring, internal rating, expert scores, bureau scores, external •  rating, etc., for which clients and products at what intervals Credit Approval • Define approval process by product · Delegated authorities · Approval limits · Verifications and operational risk controls Credit Monitoring • For various client and product groups, define: · Monitoring methods · Responsible parties · Specified monitoring intervals Portfolio • Size concentration measures and limits Management • Industry and geographic diversification • Expected Loss (EL)/Unexpected Loss (UL) calculation • Stress-testing • Portfolio risk monitoring and controlling • Early identification of negative developments, adverse trends GE T TING REPA ID IN A S SE T FIN A NCE 73 Arrears Management • Arrears management organization • Arrears analytics • Responsible collections practices · Efficient collections actions · Remote switch-off · Repossession · Restructuring and work-out of lease contracts • Provisioning Measuring KRIs at a company level to be set, targeted on a quarterly or monthly basis, monitored by the CRO, and Performance reported to Executives and Board GE T TING REPA ID IN A S SE T FIN A NCE 74 A NNE X 3 CALCUL ATING EFFECTIVE INTEREST R ATES Typically, companies offer cash sales as well as lease sales in parallel. Lease sale is typically more expensive when purely summing up the installments and the down payment. However, so far none of the assessed companies has implemented a consistent risk accounting or pricing approach for this. In MS Excel, effective interest rates are calculated as follows: NominalMonthlyInterestRate: = RATE (LeaseTermInMonths,MonthlyContractualInstallment, – CashPrice + InitialDeposit,0) = IRR (ArrayOfMonthlyPayments) EffectiveInterestRate p.a.: = (1 + NominalMonthlyInterestRate) ^ 12-1 Assume a product with the following features: • CashPrice = 10,000 • DownPayment | InitialDeposit = 2,500 • Term = 12 months • MonthlyContractualInstallment = 750 The cashflows from a customer’s point of view are as follows (in thousand): M0 M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 Sum 7.5 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -1.5 Note that the M0 cashflow is +10,000 – 2,500 = 7,500. Of course, the positive number in reality is not a real cashflow, as the customer receives the asset, instead. = IRR (M0:M12) returns 2.9% >> nominal monthly interest rate. = RATE (12,750,-10000+2500,0) also returns 2.9%. = [1+RATE (12,750,-10000+2500,0)] ^12-1 returns 41.3% >> effective yearly interest rate. GE T TING REPA ID IN A S SE T FIN A NCE 75 Check: Real lease cashflows from a customer’s point of view: M0 M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 Sum -2.5 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -.75 -11.5 Customer cashflows discounted with the calculated effective yearly interest rate (41.3%): M0 M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 Sum -2.5 -.729 -.708 -.688 -.668 -.649 -.631 -.613 -.596 -.579 -.562 -.546 -.531 -10.0 Formula for M4: MonthlyContractualInstallment / (1+ieff)(month/12) = 750 / (1+41.3%)(4/12) = -.688 Summing up the real lease cashflows discounted by the effective interest rate gives the cash price of 10,000. GE T TING REPA ID IN A S SE T FIN A NCE 76 A NNE X 4 DATA DICTIONARY We propose to think about a data dictionary, which may start as a compilation and grouping of all potentially interesting data elements, and their features, such as • Element name • Element description • Element type (date, integer, free string, dropdown, [Y/N], etc.) • Element Purpose (primary and secondary – e.g., KYC requirements, customer default prediction, etc.) • Entity/Property of (e.g., client, address, home, etc.) • Gathering/calculation logic • Verification logic • Data owner (e.g., sales agents, risk manager, etc.) • Data recording and update (when collected, how often, etc.) • Meta data recorded (e.g., time stamps, time needed to complete surveys, etc.) This dictionary should not be developed exclusively by the Risk Department, but understood as a joint exercise between all operational units. The ultimate goal of such task is to establish an advanced and to the extent possible complete and correct database to serve multipurpose analytical tasks, such as credit risk assessment, estimating marketing potential, fraud detection, measurement of recovery success, etc. GE T TING REPA ID IN A S SE T FIN A NCE 77 A NNE X 5 NPV REPOSSESSIONS Assume a company has two customers A and B, both interested into leasing appliance X. X has to be imported from China, which creates costs for the good itself (purchase price) as well as expenses for transportation, customs, storage, etc. (altogether “costs of goods sold” – COGS). Such appliance then is sold (on lease) to customer A for a price of P, which is defined as COGS * (1 + GM) with GM being the gross margin (incl. lease interest). Assume A defaults and X A is repossessed from her. Instead of selling a new appliance (also to be imported from China, etc.), the just repossessed item XA is sold to customer B (potentially with a discount, as this is not a brand-new item anymore). What is important to understand is that customer B was not “created” because of A’s default – instead B would have received X in any case. What the company saves in this special case is just the COGS, which accordingly must be the base value for the recovery calculation; hence (1 + GM) shall not be considered as recovery, as this part would have been “earned” independently from A’s default. The expected recovery calculation may look then as follows: NPV Repo = UnitCost • RedeplRate – RefurbCosts – ContrBNValue • RedeplRate • ResaleDisc RepoDate – DefaultDate + StorageTime 1+ ( effInterest 12 ) 30 with • UnitCost = COS incl. shipping, customs, etc. • RepoDate = Repossession date • RedeplRate = Redeployment rate, hence the probability • DefaultDate = Default date that the item will be sold again • StorageTime = Storage time, the time between • RefurbCosts = Refurbishment costs, hence the repossession and re-sale in days (estimated) costs to bring the item back to a saleable condition The redeployment rate may differ from item to item and is • ContrBNValue = Contractual buy-now value, hence the certainly a function of the general condition of the item, cash price of a brand-new item of the same kind the age and particularly the offered resale discount. As confirmed by multiple SHS providers, the redeployment • ResaleDisc = Resale discount, hence the (estimated) rate often decreases for items that have been repossessed percentaged sale discount of the repossessed item in from another customer before (hence on which two comparison to a brand-new item of the same kind different customers defaulted already). A proper recovery • effInterest = Effective annual interest rate of the lease logic should incorporate that decline. contract GE T TING REPA ID IN A S SE T FIN A NCE 78 A NNE X 6 MULTIPLYING TR ANSITION MATRICES Multiplying a matrix built on i months with one built on j months would result in a matrix of i+j months. Assume, one only has a one-month matrix (established empirically), multiplying this with itself gives a two months forecast. That one multiplying with itself results in a four months matrix, which can easily used to calculate eight months, etc. The following illustrations exemplify the mathematical principle of matrix multiplication – on paper as well as in MS Excel with the =MMULT() array function:43 ( 5 7 6 8 ) ( 1 2 3 4 ) ( 19 22 43 50 ) ( 1 • 5 + 2 • 7 3 • 5 + 4 • 7 1•6+2•8 3•6+4•8 ) It may make sense to apply that logic for different customer segments, regions or other features, should there be significant deviations in the results. 43 Note, that matrices can only be multiplied if the number of columns of the first one equals the number of rows of the second one. Therefore, make sure that your input transition matrix has an equal number of rows and columns. The typically means to manually include one line for “Settled” and one for “Write offs”, which both contain one value 100% and all others zero (compare the transition matrix example above – lines “Settled” and “w/o”). GE T TING REPA ID IN A S SE T FIN A NCE 79